50 results about "Time series similarity" patented technology

The present invention discloses an embedded index based hydrological time series similarity searching method. The method is carried out by the following steps: calculating corresponding embedded index vectors of each position in an original time series in an off-line preparation stage, wherein in the off-line preparation stage, the hydrological time series flood peak segmentation, serial clustering, initial reference sequence set generation, reference set training and time series embedded index calculation are realized; and calculating the index vectors by using a query sequence and a reference set sequence in an on-line searching stage, performing searching in an embedded index Euclidean vector space of the original series to find a relatively similar point as a candidate point set, refining candidate points, and then performing original DTW (dynamic time warping) measurement to find a final similar sequence. According to the embedded index based hydrological time series similarity searching method, the similarity searching is mapped to the Euclidean vector space to perform searching, thereby greatly improving the search efficiency.

The invention discloses a multi-measurement time series similarity analysis method applicable to k-neighbor inquires of a time series. A multi-single-similarity-measurement method is chosen according to the analysis requirement, each single similarity measurement is used to analyze and inquire an m-neighbor sequence or subsequence of the sequence, pruning the m-neighbor sequence or subsequence under each similarity measurement to obtain a candidate similarity sequence or subsequence, and combining the candidate similarity sequence or subsequence by using a multiple-classifier combination method with advantage weight to obtain the k-neighbor sequence of the inquired sequence. Compared with the single similarity measurement, the similarity analysis of combined multiple measurements can obtain a more comprehensive analysis result. The multiple-classifier combination method with advantage weight regulates the ranking score according to the difference of the similarity distance between the adjacent candidate similarity sequence or subsequence and the inquired sequence while using a BORDA counting method for reference, so as to reflect the specific difference of similarity of the candidate similarity sequence or subsequence.

A location estimation method using label propagation. The achieved location estimation method is robust to variations in radio signal strengths and is highly accurate by using the q-norm (0<q<1), especially, for calculating the similarities among radio signal strength vectors. The accuracy in location estimation is further improved by putting more importance on the time-series similarities. Specifically, the time-series similarity is calculated by using time-series values indicating the temporal order of radio signal strengths during the measurement. If the time-series similarity is larger than the similarity between the radio signal strength vectors, the time-series similarity is preferentially used. The exponential attenuation function can also be used for calculating the similarities, instead of the q norm (0<q<1).

The invention discloses a time-series similarity measurement method based on segmented statistical approximate representation. The method comprises the steps of feature extraction and dynamic pattern matching. First, a time series is segmented into sub series, the various statistical features of the sub series are sequentially extracted, and local pattern feature vectors are constructed; then the distance between the local pattern feature vectors is calculated by the weighted Euclidean distance, local pattern matching is achieved, the matched local pattern is used as the sub program of a dynamic programming algorithm, and global pattern matching is achieved. The method is superior to other measurement methods by a large degree on the aspects of measurement precision and calculation efficiency, and plays an important role in daily activities and industrial production of people, for example, financial transactions, traffic control, air quality and temperature monitoring, industrial flow monitoring, medical diagnosis and the like. Large scale sampling data or high-speed dynamic data flow is subjected to similarity-based search, classification, clustering, prediction, abnormal detection, on-line pattern recognition and the like.

The invention discloses a financial time series similarity query method based on K-chart expression. The method comprises the following steps of feature extraction, index construction and query processing. The method comprises the following concrete steps of firstly, extracting basic mode and classic mode features for a financial time series based on K-chart expression, and respectively translating the basic mode and classic mode features into a basic string and a classic string; secondly, respectively constructing reverse indexes on the basic string and the classic string; for each query sequence, after the basic mode and classic mode features are extracted through the same way, respectively querying the two constructed reverse indexes to acquire two candidate sets, and then carrying out intersection operation to obtain a final candidate set; obtaining a final query result through follow-up processing. The financial time series similarity query method based on K-chart expression can effectively realize nearest neighbor query, has higher measurement precision and query efficiency, has favorable extensibility for time series length, nearest neighbor query scale and data set scale, and can play a significant role in the widened electronic finance trade market.

