An active power distribution network intelligent transfer source and load balancing analysis method and system

Abstract

The application discloses an active power distribution network intelligent transfer source and load balance analysis method and system, relates to the technical field of intelligent transfer, and specifically comprises the following steps: step one, monitoring the power load and distributed photovoltaic access in the active power distribution network; step two, classifying and sorting the real-time operation basic data collected through monitoring, and performing prediction analysis on the operation basic data at the next time point; step three, setting a target function and a safety constraint condition based on the intelligent transfer target of the active power distribution network; step four, performing simulation experiments on the transfer path of the active power distribution network based on power CPS joint simulation, obtaining a generated transfer method; and step five, analyzing the source and load balance level of the active power distribution network under the generated transfer method, and constructing a visual intelligent transfer diagram. In the case of ensuring the source and load balance, the optimal transfer method of maximizing the transferred load, minimizing the switch operation times and shortening the transfer time is realized.

Description

Technical Field

[0001] This invention belongs to the field of intelligent power transfer technology, specifically relating to an intelligent power transfer source-load balance analysis method and system for active power distribution networks. Background Technology

[0002] In the process of building a new power system with new energy sources as the mainstay, the power grid structure and operating characteristics will undergo significant changes, with the most prominent changes occurring in the distribution network. New energy sources, especially distributed photovoltaic (PV), are experiencing explosive growth, and the distribution network will become a crucial channel for PV grid connection. In areas with high PV penetration, the power generation capacity fluctuates intermittently and significantly due to weather conditions, posing a considerable challenge to distribution network transfer methods. If the impact of distributed PV output is not fully considered, problems such as heavy overload and voltage exceeding limits in transfer lines can easily occur. Currently, there are no effective technical means to analyze and evaluate the source-load balance level of transfer methods involving distributed PV lines in active distribution networks, resulting in difficulties in the integration and absorption of distributed PV and making it difficult to obtain a transfer method that achieves the transfer objective while ensuring source-load balance. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent source-load balance analysis method and system for active power distribution networks, which solves the technical problem of difficulty in ensuring source-load balance while achieving the purpose of power transfer in the prior art.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for intelligent power transfer and load balance analysis in active distribution networks includes the following steps:

[0006] Step 1: Monitor the power load and distributed photovoltaic access in the active power distribution network;

[0007] The active power distribution network is monitored in real time using data acquisition devices such as smart meters and photovoltaic inverters. A time point is set every *a* unit of time, and basic operational data of the active power distribution network is collected at each time point. The types of operational data include: the power load of each branch node of the active power distribution network, the power generation of distributed photovoltaic systems, and the operating status of each branch of the active power distribution network at different times. To prevent the collected data from containing noise and missing values, relevant processing is performed, including: using filters to remove noise and using interpolation to fill in missing values.

[0008] Step 2: Classify and sort the collected real-time operational data, and perform predictive analysis on the operational data for the next time point;

[0009] S201. Classify and sort the basic operational data according to its type and time sequence; the resulting basic operational data set is represented as follows:

[0010]

[0011] Where X j Y represents the load set of the j-th branch, where j represents the j-th branch; k n represents the power generation of distributed photovoltaic systems. j This indicates that the active power distribution network includes a total of n j There are n branches, where k represents the k-th distributed photovoltaic system, and n k This indicates that the active power distribution network includes a total of n k One distributed photovoltaic; Z j This represents the set of operating states of each branch in an active distribution network.

[0012] Furthermore, X j ={X j (1), X j (2), ..., X j (t)}, where X j (t) represents the load of branch j at time t;

[0013] Furthermore, Y k ={Y k (1), Y k (2), ..., Y k (t)}, where Y k (t) represents the power generation of distributed photovoltaic k at time t;

[0014] Furthermore, Z j ={Z j (1), Z j (2), ..., Z j (t)}, where Z j (t) represents the working state of branch j at time t, where the working state is divided into normal working state and stopped working state. The Ri value of normal working state is 1, and the Ri value of stopped working state is 0.

[0015] S202. Trend extrapolation is used to predict and analyze the basic data. For the set {a(1), a(2), ..., a(t)}, a polynomial model is used for fitting and prediction, where the nth-order polynomial model is shown below:

[0016]

[0017] Where k1, ..., k n Here, represents the fitting coefficients of the polynomial model, and b is a constant. Let represent the fitted predicted value of the model at time t, and n represent the order of the polynomial model. Machine learning algorithms are used to simulate polynomial models of different orders, and the prediction objective function of the polynomial model is set. The fitting coefficients that satisfy the prediction objective function at different orders are determined by machine learning algorithms. The specific prediction objective function is shown below:

[0018]

[0019] Where N represents the number of time points, and a(t) represents the true value at time t. Based on this objective function, the fitting coefficients that minimize the polynomial fitting error are obtained, thus yielding polynomial models of different orders.

