69results about How to "Low data requirements" patented technology

The invention discloses a wind power forecasting method based on a genetic algorithm optimization BP neural network, comprising the steps: acquiring forecasting reference data from a data processing module of a wind power forecasting system; establishing a forecasting model of the BP neural network to the reference data, adopting a plurality of population codes corresponding to different structures of the BP neural network, encoding the weight number and threshold of the neural network by every population to generate individuals with different lengths, evolving and optimizing every population by using selection, intersection and variation operations of the genetic algorithm, and finally judging convergence conditions and selecting optimal individual; then initiating the neural network, further training the network by using momentum BP algorithm with variable learning rate till up to convergence, forecasting wind power by using the network; and finally, repeatedly using a forecasted valve to carry out a plurality of times of forecasting in a circle of forecast for realizing multi-step forecasting with spacing time interval. In the invention, the forecasting precision is improved, the calculation time is decreased, and the stability is enhanced.

A system for locating a wireless device involves the use of the measured signal strength of various base stations in the building or outdoor area under analysis. A topological map of the building or outdoor area under analysis is created. The map is divided into cells, and signal intensities are collected in each cell. For each cell, the signal from a particular base station is fit to a statistical distribution, such as a Gaussian distribution, and the parameters of the statistical distribution are estimated. After a device obtains a set of signal strength measurements, a probabilistic technique is employed to estimate the probability of the existence of the measurements in each of the cells of the building or area under analysis. The estimated location is the cell with the highest probability. A mobile user is tracked with the use of a Markov chain and the system can be calibrated to account for equipment and environmental variations.

The invention relates to a method for predicting a low-porosity and low-permeability clasolite effective reservoir at a high-diagenesis stage and belongs to the technical field of oil and gas exploration and development. The method is characterized in that a geological process response relationship is established through carrying out analysis, simulation and quantitative evaluation on a geological process affecting porosity, the original pore preservation and induced pore increase / decrease conditions of a reservoir are determined, thus, the current porosity of a low-porosity and low-permeability reservoir at the high-diagenesis stage is quantitatively evaluated, the development of the effective reservoir is predicted, and a basis is provided for reservoir evaluation. The method for predicting the low-porosity and low-permeability clasolite effective reservoir at the high-diagenesis stage has the advantages that the predictability is strong, the distribution of various clasolite effective reservoirs at different stages of high-diagenesis can be accurately predicted, the fine prediction on the distribution of effective reservoirs of a research target section can be achieved, and the predicted unit plane can reach a hectometer level; the theoretical basis is solid, and the condition of over reliance on the data of sampled points in the traditional method is avoided; the quantitative performance is good, the applicability is wide, and the information requirements are relatively low; and the evaluation, with different information foundations, on different exploration and development stages is facilitated.

The invention provides a non-invasive continuous blood pressure measurement method based on deep learning. The measurement method mainly comprises a pulse wave signal preprocessing method, a deep neural network model establishing method and a blood pressure measuring method. Pulse wave signals input to a neural network are subjected to preprocessing, filtering and noise reduction and normalizationprocessing, so that an established model is more accurate and has better adaptability. The method uses a deep neural network, subjective feature extraction and complex mathematical modeling problemsare avoided, and the network can conduct blood pressure measurement on different forms of pulse waves of different individuals through a large number of data training networks and can be applied to awide range of blood pressure prediction scenarios after the network is trained for once. Compared with the prior art, the objectivity is high, the model robustness is strong, and the measurement method is suitable for integration in home medical monitoring equipment and is also suitable for blood pressure monitoring of a wearable device.

The invention provides a method for estimating the non-point source pollution load of a northern plain farmland area based on rainmaking experiments, and belongs to the field of agricultural non-point source pollution loading capacity estimation. The method mainly comprises the following steps: 1), field scale: carrying out rainmaking simulated experiments, monitoring rainfall capacity, rainfall duration and runoff coefficient, monitoring pollutants of collected runoff water, mastering surface runoff and pollutant output characteristics of a farmland in a research plot, confirming a localized output coefficient, and establishing localized non-point source output characteristic database; and 2), catchment scale: establishing a perfect output model database according to related data, and carrying out non-point source pollution load model estimation. The non-point source pollution loading capacity (including surface generation loading capacity, flowing into river loading capacity and entering into receiving water loading capacity) of the farmland area is calculated according to a formula; and the spatial distribution is analyzed. According to the method, accuracy for non-point source pollution load estimation of the farmland area can be improved, and decision support is provided for non-point source pollution identification and pollution control.

