69 results about "Indicator vector" patented technology

The invention relates to a method of qualitatively evaluating a digital audio signal. It calculates a quality indicator consisting of a vector associated with each time window in real time, in continuous time, and in successive time windows. For example, the generation of a quality indicator vector calculates, for a reference audio signal and for an audio signal to be evaluated, the spectral power density of the audio signal, the coefficients of a prediction filter, using an autoregressive method, a temporal activity of the signal or the minimum value of the spectrum in successive blocks of the signal. To evaluate the deterioration of the audio signal, the method may calculate a distance between the vectors of the reference audio signal and the audio signal to be evaluated associated with each time window.

The invention discloses a large-fluctuation power grid load short-term forecasting method. The method comprises the following steps of: first, screening historical data samples by a form-factor method, determining the grading standards of sequences by utilizing a hierarchical clustering method, and determining the states of load values at each moment; calculating the state probability of a load in a period to be tested, selecting the state corresponding to the maximum state probability as the state of the load in a forecast period, wherein the interval mid-value of the state i is the forecast value at a forecast point. By the method for screening historical data vectors by utilizing an improved shape factor, comprehensively considering various influencing factors at similar moments, and accurately screening the historical data according to the characteristics of an impact load and a small hydropower load, the historical data amount required in a calculation process is reduced; collinear indicator vectors are found; and a forecasting result is simulated and calculated by a matrix laboratory. The data processing defects in the conventional algorithm are overcome; the forecasting precision is improved; and design basis is provided for the forecasting of a large-fluctuation load.

An observation lock detector block receives I&Q correlations and generates error measurement vectors, each including phase, frequency, code, pseudorange, and covariance residual estimates, and generates validity indicator and confidence indicator vectors, one of which error measurement vectors is an output as a selected error measurement vector with a respective validity indicator and confidence indicator vector, both of which selected vectors are communicated to a navigation solution processor in a ultratightly coupled navigation system for providing improved estimations of a navigation solution, the indicators indicating the usefulness of measurement errors, the prefilter detector block comprising discriminators, sequential filters, observation lock detectors and a measurement selector for selecting measurement vector based on phase, frequency, and code discriminations, the measurement errors being residual estimates that are indicated as valid or invalid for dynamic propagation of the residuals in a navigation receiver for improved performance, especially in low signal-to-noise environments, such as, for use in an ultratight GPS navigation system.

The invention discloses a fraud number identification method based on a space-time diagram, and the method comprises the steps: generating a telecommunication network through a full call record, constructing the space-time diagram for identifying a fraud number, and continuously intercepting a plurality of time-space subgraphs from the diagram according to a time interval T through employing a Snapshot method; calculating a call feature index vector of each user in the time period corresponding to each time-space sub-graph based on the call record of the user in the time period corresponding to each time-space sub-graph, and extracting a neighbor user having the same called party as each user from the time-space sub-graph for each user; constructing and training a space-time diagram neuralnetwork for identifying fraud numbers; and inputting the call feature index vectors of the target user to be identified and all neighbor users thereof in all time-space sub-graphs into a space-time graph neural network, and judging whether the number of the target user to be identified is a suspected fraud number according to the output. The invention belongs to the technical field of information, and can dynamically capture the change of a number calling behavior so as to realize accurate identification of a fraud number.

The invention discloses a rotating machine fault diagnosis method based on LOF-Kurtogram. The method comprises the following steps: firstly, segmenting a mechanical monitoring signal into a pluralityof data sections by utilizing a sliding time window with a fixed length; extracting feature index vectors such as a time domain and a time-frequency domain of each data sections; setting an initial value of a parameter k in an LOF algorithm, calculating a local abnormal factor value of each data section, screening out abnormal sections based on a 3 sigma criterion, and removing and cleaning the abnormal sections; and then inputting the cleaned data into the Kurtogram to obtain a filtered data section, solving an envelope spectrum of the filtered data section, and searching a fault characteristic frequency to complete fault diagnosis. According to the method, a traditional Kurtogram algorithm is improved on the basis of local abnormal factors, data diagnosis under large impact noise causedby environmental interference is achieved, the anti-noise processing capacity of mechanical fault data is improved, and the method has a more ideal effect on diagnosis of mechanical faults.

The invention relates to a method of qualitatively evaluating a digital audio signal. It calculates a quality indicator consisting of a vector associated with each time window in real time, in continuous time, and in successive time windows. For example, the generation of a quality indicator vector calculates, for a reference audio signal and for an audio signal to be evaluated, the spectral power density of the audio signal, the coefficients of a prediction filter, using an autoregressive method, a temporal activity of the signal or the minimum value of the spectrum in successive blocks of the signal. To evaluate the deterioration of the audio signal, the method may calculate a distance between the vectors of the reference audio signal and the audio signal to be evaluated associated with each time window.