[0020] S203. Fitting analysis is performed between high-order and low-order polynomial models. Polynomial models of different orders have different advantages. Low-order polynomial models have smaller errors when the elements of the set change steadily, but larger errors when the elements change rapidly. High-order polynomial models have smaller errors when the elements change rapidly, but larger errors when the elements change steadily. By fitting low-order and high-order polynomials using machine learning, overfitting caused by excessively high orders can be avoided, and the advantages and disadvantages of both low-order and high-order polynomial models can be neutralized to obtain the final prediction model. The specific method is as follows:

[0021] In polynomial models of orders 1 to n, those with orders between (n1, n2) are classified as low-order polynomials, and those with orders between (n3, n4) are classified as high-order polynomials, where n3 > n2. A low-order polynomial model and a high-order polynomial model are randomly selected and denoted as follows: and Machine learning is used to fit and predict the two polynomial models, and the weights of the predicted fitting results for the two polynomial models are as follows: and The error is reduced by optimizing the function, which is shown below:

[0022]

[0023] The final polynomial fitting result is obtained by optimizing the function, using weight proportions. and The obtained fitting model is By comparing the prediction errors of all fitted models, the fitted model with the smallest prediction error is selected as the final prediction model.

[0024] Step 3: Based on the intelligent power transfer target of the active distribution network, set the objective function and security constraints; considering topological constraints, node voltage constraints, distributed photovoltaic power output constraints and branch capacity constraints, to achieve the power transfer target of maximizing the transferred load, minimizing the number of switching operations and minimizing the power transfer time.

[0025] Furthermore, topological constraints:

[0026] g∈T o ;

[0027] Where g represents the topology of the active distribution network after intelligent power transfer, and T o It is a collection of active power distribution network topologies.

[0028] Node voltage constraints:

[0029]

[0030] in This represents the actual voltage at the i-th node of the j-th branch. and Let represent the lower limit and upper limit of the voltage at the i-th node of the j-th branch, respectively.

[0031] Distributed photovoltaic power output constraints:

[0032]

[0033] Where p kmin With p kmax These represent the lower and upper limits of the active power of distributed photovoltaic (PV) system k, respectively. This represents the working state coefficient of distributed photovoltaic (PV) module k. The coefficient is 1 when k is working and 0 when k stops working.

[0034] Branch capacity constraints:

[0035] S j ≤S jmax ;

[0036] Where S j S represents the actual power flowing through the j-th branch. jmax This represents the maximum power allowed to flow through the j-th branch;

[0037] The target function for the transfer is determined as follows:

[0038]

[0039] Where x represents a time period, which is divided based on the transfer method. Each time period completes a portion of the transfer path, t x c represents the duration of the x-th time interval. xu represents the number of times the switch is turned on and off in the x-th time period. x Let t represent the load transferred during the x-th time period, and t′ represent the network delay, which is the information transmission delay for an active distribution network to perform a switching operation.

[0040] Step 4: Conduct simulation experiments on the transfer path of the active distribution network based on the co-simulation of the cyber-physical system (hereinafter referred to as the power CPS system) to obtain the proposed transfer method;

[0041] A power system simulation model of an active distribution network is established to simulate its normal physical operation, such as distributed photovoltaic (PV) power generation and distribution. A communication system simulation model of the active distribution network is also established to simulate its normal information communication operation, such as data transmission. Furthermore, based on the structure of the active distribution network, the embedded components connecting the information and physical sides are simulated. Simulation parameters are set based on the simulation model, including the network topology, distributed PV type, and load node location. Simulation experiments are conducted using a power CPS system, combined with simulation models of the power system, communication system, and their interface embedded components, based on security constraints, the power transfer objective function, and predicted operating data.

[0042] By analyzing the power generation of distributed photovoltaic (PV) systems in the active power distribution network and the load demand of each node, transfer nodes are selected. Different transfer nodes correspond to different transfer paths. A transfer method includes multiple transfer paths. Based on the transfer objective function, the target offset value of the transfer path is set, as shown below:

[0043]

[0044] Where G(h,v) represents the target offset value of the transfer path from the starting node h to the ending node v, z h With z v This indicates the working status of nodes h and v. The path is evaluated by the target offset value of the transfer path. The smaller the target offset value, the more the transfer path meets the transfer target. Transfer methods with a target offset value of 0 are removed. The target offset values ​​of the transfer paths contained in the remaining transfer methods are accumulated and denoted as the total target offset value Gi(s), where s represents the s-th transfer method. Transfer methods with the highest total target offset value are selected as the proposed transfer methods.

[0045] Step 5: Analyze the source-load balance level of the active distribution network under the proposed power transfer method, and construct a visualized intelligent power transfer map;

[0046] Multiple simulation experiments were conducted at different time periods, and data from each experiment were collected to calculate the total power generation of distributed photovoltaic (PV) systems and the load demand of each node in the active power distribution network. Based on the simulation data, the output of distributed PV systems and the load of the active power distribution network system were obtained. The source-load balance level of the active power distribution network was evaluated at different time periods. Based on the actual load of each node at different time periods and the maximum load that each node can withstand, potential unbalanced nodes and branch areas with overload were identified. Based on the proposed power transfer method, the load transfer situation under the simulation experiment was observed. Furthermore, this study analyzes the source-load balance level of the active distribution network after power transfer. For cases of source-load imbalance in the active distribution network branches after power transfer, methods for adjusting the output characteristics of distributed photovoltaic power and the topology of the active distribution network are added to the power transfer method to reduce the occurrence of source-load imbalance in the active distribution network branches and improve the feasibility of the power transfer method. By collecting and analyzing conventional parameter data such as distributed photovoltaic power generation, load of each branch of the active power grid, power structure, and load characteristics, the source-load balance level of the power transfer method is evaluated. Based on the dispatching capacity of the active distribution network, a source-load balance rate is set, as shown in the following formula:

[0047]

[0048] Where N represents the existence of N time points, k and j represent the k-th distributed photovoltaic and the j-th branch corresponding to that distributed photovoltaic, y k x represents the power type coefficient of distributed photovoltaic (PV) k. j X represents the load characteristic coefficient of the j-th branch. These coefficients can be determined based on specific circumstances and requirements, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. j (t) represents the load on branch j at time t, Y k (t) represents the power generation of distributed photovoltaic k at time t, Br(k) represents the failure rate of distributed photovoltaic k under the evaluated transfer method, and Ch(j) represents the load mutation rate of branch j under the evaluated transfer method. The source-load balance rate of all branches under the transfer method is calculated and then summed to obtain the total source-load balance rate Qi(s) of the transfer method.

