基于XGBoost建模和Change-Point残差处理的风电机组齿轮箱温度预警

风电机组具有单机容量小,台数多,状态信息复杂的特点;基于SCADA数据的风电机组齿轮箱预警普遍存在预测精度和计算效率低,资源占用高的问题,选取有效方法以克服上述问题成为齿轮箱预警的关键。极限梯度提升(eXtreme Gradient Boosting,XGBoost)算法本身结合多线程、数据压缩、分片的形式,其在处理大数据量样本时仍具有较高的通用性。采用XGBoost算法建立齿轮箱正常工作状态的温度模型,使用XGBoost模型与其它4类回归预测模型分别进行齿轮箱温度预测模型实验,结果表明XGBoost模型在齿轮箱温度预测中综合性能要优于其它4类模型。当齿轮箱发生故障时,模型的样本特性会发生异常变化,导致模型的预测残差发生明显改变。为提高齿轮箱故障预警的可靠性,构建了新的Change-Point算法,利用Change-Point技术进行XGBoost模型预测残差序列分析,可定位齿轮箱状态劣化时间点,回溯状态劣化原因。

基于Change-Point的风电数据挖掘算法研究

风电机组运行过程中,一些故障导致设备状态发生改变,状态的改变发生在一个持续的时间序列中,找到变化点的时间对于故障回溯及根本原因分析具有重要价值。该文研究风电信号及状态时序变化的特点,引入统计学中的Change-Point算法,通过划分不同置信区间求取置信度方法解决奇异变点的不确定度问题。通过实验对算法进行验证,得出以下结论:Change-Point算法能够有效挖掘到历史数据中的一维及二维模型数据的变化,并给出变点;Change-Point算法思想是挖掘出数据本身的规律性,不受其他条件限制,因此可广泛应用于风电机组数据采集与监视控制(supervisory control and data acquisition,SCADA)系统变量数据挖掘中的问题回溯,快速定位SCADA数据状态变化点。

Time Sequence Change-Point Model of Electrostatic State Parameters of Aircraft Engine

Electrostatic monitoring technology of particle charging information can facilitate online monitoring of aero-engine,which effectively enhances engine fault diagnosis and health managements.Unlike traditional engine state monitoring technologies,aircraft engine monitoring by gas path electrostatic monitoring not only covers the predicted information source itself,but also detects the information that can provide an early warnings for initial fault states through gas path charging levels.This paper establishes a non-stationary time sequence change-point model for anomaly recognition of electrostatic signals based on change-point theory combined with difference method of non-stationary time series.Finally,electrostatic induction data were utilized by the engine life test for a particular aircraft to validate the proposed algorithm.The results indicate that the activity level and the event rate were0.5—0.8(nc)and 50%,respectively,which were far greater than 4—12(pc)and 0—4% under normal working conditions of the engine.

Stochastic differential equation software reliability growth model with change-point

This paper presents software reliability growth models(SRGMs) with change-point based on the stochastic differential equation(SDE).Although SRGMs based on SDE have been developed in a large scale software system,considering the variation of failure distribution in the existing models during testing time is limited.These SDE SRGMs assume that failures have the same distribution.However,in practice,the fault detection rate can be affected by some factors and may be changed at certain point as time proceeds.With respect to this issue,in this paper,SDE SRGMs with changepoint are proposed to precisely reflect the variations of the failure distribution.A real data set is used to evaluate the new models.The experimental results show that the proposed models have a fairly accurate prediction capability.

Change-Point Detection for General Nonparametric Regression Models

A number of statistical tests are proposed for the purpose of change-point detection in a general nonparametric regression model under mild conditions. New proofs are given to prove the weak convergence of the underlying processes which assume remove the stringent condition of bounded total variation of the regression function and need only second moments. Since many quantities, such as the regression function, the distribution of the covariates and the distribution of the errors, are unspecified, the results are not distribution-free. A weighted bootstrap approach is proposed to approximate the limiting distributions. Results of a simulation study for this paper show good performance for moderate samples sizes.

A Bivariate Software Reliability Model with Change-Point and Its Applications

Testing-time when a change of a stochastic characteristic of the software failure-occurrence time or software failure-occurrence time-interval is observed is called change-point. It is said that effect of the change-point on the software reliability growth process influences on accuracy for software reliability assessment based on a software reliability growth model (SRGM). We propose an SRGM with the effect of the change-point based on a bivariate SRGM, in which the software reliability growth process is assumed to depend on the testing-time and testing-effort factors simultaneously, for accurate software reliability assessment. And we discuss an optimal software release problem for deriving optimal testing-effort expenditures based on our model. Further, we show numerical examples of software reliability assessment based on our bivariate SRGM and estimation of optimal testing-effort expenditures by using actual data.

