43results about How to "The solution is not easy to obtain" patented technology

The invention proposes an enterprise field supplier recommending method based on a BP neural network, and belongs to the technical field of data analysis and data mining. A neural network and a supplier index system can be updated with the increase of bidding data, which ensures the effectiveness of decisions and the change of trend. The method overcomes the difficulties as follows: the total score of a supplier is not easy to get, and the data set for prediction is not real and objective enough as suppliers are graded by means of direct introduction for a general method. The method is combined with the actual situation. The prediction results of the index system and a neural network model are fully fused, and a decision space is set aside for experts. It is thus ensured that supplier selection can be both subjective and objective. The supplier index system built in the invention starts from both suppliers and supplied products. The index capacity is expanded. The degree of preference to suppliers is considered. A more comprehensive and applicable index system is formed.

The invention discloses a voice conversion method based on adaptive Gaussian clustering under a non-parallel text condition, and belongs to the technical field of voice signal processing. The method comprises the steps: firstly carrying out the voice feature parameter alignment of non-parallel linguistic data through a method based on the combination of unit selection and sound channel length normalization; secondly carrying out the training of an adaptive Gaussian mixed model and bilinear frequency bending and amplitude adjustment, and obtaining a conversion function needed by voice conversion; finally achieving the high-quality voice conversion through the conversion function. The method overcomes the limit that the training stage requires the parallel linguistic data, achieves the voice conversion under the non-parallel text condition, is higher in adaptability and universality, employs the adaptive Gaussian mixed model to replace a conventional Gaussian mixed model, solves a problem that a Gaussian mixed model is not precise in voice feature parameter classification, combines the adaptive Gaussian mixed model with the bilinear frequency bending and amplitude adjustment, and is better in conversion personality similarity and voice quality.

The invention provides a Tibetan TTVS system realization method. The Tibetan TTVS system realization method includes the following steps of: defining 84 FDP feature points according to an MPEG-4 standard; establishing a 3D human facial initial model, and performing texture mapping on the 3D human facial initial model; describing Tibetan mouth shapes through FAP parameters defined by the MPEG-4 standard, photographing lips of a Tibetan speaker through a camera, and establishing a Tibetan mouth shape library; obtaining a 3D human facial model based on the 3D human facial initial model in combination with the FDP parameters and the Tibetan mouth shape library, and establishing a 3D human facial model library; utilizing a prepared Tibetan corpus, and then performing clustering on HMMs to obtain an HMM library; and firstly performing text analysis on a Tibetan text to obtain context relevant labels and a phoneme sequence after the Tibetan text is input into the Tibetan TTVS system, generating acoustic parameters through a parameter generation algorithm, and finally synthesizing Tibetan speech through an STRAIGHT algorithm. The advantages of the Tibetan TTVS system realization method are that: synchronized playing of a 3D human facial animation and the synthesized Tibetan speech can be achieved.

The invention discloses an electric vehicle energy consumption prediction method and system based on big data fusion. The method includes firstly, performing decoding, cleaning, interposition on historical data of vehicle operation, dividing the processed vehicle operation data into driving state data and charging state data; and carrying out data fusion on the driving state data and the corresponding environment information data, thereby obtaining an energy consumption level prediction parameter set required by model training and an energy consumption prediction process, and solving a problemthat parameters required by an electric vehicle in an energy consumption prediction process are not easy to obtain. Wherein the energy consumption level prediction parameter set comprises complex driving conditions and driving environment parameters in the actual driving process of the vehicle, the energy consumption level prediction parameter set is used for training an energy consumption levelprediction model of the electric vehicle, and the obtained trained energy consumption level prediction model of the electric vehicle can predict the energy consumption level of the electric vehicle more accurately.

The invention provides a sensor array steady adaptive beamforming method comprising the following steps: using array observation data covariance matrix to calculate orientation vector; calculating array observation data space power spectrum according a sample covariance matrix and the orientation vector; determining an orientation section of desired signals and interference signals according to a spectrum peak position of the space power spectrum; averaging space power spectrums outside the orientation section, and calculating an average noise power; using a convex optimization method to obtain reconstructed a signal covariance matrix and an interference plus noise covariance matrix; calculating a product matrix of the said two matrixes, and using the feature vectors corresponding to the maximum feature value to obtain beamforming device factors. The method can effectively solve the performance saturation problems in the adaptive beamforming process, so the adaptive beamforming is high in stability, and easy to be applied.

