Showing papers on "Dynamic demand published in 2018"

Multi-Agent Deep Reinforcement Learning for Dynamic Power Allocation in Wireless Networks

Abstract: This work demonstrates the potential of deep reinforcement learning techniques for transmit power control in wireless networks. Existing techniques typically find near-optimal power allocations by solving a challenging optimization problem. Most of these algorithms are not scalable to large networks in real-world scenarios because of their computational complexity and instantaneous cross-cell channel state information (CSI) requirement. In this paper, a distributively executed dynamic power allocation scheme is developed based on model-free deep reinforcement learning. Each transmitter collects CSI and quality of service (QoS) information from several neighbors and adapts its own transmit power accordingly. The objective is to maximize a weighted sum-rate utility function, which can be particularized to achieve maximum sum-rate or proportionally fair scheduling. Both random variations and delays in the CSI are inherently addressed using deep Q-learning. For a typical network architecture, the proposed algorithm is shown to achieve near-optimal power allocation in real time based on delayed CSI measurements available to the agents. The proposed scheme is especially suitable for practical scenarios where the system model is inaccurate and CSI delay is non-negligible.

Intelligent Residential Energy Management System for Dynamic Demand Response in Smart Buildings

Abstract: The advancements in renewable energy technologies direct the power sector to focus on power generation from renewable energy resources (RER) as an alternative solution for meeting the future demand. Nowadays residential buildings are becoming smarter with wide use of smart appliances, integration of information and communication technology, and in-house power generation using RER. In this paper, an intelligent residential energy management system (IREMS) for prosumers of smart residential buildings is proposed, and its benefits are demonstrated through a case study. The primary objective of IREMS is reduction in electricity bills while maintaining the power demand under the maximum demand limit subjected to the various constraints governing the operation of household loads and RER. The IREMS achieves the objective by scheduling the schedulable loads during low pricing intervals while considering the operational dynamics of nonschedulable loads and availability of RER. IREMS also manages the battery energy storage in such a way so as to reduce the power dissipated through the dump load when excess power is available from RER due to the utility-defined power export limit to grid. Further, an optimal resources sizing algorithm is used to choose the size of RER and battery storage for the effective utilization of available renewable energy.

Dynamic Load Altering Attacks Against Power System Stability: Attack Models and Protection Schemes

Abstract: Dynamic load altering attacks (D-LAAs) are introduced as a new class of cyber-physical attacks against smart grid demand response programs The fundamental characteristics of D-LAAs are explained Accordingly, D-LAAs are classified in terms of open-loop versus closed-loop attacks, single-point versus multi-point attacks, the type of feedback, and the type of attack controller A specific closed-loop D-LAA against power system stability is formulated and analyzed, where the attacker controls the changes in the victim load based on a feedback from the power system frequency A protection system is designed against D-LAAs by formulating and solving a non-convex pole-placement optimization problem Uncertainty with respect to attack sensor location is addressed Case studies are presented to assess system vulnerabilities, impacts of single-point and multi-point attacks, and optimal load protection in an IEEE 39 bus test system

Intelligent Demand Response Contribution in Frequency Control of Multi-Area Power Systems

Abstract: Frequency control is one of the most important issues in a power system due to increasing size, changing structure and the complexity of interconnected power systems. Increasing economic constraints for power system quality and reliability and high operational costs of generation side controllers have inclined researchers to consider demand response as an alternative for preserving system frequency during off-normal conditions. However, the main obstacle is calculating the accurate amount of load related to the value of disturbances to be manipulated, specifically in a multi-area power system. Dealing with this challenge, this paper makes an attempt to find a solution via monitoring the deviations of tie-line flows. The proposed solution calculates the magnitude of disturbances and simultaneously determines the area where disturbances occurred, to apply demand response exactly to the involved area. To address communication limitations, the impact of demand response delay on the frequency stability is investigated. Furthermore, this paper introduces a fuzzy-PI-based supervisory controller as a coordinator between the demand response and secondary frequency control avoiding large frequency overshoots/undershoots caused by the communication delays. To evaluate the proposed control scheme, simulation studies are carried out on the 10-machine New England test power system.

