Data network traffic analysis and optimization strategy of real-time power grid dynamic monitoring system for wide-frequency measurements

0 Introduction

With rapid industrial development, the demand for electricity in various areas has also been increasing [1].The power electronic equipment used in a power grid is becoming increasingly diversified, and electrical quantity signals are becoming progressively abundant.In addition, the scale of the data network of a power system is expanding.Wide-area measurement systems (WAMSs) have been commonly used and developed, prompting increasing number of applications and demands based on dynamic measurement data [2].With the use of wide-frequency measurement devices, the monitoring range of real-time dynamic monitoring systems for a power grid has been significantly expanding.

The real-time dynamic monitoring system used for the domestic power grid of China has been rapidly developing since 2000.In 2006, the State Grid Corporation of China led the release of an enterprise standard, Q/GDW131-2006 [3].Based on the actual needs of the power system of China, this standard specifies in detail the format, conversation traffic,and technical performance of the communication protocol for a phasor measurement unit (PMU), which has been subsequently extensively used.Since then, the power system information network of China has continued to expand, and the power system has gradually strengthened its dependence on the information system.Moreover, traffic monitoring of the data network of the real-time dynamic monitoring system has become increasingly important.In 2011, the WAMS and Time Synchronization Standards Working Group optimized and improved the communication protocol of a WAMS based on a large amount of data and experience summaries from on-site operations, and released GB/T26865.2-2011 [4].Using a unified dispatching and control platform, the State Grid Corporation of China accesses and shares the operating data of different state grids based on a state, regional, and provincial three-tier structure.Thus,strengthening the traffic analysis, monitoring, and control of the data network has gradually become a key prerequisite for achieving stable access and efficient sharing of power system data [5].

At present, the PMUs operating in the domestic power grid of China mainly collect 50-Hz power frequency electrical quantity information.With the employment of power electronic equipment such as new energy generation,ultrahigh-voltage (UHV) DC, and voltage source converter(VSC) transmission, the nonpower frequency electricity volume in the power grid has significantly increased.This directly harms the electrical equipment in the power grid and causes wide-band multimode oscillations, thus making the power grid stability problematic [6-7].To deal with this problem, the PMUs in China have been improved and upgraded, to compensate the absence of an oscillation monitoring function, which has been applied in engineering based on theoretical studies [8-12].However, owing to the complexity of the user-side load in the distribution network, numerous high-frequency harmonics and interharmonics occur in the power grid for a long time.The typical harmonics in the current power system of China have been determined to gradually expand from less than 2 kHz to 9 kHz or even wider [13-15].Therefore, simple improvements and upgrades cannot effectively manage its wide-frequency measurement needs.To realize realtime monitoring of high-frequency electrical information,research institutions and manufacturers have developed wide-frequency measurement devices with a sampling frequency of 12.8 kHz.The objectives were to accurately measure and record mechanical and electrical transient processes and perform traffic analysis and control of highfrequency electrical information [16].Simultaneously,the communication protocol has been expanded for widefrequency electrical information transmission, to meet the application requirements of the main and distribution networks.

The system architecture and data flow transmission process of a WAMS are shown in Fig.1.The PMU network is a distributed local area network [17].PMU collection devices receive high-precision unified timing signals from a substation.They collect high-speed synchronous electrical quantity information and calculate phasor quantities [18].A phasor data concentrator calibrates the phasor information of multiple acquisition devices at the substation.The timecalibrated phasor information is collected into PMU data frame messages and encrypted by a longitudinal encryption device for transmission.Subsequently, the real-time data and files stored in the transmission process are analyzed and estimated.Different levels of the flow characteristics of the power system data network are obtained for the transmission guidance of the power system services.

Fig.1 WAMS architecture and data flow diagram

Considering the new demand for real-time dynamic monitoring of the power system of China, offline data transmission burden has increased.The main drawback of the existing communication network is the insufficient bandwidth [19], which cannot satisfy the increase in the high-frequency electrical signals and the sampling rate of the wide-frequency measurement device.

Therefore, this study systematically analyzes the current network traffic of wide-frequency measurement devices, and further establishes a data network optimization strategy for the present real-time dynamic monitoring system.The optimized plan is designed to overcome the weaknesses of the underlying communication protocol,source data, and data transmission process in the network.The main contributions of the study are summarized as follows:

1) Real-time traffic monitoring and analysis are conducted on the data network of the power information system of China for wide-frequency measurements.

