Fundamental Concept
SCADA：Supervisory control and data acquisition数据采集与监视控制系统
Parameters being monitored include
generator rpm, generator bearing temperatures, oil temperature, pitch angle,
yaw system, wind speeds
GUI： graphical user interfaces 图形用户界面
professional phrases
wind curtailment 弃风现象
outlier 离群值/异常值
power grid 电力公司
histogram 直方图
Regression model回归模型
DPI：Stands for “Dots Per Inch.” DPI is used to measure the resolution of an image both on screen and in print.
Raw Data Description
Program 1 2018 Scada data of A Wind Turbine
分析对象：风力发电机功率
目标：构造风力发电输出功率曲线模型
模型影响因素： 数据筛选处理方式、风机功率曲线，最优化决策选择
解决方法：regression models 回归分析模型
objective functions 优化目标函数
风力发电特点：andomness and intermittency 随机性和间歇性
风力风速-输出功率曲线图
Note：A typical wind turbine power curve has three main characteristic speeds:
- cut-in (Vc ); 2) rated (Vr ); and 3) cut-out (Vs) speeds.
1）启动风速; 空载转速;2）额定速度 3）切出风速（考虑风机安全，停止运行的速度）
实际风速-功率曲线影响因素
the empirical power curves could be substantially different from the theoretical ones due to
- the location of the turbine,
- air density,
- wind velocity distribution,
- wind direction,
- mechanical and control issues,
- uncertainties in measurements.
数据说明
In Wind Turbines, Scada Systems measure and save data’s like wind speed, wind direction, generated power etc. for 10 minutes intervals. This file was taken from a wind turbine’s scada system that is working and generating power in Turkey.
The data’s in the file are:
Date/Time (for 10 minutes intervals)
LV ActivePower (kW): The power generated by the turbine for that moment
Wind Speed (m/s): The wind speed at the hub height of the turbine (the wind speed that turbine use for electricity generation)
TheoreticalPowerCurve (KWh): The theoretical power values that the turbine generates with that wind speed which is given by the turbine manufacturer, and
Wind Direction (°): The wind direction at the hub height of the turbine (wind turbines turn to this direction automaticly)
模型建立路线
1、数据处理
2、功率模型建立
Preprocessing of data
man-made approaches,
statistic-based models
learning-based models.
Four-step method for data preprocessing
validity check, 数据有效性检查
data scaling, 数据范围缩放
missing data preprocessing, 遗失数据预处理
lag removal.？？？？？
three kinds of representative outliers
1、The first type ： the value of wind power tends to zero when the corresponding wind speed is greater than the cut-in wind speed but less than the cut-out wind speed.
Cause：Wind turbine maintenance and wind curtailment .
2、The second type ：outlier are also far from their ideal power data on the power curve, but do not equal to zero.
Cause： wind curtailment and dirt, bugs, or ice on blades, pitch malfunction and other factors.
3、The final type： the value of wind power exceeds the physical limitation of the wind turbine.
error1：
Solving：新增 pip install windrose
Python Package
matplotlib
pyplot—— drawing
basic structure
# coding=utf-8
from matplotlib import pyplot as plt# 导包
x=range(2,26,2)
y=[15,13,14,6,16,19,22,9,11.5,16,19,21]
# 绘图，展示图形
plt.plot(x,y)
plt.show()
问题1：matplotlib中文字体不显示
设置中文字体
from matplotlib.font_manager import FontProperties
font= FontProperties(fname=r"C:\Windows\Fonts\simhei.ttf", size=10) # 此文件路径为windows自带的字体路径，需要什么字体都可以此路径下找
plt.xticks(list(x)[::3],_xtick_labels[::3],rotation=(45),fontproperties=font)#rotation 旋转坐标刻度角度
# 注意fontproperties=font，其中这个font是用户自定义的预使用字体代号
问题2：报错
实际上是指x,y数据个数没有对应，注意：range（a,b)是娶不到b的！！！！
加入水平、垂直参考线
import matplotlib.pyplot as plt
plt.axhline(y=0.5, color='r', linestyle='-')
plt.show()
numpy
reshape/flatten
random
round
dtype
(24,1)与（24，）不一样，前者为二维数组，后者为一维
Axis
一维数组 0轴
二维数组 0、1轴
三维数组 0、1、2轴
numpy 读取数据
CSV type
预测
KNeighborsRegressor：通过找出一个样本的k个最近邻居，将这些邻居的某个（些）属性的平均值赋给该样本，就可以得到该样本对应属性的值。
DecisionTreeRegressor：回归树叶节点的数据类型是连续的，而分类树叶节点的数据类型是离散的。
回归树叶节点是一个个具体的值，而分类树叶节点是依据训练样本类别确定的预测类别。
回归树的叶节点返回的是“一团”训练数据的均值，而不是具体的、连续的预测值。
ExtraTreesRegressor()：
EDPR Wind Farm
professional phrases
met mast.（测风塔）
long-term projected life span; 长时间的预计寿命
cumulative fatigue 累积疲劳
down time 停机时间
监督学习(supervised learning)
Thesis Learning-Machine Learning for Predictive Maintenance on Wind Turbines
Core aim
1、predictive maintenance on wind turbines
2、big data analysis
3、machine learning algorithms
The main objective of this thesis project is to enhance the current body of knowledge regarding
predictive maintenance on wind turbines by implementing a big data analysis on a real world data set.
wind turbine
subsystem failure
big data life cycle
风力发电机常用故障诊断方法
风机齿轮箱故障诊断技术研究综述
周泽坤
( 上海电机学院，上海 200240)
数据预处理
依据四分位法的形式将运行过程中由于系统组件所带来的自身问题
进行数据剔除
基于深度变分自编码网络融合SCADA数据的
风电齿轮箱故障预警