The invention discloses time sequence similarity value acquisition method and system. The method and the system are both applied to a time sequence set. The time sequence set includes at least two time sequences, optional one time sequence is used as a target time sequence which is split to obtain at least two time subsequences, the time subsequences are distributed to different nodes in different server clusters respectively, other time sequences are not split and can be distributed to different nodes in the different server clusters, a bent path of each time subsequence and bent paths of other time sequences in the time sequence set are acquired respectively, the values of similarity of the target time sequence to the other time sequences in the time sequence set are determined according to the bent paths, and accordingly similarity values of the time sequences can be acquired simultaneously and concurrently. Therefore, operating efficiency is improved, and the method and the system are especially applicable to acquisition of similarity values of ultra-long time sequences.

The invention relates to a vegetation change occurrence time detection method based on time series similarity, comprising: first, establishing multi-year time-space continuous vegetation index time series data of a research area, calculating JM distance between vegetation index time series curves of each past year and the initial year pixel by pixel, year by year, and generating a time series curve for the JM distance of each past year and the original year; fitting the time series curve for the JM distance of each past pear and the original year by using a logistic model, and acquiring time parameter from logistic model parameters so as to automatically extract vegetation change time. In the method, the JM distances of vegetation index time series curves between the past years and the initial year are used to indicate time series similarity, and vegetation change time is acquired from change law of yearly time series similarity. The method is effective in detecting changes of time series curves in terms of amplitude, frequency and the like, the complex step of decomposing original spectral index time series data is avoided, and the problem that it is difficult to extract indexes directly from original spectral index time series data to provide comprehensive characterization of vegetation changes is solved.

The invention discloses a short-term optical power prediction method based on time series similarity. First, photovoltaic power station meteorological data and equipment operation data which are acquired in real time are repaired, unreasonable values or missing data are filled, and then all photovoltaic arrays are divided into several categories through power data analysis. Power curves of each category of photovoltaic arrays under the similar meteorological condition at the same period of the last year are found out according to meteorological prediction data, and future power is predicted in a weighted average manner. By adoption of big data analysis, optical power prediction precision is improved by comprehensive utilization of historical data and meteorological prediction data, thereby facilitating stationary operation of the power grid.

The invention discloses a time series data graphics analysis method based on automatic coding technology with packet loss. The time series data graphics analysis method comprises the following steps: 1) data preprocessing: converting time series data into a specific image format; 2) pre-training: extracting the graphic features of a time series through the automatic coding technology with the packet loss; 3) classifier training: carrying out classifier training to coding machine weight and a training sample class identifier in a pre-training process; and 4) application: realizing the functions of similarity matching and classification of the time series by utilizing the trained classifier. A defect that a traditional time series analysis method is very sensitive to data change since the traditional time series analysis method pays attention to the data feature of the time series is overcome, and a visual processing method of the time series data by people is simulated. On an aspect of the similarity matching, the invention exhibits high accuracy and low time complexity. In classification, high classification precision is guaranteed, and the invention also exhibits good universality and robustness to different types of time series data.

The invention discloses a method for measuring the space-time multi-variant hydrological time series similarity. The method comprises firstly rasterizing the original rainfall data of a flood to generate a rainwater distribution matrix map in each hour; and then calculating the 2D-DTW distance between two rainfall distribution matrix sequences. The method includes the similarity calculation methodof the two rainfall distribution matrices and the similarity measurement method for the rainfall distribution matrix sequence. By using the obtained distances between a standard template rainfall distribution matrix sequence and a test template rainfall distribution matrix sequence, it is determined that which one or more test floods are most similar to the standard template flood hydrological process. The test flood data most similar to the template flood hydrological process can be output.

The invention is suitable for the performance data analyzing field and provides a performance data dependency analyzing method and a performance monitoring system. The method comprises: a step of collecting performance data of a plurality of indexes and storing the performance data to build a database; a step of extracting a performance data of one index in the database and building a time series according to a time collecting sequence; a step of extracting a graphic feature vector of the time series; a step of building a graphic feature index for the graphic feature vector; a step of building an object time series; a step of extracting a target image feature vector of the object time series; and a step of querying the graphic feature vector similar to the target graphic feature vector in the graphic feature index, and ranking and outputting the queried result. According to the invention, the method can realize big data analytics of the performance data via a time series similarity search technology, so that the variable characteristics on a timer shaft can be intuitively reflected and a manager can directly analyze the different factors of generating the performance problems via the queried result.