2) Corresponding network optimization strategies are designed to solve the current data transmission problems of the communication network.

3) Based on the traffic analysis results and the optimization strategies of the present real-time dynamic monitoring system, a complete data network optimization architecture is designed.It can optimize the structure and data of the wide-frequency measurement power system.

1 Wide-frequency measurement transmission protocol

The data transmission protocol defined in GB/T26865.2 for the original real-time dynamic monitoring system has high data model scale and transmission efficiency.However,it is mainly defined for the transmission of power frequency data.It only uploads 50-Hz fundamental electrical volume data, without considering the transmission of harmonic and interharmonic data.Therefore, it cannot be directly applied to the transmission of wide-frequency measurement data.It is necessary to supplement and extend the existing communication protocol.

Wide-frequency measurement devices have significantly expanded the monitoring range of the real-time dynamic monitoring system of the power grid.The wide-frequency electrical quantities include fundamental waves and the harmonic and interharmonic components generated by inverter-based devices.The monitoring frequency of the wide-frequency measurement devices has been expanded to 0-2500 Hz.However, increase in the amount of real-time monitoring data increases the burden on the communication network.

To realize flexible and efficient transmission of lowfrequency, power-frequency, and harmonic and interharmonic synchronous phasors, this study extends the existing transmission protocol of the wide-frequency measurement devices.The extended transport protocol defines harmonic and interharmonic configuration and data frames as well as the associated command frames.The transmission process of the harmonic and interharmonic data is similar to that of the fundamental wave phasor.It shares the phasor and analog names with the fundamental wave data.The extended transmission protocol can distinguish specific signals only by serial number, without defining various harmonic and interharmonic signal names.Therefore, it can significantly facilitate project implementation and master station application.

1.1 Transmission protocol expansion scheme

The transmission protocol of the wide-frequency measurement devices conforms to the original transmission control protocol/internet protocol (TCP/IP) communication mode of the PMU transmission protocol.The configurations of the management, data, and offline file transmission port are not required to be changed.It only needs to expand the number of the GB/T 26865.2 communication protocol configuration (CFG) transmission channels and add a new harmonic/inter-harmonic phasor channel configuration.Channel time division multiplexing noncontinuous frequency hopping transmission timing is used for the data transmission, as shown in Figs.2 and 3.Each data frame contains only one harmonic/interharmonic frequency point data point.The interval between two adjacent harmonic/interharmonic data points is 40 ms.The transmission of 50 sets each of harmonic phasor and interharmonic dominant component data is completed every second.When the real-time data reporting rates are set as 25 fps, 50 fps, and 100 fps, the content of the transmitted data is adjusted accordingly [21].

1.2 Description of transmission frame structure of wide-frequency measurement devices

The wide-frequency measurement devices use TCP/IPbased transmission frames to send data.The frame types mainly include data and configuration frames.Each frame structure contains inherent and variable frame contents.The inherent frame length is provided in Table 1.

Table 1 Inherent frame length of transmission frame of wide-frequency measurement devices

?

Fig.2 Sequential data transmission of harmonic components

Fig.3 Sequential data transmission of interharmonic components

The variable frame length is related to the number of collectors configured in the field and the number of access intervals.For a single bay, the frame length of the information object contained in the data frame is provided in Table 2.

Table 2 Data frame length of single bay information object in data frame

Information object Number of channels Channel length (B) Remark Fundamental frequency phasor 8 4 Ua/Ub/Uc/U1/Ia/Ib/Ic/I1 Fundamental frequency analog 26 2 P/Q/F/ROCF/SFN/SFP Harmonic phasor 6 4 Ua/Ub/Uc/Ia/Ib/Ic Harmonic analog 6 2 six-channel harmonic frequency Interharmonic phasor 6 4 Ua/Ub/Uc/Ia/Ib/Ic Interharmonic analog 6 2 six-channel inter-harmonic frequency

2 Analysis of network traffic of widefrequency measurement device

Based on the expanded transmission protocol of the wide-frequency measurement devices, this study analyzes and calculates the traffic of the application layer in data transmission.The wide-frequency measurement devices are deployed in sets.Each set of the wide-frequency measurement devices is equipped with a maximum of eight acquisition devices, and each acquisition device can collect AC electrical quantities in four bays.The wide-frequency measurement devices are compatible with the PMUs, which store dynamic files and continuous recording files every minute.

