基于能量加权高斯过程回归的锂离子电池健康状态预测

针对容量再生现象影响锂离子电池健康状态预测(SOH)建模精度的问题,提出一种经验模态分解(EMD)的能量加权高斯过程回归(EWGPR)方法。该方法将容量再生现象看作是锂离子电池容量衰减过程的能量凸现,利用EMD分解获得样本的能量分布情况,根据能量情况计算每个样本的权重,进而建立基于能量加权高斯过程回归的锂离子电池SOH预测模型。基于NASA锂电池数据集的仿真实验结果表明,EWGPR方法比基本GPR算法具有更高的精度和适应性,单步预测和多步预测的均方根误差(RMSE)分别减少了3%和10%。

基于PF-GPR算法的锂离子电池剩余使用寿命预测

针对锂离子电池退化过程中不可避免的容量再生现象建立了电池退化模型,提出了融合粒子滤波(PF)和高斯过程回归(GPR)的电池剩余使用寿命(RUL)预测算法。仿真实验结果表明,所提出的算法能够实现准确的锂离子电池RUL预测。

锂离子电池剩余寿命预测方法综述

准确预测锂离子电池的剩余寿命对提高设备的安全性和降低设备的维护成本具有重要意义。针对锂离子电池剩余寿命预测方法的研究现状进行分析,归纳总结了基于模型的方法、基于数据驱动的方法和基于融合的方法,对比分析了不同方法的优缺点。然后总结归纳了锂离子电池剩余寿命预测在实际应用中常见的两个问题,最后分析了未来的发展趋势和挑战。

Electricity Generation in Turkey by Geothermal Energy Resources

Electricity generation generally is made in thermal, hydro, geothermal power plants and windfarms/windparks. Because of some advantages such as renewability, low-cost, clean, safe and naturality geothermal and wind energy will have been electricity generation source in the near future. Turkey has hot water springs suitable for electricity generation between 130-242 ~C with natural vapor and hydrothermal alterations in connection with grabens limited to active faults and diffuse young volcanism in Western Anatolia Region. Other renewable energy and electricity generation resource is wind energy. In Turkey electricity generation is made by windfarms/windparks. These parks/farms are generated 1,414.55 MW electricity. The year 2010 electricity general total installed capacity in Turkey is about 49,524.1 MW. According to the total installed capacity, thermal power plants have 65.18%, hydro power plants have 31.97%, geothermal power plants and wind farms have 2.85%. Electricity generation generally was obtained from 15 thermal power plants. In this study high temperature geothermal fields and windparks/windfarms in Turkey which are suitable for electricity generation potential were investigated.

A Nodal Electricity Market Design Based on Regulated Access to Transmission Network

Accomodation of power system constraints with the market mechanism is encountered as a major challenge along the way toward implementation of different electricity market designs. Allocation of fixe or flow-dependent inter-zone trading capacities by the PX （power exchange） can not be accepted unreservedly. The paper is meant to show that a nodal electricity market design that is based on bids for local energy and a regulated transmission access including allocation of ＂entry-exit＂ transmission capacity would be the desired solution. The market players could easily optimize their portfolio while the TSOs （transmission system operators） are requested to mobilize the network＇s in-built flexibility to increase the cross zonal capacity. In the proposed market design, the PX＇s allocation of trading capacity is clearly separate from the TSO＇s management of the power system operational constraint5. Clear operator roles would enlarge access to electricity market as well as market integration of variable RESs （renewable energy sources） that are critically dependent on short notice access to regionat markets,

基于经验模态分解的锂离子电池健康状态预测

电池健康状态(State of health, SOH)预测是确保电子系统运行可靠性和安全性的关键因素。为了准确地预测锂离子电池SOH的整体退化趋势和局部容量再生现象,提出一种将经验模态分解(Empiricalmodedecomposition,EMD)与门控循环单元(Gated recurrent unit, GRU)和差分自回归移动平均模型(Autoregressive integrated moving average model, ARIMA)相融合的锂离子电池SOH预测方法。首先,利用EMD将电池原始SOH序列进行多尺度分解,并通过计算分解子序列的连续均方误差找到高低频分界点;然后,GRU用于预测具有强烈数据波动的高频子序列,ARIMA用于预测剩余的低频子序列和残差;最后,将每个子序列的预测结果进行叠加以获得最终预测结果。试验结果表明,与其他文献中预测方法相比,基于经验模态分解的融合模型具有更高的预测精度,可以更好地捕捉电池SOH整体退化趋势和局部容量再生特性。

一种多时间尺度下的锂离子电池剩余寿命混合预测方法

针对锂离子电池储能系统寿命预测不准确问题,提出了一种多时间尺度下的电池剩余寿命混合预测方法。首先,通过带自适应噪声的完全集成经验模态分解(CEEMDAN),将原始数据分解为多个低频和高频分量。接着,利用GS-LM模型对其低频分量进行拟合预测,再将分解后的高频分量作为LSM-STW模型(滑动时间窗口和长短期记忆神经网络)的输入,得到高频分量预测结果。最后,将所有预测结果相加,得到最终的锂离子电池剩余寿命预测。实验结果表明,所提出的预测方法可以准确地预测锂离子电池容量退化趋势,不同预测起点对预测结果影响较小,并能够捕捉电池容量的再生现象。