[0049] Set a total source load balance threshold Kj, and eliminate power transfer methods with a total source load balance rate greater than or equal to this threshold. For power transfer methods with a total source load balance rate less than Kj, set a source load safety value YH(s), as shown in the following formula:

[0050] YH(s)=[W Qi Qi(s) + W Gi Gi(s)]*Qi(s)*Gi(s);

[0051] Among them W Qi With W GiThese represent the weighting coefficients for the total source-load balance rate and the total target offset, respectively. In this embodiment, W is set to... Qi With W Gi The values ​​are 0.6 and 0.4.

[0052] By comparing the magnitudes of YH(s), the transfer method with the smallest YH(s) value is selected as the optimal transfer method.

[0053] Based on a visualization system, GIS technology is used to create intelligent power transfer maps, adding map structure units that include basic geographic information and power distribution equipment structure information of the active power distribution network. A spatial model of the active power distribution network is constructed, and each path of the proposed power transfer method is identified. A view transformation model is added, providing functions such as view rotation, panoramic zoom-in, and zoom-out to facilitate browsing of details in the intelligent power transfer map. The spatial model of the active power distribution network constructed using GIS technology displays the load distribution and switch status information of each power transfer path. The visualization system can display node voltage, branch current, and power flow monitoring information in real time, and intuitively present the source-load balance level in the form of charts and heat maps. It can also identify areas or equipment with potential source-load imbalance and set prominent early warning signs for these areas or equipment in the intelligent power transfer map, providing early warnings for possible equipment overload and voltage exceeding limits, thereby improving the power transfer safety of the active power distribution network.

[0054] This invention also provides an intelligent power transfer and load balance analysis system for active power distribution networks, comprising:

[0055] Active power distribution network data monitoring module: monitors the basic operational data of the active power distribution network;

[0056] Run the basic data prediction module: classify and sort the basic data, and perform predictive analysis on it;

[0057] Safety constraint and objective function establishment module: sets safety constraints and transfer objective functions;

[0058] Transfer method generation module: Set the target offset value of the transfer path to obtain the proposed transfer method;

[0059] The optimal transfer method visualization module: obtains the optimal transfer method and displays it visually.

[0060] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0061] 1. This invention fits a high-order polynomial model to a low-order polynomial model. The resulting fitted model includes both trend extraction of the data and fine fitting of local changes in the data. This method improves the accuracy of prediction while maintaining the stability of the prediction.

[0062] 2. This invention sets the target offset value of the transfer path based on power CPS co-simulation, selects the transfer method that meets the transfer objective function, and eliminates the influence of abnormal normal branches on the transfer path by defining the working state of active distribution network branches.

[0063] 3. Based on conventional parameter data such as distributed photovoltaic power generation, load of each branch of the active power grid, power supply structure, and load characteristics, this invention sets a source-load balance rate, analyzes the source-load balance level of the active distribution network under the proposed transfer method through the source-load balance rate analysis, and obtains the optimal transfer method by comprehensively considering the source-load balance level of the active distribution network and the transfer objective function. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 A flowchart of an intelligent power transfer source-load balance analysis method for active power distribution networks is shown.

[0066] Figure 2 A flowchart illustrating the method for predictive analysis of operational baseline data is shown.

[0067] Figure 3 A structural diagram of an intelligent power supply and load balance analysis system for active power distribution networks is shown. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Example 1

[0070] like Figure 1 , Figure 2 As shown in the figure, the intelligent power supply and load balance analysis method for active distribution networks provided in this embodiment specifically includes the following steps:

[0071] Step 1: Monitor the power load and distributed photovoltaic access in the active power distribution network;

[0072] The active power distribution network is monitored in real time using data acquisition devices such as smart meters and photovoltaic inverters. A time point is set every *a* unit of time, and basic operational data of the active power distribution network is collected at each time point. The types of operational data include: the power load of each branch node of the active power distribution network, the power generation of distributed photovoltaic systems, and the operating status of each branch of the active power distribution network at different times. To prevent the collected data from containing noise and missing values, relevant processing is performed, including: using filters to remove noise and using interpolation to fill in missing values.

[0073] Step 2: Classify and sort the collected real-time operational data, and perform predictive analysis on the operational data for the next time point;

[0074] S201 classifies and sorts the basic operational data according to its type and time order, resulting in a set of basic operational data represented as follows:

[0075]

[0076] Where X j Y represents the load set of the j-th branch, where j represents the j-th branch; k n represents the power generation of distributed photovoltaic systems. j This indicates that the active power distribution network includes a total of n j There are n branches, where k represents the k-th distributed photovoltaic system, and n k This indicates that the active power distribution network includes a total of n k One distributed photovoltaic; Z j This represents the set of operating states of each branch in an active distribution network.

