70 results about "Normalization model" patented technology

Systems and methods are provided for detecting an anomaly in a computer that is part of a population of networked computers. Snapshots are received from a plurality of computers within the population of computers, where individual snapshots include a state of assets and runtime processes of a respective computer. An asset normalization model is generated from the snapshots and serves as a baseline model for detecting an anomaly in the state of assets and runtime processes of a respective computer. A snapshot from at least one of the computers is compared to the asset normalization model in order to determine whether an anomaly is present in a state of static assets and runtime processes of the at least one of the computers.

The invention provides a control method of a grid friendly type distributed power source based on hybrid energy storage. The method comprises the steps of optimally designing a topology structure of the distributed power source, controlling a direct current power supply system in a coordinating mode through a direct current (DC)/DC converter controlling a model normalization model, and self-adaptively controlling a grid of an alternating current (AC) side DC /AC converter. The distributed power source not only can be connected in a large grid to operate so as to reduce the influence of intermittent renewable energy power generation grid connection on the grid and provide support of voltage and frequency for the grid, but also can be connected into a micro-grid to operate as a network unit of the off-network type micro-grid to maintain stability of voltage and frequency of the off-network type micro-grid. When an outer grid breaks down, the distributed power source can realize switch between a grid mode and an island mode so as to improve power supply reliability of important loads in the system. The distributed power source improves technical performance and economic performance of the whole system through matching utilization of an energy type system and a power type energy storage system.

The invention relates to a robust continuous emotion tracking method based on deep learning. The method comprises the steps that (1) a training sample is constructed, and a normalization model and a continuous emotion tracking model are trained; (2) an expression image is acquired and preprocessed, the expression image obtained after being preprocessed is sent to the trained normalization model, and an expression picture with standard illumination and a standard head posture is obtained; (3) a standard image obtained after normalization is used as input of the continuous emotion tracking model, expression-related features are automatically extracted and input through the continuous emotion tracking model, and a tracking result of a current frame is generated according to time sequence information; and the steps (2) and (3) are repeated till a whole continuous emotion tracking process is completed. The method based on deep learning is adopted to construct an emotion recognition model so as to realize continuous emotion tracking and prediction, the method has robustness on illumination and posture changes, and the time sequence information of expressions can be fully utilized to track the emotion of a current user more stably based on historical emotion features.

Normalized data models are programmatically generated from a combination of product configuration model data, product configuration engine runtime validation, normalized data mappings, and settings files declaring the scope of model content. A master model generation process effectively transforms conventional configuration data into normalized configuration data. The normalized configuration data allows a user to, for example, conduct comparative product configurations. In one embodiment, a normalized model generation process generates normalized data model representing attributes and normalized features of a product. In one embodiment, the normalized configuration data model is then added to in-memory data structures used during runtime contextual configuration analysis, thus reducing the total number of data items preserved as efficiencies result from eliminating duplication and effective use of search structures. In-memory representation of the normalized configuration data model can then be serialized to disk as a file to be loaded for runtime use in a deployment.

The invention provides a coordination control method for common direct current bus mixing energy storage systems. The method includes the steps of building topological structures of the mixing energy storage systems, coordinately controlling different types of the energy storage systems by means of a direct current (DC) / DC converter control mode normalization model, and coordinately controlling the energy storage systems in the same type and with different capacity. According to the coordination control method for the common direct current bus mixing energy storage systems, by means of coordinate control for a lithium battery energy storage system and a super capacitor energy storage system, reasonable utilization of the different types of energy storage systems is realized, and therefore, the response speed of the systems is improved, service life of the energy-type energy storage systems is prolonged, and technical performance and economical performance of the whole system are improved.

Verification of external interfaces of cores on system-on-chip (SOC) designs frequently entails the purchase of costly standardized software models to test the external interfaces. Typically, the standardized models provide more functionality than is needed. Instead of standardized models, test models may be developed and utilized, but this also incurs cost and delay. The present invention provides an efficient and economical alternative. A mirror interface, or copy of the external interface undergoing verification, is used with a standardized control mechanism to verify the external interface. Because all interface I/O connections can thereby be utilized, a cost-effective and highly reusable way of verifying such interfaces is provided.

The invention provides a quantity calculation method and device. The method comprises the steps that all initial models related to quantity calculation are obtained; model conversion is performed on the initial models so as to convert the initial models into a plurality of standardized models; all the standardized models are integrated so as to form a total quantity calculation model; a quantity result is obtained on the basis of the total quantity calculation model, and a quantity bill is generated. Compared with the prior art, the manual quantity calculation process performed by a construction budget worker is omitted, quantity calculation does not need to be performed by the construction budget worker and extra and independent modeling does not need to be performed, so that the working efficiency of the construction budget worker is improved, the quantity calculation data accuracy is ensured, meanwhile, quantity calculation data are effectively maintained and stored, and the overall building life cycle utilization rate of all the initial models such as BIM models is increased.