A method and system for producing a printed text. A system operates in accordance with the method to: receive a message signal created by an author and representative of the semantic content of a printed text; produce a text signal in response to the message signal; analyze the message signal to determine a non-semantic indicator of the author's state of mind; determine a non-semantic characteristic of the printed text as a function of the determined non-semantic indicator; and printing the printed text in response to the text signal and the determined characteristic. The message signal can be a voice signal. A physiological signal such as pulse rate or variations in the pace, volume, tremulation, or average wavelength of the author's speech can also be used to determine the author's state of mind. In another embodiment of the invention, a system operates to input a message signal; generate a text signal representative of the printed text; analyze the message signal to determine a non-semantic indicator of the author's state of mind; map a state of mind indicator vector comprising the determined non-semantic indicator into an actual state of mind vector; and determine a non-semantic characteristic of the printed text as a function of the actual state of mind vector; and to print the text in accordance with the text signal and the determined non-semantic characteristic.

The invention discloses an indoor location method and device based on non-automatically deployed APs (Access Points). The method comprises the steps of collecting location fingerprint data offline based on existing APs of an existing surrounding; carrying out processing on the location fingerprint data collected in an offline stage; sorting and screening the APs; processing offline data by employing a terminal difference solution algorithm, thereby obtaining a final offline location fingerprint database for assisting indoor location judgment; after fixed value filling is carried out on the corresponding APs of the offline location fingerprint databases according to real-time data, through combination of online real-time received signal strength indicator vectors and the offline location fingerprint database, calculating a Euclidean distance; and calculating a final location result by employing a WKNN algorithm. Compared with the prior art, the indoor location system provided by the invention has the advantages that it is unnecessary to deploy the APs additionally, a great deal of base station plan and deployment work can be avoided, the location stability of the system is improved, and the system has better application value in practical engineering.

The invention belongs to the technical field of computers, and particularly relates to a data analysis method, a computer readable storage medium and terminal equipment. The method comprises the following steps of: firstly crawling financial data of a to-be-analyzed enterprise through a search engine; extracting appointed financial indexes from the financial data to form a first index vector; inputting the first index vector into a preset first neural network model to carry out processing so as to obtain a first output vector; inputting a second index vector selected from a preset template database into a preset second neural network model to carry out processing so as to obtain a second output vector; calculating a similarity between the first output vector and the second output vector; and finally determining a financial mode represented by a template index vector corresponding with the similarity with the maximum numerical value as a financial node of the to-be-analyzed enterprise.Through the method, the traditional artificial analysis is replaced by intelligent analysis of neural network models, so that the analysis cost is greatly reduced.

A method for synthetically evaluating a river water body includes such steps as establishing three-level index evaluation system, wherein the three-level index evaluation system comprise a first-levelindex, a second-level index and a third-level index, and the three-level index comprises a plurality of grades S2, quantitatively evaluating the degree of black and odor of the river water body by the matter element analysis method, comprising establishing an index evaluation matrix of three indexes, normalizing the index evaluation matrix and the index vector corresponding to the index evaluation matrix, and determining the correlation degree according to the index evaluation matrix and the index vector after the normalizing treatment; step S3, through the hierarchy-Delphi analysis method, determining the weights of the three-level indexes, establishing the expert evaluation matrix by setting the expert evaluation table, calculating the included angles of the evaluation vectors, and correcting the weights by using the normalized consistency of the expert evaluation.

The invention discloses an advertisement service user value evaluation method, device and equipment and a storage medium, and the method comprises the following steps: obtaining an association index of a user and an advertisement service, and building an index vector through the association index; calculating the association index by utilizing an analytic hierarchy process to obtain correspondingweight vector of the association index; and calculating a user value score according to the index vector and the weight vector. In combination with an actual scene of an advertisement information flowservice, a traditional RFM model is improved, L and P factors are added, a new model is constructed, and meanwhile, AHP empowerment is combined, so that a model result is more accurate and closely close to a service scene and conforms to a service rule, the user value subdivision is more accurate, and a layering effect is better; and the one-sided shortcomings of the traditional RFM model are scientifically improved, and meanwhile, the personal decision error is also improved by the AHP weighting, and the model score is more comprehensively calculated, so that the service decision is more scientific.

