http://orcid.org/0000-0002-0045-7620
Figures
Abstract
This paper proposes an innovative internet of things (IoT) based communication framework for monitoring microgrid under the condition of packet dropouts in measurements. First of all, the microgrid incorporating the renewable distributed energy resources is represented by a state-space model. The IoT embedded wireless sensor network is adopted to sense the system states. Afterwards, the information is transmitted to the energy management system using the communication network. Finally, the least mean square fourth algorithm is explored for estimating the system states. The effectiveness of the developed approach is verified through numerical simulations.
Citation: Rana MM (2017) Least mean square fourth based microgrid state estimation algorithm using the internet of things technology. PLoS ONE 12(5): e0176099. https://doi.org/10.1371/journal.pone.0176099
Editor: Xiaosong Hu, Chongqing University, CHINA
Received: December 18, 2016; Accepted: April 5, 2017; Published: May 1, 2017
Copyright: © 2017 Md Masud Rana. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper.
Funding: The author received no specific funding for this work.
Competing interests: The author has declared that no competing interests exist.
1 Introduction
Due to environmental consciousness and energy crisis, the distributed energy resources (DERs) such as solar panels are integrated into the grid. As their generation pattern is intermittent in nature, it is required to monitor them closely. In order to monitor them, the internet of things (IoT) is the potential infrastructure as it provides the endless support and wide-area connectivity. The main technologies for the IoT are machine-to-machine interfaces, wireless sensor network (WSN) and embedded sensor and actuator schemes that facilitates sensing, control interaction as well as networking [1]. Obviously, an unreliable communication network exits between sensors and energy management system (EMS). The function of the EMS is to estimate the system states and maintain the grid stability. Driven by these motivations, this paper proposes a least mean square based state estimation algorithm considering the packet dropouts in the measurement under the umbrella of the IoT framework.
Extensive investigations on power system state estimations have been carried out in the literature. To begin with, the least mean squares (LMS) based power system dynamic state estimation is explored in [2, 3]. Unfortunately, it is very difficult to tune the step size parameter of this LMS scheme. The normalized LMS (NLMS) algorithm is proposed for state estimations [2, 4]. This NLMS process normalises the input signal so that the step size parameter can guarantee the stability of this algorithm. Afterwards, the variable step size based LMS algorithm for system state estimation is proposed in [5, 6]. This approach provides an optimum convergence rate but there is a trade-off between rate and performance. An artificial intelligence based dynamic state estimation is presented in [7]. This technique does not require a mathematical model. However, in stability analysis of large-scale power systems, it is often preferable to have an exact model. From this perspective, the physical model based Kalman filter (KF) algorithm is widely used for dynamic state estimations [8–10]. Unfortunately, the computational complexity is extensive compared with the LMS algorithm. A semidefinite programming based optimal energy management system is proposed in [11]. The smart home is designed considering the photovoltaic systems and energy storage with their uncertainties. Finally, a semidefinite programming problem is formulated to optimize the electric power allocation in the battery, home load demand, photovoltaic power supply and utility grid. All of the aforementioned papers assumed that the communication is perfect between sensors and EMS. In reality, there are a lossy communication channels between them. The main contributions of this paper are summarized as follows:
- The considered environmentally-friendly renewable microgrid is expressed as a state-space model. Then the IoT based wireless sensor network is adopted to transfer the sensing state information to the energy management system where the measurements are lost over unreliable communication channels. For long distance data transmission, the IoT based digital communication scheme is proposed.
- The least mean square fourth algorithm is proposed for estimating the system states over communication channels. The fourth error function is minimized with respect to the predicted states which leads to a new estimation algorithm in the context of the IoT based smart grid communication.
The rest of the article is organised as follows. In Section 2 presents the microgrid state-space model, and its estimation algorithm is in Section 3. The simulation results and conclusion are in Section 4 and 5, respectively.
Notation: Bold face lower and upper case letters are used to represent vectors and matrices, respectively; superscripts x′ denotes the transpose of x and I is the identity matrix.
