171 results about "Spatiotemporal correlation" patented technology

The invention discloses a traffic state prediction method for the urban road network based on key road sections, which is characterized by comprising the steps of first, carrying out data preprocessing; second, establishing a spatial weight matrix of the road network; third, establishing a time correlation matrix; fourth, recognizing key road sections by using a time-space correlation matrix; andfifth, establishing a deep convolution neural network, predicting the state of the road network in the future, and carrying out evaluation on a prediction model. The traffic state prediction method predicts the urban traffic flow state from a level of the wide-range road network, thereby being conducive to guiding the traffic flow from a macroscopic perspective, and fully exploring time-space correlation characteristics of the traffic flow. The key road sections in the road network are recognized, so that the training time of the model can be greatly reduced compared with a method of taking historical states of all road sections as input data, and the prediction efficiency is improved; and the convolution neural network is adopted to serve as the prediction model, and the prediction resultis also more accurate.

The invention discloses an online car-hailing service demand prediction method based on a deep neural network. The method comprises the following steps: S1, dividing a region of a city road network, and calculating and counting the network car-hailing demand of each region; S2, designing a prediction model of the demand of the network car-hailing: based on the deep neural network, learning and training the spatio-temporal characteristics of the demand of the network car-hailing, combining the influence factors of the weather factors on the willingness of the network car-hailing users, and predicting the regional demand of the next period of time; S3, according to the historical data, training the prediction model of regional car call demand, and then predicting the demand of network car call in each region by using the trained model. By combining external factors and spatio-temporal correlation, this method uses depth neural network to express the complex non-linear spatio-temporal correlation characteristics of the demand for network car-hailing, and can achieve high prediction accuracy.

The invention discloses an alarm correlation analysis method in a communication network, targeting topology research in a tree-hierarchy structure network. The method comprises the following steps: in accordance with the space-time relativity of malfunctions occurring to network nodes, defining space-time relativity of upper layer network nodes in the tree-hierarchy structure network; based on the space-time relativity of the upper layer network nodes, clustering upper layer nodes in the tree-hierarch network; based on clustering result, dividing the total alarm database into a plurality of sub alarm databases; based on the attributes of alarm items, such as alarm occurrence frequency, alarm priority and alarm failure type, determining weight of each alarm item; utilizing the weighted Apriori correlation rules algorithm, conducting correlation rules mining on each sub alarm database. The method of the invention aims at addressing the problems of alarm relativity analysis of tree-hierarchy structure, and can effectively mine alarm correlation rules of interest among a large amount of alarm information.

A method for assisting the navigation of an aircraft comprises: assembly by families of data by a processing unit of predetermined data and acquired data, including meteorological data, the families being predefined, each data value being associated with a time window of validity; formatting the data of the families to associate with each data value a type of graphical representation as text or a scalar, vector, surface, or volume; selection of families of data to be displayed; choice of a display time window for each family of data to be displayed; spatio-temporal discretization of the trajectory; spatio-temporal correlation of the discretized trajectory with each family of data as a function of the time windows of validity, in the display time windows of the family, to extract a sub-assembly from each family of data, by the processing unit; display of the sub-assemblies in a single representation on the same display screen.

The present invention describes a speech enhancement method using microphone arrays and a new iterative technique for enhancing noisy speech signals under low signal-to-noise-ratio (SNR) environments. A first embodiment involves the processing of the observed noisy speech both in the spatial- and the temporal-domains to enhance the desired signal component speech and an iterative technique to compute the generalized eigenvectors of the multichannel data derived from the microphone array. The entire processing is done on the spatio-temporal correlation coefficient sequence of the observed data in order to avoid large matrix-vector multiplications. A further embodiment relates to a speech enhancement system that is composed of two stages. In the first stage, the noise component of the observed signal is whitened, and in the second stage a spatio-temporal power method is used to extract the most dominant speech component. In both the stages, the filters are adapted using the multichannel spatio-temporal correlation coefficients of the data and hence avoid large matrix vector multiplications.

