Method for predicting and optimizing line loss of three-phase unbalanced low-voltage transformer area
By establishing a line loss sensitivity matrix and prediction model, and using the active and reactive output increments of the intelligent soft switch for model prediction control, the problem of slow equipment response speed in the active distribution network in response to the changes of distributed power supplies is solved, and the rapid and real-time optimization control of line loss in three-phase unbalanced low-voltage platform area is achieved, and the distribution network voltage quality and optimization effect are improved.
Patent Information
- Application Number
- CN202411607819.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-12
- Publication Date
- 2025-03-18
AI Technical Summary
When the existing active distribution network responds to the rapid output changes of distributed power, the equipment response speed is slow and the output cannot be adjusted in real time; the existing real-time optimization methods are difficult to adjust the optimization strategy when the output and load of new energy are fluctuating rapidly, resulting in poor optimization results; the SOP real-time control method has problems such as uncertain external conditions and nonlinear control results during the line loss optimization process of three-phase unbalanced platform area.
By establishing a linear loss sensitivity matrix and prediction model, using the active and reactive output increments of the intelligent soft switch as the control variables, the model prediction control method is used to regulate the active and reactive transmission power of the SOP, rolling optimization of the line loss of the three-phase unbalanced platform area, and correct the prediction model deviation during the optimization process.
It realizes rapid and real-time optimization control of line losses in three-phase unbalanced low-voltage platform areas, improves the distribution network voltage quality, enhances the response ability to distributed power supplies, and improves the optimization effect.
Smart Images
Figure CN119651543A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for predicting and optimizing line loss in a three-phase unbalanced low-voltage area, which is used in the field of line loss control. Background Art
[0002] In medium and low voltage distribution networks, the form of power loads is flexible and complex, and the disordered access of a large number of single-phase loads causes the problem of three-phase imbalance in the network. For the grid side, three-phase imbalance will increase network losses and reduce economic efficiency. For the load side, the zero-sequence magnetic flux generated by three-phase imbalance will lead to a series of problems such as increased equipment losses, reduced efficiency, overheating, and shortened life. In recent years, smart soft switches (soft open point, SOP) have been widely used in active distribution networks due to their continuous power control and flexible control methods. As a fully controlled power electronic device, SOP can replace some traditional interconnecting switches in the distribution network, accurately, quickly and flexibly control the active and reactive power of the connected feeders, and realize the function of active distribution network optimization.
[0003] Traditional statistical line loss calculation methods mainly include average current method, root mean square current method, maximum load loss hour method, loss factor method, etc. Since the traditional theoretical line loss calculation principle is simple and easy to calculate, it has been widely used in the domestic distribution network line loss. However, with the continuous improvement of the requirements for line loss calculation results, this traditional calculation method has assumptions and estimates, and the calculation results are difficult to meet the requirements of modern enterprises for line loss management. At present, there are many methods for calculating line loss, such as network loss calculation method based on load measurement and theoretical calculation, and a Newton-Raphson method that uses load power to calculate low-voltage distribution station area line loss. They are all theoretical calculation methods based on the actual power meter reading of users, and are highly dependent on network structure, line model and length information. Regression analysis methods are also used in the calculation of substation line loss, but the regression equation is difficult to determine and the calculation accuracy is low. The SOP local control method based on multi-layer perceptron, its model training is completely based on historical data, and it cannot be effectively optimized when the operating conditions exceed the sample set. The above methods perform real-time adjustments through the voltage-reactive power droop curve, and have limited control capabilities and are all passive optimization methods. It is difficult to adjust the optimization strategy when the output of renewable energy fluctuates rapidly, resulting in poor optimization effects.
[0004] The present invention mainly aims to solve the following three problems:
[0005] 1. At present, the active distribution network controls primary equipment such as on-load voltage regulators and switchable capacitor banks. These primary equipment have slow response speeds and cannot continuously adjust output, making it difficult to respond to rapid output changes of distributed power sources in real time.
