148results about How to "Guaranteed forecast accuracy" patented technology

The present invention relates to the social network analysis field, in particular to a microblog popularity degree prediction method based on a user and microblog theme and a microblog popularity degree prediction system based on the user and microblog theme. The method comprises the steps of obtaining the microblog data and the user data in a preset time period, obtaining the user attribute characteristics and the microblog theme characteristics according to the microblog data and the user data, carrying out the normalization processing on the user attribute characteristics, carrying out the user clustering on the processed user characteristics, and obtaining the user class information according to a clustering result; according to the microblog theme characteristics and the user class information, obtaining a forwarding characteristic of the user clustering under a microblog theme, and calculating a weight coefficient under the microblog theme of the user clustering; according to the microblog theme characteristics, the user attribute characteristics and the weight coefficient, constructing a microblog popularity degree prediction model, and predicting the microblog popularity degree according to the microblog popularity degree prediction model.

The invention discloses a consistency maintenance device for a multi-kernel processor and a consistency interaction method, mainly solving the technical problem of large directory access delay in a consistency interaction process for processing read-miss and write-miss by a Cache consistency protocol of the traditional multi-kernel processor. According to the invention, all kernels of the multi-kernel processor are divided into a plurality nodes in parallel relation, wherein each node comprises a plurality of kernels. When the read-miss and the write-miss occur, effective data transcription nodes closest to the kernels undergoing the read-miss and the write-miss are directly predicted and accessed according to node predication Cache, and a directory updating step is put off and is not performed until data access is finished, so that directory access delay is completely concealed and the access efficiency is increased; a double-layer directory structure is beneficial to conversion of directory storage expense from exponential increase into linear increase, so that better expandability is achieved; and because the node is taken as a unit for performing coarse-grained predication, the storage expense for information predication is saved compared with that for fine-grained prediction in which the kernel is taken as a unit.

The invention discloses a real-time prediction method of mine gas concentration in short and medium terms based on radial basis function neural network integration. The method comprises the following steps of: taking mine gas concentration data as a chaotic time series to construct a plurality of prediction sub-models of radial basis function (RBF) neural networks, and taking a weighted mean of synchronous prediction results of all prediction sub-models as an integrated prediction value to realize prediction model initializtion of RBF neural network integration; then realizing prediction of the gas concentration in the range of from a short term to a medium term through setting an integrated capacity parameter (the integrated capacity parameter is also equal to an RBF network prediction step-length); and obtaining a new prediction sub-model by utilizing an incremental training mode aiming at the characteristics that gas concentration information is continuously collected, and realizing updating of the RBF neural network integration according to a first in first out queue sequence so as to improve real-time prediction precision of the gas concentration, therefore, a proper compromise can be obtained between prediction range and prediction precision requirements, and the technical requirement on a mine gas information management system is satisfied.

The invention discloses a method for fast predicting organic pollutant n-caprylic alcohol / air distribution coefficient based on molecular structure, belonging to the technical field of quantifying structure / active relationship (QSAR) facing to the environmental risk evaluation. The method is characterized of comprising the steps of: adopting the molecular structure of atomic center fragment characterization compound; and screening the atomic center fragment combination by means of stepwise regression and partial least-squares regression, to build a group contribution model for predicting KOA.The internal authentication and the external authentication improves that the built KOA group contribution model has stability and predicting capability, and a range and distance method and a probability density method express the application domain of the group contribution model, thereby defining the application range of the model and guaranteeing the predict accuracy. The method has the effectsand benefits of being capable of fast predicting the KOA of the high flux compound, obtaining the KOA with low cost, being helpful for obtaining the high flux KOA data, and having a significant meaning for the environment supervision and the risk evaluation of chemicals.

A power load forecasting method comprises the following steps: steps 1, trialing all annual load characteristic indexes and a plurality of influence factors and establishing quantitative relations between all the annual load characteristic indexes and the plurality of influence factors; historical data includes annual load characteristic index data of the past years and influence factor data; step 2, forecasting all the annual load characteristic indexes of the years to be forecasted through the quantitative relations obtained in step 1, and forecasting the annual energy output of the years to be forecasted according to the forecasted annual load characteristic indexes; step 3, distributing the forecasted annual energy output to every mouth. The annual energy output and the mouth energy output can be forecasted through the method, the plurality of influence factors are considered as a whole, the forecasting accuracy is high, and a reliable foundation is provided for planning of power utilization plans. An improved grey forecasting method is used for calculation, so that the efficiency is high.