The invention relates to a short-time traffic flow prediction method considering spatial-temporal correlation. The influence of temporal correlation on the traffic flow of a target detection point is considered, and a short-time traffic flow temporal correlation prediction value is acquired; the spatial correlation of the object traffic flow is analyzed and researched by using a hierarchical clustering method, and multiple key spatial correlation points are determined; the influence of the traffic flow of the spatial correlation points on the traffic flow of the target detection point is considered, and a short-time traffic flow spatial correlation prediction value is acquired; the temporal correlation prediction value, the spatial correlation prediction value and the prediction value of the present method are integrated by using an "entropy method" so that the final prediction result of the short-time traffic flow of the target detection point is generated; and the prediction error is evaluated and analyzed according to the prediction result of the traffic flow and the actual traffic data. According to the method, the defect of the present method that the spatial-temporal characteristics cannot be fully utilized can be overcome, and the spatial-temporal correlation prediction result and the prediction result of the present method can be further integrated so that the accuracy of the short-time traffic flow prediction result can be effectively enhanced.

The invention discloses a dynamic probabilistic power flow (PPF) calculating method considering wind speed predication error temporal-spatial coherence. The method is to analyze the node voltage and dynamic probability distribution of branch power flow of a wind power station built power system, so as to enable operators to analyze a system state conveniently. The method comprises the steps of describing the input variable predication error process according to autocorrelation coefficient stationary process; directly fitting to obtain the predication error distribution on the basis of nonparametric kernel density estimation and according to historical predication error data; performing Nataf transformation technology on the basis of the iso-probability transformation theory to obtain an error sample of temporal-spatial coherence; performing dynamic PPF calculation by the monte carlo simulation method on the basis of latin hypercube sampling so as to obtain the node voltage amplitude value and the digital characteristics and dynamic probability distribution of the branch power flow. By adopting the method, the node voltage and the dynamic probability distribution of the branch power flow can be effectively analyzed; the method has the advantages of being accurate in result and convenient to realize.

The invention relates to a method and a device for predicting airport traffic congestion based on a long short-term memory (LSTM) model. The method comprises the steps of acquiring traffic condition information and airport flight takeoff and landing information of a road within a preset range around an airport as well as aviation meteorological information within a preset range around the airport;inputting the traffic condition information, the flight takeoff and landing information and the aviation meteorological information into the LSTM model; acquiring an output result of the LSTM model,wherein the output result is a predicted congestion index of the road within the preset range around the airport in a future time period. The method is based on the consideration of space and time effects, and the aviation meteorological information is added; intra-regional traffic is regarded as a space-time related system; the prediction result is obtained based on the LSTM model, so that the accuracy of the road congestion index prediction within the preset range around the airport is further improved.

The invention discloses a short-time traffic flow prediction method based on spatial-temporal correlativity. The method comprises the following steps: determining an intraday traffic flow data matrixcontaining a target prediction point and a monitoring point; performing convolution processing and gating circular unit processing on the intraday traffic flow data matrix to acquire spatial-temporalcharacteristics of the traffic flow data; determining the previous day traffic flow data matrix and a last week traffic flow data matrix based on the traffic flow data of the target prediction point and the monitoring point at the same moment of the previous day and the same moment of the same day of the last week, and processing by adopting a bidirectional gating circular unit to acquire periodiccharacteristics of the traffic flow data; performing fusion on the spatial-temporal characteristics of the traffic flow data and the periodic characteristics of the traffic flow data, inputting intoa predictor, thereby predicting the traffic flow of the target prediction point. By applying the method disclosed by the invention, the problem that the short-time traffic flow prediction is complex in model and slow in prediction speed can be solved.

The far distance identity identifying method based on walk includes two processes of training and identifying. The training process includes the steps of obtaining training walk sequence, spatial contour cutting, shape and distance signal extraction, main component analysis, personalized physique characteristic extraction to extract visible personalized characteristic as additional characteristic for final vertification of walk classification, and obtaining trained walk data base. By means of statistic pivot element analysis method, far distance identify identifying system based on walk behavoir is realized. The present invention utilizes also improved background reduction method, characteristic space transformation of motion contour, and correlation identification or nearest neighbour rule in the characteristic extraction.

The invention discloses a short-time traffic flow prediction method based on spatial-temporal correlation. The method comprises the steps: selecting a road segment, the traffic flow of which needs tobe predicted, and interruption points in the road segment; selecting the short-time traffic flow historical data of all interruption points in the selected road segment, determining a prediction timeperiod of short-time traffic flow prediction, and verifying whether the historical traffic flow data of the interruption points is periodic or not; dividing a data set into a training data set and a test data set after the normalization of the traffic flow data through a normalization method; performing the prediction analysis of the test data set through an SARIMA model to obtain an initial prediction result; taking the prediction result as an input feature, and substituting the input feature into a random forest model to obtain a final prediction result; comparing the test data set with final prediction data, and analyzing an error. The method enables the flow data to be decomposed into a periodic part with an apparent trend and a random fluctuation part for analysis, and improves the prediction precision of the traffic flow data.