The Power of Change-Point Test for Two-Phase Regression

In this paper, the roughness of the model function to the basis functions and its properties have been considered. We also consider some conditions to take the limit of the roughness when the observations are i.i.d. An explicit formula to calculate the power of change-point test for the two phases regression through the roughness was obtained.

Component-based software reliability analysis incorporating imperfect debugging and change-point

Against the deficiencies of component-based software(CBS) reliability modeling and analysis,for instance,importing too many assumptions,paying less attention to debugging process without considering imperfect debugging and change-point(CP) problems adequately,an approach of CBS reliability process analysis is proposed which incorporates the imperfect debugging and CP.First,perfect/imperfect debugging and CP are reviewed.Based on the queuing theory,a multi-queue multichannel and infinite server queuing model(MMISQM) is presented to sketch the integration test process of CBS.Meanwhile,considering the effects of imperfect debugging and CP,expressions for fault detection and correction are derived based on MMISQM.Numerical results demonstrate that the proposed model can sketch the integration test process of CBS with preferable performance which outperforms other models.

Change-Point Analysis of Survival Data with Application in Clinical Trials

Effects of many medical procedures appear after a time lag, when a significant change occurs in subjects’ failure rate. This paper focuses on the detection and estimation of such changes which is important for the evaluation and comparison of treatments and prediction of their effects. Unlike the classical change-point model, measurements may still be identically distributed, and the change point is a parameter of their common survival function. Some of the classical change-point detection techniques can still be used but the results are different. Contrary to the classical model, the maximum likelihood estimator of a change point appears consistent, even in presence of nuisance parameters. However, a more efficient procedure can be derived from Kaplan-Meier estimation of the survival function followed by the least-squares estimation of the change point. Strong consistency of these estimation schemes is proved. The finite-sample properties are examined by a Monte Carlo study. Proposed methods are applied to a recent clinical trial of the treatment program for strong drug dependence.

Change-point estimation for censored regression model

In this paper, we consider the change-point estimation in the censored regression model assuming that there exists one change point. A nonparametric estimate of the change-point is proposed and is shown to be strongly consistent. Furthermore, its convergence rate is also obtained.

Monitoring Mean and Variance Change-Points in Long-Memory Time Series

This paper proposes two ratio-type statistics to sequentially detect mean and variance change-points in the long-memory time series.The limiting distributions of monitoring statistics under the no change-point null hypothesis,alternative hypothesis as well as change-point misspecified hypothesis are proved.In particular,a sieve bootstrap approximation method is proposed to determine the critical values.Simulations indicate that the new monitoring procedures have better finite sample performance than the available off-line tests when the change-point nears to the beginning time of monitoring,and can discriminate between mean and variance change-point.Finally,the authors illustrate their procedures via two real data sets:A set of annual volume of discharge data of the Nile river,and a set of monthly temperature data of northern hemisphere.The authors find a new variance change-point in the latter data.

MAXIMUM SCORE CHANGE-POINT ESTIMATION IN BINARY RESPONSE MODEL

This paper is concerned with the estimation of change-point in a binary response model with the assumption that the conditional median of the error term, given the explanatory variable, is zero. We construct an estimation of change-point based on the maximum score function and give its exponential convergence rate under some mild conditions.

Asymptotic and Bootstrap Tests for a Sequential Change-Point of Panel

We design monitoring procedures for the common change-point in a sequential panel data model. An asymptotic method and two new bootstrap methods are proposed to obtain critical values. We establish the asymptotic validity of the proposed bootstrap procedures. In simulation studies the empirical test size and the empirical test power values are investigated to show that the three tests are valid and have their own applications.At the same time, the estimations of an unknown change-point are obtained by using the proposed test statistic with these three methods.

Change-Point Estimates in Longitudinal Binary Data

Most change-point models assume that the response is continuous or cross sectional binary. However, in many public health problems, the data is longitudinal binary. There are few studies of change-point problems for longitudinal outcomes. This paper describes a flexible change-point model which includes random-effects and takes into account the difference between various individuals in longitudinal binary data. A transition function is used to make the linear-linear logistic model differentiable at the change-point. The location of the change-point is estimated using the maximum likelihood method. Adjustment of the transition parameter from zero to one controls the sharpness of the transition. The performance of this estimation procedure is illustrated with simulations using SAS/proc nlmixed and a detailed analysis of data relating hormone levels and ovary functions based on data from the Obstetrics and Gynecology Hospital, Medical Center of Fudan University.

