基于Matlab实现电动汽车充电负荷能量分解算法仿真（源码+数据+说明文档+报告）.rar

共21个文件

m：14个

txt：3个

pdf：1个

版权申诉

matlab

5星 · 超过95%的资源 114 浏览量 2023-07-19 10:21:43 上传评论收藏 629KB RAR 举报

资源推荐

资源详情

资源评论

收起资源包目录

基于Matlab实现电动汽车充电负荷能量分解算法仿真（源码+数据+说明文档+报告）.rar （21个子文件）

基于Matlab实现电动汽车充电负荷能量分解算法仿真（源码+数据+说明文档+报告）

IECON2014_Codes

pitFilter.m 640B

说明文档.txt 738B

getHeight.m 270B

findType.m 2KB

localNoiseAmplitude.m 1KB

detectDryer.m 6KB

getSignalWidth.m 878B

bumpTrainFilter.m 5KB

dryerFilter.m 1KB

getSignal.m 469B

estEV.m 23KB

spikeFilter.m 951B

getSegment.m 935B

报告.pdf 177KB

EV_Demo.m 2KB

Data

数据说明.txt 1KB

3367_2013_05.mat 426KB

ACdetector.m 1KB

注意事项.txt 129B

图片.jpg 13KB

说明文档.doc 51KB

Training-Free Non-Intrusive Load Monitoring of

Electric Vehicle Charging with Low Sampling Rate

Zhilin Zhang

∗

, Jae Hyun Son, Ying Li, Mark Trayer, Zhouyue Pi

Samsung Research America – Dallas

1301 E. Lookout Drive, Richardson, TX 75082, USA

∗

Email: [email protected]

Dong Yoon Hwang, Joong Ki Moon

Smart Home Solution Lab

Samsung Electronics Inc.

Suwon, Kyeong-gi-do, Korea

Abstract—Non-intrusive load monitoring (NILM) is an im-

portant topic in smart-grid and smart-home. Many energy

disaggregation algorithms have been proposed to detect various

individual appliances from one aggregated signal observation.

However, few works studied the energy disaggregation of plug-

in electric vehicle (EV) charging in the residential environment

since EVs charging at home has emerged only recently. Recent

studies showed that EV charging has a large impact on smart-

grid especially in summer. Therefore, EV charging monitoring has

become a more important and urgent missing piece in energy

disaggregation. In this paper, we present a novel method to

disaggregate EV charging signals from aggregated real power

signals. The proposed method can effectively mitigate interference

coming from air-conditioner (AC), enabling accurate EV charging

detection and energy estimation under the presence of AC power

signals. Besides, the proposed algorithm requires no training,

demands a light computational load, delivers high estimation

accuracy, and works well for data recorded at the low sampling

rate 1/60 Hz. When the algorithm is tested on real-world data

recorded from 11 houses over about a whole year (total 125

months worth of data), the averaged error in estimating energy

consumption of EV charging is 15.7 kwh/month (while the

true averaged energy consumption of EV charging is 208.5

kwh/month), and the averaged normalized mean square error

in disaggregating EV charging load signals is 0.19.

Keywords—Non-intrusive load monitoring (NILM); Electric

Vehicle (EV); Smart Grid; Energy Disaggregation

I. INTRODUCTION

Non-intrusive load monitoring (NILM) or non-intrusive

appliance load monitoring (NIALM) is an important solution

to realize smart-grid and smart-home energy management

beneﬁts. It aims to estimate operation status and energy

consumption of individual electronic appliances by monitoring

aggregated current/voltage/power signals in the main circuit

panel of a house or a building [1]–[3].

Electric vehicle (EV) charging is becoming an important

load element for smart grid analysis [4]–[6] although home

charging EVs recently entered the market. Due to the growing

number of the EV customers, a utility might start to experience

non-marginal impacts on parts of its distribution system.