The disclosure includes a method for traffic flow prediction based on data mining on spatio-temporal correlations. The method includes establishing a prediction model, data mining on spatio-temporal correlations, and traffic flow prediction based on spatio-temporal correlated data. The prediction model can be a linear regression model with multiple variables. The data mining on spatio-temporal correlations is based on a multi-factor linear regression model and by means of the optimization method in terms of sparse representation. The data from the spatio-temporal correlated sensors that are relevant to the prediction task are determined automatically. The traffic flow prediction based on spatio-temporal correlated data refers to that the prediction is performed with the input to the prediction model to be the data from the spatio-temporal correlated sensors.

The invention discloses a crowd sensing excitation method for random participation based on a crowd sensing system. The crowd sensing system comprises a control platform and a plurality of mobile users. According to the method, first, the control platform calculates task returns according to relative heat of sensing tasks; second, the control platform evaluates data quality of all the sensing tasks, measurement times needed by the sensing tasks are calculated according to a data quality increment in single measurement, decreasing steps of measurement returns are determined, and returns to participants in next measurement are further calculated; and last, the control platform publishes tasks information, the mobile users select the sensing tasks, execute measurement and upload sensing data, and the control platform sends corresponding returns to the mobile users. Through the method, limitations on the crowd sensing system by random participation and time-space correlation are better solved, the excitation demand for participating crowd sensing of the mobile users is met, the degree of balance of user participation is increased while the degree of completion of the crowd sensing tasks is increased, and meanwhile the total income of each participant is maximized.

A CPU may select a variable from a variable set as a dependent variable. The variable set may be part of the data structure that includes a plurality of vector values, a vector value associated with a variable set of n number of variables, and each variable of the variable set having a variable value. The number of dependent variable steps for the dependent variable may be determined. The number of the vector values in a dependent variable step is determined as being number of independent variables. A function is mapped to a plurality of thread processors, and each thread processor is assigned for the function to be performed on each one of the independent variables for each of the dependent variable steps.

The invention discloses an acquiring method for a dynamic random optimal power flow of a power system for a wind-containing power field. According to the acquiring method, the dynamic random optimal power flow of the power system under the influence of the randomness of wind speed and load as well as temporal and spatial correlation is realized. The acquiring method comprises the following steps: firstly, establishing a dynamic probability model of wind speed, and analyzing the temporal and spatial correlation of the wind speed; secondly, calculating by adopting a deterministic dynamic optimal power flow based on an original dual and decoupling interior point method to obtain an optimal dispatching scheme; secondly, under the guidance of the dispatching scheme, solving the dynamic random optimal power flow considering the correlation based on a cumulants method to obtain probability distribution of a state variable, and adjusting the upper and lower bound of chance constraint according to the probability distribution; finally, performing iterative computation to solve a group of optimal dispatching schemes meeting all the chance constraints. According to the acquiring method disclosed by the invention, the optimal power flow of the power system under the random influence of an input variable can be effectively processed. The acquiring method has the advantages of accurate result and convenience for realization. The obtained result has certain guiding significance for dispatching persons.

The invention relates to a traffic flow prediction method based on a Transform space-time diagram convolutional network, and belongs to the technical field of traffic flow prediction, and the method comprises the following steps: constructing a static adjacency matrix according to detectors deployed in a road network and the connectivity and Euclidean distance between the detectors; merging the traffic flow original data collected by the detector according to a specified time interval; performing normalization processing on the data set by adopting a maximum-minimum method, constructing a traffic flow space-time diagram, and dividing the data set into a training set and a test set; a convolutional network prediction model based on the Transform space-time diagram is constructed; training a prediction model by taking the training set data as input; and performing traffic flow prediction on the test set by using the trained space-time diagram convolutional network prediction model, and performing evaluation analysis on a prediction error according to a prediction result and actual traffic data. Compared with a traditional method, the method has the advantages that the spatial-temporal correlation in the traffic flow data can be effectively extracted, information in the traffic flow data can be more fully mined, and the prediction precision is improved.