The invention relates to a method for distinguishing gene mutation types an individual tumor sample based on second-generation sequencing. A tumor tissue sample and a normal tissue sample are used forlibrary construction and NGS sequencing respectively, strand bias and different types of base frequencies of mutation sites stored in an intermediate file BAM for biological information analysis of the tumor tissue sample are analyzed, the quality of base comparison and the frequency of noise are used as the training characteristics of machine learning, meanwhile, type information of corresponding mutation sites of the normal tissue sample is paired to serve as a prediction mutation type, a classification prediction model is constructed to distinguish somatic mutation from germline mutation,the model is used to distinguish somatic mutation from germline mutation, the detection efficiency is high, the specificity is high, and after the model is established, the individual tumor sample canbe used for NGS sequencing and mutation detection, the detection cost of a normal or cancer sample can be well saved, and meanwhile, the problem that normal tissues of tumor patients with specific types are difficult to obtain can be solved.

The invention discloses an adaptive robustness output feedback control method for a controlled emission platform, belonging to the electro mechanical servo control field. The adaptive robustness output feedback control method for the controlled emission platform comprises steps of establishing a mathematic model of the controlled emission platform, designing adaptive robustness controller for outputting the feedback, and performing a test on the stability of the adaptive robustness controller outputting the feedback. The adaptive robustness output feedback control method for the controlled emission platform performs estimation on inter-framework coupling coefficients of the controlled emission platform, effectively compensates the inter-framework coupling interference torque of the controlled emission platform, effectively observes the system velocity quantity by utilizing an high order sliding mode observer, solves the problem that the velocity quantity is not easy to obtain through the traditional ARC control method, and guarantees the good control performance of the controlled emission platform.

The invention relates to an observation method of domain structure of heat-proof neodymium-iron-boron magnet. The solution of NaOH is slowly instilled into the solution of FeCl2 and FeCl3, the solution is agitated and filtered to obtain the power of Fe3O4, the power of Fe3O4, sodium oleate and suitable HCl are added into the distilled water, the solution is agitated and diluted to obtain the hydromagnetic of Fe3O4; the sample to be measured is electro-polished in the electrolyte of CrO3 and 85 % phosphoric acid; The hydromagnetic of Fe3O4 is placed on the surface of sample on the electromagnet, and covers the coverglass. When the power is dried, the domain structure can be observed by microscope. The cost of the method in the invention is low, the operation is simple, and it is easy to spread for application.

The embodiment of the invention provides an entity alignment method and system based on a hierarchical attention mechanism. The method comprises the following steps: dividing target entity data to besubjected to entity alignment in a knowledge graph, and obtaining a word-level entity data vector and a sentence-level entity data vector; obtaining word vector similarity between the word-level entity data vectors according to an attention mechanism, and obtaining sentence vector similarity between the sentence-level entity data vectors according to the attention mechanism; and according to the word vector similarity, obtaining a distribution weight parameter vector matrix of the word-level entity data vector, and according to the sentence vector similarity and a Jaccard coefficient formula,obtaining a distribution weight parameter vector matrix of the sentence-level entity data vector so as to carry out entity alignment on the target entity data. According to the embodiment of the invention, the entity alignment accuracy is improved, so that the entity vector is easier to generate, and the problem that prior information is not easy to obtain in the entity alignment process is effectively solved.

A method for calculating mass weighted mean mainly comprises five steps and specifically comprises the first step of adoption of a measuring point arrangement method, isentropic exponent value k calculation, speed coefficient lambda and flow function q (lambda) calculation, approximate mass flow calculation and parameter mass weighted mean calculation. In the measuring point arrangement method, it is required that total pressure measuring points and total temperature measuring points are arranged uniformly; boundaries of the responsible area of each measuring point are selected on a center line of the two adjacent measuring points, and it is required that the number of the total temperature measuring points and the number of the total pressure measuring points arranged in the same area are equal. The method for calculating the mass weighted mean has the advantages that calculation is easy, precision is high, calculation of the mass weighted mean of related pneumatic parameters in a uniform flow field can be achieved, and the problem that the mass flow at the measuring points is not easy to acquire is solved. The calculation method is simple, precision is high, and calculation of the mass weighted mean of the related pneumatic parameters in all flow fields can be achieved.