An anomaly detection method for mechanical monitoring data based on kernel estimation LOF is disclosed. Firstly, a mechanical monitoring signal is divided into a plurality of data segments by using asliding window with a fixed length. Extracting feature index vectors of each data segment; setting the initial value of parameter k in LOF algorithm and calculating the local anomaly factor value of each data segment, according to Nadaraya-Watson kernel estimation method, we can get the estimation value sequence of local anomaly factor; then the k-value is iterated and the threshold value S is setto realize the adaptive selection of k-value. The length of the sliding window is iterated. When the length of the sliding window is iterated to the preset value, the sequence containing the maximumlocal anomaly factor estimation value is selected. Finally, the threshold value T is calculated, the abnormal segment is screened out, and the abnormal segment detection is completed. The invention improves the traditional LOF algorithm based on the kernel estimation, realizes the adaptive selection of the parameter k and the length of the abnormal section, improves the ability of the algorithm todetect the abnormal section of the mechanical monitoring data, and has an ideal effect on the quality improvement of the mechanical monitoring data.

The invention discloses a rotary machine degradation trend prediction method based on steady subspace exogenous vector autoregression, which comprises the following steps of: firstly, carrying out first stationary subspace decomposition on denoised multi-channel signals to extract vibration stationary components; extracting time-domain and frequency-domain degradation characteristic quantities andobtaining a high-dimensional degradation index vector group through characteristic fusion; performing second stationary subspace decomposition and differential operation on the high-dimensional degradation index vector group in the time domain and the frequency domain to extract weak stationary components in the degradation indexes as rotary machine degradation indexes; carrying out stability test and impulse response analysis on the degradation indexes, determining endogenous and exogenous variables and model orders, determining vector autoregression model parameters through maximum likelihood estimation, and finally carrying out degradation trend estimation on the rotary machine at different prediction starting points. The degradation trend prediction model obtained by the method not only has good generalization ability under small sample learning, but also is rapid in calculation and strong in releasability.

An improved method of graph-based segmentation of objects in images uses the property of rectilinear shape classes which optimize the ratio of specific metrics, that can be expressed as Laplacian matrices applied to indicator vectors. A relaxation of the binary formulation of this problem allows a solution via generalized eigenvectors. This segmentation algorithm incorporating shape information requires no initialization, is non -iterative and finds a steady-state (i.e., global optimum) solution. The method is generally applicable to segmentation of rectilinear shapes.

The invention provides an internet AS deduction and router division method based on SVI. Firstly, a mathematical model is established, and the relation between nodes, community indicator vectors and subordinate vectors is determined; secondly, a prior probability model about the subordinate vectors and observation values is established; thirdly, the prior probability model is solved through the SVI algorithm, the degree of membership of the nodes is worked out, and the AS division result of the nodes is obtained. The invention further provides a method for checking the division result.

The invention relates to a vector method-based method for setting the parameter of a fractional order PI lambda controller and belongs to the technical field of frictional order automatic control. The method comprises the steps of writing a controlled object into a controlled object vector mode by combining the mathematical model transfer function of a given controlled object system, writing a performance index vector mode, utilizing a correcting vector L to correct the controlled object vector P to achieve the performance index vector G, and solving the transfer function of the controller, wherein the correcting vector and a controller vector form a triangle in a complex plane. The vector method-based method for setting the parameter of the fractional order PI lambda controller has the advantages that the calculation amount is reduced, and the setting process is simple and easy to understand. When the controlled object is changed, people can mechanically apply the subsequent formula by only computing the module value, the phase angle and the angle speed of the controlled object, so the process of repeated equation derivation is reduced, and the multi-solution problem existing in the matlab (matrix laboratory) solving is solved.

An observation lock detector block receives I&Q correlations and generates error measurement vectors, each including phase, frequency, code, pseudorange, and covariance residual estimates, and generates validity indicator and confidence indicator vectors, one of which error measurement vectors is an output as a selected error measurement vector with a respective validity indicator and confidence indicator vector, both of which selected vectors are communicated to a navigation solution processor in a ultratightly coupled navigation system for providing improved estimations of a navigation solution, the indicators indicating the usefulness of measurement errors, the prefilter detector block comprising discriminators, sequential filters, observation lock detectors and a measurement selector for selecting measurement vector based on phase, frequency, and code discriminations, the measurement errors being residual estimates that are indicated as valid or invalid for dynamic propagation of the residuals in a navigation receiver for improved performance, especially in low signal-to-noise environments, such as, for use in an ultratight GPS navigation system.