Particularly, the gravity of this impact will depend on at what

time, for how long, with what utility rate, and in what season

these EVs are being charged [7]. Therefore, it is necessary to

solicit the importance and urgency of monitoring EV charging

load via energy disaggregation.

Another usage of monitoring EV charging load is to

provide house owners the monthly energy consumption of EV

0 500 1000

Power (kw)

Time Index (min.)

(a) EV Power Signal

200 400 600 800 1000 1200 1400

0.5

1.5

2.5

3.5

Power(kw)

Time Index(min.)

(b) AC Power Signal

Fig. 1. (a) An EV power signal. (b) An AC power signal exhibiting two

kinds of waveform patterns, i.e. spike trains and lumps.

charging. This monthly feedback information can help house

owners in bill-management and travel-management in the same

way as monthly gas bill and conventional monthly electricity

bill [8].

There are many algorithms available for the energy disag-

gregation of various residential appliances [1]–[3], [9]–[14],

such as hidden Markov model(HMM) algorithms [10], [11],

[13]. However, these algorithms were not speciﬁcally designed

for EV charging, and they require extensive training and a large

computational load. Therefore, for practical implementation

where simultaneously monitoring tens of thousands houses is

required, those algorithms may not be an attractive solution.

In this paper, a novel algorithm for energy disaggregation

of EV charging is presented. It has several desired advantages.

(1) It can mitigate the interference coming from air-conditioner

(AC) power signals. Thus, it could be very helpful for smart

grid load analysis and management during peak load time in

summer. (2) It does not require training, which is an highly

attractive feature toward practical implementation. (3) It de-

mands a light computational load, thus suitable for monitoring

tens of thousands residual houses in large scale. (4) It works

well for data sampled at 1/60 Hz, which aligns with the data

provision capability of many smart-meters. Experiments based

on real-world power data showed that it exhibits far better

performance than state-of-the-art algorithms.

II. C

HALLENGES

One big challenge of disaggregating an EV charging load

from aggregated power signals is mitigating interference from

AC. As shown in Fig.1(a), an EV charging load signal can be

characterized as a square wave of a high amplitude (higher

than 3 kW) and a long duration (longer than 30 minutes but

generally shorter than 200 minutes) [5]. AC power signals

usually exhibit two kinds of waveform patterns. One pattern

resembles a spike train with very short durations (e.g. the

train of waves from the 1st to 700-th minutes in Fig.1(b)).

Another waveform pattern resembles a rectangular waveform

of a high and slowly ﬂuctuating amplitude and a long duration

(e.g. the two lumps from the 700-th to 1200-th minutes in

Fig.1(b)). This waveform pattern can seriously affect disag-

gregation performance of EV charging load signals due to

the difﬁculty of distinguishing the AC waveform pattern from

EV charging load signals, especially in the presence of other

appliances’ power signals and highly ﬂuctuating residual noise.

For notational convenience, this kind of AC waveforms will

be called as AC lumps.

Another challenge lies with the aggregated data themselves

of being real power signals sampled at 1/60 Hz. At this

sampling rate, many useful appliance signatures [15]–[17] such

as transient characteristics available from high sampling rates

no longer exist, which limits pattern recognition tools to render

accurate disaggregation results.

The third challenge is the lack of ground-truth of EV

charging load signals for each house and its large variation

across different houses. To obtain the ground-truth of EV

charging load signals in a given house, it requires to install

sub-meter sensors to record these signals. However, it is

unpractical to install such sub-meter sensors in every house.

Thus, when disaggregating EV charging load from aggregated

power signals in a given house, there is no training set (i.e.,

a collection of ground-truth of EV charging load signals in

the house) available to train an algorithm. On the other side,

EV charging load signals have large variation across different

houses. For example, EV charging load signals could have

different amplitudes (although always higher than 3 kW), dif-

ferent width (i.e., charging duration), and different appearance

time from house to house. As a result, an algorithm working

well for a given house may perform poorly for another house.