An environmental footprint assessment method and system for agricultural products, and the method comprises the following steps: obtaining agricultural original activity data of a target area through an Internet of Things technology, the agricultural original activity data including direct natural resource consumption data, direct emission data, secondary energy consumption data and resource material consumption data; acquiring background environment data of the target area; and setting an initial LCIA model, and improving the LCIA model by using the background environment data to obtain an LCIA evaluation model. According to the application, the Internet of Things platform technology is utilized to solve the problem that agricultural production LCI list data is missing, especially the first-hand primary activity data is not easy to obtain, and particularly, various types of sensors are installed on the first site of agricultural production activities to collect first-hand LCI primary activity data; so that the problem of data deviation caused by the fact that the LCI data is based on the secondary data type at present is solved.

The invention relates to a novel synthesis method for cis-octahydropyrrolo[3,2-b]pyrrole, and mainly solves the technical problems that raw materials are expensive, reaction conditions are harsh, a reaction route is long, the reaction operation is inconvenient, and a large number of isomers are generated through ring closure in the conventional cis-octahydropyrrolo[3,2-b]pyrrole preparation process. The method comprises the following steps of: performing alkylation reaction of industrially available 3-pyrrolidone serving as an initiative raw material and an allyl compound directly under the action of acid to obtain a 3-tetrahydro-pyrrolidone substitute; performing reductive amination reaction to obtain a 3-aminopyrrolidine derivative; performing protective group adding reaction; and performing ozonization reduction ring-closing reaction and dehydroxylation reaction on the 3-aminopyrrolidine derivative added with the protective group to obtain a corresponding cis-octahydropyrrolo[3,2-b]pyrrole derivative. The target product and important intermediates can be widely applied to synthesizing various pyrrole derivatives.

The invention discloses a hybridoma cell line of Escherichia-coli O157:H7 resistant monoclonal antibody. By adoption of the hybridoma cell line, the problem that the monoclonal antibody is not easily obtained; by passage for many years and times, the monoclonal antibody can be stably secreted, and the secreted monoclonal antibody is applied to a liquid-phase chip and a colloidal gold method to detect Escherichia-coli O157:H7 in foods, and has the advantages of good sensitivity, specificity, stability and repeatability and the like. Pathogens also can be detected under the condition of low concentration, so that the detection sensitivity and the detection rate are further increased and the leaking detection and the false detection are prevented strictly.

A novel direct current power circuit based on a choke transformer comprises a power taking power unit (10) based on a choke transformer, a filtering circuit (20), a charger unit (30) and a stabilizing output unit (40). The choke transformer is utilized to acquire power, and an electric control means is finally utilized to obtain an expected low voltage power supply to meet the power requirements of various electric devices. The circuit has the charging storage capability, a special low voltage cable is not required to be laid, auxiliary devices like a support are not required to be erected, and a resolution scheme is flexible and easy to achieve, low in cost and less in maintenance. The circuit is simple, reliable and practical in structure. The system has high reliability and high efficiency, solves the problem that the low voltage power supply is not easy to acquire due to the fact that a cable is inconvenient to lay along the existing electrical railway, the optimum direct current power supply is provided for an electric device, the integral reliability and information level of the electrical railway system are improved by means of investment of the device, and the control and monitoring means are flexible.

The invention relates to Er<3+>, Pr<3+> and Tm<3+>-doped yttrium lanthanum oxidetransparent ceramic capable of being applied to intermediate infrared 2.7-mu m waveband lasers and a preparation method of the transparent ceramic and belongs to the field of special ceramic material manufacturing technologies. The method is characterized in that high-purity Er2O3, Pr6O11, Tm2O3, La2O3 and Y2O3 nano-powder is used as raw materials, and an Er<3+>, Pr<3+> and Tm<3+>-doped yttrium lanthanum oxidetransparent ceramic material is prepared with a solid-state sintering method under the condition of the lower temperature (1550-1650 DEG C). According to the Er<3+>, Pr<3+> and Tm<3+>-doped yttrium lanthanum oxidetransparent ceramic prepared with the method, tests prove that the ceramic has a broader emission bandnear 2.7 mu m and can be used as the2.7-mu m wavebandintermediate infrared laser material.

