Novel power distribution system parallel simulation method based on balanced network division strategy

Abstract

The invention discloses a novel power distribution system parallel simulation method based on a balanced network division strategy, and the method comprises the steps: 1, building a two-tuple mathematical model of a power network topology structure based on an undirected graph segmentation theory, and constructing a node and line calculation complexity value set of the power network topology structure based on the two-tuple mathematical model; step 2, taking a transmission line as a network division point of the power network topology structure, and establishing a network division scheme of the power network topology structure by adopting a combination generation algorithm; then combining the unbalance degree evaluation index of each sub-network scheme, and performing traversal by using a depth-first search algorithm to obtain a balanced sub-network strategy of the power network topology structure; and step 3, realizing natural decoupling of each sub-network based on a Bergeron line decoupling model, and further realizing a parallel simulation method for load balancing of each sub-network. A simulation result shows that the network division strategy can fully consider the computing resource allocation condition of each CPU core, and the parallel simulation efficiency can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of electromagnetic transient parallel simulation, and in particular relates to a novel parallel simulation method for a power distribution system based on a balanced network distribution strategy. Background Art

[0002] With the expansion of distribution networks and the integration of new energy devices, power network topologies are becoming increasingly complex. Traditional serial simulation methods face a bottleneck of low simulation efficiency when handling large-scale systems. Parallel simulation technology significantly improves computational efficiency and shortens simulation time by dividing tasks into multiple subtasks and solving them in parallel using multi-core processors.

[0003] However, uneven distribution of subtasks can overload some computing units, increasing synchronization wait time and thus affecting overall simulation performance. Therefore, balanced resource allocation is a key research topic. By implementing appropriate network partitioning strategies and optimization algorithms, we can ensure a balanced distribution of computational load across subtasks, helping to improve simulation efficiency. Summary of the Invention

[0004] This paper abstracts the power network topology into a binary mathematical model, representing the quantified values of the simulation computational complexity using two sets. Combining the imbalance indicators of the network partitioning scheme, a combinatorial generation algorithm and a depth-first search algorithm are applied to solve a set of balanced network partitioning strategies. Based on this, a transmission line Bergeron model is constructed to decouple the system, thereby achieving a parallel simulation method with balanced distribution of subtasks.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] A new parallel simulation method for power distribution system based on balanced network distribution strategy includes the following steps:

[0007] Step 1: Based on the undirected graph partitioning theory, a binary mathematical model of the power network topology structure is established, and based on the binary mathematical model, a set of calculation complexity values of nodes and lines of the power network topology structure is constructed;

[0008] Step 2: Using transmission lines as sub-network points of the power network topology, a combination generation algorithm is used to establish a sub-network scheme for the power network topology; then, combining the imbalance evaluation index of each sub-network scheme, a depth-first search algorithm is applied to traverse and obtain a balanced sub-network strategy for the power network topology;

[0009] Step 3: Based on the Bergeron line decoupling model, a natural decoupling of each sub-network is achieved, thereby realizing a parallel simulation method for load balancing of each sub-network.

[0010] Preferably, the binary mathematical model G(V, L) of the power network topology is expressed as:

[0011]

[0012] Where: V represents the node set; n represents the number of nodes; L represents the set of lines, v p 、v q Represents different nodes connected at both ends of the line.

[0013] Preferably, the simulation calculation amount of each node in the power network topology is assigned a value V m , the simulation calculation value of each line is assigned L m It is expressed as follows:

[0014]

[0015] Where: x g is the computational complexity value of the g-th node; y z The computational complexity of the z-th line.

[0016] Preferably, the imbalance evaluation index of each network division scheme is calculated according to the following formula:

[0017]

[0018] Where: N represents the number of subnets, w1, w2…w N Represent the simulation computation amount of N sub-networks respectively.

[0019] The new parallel simulation method for power distribution system based on balanced network strategy according to claim 4 is characterized in that the simulation calculation amount w of any sub-network i i It is obtained by the following method: Calculate the complexity value x for the nodes included in the sub-network g And the circuit calculation complexity value y z Sum the result, and then divide the sum by the simulation step size H to get the simulation calculation amount w of each sub-network. i .

[0020] Preferably, the Bergeron line decoupling model is expressed as:

[0021]

[0022] Where: Z is the wave impedance, τ is the transmission delay; i k Indicates the current at the beginning of the line, u k Indicates the voltage at the beginning of the line, u m Indicates the voltage at the end of the line, i m Indicates the current at the end of the line.

[0023] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects:

[0024] This paper provides an effective balanced partitioning strategy for parallel simulation of distribution networks, which can shorten the synchronization wait time of each subtask. Based on this strategy, a simple implementation algorithm is proposed. This strategy and algorithm can effectively partition distribution networks of any topology, significantly improving the overall performance of parallel simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of steps 1 to 3 in the present invention.