The invention discloses a system vulnerability discovery method and system based on a topology transfer relation. According to a relationship map of the number of nodes and a structure description system, simplification is conducted to form a vulnerability association map; according to a physical relation between the system nodes in the system, the transfer strength and probability of crash behaviors between the nodes are analyzed, some weak transfer relations are ignored, and a transfer relation matrix of the crash behaviors of the system is calculated; on the basis of the transfer relation matrix, key index vectors of the system are calculated; tightness indexes of an original system map are used for correcting the key indicator vectors of the system; on the basis of the key indicator vectors, hierarchical clustering and conclusion analysis are conducted on the system. The idea of combining the topology transfer relation, the coupling strength and a local structure is fully utilized,the depth and breadth of influence of key resources on the system can be fully reflected, and weak links of the system can be quickly found out.

The invention provides a new-situation load potential prediction method for a rural area. The method comprises the following steps that: according to family characteristics, a government input situation, market characteristics and a natural characteristic, analyzing an impact factor, and preliminarily constructing input-output index; utilizing SPSS (Statistic Package for Social Science) analysis software to screen the input-output index; according to a screening result, determining a final input-output index system; carrying out data processing on an input-output index vector; constructing rural electric energy to replace a recent potential analysis model; and applying a DEAP calculation tool to replace the recent potential analysis model and the input-output index system with the electricenergy of the rural area for prediction. The method has the beneficial effects that the research of the potential model replaced with the electric energy can more scientifically and accurately research and analyze rural area electric energy replacement work; and through the combination of a rural electric energy replacement potential value, the more reasonable rural electric energy replacement priority of each area is designed, and the influence of city haze for the life and production of residents can be more quickly gotten rid.

The invention discloses an evaluation method of internet dissemination influences of a toy defect event. The evaluation method of the internet dissemination influences is provided thereby on the basisof breadth and influence depth concepts of news dissemination. Different dissemination channels of micro-blogs and web media are comprehensively considered, and related keywords formed by clusteringof historical toy defect events are used for keyword extraction, and include an event range to the greatest extent; and dissemination influence values are calculated in sub-platform dissemination influence calculation of the micro-blogs and the web media according to different indicator vectors reflecting dissemination breadth and depth. Compared with a single indicator, different indicators can very sensitively reflect defect complaints of certain high-comment or high-quality users. A Sigmond function is used as a standardization manner, and a discrimination degree is higher. The evaluation method of the dissemination influences of the invention has higher practical values in reality.

The invention relates to a method for evaluating fouling of the passages of a spacer plate (10) of a shell-and-tube heat exchanger (11), wherein said passages (12a, 12b) are provided along the tubes (11) to allow fluid to flow through the spacer plate (10), wherein, for each of a plurality of at least one passage (12a, 12b): at least one measurement of a parameter dependent on fouling or the presence of magnetite is made in the vicinity of the passage, using an eddy current probe; this measurement is used to derive at least one indicator of the fouling of said passage, characterized in that said fouling is evaluated by comparing a set of one or more fouling indicator vectors of dimension of at least two, built from the fouling indicators thus obtained with a plurality of fouling indicator vector sets contained in a database, wherein each of said indicator vector sets is associated with a quantitative descriptor of fouling.

The invention relates to a construction method of an enterprise production safety early warning system adopting gray prediction, belonging to the information technology field. Through the gray prediction of early warning and prediction technology, the dynamic release of enterprise production safety status and development trend of the forecast. The method comprises the following steps: business data collection, generation of an early warning indicator vector, generation of an early warning value by an early warning model, generation of a prediction value by a gray prediction model, and generation of an early warning prediction comprehensive report. Not only the early warning indexes and their quantitative values are collected, but also an important standard is provided for the overall evaluation of the work level of production safety in enterprises.

The invention discloses an enterprise data privacy protection evaluation method based on a hierarchical model, and the method comprises the following steps: evaluating the maturity score of each environment factor in multiple fields, and obtaining an enterprise data privacy maturity self-evaluation initial score; constructing an enterprise data privacy protection hierarchical model, assigning an index weight and performing initial processing to obtain an index subjective weight; standardizing the initial subjective index value to obtain a standard index; on the basis of the Shannon entropy theory, calculating entropy values of standard indexes, and determining objective weights of the indexes; in combination with the index objective weight, calculating a subjective and objective index comprehensive weight and processing to obtain an index vector; and obtaining an enterprise data privacy maturity score of each environmental factor according to the enterprise data privacy maturity self-evaluation initial score and the index vector. According to the invention, the maturity analysis model of the enterprise data privacy is constructed, so that enterprises are helped to clarify involvedlinks which easily cause data privacy leakage in a real business environment, the privacy of users is protected, and the security level of the enterprises is enhanced.