The invention relates to coarse-fine-particle mixed soil and a preparation method thereof, and belongs to the technical field of civil construction. The coarse-fine-particale mixed soil is prepared from the following raw materials: 77-95 parts by weight of sand gravel with a particle diameter of 0.1-20mm, 11-16 parts by weight of powder clay with a particle diameter of less than 0.1 mm, and 4-6 parts by weight of a gel material. The invention also discloses the preparation method of the coarse-fine-particle mixed soil. According to the coarse-fine-particle mixed soil and a preparation method thereof, the preparation method is simple in process, easy in operation and loose in condition limitation. The coarse-fine-particle mixed soil is prepared from the sand gravel and the powder clay; theraw materials are easily obtained and are abundant in supply; and the prepared mixed soil has high compactness, is easily molded, and has better cementation performance among particles, so that the problem that in-situ soil sample of a deep covering layer is not easy to obtain can be well solved.

The invention relates to an oral medication preparation for the prevention and treatment of diseases of digestive canal infection containing pig IgE, wherein the IgE is isolated from the pig serum with biological activity. The process for isolation comprises separating serum from pig blood, removing the impure protein in the serum, removing the micromolecular impurity substance through concentration and carrying out virus deactivation. The oral medication preparation provided by the invention can also contain the pharmaceutically acceptable findings.

The invention provides a hexagrammos otakii population dynamic DEB-IBM model and a modeling method. The model comprises two types of model entities including individuals and an environment; the individuals are divided into embryos, juvenile fishes, male adult fishes and female adult fishes according to ages and genders, wherein the life history characteristics of the individuals are represented by four state variables in a hexagrammos otakii DEB model, namely a structure, L, cm, energy reserve, E, J, development, EH, J and breeding cache, ER, J; and according to the environment, the state variables comprise a temperature time sequence and a food concentration time sequence. The modeling method comprises the following steps of (1) defining the state variables and parameters; (2) performing a model initialization process; (3) establishing a running program; (4) adding a sub-model; and (5) running the model. According to the model and the modeling method, energy is used as flux in a model framework and penetrates through the whole life cycle and population changes, so that the problem that the parameters are not easy to obtain is solved to a great extent, and the coupling process of the fish population model and an external environment module is optimized.

The invention discloses a classification adversarial fraud detection method based on class clear representation, which adopts an improved auto-encoder and an improved generative adversarial network, and comprises four stages: a first stage: extracting preliminary features based on normal behavior data, and extracting data features thereof by using the improved auto-encoder; a second stage: modifying a target function of the original generative adversarial network to obtain an improved generative adversarial network, and generating pseudo-abnormal behavior data by using the improved generative adversarial network; a third stage: inputting the normal user behavior data and the pseudo-abnormal behavior data into an improved auto-encoder for training together, and extracting final features from the normal behavior data by using the trained encoder; and a fourth stage: training the improved generative adversarial network by using the finally extracted normal behavior data features, and detecting fraud by using a discriminator obtained after training as a fraud detector. According to the method, the accuracy and the stability of fraud detection are remarkably improved.

The invention belongs to the technical field of fuel cells, particularly relates to an electrochemical reaction power generation system, and solves the problems that an existing electrochemical reaction power generation system adopting gaseous fuel needs high pressure, and a fuel storage tank is large in size, not easy to obtain and poor in safety. An independent constant-temperature environment fuel concentration detection branch is used, the concentration detection stability and accuracy are high and is not influenced by external environment temperature, has the capability of directly inputting fuel with the concentration of 100% into the system for power generation reaction, has an automatic operation strategy of online reduction of an oxidation state of an electrochemical catalyst into a metal state, greatly improves the catalytic activity of the electrochemical catalyst, and greatly prolongs the service life of the electrochemical catalyst. According to the power generation device, the fuel concentration in the internal fuel circulation system is automatically adjusted by detecting the environment temperature and humidity, so that the system can be stably operated under different temperature conditions, and the self-starting capacity of an extremely-low-temperature system without external environment control facility support is achieved.