2 Modelling the microgrid
One of the vital characteristics of the smart grid is to integrate the renewable resources into the electricity network. To demonstrate, the schematic diagram of an isolated studied microgrid is illustrated in Fig 1 [12–14]. It can be observed that the microgrid is connected with the grid/another microgrid through the point common coupling points (PCCs). Basically, the energy generating unit of a microgrid entails of AC/DC voltage sources, DC/AC converter and series RLC filter. The load is connected to the PCC, and the local loads are equivalent to resistive, capacitive and constant power load (CPL). After applying Kirchhoff’s voltage and Kirchhoff’s current laws, the dynamic of a microgrid in the dq-frame is expressed as follows [12–14]:
(1)
(2)
(3)
(4)
(5)
(6)
Here, i, R, L, C, ω, vd, and vt are the current, resistor, inductor, capacitor, frequency, PCC voltage and line voltage, respectively. For instance, itd and itq represent the line current in the d-q frame.
Usually, the CPL is connected to the AC bus via the rectifier circuit, RLC filter and converter. To clarify, Fig 2 shows the schematic diagram of a CPL in the considered microgrid. The dynamic equation of RLC dc-link filter is expressed as follows:
(7)
(8)
Here, i, Pin, v, L, and C represent the maximum current, converter/inverter input power, peak voltage, inductor and capacitor, respectively. For example, idc and vout are the RLC filter current and output voltage, respectively. With appropriate firing angle and switching function in the rectifier circuit, the CPL line current iCPLd for Eq (5) is given by [15]:
(9)
(10)
(11)
Here, the maximum magnitude of the stitching function Sw is equal to
[13].
Now the above nonlinear set of system equations is linearised around the operating point x⋆ i.e., where
is the operating point [12, 13]. After simple algebraic manipulations, the continuous-time state-space model of a microgrid is expressed as follows:
(12)
where x = [Δitd Δitq ΔiLd ΔiLq Δvd Δvq Δidc Δvout],
,
,
and u = δPin. The symbol n is the process uncertainty. The process noise n is the zero mean Gaussian distribution [16–18] whose covariance matrix is Q. The system is expressed as a disrate-time state-space model as follows:
(13)
where A = I + ΔtAc, B = ΔtBc, and Δt is step size parameter.
In order to monitor the microgrid farm, the service provider deploys the IoT elements such as sensors around the microgrids. The IoT infrastructure for sensing and estimating the system state is illustrated in Fig 3 [19, 20]. Basically, the voltage and current sensors are deployed to measure the microgrid states. Then it transmits to the energy management system through the internet which adds noise. The received signal at the energy management system is given by:
(14)
where C(k) is the observation matrix, w(k) is the noise which considers zero mean Gaussian distribution with R(k) covariance matrix, λ(k) is the packet loss parameter with
Here, γ(k) is the packet arrival probability [21–23].
In order to transmit the sensing information to the EMS, the sensing information is quantized to obtain the bit information [19, 20]. This is due to the fact that the digital modulation scheme is used for long distance signal transmission. So, the bit information is modulated using the binary phase shift keying (BPSK) to obtain the modulated signal m(k). The modulated signal is then transmitted through the internet, which causes the signal distortions. The received signal at the control center is written as follows.
(15)
where n(k) is the channel noise.
3 Proposed estimation algorithm
According to the steepest decent approach, the state updated rule is given by [2, 3, 24]:
(16)
Here,
is the predicted states,
is the previous estimated states, μ is the scalar step size parameter, J(k) is the objective function to be minimized and ΔJ(k) is the gradient decent. The cost function is defined by [2, 3]:
(17)
Here, en(k) is the n-th error function, which is defined as follows:
(18)
Here,
and Hn is the n-th row of the observation matrix. The minimization of the cost function with respect to the predicted states lead to the following expression:
(19)
From Eq (16) and Eq (19), the update state estimation is expressed by [2, 3]:
(20)
Here, K is the user defined matrix, which depends on the specific application [2, 3]. In order to obtain the consistent estimation performance, the following normalization version is proposed:
(21)
Here, ϵ is the small positive value to avoid division by zero [25]. It can be seen that the proposed least mean square fourth algorithm is very simple, and it does not need to know the observation noise covariance as opposed to the KF scheme. When the communication is involved then
is equivalently replaced by the dequantized output sn(k) i.e.,
. This is due to the fact that the depending on the designed communication scheme, the infrastructure is equivalently reconstructed
which is also considered by different application domain papers in [26, 27].