Group LASSO for Change-points in Functional Time Series

Multiple change-points estimation for functional time series is studied in this paper.The change-point problem is first transformed into a high-dimensional sparse estimation problem via basis functions.Group least absolute shrinkage and selection operator(LASSO)is then applied to estimate the number and the locations of possible change points.However,the group LASSO(GLASSO)always overestimate the true points.To circumvent this problem,a further Information Criterion(IC)is applied to eliminate the redundant estimated points.It is shown that the proposed two-step procedure estimates the number and the locations of the change-points consistently.Simulations and two temperature data examples are also provided to illustrate the finite sample performance of the proposed method.

Smoothed Maximum Score Change-point Estimation in Binary Response Model

This paper studies the large sample properties of the change point estimates in binary response models. The estimate is obtained by maximizing the smoothed score function when the median of the latent error variable is assumed to be zero. An exponential convergence rate of the change point estimate is also established.

Reliability Evaluation of Two-Phase Degradation Process with a Fuzzy Change-Point

For some products,degradation mechanisms change during testing,and therefore,their degradation patterns vary at different points in time;these points are called change-points.Owing to the limitation of measurement costs,time intervals for degradation measurements are usually very long,and thus,the value of change-points cannot be determined.Conventionally,a certain degradation measurement is selected as the change-point in a two-phase degradation process.According to the tendency of the two-phase degradation process,the change-point is probably located in the interval between two neighboring degradation measurements,and it is a fuzzy variable.The imprecision of the change-point may lead to the incorrect product’s reliability evaluation results.In this paper,based on the fuzzy theory,a two-phase degradation model with a fuzzy change-point and a statistical analysis method are proposed.First,a two-phase Wiener degradation model is developed according to the membership function of the change-point.Second,the reliability evaluation is carried out using maximum likelihood estimation and a fuzzy simulation approach.Finally,the proposed methodology is verified via a case study.The results of the study show that the proposed methodology can achieve more believable reliability evaluation results compared with those of the conventional approach.

Ovarian Cancer Screening Based on Mixture Change-Point Model

Ovarian cancer is one of the most deadly female genital malignant tumors in many regions while an effective early screening strategy can save numerous lives.CA125 and HE4 are tumor markers validated efficacious as well as most commonly used in recent screening research of ovarian cancer.In this paper,the authors construct a change-point and mixture model on the basis of longitudinal CA125 and HE4 levels and estimated parameters using maximum likelihood method with the preclinical duration assumed right-censored,which is more adaptive and yields comparable results in comparison to the Bayesian approach raised by Skates.Consistency of estimators is proved.The authors also run a 5-year simulation of sequential screening by calculating the risk of cancer and hypothesis testing the true incidence time respectively.Results show that diagnosis based on hypothesis test performs better in early detection.

Parameter optimization of complex network based on the change-point identification

This paper proposes a novel method for the parameter optimization of complex networks established through coarsening and phase space reconstruction.Firstly,we identify the change-points of the time series based on the cumulative sum(CUSUM)control chart method.Then,we optimize the coarse-graining parameters and phase space embedding dimension based on the evolution analysis of the global topology index(betweenness)at the mutation point.Finally,we conduct a simulation analysis based on real-time data of Chinese copper spot prices.The results show that the delay of the copper spot prices in Chinese spot market is 1 day,and the optimal embedding dimension of the phase space reconstruction is 3.The acceptance level of the investors towards the small fluctuations in copper spot prices is 0.2 times than the average level of price fluctuations,which means that an average price fluctuation of 0.2 times is the optimal coarse-graining parameter.

Asymptotic Distribution of the Jump Change-Point Estimator

The asymptotic distribution of the change-point estimator in a jump change- point model is considered. For the jump change-point model Xi - α ＋ θ{[nτ0] 〈 i ≤ n} ＋ εi, where εi （i = 1,…. , n） are independent identically distributed random variables with Eεi -= 0 and Var（εi） 〈 ∞, with the help of the slip window method, the asymptotic distribution of the jump change-point estimator τ is studied under the condition of the local alternative hypothesis.