The invention provides a vehicle running track reconstruction method based on multiple probability matching under sparse sampling, which is characterized in that a historical data statistics sparse sampling point tolerance distribution is used, and a search area is determined; then a candidate match object (road section or intersection) is searched in a region of search, and can be divided into various types according to the characteristics of the candidate object, if no match object is in the search area, the sampling point is not considerate, if only one object is in the search area, then the sampling point couples to the only object, if various candidate objects is in the search area, a double layer probability matching model is used for further processing; the double layer probability matching model can calculate the coupling probability of each possible track according to matching probability of the sampling point and the selection probability of the reasonable path, and the track with utmost possible probability can be selected for being taken as a reconstruction track of the sparse sampling point. The vehicle running track reconstruction method can reduce the matching error of the sparse sampling data, and can effectively increase the precision and speed of reconstructed vehicle running track in a complex road net.

A lithographic method and apparatus used to pattern an object. An illumination system supplies a beam of radiation. An array of individually controllable elements patterns the beam. The patterned beam is projected by a projection system on to a substrate supported on a substrate table. Individual elements of the array are periodically addressed to load an image frame on to the array appropriate to the pattern to be imparted to the beam at any given instant in time taking into account the image to be projected onto the substrate. In any given frame loading operation, only those individual controllable elements that must change state are addressed to reduce the amount of data required to be transferred during the frame loading operation.

A method for use in video compression is disclosed. In particular, the claimed invention relates to a method of more efficient fractional-pixel interpolation in two steps by a fixed filter (240) and an adaptive filter (250) for fractional-pixel motion compensation.

The invention discloses a pipe network leak detecting method in combination with resistance identification, relates to pipe network leak detecting methods and aims to solve the problems that passive leakage control methods mainly need a large number of instruments, equipment and manpower and are hardly combined with automatic monitored control systems, and large deviation of optimization results due to a small number of samples and long consuming time exist in artificial neural network methods. The pipe network leak detecting method in combination with resistance identification includes that firstly, pressure observed values of flow nodes of part of pipe sections are used as known conditions; an equation set containing pipe section resistance information is established; resistance results are expressed through generalized inverse solution of the equation set. Then, a pipeline where leakage possibly occurs is divided into a plurality of different areas; in each area, observed values of operation parameters of an edge pipe network are used as known conditions, virtual nodes are introduced to represent leakage points, and specific positions and leakage flow of the virtual nodes are determined by the aid of genetic algorithm optimization; accordingly, leakage is positioned and quantified. Then sequential manual troubleshooting and re-checking is performed. The pipe network leak detecting method in combination with resistance identification is applicable to the field of pipe network leak detection.

A method for use in video compression is disclosed. In particular, the claimed invention relates to a method of more efficient fractional-pixel interpolation in two steps by a fixed filter (240) and an adaptive filter (250) for fractional-pixel motion compensation.

The invention discloses a robust optimization-based emergency resource mobilization and transportation scheduling plan generation method, including step 1: considering the demand of the disaster area and the uncertainty factors of transportation time, and establishing the uncertainty parameters for the integration of emergency resource mobilization and transportation scheduling An optimization model, wherein the uncertainty parameter optimization model includes a total cost optimization objective and constraint conditions; Step 2: using a robust optimization method to process the total cost optimization objective and the constraint conditions under the uncertainty parameter conditions into a robust Stick corresponding total cost optimization objective and robust corresponding constraint conditions; Step 3: Solve the robust corresponding total cost optimization objective based on robust corresponding constraint conditions; Step 4: Output emergency resource mobilization and transportation scheduling plan. It has the advantage of being able to quickly and effectively generate emergency resource mobilization plans such as rescuers, transportation vehicles, and medical services, and transportation dispatch plans for rescuers and seriously ill patients after a major natural disaster occurs.

A multi-attribute inference method of social network users based on a variational automatic encoder comprises the following steps: preprocessing online social network data, and constructing a user attribute network; constructing an attribute deduction model which comprises a user variation automatic encoder, an attribute variation automatic encoder and a discriminator, encoding input data by the model to obtain potential representation of the user and attribute information, and reconstructing a complemented user attribute matrix through the potential representation of the user; training a model through an adversarial training mode, so that the obtained potential representation of the user contains more complete attribute information; inputting the to-be-completed user attribute data and the friend relationship among the users into the model, wherein the output user attribute matrix represents the probability that the users have different attributes. The method can be used for complementing the user attribute data in the online social network, so that the complete user portrait is obtained, the required data is easy to obtain, the calculation complexity is low, the attributes can bequickly inferred in the complex network, and meanwhile, the accuracy in most attribute prediction is very high.