[0077] Furthermore, X j ={X j (1), X j (2), ..., X j (t)}, where X j (t) represents the load of branch j at time t;

[0078] Furthermore, Y k ={Y k (1), Y k (2), ..., Y k (t)}, where Y k (t) represents the power generation of distributed photovoltaic k at time t;

[0079] Furthermore, Z j ={Z j (1), Z j (2), ..., Z j (t)}, where Z j(t) represents the working state of branch j at time t, where the working state is divided into normal working state and stopped working state. The Ri value of normal working state is 1, and the Ri value of stopped working state is 0.

[0080] S202 uses trend extrapolation to predict and analyze the basic operational data. For the set {a(1), a(2), ..., a(t)}, a multinomial model is used for fitting and prediction. The nth-order multinomial model is shown below:

[0081]

[0082] Where k1, ..., k n Here, represents the fitting coefficients of the polynomial model, and b is a constant. Let represent the fitted predicted value of the model at time t, and n represent the order of the polynomial model. Machine learning algorithms are used to simulate polynomial models of different orders, and the prediction objective function of the polynomial model is set. The fitting coefficients that satisfy the prediction objective function at different orders are determined by machine learning algorithms. The specific prediction objective function is shown below:

[0083]

[0084] Where N represents the number of time points, and a(t) represents the true value at time t. Based on this objective function, the fitting coefficients that minimize the polynomial fitting error are obtained, thus yielding polynomial models of different orders.

[0085] S203 performs fitting analysis on high-order and low-order polynomial models. Polynomial models of different orders have different advantages. Low-order polynomial models have smaller errors when the elements of the set change steadily, but larger errors when the elements change rapidly. High-order polynomial models have smaller errors when the elements change rapidly, but larger errors when the elements change steadily. By fitting low-order and high-order polynomials using machine learning, overfitting caused by excessively high orders can be avoided, and the advantages and disadvantages of both low-order and high-order polynomial models can be neutralized to obtain the final prediction model. The specific method is as follows:

[0086] In polynomial models of orders 1 to n, those with orders between (n1, n2) are classified as low-order polynomials, and those with orders between (n3, n4) are classified as high-order polynomials, where n3 > n2. A low-order polynomial model and a high-order polynomial model are randomly selected and denoted as follows: and Machine learning is used to fit and predict the two polynomial models, and the weights of the predicted fitting results for the two polynomial models are as follows: and The error is reduced by optimizing the function, which is shown below:

[0087]

[0088] The final polynomial fitting result is obtained by optimizing the function, using weight proportions. and The obtained fitting model is By comparing the prediction errors of all fitted models, the fitted model with the smallest prediction error is selected as the final prediction model.

[0089] Step 3: Based on the intelligent power transfer target of the active distribution network, set the objective function and security constraints; considering topological constraints, node voltage constraints, distributed photovoltaic power output constraints and branch capacity constraints, to achieve the power transfer target of maximizing the transferred load, minimizing the number of switching operations and minimizing the power transfer time.

[0090] Furthermore, topological constraints:

[0091] g∈T o ;

[0092] Where g represents the topology of the active distribution network after intelligent power transfer, and T o It is a collection of active power distribution network topologies.

[0093] Node voltage constraints:

[0094]

[0095] in This represents the actual voltage at the i-th node of the j-th branch. and Let represent the lower limit and upper limit of the voltage at the i-th node of the j-th branch, respectively.

[0096] Distributed photovoltaic power output constraints:

[0097]

[0098] Where p kmin With p kmax These represent the lower and upper limits of the active power of distributed photovoltaic (PV) system k, respectively. This represents the working state coefficient of distributed photovoltaic (PV) module k. The coefficient is 1 when k is working and 0 when k stops working.

[0099] Branch capacity constraints:

[0100] S j ≤S jmax ;

[0101] Where S j S represents the actual power flowing through the j-th branch. jmaxThis represents the maximum power allowed to flow through the j-th branch;

[0102] The target function for the transfer is determined as follows:

[0103]

[0104] Where x represents a time period, which is divided based on the transfer method. Each time period completes a portion of the transfer path, t x c represents the duration of the x-th time interval. x u represents the number of times the switch is turned on and off in the x-th time period. x Let t represent the load transferred during the x-th time period, and t′ represent the network delay, which is the information transmission delay for an active distribution network to perform a switching operation.

[0105] Step 4: Conduct simulation experiments on the transfer path of the active distribution network based on the power CPS co-simulation to obtain the proposed transfer method;

[0106] A power system simulation model of an active distribution network is established to simulate its normal physical operation, such as distributed photovoltaic (PV) power generation and distribution. A communication system simulation model of the active distribution network is also established to simulate its normal information communication operation, such as data transmission. Furthermore, based on the structure of the active distribution network, the embedded components connecting the information and physical sides are simulated. Simulation parameters are set based on the simulation model, including the network topology, distributed PV type, and load node location. Simulation experiments are conducted using a power CPS system, combined with simulation models of the power system, communication system, and their interface embedded components, based on security constraints, the power transfer objective function, and predicted operating data.