The invention discloses a normalization method for fatigue characteristics of an asphalt mixture under different stress states. The strength yield surface of the asphalt mixture based on a yield criterion is established by combining strength and fatigue test at different loading speeds under different stress states with a Desai strength yield surface model, a fatigue characteristic analysis technology based on the yield criterion is provided, and a normalization model for the fatigue characteristics of the asphalt mixture in different stress states is obtained. The fatigue characteristic analysis technology and the normalization model can be used to eliminate a difference between the fatigue test results of the asphalt mixture under different test conditions and make up for the deficiencyof no accurate evaluation of the fatigue performances of the asphalt mixture in traditional S-N fatigue equation; and the method realizes the unified expression of the fatigue performances of the asphalt mixture under different test conditions, and provides a theoretical, technologic and technical basis for realizing the scientific transformation from material fatigue to structural fatigue.

The invention discloses a method for constructing a testing environment of a satellite control system. The method comprises the following steps of classifying component parts of a satellite control system to be tested, establishing a standardized model of each part and performing modular packaging; classifying input and output ports in each standardized model according to respective signal type, and counting various signals of a single part model; performing number maximized configuration on hardware interfaces of test equipment according to types and numbers of parts possibly used by different satellite control systems; and fixedly connecting hardware interface signals of the test equipment and modular models, and fixedly connecting part models and a dynamical model to form a closed loop. By the method, the current situation of constructing the testing environment of the satellite control system by 'customizing for each satellite' is changed, the development speed is improved, the efficiency of constructing the testing environment is improved by modularizing and standardizing the interfaces, the method is convenient to use, and manpower, material resources and time cost are saved.

The invention relates to the field of voice synthesis, and discloses a voice conversion text method, a device thereof, a computer device and a storage medium. The method includes: acquiring voice information sent by a client; detecting a language type of the voice information; acquiring a voice normalization model matching the language type, and processing the voice information by using the voicenormalization model to generate a standardized voice message; obtaining a voice conversion text model matching the language type, and processing the standardized voice information by using the voice conversion text model to generate text information. The method provided by the invention can realize the translation of the dialect voice.

The invention provides a diagnostic determination method for a spacecraft control system under influence of noise. First, a multivariate probability distribution statistic model of the spacecraft control system is built through a standard model and an equivalent space method; then a detectability index is obtained; and fault detectability is determined; finally, an isolability index is obtained, and the fault isolability is determined. According to the diagnostic determination method, influence of process noise, observation noise and other interference factors on diagnostic performance are fully taken into consideration, fault diagnosis algorithms are not required to be designed, and the fault detectability and the isolability can be determined according to system information including kinetics, kinematics, a controller model and the like of the spacecraft control system and the configuration and installation condition of a sensor and an actuator, the fault diagnosis process of the spacecraft control system is simplified, meanwhile, a theoretical basis can be provided for design of a diagnosis algorithm, the design method of the spacecraft control system is optimized, and the controllability of the spacecraft control system is improved.

The invention discloses an automated testing device for a power distribution terminal. The automated testing device is characterized by comprising an OMAP module, an FPGA module, a display module used for man-machine interaction, wherein the OMAP module and the FPGA module communicate with each other through a data bus, the FPGA module is connected with a telemetering input testing interface and carries out processing on tasks at a telemetering input side, the OMAP module is respectively connected with a remote signaling input testing interface and a remote control input testing interface and carries out processing on tasks at a control side, and a standard model of the automated testing device is used for constructing a plurality of the power distribution terminals into a plurality of logical devices according to a modeling principle of an IEC61850 standard, so as to achieve function tests of the power distribution terminal. The automated testing device for the power distribution terminal solves the problems that the existing state grid typical design power distribution terminal is low in testing efficiency, the field maintenance workload is great and the like.

The invention discloses a railway and wind power plant environment wind speed machine learning prediction method. The method includes: selecting various neural network models; establishing 100 wind speed prediction characteristic pre-selection models for each wind speed; carrying out mean square error analysis and correlation analysis; selecting 10 models with good performance and strong feature independence; establishing a plurality of wind speed prediction integrated models; integrating the wind speed prediction characteristics; finally, establishing a wind speed prediction normalization model; carrying out Normalization processing on the wind speed integrated value, calculating the correlation between the predicted wind speed vector and the wind speed vector of the training sample, restoring and predicting the wind speed value through wind speed noise at the corresponding moment with high correlation. The method can effectively predict the non-stable wind speed and has the accurateprediction effect on complex and non-linear wind speed values.

The invention provides an auxiliary diagnosis method and system based on deep learning of electroencephalograms. The auxiliary diagnosis method and system are used for solving a problem that diagnosisaccuracy of epilepsy diseases is not high. The method comprises the following steps: S10: collected electroencephalogram sample data is obtained, the electroencephalogram sample data is integrated into a preset normalization model, and standardized electroencephalogram integer data is obtained; S10: according to a preset word embedding model, the standardized electroencephalogram integer data isconverted into a word embedding vector; S30: features of the word embedding vector are extracted according to a preset deep learning model, and the extracted features are subjected to time stamping operation and identifying and diagnosing operation; S40: according to the time stamping operation and the identifying and diagnosing operation, disease attack probability is output, and electroencephalogram sample data of which the disease attack probability exceeds preset probability is distinguished. Via the auxiliary diagnosis method and system disclosed in the invention, electroencephalograms ofa patient can be automatically diagnosed via a training model, time zones of epileptic seizures in the electroencephalograms can be automatically identified and marked, probability of contracting diseases can be obtained, work efficiency of clinicians can be improved, and diagnosis efficiency can be increased.