Energy-Aware Dynamic Resource Allocation in UAV Assisted Mobile Edge Computing Over Social Internet of Vehicles

Abstract: Social Internet of Vehicles (SIoV) is a new paradigm that enables social relationships among vehicles by integrating vehicle-to-everything communications and social networking properties into the vehicular environment. Through the provision of diverse socially-inspired applications and services, the emergence of SIoV helps to improve the road experience, traffic efficiency, road safety, travel comfort, and entertainment along the roads. However, the computation performance for those applications have been seriously affected by resource-limited on-board units as well as deployment costs and workloads of roadside units. Under such context, an unmanned aerial vehicle (UAV)-assisted mobile edge computing environment over SIoV with a three-layer integrated architecture is adopted in this paper. Within this architecture, we explore the energy-aware dynamic resource allocation problem by taking into account partial computation offloading, social content caching, and radio resource scheduling. Particularly, we develop an optimization framework for total utility maximization by jointly optimizing the transmit power of vehicle and the UAV trajectory. To resolve this problem, an energy-aware dynamic power optimization problem is formulated under the constraint of the evolution law of energy consumption state for each vehicle. By considering two cases, i.e., cooperation and noncooperation among vehicles, we obtain the optimal dynamic power allocation of the vehicle with a fixed UAV trajectory via dynamic programming method. In addition, under the condition of fixed power, a search algorithm is introduced to derive the optimized UAV trajectory based on acceptable ground-UAV distance metric and the optimal offloaded data size of the vehicle. Simulation results are presented to demonstrate the effectiveness of the proposed framework over alternative benchmark schemes.

Energy Efficient Dynamic Resource Optimization in NOMA System

Abstract: Non-orthogonal multiple access (NOMA) with successive interference cancellation (SIC) is a promising technique for next generation wireless communications. Using NOMA, more than one user can access the same frequency-time resource simultaneously and multi-user signals can be separated successfully using SIC. In this paper, resource allocation algorithms for subchannel assignment and power allocation for a downlink NOMA network are investigated. Different from the existing works, here, energy efficient dynamic power allocation in NOMA networks is investigated. This problem is explored using the Lyapunov optimization method by considering the constraints on minimum user quality of service and the maximum transmit power limit. Based on the framework of Lyapunov optimization, the problem of energy efficient optimization can be broken down into three subproblems. Two of which are linear and the rest can be solved by introducing Lagrangian function. The mathematical analysis and simulation results confirm that the proposed scheme can achieve a significant utility performance gain and the energy efficiency and delay tradeoff is derived as $[{\mathrm{ O}}(1/V),{\mathrm{ O}}(V)]$ with $V$ as a control parameter under maintaining the queue stability.

Demand Response of Ancillary Service From Industrial Loads Coordinated With Energy Storage

Abstract: As one of the featured initiatives in smart grids, demand response is enabling active participation of electricity consumers in the supply/demand balancing process, thereby enhancing the power system's operational flexibility in a cost-effective way. Industrial load plays an important role in demand response because of its intense power consumption, already existing advanced monitoring, and control infrastructure, and its strong economic incentive due to the high energy costs. As typical industrial loads, cement plants are able to quickly adjust their power consumption rate by switching on/off the crushers. However, in the cement plant as well as other industrial loads, switching on/off the loading units only achieves discrete power changes, which restricts the load from offering valuable ancillary services such as regulation and load following, as continuous power changes are required for these services. In this paper, we overcome this restriction of poor granularity by proposing methods that enable these loads to provide regulation or load following with the support of an onsite energy storage system.

Virtual inertia control of PV systems for dynamic performance and damping enhancement of DC microgrids with constant power loads

Abstract: DC bus voltage in islanded DC microgrids (MGs) is prone to power fluctuations of sources and loads. This is due to the lack of generational inertia, which is caused by high penetration of converter-based renewable energy sources (RESs). With the growing penetration of RESs, especially photovoltaic arrays (PVAs), they are required to provide grid support services such as inertial response in a similar way to conventional synchronous generators. Here, a virtual inertia control (VIC) is proposed for PVAs to enhance the inertia of a hybrid PVA-battery DC MG. The proposed VIC employs active power control of PVAs to provide virtual inertial response (VIR) without using any high-power energy storage system such as supercapacitors. An adaptive virtual inertia gain is introduced to achieve dynamic power sharing between the PVAs that provide VIR. Impedance-base stability analysis is utilised to study the impact of the virtual inertia gain on the system stability and to show the impact of the proposed VIC on improving the stability margin of the DC MG in the presence of destabilising constant power loads (CPLs). Finally, simulation results are presented to verify the effectiveness of the proposed method in dynamic performance and damping enhancement of the DC MG and reducing the high-current stress on the battery.