The invention discloses a trend segmentation similarity-based airport noise monitoring point exception identification method and belongs to the technical field of airport noise monitoring point exception analysis. The method comprises the steps of firstly obtaining noise monitoring data of monitoring points around an airport by utilizing a monitoring device; secondly preprocessing the monitoring data and creating a standard noise time series data set; thirdly performing dimension reduction expression on noise time series of the monitoring points by using a trend segmentation-based time series expression method; fourthly by utilizing a trend segmentation-based similarity measurement method, measuring noise time series similarity between the monitoring points, and establishing a similarity matrix; fifthly finding out first k monitoring points with relatively high similarity with each monitoring point, and creating a similar monitoring point set; and finally measuring the similarity between new noise time series of the monitoring points and new noise time series of associated monitoring points, and if the similarity is remarkably changed, determining that the monitoring points are exceptional. According to the method, the monitoring point exception can be accurately identified, so that the timeliness and validity of airport noise monitoring point maintenance are effectively improved.

The invention discloses a time series similarity searching method based on segment weight, which comprises the following steps: (1) segmenting a query sequence q by adopting an important turning pointof the time series; (2) establishing piecewise Euclidean distance with weights as similarity measure of time series; (3) performing k-nearest neighbor similarity querying on the q, searching for thek most similar sequences of the query sequence q; (4) ending the query if the user is satisfied with the query result obtained in the step (3); If the user is not satisfied with the query result obtained in the step (3), marking the result and entering the step (5); (5) allowing the system to update the weights of the segments by using the sequence of user tags and returning to step (2). The invention automatically updates the weights through user feedback, adaptively learns the attention degree of the user to different segments, can improve the accuracy of similarity measurement, and furtheroptimizes the search result.

The invention discloses a circuit breaker state evaluation method. The method comprises the steps: forming to-be-tested data for representing the characteristics of a circuit breaker through the characteristic time points of a circuit breaker closing current curve; obtaining the data of the circuit breaker with a fault and the type of the fault, and dividing the data into three fault data clusters; calculating the relative proximity of the to-be-tested data and the fault data clusters according to the big data clustering idea, and dividing device states into a healthy state, a latent fault state or a fault state; judging the fault type of the device with the fault based on the above, and calculating a health score of health equipment based on a fault type correlation weight, and obtainingthe predicted fault development time of a latent fault device through a time series similarity analysis method. The method can predict the time of transition of a state to the fault state more accurately, can achieve the finding of a hidden risk of the device before the fault occurs, and facilitates the early detection and resolution of equipment hazards.

The invention relates to a multivariate time series similarity measuring method oriented to the ocean field. The similarity measuring method comprises the following steps that S1, typhoon data is collected; S2, the typhoon data is preprocessed, wherein typhoon attribute screening and data supplementing are included; S3, the typhoon data is described, wherein movement direction showing and typhoon time series showing are included; S4, the typhoon data is subjected to similarity measurement, wherein typhoon attribute weight calculation, W-DTW distance calculation and W-DTW distance judgment are included ; S5, similar typhoon is output. The method has the advantages that whether two ocean time series with dynamics, spatiality, predictability and multiple attributes are similar or not is judged; the development trend of a current ocean event is judged based on an occurring ocean event; for ocean disasters, a convenient auxiliary decision can be provided for relevant departments, protection measures are taken, and economic losses and casualties caused by the ocean disasters are reduced.

The invention discloses a time-series similarity measurement method capable of adapting to missing data. According to the method, data pairs are extracted from two original time series in pairs and are divided into five types according to data missing conditions, and first-order similarity intervals are calculated respectively; intervals are extracted from the first-order similarity intervals in pairs, second-order similarity is figured out, and second-order similarity vector quantities are obtained; at last, the second-order similarity vector quantities are averaged to obtain final similarity of the two time series. The method can adapt to multiple scenes, is simple and does not have any requirement for data integrity.

The invention discloses an oil immersed transformer state evaluation method. The method includes constructing to-be-measured data to characterize the characteristics of an oil immersed transformer byusing the oil and gas data of the oil immersed transformer; obtaining fault oil immersed transformer data, and dividing six fault data clusters according to fault types; calculating the relative proximity of the to-be-measured data and the fault data clusters according to big data clustering thought; dividing an equipment state into a health state, a latent fault state or a fault state; judging the fault types of fault equipment on this basis, and calculating the health scores of the health equipment based on the fault type associating with weight; and obtaining the predicting fault development time of latent fault equipment through a time series similarity analysis method. Thus, the time inverting into the fault state can be accurately predicted, the hidden risks of the equipment can be found before fault occurs, and early detection and solution of equipment hidden danger can be realized.