4 Numerical simulation results
The effectiveness of the developed approach is verified through the numerical simulation. After modelling the microgrid, the developed scheme is applied to the microgrid state estimations. To do this, the simulation parameters are described in Table 1 [12, 13], where Matlab is used as a simulator. The process and measurement noises are considered as zero mean white Gaussian noises with variances Q = 0.00001I and R = 0.00012I, respectively. The considered value of K is 0.16 × eye(8). The sampling step size parameter is 0.000001. The system is considered as a balanced load with unregulated microgrid.
The proposed algorithm is tested on the microgrid with and without communication delays. The simulation results without delays are presented in Figs 4–9. First of all, Figs 4 and 5 show the direct-quadrature (dq) axis filter currents versus time slots. It can be seen that the proposed scheme is able to estimate these states within 0.00015-0.0002 seconds (time slot × step size = 200 × 0.000001). This is due to the fact that the developed approach is able to perfectly rectify the estimation errors as time proceeds. Moreover, the simulation results in Figs 6 and 7 illustrate the dq-frame load currents versus time. It can be observed that the actual states are matched with the estimated states within a short time. Furthermore, the point of common coupling voltages in the dq-frame are illustrated in Figs 8 and 9. Even if the common coupling voltages are changed with the local loads and renewable energy generation patterns, but the proposed algorithm can effectively track the system states. Other microgrid state responses have similar kind of estimation performance. Note the small fluctuations due to the Gaussian noises and packet losses.
In practical situation, there may be the communication delays when sensing the system states. Considering 2 sample communication delays in the observation, the rectifier output current and voltage are demonstrated in Figs 10 and 11, respectively. It is clearly seen that the actual system states match the estimated states within 0.0004 seconds. As the estimation algorithm is designed without communication delays, so the proposed scheme will need more time to estimate the microgrid states as expected.
5 Conclusion and future work
This paper proposes a least mean square fourth algorithm considering the packet dropouts. First of all, the microgrid incorporating the renewable distributed energy resources is represented by a state-space model. Then the innovative IoT infrastructure is proposed for transferring the microgrid information to the energy management system for estimating the system states. The effectiveness of the developed approach is verified through numerical simulations. It can be seen that the developed approach is very effective in estimating the system states. The future work involves applying the suitable control algorithm so that the system will be stabilized.
Author Contributions
- Conceptualization: MMR.
- Data curation: MMR.
- Formal analysis: MMR.
- Funding acquisition: MMR.
- Investigation: MMR.
- Methodology: MMR.
- Project administration: MMR.
- Resources: MMR.
- Software: MMR.
- Supervision: MMR.
- Validation: MMR.
- Visualization: MMR.
- Writing – original draft: MMR.
- Writing – review & editing: MMR.
References
- 1. Kumarage H, Khalil I, Alabdulatif A, Tari Z, Yi X. Secure data analytics for cloud-integrated internet of things applications. IEEE Cloud Computing. 2016;3(2):46–56.
- 2. Ahmed A, Moinuddin M, Al-Saggaf UM. State space least mean fourth algorithm for dynamic state estimation in power systems. Arabian Journal for Science and Engineering. 2015; p. 1–17.
- 3. Malik MB, Salman M. State-space least mean square. Digital Signal Processing. 2008;18(3):334–345.
- 4. Ni J, Li F. A variable step-size matrix normalized subband adaptive filter. IEEE Transactions on Audio, Speech, and Language Processing. 2010;18(6):1290–1299.
- 5. Zhang F, Wang Y, Ai B. Variable step-size MLMS algorithm for digital predistortion in wideband OFDM systems. IEEE Transactions on Consumer Electronics. 2015;61(1):10–15.
- 6. Cho H, Kim SW. Variable step-size normalized LMS algorithm by approximating correlation matrix of estimation error. Signal Processing. 2010;90(9):2792–2799.
- 7.