Dynamic Power Management System Employing a Single-Stage Power Converter for Standalone Solar PV Applications

Abstract: This paper presents a dynamic power flow management system for a solar photovoltaic (PV) system employing a single-stage single-inductor-based dual-input/output dc–dc converter feeding standalone dc loads backed up by a rechargeable battery A time-sharing voltage-mode control scheme has been proposed for power flow management between solar PV, a battery, and a standalone dc load, and it also maintains a constant dc load voltage and performs maximum power point tracking operation of solar PV The implementation of the control scheme has been described in detail The steady-state performance of the single-stage converter has been explained with the relevant analytical expressions derived along with the characteristics A state-space average model was developed for simulating the transient behavior and validating the working of the system for step changes in the input solar PV power and the dc loads A hardware prototype of the proposed system has been fabricated, and the proposed controller has been implemented using the dSPACE DS1103 real-time interface board The working of the proposed scheme for different levels of input solar insolation and dc load power demand has been satisfactorily demonstrated, and the corresponding results are also provided

Communication Modeling of Solar Home System and Smart Meter in Smart Grids

Abstract: The future energy networks are envisioned to be green and clean with high penetration of renewable energy-based generators. The most promising type is solar energy which has immense potential around the globe. Solar home systems (SHSs) with rooftop solar panels are proliferating in urban cities as well as in distant rural areas. Possible interaction of SHS with utility grid will result in dynamic power flow which is a huge challenge for power utility authorities and consumers. The smart meters (SMs) are being deployed to make this possible through bi-directional energy and information exchange. In order to address this need, this paper develops the communication models of SHS and SM based on the IEC 61850 standard. These models provide standardized approach to these technologies and facilitate a series of functionalities, such as power flow control, demand response, and other ancillary services, using configured message exchange. The detailed models, their use cases and the messages are studied in detail. Furthermore, extensive simulations are run with riverbed modeler to investigate the dynamic performance. IEC 61850-based models of SHS and SM are implemented, message frames are developed according to use cases, and the functionalities mentioned earlier are run as scenarios. Finally, the performances of different communication technologies have been analyzed to estimate their adequacy for smart grid implementations.

A Two-Stage Approach for Network Constrained Unit Commitment Problem With Demand Response

Abstract: This paper proposes a two-stage formulation for the day-ahead energy scheduling problem with demand response (DR). The first stage solves a network-constrained unit commitment problem with DR, to determine the hourly net demand changes (i.e., difference between final and initial demand values) happening at each DR bus 1 along with the unit commitment schedule and ac load flow solution. Here, the objective is to maximize the social welfare which is expressed as the total utility of the demand side minus the total generation cost. The second stage solves an incentive or penalty minimization problem to determine the demand shifting and demand curtailment across the 24-h period at each DR bus, offering DR, based on the hourly net demand changes obtained during the first stage. The proposed formulation shows how demand shift and demand curtailment happening at different DR buses can be traced back to the hourly net demand changes occurring at the system level. The results, presented for a six-bus system and IEEE 118 bus system, show the benefits of including DR into the network-constrained unit commitment problem according to the proposed formulation. 1 Any bus which is capable of offering DR will be referred to as DR bus. It should not be mistaken for PQ bus referred in the load flow analysis.

Dynamic Demand Response Using Customer Coupons Considering Multiple Load Aggregators to Simultaneously Achieve Efficiency and Fairness

Abstract: This paper discusses the feasibility of using customer coupon demand response in meshed secondary networks. Customers are rewarded by coupons to achieve the objective of optimal operation cost during peak periods. The interdependence of the locational marginal price and the demand is modeled by an artificial neural network. The effect of multiple load aggregators participating in customer coupon demand response is also investigated. Because load aggregators satisfy different proportions of the objective, a fairness function is defined that guarantees that aggregators are rewarded in correspondence with their participation toward the objective. Energy loss is also considered in the objective as it is an essential part of the distribution system. A dynamic coupon mechanism is designed to cope with the changing nature of the demand. To validate the effectiveness of the method, simulations of the proposed method have been performed on a real heavily-meshed distribution network in this paper. The results show that customer coupon demand response significantly contributes to shaving the peak, therefore, bringing considerable economic savings and reduction of loss.

