141 results about "Time series dataset" patented technology

Systems and methods for time series data error detection, correction, and transformation may detect gaps and anomalies in time series data, such as from a meter device, and may correct the gaps and adjust the anomalies prior to long-term record storage. Data forecasting may be used to correct the errors in the time series data. The error corrected data may be regarded as an actual set of time series data and become a base data set against which additional heuristic projections are generated. In addition, the time series data may be transformed into any number of physical and virtual device hierarchies that represent the underlying data source configurations, and may then be stored in an analytical database for further analysis. The hierarchies may be irregular and may change over time.

The invention discloses a time series data storage method and apparatus, a time series data query method, apparatus and system, and an electronic device. The time series data storage method comprises the steps of acquiring to-be-stored time series data as original time series data, wherein attributes of the original time series data include a data identifier attribute, a time attribute and a data value attribute; performing statistics processing on the original time series data with the same data identifiers and the time in a preset time interval according to at least one preset time interval, forming a time series data statistics value of at least one time granularity based on the same data identifiers, and storing the original time series data and the time series data statistics values of all the time granularities in a distributed database. By adopting the method, the data storage amount can be effectively reduced, so that the effects of saving storage space and increasing data query speed are achieved; and the time series data is stored in the distributed database, so that the extensibility of the storage space can be effectively improved.

The invention provides a seed maize field identification method and system based on multi-source and multi-temporal high resolution remote sensing data. The method comprises the steps of 1) getting the image of No. 1 multi-temporal high resolution Wild Field View (WFV) for the monitored maize field in maize growth season and the No. 2 high resolution panchromatic band image for key growth period; pre-processing the image of No. 1 multi-temporal high resolution Wild Field View (WFV) and getting xxx of No. 1 multi-temporal high resolution Wild Field View normalized difference vegetation index WFV NDVI Time Series Dataset and well-aligned High-Score 2# Panchromatic Band; S3: Application of Object-oriented Classification Method for Processing of WFV NDVI timing dataset of High-Score 1# in Maize Growing Season, to identify the cornfield block in the said monitoring area according to the phenological differences between crops; S4: to identify the seed maize field in the monitoring area based on the block acquired by S3 and according to the difference in spectrum and texture information between the seed maize field and the growing maize field on High-Score 2 panchromatic wave band. The present invention provides an accurate, economic and objective method for remote sensing and monitoring of seed maize breeding.

The invention discloses a runoff prediction method based on an attention mechanism and LSTM by using a deep learning algorithm. The method comprises the following steps: firstly, collecting influence characteristics related to runoff in a drainage basin, then constructing a time sequence data set corresponding to the characteristics and runoff, obtaining a runoff prediction model based on an attention mechanism and LSTM through training, and predicting the subsequent runoff according to the obtained runoff prediction model. Meanwhile, some short-term important features are ignored when the LSTM memorizes a long-term sequence mode, so that an attention mechanism is added, key elements in a runoff sequence are selectively concerned, the capability of capturing effective features by the LSTM is improved, and the prediction precision is relatively high. In addition, a deep learning method driven by data is used, dependence on a hydrological and physical mechanism in a drainage basin is reduced, and the application range of the model is effectively expanded.

A computer-implemented method of generating, from time-series data, a time-series of data sets for predictive analysis, comprises dividing the time-series data into evenly-sized overlapping segments of data, generating an image representing data for each segment, using the time-series data to determine a trend associated with each image, and storing each of the generated images and its associated trend as a data set. In some embodiments of the method the image from each stored data set is transformed into numerical vectors through a feature extraction process using a pre-trained convolutional neural network. The numerical vectors are stored in association with the data set, and the data sets and associated numerical vectors are used to predict the trend for a new time-series image which has been generated from any time-series data.

A method detects anomalies in time series data, wherein the time series data is multivariate, by partitioning time series training data into partitions. A representation for each partition in each time window is determined to form a model of the time series training data, wherein the model includes representations of distributions of the time series training data. The representations obtained from partitions of time series test data are compared to the model to obtain anomaly scores.

The anomaly diagnosis system includes the state measure calculator acquiring sensor data from sensors in a machine facility as time series data; an approximation formula calculator calculating a state measure being an index indicating a state of the machine facility, such as anomaly and a performance by a statistical method in which the time series data is used as learned data; and a state measure estimating unit estimating the state measures until future time using the approximation formula. Whenever the latest time series data is acquired, the reference period in which the time series data corresponding to the state measure referred to calculate the approximation formula by the reference period setting unit, is successively extended by addition of time when the latest time series data is acquired. The approximation formula calculator calculates the approximation formula using the state measure of the time series data acquired in the reference period.

