Automatic Topology Identification of Distribution Network Areas: Technical Analysis and Practical Application

01

Work Background

Against the backdrop of the national "3060 Strategy" to build a new power system with new energy as the mainstay, Zhejiang Province responded to the National Energy Administration's pilot project for county-wide rooftop distributed photovoltaic (PV) construction, selecting 30 counties (cities, districts) for pilot demonstrations. In supporting the county-wide rooftop distributed PV pilot project, the low-voltage distribution network has become the main battleground for implementing the State Grid Corporation's strategy of "building a world-leading energy internet enterprise with Chinese characteristics" and Zhejiang Company's strategy of "building a diversified, integrated, and highly resilient power grid under the energy internet model." Based on the guiding principles of the "Notice of the State Grid Equipment Department on Issuing the 2021 Work Plan for the Construction and Application Improvement of Smart Distribution Areas" (Equipment Distribution [2021] No. 37) issued by the State Grid Corporation headquarters, the construction of distribution IoT distribution areas utilizes core technology equipment— the smart distribution area terminal— to achieve panoramic monitoring of connected equipment and distributed power sources (PV), thereby supporting the company's business application needs for large-scale consumption and lean operation and maintenance of distributed power sources.

Recently, the State Grid Corporation of China held a meeting to promote the construction of county-wide photovoltaic power generation, pointing out that for distribution transformer areas already connected to distributed photovoltaic systems, full coverage of smart integrated terminals is required. By the end of 2021, the deployment of smart terminals in 90,000 distribution transformer areas across the province will be completed. To strongly support the construction of county-wide rooftop photovoltaic systems, Zhejiang Company plans to carry out large-scale demonstration construction of distribution IoT network extension and application in 11 prefecture-level companies across the province in 2022 (one power supply station will be selected in each prefecture-level company to achieve full coverage). In order to better support the lean operation and maintenance management of distribution transformer areas after the connection of county-wide rooftop low-voltage distributed photovoltaic systems, it is necessary to realize the panoramic status perception of distribution IoT distribution transformer areas based on smart integrated terminals; among them, topology automatic identification technology is the most core basic application for realizing the panoramic status perception of distribution IoT distribution transformer areas, and its importance is self-evident.

02

Overview of Automatic Topology Identification Technology

Distribution network voltage levels are divided into high-voltage distribution voltage (110kV, 63kV, 35kV), medium-voltage distribution voltage (10kV), and low-voltage distribution voltage (0.4kV). A low-voltage distribution area refers to the power supply range or region of a 10kV/0.4kV distribution transformer. Therefore, the topology of a distribution area mainly focuses on identifying the topological relationships at the 0.4kV voltage level, i.e., identifying the "transformer-customer" relationship (here, "customer" can specifically refer to low-voltage users). Due to lagging technology and management, the construction and maintenance of low-voltage distribution networks lag significantly behind those of transmission networks. Distribution areas involve numerous assets with complex connections. Information on assets such as transformers, meter boxes, and electricity meters, as well as their interconnections, needs to be accurately entered into an information system. Previously, this was often done manually, which was time-consuming and difficult to guarantee quality. Furthermore, with the passage of time, the implementation of rural power grid upgrades, meter box replacements, and other construction projects has led to frequent changes to equipment and topology diagrams within the distribution area.

Low-voltage topology error

Main manifestations

● The “household-transformer” correspondence is incorrect, that is, the file division is incorrect, and the meter number that is not in this transformer area (usually a neighboring transformer area) is loaded into the concentrator of this transformer area, which affects the accuracy of the transformer area line loss calculation.

● “User-to-line” means that the relationship between the user and the branch line is missing, and the line connection information between the user and the distribution transformer is lacking. When a fault occurs in the distribution area, it is impossible to quickly and accurately determine the fault section and the power outage section.

Causes of topology errors in low-voltage distribution areas

1

Errors in user meter records during the construction of low-voltage distribution areas caused errors in the low-voltage power distribution topology.