A location estimation method using label propagation. The achieved location estimation method is robust to variations in radio signal strengths and is highly accurate by using the q-norm (0<q<1), especially, for calculating the similarities among radio signal strength vectors. The accuracy in location estimation is further improved by putting more importance on the time-series similarities. Specifically, the time-series similarity is calculated by using time-series values indicating the temporal order of radio signal strengths during the measurement. If the time-series similarity is larger than the similarity between the radio signal strength vectors, the time-series similarity is preferentially used. The exponential attenuation function can also be used for calculating the similarities, instead of the q norm (0<q<1).

The invention provides a harmonic source tracing method based on an algorithm with dynamic planning time series similarity, comprising: obtaining, from an electric energy quality monitoring system, harmonic voltage monitoring data in a period of time collected by an electric energy quality monitoring terminal installed on a certain bus; obtaining, from a power consumption information collecting system, power consumption active power data in the time period of all users supplied by the bus; processing the harmonic voltage monitoring data and the power consumption active power data by using a data standardization method; and solving the correlation between the power consumption active power series data of each user and the harmonic voltage series data of the common connection points by usingan algorithm that solves the time series data similarity based on the dynamic planning principle, and carrying out the harmonic source tracing. The method provided by the invention fully mines the value of smart meter data deployed on a large scale, and infers possible users that cause harmonic problems by data correlation analysis results of the power consumption condition and the harmonic condition of the users, thereby providing a basis for accurate harmonic responsibility division, future quality-based pricing, and precise control.

The invention discloses a time sequence similarity calculation method and system based on deep learning, and a medium. The time sequence similarity calculation method based on deep learning comprisesthe implementation steps of 1) obtaining time sequence data of two equal-length time periods; and 2) inputting the time series data of the two equal-length time periods into a pre-trained neural network model based on deep learning to obtain the similarity between the time series data of the two equal-length time periods. According to the invention, the advantages of various traditional measurement methods are integrated; the effect is better than that of each traditional measurement method in the aspect of time sequence similarity measurement; according to different requirements and differentdata sets, the same method can be used to learn a data similarity measurement method suitable for different fields, and for different problems, a similarity calculation method does not need to consider inherent characteristics of data.

The invention discloses a time series data graphics analysis method based on automatic coding technology with packet loss. The time series data graphics analysis method comprises the following steps: 1) data preprocessing: converting time series data into a specific image format; 2) pre-training: extracting the graphic features of a time series through the automatic coding technology with the packet loss; 3) classifier training: carrying out classifier training to coding machine weight and a training sample class identifier in a pre-training process; and 4) application: realizing the functions of similarity matching and classification of the time series by utilizing the trained classifier. A defect that a traditional time series analysis method is very sensitive to data change since the traditional time series analysis method pays attention to the data feature of the time series is overcome, and a visual processing method of the time series data by people is simulated. On an aspect of the similarity matching, the invention exhibits high accuracy and low time complexity. In classification, high classification precision is guaranteed, and the invention also exhibits good universality and robustness to different types of time series data.

The invention discloses a principal-component analysis-based construction method of a multivariate hydrological time series matching model. Combined-model construction of multivariate hydrological time series similarity matching is carried out on the basis of principal-component analysis (PCA) and dynamic-time-warping (DTW) methods. The construction method includes: firstly, carrying out isomorphic processing on original data, wherein a Z-score standardization method is adopted; then carrying out piecewise aggregate approximation (PAA) processing on the processed data, and carrying out PCA processing on the data after PAA processing, wherein dimension reduction of the data in both a time dimension and a variable dimension is realized after the two times of processing; and finally, using aweighted DTW method for similarity matching, and obtaining a time series, which is most similar to a given time series, by matching. The construction method improves accuracy and time efficiency of similarity matching, provides services for hydrological forecasting and hydrological data analysis, and has higher application values for needs of water conservancy informatization and water conservancymodernization.

The present invention belongs to the field of electric power technology, in particular, a harmonic source tracing method based on a dynamic programming time series similarity algorithm. Aiming at the problem of poor accuracy, the following scheme is proposed, including the following steps: S1: extract a certain time in a certain day and a method in the system All harmonic data at this time every day; S2: Extract the same time period of the day and all unit time power consumption in the system at this time every day; S3: Process all harmonic data and power consumption per unit time at a certain time on a certain day to form Change curve; S4: weighted average of all harmonic data and power consumption per unit time at a certain time each day to form a change curve; S5: respectively subtract the data of a certain day from the weighted average of harmonic data and power consumption per unit time , get the absolute value and then calculate the average value; S6: use the sequence similarity algorithm to calculate and solve, and trace the source of harmonics according to the calculated value. The present invention monitors and judges the harmonic data more comprehensively and accurately, and the judgment method is simple and effective.

