PMU-based linear state estimation of Lausanne subtransmission network: Experimental validation

https://doi.org/10.1016/j.epsr.2020.106649Get rights and content

Highlights

  • Real-time linear power system state estimation based on synchrophasors.

  • PMU-based weighted least squares (WLS) and least absolute value (LAV) estimators.

  • State estimation accuracy and time latency assessment.

  • Bad data and network parameter error detection.

  • Experimental validation and implementation details in a real subtransmission network.

Abstract

The paper presents the implementation details and the experimental validation of a linear state estimator based on synchrophasor measurements in a real subtransmission network. The data are provided by 15 phasor measurement units (PMUs) installed in the 125-kV grid of the city of Lausanne, Switzerland. The PMU-based monitoring infrastructure, the telecommunication and the phasor data concentrator are described in detail. We compare the performance of two different state estimation methods, i.e., the linear weighted least squares and the least absolute value estimator, with a focus on the influence of the line parameter errors on the state estimates. We further analyze the latency contribution associated to each step of the measurement chain, in order to validate its appropriateness to serve time-critical power system applications.

Introduction

In the last decades, state estimation (SE) has become a core power system situation awareness functionality for network operators. Typically, power system state estimators consist in a non-linear weighted least squares (WLS) process, that computes the most likelihood system-state through the statistical processing of real and pseudo-measurements. The accuracy and reliability of the state estimates are exploited by various functionalities, like voltage control, contingency and stability analysis, security assessment, and protections tuning.

One of the main features of SE is the capability to detect, identify, and eliminate large measurement errors, also called bad-data. Most state estimators (e.g., the WLS) need to be coupled with a bad-data processing method, such as the well-known largest normalized residual (LNR) method [1]. By contrast, the so-called robust estimators, e.g., the least absolute value (LAV), possess an intrinsic bad-data rejection property, thus there is no need to employ a separate bad-data processor [2]. Also, the vulnerability to leverage measurements can be eliminated by strategic scaling without affecting the state estimates, as demonstrated in Göl and Abur [2]. In terms of computational time, LAV becomes competitive with WLS, in particular in presence of bad data.

The performance of the aforementioned estimators can be improved if the measurements adopted to estimate the system state are entirely composed of synchrophasors provided by Phasor Measurement Units (PMUs). Indeed, a state estimator that uses only synchrophasor measurements has the following benefits compared to conventional estimators:

  • 1.

    The measurements are a linear function of the state variables, which leads to a linear state estimator (LSE) consisting in a non-iterative algorithm with low computational time;

  • 2.

    The measurements are time-stamped, so that they can be time-aligned at the data collection point. This ensures that every set of measurements given to the LSE is composed of measurements taken at the same instant;

  • 3.

    As the PMUs directly measure the phase-angle, there is no need to choose a reference bus, at which the voltage phase-angle is fixed to an arbitrarily selected value [3];

  • 4.

    The very high PMU streaming-rate (tens of frames-per-second (fps)) leads to a state estimator characterized by high refresh-rate and low latency, which is called real-time state estimation (RTSE);

  • 5.

    PMUs measure by default the current and voltage phasors in each phase separately, so that three-phase SE can be used irrespectively of the network-parameter symmetry and power-flow balance.

In the current literature, few papers have described the implementation details and experimental results of LSEs in real power systems using PMUs, e.g., [4], [5], [6]. These previous contributions are limited to the description of the SE method and implementation aspects, lacking a thorough analysis of the SE results. In this paper, we present the SE results obtained using the LAV and the LWLS estimators for the 125-kV subtransmission network of Lausanne city (Switzerland) that is operated by the utility Service Industriels de Lausanne (SiL). The originality of this work lies in a detailed performance assessment of the SE results provided by LSEs that use real measurements taken in a real power network extensively equipped with PMUs. First, we present and discuss the results of the LWLS and LAV estimators. Second, we perform a latency assessment of every element of the chain comprising the PMU, the telecom network, the Phasor Data Concentrator (PDC) and the SE process.

The paper is structured as follows. Section 2 describes the topology of the network and of the PMU placement. Section 3 provides implementation details of the sensing infrastructure and of the synchrophasor network. Section 4 formulates the studied SE methods. Section 5 assesses the accuracy of the employed SE methods and the time-latency of the overall SE process. Section 6 concludes the paper.