In summary, a practical algorithm should work well for

various houses and every season (especially the summer), and

should not require training sets. But due to the above issues,

to achieve high disaggregation accuracy of EV charging load

is truly challenging.

III. P

ROPOSED ALGORITHM

This section will describe the proposed algorithm in detail.

A one-day aggregated power signal (taken from the Pecan

Street Database [18]) is chosen for illustration purpose (see

Fig.2(a)).

A. Step 1: Thresholding the Aggregated Signal

For a given aggregated signal x(t), ﬁrst, a threshold T

low

is applied to obtain a rough estimate of the EV charging load

signal:

(t)=



x(t) x(t) ≥ T

low

0 x(t) <T

low

(1)

where T

low

 max



2.5,

2|x(k)>2|



k: x(k)>2

x(k)



, and

|x(k) > 2| counts the number of sampling points whose am-

plitude is larger than 2 kW. After the initial thresholding (see

Fig.3(a)), the segments information of x

(t) can be obtained

200 400 600 800 1000 1200 1400

Aggregated Signal

(a)

Power (kw)

200 400 600 800 1000 1200 1400

Ground−Truth of AC

(b)

Power (kw)

200 400 600 800 1000 1200 1400

Ground−Truth of EV (12.0 kwh)

(c)

Power (kw)

200 400 600 800 1000 1200 1400

Estimated EV (12.7 kwh)

(d)

Power (kw)

Time Index (min.)

Fig. 2. Energy disaggregation by the proposed algorithm. (a) An aggregated

signal of one day. (b) The ground-truth of AC. (c) The ground-truth of

EV (energy consumption is 12.0 kwh). (d) The estimated EV power signal

(estimated energy consumption is 12.7 kwh). The energy estimation error

(deﬁned in Section IV) is 5.8%, and the MSE is 0.178.

200 400 600 800 1000 1200 1400

Signal After Thresholding in Step 1

(a)

Power (kw)

200 400 600 800 1000 1200 1400

Ground−Truth of EV

(b)

Power (kw)

200 400 600 800 1000 1200 1400

Result After Spike−Train Filter

(c)

Power (kw)

200 400 600 800 1000 1200 1400

Result After Using Large Fixed Value (20)

(d)

Power (kw)

200 400 600 800 1000 1200 1400

Result After Using Large Fixed Value (90)

(e)

Power (kw)

Time Index (min.)

Fig. 3. Results after using the spike-train ﬁlter and the ﬁx-value thresholding

method. (a) The aggregated signal after thresholding in Step 1. (b) The ground-

truth of EV power signal, which has two waves. (c) The result after using

our spike-train ﬁlter. (d) The result after thresholding using a small ﬁxed

value (20). There were many short AC spikes unremoved. (e) The result after

thresholding using a large ﬁxed value (90). Note that the ﬁrst EV wave (from

the 510-th to 550-th minute) was also removed.

such as the locations of a start-point and an end-point of each

segment.

B. Step 2: Filtering the Spike-Train

Many AC spikes are present in the thresholding result

(Fig.3(a)), which need to be removed. One may consider

setting a threshold to remove all spikes whose duration is

shorter than the threshold. However, it is not easy to ﬁnd a

suitable duration threshold to remove all these spikes due to

the varying nature of an AC spike duration (see Fig.2(b)).

Note that the duration of AC spikes gradually increases

from morning to later afternoon and gradually decreases from

later afternoon to midnight. Based on this observation, the

following ﬁlter is designed to remove these spikes.

It ﬁrst ﬁnds segments with duration shorter than T

seed

=20

(minutes), which are called ‘seeds’ and labeled as ‘spikes to

remove’. Then, from each ‘seed’, the ﬁlter searches the nearest

segment forwardly, checking whether the segment’s duration

is shorter than D  (1 + η)D

cur

and whether the gap between

the ‘seed’ and the nearest segment is no more than 3D

cur

where D

cur

is the duration of the current ‘seed’ and η is a

duration extension parameter (η =1.2 in our algorithm). If this

search condition meets, this nearest segment will be labeled as

‘spikes to remove’ and will be set as a new ‘seed’. Now, using

this new ‘seed’, the ﬁlter repeats the same forward segment

searching to the nearest segment, checking its condition in the

same prescribed manner. If the search condition is not met,

then jump to another ‘seed’ and check its nearest segment

forwardly as before. Similarly, the ﬁlter searches backwardly

as well. In the end, after completing the whole search range,

all segments labeled as ‘spikes to remove’ are removed from

(t).