The invention discloses an intra-frame quick selection method based on a non-separable secondary transform mode. By means of the spatial-temporal correlativity of coding units (CU) which are located at the same location and different depths in a video sequence, an index value of an optimal NSST mode is predicted in advance, the unnecessary index cycling process is skipped, the time-consuming NSSTmode selection process in the coding process is avoided, and on the condition that it is guaranteed that the video subjective quality reduction is negligible, the computing complexity of a coder is reduced, the coding time is shortened, and the coding efficiency is improved. Meanwhile, the scheme is simple, easy to operate and beneficial to industrial popularization of a new-generation video coding standard.

The invention discloses a pumped storage unit vibration trend prediction method, which comprises the following steps of: firstly obtaining historical and real-time data of a unit online when it is innon-stationary vibration; transmitting data to a user terminal; performing time domain analyzing on vibration signals by utilizing experiential wavelet decomposition; extracting comprehensive characteristics of energy entropy and singular value; performing correlation analysis with unit operation working condition after performing discretization processing on signal characteristic data set according to a certain rule; performing frequent item mining by utilizing Apriori algorithm; analyzing time-space correlation of data characteristics and unit faults; dividing a unit safe operation area through correlation analyzing results; finally constructing a time series model, and predicting development trend in the future limited time through time series trend prediction method, so that the unit operate state trend is predicted and evaluated, and technical support is provided for implementing unit state overhaul. According to the pumped storage unit vibration trend prediction method, the trendcan be accurately predicted, evaluation index is comprehensive and it is convenient to evaluate.

The invention discloses a two-stage robust unit combination method considering spatiotemporal correlation of uncertainty prediction error. The method comprises the following steps that firstly, basedon complete historical data, a polyhedron uncertainty set capable of describing the spatiotemporal correlation linearity of the uncertain amount (wind power and load) is established, and an analysis relation between the confidence probability and the uncertainty set is given; then based on the polyhedron set, a two-stage robust safety constraint unit combination model with the lowest operating cost in a prediction scene as a target is established, and a generalized Benders decomposition method is adopted for solving. Finally, a large amount of historical wind power and load data are counted toconstruct an uncertainty set taking the spatiotemporal correlation into consideration, and testing is carried out on the basis of an improved IEEE-30 and IEEE-118 node system. The result shows that the polyhedron set considering the spatiotemporal correlation can effectively reduce the conservativeness of robust optimization, and meanwhile, the combination safety and economical efficiency of therobust safety constraint unit are ensured.

The invention discloses a traffic jam prediction method based on deep learning and fuzzy clustering. The method comprises the following steps: (1) obtaining three-parameter original data of road traffic flow detected by a detector; (2) removing abnormal data from the data by using a threshold method, replacing the abnormal data by using a moving average method, and complementing missing data to obtain complete traffic time series data; aggregating the data to obtain a proper time interval, and finally normalizing the aggregated data; (3) mining spatial-temporal correlation according to historical data, performing training by using a CNN and a GRU to extract traffic flow spatial-temporal features, and performing future moment parameter prediction by using the trained network; (4) performingfuzzy clustering by using the historical data in the step (2), and calculating a membership degree to obtain a clustering center; and (4) judging the traffic state of the prediction time period according to the membership degree of the traffic data predicted in the step (3), thereby achieving the purpose of predicting traffic congestion. According to the invention, the traffic jam can be predicted more accurately.

The invention provides an internet-of-things tampering invasion detection method in combination with abnormal monitoring. Aiming at a periodic environment monitoring network, the technical problem that the real perception data is tampered by malicious node and the credibility of the data is damaged is solved. The method comprises a local abnormal monitoring step, a neighbor invasion detection step and a global invasion detection step. The space-time relativity of the data in the network is considered, and mutual supervision is performed by using the openness of a channel usable for a node and a neighbor node, so invasion detection between neighbors is realized, and the space relativity is detected by using the global invasion detection with a suspicious abnormal event node as the center; by using the result, on one hand, the monitoring and the positioning of the abnormal event are realized, and on the other hand, the invasion node close to the abnormal event can be reversely detected.

The invention relates to a traffic data tensor filling method based on spatiotemporal constraints. The method comprises the steps of obtaining incomplete traffic data, and constructing a traffic flowdata tensor; building a tensor filling model based on factor decomposition; constructing a spatiotemporal constraint tensor by analyzing spatiotemporal characteristics of the data, and optimizing thetensor filling model based on the factor decomposition; and performing spatiotemporal constraint tensor filling, and restoring original traffic data. According to the traffic data tensor filling method based on the spatiotemporal constraints, provided by the invention, the tensor filling method based on the factor decomposition is applied to the field of traffic data restoration, and the spatiotemporal correlation and the low-rank characteristic of the traffic data are fully excavated, so that the precision of restoring the complete traffic data is improved.