The embodiment of the invention provides an index evaluation method and device. The method includes the steps of acquiring an instruction stream of a test program, and splitting the instruction streaminto N instruction segments with an equal length; determining feature matrices A, mean column vectors B and weight vectors W corresponding to the N instruction segments; according to the feature matrices A, the column vectors B, the weight vectors W and a plurality of optimized objectives, determining a target optimization model of the test program, and solving an optimized column vector of the target optimization model; determining number vectors and coefficient vectors of the instruction segments corresponding to the optimized column vector; simulating the instruction segments correspondingto the number vectors in a simulation device respectively to obtain feature index vectors of the instruction segments corresponding to the number vectors; determining inner products of the feature index vectors and the coefficient vectors as characteristic indexes for evaluating the test program.

A method for 3D volume segmentation in digitized medical images includes providing a digitized medical image defined on an N-dimensional lattice, obtaining an oriented closed 2-dimensional contour on one or more 2-dimensional slices from said image, defining face weights w from said image intensities by forming a dual lattice of said image lattice, minimizing∑i⁢wi⁢zisummed over all faces subject to the constraint Bz=r wherein vector z is an indicator vector indicating whether a face is present in a minimum-weight surface, vector r is an indicator vector of said 2-D contour, and wherein B is an edge-face incidence matrix, and segmenting said 3-D image into distinct regions separated by the minimum-weight surface indicated by said vector z.

The embodiment of the invention provides a keyword extraction method and device based on graph and word and sentence collaboration. The method comprises the following steps of: based on candidate words and sentences obtained from a text of a to-be-extracted keyword, calculating the diffusivity corresponding to each two candidate words; calculating a first weight of an edge between every two candidate words in the undirected weighted graph based on the diffusivity, calculating a first index of each candidate word based on the first weight; calculating a second weight of an edge between every two sentences in the directed weighted graph, calculating a second index of each sentence based on the second weight, and obtaining a first index vector of the sentence based on a second index of the sentence, constructing a second index vector of each candidate word containing a fifth index of the candidate word based on the first index of each candidate word and the obtained first index vector, and extracting a keyword in the text based on a size sequence of the fifth indexes in the second index vector. According to the embodiment of the invention, the accuracy of extracting the keywords in the text can be improved.

The invention relates to the technical field of big data, and discloses a data processing method based on an electric power big data platform, which comprises the following steps: acquiring electric power data, and carrying out missing value processing and noise data processing on the electric power data; according to the processed electric power data, calculating to obtain an electric power data index, and assigning an electric power data index weight by using an electric power data index weighting method; detecting the weighted power data indexes by using an abnormal power utilization detection method based on machine learning; clustering the power data index data by using an improved K-means algorithm, and taking a clustered power data index vector as a power data feature vector; and taking the electric power data feature vector as model input, and performing fault diagnosis of the electric power system on the electric power data feature vector by using an improved LSTM network model. The invention further provides a data processing system based on the electric power big data platform. According to the invention, power data processing is realized.

The invention relates to an analysis method for quantifying a physiological state of the human body, which comprises the following steps: finding n index vectors closest to an input vector in an index database G, and forming a matrix D by the n index vectors; finding an optimization parameter w; solving an estimation vector Vest by utilizing the optimization parameter w, wherein Vest is equal to Dw; for each index vector xj of the matrix D, finding n index vectors closest to the index vector xj, forming a matrix Dj, and finding an optimization parameter wj; solving an estimation vector Vjest by using the optimization parameter wj, wherein Vjest is equal to Djwj; calculating a difference value between the input vector and an estimation vector thereof as a first difference value Vre; calculating a difference value between each index vector xj in the matrix D and an estimation vector Vjest thereof, and taking the difference value as a second difference value; all the second difference values form a difference value library Gre; and calculating a state change index according to the distribution condition of the first difference value Vre in the difference value library Gre.

The invention discloses a network performance increment optimization evaluation method oriented to simulation of a power communication system. The network performance increment optimization evaluationmethod comprises the steps that performance indicator data of the power communication system are preprocessed, and the indicator vector corresponding to each performance indicator datum is obtained;the products of the local densities of all the indicator vectors and the extreme distances are subjected to descending sort, a first preset number of the indicator vectors are selected as central points of all class clusters, and the remaining indicator vectors are divided into the corresponding class clusters to be clustered according to the relative distances; multiple indicator vector representations in each class cluster are determined, and the representative vectors are selected; the representative vectors in each class cluster and the increment simulation performance indicator data are combined to form a new to-be-clustered sample set, and re-clustering is conducted according to the relative distances between the increment simulation performance indicator data and the representativevectors to update the representative vectors; and according to the representative vectors updated after re-clustering, evaluation is conducted through the network performance increment optimization evaluation method of the power communication system.