[0006] 2. The existing real-time optimization method uses a droop curve, which makes it difficult to adjust the optimization strategy when the output and load of renewable energy fluctuate rapidly, resulting in poor optimization results.
[0007] 3. The existing SOP real-time control method has problems such as uncertain external conditions and nonlinear control results in the process of line loss optimization in three-phase unbalanced substations. Summary of the invention
[0008] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for predicting and optimizing line losses in three-phase unbalanced low-voltage substations. The distribution network line loss sensitivity is obtained by combining the distribution network topology and long-term measurement data, and a prediction model for distribution network line losses is established. The model predictive control method is adopted in the real-time optimization stage. By adjusting the active and reactive transmission power of the SOP, the line losses in the three-phase unbalanced substations are rolling optimized, and the prediction model deviation is corrected during the optimization process.
[0009] A technical solution to achieve the above purpose is: a method for predicting and optimizing line loss in a three-phase unbalanced low-voltage area, comprising the following steps:
[0010] Step 1: Establish a line loss sensitivity matrix and a line loss prediction model, taking the active and reactive output increments of the intelligent soft switch as control variables;
[0011] Step 2: Taking the intelligent soft switch operation cost, the distribution network three-phase imbalance and the operation loss as the minimum objective function, and the node voltage and power meeting the operation requirements as the constraint, a rolling optimization model is established and the control variable sequence for the next N time periods is solved;
[0012] Step 3: only issue the first instruction of the control variable sequence, i.e., the active and reactive transmission power of the intelligent soft switch at time k+1;
[0013] Step 4, measure the actual power of the system at time k+1 as the initial value at time k+1, use the error as the correction amount of the prediction model, input it into the correction model, perform a new round of optimization, obtain a new line loss sensitivity matrix, and return to step 1.
[0014] Furthermore, in step 1, the line loss sensitivity matrix is established as follows:
[0015] The active power loss on the low-voltage power line is shown in formula (1):
[0016]
[0017] in represents a three-phase line, j is the node number, They are the resistance of the three-phase line and the active and reactive power flowing through the line. is the line voltage, which is approximately the voltage of the upstream node of the line, and T is the set of distribution network nodes;
[0018] The node voltage can be approximately regarded as a constant under a small power disturbance. After the first-order Taylor expansion of equation (1), the change in active line loss can be obtained as shown in equation (2):
[0019]
[0020] ΔP loss is the change in distribution network loss, They are the active power injected into the node, the reactive power injected, and the voltage injected. ΔP and ΔQ are the changes in the active power and reactive power injected into the node. is the sensitivity of line loss to changes in active power and reactive power, which can be obtained using only quantities related to feeder voltage and distribution network impedance, as shown in equations (3) and (4).
[0021]
[0022] P j , Q j is the active and reactive power flowing through the line, U j is the line voltage, R j is the node resistance.
[0023] Furthermore, in step 1, the line loss prediction model is a power grid line loss prediction model established based on the long-term optimization results and local measurement data for a certain low-voltage substation topology and operating state. At time k, the distribution network voltage and line loss value at the future k+i time after the intelligent soft switch transmission power changes are predicted, and the control variable of the loss prediction model at time k is recorded as Δu(k)=[ΔP(k),ΔQ n (k),ΔQ m (k)], where ΔP(k), ΔQ n (k) and ΔQ m (k) represents the active transmission power of the intelligent soft switch and the reactive transmission power at both ends, respectively, and m and n are the node numbers at both ends of the intelligent soft switch.
[0024] Furthermore, the line loss prediction model is as follows:
[0025]
[0026] is the sensitivity matrix of distribution network line loss to control variables;
[0027]
[0028] Where i = 1,…,N p -1,N p represents the prediction step length; Ploss (k) represents the distribution network loss at time k; P loss (k+i|k) represents the loss value predicted at time k at future time k+i; Δu(k+i|k) represents the control variable increment predicted at time k at future time k+i, which is the control variable of the predictive control model.