Deep Reinforcement Learning for Distributed Dynamic Power Allocation in Wireless Networks.

Abstract: This work demonstrates the potential of deep reinforcement learning techniques for transmit power control in emerging and future wireless networks. Various techniques have been proposed in the literature to find near-optimal power allocations, often by solving a challenging optimization problem. Most of these algorithms are not scalable to large networks in real-world scenarios because of their computational complexity and instantaneous cross-cell channel state information (CSI) requirement. In this paper, a model-free distributed dynamic power allocation scheme is developed based on deep reinforcement learning. Each transmitter collects CSI and quality of service (QoS) information from several neighbors and adapts its own transmit power accordingly. The objective is to maximize a weighted sum-rate utility function, which can be particularized to achieve maximum sum-rate or proportionally fair scheduling (with weights that are changing over time). Both random variations and delays in the CSI are inherently addressed using deep Q-learning. For a typical network architecture, the proposed algorithm is shown to achieve near-optimal power allocation in real time based on delayed CSI measurements available to the agents. This work indicates that deep reinforcement learning based radio resource management can be very fast and deliver highly competitive performance, especially in practical scenarios where the system model is inaccurate and CSI delay is non-negligible.

A Novel Dynamic Power Routing Scheme to Maximize Loadability of Islanded Hybrid AC/DC Microgrids Under Unbalanced AC Loading

Abstract: This paper proposes a novel dynamic power routing (DPR) scheme for hybrid ac/dc microgrids operating in islanded mode, where unlike in grid-connected microgrids, local generation adequacy is crucial for proper system operation. The unbalanced nature of ac distribution networks limits the microgrid loadability in the sense that loads must be shed from heavily loaded phases, even if the connected distributed generators (DGs) have not reached their total three-phase capacity limits. The main challenge is to exploit the available resources by routing the power between the ac subgrid phases, thereby minimizing load shedding. The proposed method utilizes the interlinking converters between the ac and dc sides of hybrid ac/dc microgrids to provide this functionality. A supervisory controller implements a DPR-based optimal power flow (OPF) algorithm to allow full loadability of the islanded network. The formulated OPF problem is solved analytically using an interior point method that has proved to be computationally cost-effective. Many case studies are conducted to address the unbalance problem and to validate the effectiveness of the proposed strategy against conventional methods, which are based solely on optimal DG droop settings.

Improved Optimal Decentralized Load Modulation for Power System Primary Frequency Regulation

Abstract: Nowadays the interest in smart load technologies for primary frequency regulation is spurred due to the increasing penetration of renewable energy resources. In this paper, an improved optimal load control (improved OLC) is introduced by applying a multiobjective optimization-based gain-tuning method to the conventional OLC approach. The objective is to minimize the frequency nadir, time response, steady-state error, total load shed, and aggregated disutility of controllable loads subject to power balance over the network. Simulation results indicate that enabling a multiobjective optimization-based gain-tuning procedure in the OLC approach can provide better power system frequency regulation. Time-domain analysis confirms the superior performance of improved OLC in terms of frequency nadir (Hz), steady-state error (Hz), control effort, and NERC-based performance metrics (MW/0.1 Hz), with detailed load and wind farm models. Furthermore, small-signal analysis demonstrates that the improved OLC enhances the system closed-loop performance and stability margins by increasing the damping ratio of the system's critical modes.

SmoothOperator: Reducing Power Fragmentation and Improving Power Utilization in Large-scale Datacenters