The embodiment of the invention discloses a bandwidth prediction method and device, equipment and a storage medium, and the method comprises the steps: obtaining an uplink receiving bandwidth prediction value queue of a receiving end; obtaining a current sending rate, a first round trip delay queue and a packet loss rate queue of the sending end and an uplink sending bandwidth prediction value of a previous period; wherein the round-trip delay in the first round-trip delay queue is determined according to the sending time that the sending end sends the test data packet in each period and the time that the sending end receives the test response data packet from the receiving end; and determining the uplink transmission bandwidth predicted value of the current period according to the uplink receiving bandwidth predicted value queue, the current transmission rate, the first round-trip delay queue, the packet loss rate queue and the uplink transmission bandwidth predicted value of the previous period. According to the method, the network bandwidth can be accurately predicted, and network resources can be saved.

The invention relates to a rapid collective Kalman filtering assimilating method for real-time data of high-frequency observation data. The method comprises: collecting the high-frequency observation data and controlling the quality; calculating an observation error covariance matrix; obtaining the error covariance matrix of background fields by calculating a forecast trend, i.e. the difference value of the adjacent background fields; utilizing the covariance matrix, the error covariance matrix of the background fields, the observation data and the background fields currently obtained by the calculation of a marine numerical model so as to carry out the real-time assimilation on the observation data of different moments, assigning the updated analysis field to the initial field of the next-moment integral and continuously forecasting forwards; and repeating the operations, thus realizing the real-time assimilation on the high-frequency observation data of different moments in the integral course. The assimilating method has the advantages that the real-time assimilation of the high-frequency observation data is realized; the assimilation efficiency of the data is enhanced; the defect that a large amount of collective models are simultaneously operated in the implementation course of the traditional EnKF (ensemble kalman filter) is overcome; the problem of non-convergence is avoided; and the purposes of accurate numerical simulation and marine forecasting are reached.

The invention discloses a course recommendation method and system based on a graph convolutional neural network and dynamic weight, and the method comprises the steps: obtaining a score value of a user for each course, carrying out the preprocessing of the score value, and obtaining a first user-course matrix; constructing a graph convolutional neural network according to the first user-course matrix; generating a user embedding vector and a course embedding vector according to the graph convolutional neural network; predicting a score value of an unscored course in the first user-course matrix according to the user embedding vector and the course embedding vector; filling the predicted score value into the first user-course score matrix to obtain a second user-course score matrix; and performing sequence pattern mining on the second user-course scoring matrix to obtain a recommended course sequence of each user. The method improves the prediction speed, guarantees the accuracy of a prediction result, and can be widely used in the technical field of deep learning.

The present invention discloses artificial neural network technological method of forecasting the terminal of medium-sized and small converter. For steel-making converter system, there are many factors, such as furnace life, gun position, sputtering, etc. to affect the terminal carbon content and temperature in non-linear relation hard to describe mathematically. The present invention applies neural network technology in the control system, and can monitor and forecast the non-linearity, non-determinacy and complexity effectively to forecast the terminal temperature and terminal carbon content of converter accurately.

The invention relates to a traffic accident rate prediction system based on on-line variational Bayesian support vector regression, which comprises a data preprocessing module, an online variational Bayesian support vector regression model building module, an online variational bayesian support vector regression model training module, and an online variational bayesian support vector regression model prediction module. This method effectively solves the problem that the traditional support vector regression model predicts the speed of traffic accident rate is slow, the prediction result is inaccurate, and it is difficult to solve the problem on line and show its practical value.

The invention discloses a product potential user mining method and device. The product potential user mining method comprises the following steps: collecting user data to be predicated; for the user data to be predicated, carrying out prediction of potential users by utilizing a prediction model pre-established based on a machine learning algorithm; and outputting a prediction result. According tothe technical scheme, product potential users are obtained through the prediction model pre-established based on the machine learning algorithm, so that compared with a manual screening mode, the method greatly improves screening efficiency of the potential users, and effectively saves labor and time cost.