To prevent from removing a segment with a very large

duration, one can adjust a threshold T

spike

such that all

removed segments have duration no more than T

spike

. For the

proposed algorithm, it is set as T

spike

= 90(minutes).

Note that the ﬁlter does not remove all segments which

have duration no more than T

spike

. It removes a segment

only if its duration does not increase sharply compared to

its surrounding segments’ duration. If a segment with a long

duration is surrounded by very short segments, even if this

long segment has duration shorter than T

spike

, it will not be

removed. The reason is that this segment could potentially

indicate a waveform of EV, dryer, or oven. So, it requires

further examination.

Fig.3 shows one example that the proposed spike-train

ﬁlter removed all AC spikes. While using a ﬁxed threshold

value to remove these AC spikes, several single spikes were

unremoved due to a small threshold (Fig.3(d)), or a part of

an EV power signal was removed mistakenly due to a large

threshold (Fig.3(e)).

C. Step 3: Removing Residual Noise

Residual noise refers to the mixture of errors from ﬂuctu-

ation of power signals, loss in power lines, and power signals

of appliances with a low amplitude. With location information

of each segment obtained in Step 1, the amplitude of residual

noise can be estimated around each segment. For each segment,

using the minimum value of N

points immediately before the

segment and the minimum value of N

points immediately

after the segment, the amplitude of the local residual noise

can be estimated by averaging the two minimum values. The

residual noise removal can be obtained by subtracting the

segment by its associated local residual noise amplitude. In

our algorithm N

= N

=5.

D. Step 4: Classifying the Type of Each Segment

At this point, there are only a few remaining segments

in the ﬁltered aggregated signal. And every segment can be

classiﬁed into one of three types.

Type 0: The segment belongs to a dryer/oven waveform,

or belongs to an EV waveform fully overlapping

with a dryer/oven waveform which has almost the

same duration as the EV waveform. For the former

case, the segment can be simply removed since it

is not an EV waveform. For the latter case, the

segment should have very high amplitude since a

dryer/oven waveform has high amplitude like an

EV waveform (generally higher than 5 kW).

Type 1: The segment belongs to an EV waveform, or an

AC lump, or an EV waveform overlapping with

waveforms of non-AC appliances with relatively

shorter durations, or an AC lump overlapping with

waveforms of other appliances. One can calculate

the approximate width and height of the segment,

decide whether it is an EV waveform, and then

reconstruct the EV waveform.

Type 2: The segment belongs to an EV waveform over-

lapping with an AC waveform, which is probably

also overlapping with other appliances’ wave-

forms. For example, the ﬁrst two segments shown

in Fig.3(c) are respectively an EV waveform

overlapping with an AC spike train and an EV

waveform overlapping with an AC lump and a

dryer waveform.

To classify a given segment S(t) after Step 3, the following

cumulative counting function is calculated:

f(c)=S(t) >c (2)

where c is an amplitude threshold from 0 to max(S(t)), and

the operator S(t) >c counts the number of sampling points

in S(t) with an amplitude greater than c. For example, if c =0,

then f(c) is the total number of all nonzero samples in the

segment. If c = max(S(t)), then f(c)=0.

When calculating the gradient of the cumulative function

f(c), one can ﬁnd that there are two prominent peaks for Type

2 segments. This is because both an AC waveform and an EV

waveform can be approximated as square waves, and a square

wave can result in sharp drop in f(c) when c is equal to the

height of the square wave. Similarly, there is one

prominent

peak in the gradient of f (c) for Type 1 segments and no

prominent peak for Type 0 segments. Thus, the number of

prominent peaks in the gradient of the function suggests which

type an observing segment belongs to.

