Time Series Anomaly Detection for KPIs Based on Correlation Analysis and HMM

KPIs (Key Performance Indicators) in distributed systems may involve a variety of anomalies, which will lead to system failure and huge losses. Detecting KPI anomalies in the system is very important. This paper presents a time series anomaly detection method based on correlation analysis and HMM. Correlation analysis is used to obtain the correlation between abnormal KPIs in the system, thereby reducing the false alarm rate of anomaly detection. The HMM (Hidden Markov Model) is used for anomaly detection by finding the close relationship between abnormal KPIs. In our correlation analysis of abnormal KPIs, firstly, the time series prediction model (1D-CNN-TCN) is proposed. The residual sequence is obtained by calculating the residual between the predicted value and the actual value. The residual sequence can highlight the abnormal segment in each data point and improve the accuracy of anomaly screening. According to the obtained residual sequence, these abnormal KPIs are preliminarily screened out from the historical data. Next, KPI correlation analysis is performed, and the correlation score is obtained by adding a sliding window onto the obtained anomaly index residual sequence. The correlation analysis based on the residual sequence can eliminate the interference of the original data fluctuation itself. Then, a correlation matrix of abnormal KPIs is constructed using the obtained correlation scores. In anomaly detection, the constructed correlation matrix is processed to obtain the adaptive parameters of the HMM model, and the trained HMM is used to quickly discover the abnormal KPI that may cause a KPI anomaly. Experiments on public data sets show that the method obtains good results.

Instability of the pseudoautosomal boundary in house mice

Faithful segregation of homologous chromosomes at meiosis requires pairing and recombination. In taxa with dimorphic sex chromosomes, pairing between them in the heterogametic sex is limited to a narrow interval of residual sequence homology known as the pseudoautosomal region (PAR). Failure to form the obligate crossover in the PAR is associated with male infertility in house mice (Mus musculus) and humans. Yet despite this apparent functional constraint, the boundary and organization of the PAR is highly variable in mammals, and even between subspecies of mice. Here we estimate the genetic map in a previously-documented expansion of the PAR in the Mus musculus castaneus subspecies and show that the local recombination rate is 100-fold higher than the autosomal background. We identify an independent shift in the PAR boundary in the Mus musculus musculus subspecies and show that it involves a complex rearrangement but still recombines in heterozygous males. Finally, we demonstrate pervasive copy-number variation at the PAR boundary in wild populations of M. m. domesticus, M. m. musculus and M. m. castaneus. Our results suggest that the intensity of recombination activity in the PAR, coupled with relatively weak constraints on its sequence, permit the generation and maintenance in the population of unusual levels of polymorphism of unknown functional significance.

Combination forecast model for concrete dam displacement considering residual correction

In conventional dam displacement monitoring models, forecast precision is below the standard, the fitting residual sequence contains chaotic components, and information mining of dam prototype observation data is limited. In consideration of the chaotic characteristics of the fitting residual sequence in regression model, the multi-scale wavelet analysis is used to decompose and reconstruct the residual sequence in this study; back propagation neural network and autoregressive integrated moving average model are used to forecast the reconstructed residual sequence by identifying the high-frequency and low-frequency characteristics of signals. By superimposing the residual forecast value with the forecast value of regression model, the combination forecast model for concrete dam displacement considering residual correction is proposed. Examples show that, compared with conventional models, the proposed combination model is better in fitting precision and convergence speed. Forecast capability is significantly improved for dam displacement forecast when effective components contained in residual sequence are considered. A new method of displacement forecast for high slope and other hydraulic structures is presented.

Modeling of the Trajectory of the Level Fluctuations in Lake Naroch

As a result of the conducted investigations of the level fluctuations in lake naroch the initial data are divided into 3 components: a polynomial regression that makes it possible to find out an independent on time law of trajectory, a periodic component of sinusoidal type and a residual sequence of independent random quantities. Modeling of the trajectory fluctuations is based on the deterministic part, consisting of the regression of the 8th order and periodic component, and also the random part, consisting of independent equally distributed quantities. Using this model it can be modeled the trajectory of the level fluctuations in lake naroch. The modeled trajectory by 200 years long demonstrated the probability of exceeding of maximum annual level, equal 1 per cent.

Combined Forecast of Calibration Interval Based on Linear Trend Model and Neural Network

For realizing the dynamic optimization of measuring instrument calibration interval, predicting the history calibration data by modeling. First the improved moving average method is used to modeling and to predict the development trend of parameters. On the basis of this, BP network is used to compensate the predicted residual sequence, so as to get more accurate forecasts. Then improved MA - BP prediction model is given to optimize the calibration interval dynamically. The model is verified through experiments. The results show that the model has higher prediction precision and better universality.

Time Series Analysis of Pulmonary Tuberculosis Incidence: Forecasting by Applying the Time Series Model

The main objective of this study is to identify the stochastic autoregressive integrated moving average (ARIMA) model to predict the pulmonary tuberculosis incidence in Qianan. Considering the Box-Jenkins modeling approach, the incidence of pulmonary tuberculosis was collected monthly from 2004 to 2010. The model ARIMA(0,1,1)12 was established finally and the residual sequence was a white noise sequence. Then, this model was used for calculating dengue incidence for the last 6 observations compared with observed data, and performed to predict the monthly incidence in 2011. It is necessary and practical to apply the approach of ARIMA model in fitting time series to predict pulmonary tuberculosis within a short lead time.