The invention relates to a cloud shadow identification method supported by priori data, in particular to a cloud shadow automatic detection method based on a real surface, and the specific process isshown as figure 1. Firstly, utilizing MOD09 A1 data to construct a real surface reflectance data set of a visible light near infrared band one scene every month; then, based on a 6S radiation transmission model; assuming that the ground pixels are uniform and have a Lambert surface; simulating the surface reflectance under the conditions of different atmospheric modes, aerosol types, observation geometries and aerosol optical thicknesses under the condition of a near infrared band of visible light under a sunny condition; according to the change condition of the apparent reflectance, constructing a lookup table of the apparent reflectance change based on the parameters, and nonlinearly fitting a minimum value of the apparent reflectance change by using a least square method according to lookup table data to obtain a cloud shadow detection algorithm of the to-be-detected sensor.

The invention provides a coupling method of a lumped hydrological model and a two-dimensional hydrodynamic model, 1, data processing: acquiring and proces that basic data required by the lumped hydrological model and the two-dimensional hydrodynamic model, wherein the computing grid of the hydrodynamic model and the grid data of the land use raster are adopt with the grid data of the DEM raster data of the elevation data, and the grid data of the DEM raster data of the land use raster are used with the DEM raster data; 2, constructing a hydrological model and calculating: construct a lumped hydrological model based on basic data; 3, a coupling method: adopting a downscaling method to make that obtain runoff process consistent with the temporal and spatial scales of the hydrodynamic model;4, constructing and calculating that hydrodynamic model: reading the downscaled runoff grid data into the hydrodynamic model for calculation, and obtaining the spatial distribution of flood eigenvalue. The invention has relatively low data requirements and can be applied to regions with little or no data. The coupling part is simple and reduces the errors and uncertainties in numerical computation.

A system for detecting moving objects within a predetermined geographical area is provided. The system is designed to convey object movement information from an airborne surveillance platform to a ground-based operator station with reduced data transmission. This is accomplished by computer processing image data on the surveillance platform prior to transmitting data to the ground station. First, the system constructs a 3D model of the area under surveillance, for example, by obtaining many different views of the area using an aircraft. One 3D model is maintained at the surveillance platform, and another is transmitted to the ground station. During a surveillance mission, a succession of relatively low data, 2D images are created and aligned with the surveillance platform's 3D model. The alignment reveals differences in the images (tracking data) which is then transmitted to the ground station for use with the ground station's 3D model to resolve object movement information.

The invention provides a user terminal positioning method and system, an electronic device and a storage medium. The user terminal positioning method comprises the steps of determining target cells corresponding to target sampling points in an LTE network according to MRO data; if all target cells are outdoor cells, determining whether the number of target sampling points that do not coincide is less three; if yes, acquiring primary positioning information of the target sampling points by using an outdoor positioning method and a distance between the target sampling point and the correspondingtarget cell; if not, performing three-point positioning on the target sampling points according to a least square method so as to obtain the primary positioning information of the target sampling points; and if the primary positioning information of the target sampling points has an error greater than a preset error threshold, performing error regulation on the primary positioning information byusing a machine learning method, so as to obtain secondary positioning information of the target sampling points. Through adoption of the method, the user in the wireless network can be positioned accurately, data requirements are lowered effectively, and the positioning process is reliable and fast.

The invention discloses an image splicing fusion system in the technical field of video image processing. The image splicing fusion system comprises an image collection module which is in electrical output connection with an image processing module, the image processing module is electrical output connection with a video flow server, a GPRS (general packet radio service) network unit is in electrical output connection with an image splicing fusion unit, and a wireless communication module is in electrical output connection with an equipment terminal. Detection and tracking of a moving target in a splicing image are implemented through the image splicing fusion unit, display through multiple video windows is avoided, an observer is enabled to concentrate on observing an image, and receiving multiple video flows is changed into receiving one video flow, so that resources of a computer are saved, and operand of the image processing module is reduced; splicing imaging is performed after processing through a fusion algorithm. The image splicing fusion system is high in fault tolerance and openness, short in processing time, low in data requirement and high in analysis capability.