[0029] Furthermore, in step 2, the objective function is to minimize the intelligent soft switch operation cost, the three-phase imbalance of the distribution network, and the operation loss:
[0030]
[0031] Where ΔP ope With ΔP loss are operating cost and loss cost respectively, and α is the line loss coefficient.
[0032] Furthermore, when the voltage deviation is large, the line loss coefficient will decrease, reducing the weight of the line loss term in the objective function. The line loss coefficient is specifically shown in the following formula:
[0033]
[0034] V dev is the voltage deviation.
[0035] Furthermore, in step 2, the rolling optimization model is specifically:
[0036] At the same time, considering the impact of node injected active power ΔP and reactive power ΔQ on the distribution network operation state, the active and reactive transmission power of the intelligent soft switch is used as the control variable Δu as shown in formula (8):
[0037] Δu i =[ΔP i ΔQ m,i ΔQ n,i ] (8)
[0038] Optimize the operation cost ΔP in the objective function ope and loss cost ΔP loss As shown below,
[0039]
[0040]
[0041] Where: P p , P q They are the control cost of active and reactive power by adjusting the intelligent soft switch and the loss cost of line loss, T represents the matrix conversion, and f v is the voltage unbalance objective function, is the value at node i Phase voltage value, A, B, C represent different phases, and N is the number of distribution network nodes.
[0042] Furthermore, due to the high efficiency of intelligent soft switching, assuming that P p =P q relationship.
[0043] Furthermore, in step 2, the constraints that the node voltage meets the operation requirements are specifically:
[0044] Voltage upper and lower limit constraints: V min With V max are the lower and upper limits of the node voltage at both ends of the intelligent soft switch, as shown in formula (13),
[0045] V min (k+i)≤V a,b,c (k+i|k)≤V max (k+i) (13)
[0046] Considering the operating state limitation and maximum capacity S of the intelligent soft switch max And the upper limit of active and reactive output P max , Q max , the transmission power constraint of the intelligent soft switch is as follows,
[0047]
[0048] Furthermore, in step 4, the correction model is input and a new round of optimization is performed as follows:
[0049] When a new round of predictive control is carried out, it is necessary to update the injected power items of the nodes at both ends of the smart soft switch in the sensitivity matrix according to the change of the working state of the smart soft switch to obtain a new loss sensitivity matrix. If the distribution network is in a stable operating state, the deviation caused by the change of the working state of other nodes in the distribution network is ignored, and the line loss sensitivity is corrected according to the change of the working state of the smart soft switch, as shown below.
[0050]
[0051]
[0052] R is the line resistance where the node is located, and U is the node voltage.
[0053] This patent aims at optimizing the line loss in three-phase unbalanced low-voltage substations, taking SOP transmission power as the control object. When the distribution network operation conditions change, the control strategy can be actively changed to ensure the optimization effect.
[0054] Compared with the existing technical solutions, the advantages of the present invention are listed as follows:
[0055] 1. Introduce SOP to optimize line loss in low-voltage areas, with continuous power control and flexible control methods, to achieve accurate, fast and flexible control of active and reactive power of connected feeders.
[0056] 2. During the optimization process, only the electrical information of the nodes at both ends of the SOP needs to be collected. By optimizing the voltage of key nodes, the distribution network voltage can be quickly and real-time optimized and controlled to improve the overall voltage quality;
[0057] 3. Simultaneously regulate the active and reactive power transmitted by SOP, giving full play to the regulation capability of intelligent soft switch. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] Figure 1 The present invention is a method flow chart of a method for predicting and optimizing line losses in three-phase unbalanced low-voltage areas. DETAILED DESCRIPTION
[0059] In order to better understand the technical solution of the present invention, the following is a detailed description through specific embodiments:
[0060] See also Figure 1 The present invention provides a method for predicting and optimizing line loss in a three-phase unbalanced low-voltage area, comprising the following steps:
[0061] Step 1: Establish a line loss sensitivity matrix and a line loss prediction model, taking the active and reactive output increments of the intelligent soft switch as control variables.