A method and apparatus for storing time series data, in which a data structure is defined comprising a plurality of consecutive slices, each slice corresponding to a time period, time series data is stored in the data structure as runs of time series data entries each having a fixed, predetermined size and being associated with a time, each run of time series data entries comprising a number of consecutive time series data values, wherein each run of time series data is stored in a slice corresponding to the time period of the times associated with the time series data entries of the run, and each run of time series data is associated with metadata identifying the duration of the run and an offset of a start time of the run from a start time of the slice in which the run is stored.

The information use frequency prediction device is provided with a comparison section for shifting a second time series data related to the use frequency of a first keyword, sequentially on the time base in a unit of predetermined time, based on a first time series data related to the use frequency of a second keyword input while searching over the Internet, and calculating a correlation coefficient between the first time series data and the second time series data, for each unit of predetermined time. Moreover, a prediction section specifies a pair of the first time series data and the second time series data, corresponding to the correlation coefficient having the highest value, of a plurality of correlation coefficients calculated by the comparison section and for predicting the use frequency of the first time series data constituting the pair, based on the second time series data.

Provided is a method and system for detecting changes in models applied to analyze time series data. A method includes receiving at a processor a data stream transmitted from a sensor configured to measure an operating parameter of a component being monitored, wherein the data stream comprises at least time series data. The method also includes analyzing the data stream to identify a sequence of interest in the time series data, searching metadata stored separately for an appropriate model to apply to the time series data, and selecting the appropriate model. Information about the selected appropriate model is carried forward with the time series data.

A computer-implemented method, system, and a computer readable article of manufacture identify local patterns in at least one time series data stream. A data stream is received that comprises at least one set of time series data. The at least one set of time series data is formed into a set of multiple ordered levels of time series data. Multiple ordered levels of hierarchical approximation functions are generated directly from the multiple ordered levels of time series data. A set of approximating functions are created for each level. A current window with a current window length is selected from a set of varying window lengths. The set of approximating functions created at one level in the multiple ordered levels is passed to a subsequent level as a set of time series data. The multiple ordered levels of hierarchical approximation functions are stored into memory after being generated.

The invention relates to a multi-scale region flood disaster risk remote sensing assessment method and system. The method comprises the steps of cooperatively extracting feature information of a flood disaster water body by utilizing multi-source multi-platform remote sensing data to form space distribution data of the water body; by analyzing critical features of the flood disaster water body, and in combination with environmental factors in a historical flood disaster occurrence and development process, establishing a flood disaster long-time series data set based on the critical features of the water body; and then according to a reasoning algorithm, performing multi-scale region flood disaster risk level assessment from a disaster flooding range and a disaster loss degree to generate a space distribution result of flood disaster risk assessment. The method has a very important effect for enhancing disaster early warning, avoiding risk and the like, can reduce the loss caused by flood disasters maximally, and is an important scientific basis of performing disaster risk management and decision-making.

The invention provides a time series data storing and querying method and a storing and processing platform, and relates to the technical field of databases. The time series data storing and queryingmethod and the storing and processing platform are applied to a database, and the database comprises an indexing file and a data file. The time series data storing method comprises the following steps: receiving time series data to be stored and compressing the time series data to obtain compressed time series data; writing the compressed time series data to the tail of a data file; determining the offset, timestamp and data length of the compressed time series data; and when the compressed time series data are written successfully, constructing an index entry according to the timestamp, offset and data length, and writing the index entry to the tail of an index file. An embodiment of the invention aims to provide the time series data storing and querying method and the storing and processing platform so as to rapidly store, query and process a bulk of time series data, and time series database service which is high in disaster tolerance ability, convenient to use and efficient is provided.

One embodiment of the invention disclosed herein provides techniques for detecting and remediating an outlier server in a distributed computer system. A control server retrieves a group of time-series data sets associated with a first time period, where each time-series data set represents a performance metric for a different server in a group of servers. The control server generates a cluster that includes two or more of the time-series data sets, where the performance metric for each server that is associated with one of the time-series data sets in the cluster is within a threshold distance from the performance metric for the servers that are associated with the other time-series data sets in the cluster. The control server determines that a particular time-series data set corresponds to a server included in the group of servers and is not included in the cluster, and marks the server as an outlier server.

The invention discloses a method for classifying multi-variable time sequences based on a deep long / short-term memory neural network. The method includes the steps of by selecting long / short-term memory neuron structures as hidden layer neuron structures of a recurrent neural network, overlapping long / short-term memory neurons, and designing a deep layer recurrent neural network classification framework, thereby achieving the aim of improving the classification accuracy of multi-variable time sequence data. It is found through experimental comparison that accuracy is higher compared with an existing classification model and universality is achieved on time sequence data set classification tasks in multiple fields.