[0107] By analyzing the power generation of distributed photovoltaic (PV) systems in the active power distribution network and the load demand of each node, transfer nodes are selected. Different transfer nodes correspond to different transfer paths. A transfer method includes multiple transfer paths. Based on the transfer objective function, the target offset value of the transfer path is set, as shown below:

[0108]

[0109] Where G(h,v) represents the target offset value of the transfer path from the starting node h to the ending node v, z h With z vThis indicates the working status of nodes h and v. The path is evaluated by the target offset value of the transfer path. The smaller the target offset value, the more the transfer path meets the transfer target. Transfer methods with a target offset value of 0 are removed. The target offset values ​​of the transfer paths contained in the remaining transfer methods are accumulated and denoted as the total target offset value Gi(s), where s represents the s-th transfer method. Transfer methods with the highest total target offset value are selected as the proposed transfer methods.

[0110] Step 5: Analyze the source-load balance level of the active distribution network under the proposed power transfer method, and construct a visualized intelligent power transfer map;

[0111] Multiple simulation experiments were conducted at different time periods, and data from each experiment were collected to calculate the total power generation of distributed photovoltaic (PV) systems and the load demand of each node in the active power distribution network. Based on the simulation data, the output of distributed PV systems and the load of the active power distribution network system were obtained. The source-load balance level of the active power distribution network was evaluated at different time periods. Based on the actual load of each node at different time periods and the maximum load that each node can withstand, potential unbalanced nodes and branch areas with overload were identified. Based on the proposed power transfer method, the load transfer situation under the simulation experiment was observed. Furthermore, this study analyzes the source-load balance level of the active distribution network after power transfer. For cases of source-load imbalance in the active distribution network branches after power transfer, methods for adjusting the output characteristics of distributed photovoltaic power and the topology of the active distribution network are added to the power transfer method to reduce the occurrence of source-load imbalance in the active distribution network branches and improve the feasibility of the power transfer method. By collecting and analyzing conventional parameter data such as distributed photovoltaic power generation, load of each branch of the active power grid, power structure, and load characteristics, the source-load balance level of the power transfer method is evaluated. Based on the dispatching capacity of the active distribution network, a source-load balance rate is set, as shown in the following formula:

[0112]

[0113] Where N represents the existence of N time points, k and j represent the k-th distributed photovoltaic and the j-th branch corresponding to that distributed photovoltaic, y k x represents the power type coefficient of distributed photovoltaic (PV) k. j X represents the load characteristic coefficient of the j-th branch. These coefficients can be determined based on specific circumstances and requirements, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. j (t) represents the load on branch j at time t, Y k(t) represents the power generation of distributed photovoltaic k at time t, Br(k) represents the failure rate of distributed photovoltaic k under the evaluated transfer method, and Ch(j) represents the load mutation rate of branch j under the evaluated transfer method. The source-load balance rate of all branches under the transfer method is calculated and then summed to obtain the total source-load balance rate Qi(s) of the transfer method.

[0114] Set a total source load balance threshold Kj, and eliminate power transfer methods with a total source load balance rate greater than or equal to this threshold. For power transfer methods with a total source load balance rate less than Kj, set a source load safety value YH(s), as shown in the following formula:

[0115] YH(s)=[W Qi Qi(s) + W Gi Gi(s)]*Qi(s)*Gi(s);

[0116] Among them W Qi With W Gi These represent the weighting coefficients for the total source-load balance rate and the total target offset, respectively. In this embodiment, W is set to... Qi With W Gi The values ​​are 0.6 and 0.4.

[0117] By comparing the magnitudes of YH(s), the transfer method with the smallest YH(s) value is selected as the optimal transfer method.

[0118] Based on a visualization system, GIS technology is used to create intelligent power transfer maps, adding map structure units that include basic geographic information and power distribution equipment structure information of the active power distribution network. A spatial model of the active power distribution network is constructed, and each path of the proposed power transfer method is identified. A view transformation model is added, providing functions such as view rotation, panoramic zoom-in, and zoom-out to facilitate browsing of details in the intelligent power transfer map. The spatial model of the active power distribution network constructed using GIS technology displays the load distribution and switch status information of each power transfer path. The visualization system can display node voltage, branch current, and power flow monitoring information in real time, and intuitively present the source-load balance level in the form of charts and heat maps. It can also identify areas or equipment with potential source-load imbalance and set prominent early warning signs for these areas or equipment in the intelligent power transfer map, providing early warnings for possible equipment overload and voltage exceeding limits, thereby improving the power transfer safety of the active power distribution network.

[0119] Example 2

[0120] like Figure 3 As shown in the figure, the intelligent power supply and load balance analysis system for active power distribution networks provided in this embodiment specifically includes the following:

[0121] Active power distribution network data monitoring module: Real-time monitoring of the active power distribution network. A time point is set every 'a' unit of time. At each time point, the basic operating data of the active power distribution network is collected. The types of basic operating data include: the power load of each branch node of the active power distribution network, the power generation of distributed photovoltaic, and the working status of each branch of the active power distribution network in each time period. The collected data is processed accordingly.

[0122] The basic data prediction module for running operations: The basic data is categorized and sorted according to its type and time sequence, resulting in a set of basic data represented as follows:

[0123]

[0124] Where X j Y represents the load set of the j-th branch, where j represents the j-th branch; k n represents the power generation of distributed photovoltaic systems. j This indicates that the active power distribution network includes a total of n j There are n branches, where k represents the k-th distributed photovoltaic system, and n k This indicates that the active power distribution network includes a total of n k One distributed photovoltaic; Z j This represents the set of operating states of each branch in an active power distribution network.