[0062] In order to establish the optimization problem control variables Δu and loss P loss The prediction model between the two needs to introduce the sensitivity matrix to the voltage and line loss changes caused by the change of node injection power, and establish a linear prediction model. The active power loss on the power line in the low-voltage area is shown in formula (1).
[0063]
[0064] in represents a three-phase line, j is the node number, They are the resistance of the three-phase line and the active and reactive power flowing through the line. is the line voltage, which is approximately the voltage of the upstream node of the line, and T is the set of distribution network nodes.
[0065] The node voltage can be approximately regarded as a constant under a small power disturbance. After performing a first-order Taylor expansion on formula (1), the change in active line loss can be obtained as shown in formula (2).
[0066]
[0067] ΔP lossis the change in distribution network loss, They are the active power injected into the node, the reactive power injected, and the voltage injected into the node respectively. ΔP and ΔQ are the changes in the active power and reactive power injected into the node respectively. It is the sensitivity of line loss to changes in active power and reactive power, and can be calculated using only quantities related to feeder voltage and distribution network impedance, as shown in equations (3) and (4).
[0068]
[0069] P j , Q j is the active and reactive power flowing through the line, U j is the line voltage, R j is the node resistance.
[0070] For a certain low-voltage substation topology and operating state, a power grid line loss prediction model is established based on the long-term optimization results and local measurement data. At time k, the distribution network voltage and line loss values at the future k+i time after the SOP transmission power changes are predicted. The control variable of the loss prediction model at time k is recorded as Δu(k)=[ΔP(k),ΔQ n (k),ΔQ m (k)], where ΔP(k), ΔQ n (k) and ΔQ m (k) represents the active transmission power of SOP and the reactive transmission power at both ends, m and n are the node numbers at both ends of SOP. The prediction model of line loss is shown below.
[0071]
[0072] is the sensitivity matrix of distribution network loss to control variables.
[0073]
[0074] Where i = 1,…,N p -1, N p represents the prediction step length; P loss (k) represents the distribution network loss at time k, which is obtained through superior communication every 15 minutes; P loss (k+i|k) represents the loss value predicted at time k at future time k+i; Δu(k+i|k) represents the control variable increment predicted at time k at future time k+i, which is the control variable of the predictive control model.
[0075] Step 2: Taking the intelligent soft switch operation cost, the three-phase imbalance of the distribution network and the minimum operation loss as the objective function, and the node voltage and power meeting the operation requirements as the constraint, a rolling optimization model is established and the control variable sequence for the next N time periods is solved.
[0076] During the rolling optimization process, the SOP operation cost and the line loss in the low-voltage area are minimized while ensuring that the voltages at both ends of the nodes meet the operating requirements. Based on the long-term optimization results, the real-time optimization objective function based on model predictive control is established as shown in formula (7).
[0077]
[0078] Where ΔP ope With ΔP loss are the operating cost and loss cost respectively, and α is the line loss coefficient. Different from the existing real-time optimization model based on droop control, the present invention proposes a voltage optimization model based on model predictive control (MPC), which can simultaneously consider the impact of node injected active power ΔP and reactive power ΔQ on the distribution network operation state, and take the active and reactive transmission power of SOP as the control variable Δu as shown in formula (8).
[0079] Optimize the operation cost ΔP in the objective function ope and loss cost ΔP loss As shown below.
[0080]
[0081] Where: P p , P q They are the control cost of active and reactive power by adjusting the intelligent soft switch and the loss cost of line loss, T represents the matrix conversion, and f v is the voltage unbalance objective function, is the value at node i Phase voltage value, A, B, C represent different phases, N is the number of distribution network nodes. Due to the high efficiency of intelligent soft switch, assuming P p =P q relationship.
[0082] In order to ensure that the voltage of the nodes at both ends of the SOP operates in the optimal range, different control strategies are adopted according to the different states of the nodes. The control measures are switched by controlling the value of the line loss coefficient α in the objective function. dev When it is larger, the line loss coefficient will decrease, reducing the weight of the line loss term in the objective function, as shown in formula (12).