2.1 Real-time traffic analysis

When a wide-frequency measurement device only transmits PMU dynamic real-time dynamic data, the data flow change is as shown in Fig.4.Based on Table 3,with the variation in the number of access bays and the transmission frame rate, the traffic changes from 2.7 kB/s to 269.3 kB/s.

Fig.4 Dynamic data transmission flowchart (single-channel)

Table 3 Calculation of PMU data transmission flow of widefrequency measurement devices

Access bay Number of acquisition devices PMU realtime data -25 Hz (kB/s)PMU realtime data -50 Hz (kB/s)PMU realtime data -100 Hz (kB/s)1 1 2.7 5.5 10.9 4 1 8.9 17.8 35.5 8 2 17.2 34.5 68.9 12 3 25.6 51.2 102.3 16 4 33.9 67.9 135.7 20 5 42.3 84.6 169.1 24 6 50.6 101.3 202.5 28 7 59.0 118.0 235.9 32 8 67.3 134.7 269.3

Fig.5 shows the change in the frame rate of the wide-frequency data transmission traffic including the fundamental frequency and the harmonics and interharmonics with the number of bays.As noticeable from Table 4, the data flow varies from 4.5 kB/s to 494 kB/s.The wide-frequency data transmission traffic is 1.64-1.84 times the dynamic data traffic, which increases with the number of bays, regardless the reporting rate.

Table 4 Calculation of wide-frequency data transmission flow of wide-frequency measurement device

?

Fig.5 Wide-frequency data transmission flowchart(single-channel)

The calculation and analysis summarized in Tables 3 and 4 are based on single-channel communication.For UHV substations, the wide-frequency measurement devices need to communicate with the state, regional, and provincial dispatching master stations (including their main and backup dispatching systems).Moreover, at maximum, six sets of communication channels need to be established.The dynamic data transmission traffic is 16.2-1615.8 kB/s, and the wide-frequency data transmission traffic is 27-2964 kB/s.

2.2 File transfer traffic analysis

The continuous recording files have the largest file sizes.Based on the completion of a single file transmission task within a specified time, the traffic of a single master station calling file is calculated, and the results are provided in Table 5.Considering the number of intervals and the number of acquisition devices, with a compression rate of 0.6, the storage size of a continuous recording file of 6.4 kHz in 1 min is 4.4-98.4 MB.

Table 5 Single file call traffic estimation

Continuous recording data for 6.4 kHz storage in 1 min (MB)Required bandwidth for querying in 30 s(MB/s)Required bandwidth for querying in 60 s(MB/s)4.4 0.14667 0.0733 12.3 0.41000 0.2050 16.7 0.55667 0.2783 29.0 0.96667 0.4833 41.3 1.37667 0.6883 53.6 1.78667 0.8933 65.9 2.19667 1.0983 78.2 2.60667 1.3033 90.5 3.01667 1.5083 98.4 3.28000 1.6400

2.3 Experimental results

Based on the plant equipment configuration and bay access conditions, this study analyzes and calculates the application layer message traffic of a wide-frequency measurement device in data transmission.In the case of a single channel, the dynamic data transmission traffic of a wide-frequency measurement device is 2.7-269.3 kB/s,

whereas the traffic with six channels is 16.2-1615.8 kB/s.The wide-frequency data transmission traffic for a singlechannel case is 4.5-494 kB/s, and the traffic with six channels is 27-2964 kB/s.For file transfer, the data traffic requirement is 0.15-3.28 MB/s and the completion time is within 30 s.

In view of the traffic analysis results of the widefrequency measurement devices, this study analyzes a UHV substation, state direct dispatching power plant, 500-kV power plant as well as substations, and 220-kV and below power plants as well as substations.Based on the current transmission protocol, the real-time data are considered at a maximum of 100 fps, and the recommendations for the bandwidths of the different types of substations or power plants are listed in Table 6.