The invention discloses a multiphase flow (the below mentioned multiphase flow specially means three-phase flow and above three-phase flow) capacitance normalization method used for electrical capacitance tomography. Present normalization methods only aim at two-phase flow, and only take the capacitance value in the full field distribution of two dielectric constant phases into consideration, and a two-phase flow normalization model in three or above different dielectric constant phases existing in the multiphase flow, which only takes the capacitance values in the highest phase and lowest phase full field distribution case into consideration and neglecting the capacitances in the case of full field distribution of other dielectric constant phases, causes normalization error accumulation and affects the precision of an image reconstructing integrity. The multiphase flow capacitance normalization method used for electrical capacitance tomography, provided by the invention, considers the influences of the capacitances in the case of full field distribution of different dielectric constants on the measurement values of measured capacitances, and improves the precision of image reconstructing algorithm.

An industrial cluster innovation index calculation method includes the steps that a small town innovation index initial set and sample index original data are obtained to build a sample index data matrix; correlation coefficients between indexes of the sample matrix are calculated for collinearity diagnostics, secondary index screening is conducted in cooperation with the index guiding principle, and an index final evaluation system is obtained; a research object data matrix is built according to the index final evaluation system, an index threshold is further set, statistical processing and cleaning are conducted on researched data, and an index original data matrix is output; a data standardization formula is set, and a data standardization matrix is output by means of calculation of the index original data matrix through a linear standardization model; according to an AHP weight calculation formula, an index weight matrix is obtained; the data standardization matrix and the index weight matrix are substituted into a small town innovation index model to obtain a final index result, and ranking of various indexes is generated.

The invention discloses a marketing and distribution data consistency verification method based on the IEC61970/61968 CIM standard. The method respectively builds equipment relation models of a distribution side and a marketing side based on the IEC61970/61968 CIM standard; and consistency verification for basic data of the distribution side and the marketing side is performed according to the models. According to the method, standardization modeling management of the basic data of the distribution side and a production side can be realized via the relation models, data unified management and model data exchange are facilitated, data consistency can be verified via the data consistency verification method based on the standardized model, the data quality and data reliability are improved via fully usage of the redundancy of the data of a power distribution network, and basis and convenience are provided for the realization of typical application of various marketing and distribution scenes.

The invention provides a face image normalization method based on an auto-encoding network, which comprises the following steps: step 1, constructing a training data set of the auto-encoding network,and preprocessing the training data set; 2, constructing a coding network and a decoding network of the self-coding network, wherein the coding network is constructed based on a Resnet34 module, and the decoding network is composed of a deconvolution module; 3, using an L1loss loss function to measure the difference between the noise face and the normalized face, using a cross entropy loss function to measure the face image classification loss, and using the result of weighted summation of the two losses as the final loss function of the self-encoding network; step 4, training the preprocessedtraining data set on a self-encoding network to obtain a trained face normalization model; and step 5, inputting the to-be-processed face image into the face normalization model to complete face image normalization. By the adoption of the technical scheme, faces can be normalized in batches at a time without any prior information, compared with other face image restoration algorithms, the methodhas the advantages of being simple, convenient, easy to implement, high in efficiency and good in effect, face image identity information before and after normalization is kept consistent, has the advantages of being invariant in feature and high in accuracy and is of great significance to a face recognition algorithm.

The invention discloses a database server performance evaluation implementation method based on Informix. The C programming language and the SQL are used for development, the method is compatible with host systems of different structures, and a performance evaluation standard model is provided for an Informix data system; the performance evaluation standard model mainly comprises a starting module, a pressure generating module, a result recording module and a performance monitoring module; the performance evaluation method achieves standardization and routinization of an Informix pressure testing model, meanwhile avoids harmful effects on later performance evaluation of complex SQL statements and programmer levels, is convenient to transplant, and is compatible with the host systems of different structures.

The invention relates to a high-resolution image landslide automatic detection method based on multi-level perception feature progressive self-learning, and the method comprises the steps: 1, dividingfeature levels based on perception depth through employing a high-resolution image of a landslide region in order to support the feature normalization mapping of perception level progressive; 2, establishing a scale normalization model with progressively enhanced features, mapping perception feature elements of each level by taking spatial scales and dimensions as carriers, and generating a multi-level feature map with highly organized semantic information; 3, constructing a progressive self-learning regionalized network integrated with multi-level feature map constraints, and generating a landslide target-oriented detection network through end-to-end training; and d) inputting to-be-analyzed target high-resolution image data to the detection network, performing targeted detection of thelandslide target from the perspective of gradually enhancing image features, and finally outputting landslide target image representation. The method overcomes the defect of single understanding of complex scene images in the prior art, enhances the association ability between the features, and enables the detection result to be more accurate.