[0083]
[0084] Among them, the voltage deviation V dev The voltage deviation level of the distribution network is evaluated as shown in the following formula.
[0085]
[0086] and U thr are the upper and lower limits of the allowable voltage, n is the number of distribution network nodes, and t is the optimization round.
[0087] By adding upper and lower voltage limits to constrain V min With V max , to optimize the voltage of the nodes at both ends of the SOP, as shown in formula (13).
[0088] V min (k+i)≤V a,b,c (k+i|k)≤V max (k+i) (13)
[0089] Consider the operating state limit and maximum capacity S of SOP max And the upper limit of active and reactive output P max , Q max , the transmission power constraints on SOP are as follows.
[0090]
[0091] Step 3: only issue the first instruction of the control variable sequence, that is, the active and reactive transmission power of the intelligent soft switch at time k+1.
[0092] Step 4, measure the actual power of the system at time k+1 as the initial value at time k+1, use the error as the correction amount of the prediction model, input it into the correction model, perform a new round of optimization, obtain a new line loss sensitivity matrix, and return to step 1.
[0093] Since the sensitivity matrix is generated based on long-time scale measurement results, when a new round of predictive control is carried out, the injected power items of the nodes at both ends of the SOP in the sensitivity matrix need to be updated according to the change of the SOP working state to obtain a new loss sensitivity matrix. In addition, the operation of the distribution network is constantly changing. When performing real-time control, it is necessary to correct the new round of prediction model according to the deviation of the previous round of prediction control to improve the prediction accuracy of the grid voltage and line loss prediction model.
[0094] If the distribution network is in stable operation, ignoring the deviation caused by changes in the working status of other nodes in the distribution network, the line loss sensitivity is corrected according to the change in the SOP working status, as shown below.
[0095]
[0096] R is the line resistance where the node is located, and U is the node voltage.
[0097] As a new type of power electronic device, SOP has highly controllable real-time transmission power, and the use of model predictive control (MPC) for control can achieve better optimization effects. The three-phase unbalanced calculation has very high real-time requirements. The real-time optimization method usually only requires local measurement information to complete the control and realize the line loss optimization of the low-voltage substation. The present invention proposes a model predictive control method based on line loss sensitivity. By constructing a prediction model, the impact of the SOP power change on the operating state of the substation is determined. During the optimization process, the present invention only needs to collect the electrical information of the nodes at both ends of the intelligent soft switch, and by optimizing the three-phase voltage of the key nodes, it can realize the rapid and real-time optimization control of the line loss in the three-phase unbalanced substation.
[0098] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.
Claims
1. A method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas, characterized in that: The steps include: Step 1: Establish a line loss sensitivity matrix and a line loss prediction model, taking the active and reactive output increments of the intelligent soft switch as control variables; Step 2: Taking the intelligent soft switch operation cost, the distribution network three-phase imbalance and the operation loss as the minimum objective function, and the node voltage and power meeting the operation requirements as the constraint, a rolling optimization model is established and the control variable sequence for the next N time periods is solved; Step 3: only issue the first instruction of the control variable sequence, i.e., the active and reactive transmission power of the intelligent soft switch at time k+1; Step 4, measure the actual power of the system at time k+1 as the initial value at time k+1, use the error as the correction amount of the prediction model, input it into the correction model, perform a new round of optimization, obtain a new line loss sensitivity matrix, and return to step 1.
2. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1 is characterized in that: In step 1, the line loss sensitivity matrix is established as follows: The active power loss on the low-voltage power line is shown in formula (1): in represents a three-phase line, j is the node number, They are the resistance of the three-phase line and the active and reactive power flowing through the line. is the line voltage, which is approximately the voltage of the upstream node of the line, and T is the set of distribution network nodes; The node voltage can be approximately regarded as a constant under a small power disturbance. After the first-order Taylor expansion of equation (1), the change in active line loss can be obtained as shown in equation (2): ΔP loss is the change in distribution network loss, They are the active power injected into the node, the reactive power injected, and the voltage injected. ΔP and ΔQ are the changes in the active power and reactive power injected into the node. is the sensitivity of line loss to changes in active power and reactive power, which can be obtained using only quantities related to feeder voltage and distribution network impedance, as shown in equations (3) and (4). P j , Q j is the active and reactive power flowing through the line, U j is the line voltage, R j is the node resistance.
3. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1 is characterized in that: In step 1, the line loss prediction model is a grid line loss prediction model established based on the long-term optimization results and local measurement data for a certain low-voltage substation topology and operating state. At time k, the distribution network voltage and line loss value at the future k+i time after the intelligent soft switch transmission power changes are predicted. The control variable of the loss prediction model at time k is recorded as Δu(k)=[ΔP(k),ΔQ n (k),ΔQ m (k)], where ΔP(k), ΔQ n (k) and ΔQ m (k) represents the active transmission power of the intelligent soft switch and the reactive transmission power at both ends, respectively, and m and n are the node numbers at both ends of the intelligent soft switch.
4. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 3 is characterized in that: The line loss prediction model is as follows: is the sensitivity matrix of distribution network line loss to control variables; Where i = 1,…,N p -1, N p represents the prediction step length; P loss (k) represents the distribution network loss at time k; P loss (k+i|k) represents the loss value predicted at time k at future time k+i; Δu(k+i|k) represents the control variable increment predicted at time k at future time k+i, which is the control variable of the predictive control model.
5. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1, characterized in that: In step 2, the objective function is to minimize the operation cost of the intelligent soft switch, the three-phase imbalance of the distribution network, and the operation loss: Where ΔP ope With ΔP loss are operating cost and loss cost respectively, and α is the line loss coefficient.
6. A method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 5, characterized in that: When the voltage deviation is large, the line loss coefficient will decrease, reducing the weight of the line loss term in the objective function. The line loss coefficient is specifically shown in the following formula: V dev is the voltage deviation.
7. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1, characterized in that: In step 2, the rolling optimization model is as follows: At the same time, considering the impact of node injected active power ΔP and reactive power ΔQ on the distribution network operation state, the active and reactive transmission power of the intelligent soft switch is used as the control variable Δu as shown in formula (8): Δu i =[ΔP i ΔQ m,i ΔQ n,i ](8) Optimize the operation cost ΔP in the objective function ope and loss cost ΔP loss As shown below, Where: P p , P q They are the control cost of active and reactive power by adjusting the intelligent soft switch and the loss cost of line loss, T represents the matrix conversion, and f v is the voltage unbalance objective function, is the value at node i Phase voltage value, A, B, C represent different phases, and N is the number of distribution network nodes.
8. A method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 6, characterized in that: Since the efficiency of intelligent soft switch is high, assuming P p =P q relationship.
9. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1, characterized in that: In step 2, the constraints that the node voltage meets the operation requirements are as follows: Voltage upper and lower limit constraints: V min With V max are the lower and upper limits of the node voltage at both ends of the intelligent soft switch, as shown in formula (13), V min (k+i)≤V a,b,c (k+i|k)≤V max (k+i) (13) Considering the operating state limitation and maximum capacity S of the intelligent soft switch max And the upper limit of active and reactive output P max , Q max , the transmission power constraint of the intelligent soft switch is as follows, 10. The method for predicting and optimizing line loss in three-phase unbalanced low-voltage areas according to claim 1, characterized in that: Step 4: Input the correction model and conduct a new round of optimization. When a new round of predictive control is carried out, it is necessary to update the injected power items of the nodes at both ends of the smart soft switch in the sensitivity matrix according to the change of the working state of the smart soft switch to obtain a new loss sensitivity matrix. If the distribution network is in a stable operating state, the deviation caused by the change of the working state of other nodes in the distribution network is ignored, and the line loss sensitivity is corrected according to the change of the working state of the smart soft switch, as shown below. R is the line resistance where the node is located, and U is the node voltage.