70results about How to "Strong theoretical foundation" patented technology

The invention discloses a label propagation anti-fraud detection method and system based on an enterprise relation map, belongs to the field of financial credit, and aims to solve the technical problem of how to effectively analyze complex network data to discover valuable information and further mine fraud risks embodied by complex network relations, and the method comprises the following steps of: s1, establishing an enterprise blacklist library; s2, constructing a relational graph: screening related tables and fields listed in the relational graph in a relational database, and extracting arelational database object entity and an entity relationship; and S3, performing anti-fraud detection on the enterprise based on the self-built blacklist library and the enterprise relationship graph:based on blacklist library identification relationship graph blacklist nodes, extracting blacklist node connection subgraphs, identifying fraud enterprise nodes in the connection subgraphs by applying a label propagation algorithm, and estimating the enterprise anti-fraud probability. The system comprises an enterprise blacklist library establishment unit, a relation graph construction unit and an anti-fraud detection unit.

The invention discloses a method for refining grains of large medium high alloy steel forgings in the technical field of metal heat treatment. The method comprises the following steps of: austenitizing the forged forgings, cooling to tip temperature of a perlitic transformation area, and isothermally keeping temperature or fluctuantly keeping temperature to realizing isothermal decomposition of pearlite; and cooling to room temperature, and performing the secondary austenitizing process to realize recrystallization to refine the grains again. The method solves the problem that the traditional multiple normalizing process in actual production has instable effect of refining the grains of the large forgings, and a serious phenomena of coarse grains and mixed grains. By incomplete isothermal balanced decomposition of austenite, structure inheritance is cut off and eliminated, average austenite grain size above ASTM No.5 grade is achieved, the structure state of the large forgings is improved, the ultrasonic inspectability of the large forgings is improved; meanwhile, working hours are greatly shortened, the energy consumption is reduced, and the cost is saved.

The invention relates to a thin-walled component arc wire filling additive manufacturing temperature field predicting method. The method includes the steps that a stacking condition of a thin-walled component is determined; a thermal cycle curve is measured; the shape of any stacking layer molten pool above a second stacking layer is observed; a model is built and grids are divided; a heat dissipation boundary condition and a heat source parameter are loaded; a model temperature field is calculated, a thermal cycle parameter is extracted, the length of the stacking layer molten pools is measured, and compared with test data, the maximum error of thermal cycles and size errors of the molten pools are all smaller than preset values through a fine-tuning heat source model shape parameter; the fine-tuning model shape parameter is loaded in a numerical model of the thin-walled component, and arc wire filling additive manufacturing temperature field predicting is completed. The numerical calculation method is adopted, by comparing the thermal cycle parameters on a substrate with the sizes of the molten pools of the stacking layers, thin-walled component arc wire filling additive manufacturing temperature field predicting is accurately completed, and a theoretical basis is provided for solving residual stress and deformation problems caused in the thin-walled component arc wire filling additive manufacturing process.

The invention provides a minimum miscible phase pressure computing method of low-permeability reservoir CO2 flooding. The minimum miscible phase pressure computing method of low-permeability reservoirCO2 flooding comprises the following steps that minimum miscible phase pressure and underlying parameters of four or more samples are obtained by adopting a slim-tube displacement experimental method; the underlying parameters of the four or more samples are substituted through a theoretical formula method to calculate and obtain the minimum miscible phase pressure ; the minimum miscible phase pressure error of a same sample is obtained by comparing the slim-tube displacement experimental method and the theoretical formula method, and empirical constants are obtained by substituting the underlying parameters of four selected samples into a logarithm formula and establishing an augmented matrix; and a computational formula model is obtained by fitting the minimum miscible phase pressure ina low permeability reservoir and the underlying parameters. According to the minimum miscible phase pressure computing method of the low-permeability reservoir CO2 flooding, the minimum miscible pressure determined by the slim-tube displacement experimental method is more accurate, the theoretical formula method is more convenient and fast, and the cost is lower; at the same time, calculation requirements of the low-permeability reservoir in the Northern Jiangsu Basin can be met, a new derivation computational formula is more accurate in computational accuracy, the theoretical basis is higher, the labor intensity is low, and the minimum miscible pressure computing method of the low-permeability reservoir CO2 flooding is suitable for the use of low-permeability reservoir in the Northern Jiangsu Basin.