[0125] Furthermore, X j ={X j (1), X j (2), ..., X j (t)}, where X j (t) represents the load of branch j at time t;

[0126] Furthermore, Y k ={Y k (1), Y k (2), ..., Y k (t)}, where Y k (t) represents the power generation of distributed photovoltaic k at time t;

[0127] Furthermore, Z j ={Z j (1), Z j (2), ..., Z j (t)}, where Z j (t) represents the working state of branch j at time t, where the working state is divided into normal working state and stopped working state. The Ri value of normal working state is 1, and the Ri value of stopped working state is 0.

[0128] For the set {a(1), a(2), ..., a(t)}, a multinomial model is used for fitting and prediction, where the nth-order multinomial model is shown below:

[0129]

[0130] Where k1, ..., k n Here, represents the fitting coefficients of the polynomial model, and b is a constant. Let represent the fitted predicted value of the model at time t, and n represent the order of the polynomial model. Machine learning algorithms are used to simulate polynomial models of different orders, and the prediction objective function of the polynomial model is set. The fitting coefficients that satisfy the prediction objective function at different orders are determined by machine learning algorithms. The specific prediction objective function is shown below:

[0131]

[0132] Where N represents the number of time points, and a(t) represents the true value at time t. Based on this objective function, the fitting coefficients that minimize the polynomial fitting error are obtained, thus yielding polynomial models of different orders.

[0133] Fitting analysis is performed between higher-order and lower-order polynomial models. Among polynomial models of orders 1 to n, those with orders between (n1, n2) are classified as lower-order polynomial models, and those with orders between (n3, n4) are classified as higher-order polynomial models, where n3 > n2. A lower-order polynomial model and a higher-order polynomial model are randomly selected and denoted as follows: and Machine learning is used to fit and predict the two polynomial models, and the weights of the predicted fitting results for the two polynomial models are as follows: and The error is reduced by optimizing the function, which is shown below:

[0134]

[0135] The final polynomial fitting result is obtained by optimizing the function, using weight proportions. and The obtained fitting model is By comparing the prediction errors of all fitted models, the fitted model with the smallest prediction error is selected as the final prediction model.

[0136] Security constraints and objective function establishment module: Security constraints include: topology constraints, node voltage constraints, distributed photovoltaic power output constraints, and branch capacity constraints;

[0137] Furthermore, topological constraints:

[0138] g∈T o ;

[0139] Where g represents the topology of the active distribution network after intelligent power transfer, and T o It is a collection of active power distribution network topologies.

[0140] Furthermore, node voltage constraints:

[0141]

[0142] in This represents the actual voltage at the i-th node of the j-th branch. and Let represent the lower limit and upper limit of the voltage at the i-th node of the j-th branch, respectively.

[0143] Furthermore, the output constraints of distributed photovoltaic power:

[0144]

[0145] Where p kmin With p kmax These represent the lower and upper limits of the active power of distributed photovoltaic (PV) system k, respectively. This represents the working state coefficient of distributed photovoltaic (PV) module k. The coefficient is 1 when k is working and 0 when k stops working.

[0146] Furthermore, branch capacity constraints:

[0147] S j ≤S jmax ;

[0148] Where S j S represents the actual power flowing through the j-th branch. jmax This represents the maximum power allowed to flow through the j-th branch;

[0149] The target function for the transfer is determined as follows:

[0150]

[0151] Where x represents a time period, which is divided based on the transfer method. Each time period completes a portion of the transfer path, t x c represents the duration of the x-th time interval. x u represents the number of times the switch is turned on and off in the x-th time period. x Let t represent the load transferred during the x-th time period, and t′ represent the network delay, which is the information transmission delay for an active distribution network to perform a switching operation.

[0152] The proposed module for the transfer method includes: establishing a power system simulation model of the active distribution network to simulate its normal physical operation; establishing a communication system simulation model of the active distribution network to simulate its normal information communication operation; and simulating the embedded components used to connect the information and physical sides of the active distribution network based on its own structure. Simulation experiments are then conducted using a power CPS system, combining the simulation models of the power system, communication system, and their interactive interfaces with the embedded components.

[0153] By analyzing the power generation of distributed photovoltaic (PV) systems in the active power distribution network and the load demand of each node, transfer nodes are selected. Different transfer nodes correspond to different transfer paths. A transfer method includes multiple transfer paths. Based on the transfer objective function, the target offset value of the transfer path is set, as shown below:

[0154]

[0155] Where G(h,v) represents the target offset value of the transfer path from the starting node h to the ending node v, z h With z v This indicates the working status of nodes h and v. The path is evaluated by the target offset value of the transfer path. The smaller the target offset value, the more the transfer path meets the transfer target. Transfer methods with a target offset value of 0 are removed. The target offset values ​​of the transfer paths contained in the remaining transfer methods are accumulated and denoted as the total target offset value Gi(s), where s represents the s-th transfer method. Transfer methods with the highest total target offset value are selected as the proposed transfer methods.

