Automatic Control of Low Voltage Load in Power Systems Based on Deep Learning

There are many voltage‐dependent loads (VDLs) in power systems. When the unit reserve supply is low, the transmission network adjusts the voltage of grid‐connected nodes to provide additional resources. Many reactive power devices with different regulatory performance levels and timescales must be coordinated to employ flexible VDLs to the maximum extent, and consequently, a transient stability‐constrained optimal power flow (TSCOPF) problem with mixed integers and multiperiod coupling must be solved. A multitimescale scheduling ‘day‐ahead‐intraday‐real‐time’ framework is designed for the optimal economic scheduling of VDLs. In the day‐ahead stage, the optimal scheduling period of the discrete reactive power control devices is determined. In the intraday stage, a rolling optimization model based on model predictive control is established, and the specific action times of the discrete reactive power devices are determined. In the real‐time stage, the optimal regulation strategy of the continuous reactive power control device is optimized by rolling. Then, to consider the speed and accuracy of the solution at the same time, a continuous sensitivity correction method is adopted to simplify and solve the original model. Finally, with a modified IEEE‐39 system and a modified IEEE‐118 system as examples, it is verified that the multitimescale scheduling framework maximizes the exploitation of the flexible resources of VDLs to ensure voltage security and stability and improve the economy of system operation under different scenarios.

Harmonic analysis studies have become an important component of power system analysis and design on recent years. They are used to quantify the distortion in voltage and current waveform and avoid them to propagate through the AC network. Harmonics are created mainly by nonlinear loads in power systems. Thus, in this paper are analyzed the harmonics created by industrial loads and other electronic devices and their effects.

Medium-term forecasting is an important category of electric load forecasting that covers a time span of up to one year ahead. It suits outage and maintenance planning, as well as load switching operation. We propose a new methodol-ogy that uses hourly daily loads to predict the next year hourly loads, and hence predict the peak loads expected to be reached in the next coming year. The technique is based on implementing multivariable regression on previous year's hourly loads. Three regression models are investigated in this research: the linear, the polynomial, and the exponential power. The proposed models are applied to real loads of the Jordanian power system. Results obtained using the pro-posed methods showed that their performance is close and they outperform results obtained using the widely used ex-ponential regression technique. Moreover, peak load prediction has about 90% accuracy using the proposed method-ology. The methods are generic and simple and can be implemented to hourly loads of any power system. No extra in-formation other than the hourly loads is required.

The global energy issues have become increasingly prominent in recent years, photovoltaic power generation as a renewable energy use pattern is widely used, but a large number of photovoltaic power generation to the grid is a big negative impact, it is necessary to predict the output power of the photovoltaic. Power system load forecasting is the reference and safeguard of the power system operation. This article analyzes the main point of the prediction of photovoltaic power system and power load system, then introduces the support vector machine (SVM) based on quantum particle swarm optimization (QPSO) to do the prediction. And then this paper proposes a generalized system load prediction system containing the photovoltaic power system.

The behavior of the load model in power system stability studies has been shown by various investigators to be a very important factor in determining the stability of power system elements. Load models which are presently available in typical stability programs make it difficult if not impossible to model the actual dynamics of power system load. Present load models allow easy static representations, but require detailed individual dynamic models such as induction or synchronous motor models. Thus a simple load model that can represent power system composite load dynamics will be a valuable tool in the stability study Presented herein is a second order state space load model which can represent power system composite load dynamics. In addition, the parameters of this state space load model can be identified from actual system load measurements by a weighted least squares parameter identification process. Several authors recognized the significance of load behavior to changes in voltage as long as 25 years ago. However, most of the attention in stability studies has been directed toward improved generator and control system models. Until recently, little modeling has been considered necessary for power system loads since the effects of load models in stability studies were considered only secondary. At present the load representation in typical stability programs is limited to a static load model where the load is represented as constant MVA, constant current, constant impedance, or some combination of the three.

Comprehensive reliability assessment of substation originated outages using the Monte Carlo simulation technique

In the context of the “dual carbon” goal, a high proportion of renewable energy connected to the grid to replace traditional high-carbon units has become a general trend. Taking the high proportion of renewable energy power system as the research object, the medium and long term planning of the installed capacity of source, grid, load and storage of power system is carried out considering the two dimensions of investment and operation maintenance, and the optimization model of source, grid, load and storage of power system is established. A low-carbon dispatching model considering carbon emission and demand response of distribution network is also proposed. Firstly, considering the characteristics of distribution network receiving power from the main network, the carbon emission model of distribution network based on carbon emission flow theory is constructed, and the low-carbon index of distribution network is proposed. Secondly, the improved dynamic carbon emission factor of distribution network is proposed, which is used as a guide signal to provide reference for low-carbon electricity consumption of distribution network users. According to the characteristics of users in the distribution network, the industrial demand response model aiming at low dispatch cost and low carbon and the residential demand response model aiming at electricity consumption satisfaction and low carbon satisfaction are constructed respectively.

