Compliance verification and probabilistic analysis of state-wide power quality monitoring data

1 Introduction

With the rapid development of power system in China,power quality has become a critical issue that draws attentions of both grid operators and power consumers in recent years. Till now, several regional power quality monitoring systems have been built in China. However,power quality data is hardly integrated and effectively utilized, due to lack of communication protocols, and unified specifications for data representation.

Aiming to assess the power qualites of power systems at all voltage levels in the state scale [1-6], as well as to analyze the impact of disturbing sources on the grid, State Grid Corporation of China (SGCC) has built a state-wide Power Quality Monitoring and Analysis System (PQMAS)using big data technologies.

In PQMAS, over 10,000 monitoring points are involved, whose voltage levels range from 10 kV to 750 kV. The objectives of PQMAS include compliance verification and power quality performance analysis. The system conducts limit comparisons for all PQ parameters referred to in current national standards. In addition, the PQ performance is evaluated using verification results. Based on the large volume of data collected, PQMAS will embed functions of trouble shooting and advanced data analysis in the near future.

2 System Architecture

PQMAS has three structural levels - PQ monitors,provincial monitoring stations and the national monitoring station [7,8]. Provincial stations obtain PQ data from monitors and collect account information of monitoring points and equipment from existing SGCC Power Production Management System (PMS). Monitoring and account data is then transmitted through the Unified Data Exchange Platform (UDEP) to national station, where statistics and analysis are mainly carried out.

The national station is built upon the standard big data platform of SGCC [8,9]. Big data technologies are used to deal with massive PQ data which is approximately 20TB monthly. Computation tasks are mostly carried out using Hadoop Hive and Hadoop MapReduce. The former is suitable for statistical computation and data query, and the latter is for off-line advanced analysis. Furthermore, to guarantee computation efficiency, Spark memory is utilized for high-performance computing tasks.

To make sure synchronous monitoring, PQ monitors use both within-station and network clock synchronization method [7,10-15]. Improved IEC 61850 communication protocol is adopted between PQ monitors and provincial stations.

As shown in Fig.1, the major modules in the national station are described as follows:

ü Interface server – Communication with provincial stations

ü Web server - Allocation of Web App software

ü Analysis result storage - Storage of analysis results and accounting data

ü Big data platform - Storage and analysis of collected PQ data

As in the provincial stations, the major components include:

Fig.1 Illustration of system architecture

ü Computation server – To reformat PQ data into four-tuple data format

ü Interface server - Communication with national stations and with PMS

ü Web server – To allocate Web App software

ü Query server - Short-term storage of collected data and query service for account data

ü Task processing server- Message information caching and analysis

Both critical and problematic points are considered in selecting monitoring locations. Since the distributed generations and loads with disturbing emissions raise increasing concerns related to harmonics, unbalance,and flicker, a number of monitors are installed at the interconnection points of the grid and these interference sources.

3 Unified Data Format

PQ parameters transmitted to national station include 3-min monitoring data obtained by integrating 10-cycle measurement made by monitors [8,9]. As to the data format, JSON is used for PQ parameters and COMTRADE for real-time waveform [16].

In order to reserve all the measurement and integration information embedded in the data, a self-described fourtuple data structure is proposed. The data structure provides a unified representation for measurement time and computation methods, thus improving efficiency for data transmission and re-utilization.

In detail, the data structure consists of four elements -monitoring point ID, data description, measurement time and value. Among the four elements, data description is a composite code consisting of a number of fixed-length coding blocks, including PQ index name, phase, precision,integration method, etc.

Fig.2 Description of four-tuple data structure

4 Compliance verification analysis

In PQMAS, monitored parameters include not only current and voltage RMS, but also all PQ parameters mentioned in national standards, i.e. frequency and voltage deviation, harmonic/inter-harmonic voltage/current,voltage unbalance factor, flicker, voltage sag, voltage swell and temporary interruption [8]. The thresholds for these parameters are described below.

4.1 Frequency deviation

F represents frequency in Hz and Lf indicates whether frequency is within the limit of the standard. Lf is true if F accords with the standard, and it is false if F exceeds the limits. The meaning of true or false also applies in the following sections.