[0156] The optimal power transfer method visualization module collects and analyzes conventional parameter data such as distributed photovoltaic power generation, load of each branch of the active power grid, power structure, and load characteristics. It evaluates the source-load balance level of the power transfer method and sets the source-load balance rate based on the dispatching capability of the active distribution network. The specific formula is shown below:

[0157]

[0158] Where N represents the existence of N time points, k and j represent the k-th distributed photovoltaic and the j-th branch corresponding to that distributed photovoltaic, y k x represents the power type coefficient of distributed photovoltaic (PV) k. j X represents the load characteristic coefficient of the j-th branch. These coefficients can be determined based on specific circumstances and requirements, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. j (t) represents the load on branch j at time t, Y k(t) represents the power generation of distributed photovoltaic k at time t, Br(k) represents the failure rate of distributed photovoltaic k under the evaluated transfer method, and Ch(j) represents the load mutation rate of branch j under the evaluated transfer method. The source-load balance rate of all branches under the transfer method is calculated and then summed to obtain the total source-load balance rate Qi(s) of the transfer method.

[0159] Set a total source load balance threshold Kj, and eliminate power transfer methods with a total source load balance rate greater than or equal to this threshold. For power transfer methods with a total source load balance rate less than Kj, set a source load safety value YH(s), as shown in the following formula:

[0160] YH(s)=[W Qi Qi(s) + W Gi Gi(s)]*Qi(s)*Gi(s);

[0161] Among them W Qi With W Gi These represent the weighting coefficients for the total source-load balance rate and the total target offset, respectively. In this embodiment, W is set to... Qi With W Gi The values ​​are 0.6 and 0.4.

[0162] By comparing the magnitudes of YH(s), the transfer method with the smallest YH(s) value is selected as the optimal transfer method.

[0163] Based on a visualization system, GIS technology is used to create intelligent power transfer maps, adding map structure units that include basic geographic information and power distribution equipment structure information of the active power distribution network. A spatial model of the active power distribution network is constructed, and each path of the proposed power transfer method is identified. A view transformation model is added, providing functions such as view rotation, panoramic zoom-in, and zoom-out to facilitate browsing of details in the intelligent power transfer map. The spatial model of the active power distribution network constructed using GIS technology displays the load distribution and switch status information of each power transfer path. The visualization system can display node voltage, branch current, and power flow monitoring information in real time, and intuitively present the source-load balance level in the form of charts and heat maps. It can also identify areas or equipment with potential source-load imbalance and set prominent early warning signs for these areas or equipment in the intelligent power transfer map, providing early warnings for possible equipment overload and voltage exceeding limits, thereby improving the power transfer safety of the active power distribution network.

[0164] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0165] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for intelligent power supply and load balancing analysis in an active distribution network, characterized in that, include: Step 1: Monitor the power load and distributed photovoltaic access in the active power distribution network; Real-time monitoring of basic operational data of the active power distribution network is achieved through data acquisition equipment. Step 2: Classify and sort the collected real-time operational data, and perform predictive analysis on the operational data for the next time point; The trend extrapolation method is adopted, and prediction analysis is performed by fitting the high-order polynomial model and the low-order polynomial model of the trend extrapolation method. The trend extrapolation method is used to predict and analyze the basic operational data. The specific method is as follows: For sets Using the formula A multinomial model is used for fitting and prediction, where, ,..., Here, represents the fitting coefficients of the polynomial model, and b is a constant. Let represent the fitted prediction value of the model at time t, and n represent the order of the polynomial model, using the formula... Set the objective function for prediction, where N represents the number of time points. The true value at time t is represented by the fitting coefficients that satisfy the prediction objective function at different orders are determined by machine learning algorithms, thereby obtaining polynomial models of different orders. Step 3: Based on the intelligent power transfer target of the active distribution network, set the objective function and security constraints; considering topology constraints, node voltage constraints, distributed photovoltaic power output constraints and branch capacity constraints, the power transfer target is to maximize the power transfer load, minimize the number of switching operations and minimize the power transfer time. Step 4: Conduct simulation experiments on the transfer path of the active distribution network based on the joint simulation of the power CPS system to obtain the proposed transfer method; through simulation experiments, set the target offset value of the transfer path to obtain the proposed transfer method. Step 5: Analyze the source-load balance level of the active distribution network under the proposed transfer method, and construct a visualized intelligent transfer map; by collecting and analyzing relevant data, obtain the total source-load balance rate of the implemented transfer method, and then combine it with the total target offset value to obtain the optimal transfer method.

2. The intelligent power supply and load balance analysis method for active distribution networks according to claim 1, characterized in that, The data is categorized and sorted according to its type and time sequence. The specific method is as follows: use This represents the underlying data set for operation, where Let j represent the load set of the j-th branch, where j represents the j-th branch; This represents the set of power generation from distributed photovoltaic (PV) system k, which includes a total of [number of units]. Branch road and One distributed photovoltaic; This represents the working state of the j-th branch, where t represents the time t, where the load of each branch consists of the loads of all nodes in that branch, and the power generation at each time is a set. The working state of branch j at each moment is determined by... This indicates that the working status is divided into normal working status and stopped working status. The normal working status... A value of 1 indicates the device is in a stopped working state. The value is 0.

3. The intelligent power supply and load balance analysis method for active distribution networks according to claim 1, characterized in that, The fitting analysis is performed between the higher-order polynomial model and the lower-order polynomial model. The specific method is as follows: In polynomial models of orders 1 to n, the order is set to be in the interval [missing information]. The ones between are of lower order, within the interval The ones between belong to the higher order. Choose a low-order polynomial model and a high-order polynomial model, denoted as [model name missing], ... and Using the formula Will and Perform fitting, where and Representing the model respectively and The fitting weight ratio, Using the formula To obtain the optimal fitting weight ratio and Based on the fitting weight ratio, the model is fitted, and the prediction error of all fitted models is compared. The fitted model with the smallest prediction error is selected as the final prediction model.