The invention relates to a service bandwidth adjustability prediction based heterogeneous wireless network load balancing method. The service bandwidth adjustability prediction based heterogeneous wireless network load balancing method comprises the steps of 1) establishing a seasonal moving average differential model; 2) calculating a prediction step length, obtaining a predicting moment of a call arrival rate and predicting the call arrival rate of the heterogeneous wireless network in the current time bucket through the seasonal moving average differential model so as to obtain the call arrival rate at the predicting moment; 3) selecting a target network to which a service is switched by adopting an SAW algorithm or by adopting an improved SAW algorithm based on the call arrival rate at the predicting moment; and 4) after the service is switched to the target network, establishing a power and exponential function transfer curve based on the call arrival rate at the predicting moment and adjusting a service access bandwidth nonlinearly. Compared with the prior art, the service bandwidth adjustability prediction based heterogeneous wireless network load balancing method has the advantages of balanced active load, accurate and scientific prediction algorithm, predictable bandwidth adjustment algorithm, rapid and effective load control, decreased fluctuation of load, low call blocking rate and the like.

The invention discloses a power system vibration mode matching method based on a matrix perturbation theory. The method comprises the following steps: first, calculating to obtain a characteristic value and a characteristic vector corresponding to an initial system mode before parameter change by applying characteristic value analysis, and then calculating approximate values corresponding to the system mode of the characteristic value and the characteristic vector after the parameters change by utilizing the matrix perturbation theory; then, determining the corresponding relationship between the approximate values of the characteristic value and the characteristic vector before and after the parameters change; later on, finding a mode which is closest to the approximate values from exact values of the system mode after the parameters change, establishing the corresponding relationship between the exact values and the approximate values of the system mode after the parameters change; finally, establishing the matching and corresponding relationship between the system modes before and after the parameters change according to a contact link between the system modes after and before the approximate values of the system modes serving as parameters change. The method has the beneficial effects that the principle is simple and clear, the discrimination is fine and the criteria are complete.

The invention provides a design method for an output feedback controller with a time varying sampling period. The method comprises the steps of setting time delay of a network control system to be a constant value and to be less than a sampling period, wherein a time varying sampling period is bounded, packet loss meets a Bernoulli random sequence and the probability is known; discretizing an original linear time invariant network control system and obtaining a discrete time varying system with uncertain items through conversion; obtaining the maximum singular value of a matrix and a conjugate transpose matrix through solving to obtain parameters in the discrete time varying system; according to state variables received by a controller and the obtained parameters in the discrete time varying system, building a dynamic output feedback controller; combining the built linear matrix inequality and the Lyapunov stability principle, solving the matrix in the feedback controller to determine and obtain a dynamic output feedback controller. The method takes into consideration the instability of sampling periods of network control systems, has the advantages of clear design thought, high stability and simple operation and can be applied to engineering practices.

The invention discloses a mixed text feature word extraction method comprising the steps of performing word segmentation on a text by using a word segmentation technology; removing stop words by matching a stop word list; acquiring a series of word position and property weight values according to research statistics; extracting the first word c(w1), and the first m-1 words of which the RE (ci,c(w1) values are bigger according to information quantity of the words in the text based on the abovementioned two factors; building a word network model based on a word semantic similarity method; and at last finding a text feature word set meeting an important degree condition according to a neighborhood method. Compared with traditional text feature word extraction method, the method provided by the invention has higher accuracy, harsher conditions, and higher application value; according to the method, the feature of the text can be extracted under the condition that only a text condition is provided without a text set or without advanced classification, contribution degrees of the different words to a text thought are computed, and a good theoretical basis is provided for subsequent text similarity and text clustering.

The invention discloses a method for solving a text similarity based on the Gini index. The method comprises the following steps: performing text word segmentation processing by use of the word segmentation technology, matching with a stop word list to perform a stop word elimination operation on a vocabulary, and obtaining a series of vocabulary positions and word characteristic weighted values according to the research statistics; collecting and reducing dimensions of the text vocabulary by use of a target weight function as shown in description, combining the vocabularies with high similarity according to the semantic similarity, collecting and reducing the dimensions of above characteristic words again, and solving the inter-textual similarity by use of the similarity between the vectors. Compared with the traditional text characteristic vocabulary extracting method, the method disclosed by the invention is higher in accuracy, better in application vale, and good in data processing effect; the defects of an information gain method are overcome, the result is more suitable for the experience value, the text characteristic vocabulary high-dimensional spare problem and the problem of the synonyms and polyseme are solved, the contribute degrees of different vocabularies to the text thought are computed, and the good theory basis is provided for the subsequent text similarity and text clustering.