The invention discloses a method for calculating time sequence similarity in the technical field of computer information technology processing. The method includes the steps of dividing a time sequence S1 to be compared and a time sequence S2 to be compared into time subsequences S1 (i) and time subsequences S2 (i) respectively with the same mode, setting the weight (wi) of each time subsequence S1 (i) and the weight (wi) of each time subsequence S2 (i), calculating distances between the corresponding time subsequences, and calculating the similarity of the time sequence S1 and the time sequence S2 according to the distances between the corresponding time subsequences and the weights of the time subsequences. The method can better reflect the similarity of shapes of the time sequences and is low in complexity and quick in judgment.

The invention relates to a multivariate time series similarity measuring method oriented to the ocean field. The similarity measuring method comprises the following steps that S1, typhoon data is collected; S2, the typhoon data is preprocessed, wherein typhoon attribute screening and data supplementing are included; S3, the typhoon data is described, wherein movement direction showing and typhoon time series showing are included; S4, the typhoon data is subjected to similarity measurement, wherein typhoon attribute weight calculation, W-DTW distance calculation and W-DTW distance judgment are included ; S5, similar typhoon is output. The method has the advantages that whether two ocean time series with dynamics, spatiality, predictability and multiple attributes are similar or not is judged; the development trend of a current ocean event is judged based on an occurring ocean event; for ocean disasters, a convenient auxiliary decision can be provided for relevant departments, protection measures are taken, and economic losses and casualties caused by the ocean disasters are reduced.

The invention discloses a time sequence similarity query method based on inverted indexes. The method comprises steps of index building and query processing, firstly, a real value type time sequence is converted into a discrete character string through symbol aggregate approximation representation, then a characteristic subsequence is extracted, codes are stored by vector approximation files, the subsequence is converted into word insertion inverted indexes with two types of granularity, and multi-granularity time sequence inverted indexes are built. According to the time sequence similarity query method based on the inverted indexes, an efficient two-stage filtration query method is designed for the indexes, k nearest neighbor similarity query can be realized, on the premise that a higher precision ratio is guaranteed, query time overhead is shorter, and good extendibility for the time sequence length, k nearest neighbor similarity query scale and data set scale is achieved; and the method can play an important role in daily activities and industrial production such as real-time query of stock volatility, on-line pattern recognition of sensor data flow and the like.

The invention discloses time sequence similarity value acquisition method and system. The method and the system are both applied to a time sequence set. The time sequence set includes at least two time sequences, optional one time sequence is used as a target time sequence which is split to obtain at least two time subsequences, the time subsequences are distributed to different nodes in different server clusters respectively, other time sequences are not split and can be distributed to different nodes in the different server clusters, a bent path of each time subsequence and bent paths of other time sequences in the time sequence set are acquired respectively, the values of similarity of the target time sequence to the other time sequences in the time sequence set are determined according to the bent paths, and accordingly similarity values of the time sequences can be acquired simultaneously and concurrently. Therefore, operating efficiency is improved, and the method and the system are especially applicable to acquisition of similarity values of ultra-long time sequences.

The invention relates to a vegetation change occurrence time detection method based on time series similarity, comprising: first, establishing multi-year time-space continuous vegetation index time series data of a research area, calculating JM distance between vegetation index time series curves of each past year and the initial year pixel by pixel, year by year, and generating a time series curve for the JM distance of each past year and the original year; fitting the time series curve for the JM distance of each past pear and the original year by using a logistic model, and acquiring time parameter from logistic model parameters so as to automatically extract vegetation change time. In the method, the JM distances of vegetation index time series curves between the past years and the initial year are used to indicate time series similarity, and vegetation change time is acquired from change law of yearly time series similarity. The method is effective in detecting changes of time series curves in terms of amplitude, frequency and the like, the complex step of decomposing original spectral index time series data is avoided, and the problem that it is difficult to extract indexes directly from original spectral index time series data to provide comprehensive characterization of vegetation changes is solved.