4. The intelligent power supply and load balance analysis method for active distribution networks according to claim 1, characterized in that, The specific method for setting security constraints and the transfer objective function is as follows: pass This represents a topology constraint, where g is the topology of the active distribution network after intelligent power transfer is implemented. For a set of active distribution network topologies, using inequalities Represents node voltage constraints, where This represents the actual voltage at the i-th node of the j-th branch. and Let the lower and upper limits of the voltage at the i-th node of the j-th branch be represented, respectively. Using the inequality... This represents the output constraint of distributed photovoltaic power, where and These represent the lower and upper limits of the active power of distributed photovoltaic (PV) system k, respectively. This represents the working state coefficient of k; the coefficient is 1 when k is working and 0 when k stops working; using the inequality... This represents the branch capacity constraint, where This represents the actual power flowing through the j-th branch. This represents the maximum power allowed to flow through the j-th branch; Using formula This represents the objective function for the transfer of resources, where x represents the x-th time period. This represents the duration of the x-th time interval. This represents the number of times the device is switched on and off during the x-th time period. This represents the load transferred during the x-th time period. This refers to network latency, specifically the information transmission delay during a single switching operation in an active power distribution network.

5. The intelligent power supply and load balance analysis method for active distribution networks according to claim 1, characterized in that, Set the target offset value for the transfer path to obtain the proposed transfer method. The specific method is as follows: By constructing simulation models of the power system, communication system, and embedded components of an active distribution network, and conducting simulation experiments through a power CPS system, real-time data on the power generation of distributed photovoltaics and the load demand of each node in the active distribution network are obtained. Transfer nodes are then identified, and formulas are used to... Obtain the target offset value of the transfer path, where This represents the target offset value of the transfer path from the starting node h to the ending node v. and The working status of nodes h and v is represented. The target offset values ​​of all transfer paths of the transfer method are calculated and accumulated to obtain the total target offset value Gi(s), where s represents the s-th transfer method. The transfer method with the smallest total target offset value Gi(s) is selected as the proposed transfer method.

6. The intelligent power supply and load balance analysis method for active distribution networks according to claim 5, characterized in that, By collecting and analyzing conventional parameter data on distributed photovoltaic power generation, load of each branch of the active power grid, power supply structure, and load characteristics, the overall source-load balance rate for implementing the power transfer method is obtained. The specific method is as follows: Multiple simulation experiments were conducted at different time periods. Conventional parameter data for distributed photovoltaic power generation, load on each branch of the active power grid, power supply structure, and load characteristics were collected for each simulation. These data were then analyzed using formulas. Let N represent the source-load balance rate of branch j, where N represents the existence of N time points, and k represents the distributed photovoltaic (PV) system pre-installed on branch j. This represents the power source type coefficient of distributed photovoltaic (PV) system k. This represents the load characteristic coefficient of the j-th branch. This represents the load on branch j at time t. Let represent the power generation of distributed photovoltaic k at time t, Br(k) represent the failure rate of distributed photovoltaic k under the evaluated transfer method, and Ch(j) represent the load mutation rate of branch j under the evaluated transfer method. After calculating the source-load balance rate of all branches under the transfer method, the results are summed to obtain the total source-load balance rate Qi(s) of the transfer method.

7. The intelligent power supply and load balance analysis method for active distribution networks according to claim 6, characterized in that, By comparing the overall source-load balance rate and overall target offset of the comprehensive power transfer methods, the optimal power transfer method is obtained. The specific method is as follows: Set a total source load balance threshold Kj, and eliminate power transfer methods with a total source load balance rate greater than or equal to this threshold. For power transfer methods with a total source load balance rate less than Kj, set a source load safety value YH(s), and use the formula... Represents the source load safety value, where and The weighting coefficients for the total source load balance rate and the total target offset value are respectively represented. By comparing the magnitude of YH(s), the transfer method with the smallest YH(s) value is selected as the optimal transfer method.

8. An intelligent power transfer source-load balance analysis system for active distribution networks, applied to the intelligent power transfer source-load balance analysis method for active distribution networks as described in any one of claims 1-7, characterized in that, It includes an active power distribution network data monitoring module, an operational basic data prediction module, a security constraint and objective function establishment module, a power transfer method generation module, and an optimal power transfer method visualization module; Active power distribution network data monitoring module: Real-time monitoring of basic operational data of the active power distribution network through data acquisition equipment; The basic operational data prediction module classifies and sorts the real-time operational basic data collected by monitoring, and performs predictive analysis on the operational basic data for the next time point. Safety constraint and objective function establishment module: Set safety constraint conditions including topology constraints, node voltage constraints, distributed photovoltaic power output constraints and branch capacity constraints, and set the transfer objective to maximize the transfer load, minimize the number of switching operations and minimize the transfer time; The module for generating the transfer method comprehensively considers the power system, communication system, and embedded components of the active distribution network's interaction interface, performs real-time simulation, sets the target offset value of the transfer path, and obtains the proposed transfer method. The optimal transfer method visualization module compares the total source-load balance rate and total target offset of the comprehensive transfer methods to obtain the optimal transfer method. The module then uses GIS calculations to create an intelligent transfer map corresponding to the optimal transfer method for visualization.

Images