The invention discloses a fast screen content coding method based on spatiotemporal correlation. According to the fast screen content encoding method disclosed by the invention, a CU division module based on spatiotemporal correlation firstly calculates an absolute frame difference between a current frame CU and the CU at the same position of a previous frame and divides the CUs into two categories; then the CU division module based on spatiotemporal correlation judges that the current CU terminates division in advance or judges that the current CU only performs PLT mode prediction according to the correlation between the current frame CU and a spatiotemporal adjacent CU of the current frame CU on depths and intra-frame prediction modes; a mode selection module based on spatiotemporal correlation skips a specific prediction mode by means of the mode correlation between the current CU and the spatiotemporal adjacent CU of the current CU; and a CU division module based on coding bits firstly obtains the threshold of a CU coding bit under each depth via a CU hit rate and a curve chart of the coding bits, and terminates the CU division in advance if the coding bit of the current CU issmaller than the threshold. By adoption of the fast screen content coding method disclosed by the invention, the CU division complexity is reduced, the selection process of the intra-frame predictionmodes is simplified, and the SCC coding efficiency is improved.

The invention discloses a spatio-temporal clustering method for compressive data gathering. The spatio-temporal clustering method comprises the steps of: 1, an initialization stage; 2, spatio-temporal correlation measurement; 3, network clustering; 4, compressive data gathering; 5, data reconstruction. In the network clustering step, aggregation nodes obtain clustering information according to a calculated value of spatio-temporal correlation and broadcast the information to the whole network, and intra-cluster nodes select cluster head nodes dynamically. By adopting the steps, the spatio-temporal clustering method for compressive data gathering reduces data transmission times under the condition of same reconfiguration error, balances network energy consumption, prevents a certain node from dying too early due to energy consumption, and prolongs network lifetime.

The invention provides a spatio-temporal correlation day-ahead plan load flow analysis method with uncertainty being taken into consideration. The spatio-temporal correlation day-ahead plan load flow analysis method comprises the following steps: obtaining next-day forecast wind speed of wind power plants in 96 time period within 24 hours; obtaining next-day output of the wind power plants; obtaining forecast error distribution of the next-day output of the wind power plants; carrying out independence conversion on random variables having spatio-temporal correlation; calculating semi-invariant of a load flow in each line; and determining relevant information for dispatching. The method carries out engineering algorithm processing on the forecast error distribution having correlation, thereby solving the problem that probability distribution having correlation cannot be obtained by utilizing a series expansion method directly. Analysis is carried out on the next-day load flow of each line in each time period through a Gram-Charlier series expansion method ; through such analysis method, probabilistic load flow problem can be solved conveniently and effectively; the method has a practical value; and the present series expansion method commonly used in medium / long-term probabilistic load flow analysis is expanded to short-term plan load flow analysis, thereby providing more data support for economic dispatching.

The invention relates to a self-adaptive traffic video real-time defogging method based on temporal-spatial coherence. The method comprises the steps: S1, estimating a lane space region of a time slice initial frame, a haze influence mark value T, an initial transmissivity correction value X and a transmissivity distribution condition; S2, extracting a straight line lane in a video and determining a limited lane region; S3, searching for a region with the minimum haze interference from an original image to obtain a brightest pixel value as atmospheric light intensity; S4, dividing space ranges, wherein a mark camera is arranged in each space range and calculating the haze influence mark value T, the image contrast and the initial transmissivity correction value by using an image shot by the camera and applying those to other cameras in the region; S5, obtaining the transmissivity distribution of the optimal transmissivity of each image block and optimizing the blocky transmissivity distribution by using a guide filter; and S6, solving the original image pixel value in the lane space region and reducing a haze-free image of the lane part.

Provided are a method and apparatus for detecting an abnormal movement. The apparatus includes a feature tracing unit configured to extract features of a moving object in an input image, trace a variation in position of the extracted features according to time, and ascertain trajectories of the extracted features; a topic online learning unit configured to classify the input image in units of documents which are bundles of the trajectories, and ascertain probability distribution states of topics, which constitute the classified document, by using an online learning method which is a probabilistic topic model; and a movement pattern online learning unit configured to learn a velocity and a direction for each of the ascertained topics, and learn a movement pattern by inferring a spatiotemporal correlation between the ascertained topics.