To ﬁnd prominent peaks, we search peaks with mutual dis-

tance larger than 2 kW and peak height larger than 0.2 max(g)

where g is the gradient of f. The Matlab command ﬁndpeaks

can ﬁnish this task easily. If there is one peak, the segment

is classiﬁed as Type 1 (Fig.5). If there are at least two

peaks, further calculate the area under the normalized gradient

function g

 g/ max(g). If the area is larger than 35% of

the square area with the same width and height as g

(e.g.

the green square area in Fig.4 and Fig.6), then the segment is

classiﬁed as Type 0 (Fig.4); otherwise, it is classiﬁed as Type

评论收藏

内容反馈

版权申诉

Zhuzhuxiakl

2024-05-12

感谢大佬分享的资源，对我启发很大，给了我新的灵感。
qq_40896220

2023-11-28

发现一个宝藏资源，资源有很高的参考价值，赶紧学起来~
2301_77480770

2023-12-15

感谢资源主的分享，很值得参考学习，资源价值较高，支持！

Matlab仿真实验室

粉丝: 2w+
资源: 2180

基于Matlab实现电动汽车充电负荷能量分解算法仿真（源码+数据+说明文档+报告）.rar

基于Matlab实现大型电动汽车充电行为的仿真研究（源码+数据）.rar

基于Matlab大型电动汽车充电行为的仿真研究（源码+数据）.rar

基于Matlab实现凸优化各种算法（源码+使用说明）.rar

基于Matlab和Simulink实现配电网仿真（源码+模型+报告+PPT）.rar

matlab程序-电动汽车.rar_Matlab电源规划_充电站 规划_充电站 规划_充电站matlab_配送

用matlab实现的仿真滑模各个算法研究源码+数据.zip

基于Matlab实现的Kmeans聚类算法仿真源码+数据（高分课程设计）.zip

matlab实现USV路径跟踪LOS控制算法仿真源码+全部数据（大作业课设）.zip

基于Matlab实现GPS IMU经典ESKF融合算法仿真（源码+数据+说明文档）.rar

基于Matlab实现决策树C4.5算法（源码+数据+教程）.rar

基于Matlab模拟电车充电需求.zip

基于MATLAB仿真的电动车滑行自充电系统设计.pdf

电动汽车驱动力与行驶阻力平衡图在Matlab中的程序

基于遗传算法的新能源电动汽车充电桩与路径选择MATLAB程序

基于Matlab实现UWB IMU融合算法仿真（源码+数据+说明文档）.rar

基于Matlab实现VRU和AHRS姿态融合算法仿真（源码+数据+说明文档）.rar

基于Matlab实现GNSS和INS组合导航算法仿真（源码+数据+说明文档）.rar

基于Matlab实现电机仿真源码+数据+报告（高分项目）.rar

标准全能电瓶车控制程序

基于Matlab实现粒子群算法优化微电网能量管理仿真（源码+数据）.rar

基于Matlab实现支持向量机算法（源码+数据+说明文档）.rar

基于Matlab实现svm算法仿真（源码+数据）.rar

基于Matlab实现决策树与随机森林算法（源码+数据+说明文档）.rar

基于Matlab实现金融算法股票预测仿真实战（源码+数据+说明文档）.rar

基于Python的二手房数据分析源码+数据+说明文档+分析报告（高分项目）.rar

基于有机发光二极管(OLED)的建模优化算法的matlab仿真源码+数据.zip

基于Matlab低温多效蒸馏海水淡化产水仿真（源码+数据+报告）.rar

基于MATLAB实现潮流计算仿真（完整源码+说明文档+数据）.rar

最新资源

matlab程序-电动汽车.rar_Matlab电源规划_充电站规划_充电站规划_充电站matlab_配送