Abstract: With the ever growing popularity of cloud computing and web services, Internet companies are in need of increased computing capacity to serve the demand. However, power has become a major limiting factor prohibiting the growth in industry: it is often the case that no more servers can be added to datacenters without surpassing the capacity of the existing power infrastructure. In this work, we first investigate the power utilization in Facebook datacenters. We observe that the combination of provisioning for peak power usage, highly fluctuating traffic, and multi-level power delivery infrastructure leads to significant power budget fragmentation problem and inefficiently low power utilization. To address this issue, our insight is that heterogeneity of power consumption patterns among different services provides opportunities to re-shape the power profile of each power node by re-distributing services. By grouping services with asynchronous peak times under the same power node, we can reduce the peak power of each node and thus creating more power head-rooms to allow more servers hosted, achieving higher throughput. Based on this insight, we develop a workload-aware service placement framework to systematically spread the service instances with synchronous power patterns evenly under the power supply tree, greatly reducing the peak power draw at power nodes. We then leverage dynamic power profile reshaping to maximally utilize the headroom unlocked by our placement framework. Our experiments based on real production workload and power traces show that we are able to host up to 13% more machines in production, without changing the underlying power infrastructure. Utilizing the unleashed power headroom with dynamic reshaping, we achieve up to an estimated total of 15% and 11% throughput improvement for latency-critical service and batch service respectively at the same time, with up to 44% of energy slack reduction.

Thermal-Aware Resource Management for Embedded Real-Time Systems

Abstract: With an increasing demand for complex and powerful system-on-chips, modern real-time automotive systems face significant challenges in managing on-chip-temperature. We demonstrate, via real experiments, the importance of accounting for dynamic ambient temperature and task-level power dissipation in resource management so as to meet both thermal and timing constraints. To address this problem, we propose RT-TRM, a real-time thermal-aware resource management framework. We first introduce a task-level dynamic power model that can capture different power dissipations with a simple task-level parameter called the activity factor . We then develop two new mechanisms, adaptive parameter assignment and online idle-time scheduling . The former adjusts voltage/frequency levels and task periods according to the varying ambient temperature while preserving feasibility. The latter generates a schedule by allocating idle times efficiently without missing any task/job deadline. By tightly integrating the solutions of these two mechanisms, we can guarantee both thermal and timing constraints in the presence of dynamic ambient temperature variations. We have implemented RT-TRM on an automotive microcontroller to demonstrate its effectiveness, achieving better resource utilization by 18.2% over other runtime approaches while meeting both thermal and timing constraints.

Dynamic Power Distribution System Management With a Locally Connected Communication Network

Abstract: Coordinated optimization and control of distribution-level assets enables a reliable and optimal integration of massive amount of distributed energy resources (DERs) and facilitates distribution system management (DSM). Accordingly, the objective is to coordinate the power injection at the DERs to maintain certain quantities across the network, e.g., voltage magnitude, line flows, and line losses, to be close to a desired profile. By and large, the performance of the DSM algorithms has been challenged by two factors: 1) the possibly nonstrongly connected communication network over DERs that hinders the coordination; and 2) the dynamics of the real system caused by the DERs with heterogeneous capabilities, time-varying operating conditions, and real-time measurement mismatches. In this paper, we investigate the modeling and algorithm design and analysis with the consideration of these two factors. In particular, a game-theoretic characterization is first proposed to account for a locally connected communication network over DERs, along with the analysis of the existence and uniqueness of the Nash equilibrium therein. To achieve the equilibrium in a distributed fashion, a projected-gradient -based asynchronous DSM algorithm is then advocated. The algorithm performance, including the convergence speed and the tracking error, is analytically guaranteed under the dynamic setting. Extensive numerical tests on both synthetic and realistic cases corroborate the analytical results derived.

Dynamic Power Allocation for Downlink Multi-Carrier NOMA Systems

Abstract: In this letter, the performance of optimal dynamic power allocation (PA) is analyzed for downlink multi-carrier non-orthogonal multiple access (MC-NOMA) systems. We study the PA optimization problem with a pair of users and weighted-sum-rate utility. A novel low-complexity algorithm is designed to solve the problem with a closed-form PA expression. The analytical data rate performance is derived and verified by simulation results. The numerical results indicate that a large difference in user channel states improves performance by enhancing the user diversity gain in NOMA systems. Moreover, the channel diversity gain can be achieved by multi-carrier transmission and increases with the number of subcarriers. Finally, the number of subcarriers for dynamic PA in MC-NOMA systems is discussed for practical applications.