A method and system for analyzing time series data. In an embodiment, a loop is executed and terminated upon a specified maximum number of iterations of the loop being performed or upon a difference between scores in successive iterations of the loop not being greater than a specified tolerance, wherein the score in each iteration is calculated as function of an absolute value of a difference between respective cumulative probability values of first and second cumulative probability distributions which are generated from respectively first and second time series data sets. In an embodiment, time series data is processed in a sequence of time periods, wherein a combined cumulative probability distribution is generated in each time period by combining a cumulative probability distribution of new time series data with previously combined cumulative probability distribution data according to a ratio of the number of new to previous observed values.

An unnecessary load is prevented from being applied to the system even in cases where the time series data required for the time series data analysis processing has not been collected. A time series data processing apparatus assigns an arrival time, which is a time that the time series data arrived, to the time series data sent from the data source, determines whether the requested time series data has arrived, and predicts the arrival time of the time series data, which was determined by the data arrival determination unit as not yet arrived, based on the arrival time assigned to each of the time series data.

In some aspects, computer-implemented methods of identifying patterns in time-series social-media data. In an embodiment, the method includes applying a deep-learning methodology to the time-series social-media data at a plurality of temporal resolutions to identify patterns that may exist at and across ones of the temporal resolutions. A particular deep-learning methodology that can be used is a recursive convolutional Bayesian model (RCBM) utilizing a special convolutional operator. In some aspects, computer-implemented methods of engineering outcome-dynamics of a dynamic system. In an embodiment, the method includes training a generative model using one or more sets of time-series data and solving an optimization problem composed of a likelihood function of the generative model and a score function reflecting a utility of the dynamic system. A result of the solution is an influence indicator corresponding to intervention dynamics that can be applied to the dynamic system to influence outcome dynamics of the system.

The invention discloses a time series forecasting method and system based on SVR (Support Vector Regression). The time series forecasting method based on SVR comprises the following steps of: selecting historical data from an existing time series data set and obtaining a plurality of training data sets; determining regular parameters and Gaussian nuclear parameters of SVR modules to be constructed, and constructing an SVR module corresponding to each training data set; selecting T historical data between a moment of t-T+1 to a current moment t; and under the condition that a first difference of a moment to be forecasted and the current moment is less than or equal to the number of SVR modules, selecting the SVR module corresponding to the first difference, and obtaining a forecast value of the moment to be forecasted from the T historic data by directly utilizing the SVR module. Compared with a method for obtaining the forecast value of the moment to be forecasted through multi-step forecasting in the prior art, in the method for obtaining the forecast value through a one-step forecasting disclosed by the invention, the accumulation of forecast errors is reduced, and thus the accuracy of obtaining the forecast value is improved.

A data analysis computer system is provided that receives a timeseries dataset and generates implied data from the dataset. The dataset is further vectorized to reduce the dimensionality of the data. Users provide input to identify windows of data that either positively or negatively correlate to instances of a given type of occurrence within the data. The user defined windows are converted to fixed sized windows and a machine learning algorithm constructs a model from the data. The model is used to predict instances of the given type of occurrence in newly received data. Validation of the predications may be performed.

A technique for effective classification of time series data using a rule-based wavelet decomposition approach. This method is effective in classification of a wide variety of time series data sets. The process uses a combination of wavelet decomposition, discretization and rule generation of training time series data to classify various instances of test time series data. The wavelet decomposition can effectively explore the data at varying levels of granularity to classify instances of the test time series data.

The invention discloses a remote sensing time sequence analysis-based abandoned land information extraction method. The method comprises the following steps of obtaining remote sensing image information; constructing an NDVI time sequence data set according to the acquired remote sensing image information, and constructing an NDVI continuous time sequence growth change curve of vegetation in the cultivated land range; according to the NDVI continuous time sequence growth change curve of the vegetation in the cultivated land range, extracting the parameter CV of the fluctuation intensity of theintra-year change curve of the NDVI; and extracting spatial distribution information of the abandoned land from the cultivated land range according to the CV value. According to the method, the phenological characteristic difference analysis of the abandoned land and the cultivated land can be carried out by utilizing remote sensing intra-year and inter-year continuous time sequence images, and the growth characteristics of the abandoned land are represented by normalizing the CV of the intra-year change curve fluctuation intensity of the difference vegetation index NDVI, so that the distribution information of the abandoned land in the year is obtained; and aggregating the multi-year image time sequence set to extract multi-year abandoned land spatial distribution information, thereby obtaining multi-year abandoned land evolution information.

Disclosed is an apparatus that generates automatically a characteristic pattern in time series data by clustering a plurality of time series subsequences generated from the time series data. The apparatus includes a time series subsequence generation unit that generates a plurality of time series subsequences from the time series data, a phase alignment unit that aligns a phase of the generated time series subsequence, a clustering unit that performs clustering of a plurality of the time series subsequences, each having a phase aligned, a storage apparatus that stores the pattern obtained by the clustering, and an output apparatus that outputs the stored pattern.