The present invention relates to a fuzzy system reliability design method in combination with a modal interval algorithm. The method comprises the following steps of: 1. defining operations of fuzzy numbers; 2. judging multiple events and carrying out derivation on the multiple events and the sub-events of the multiple events; 3. judging whether the monotonicity of the multiple events is consistent with the monotonicity of the sub-events of the multiple events, and separately performing modal interval analysis by using coercive optimal and partial coercive optimal theories; 4. performing calculation according to a modal interval operation rule; 5. performing calculation to obtain an interval reliability index, security possibility degree and invalidation possibility degree of a system for reliability design of the system; 6. determining whether the reliability design of the system is performed in a serial, parallel or serial-parallel manner, and if so, performing the next step, or otherwise, ending; 7. performing modal interval analysis in the same method as the steps 2 to 4 so as to obtain the accurate security possibility degree and invalidation possibility degree of the system; and 8. determining the design reliability of the fuzzy system according to an actual requirement. According to the method, the correctness of the system reliability design is improved and the calculation amount is reduced.

The invention provides a road traffic flow condition prediction method in a data sparse time period. When traffic flow data are insufficient, a transformation rule of a to-be-predicted road traffic flow in a time sequence is explored by using a time dynamic sequence supplementing method; a road traffic condition of a data sparse time period is restored by environment information feature extractionbased on a conditional random domain, so that a time evolution sequence of the road traffic flow in a period of time is obtained; matching with a historical time sequence of the road traffic flow isperformed to find out a time sequence fragment with a similar evolution trend and a traffic flow condition of a prediction time point is deduced. The test and on-site detection show that the prediction result of the traffic flow data sparse time period is basically accurate and reliable, so that defects of road traffic flow condition prediction in the data sparse time period in the prior art are effectively overcome, the traffic flow condition prediction weakness is eliminated, and the overall quality of the road traffic flow condition prediction method is improved.

The invention discloses a spatial heterogeneity analysis method for the sensitivity of the roadside parking demand to the parking fee. The distribution rule of sensitivity of the roadside parking demand to the parking price in different spatial positions is analyzed based on a geographical weighted regression model and historical data, a spatial weight matrix is adopted to describe the spatial heterogeneity of the sensitivity of the parking demand to price, the factors affecting the spatial heterogeneity are analyzed based on a generalized linear regression model, and a roadside parking demand elasticity prediction model is built. The method makes up for the defect that the parameters are globally fixed and lacks the consideration on spatial heterogeneity in previous parking demand and parking fee relationship modeling, and a relationship model between road section and resident information and the demand elasticity is built, and the demand elasticity difference between different spatial positions can be analyzed in a differentiated manner. Instances show that the spatial heterogeneity analysis method for the sensitivity of the roadside parking demand to the parking fee put forward by the invention lays a good theoretical foundation for regional dynamic parking fee charging, and is better than previous models.

The invention discloses a wind power plant equivalent value modeling method based on a main ingredient analysis method and a layered clustering algorithm. The method comprises the following steps of extracting 13 complete state variables for expressing the operation state of a wind power unit according to a built mathematical model of a built doubly feed induction generator; on the basis of a mainingredient analysis method, extracting a primary variable expressing the main oscillation features of the state variable in 13 station variables; using the primary variable as a grouping index; usinga layered cluster algorithm for performing clustering grouping on the wind power plant of the doubly feed induction generator; according to the result of the primary variable clustering, dividing thewind power plant into a plurality of machine unit groups; on the basis of the division result, building a wind power plant equivalent value model. The calculation complexity is low; the realization is easy; the application range is wider; the practical requirements of dynamic equivalent value analysis of the wind power plant can be well met; good theoretical basis is provided for analysis of theelectric power system safety and stability after the wind power integration.

The invention provides a coal bearing property logging sensitive parameter evaluating method and belongs to the field of unconventional oil and gas exploration and development. The coal bearing property logging sensitive parameter evaluating method is based on coal core scale logging and comprises extracting and establishing coal bearing property logging rapid evaluation indexes and forming an expository mode by combining natural gamma logging value and density logging value with limetone stratum density logging mean value. The coal bearing property logging sensitive parameter evaluating method achieves rapid and intuitional explanation of gas reservoir of coal seams.