Kamwa I, Baraboi B, Wamkeue R. Sensorless ANN-based speed estimation of synchronous generators: Improved performance through physically motivated pre-filters. In: Proc. of the Joint Conference on Neural Networks; 2006. p. 1710–1718.
- 8. Dini DH, Mandic DP. Widely linear modeling for frequency estimation in unbalanced three-phase power systems. IEEE Transactions on Instrumentation and Measurement. 2013;62(2):353–363.
- 9. Singh AK, Singh R, Pal BC. Stability analysis of networked control in smart grids. IEEE Transactions on Smart Grid. 2015;6(1):381–390.
- 10. Kanna S, Dini DH, Xia Y, Hui S, Mandic DP. Distributed widely linear Kalman filtering for frequency estimation in power networks. IEEE Transactions on Signal and Information Processing over Networks. 2015;1(1):45–57.
- 11. Wu X, Hu X, Moura S, Yin X, Pickert V. Stochastic control of smart home energy management with plug-in electric vehicle battery energy storage and photovoltaic array. Journal of Power Sources. 2016;333:203–212.
- 12. Karimipour D, Salmasi FR. Stability analysis of AC microgrids with constant power loads based on popov’s absolute stability criterion. IEEE Transactions on Circuits and Systems II: Express Briefs. 2015;62(7):696–700.
- 13. Chaijarurnudomrung K, Areerak K, Areerak K. Modeling and stability analysis of AC-DC power system with controlled rectifier and constant power loads. WSEAS Trans Power Syst. 2011;6(2):31–41.
- 14.
Areerak KN, Bozhko S, Asher G, Thomas D. Stability analysis and modelling of AC-DC system with mixed load using dq-transformation method. In: Proc. of the International Symposium on Industrial Electronics; 2008. p. 19–24.
- 15.
Areerak KN, Bozhko S, De Lillo L, Asher G, Thomas D, Watson A, et al. The stability analysis of AC-DC systems including actuator dynamics for aircraft power systems. In: Proc. of the European Conference on Power Electronics and Applications; 2009.
- 16. Ghahremani E, Kamwa I. Online state estimation of a synchronous generator using unscented Kalman filter from phasor measurements units. IEEE Transactions on Energy Conversion. 2011;26(4):1099–1108.
- 17. Wang X, Yaz EE. Smart power grid synchronization with fault tolerant nonlinear estimation. IEEE Transactions on Power Systems, to appear in 2016;.
- 18. Ariff M, Pal B, Singh A. Estimating dynamic model parameters for adaptive protection and control in power system. IEEE Transactions on Power Systems. 2015;30(2):829–839.
- 19. Elkhorchani H, Grayaa K. Novel home energy management system using wireless communication technologies for carbon emission reduction within a smart grid. Journal of Cleaner Production. 2016;135:950–962.
- 20. Rana M, Li L. Microgrid state estimation and control for smart grid and Internet of Things communication network. Electronics Letters. 2015;51(2):149–151.
- 21.
Su L, Chesi G. On the robust stability of uncertain discrete-time networked control systems over fading channels. In: Proc. of the American Control Conference; 2015. p. 6010–6015.
- 22.
Hu S, Yan WY. Stability of networked control systems subject to input and output packet loss. In: Proc. of the Joint 48th IEEE Conference on Decision and Control and 28th Chinese Control Conference; 2009.
- 23. Xiao N, Xie L, Qiu L. Feedback stabilization of discrete-time networked systems over fading channels. IEEE Transactions on Automatic Control. 2012;57(9):2176–2189.
- 24.
Haykin S, Widrow B. Least-mean-square adaptive filters. vol. 31. John Wiley and Sons; 2003.
- 25.
Farhang-Boroujeny B. Adaptive filters: theory and applications. John Wiley & Sons; 2013.
- 26.
Gong S, Li H, Lai L, Qiu RC. Decoding the nature encoded messages for distributed energy Generation Control in Microgrid. In: Proc. of the International Conference on Communications; 2011. p. 1–5.
- 27. Rawat DB, Bajracharya C. Detection of false data injection attacks in smart grid communication systems. IEEE Signal Processing Letters. 2015;22(10):1652–1656.