The invention belongs to the technical field of chaotic systems. The invention discloses a chaotic time sequence prediction method based on attention mechanism deep learning. The method comprises thefollowing steps: (1) constructing a chaotic time sequence data set; (2) carrying out phase space reconstruction on a chaotic time sequence; (3) training chaotic time sequence data by using an LSTM neural network model; (4) constructing a prediction-based attention mechanism model, (5) constructing an offline training model, and (6) carrying out online prediction. The chaotic time sequence prediction method based on attention mechanism deep learning is clear in model structure, has a reference value, and can be applied to the aspects of financial market prediction or energy prediction and the like of a chaotic system.

The invention relates to a vegetation index time series data reconstruction method based on wavelet multi-scale decomposition. The method is characterized by comprising the following steps: performing wavelet transformation according to the vegetation index time series data to respectively sequentially decompose the original time series data of vegetation indexes and climatic factors into corresponding half moth, month, double month, season, half a year, and time series data based on interannual scales; further selecting the climatic factor time series data under the corresponding scale according to the features of the vegetation index time series data in each scale; selecting a proper model; reconstructing time series data in different scales; finally combining the vegetation index time series data in all scales to realize the reconstruction of the original vegetation index time series data. The method has the characteristics of being high in precision and wide in applicable scope.

The invention discloses a taxi track data-based travel space-time mode identification method and system. The method comprises the following steps of carrying out the collective counting processing onthe taxi OD data points by taking a road intersection point as a center and a circular area with a set length as a radius as a research unit; clustering the taxi travel time sequence data set by adopting a density-based clustering algorithm, and adjusting the clustering parameters according to an evaluation index to obtain an optimal clustering result, wherein a product obtained by adding a time window constraint condition to the DTW and adjusting a metric value of a first-order time correlation coefficient adaptive dissimilarity index through an adjustment function is used as a final time sequence similarity measurement function; and superposing the spatial distribution of the clustering result by combining the research area base map to obtain a travel mode spatial distribution map. According to the present invention, the results of the research unit division method and the time sequence clustering method are more consistent with the general cognition of the taxi mode, and better applicability is shown for the taxi data.

Systems and methods are provided for monitoring a subject, and particularly, for inferring an underlying brain state present in absence of current conditions. In some aspects, a method for monitoring the subject is provided including steps of receiving physiological feedback from at least one sensor configured to acquire physiological information from locations associated with a subject's brain, assembling a set of time-series data using the received physiological feedback, and identifying portions of the set of time-series data that indicate a burst suppression state. The method also includes identifying a burst characteristic profile associated with a burst pattern determined from the identified portions, and comparing the burst characteristic against a reference set of burst profiles. The method further includes determining, based on the comparison, a likelihood of a brain state of the subject underlying the burst suppression state, and generating a report indicative of the likelihood of the determined brain state.

A detecting apparatus includes the following units. A pressure detecting unit detects in time series higher and lower static pressures and a differential pressure. Fluctuation calculating units calculate time series data sets of a differential pressure fluctuation, a higher static pressure fluctuation, and a lower static pressure fluctuation. Square sum calculating units calculate the square sum of the time series data sets of the differential pressure fluctuation, the square sum of the time series data sets of the higher static pressure fluctuation, and the square sum of the time series data sets of the lower static pressure fluctuation. A correlation coefficient calculating unit calculates first and second correlation coefficients. An intermediate variable calculating unit calculates an intermediate variable. A blockage rate calculating unit calculates a blockage rate. A determination unit determines a blockage of higher and lower pressure impulse lines with reference to the blockage rate.

An aspect of synchronization of time between simulation models includes receiving a specification request for generating a set of target time-series data using a set of source time-series data, obtaining specification information relating to the set of source time-series data, obtaining specification information relating to the set of target time-series data and obtaining the source time-series data; comparing the source specification information and the target specification information to determine if the set of source time-series data are time-aligned with the set of target time-series data, and converting the set of source time-series data to the set of target time-series data upon determination that time alignment is needed.

The invention discloses a water quality automatic online station high-frequency continuous observation data quality control method which comprises the following steps: acquiring first water quality observation sequence data through an acquisition module to obtain a time sequence data set; carrying out differential operation on the time sequence data set, and carrying out stability detection to obtain stable differential time sequence data; determining a statistical window value and a sliding step size value range for the data, and carrying out sliding detection according to the statistical window value and the sliding step size; carrying out abnormal value detection through a plurality of detection methods to obtain corresponding abnormal values, and combining the abnormal values to obtaina comprehensive abnormal value; reconstructing the abnormal value sequence points to obtain second water quality observation sequence data to realize quality control of the water quality observationdata; according to the method, abnormal value detection is carried out on the water quality, reconstruction is carried out, data difference processing, stability detection and statistical window and sliding detection are combined, quality control over automatic online station data of the water quality is achieved, and the method has industry popularization value and application prospects.