[0026] Figure 2 This is the improved IEEE34 node system topology diagram in the present invention

[0027] Figure 3 This is the simulation waveform diagram in the second embodiment of the present invention DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is described in detail below with reference to the accompanying drawings. It should be understood that the specific implementation examples described herein are only used to explain the present invention and are not intended to limit the present invention.

[0029] The technical solution of the present invention is further described below with reference to the accompanying drawings and specific embodiments:

[0030] Example 1

[0031] like Figure 1 As shown in FIG, a new parallel simulation method for a power distribution system based on a balanced network distribution strategy includes the following steps:

[0032] Step 1: Establish a binary mathematical model G(V, L) of the power network topology structure:

[0033] The establishment of the binary mathematical model G(V, E) specifically includes two steps, as follows:

[0034] Step 11: Establish a binary mathematical model of the power network topology:

[0035] A binary mathematical model of the power network topology is established based on the undirected graph partitioning theory.

[0036] Specifically, if there are n nodes and k lines in the power network topology, in order to describe and analyze the topology of the power network, a two-tuple mathematical model G(V, L) is introduced to represent the topology of the power network. In this two-tuple mathematical model, V represents the set of nodes and L represents the set of lines, that is:

[0037] V={1,2,……,n}

[0038] L={(v p ,v q )|v p ,v q ∈V,p≠q}

[0039] In the formula, each line connects two different nodes v p , v q .

[0040] Furthermore, to more clearly describe the circuit connections of the power network, an adjacency list data structure is often used. This data structure represents the power network by recording the nodes adjacent to each node. This method is not only intuitive but also has significant advantages in terms of storage space and query efficiency.

[0041] Specifically, for a given binary mathematical model G(V, L), its adjacency table function is expressed as:

[0042] Adj(v)={u∈V|(v,u)∈L}

[0043] Where Adj(v) is the set of all vertices adjacent to vertex v, and u is a vertex in set V.

[0044] Step 12: Construct the system's node and line calculation complexity value set:

[0045] In the electromagnetic transient simulation of power network, the simulation calculation amount of different nodes and lines varies greatly. In order to accurately reflect this effect, the set V m Indicates the simulation calculation value of each node, the set L m Indicates the simulation calculation value assignment of each line:

[0046] V m ={v g :x g |v g ∈V}

[0047] L m ={(v p , v q ):y z |(v p , v q )∈L}

[0048] Where x g is the computational complexity value of the g-th node; y z The computational complexity of the z-th line.

[0049] In the above formula, x g 、y z The value of is determined based on the number of additional nodes brought by the component. In this way, the simulation calculation amount of each node and line in the system can be intuitively reflected in the set V m and L m In this way, it is convenient to conduct a comprehensive analysis of the simulation calculation complexity of the entire distribution system after the network is divided.

[0050] Step 2: Consider the method of selecting a network partitioning strategy that balances the subtask load distribution:

[0051] Using transmission lines as the sub-network points of the power network topology, a combinatorial generation algorithm is used to establish the sub-network scheme of the power network topology. Then, combined with the imbalance evaluation index of each sub-network scheme, a depth-first search algorithm is applied to traverse and obtain the balanced sub-network strategy of the power network topology. Specifically, it includes two steps, as follows:

[0052] Step 21: List all subnets:

[0053] This invention utilizes a transmission line-based decoupling method that is independent of the grid topology, ensuring that parallel processing of sub-networks does not impose additional computational burden. Given that the transmission lines in the system serve as sub-network points, a combinatorial generation algorithm can be applied to enumerate all possible combinations of separate lines. Mathematically, this is achieved by generating all possible fixed-size subsets of the line set L.

[0054] Step 22: Calculate the imbalance index value of each sub-network based on the depth-first search algorithm:

[0055] The depth-first search algorithm can be applied to the traversal of the graph. During the search process, the algorithm will search down along every possible path until it can no longer search deeply, then it will fall back to the node just visited. Each node can only be visited once. The present invention uses the depth-first search algorithm to traverse the sub-network under each network partitioning strategy, and takes the value x of the nodes contained in the sub-network. g and line value y z On this basis, the total is divided by the simulation step size H, and the simulation calculation amount w of each sub-network is finally obtained. i , and then the imbalance index value of each network strategy is obtained by solving the following formula:

[0056]

[0057] The index values under all network partitioning strategies are sorted. While meeting the simulation accuracy requirements, the group with the smallest index value is selected as the network partitioning strategy proposed by this invention. Therefore, for network topologies of any scale, a set of network partitioning strategies that calculate load balancing distribution can be obtained according to the above method.

[0058] Step 3: Parallel simulation method based on Bergeron circuit model:

[0059] This step realizes the natural decoupling of each sub-network based on the Bergeron line decoupling model, and then realizes a parallel simulation method with load balancing of each sub-network.