With the rapid growth of new energy and flexible load, the random fluctuation of power system operation is enhanced, and the traditional voltage control scheme is limited. In view of the shortcomings of traditional reactive voltage control in matching the dual characteristics of the source and load in new power system, a multi-time scale coordinative strategy of reactive power and voltage control for frequent power flow fluctuation in new power system is proposed, and the reactive power and voltage control is carried out according to the normal and the unexpected large fluctuation of the power flow. The reactive power and voltage coordinated control mode of hour-minute level is used in the normal fluctuation of power flow. The timely adjustment of voltage deviation and the rapid suppression of voltage fluctuation are realized by MPC rolling optimization and dynamic-static reactive power resource substitution, and sufficient dynamic reactive power reserve is reserved for the system. The reactive power and voltage coordinated control mode of second level is used in the case of non-expected large fluctuation of power flow, and the fast decision-making ability of deep reinforcement learning realizes the rapid recovery of over-limit voltage. An example of a regional power grid in eastern China verifies the effectiveness of the proposed method.

The load in power system has the two characters, one is random time-variety and the other one is geographical dispersion. Both have negative effects on the accuracy of load model. In this paper, an improved fuzzy C-means (FCM) algorithm based on density function is proposed to minimize this influence. Firstly, the disadvantage of traditional FCM algorithm is analyzed to show its subjectivity, which is hard to obtain the available global optimal solution. Then, a FCM algorithm based on density function theory is applied to initialize the clustering centers. It can determine the number of load category and optimize the clustering result of power system daily load profile. The simulation result of a real power system load model verifies the accuracy and practicability of the proposed method. It can obtain the objective load category number and effectively solve the non-convergence problem in traditional FCM algorithm.

With the increasing use of constant power loads in power system, their influences on the stability of the system cannot be ignored. This paper introduces a power buffer configuration based on the boost converter topology. The power buffer functions to decouple the power system and the load under fault condition to mitigate the negative influence of the constant power on the power system. As such power buffer increases the stability of the network, and guarantees the proper operation of the load. Appropriate control schemes for the power buffer are proposed. Furthermore, simulation results of three-phase power buffer using Simulation/Matlab software are given to verify the validity of the power buffer concept

Chapter 1 - Introductory chapter: effects of power quality problems on energy efficiency of power systems

Harmonic voltages and currents are created by nonlinear loads in power systems. These harmonics can cause serious problems to the energy suppliers (utility companies) and energy users (consumers). This dissertation presents new techniques for harmonic measurement and the accompanying problem, harmonic reduction. To reduce harmonics, the location and magnitude of harmonic sources must be known a priori. However, to measure and monitor harmonic sources requires specialized instrumentation that is not generally permanently installed at most buses. In this study, a method is developed to use neural networks in conjunction with state estimation for monitoring and estimating harmonic sources with only a few permanent instruments. The method can also identify and monitor a "suspected" harmonic source that has not previously been measured. Once harmonic sources are known, the objective from an energy supplier's point of view is to minimize the combined total harmonic distortion of all buses on a distribution system. A procedure is developed for designing optimum filters on distribution feeders that have distributed and balanced or unbalanced harmonic sources and variable compensating capacitors. The design includes finding optimum filter admittances and the corresponding locations. Symmetrical components are used for the unbalanced harmonic source case. Filter realization with the lowest cost including sensitivity analysis is also discussed. The optimum filters chosen and placed in this manner may produce results superior to the conventional filter design procedures. The problem of harmonic reduction is alternatively approached as an energy user's problem. The user's objective is to eliminate his voltage harmonic distortion problem locally without considering the effects of voltage distortion at neighboring buses. The remedy is insertion of a reactor in series with the local compensating capacitor. A method is presented for finding the optimum fixed LC combination to minimize voltage harmonic distortion at a load bus while holding the displacement factor (fundamental power factor) at a desired value and constraining the total cost of the compensating equipment. Source harmonics and impedances are represented as randomly time-varying quantities.

UPFC (Unified power flow controller) is one of the most powerful and complex FACTS (Flexible AC transmission system) devices in a power system today. It is primarily used for simultaneous or independent control of real and reactive power in transmission lines for flexible, reliable and economic operation and loading of power systems. In this paper, the effectiveness of UPFC has been explored in transmission system network when subjected to three phase to ground fault condition. The comparison has been made between VSI (Voltage Source Inverter) based UPFC and ZSI (Z-Source Inverter) based UPFC. From the results so obtained, it has been established that the performance of ZSI based UPFC is superior as compared to that of VSI based UPFC.

The results of a dynamic optimization procedure in the design of load and frequency control (LFC) of power systems are presented. This design is based on the optimal linear regulator theory, accommodated to satisfy the performance objectives of the LFC in large multiarea interconnections. A proportional-plus-integral control law is considered. The paper also analyzes the influence of system and design parameters on power system performance considering two-and three-area interconnections. Finally, the problems of implementation are discussed.

In this paper, a novel meta-heuristic Colliding bodies optimization-based proportional-integral-derivative controller with derivative filter is presented to solve a single-area power system's load frequency control issues. The controller parameters are determined by employing the integral of time multiplied absolute error as the fitness function. A single-area non-reheat thermal power system is considered for establishing the efficacy of the proposed method. The robustness of the proposed controller is also ascertained by inserting perturbation in system parameters. It is observed from the simulation results that the proposed method provides improved dynamic performance over the existing methods.