Impact of CCGT Start-Up Flexibility and Cycling Costs Toward Renewables Integration

Abstract: The large-scale introduction of variable and limitedly predictable renewables requires flexible power system operation, enabled by, i.a., dynamic power plant operation, storage, demand response, and enhanced interconnections. The fast start-up capabilities of combined-cycle gas turbines (CCGTs) are crucial in this regard. However, these nonstandard operating conditions significantly reduce the lifetime of critical turbine components, as reflected in long-term service agreements (LTSAs). This should also be reflected in short-term scheduling models. In light of this challenge, we apply a unit commitment model that allows multiple start-up loading modes while accounting for the corresponding turbine maintenance costs based on LTSAs. Leveraging this model, we investigated the need for fast start-up capabilities of a set of CCGTs as part of a small-scale test system considering various shares of renewables and dynamic reserve requirements. We have found that fast starts are often cost-optimal despite their greater turbine maintenance costs and a cost reduction of around 1% is obtained when considering more costly fast start-up modes when scheduling. Furthermore, cost-optimal reserve sizing is a function of the planning frequency and is reduced by fast starting capabilities. We conclude that taking advantage of fast start-up capabilities benefits the electricity generation system and yields a significant cost reduction.

Study on Power Factor Behavior in High-Speed Railways Considering Train Timetable

Abstract: In many traction substations (TSSs) of high-speed railways, the average power factor measured in a 24-h period is usually below the required value of 0.9 that takes huge additional energy penalty from the utility companies. It is urgent to evaluate and improve the average power factor at the point of common coupling in the TSS. Thus, a dynamic power factor assessment and its sensitivity method along with detailed dynamic modeling and solving technology are presented here. The power factor behavior of the traction power supply system (TPSS) and corresponding influencing factors are fully investigated. To quantify the effects of variable factors on the average power factor, an average power factor changing rate (APFCR) index is defined and calculated. In addition, the APFCR-index method and shunt reactive compensation (SRC) technique are combined to improve the average power factor. The numerical dynamic power flow results verify the effectiveness and validity of the presented method in studying power factor issue. The dynamic power factor behaviors caused by high-speed trains, electric devices in TPSS, and train timetable have been studied thoroughly. The train timetable is the most important influential factor and the average power factor can be improved through increasing train quantity and installing SRCs. Finally, a generalized platform is proposed for a convenience assessment of various conditions based on the MATLAB and LabVIEW software.

DORA: Optimizing Smartphone Energy Efficiency and Web Browser Performance under Interference

Abstract: This paper proposes DORA — a dynamic frequency controller that maximizes the energy efficiency of smartphones subject to user satisfaction demands in the presence of memory interference stemmed from background processes and coscheduled applications. The proposed algorithm predicts the optimal energy-efficient frequency setting at runtime using staticallytrained performance, dynamic power, and leakage power models. The parameters of the models represent web page characteristics and dynamically varying architecture and system conditions. The algorithm is designed, implemented and extensively evaluated on a Google Nexus 5 smartphone using a variety of mobile web browsing workloads. The results show high prediction accuracies for the performance and power models of 97.5% and 96%, respectively. Overall, DORA improves the smartphone's energy efficiency by an average of 16% compared to the default Android frequency governor, interactive, while maintaining the desired levels of user satisfaction (web page load time).

Dynamic Power Budgeting for Mobile Systems Running Graphics Workloads

Abstract: Competitive graphics performance is crucial for the success of state-of-the-art mobile processors. High graphics performance comes at the cost of higher power consumption, which elevates the temperature due to limited cooling solutions. To avoid thermal violations, the system needs to operate within a power budget. Since the power budget is a shared resource, there is a strong demand for effective dynamic power budgeting techniques. This paper presents a novel technique to efficiently distribute the power budget among the CPU and GPU cores, while maximizing performance. The proposed technique is evaluated using a state-of-the-art mobile platform using industrial benchmarks, and an in-house simulator. The experiments on the mobile platform show up to 15% increase in average frame rate compared to default power allocation algorithms.