[0060] The Bergeron line decoupling model uses the propagation characteristics of electromagnetic waves in transmission lines and introduces historical current terms to reflect the electrical information of the opposite system, thereby achieving decoupling of the systems at both ends of the line. The circuit structure of a typical Bergeron model is as follows: Figure 2 As shown in the figure, assuming that the line length is l, its unit inductance value is L0, and its unit capacitance value is C0, then the relationship between the voltage and current at the beginning and end of the line is:

[0061]

[0062] Where Z is the wave impedance, and its value is τ is the transmission delay, and its value is

[0063] In this way, the computational tasks of the subsystems at both ends of the line can be assigned to different computational units to achieve parallel simulation of electromagnetic transients and improve simulation efficiency.

[0064] The improved IEEE34 node test feeder system is built as follows Figure 2 As shown, the system parameters are shown in Table 1, and a photovoltaic power generation unit is connected to nodes 890 and 840 respectively.

[0065] Table 1 IEEE34-node system parameters

[0066]

[0067] Representing the system as a binary tuple and assigning values to each line and node, a combination generation algorithm and a depth-first search algorithm were used to determine a set of networking strategies with the lowest imbalance index value. This revealed that the balanced network split was between nodes 858 and 832. Furthermore, based on practical engineering experience and commonly used network splitting methods, a traditional AC-DC network splitting strategy was selected, splitting the network between nodes 854 and 852. To validate the proposed new parallel simulation method for distribution systems based on a balanced network splitting strategy, a single-rate parallel simulation scenario was set up. The simulation time for each network splitting strategy at different simulation step sizes is shown in Table 2.

[0068] Table 2 Single-rate parallel simulation time consumption of IEEE 34-node system

[0069]

[0070] Table 2 shows that applying parallel simulation technology can effectively improve simulation efficiency. The balanced network partitioning strategy proposed in this paper achieves higher simulation efficiency than traditional network partitioning strategies in parallel simulation scenarios. In the example presented in this paper, with a simulation step size of 10 μs, the proposed network partitioning strategy achieves an 8.3% improvement in simulation efficiency compared to traditional network partitioning strategies, and the parallel speedup ratio reaches 1.45.

[0071] In order to verify the accuracy of the proposed network division strategy, the present invention analyzes the effective value output of the line voltage at the fault point and compares it with the high-precision serial simulation results. Figure 3 It can be seen that the output waveform of the parallel simulation under the proposed strategy is basically consistent with the standard model, and its accuracy meets the requirements of high-performance electromagnetic transient simulation.

[0072] It can be seen that the network division strategy proposed in the present invention solves the problem of uneven distribution of simulation computing resources under the traditional network division strategy, and helps to improve the efficiency of parallel simulation while ensuring the simulation accuracy.

Claims

1. A new parallel simulation method for power distribution system based on balanced network distribution strategy, characterized by: The steps include: Step 1: Based on the undirected graph partitioning theory, a binary mathematical model of the power network topology structure is established, and based on the binary mathematical model, a set of calculation complexity values of nodes and lines of the power network topology structure is constructed; Step 2: Using transmission lines as sub-network points of the power network topology, a combination generation algorithm is used to establish a sub-network scheme for the power network topology; then, combining the imbalance evaluation index of each sub-network scheme, a depth-first search algorithm is applied to traverse and obtain a balanced sub-network strategy for the power network topology; Step 3: Based on the Bergeron line decoupling model, a natural decoupling of each sub-network is achieved, thereby realizing a parallel simulation method for load balancing of each sub-network.

2. The novel parallel simulation method for power distribution system based on balanced network division strategy according to claim 1 is characterized in that: The binary mathematical model G(V,L) of the power network topology is expressed as: Where: V represents the node set; n represents the number of nodes; L represents the set of lines, v p 、v q Represents different nodes connected at both ends of the line.

3. The novel parallel simulation method for power distribution system based on balanced network division strategy according to claim 1 is characterized in that: The simulation calculation value V of each node in the power network topology m , the simulation calculation value of each line is assigned L m It is expressed as follows: Where: x g is the computational complexity value of the g-th node; y z The computational complexity of the z-th line.

4. The novel parallel simulation method for power distribution system based on balanced network division strategy according to claim 1 is characterized in that: The imbalance evaluation index of each network solution is calculated according to the following formula: Where: N represents the number of subnets, w1, w2…w N Represent the simulation computation amount of N sub-networks respectively.

5. The novel parallel simulation method for power distribution system based on balanced network division strategy according to claim 4 is characterized in that: The simulation computation amount w of any sub-network i i It is obtained by the following method: Calculate the complexity value x for the nodes included in the sub-network g And the circuit calculation complexity value y z Sum the result, and then divide the sum by the simulation step size H to get the simulation calculation amount w of each sub-network. i .

6. The novel parallel simulation method for power distribution system based on balanced network division strategy according to claim 1 is characterized in that: The Bergeron line decoupling model is expressed as: Where: Z is the wave impedance, τ is the transmission delay; i k Indicates the current at the beginning of the line, u k Indicates the voltage at the beginning of the line, u m Indicates the voltage at the end of the line, i m Indicates the current at the end of the line.

Images