2

Insufficient attention was paid to the transformer substations, and only some transformer substation information was recorded during the construction process, resulting in a lack of topological information for low-voltage transformer substations.

3

When a fault occurs in a low-voltage distribution area, the failure to record or incorrect recording of wiring adjustments during inspection and emergency repair can cause topology errors in the low-voltage distribution area.

In recent years, many technical methods have emerged in the research on transformer substation topology identification, as shown in Table 1.

Table 1 Comparative Analysis of Topology Identification Methods for "House-Transformer" Identification

While the above methods can achieve good identification of transformer substations, methods 2-4 are based on the principle of carrier communication and are affected by factors such as clock synchronization, sampling error, and crosstalk between transformer substations, resulting in relatively low identification accuracy. Furthermore, their applicable scenarios are limited by carrier communication equipment. Methods 5-7 require creating a momentary short circuit to generate huge pulses in voltage and current in order to achieve identification, which will cause huge fluctuations in grid voltage and current, potentially endangering the stable operation of user equipment and distribution network. In addition, it is difficult to miniaturize the relevant equipment, resulting in high promotion costs.

In addition, big data methods are another type of technical solution that has been extensively studied. These methods primarily analyze and solve problems based on electrical quantity data from electricity meters (voltage, current, electricity consumption, instantaneous power, etc.), using a series of heuristic optimization algorithms, stochastic simulation algorithms, least squares, regression optimization, and other algorithms to obtain the relationship between households and transformers. This type of method does not require additional investment costs; however, its accuracy in practical applications cannot be guaranteed due to limitations in the synchronization and accuracy of electrical quantity acquisition, the presence of small and zero-gain meters, electricity theft, line loss, and the challenges of data acquisition and communication. This makes widespread adoption difficult.

With the widespread adoption of county-wide rooftop photovoltaic (PV) installations, electric vehicles, distributed energy storage, and microgrids, low-voltage distribution networks have gradually evolved from passive to active network structures. This has placed higher demands on the topology information of low-voltage distribution transformer areas. On the one hand, the concept of "household" has expanded to include a broader range of "users," encompassing distributed PV, charging piles, distributed energy storage, and other newly added types of end nodes. On the other hand, key branch nodes in the low-voltage network topology, including branch boxes and meter boxes, have also become objects of topology identification. In the construction of the distribution IoT, automatic transformer area topology identification is the most crucial foundational application. It acts like a "map navigation" system for distribution network maintenance personnel, providing intuitive and real-time information on transformer area structure, operation, and fault status, thereby improving repair efficiency and effectively shortening power outage time. Therefore, designing a reliable, accurate, and robust topology identification technology solution that meets the actual business needs of Zhejiang Company is key to promoting the construction of the distribution IoT and the application of the energy internet.

03

Topology recognition technology test

1. Technology platform development

In order to test and verify topology identification technology more comprehensively and efficiently, the State Grid Zhejiang Electric Power Research Institute independently developed the first domestic testing platform for the "edge-end" interaction function of distribution IoT distribution areas, as shown in Figure 1.

Based on testing needs, this platform can connect to various new intelligent devices and sensing devices, such as charging piles, energy storage, photovoltaic systems, and low-voltage monitoring units, to verify the "edge-to-end" interaction capabilities between the connected devices and intelligent integrated terminals. By simulating the physical characteristics of actual distribution areas in a semi-realistic manner, the platform can generate complex topologies with multiple distribution areas, branches, and levels with a single click. It can flexibly construct typical distribution area scenarios in the Zhejiang power distribution network and conduct functional tests for applications such as power quality management and intelligent autonomy. Compared with traditional laboratory principle tests or field operation tests, this platform not only highly replicates the actual low-voltage distribution area operating environment but also possesses the flexibility and customizability of a laboratory testing platform. It can better quantify and evaluate the functions and performance of the tested equipment, meeting the testing and research needs of new technologies and services in the distribution IoT.