The invention provides a method for improving estimated values of lines of an SCADA (supervisory control and data acquisition) system. The method comprises steps as follows: actually measured data of the SCADA system under a current power grid topological structure are acquired, line impedance is acquired and taken as estimation parameters for estimation of the lines, and estimated values of the lines are calculated, wherein the actually measured data of the SCADA system are acquired by actually measuring voltages and phases of nodes and line power through the SCADA system; voltage classes of the lines are determined, voltage per-unit values at the head end and the tail end of each line are adjusted, the line impedance is corrected according to the line estimated values and the adjusted voltage per-unit values at the head end and the tail end of each line and set as estimation parameters for re-estimation of the lines, and the re-estimated values of the lines are obtained. With the adoption of the method, network topology mistakes and parameter mistakes can be corrected simultaneously to improve the estimated values of the SCADA system and can be distinguished according to different voltage classes, and the method has a higher theoretical basis and higher engineering value.

The invention discloses an imprecise fault identification method for power grid completeness state information reconstruction. WAMS of an RTU and a PMU is used for synchronous phasor measurement, wide-area fault information is provided, the requirements for real-time performance and globality of tidal current fingerprint collection are met as much as possible, under the condition that fault samples and information are incomplete, an imprecise probability principle is introduced for research, and the requirement for fault tolerance is met as much as possible. Besides, wide-area measurement is taken as a background, and fault information acquired by the power flow fingerprints of the RTU and the PMU is combined to an inaccuracy principle so that fault identification can be performed under the condition of power grid uncertainty.

The invention discloses a kit capable of rapidly and specifically detecting mycobacterium tuberculosis infection, and comprises three pairs of specific primers for detecting free small RNA of mycobacterium tuberculosis. Specific primers are designed for detecting free small RNA of mycobacterium tuberculosis, and free small RNA is extracted from human peripheral blood, detected, and synthesized so as to obtain cDNA. Then a realtime quantitative amplification system containing gradient-diluted cDNA, and a kit and method for rapidly and specifically detecting mycobacterium tuberculosis infection are established. The kit and method solve the problems that in the prior art the diagnosis time in a lab for patients is too long, the detection sensitivity is low, and the operation is very complicated. The provided kit is capable of being used to detect mycobacterium tuberculosis in a clinical lab, and can also be used to evaluate clinical treatment effect. The provided kit has the advantages of high sensitivity, good specificity, low cost, and convenient operation on detecting mycobacterium tuberculosis, and is capable of fully filling the requirements on rapid diagnosis in clinic or lab.

The invention discloses a rubber injection cold runner design and optimization method. The method comprises the following steps that the number and positions of pouring gates are determined accordingto parameters of required rubber products, and paths of distribution channels are determined according to the positions of the pouring gates; through a fluid motion equation, the diameter of the distribution channels are primarily determined, and pressure of the pouring gates is equal; according to an emprical formula, required inlet flow of a cold runner and the value range of the diameter of thepouring gates are determined; the inlet flow, the diameter of the pouring gates and the diameter of the distribution channels in a cold runner pouring system are adopted as variables, and an orthogonal test is designed based on an orthogonal list; a PRO / E software is adopted to conduct a three-dimensional design of a cold runner test group, and a POLYFLOW software is adopted to carry out simulation analysis on a pressure field, a velocity field and a shear rate field in the rubber flow process in the cold runner; and according to requirements for the required product, an evaluation system isestablished for simulation results. According to the method, the design cycle of the rubber injection cold runner is shortened, and rationality and reliability of cold runner design are guaranteed.

The invention provides a new keyword extraction technology. According to the technology, vocabulary position weights and word class weights are determined according to a Chinese word segmentation preprocessing process, the relevancy between two vocabularies is calculated with reference to a core vocabulary with the highest text vocabulary contribution, a multi-subject network model is constructed, an objective function is constructed to extract connection words, a cross function is utilized to fuse the connection words into the multi-subject network model, a new model graph is obtained, and then an anteposition vocabulary, namely a text keyword is extracted. The technology is high in accuracy and has higher application value, the contributions of different vocabularies to text ideology can be precisely calculated, multi-subject performance is considered, different characteristics are distinguished, and a good theoretical basis is provided for subsequent text similarity analysis and text clustering.