Table 6 Substation/power plant-side bandwidth recommendations

Type of station Number of intervals Dispatching master station Recommended bandwidth(MB/s)UHV substation(Converter station) 32 State, regional,and provincial stations 60 State direct dispatching power plant 16 State, regional,and provincial stations 30 500-kV substation 32 Regional and provincial stations 50 500-kV power plant 16 Regional and provincial stations 25 220-kV and below power plant and substation 16 Provincial station 20

3 Optimization strategy of transmission protocol for real-time dynamic monitoring system

According to the network traffic analysis results of the wide-frequency measurement devices, the communication network of the current real-time dynamic monitoring system mainly has the following problems:

1) Real-time data sharing based on concurrent acquisition significantly increases the network transmission traffic.It increases the requests to the existing network bandwidth.

2) Owing to the limitation of the communication bandwidth, using the TCP to transmit real-time data causes unnecessary waste of the network resources.

3) Real-time data transmission is characterized by strong real-time performance.Offline data transmission has the characteristics of weak real-time performance, low reliability, and generally a large transmission volume.Owing to the different types of data transmission requirements, the requirements of the transmission bandwidth of the data link are increased.

Summarizing, the current problem encountered by the communication network of the present real-time dynamic monitoring system is the significant increase in the widefrequency measurement data, which raises the requirements of the communication bandwidth.Simultaneously, the lack of bandwidth for the scheduling data network under the condition of increasing offline data transmission burden is increased.Aiming at real-time data sharing optimization,this study designs a transmission protocol optimization scheme for the real-time dynamic monitoring system from the underlying communication protocol and the source data.The specific framework is as follows.

As the WAMS network adopts a distributed structure[22], the measurement information is collected and sent to the master stations in sequence.When a certain dispatching master station needs to obtain measurement data of a certain wide-frequency measurement device deployed under other dispatching master stations, it can only be realized by a historical data exchange between the master stations.With the accelerated deployment of wide-frequency measurement devices, real-time data sharing based on concurrent collection significantly increases the network transmission traffic and raises the requirements of the existing network bandwidth.Therefore, this study designs a new online data sharing scheme to avoid concurrent collection and reduce the network bandwidth requirements for data transmission.

3.1 Bottom-level communication protocol optimization

The wide-frequency measurement transmission protocol currently adopted uses the underlying communication protocol of the TCP for both real-time and offline data transmission.Owing to the different characteristics of the TCP and the user datagram protocol (UDP) [23], this study optimizes and adjusts the underlying communication protocol to meet the transmission requirements for the realtime data sharing optimization scheme in the application scenarios of the WAMS network [24].

3.2 Wide-frequency measurement device data optimization

The optimization of the data of the wide-frequency measurement devices is involves data processing and data analysis.The insufficient bandwidth of the scheduling data network is a resource allocation problem caused by offline data transmission requirements.When the link processing capacity is saturated, the problem needs to be solved by optimizing the source data.This study uses compressed transmission to process the source data of the wide-frequency measurement devices.Moreover, it designs service and local protocols based on the principle of remote procedure call (RPC) and the application requirements for different types of transmission data, such as real-time,offline, and statistical data, effectively reducing the burden of real-time data transmission.

4 Real-time dynamic monitoring system transmission protocol optimization scheme

This study designs a new online data sharing scheme based on the characteristics of the WAMS network.

As shown in Fig.6, the current data transmission method lacks real-time performance and stability, and the overall design is complex.This study extends the message transmission structure of GB/T 26865.2-2011.The identification code, DC_IDCODE, of the data concentrator is distinguished from the identification code, PMU_IDCODE, of a PMU.Simultaneously, PMU_IDCODE is added to a data frame to accurately and independently transmit the content of the data frame to each PMU.

Fig.6 Schematic of current data interaction

As shown in Fig.7, when a wide-frequency measurement device or a PMU of a certain substation is connected to the dispatching master station, the master station adds the DC_IDCODE of the device, PMU_IDCODE, substation,and other information to the subscription list to provide external data forwarding services.Other dispatching master stations can find, subscribe, and unsubscribe the CFG and data messages of a wide-frequency measurement device or a PMU of one substation by searching the service list.After the data of this PMU or wide-frequency measurement device of one substation are sent to the communication front processor, the master station directly forwards the data message of the entire station or even the data of a single collection unit according to the subscription scenario.This data transmission method can effectively reduce the requirements of the network bandwidth.