The invention discloses an environmental factor estimation method based on an Arrhenius model. According to the method, a degradation track is drawn according to performance degradation data, an invalid threshold value is set, pseudo-failure lifetime data are obtained according to a degradation model, the obtained pseudo-failure lifetime data are taken as incomplete samples after the distribution of the pseudo-failure lifetime data is verified, and a truncation sample is obtained. An environmental factor expression formula based on the Arrhenius model is given according to the specific distribution definition of environmental factors, parameters in the environmental factor expression formula are estimated with an optimum linear estimation method, and finally, the environmental factors obtained through calculation are verified. Accordingly, the reliability of a product is estimated more accurately, simultaneously, the efficiency is high, and the cost is low.

The invention discloses an urban inland inundation risk rapid assessment method and system, and the method and system take the inland inundation occurrence speed as a risk measurement standard: for agiven inland inundation point, the shorter the time needed for reaching a dangerous ponding depth after rainfall is, the more limited the time reserved for emergency response is, and the higher the corresponding risk is. According to the scheme, calculation can be conducted only by designing rainstorm, DEM, land utilization / coverage and drainage system design standards, a complex hydrodynamic model is not involved, the requirement for drainage pipe network data is low, and calculation can be completed based on a GIS platform. According to the method and the system, the problems of high requirements on basic data and modeling capability of operators, low calculation efficiency and poor practicability of existing inland inundation risk assessment are solved.

The invention relates to a kernel density estimation-based non-invasive power load identification method. The method comprises the following steps of: selecting a common household power load as a research object, acquiring power consumption data of the research object, carrying out sub-state division and extraction power distribution; generating a household working state set according to the power distribution, and calculating simulation power consumption data under each state; carrying out kernel density estimation to obtain probability distribution reference model of each state simulation data; identifying household working state transition points in the reference models, and dividing each household working state data segments; and for each data segment, searching a household working state which is closest to the probability distribution of the data segment, and comparing the household working state with the probability distribution so as to complete an identification task. According to the method, the main data features of power load power consumption can be effectively extracted, the main data distribution features are highlighted, and the influences of random power consumption data and abnormal fluctuation are weakened, so that effect can be well decomposed in the aspect of non-invasive identification, and the method is suitable for the changing and complicated working environment of the current household power grid.

The invention relates to a post-earthquake road disaster area rapid prediction method and system and a storage medium. The method comprises the steps of dividing a research area into a plurality of slope units; obtaining a training set composed of landslide units and non-landslide units based on historical earthquake landslide distribution data; training based on the training set to obtain a landslide probability evaluation model; performing landslide probability evaluation by using the landslide probability evaluation model to obtain a danger unit; and determining a landslide range of each dangerous unit by using a physical mechanical model to obtain a disaster area prediction result of the post-earthquake road. Compared with the prior art, the invention comprises training a landslide probability evaluation model by using a machine learning method to obtain a danger unit with a relatively large landslide probability; then using a mechanical model for researching the dangerous unit, so that the rapid evaluation of the landslide influence on the road in the post-earthquake research area and the accurate space positioning of the road disaster condition are realized. The invention can carry out the rapid evaluation of a wide area space, and provides a reference for the formulation of earthquake emergency rescue measures.

The invention discloses a real image generation method based on an annotation image under unsupervised training and a storage medium. The method comprises the following steps: inputting the annotationimage into a generator to generate three output images with different sizes; using a hierarchical visual perception discriminator to obtain six discrimination results; converting the discrimination result into adversarial loss by adopting an adversarial loss function; generating a blurred picture, and then calculating the confrontation loss of a discrimination result obtained by inputting the blurred picture into the hierarchical visual perception discriminator; performing adjacent pairwise grouping on the output images, inputting the output images into a VGG19 network, and then calculating the consistent loss of the images; inputting the output picture into three semantic segmentation networks ICNet which do not share parameters, and calculating return segmentation loss; collecting finallosses obtained by the four loss values to optimize the whole network, returning to the first step when the network is not converged, and taking the optimized generator as an image generation model when the network is converged; and generating a real image from the input annotation image by adopting an image generation model.

The invention provides a water quality prediction method of a deep learning model based on a physical law and process driving. The method comprises the following steps: modifying a loss function of the deep learning model according to the physical law; using the water quality model to generate simulation time sequence data of the water quality indexes; training the deep learning model by using the simulation data to obtain a pre-trained model; adjusting and optimizing the pre-training model by using historical measured data of the water quality indexes to obtain a physical constraint and process driven deep learning model PRPGDL; and finally, predicting future water quality index data based on the PRPGDL model. Compared with a water quality model, the method needs fewer boundary conditions and parameters, and has higher prediction accuracy, speed and flexibility; compared with a deep learning model, the method has higher accuracy and universality, and needs less actual measurement data; the water quality prediction method is higher in accuracy, higher in generalization ability and applicability and less in actual measurement data demand.