The invention disclosed an improved algorithm for extracting a text feature by a small-world model. According to a Chinese word segmentation preprocessing process, vocabulary position weight and part-of-speech weight are determined, and the semantic relativity function R (c1,c2) is determined by combining a relevancy algorithm of two vocabularies on the basis of the HowNet with a vocabulary to text important degree method. Since a polymerism can construct a vocabulary semantic network model map, two parameters, including a density parameter and an edge weight parameter, are set, proper threshold values are independently set, and the threshold values are taken as text feature vocabularies if a condition is met. The algorithm exhibits better accuracy and application value, the deficiency that an information gaining method is only suitable for extracting text features of one category, the contribution degrees of different vocabularies for text thought can be accurately calculated, and meanwhile, a good theoretical basis is provided for subsequent text clustering.

The invention provides a correcting calculation method for suspended sediment runoff of a river. The correcting calculation method is characterized by comprising the following steps: firstly, calculating average sediment concentration of sections to obtain average value D_CSI and D_CSII of sediment concentrations of two sections; secondly, determining marginal distribution, and calculating joint distribution function parameters: determining marginal distribution functions F1(CSI) and F2(CSII) as well as a parameter theta of a Copula function Ctheta(u, v); thirdly, carrying out interval discrete: equally dividing a value taking interval into n parts, wherein each part h=( D_CSIImax-D_CSIImin) / n, and marking a midpoint of a subinterval [D_CSII(k), D_CSII(k+1)] as xk+1 / 2= xk+h / 2; fourthly, establishing a joint distribution model based on the Copula function; calculating to obtain u= FCSI(xCSI) and v= FCSII(xCSII) as well as an inverse function F<-1>y(v) of y and the Copula function c(u, v) according to given sediment concentrations x_CSI and x_CSII of the sections and the marginal distribution F1(CSI) and F2(CSII); fifthly, correcting the calculated sediment concentration on the basis of conditional expectation. The method has higher applicability; according to the method, the correcting calculation precision of the sediment concentration can be effectively improved, and working intensity of a hydrologic survey is reduced.

The invention relates to a determination method of a support joint allowable error of a cable-rod tensile structure. Firstly a support joint error model of the cable-rod tensile structure following definite distribution in radial, circular and vertical directions is constructed based on the stochastic theory; error sensitivity of the structure for positional deviations in the radial, circular andvertical directions of the support joint is reviewed by use of a Probabilistic Design System (PDS) module in the ANSYS software and the Latin Hypercube Sampling (LHS); on the basis, the concept of searching error allowable value and the specific search method are further put forward under the conditions that the reliability index is not less than 1.5 and the cable-rod internal force deviation is not more than 10% in the serviceability limited state and the serviceability limited state is satisfied in accordance with Unified Standard for Reliability Design of Building Structures and Technical Specification for Prestressed Steel Structures. The method is of strong theoretical foundation, capable of accurately determining the support joint allowable error of the cable-rod tensile structure and generally applicable to the support joint error calculation field of the cable-rod tensile structure.

The invention provides an improved text similarity solution method. The method comprises the following steps: performing word segmentation and unused word removal processing on two texts (the formula is as shown in the specification), calculating weights of feature vocabularies in the texts based on an information theory method, calculating position weights of the vocabularies according to the position information of the vocabularies, constructing feature value target functions (the formula is as shown in the specification) in the texts by integrating all factors, extracting two feature vocabulary vectors, and converting a text similarity problem into a vector distance problem. Compared with the traditional text similarity calculation method, the improved text similarity solution method provided by the invention has the advantages of higher accuracy and wider application range, and meanwhile provides good theoretical basis for subsequent text clustering.

The invention discloses the crack identification method and system of tight sandstone formation. The method comprises the following steps of based on deep induction logging data and medium induction logging data, acquiring a deep induction response value and a medium induction response value and acquiring a deep induction response average value and a medium induction response average value; basedon the deep induction response value, the medium induction response value, the deep induction response average value and the medium induction response average value, acquiring a first regression coefficient; based on the first regression coefficient, the deep induction response average value and the medium induction response average value, acquiring a second regression coefficient; based on the deep induction response value, the medium induction response value, the deep induction response average value, the first regression coefficient and the second regression coefficient, acquiring a doubleinduction difference coefficient; and setting a threshold, comparing the threshold and the double induction difference coefficient and identifying a crack. In the invention, through calculating the double induction difference coefficient and comparing with the threshold, a high angle crack in tight sandstone can be effectively identified, and the identification is not affected by changes in formation lithology, a physical property, oiliness and the like.