2. Status of Testing

Currently, the most mainstream topology identification technology is the micro-current injection method, which has significant advantages in effectiveness and reliability compared to other methods (such as big data analysis or signal correlation methods). However, there are some differences in the implementation schemes of the micro-current injection method, mainly including four methods: "active current injection + time-domain signal detection," "active current injection + frequency-domain signal detection," "reactive current injection + time-domain signal detection," and "reactive current injection + frequency-domain signal detection." Currently, there is no unified technical approach in Zhejiang Province, resulting in significant differences in topology identification technology and its supporting low-voltage extended sensing terminals (such as LTU low-voltage monitoring units). This directly leads to strong coupling in the interaction between edge and end devices (i.e., under the same distribution area, all TTUs and LTUs must use the same equipment, as shown in Figure 2), which is detrimental to the province-wide promotion, construction, and application.

In response to this situation, and to select the optimal topology identification technology solution that meets the application needs of Zhejiang Company's distribution IoT distribution area construction, and to achieve unified functional specifications and interoperability of the topology identification supporting terminal equipment (low-voltage monitoring unit), from September 1st to 15th, 2021, the Electric Power Research Institute conducted a special test on the topology identification solutions of four major technology suppliers in the province. Relevant technical personnel from each supplier participated and confirmed the final test results. The technical solutions involved in the evaluation are shown in Table 2.

Table 2 Comparative Analysis of Topology Recognition Technologies Participating in the Special Test

Based on the "edge-end" interaction function test platform of the distribution IoT, the Provincial Electric Power Research Institute conducted a comprehensive evaluation of the participating transformer area topology identification technology from five dimensions: "device plug-and-play, topology identification accuracy, phase identification accuracy, topology identification time, and identification anti-interference capability." During the test, "four topology structures, 15 test items, and 15 test operating conditions" were set up for the tested topology identification technology solutions. Through various operating conditions under multi-node, multi-level, and multi-element access, the topology identification technology solutions were comprehensively evaluated and tested, allowing for a more efficient, faster, and more effective understanding of the advantages and disadvantages of topology identification technology.

04

Test Result Analysis and Recommendations

Through extensive testing and comparison, a comprehensive comparison of the four submitted proposals was conducted, leading to the following conclusions:

(1) Distributed photovoltaic access, SVG reactive power compensation, charging pile access, and harmonic current injection and power phase change will affect the accuracy of topology identification algorithm;

(2) As the scale of topology nodes and topology levels increases, the time required for the topology world also increases; however, considering the change cycle of the actual transformer area topology structure, the response speed of the existing topology identification algorithm can meet the needs of business applications.

(3) When constructing a local communication network based on HPLC, there is cross-interference between adjacent stations (i.e., cross-networking occurs), which will affect the topology identification algorithm.

(4) The “active current injection + frequency domain detection” technical route showed the best performance in the entire topology identification test, with the best identification accuracy and adaptability.

Based on the above evaluation results, the Zhejiang Electric Power Research Institute, aiming to achieve the goal of "interconnection and interoperability of edge-end devices in the power distribution Internet of Things", optimized and standardized the topology identification technology (characteristic current physical characteristics), topology identification interactive data items and related functions applied to Zhejiang Company, as shown in Table 3.

Table 3. Physical Features of Topology Identification Based on Characteristic Current Injection by Zhejiang Company

05

Summarize

Transformer topology identification technology, as the core foundational application for the construction of distribution IoT based on transformer smart converged terminals, is also the basis for subsequent lean operation and maintenance management of distribution networks and energy internet business applications under cloud-edge-device collaboration. This special evaluation work has laid the foundation for the application of more new distribution IoT technologies and equipment planned for deployment in various cities and prefectures across the province, effectively promoting the digital development of the province's distribution network, meeting safer, more reliable, cleaner, and more diversified electricity demands, and fully supporting the construction of a new power system.