4.2 Voltage deviation

As the voltage measurement made at monitoring points is with voltage levels no less than 35 kV, the voltage deviation should follow the limit below.

∆U +max represents the maximum value of positive voltage deviation measurements.∆U −min represents the minimum of negative voltage deviations.U norm represents the nominal voltage. Lv indicates whether voltage deviation is within the limit of the standard.

In terms of monitoring points with voltage levels less than 35 kV, equation (2) can be rewritten as:

Also note that, if∆U−min is higher thanUnorm, the equation (2) and (3) should be changed into the following two forms, respectively.

4.3 Flicker

In terms of monitoring points with voltage levels no less than 110 kV, compliance verification for flicker can be conducted using the following equation:

Where Fk represents the measurement of long-time flicker. Lfk indicates whether flicker is within the limit of the standard.

As for monitoring points with voltage levels less than 110 kV, equation (6) can be rewritten as:

4.4 Harmonic voltage

The thresholds, defined as Thh arm _v for harmonic voltage percentage is defined separately for odd and even orders.As to odd harmonicd orders,

As to even harmonic orders, equation (8) is rewritten as:

The thresholds for THD (Total Harmonic Distortion)for voltage are also defined with respect to voltage levels,expressed as follows.

4.5 Harmonic current

The measured harmoniccurrent should be compared with converted thresholds which is calculated using equation (11).

Thh arm _c _ norm is threshold for benchmark capacity.Th harm _c represents actual harmonic current threshold for comparison. SP represents minimum short-circuit capacity of the monitoring point, andSPb ase indicates the benchmark short-circuit capacity defined in the national standard for each voltage level.

4.6 Inter-harmonic voltage

The thresholds for inter-harmonic voltage are defined below.

where THin indicates thresholds for inter-harmonic voltage and h_ ordermeans inter-harmonic order.

4.7 Unbalance factor

As to the negative-sequence voltage unbalance factor,two thresholds are defined, namely short-time and ordinary thresholds, corresponding to maximum and 95% probability value of daily measurement statistics, respectively.

where THuf indicates thresholds for negative-sequence voltage unbalance factor.

5 PQ performance assessment

Based on compliance verification results for single measurement of each monitoring point, PQ performance can be assessed for several temporal or spatial scales.

As to frequency deviation, voltage deviation and flicker,a metric named Qualification Rate (QR) is used [8], which is defined below.

QR calculates the ratio of time span with good power quality to all time stamps in operation. The resolution of time duration can be second or minute.

As to harmonic voltage and current, inter-harmonic voltage and negative-sequence voltage unbalance factor, a metric called Limit-Exceeding Rate (LER) is proposed, as expressed below.

LER calculates the ratio of limit-exceeding points(in cases of analysis for regions) or monitoring days (in cases of analysis for time periods) to all points or days in operation. Thus, PQ performance can be assessed and compared between different regions or time periods.

6 Result Presentations

A user-friendly visualization software is designed for presentations of analysis results. Analysis reports are conducted both for statistics of PQ parameters and their limit exceeding results. In detail, the minimum,maximum, average, and 95% probability value as well as limit-exceeding times are calculated on the basis of daily,monthly and yearly for all PQ parameters. Comparisons between different regions and different interference sources are also conducted. In addition, the impact of various types of interference sources on the grid is analyzed using correlation analysis tools. The analysis task is conducted automatically using computation resources of the big data platform.

The visualization tool also provides various graphical representations to better display analysis results.Histograms, scatter plots, box-plots and radar charts are all included.

7 Conclusion

This paper presented the data analysis conducted for a state-wide PQ monitoring and analysis system in China.The system provided compliance analysis with regard to a number of power quality parameters. Based on two PQ metrics, PQ parameters can be precisely integrated, thus providing a benchmarking evaluation platform for different regions and time periods. To better present the analysis results, a user-friendly visualization tool is designed. In the near future, trouble shooting and advanced data analysis tools will be included in the system.

Acknowledgements

This work was supported by the State Grid Science and Technology Project (GEIRI-DL-71-17-002).