Section snippets

Network structure and PMU placement

For the study presented in this paper, we use the measurements provided by PMUs installed in the 125-kV network that supplies the city of Lausanne. Fig. 1 shows the network structure and the PMU placement. The topology is composed of 7 buses and 10 lines. Buses #1 and #7 are connected to the higher voltage grid through step-up transformers. No zero-injection buses are present in this network. The statistics of the power absorbed by the loads are reported in Table 1.

We installed 15 PMUs and each

System architecture

This section describes the measurement infrastructure that comprises transducers, PMUs, telecommunication network and Phasor Data Concentrator (PDC).

State estimation methods and implementation

The SE process is composed of a series of functions depicted in the flow diagram of Fig. 3.

The raw measurements undergo plausibility checks in order to identify gross errors. The topology processor builds the network topology based on the status of the breakers.1 Then, an observability analysis is performed in order to check whether the entire grid is observable given the topology

State estimation results

In this section, we present and discuss the estimation values obtained with the following state estimators:

  • LWLS estimator without bad-data processing;

  • LWLS estimator that uses the LNR test to identify and remove bad data, called LWLS - LNR (the threshold of the LNR-test is set to 4);

  • LAV estimator.

In addition, we present a latency assessment of each element of the process, from the PMUs to the SE output.

Conclusion

The paper described the practical implementation and results of state estimation in the 125-kV subtransmission network of the city of Lausanne, Switzerland. We assessed the SE accuracy using two methods, the LWLS and the LAV, and we measured the time-latency of every element of the process.

We observed that the effect of line parameter errors and measurement errors can be similar, as they can both generate large measurement residuals. This phenomenon causes the LNR-test to erroneously eliminate

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (17)

  • A. Abur et al.

    Power System State Estimation: Theory and Implementation

    (2004)
  • M. Göl et al.

    LAV based robust state estimation for systems measured by PMUs

    IEEE Trans. Smart Grid

    (2014)
  • A. Abur

    Impact of phasor measurements on state estimation

    Electrical and Electronics Engineering, 2009. ELECO 2009. International Conference on

    (2009)
  • K.D. Jones et al.

    Three-phase linear state estimation using phasor measurements

    IEEE Power and Energy Society General Meeting (PES)

    (2013)
  • L. Zhang et al.

    Design, testing, and implementation of a linear state estimator in a real power system

    IEEE Trans. Smart Grid

    (2017)
  • M. Pignati

    Real-time state estimation of the EPFL-campus medium-voltage grid by using PMUs

    2015 IEEE Power Energy Society Innovative Smart Grid Technologies Conference (ISGT)

    (2015)
  • Instrument transformers part 3: Additional requirements for inductive voltage transformers, IEC Std 61869-3...
  • Instrument transformers part 2: Additional requirements for current transformers, IEC Std 61869-2...
There are more references available in the full text version of this article.

Cited by (22)

  • A comprehensive literature review of conventional and modern islanding detection methods

    2022, Energy Strategy Reviews
    Citation Excerpt :

    PMU-based detection can also ensure the control and coordination of a power system across a wide range, particularly for electrical power transmission. Because PMU can gather information for big sections and solve large-scale systems, when multiple control areas are involved in power transmission, PMU can detect a wide range without disrupting the network [131]. Big data analysis technology is an important field that holds a significant role in the electrical power industry to boost response time actively to achieve reliable power & energy systems development [132].

  • OPF-based under frequency load shedding predicting the dynamic frequency trajectory

    2020, Electric Power Systems Research
    Citation Excerpt :

    It is reasonable to assume that the network model, combining network topology and electrical parameters of the network components, is assessed in real-time in the TSO control room. The topology can be indeed constructed by collecting the breaker/switch statuses that are streamed by PMUs [21]. This includes any loss of generation that might call for UFLS action.

View all citing articles on Scopus

This work was supported by the Swiss Federal Office of Energy (OFEN) under Grant SI/501080-01. The Authors would like to thank Services Industriels de Lausanne (SiL) for the collaboration.

View full text