A lithographic method and apparatus used to pattern an object. An illumination system supplies a beam of radiation. An array of individually controllable elements patterns the beam. The patterned beam is projected by a projection system on to a substrate supported on a substrate table. Individual elements of the array are periodically addressed to load an image frame on to the array appropriate to the pattern to be imparted to the beam at any given instant in time taking into account the image to be projected onto the substrate. In any given frame loading operation, only those individual controllable elements that must change state are addressed to reduce the amount of data required to be transferred during the frame loading operation.

A computer-executable method for diagnosing trouble causes in telephone cable segments when trouble data are available only for entire cables is presented. This method diagnoses problems in the components of processes when the component location of process problems is not known. It gives a statistical relationship between component attributes and process problem rates, to facilitate assessment and improvement of policies about the process. This method may be used for diagnosing systematic problems in segments of telephone cable connections of given central offices.

The invention discloses an intersection turning overflow detection method based on a vehicle track and a long-short memory neural network. The method comprises steps of cutting a track of a vehicle sample according to a starting time point and an ending time point, and extracting a track of a corresponding signal period; then calculating speed of the vehicle at each track point according to the track; then mapping the speed to a time axis and a space axis separately, so as to obtain a speed curve of the vehicle on the time axis in the period, and then sampling to obtain a comprehensive time-space speed vector. On the other hand, a detector distributed at the inlet of a stretching section acquires speed, and then turning overflow duration in a signal period is acquired by setting a speed threshold; the comprehensive time-space speed vector and the turning overflow duration are used as parameters so as to train the long-short memory neural network; and the trained neural network can be used for detecting turning overflow.

The invention relates to an individual network security awareness situation prediction method, and belongs to the technical field of computer and information sciences. The method comprises the steps of firstly performing pre-treatment and attribute selection on network security awareness data of a common individual, then judging an individual network security awareness level, using a naive Bayes algorithm to work out each level initial probability of the individual, initial probability of the individual and a single-step transfer matrix, and build a Markov chain, and thus building a network security awareness situation change probability prediction system for crowds with different risk levels. Compared with the prior art, the method provided by the invention can select the prediction modelwith the corresponding risk level according to the data input by different individuals and predict the change probability in many years while further enhancing the prediction accuracy of the individual network security awareness situation change probability, is fast in processing speed, and can achieve the purposes of predicting the future security awareness situation change probability of the individual and nipping in the bud.

The invention relates to a method for predicting a low-porosity and low-permeability clasolite effective reservoir at a high-diagenesis stage and belongs to the technical field of oil and gas exploration and development. The method is characterized in that a geological process response relationship is established through carrying out analysis, simulation and quantitative evaluation on a geological process affecting porosity, the original pore preservation and induced pore increase / decrease conditions of a reservoir are determined, thus, the current porosity of a low-porosity and low-permeability reservoir at the high-diagenesis stage is quantitatively evaluated, the development of the effective reservoir is predicted, and a basis is provided for reservoir evaluation. The method for predicting the low-porosity and low-permeability clasolite effective reservoir at the high-diagenesis stage has the advantages that the predictability is strong, the distribution of various clasolite effective reservoirs at different stages of high-diagenesis can be accurately predicted, the fine prediction on the distribution of effective reservoirs of a research target section can be achieved, and the predicted unit plane can reach a hectometer level; the theoretical basis is solid, and the condition of over reliance on the data of sampled points in the traditional method is avoided; the quantitative performance is good, the applicability is wide, and the information requirements are relatively low; and the evaluation, with different information foundations, on different exploration and development stages is facilitated.

The invention discloses a wind power forecasting method based on a genetic algorithm optimization BP neural network, comprising the steps: acquiring forecasting reference data from a data processing module of a wind power forecasting system; establishing a forecasting model of the BP neural network to the reference data, adopting a plurality of population codes corresponding to different structures of the BP neural network, encoding the weight number and threshold of the neural network by every population to generate individuals with different lengths, evolving and optimizing every populationby using selection, intersection and variation operations of the genetic algorithm, and finally judging convergence conditions and selecting optimal individual; then initiating the neural network, further training the network by using momentum BP algorithm with variable learning rate till up to convergence, forecasting wind power by using the network; and finally, repeatedly using a forecasted valve to carry out a plurality of times of forecasting in a circle of forecast for realizing multi-step forecasting with spacing time interval. In the invention, the forecasting precision is improved, the calculation time is decreased, and the stability is enhanced.