The invention discloses a preliminary prospecting well optimum decision-making method and system. The preliminary prospecting well optimum decision-making method comprises the steps: calculating the success probability of a preliminary prospecting well; calculating the value of the preliminary prospecting well; and establishing a decision coordinate system, and judging whether the preliminary prospecting well is an optimal preliminary prospecting well or not according to the positions of the data points of the preliminary prospecting well success probability and the preliminary prospecting well value in the decision coordinate system. According to the preliminary prospecting well optimum decision-making method, on the basis of considering the success probability and the value of the preliminary prospecting well, the two-dimensional optimization model is utilized to carry out preliminary prospecting well optimization decision, and a reliable and rapid decision basis can be provided forthe preliminary prospecting well, so that the deployment quality of the preliminary prospecting well is improved, and the exploration investment is inclined to a region with good benefits.

The invention provides a cell nucleus fluorescent dye as well as a dyeing method and application thereof, and provides a novel fluorescent dye for cell nucleus dyeing. The cell nucleus fluorescent dye has good membrane permeability, high sensitivity, rapid nucleus dyeing efficiency, good light stability, good biocompatibility and longer-wavelength exciting light and emission light, and can be applied to fluorescent dyeing of cell nucleuses.

The invention relates to an unfrozen water model based on adsorption and capillary coupling effect, which is prepared by the following steps: (1) collecting a soil sample, measuring the specific surface area AS of the soil and the humidity at room temperature, and taking the humidity of the soil at room temperature as the initial water content w0; (2) establishing an unfrozen water content model, and calculating the unfrozen water content wu; the method has the advantages of simple expression, few input parameters, convenience in calculation and deeper theoretical basis.

The invention discloses a method for calculating a main influence radius of surface settlement of a salt rock underground storage, which comprises the following steps: 1) simplifying the salt rock underground storage; 2) assuming that the boundary line of the subsidence area of the salt rock underground storage is an arc with O as a circle center and R as the radius; 3) calculating the total shearforce T and the total shearing resistance S acting on the arc; 4) changing the value of the arc radius R, calculating the total shear force and the total shearing resistance on the arc correspondingto the different arc radii; 5) fitting the relation expression of the difference (T-S) of the total shear force T and the total shearing resistance S on different radius arcs with the radius R, and making the T-S=0, determining the corresponding circular arc critical radius Rm value at the moment, and the circular arc corresponding to the critical radius is the settlement area boundary line in real significance; 6) determining the main influence radius of the surface settlement of the underground repository according to the boundary line of the settlement area. By the method, the main influence radius of the surface settlement of the underground repository is obtained, and the calculation accuracy is high.

The invention relates to the technical field of gene engineering, in particular to an SSR (simple sequence repeat) molecular marker of akebia trifoliate as well as an application and a preparation method of the SSR molecular marker. Primers of the SSR molecular marker comprise one or more pairs of primers shown in sequences SEQ ID NO:1-98. The developed transcriptome SSR molecular marker of the akebia trifoliate can be used for analyzing genetic diversity of akebia trifoliate germplasm resources and lays a good theoretical foundation for molecular biology and genetic research of the akebia trifoliate. The SSR marker developed with a transcriptome high-throughput technology is high in speed, low in cost and high in efficiency.

The invention provides a new text feature vocabulary extraction method. The method comprises the following steps: performing text word segmentation processing by using the word segmentation technology, performing unused word removal processing on vocabularies by using an unused word matching table, obtaining a series of vocabulary positions and vocabulary weight values according to investigation, survey and statistics, and extracting the first word c(w1), and the first m-1 words of which the RE (ci,c(w1)) values are bigger according to information quantity of the words in the text based on the abovementioned two factors, namely obtaining a feature vocabulary vector of the text. Compared with the traditional text feature vocabulary extraction method, the new text feature vocabulary extraction method provided by the invention has the advantages of high accuracy and higher application value, features of the text can be extracted when no text set is provided or no category is classified in advance and only one text condition is provided, the construction of different vocabularies to the idea of the text is calculated, and good theoretical basis is provided for subsequent text similarity and text clustering.