PShifter: feedback-based dynamic power shifting within HPC jobs for performance

Abstract: The US Department of Energy (DOE) has set a power target of 20-30MW on the first exascale machines. To achieve one exaFLOPS under this power constraint, it is necessary to manage power intelligently while maximizing performance. Most production-level parallel applications suffer from computational load imbalance across distributed processes due to non-uniform work decomposition. Other factors like manufacturing variation and thermal variation in the machine room may amplify this imbalance. As a result of this imbalance, some processes of a job reach the blocking calls, collectives or barriers earlier and wait for others to reach the same point. This waiting results in a wastage of energy and CPU cycles which degrades application efficiency and performance.We address this problem for power-limited jobs via Power Shifter (PShifter), a dual-level, feedback-based mechanism that intelligently and automatically detects such imbalance and reduces it by dynamically re-distributing a job's power budget across processors to improve the overall performance of the job compared to a naive uniform power distribution across nodes. In contrast to prior work, PShifter ensures that a given power budget is not violated. At the bottom level of PShifter, local agents monitor and control the performance of processors by actuating different power levels. They reduce power from the processors that incur substantial wait times. At the top level, the cluster agent that has the global view of the system, monitors the job's power consumption and provides feedback on the unused power, which is then distributed across the processors of the same job. Our evaluation on an Intel cluster shows that PShifter achieves performance improvement of up to 21% and energy savings of up to 23% compared to uniform power allocation, outperforms static approaches by up to 40% and 22% for codes with and without phase changes, respectively, and outperforms dynamic schemes by up to 19%. To the best of our knowledge, PShifter is the first approach to transparently and automatically apply power capping non-uniformly across processors of a job in a dynamic manner adapting to phase changes.

Matrix-type flexible charging pile and a charging method capable of dynamically allocating power

Abstract: The invention discloses a matrix-type flexible charging pile, and a charging method capable of dynamically allocating power. The method comprises the following steps that S1) each charging terminal is connected to a corresponding electric vehicle; S2) the charging terminal receives a charging power demand of the electric vehicle, and compares the charging power demand with the total module power of a fixed power region corresponding to the charging terminal; S3) if the charging power demand exceeds the total module power of the fixed power region, the charging terminal calculates the number of additional charging modules allocated to the current section of a direct-current bus, and delivers the number of charging modules to a matrix controller; and S4) the matrix controller allocates, according to the required number of charging modules, the required number of charging modules in a dynamic power region to a corresponding direct-current bus, and at the same time, switches a module communication line to a corresponding communication bus. The charging method capable of dynamically allocating power can satisfy charging demands of electric vehicles of different energy storage capacities and different charging rates, and improves the conversion efficiency and the utilization rate of a charging device.

Active control method of large-scale wind integrated power system with enhanced reactive power support for wind speed fluctuation

Abstract: A regional power grid containing high-density wind power is generally connected to a load centre through long-distance transmission lines. Voltage problems, caused by wind speed fluctuation, in the sending-end grid have become increasingly serious. The variable speed wind turbine has been used to compensate the deficit in reactive power. However, the available methods could not meet the requirement because the controllable range of reactive power contributed by wind farms cannot be adjusted. A new idea of reactive power control realised by the active control before wind speed fluctuation was proposed to prevent the voltage variation. Firstly, the influence of wind speed fluctuation on voltage was analysed, and the power controllable range of wind turbines was studied. The principle and strategy of the active control of reactive power were proposed based on the model predictive control theory. Then, the active control model was established according to the system dynamic demand for reactive power. According to wind speed forecasting, the reactive control capability of wind turbine was excavated to meet the grid demand by adjusting active power before wind speed variation, and the reduced active power of wind farms is optimally compensated by thermal powers. Finally, the method was proven to solve voltage problems.

DEC-NoC: An Approximate Framework Based on Dynamic Error Control with Applications to Energy-Efficient NoCs

Abstract: Network-on-Chips (NoCs) have emerged as the standard on-chip communication fabrics for multi/many core systems and system on chips. However, as the number of cores on chip increases, so does power consumption. Recent studies have shown that NoC power consumption can reach up to 40% of the overall chip power [1]-[3]. Considerable research efforts have been deployed to significantly reduce NoC power consumption. In this paper, we build on approximate computing techniques and propose an approximate communication methodology called DEC-NoC for reducing NoC power consumption. The proposed DEC-NoC leverages applications' error tolerance and dynamically reduces the amount of error checking and correction in packet transmission, which results in a significant reduction in the number of retransmitted packets. The reduction in packet retransmission results in reduced power consumption. Our cycle accurate simulation using PARSEC benchmark suites shows that DEC-NoC achieves up to 56% latency reduction and up to 58% dynamic power reduction compared to NoC architectures with conventional error control techniques