Fig.7 Schematic of data interaction after data field optimization

4.1 Selection of bottom layer protocol

To meet the transmission requirements of the online data sharing solution, the underlying transmission protocol of widefrequency measurements needs to be optimized and adjusted.Real-time data transmission has the characteristics of strong real-time performance, high fault tolerance, and predictability.Offline data transmission is characterized by weak real-time performance, low reliability, and generally a large transmission volume.Therefore, the use of the TCP for real-time data transmission causes unnecessary waste of the network resources when the communication bandwidth resources are limited.The UDP and upper-layer applications can achieve data transmission by online data sharing.They can meet the requirements of real-time and offline data transmission applications simultaneously and alleviate the current shortage of insufficient network resources.

4.2 Wide-frequency measurement device data optimization implementation

1) Data compression and transmission in wide-frequency measurement device

A wide-frequency measurement device can store and transmit offline data files by compression.Subsequently,the master station decompresses and stores the data.As summarized in Table 7 and shown in Fig.8, we test and verify the performance of four mainstream compression algorithms: as gz, bz2, zip, and zstd.Taking a single acquisition device with six interval data channels as an example, different algorithms and compression parameters are selected to compress offline data.The four algorithms achieve compression ratios of approximately 30%-75%,which significantly reduces the amount of offline file data.Although compression processing increases the processing overhead at both ends of the network, it reduces the overhead of data link transmission.Therefore, it can optimize the network transmission environment.

Table 7 Data compression rate statistics under different compression algorithms

Level gzip zip bz2 zstd Compression rate Throughput Compression rate Under control Compression rate Under control Compression rate Under control 1 71.3% 37.8 71.2% 36.3 56.9% 10.4 73.3% 229.2 2 69.7% 36.9 69.6% 35.8 45.5% 11.5 63.9% 172.4 3 69.2% 36.4 69.1% 22.1 40.5% 11.7 58.3% 112.8 4 72.6% 30.7 72.5% 23.7 38.0% 11.6 55.0% 88.2 5 72.4% 27.8 72.3% 24.7 36.6% 11.5 52.4% 43.8 6 72.5% 24.4 72.4% 20.6 35.6% 11.6 52.2% 30.1 7 72.5% 22.3 72.4% 16.9 35.1% 11.2 51.9% 21.9 8 72.4% 18.3 72.3% 16.9 34.7% 10.4 51.6% 15.5 9 72.4% 18.3 72.3% 16.4 34.4% 11.0 51.6% 13.6

Fig.8 Scheme of second harmonic/interharmonic timing transmission

2) Local and service data analysis

For different types of transmission data such as realtime, offline, and statistical data, the protocol is designed for localization and services based on the RPC principle and the application requirements.As shown in Fig.9,for data that can be calculated locally in an acquisition device, such as low-frequency oscillation monitoring, sub/supersynchronous oscillation monitoring, and harmonic/interharmonic spectrum analysis, only the statistical result data are uploaded to the master station (client).This effectively reduces the amount and frequency of real-time data upload.

Fig.9 Data analysis service and modularization

The implementation process is as follows:

a) The master station (client) calls services such as lowfrequency oscillation monitoring, sub/supersynchronous oscillation monitoring, and harmonic/interharmonic spectrum analysis by local calling (interface).

b) After the master station stub receives the call instruction,it assembles the parameters.Subsequently, the master station sockets send the message to the sockets of a wide-frequency measurement device by network transmission.

c) The wide-frequency measurement device (server)stub analyzes the task after receiving it and calls the local collection device and functions of the concentrator function such as low-frequency oscillation monitoring, sub/supersynchronous oscillation monitoring, and harmonic/interharmonic spectrum analysis for calculations.

d) The wide-frequency measurement device returns the calculated statistical results to the master station stub.The wide-frequency measurement device stub packs the results and subsequently sends them to the master station sockets by the wide-frequency measurement device sockets.

e) The master station stub receives the result message and parses it.

f) The master station obtains the final results.

5 Conclusion

In this study, a detailed t raffic analysis on the data network of China for real-time power grid wide-frequency measurement is conducted.Based on the analysis results,a suitable optimization strategy is designed for the transmission protocol of the current real-time dynamic monitoring system of the power system.A complete optimization scheme is established for the underlying communication protocol, source data, and data transmission process.Compared with the current transmission protocol,the optimization scheme designed in this study can ensure real-time data transmission and stability of the system information transmission for wide-frequency measurements.Experimental results show that the proposed scheme can improve the economic and technical relevance of the power system operation.

Declaration of Competing Interest

We declare that we have no conflict of interest.