The invention provides a future working condition prediction method based on front vehicle running information under an intelligent network connection system, which belongs to the technical field of intelligent traffic. The method comprises the steps of front vehicle running working condition data acquiring, front vehicle running working condition dividing, future working condition prediction model establishing and future vehicle working condition online predicting. The working condition prediction method can accurately acquire the front vehicle working condition information closest to the future working condition information of the vehicle, and accordingly establishes a future working condition prediction model which combines a least squares support vector machine and self-regressive sliding average error correction and is provided with prediction model precision judgement to carry out future working condition prediction of the vehicle. A prediction result has the advantages of good instantaneity, high accuracy and high reference performance. The prediction method is especially suitable for a hybrid vehicle running on a fixed line.

The invention relates to a protein structure information prediction method, apparatus and device, and a storage medium, and relates to the technical field of biological information. The method comprises the following steps of: obtaining multi-sequence alignment data by performing sequence alignment query for an amino acid sequence of protein in a first database; performing feature extraction on the multi-sequence alignment data; after obtaining initial sequence features, processing the initial sequence features by a sequence feature amplification model to obtain amplification sequence featuresof the protein; and predicting structural information of the protein according to the amplification sequence features. According to the scheme, when the structural information of the protein is predicted based on artificial intelligence, the prediction efficiency of the structural information of the protein can be improved under the condition of ensuring the prediction accuracy of the structuralinformation of the protein.

The invention discloses a prediction method, a device and a processor for electricity consumption. The method comprises the steps of according to a preset prediction model, adopting historical direct supply electricity consumption data corresponding to a known electricity consumption collecting period, or the historical managing area electricity consumption data in a preset time period before the known electricity consumption collecting period, and / or the historical national electricity consumption data for prediction, acquiring the managing area electricity consumption data of each known electricity consumption collecting period and / or the prediction result of the national electricity consumption data, and acquiring the average relative error corresponding to the prediction result; then determining a calibration prediction model corresponding to the known electricity consumption collecting period according to the average relative error; finally selecting the calibration prediction model corresponding to the electricity consumption collecting period to be predicted, and acquiring the managing area electricity consumption data corresponding to the electricity consumption collecting period to be predicted and / or the prediction value of the whole social electricity consumption data by the calibration prediction model. By the method, the national electricity consumption and the managing area electricity consumption can be predicted after the direct supply electricity consumption is obtained.

The invention belongs to the technical field of urban transportation informatization, and particularly relates to a traffic jam time calculation method for a transition road segment. The method comprises the following steps that 1, a radar detector is installed and debugged; 2, the section flow of the upstream segment and the section flow of the downstream segment are acquired; 3, the traffic capacity of the transition road segment is calculated; 4, the queuing duration time is calculated; 5, the queuing dispersing time is calculated; 6, the traffic jam time is calculated. By means of the method, a traffic commanding center can remind a driver to reasonably set the car speed or reselect a travel route in advance, and then the maximum traffic capacity of the transition road segment is achieved on the premise that safe operation of the transition road segment is guaranteed. Even different emergency measures can be taken according to the different traffic jam times, so that reliable running of a whole traffic system in daily operation is guaranteed.

The invention discloses a digestive tract endoscope image processing method and device, a storage medium, equipment and a system, belongs to the technical field of artificial intelligence, and relatesto a computer vision technology and a machine learning technology. The digestive tract endoscope image processing method comprises the following steps: acquiring a to-be-detected digestive tract endoscope image; classifying the digestive tract endoscopic images to be detected based on a first model, the first model being obtained by training based on a first training data set under the constraintof a second model, the first training data set comprising a pure data set and a noise data set, and the second model being obtained by training based on a second training data set before training thefirst model; wherein the pure data set comprises sample images with consistent annotations, and the noise data set comprises sample images with inconsistent annotations, and the second training dataset is a subset of the first training data set and comprises a pure data set, and the sample images are digestive tract endoscope images. According to the digestive tract endoscope image processing method, the training data volume is increased, and meanwhile, the influence of label labeling errors on the model prediction accuracy can be reduced.

The invention discloses a fresco scaling damage assessment method based on a near-infrared hyperspectrum. The fresco scaling damage assessment method comprises the following steps: obtaining the images of a specific fresco under hundreds of continuous different spectral wavelengths by use of a near-infrared hyperspectral imaging system; observing and locating by use of naked eyes, finding out key positions under different orders of severity of the scaling damage in the fresco, marking the corresponding positions in the corresponding hyperspectral images, and extracting spectral data as a damage spectral standard library of the degrees of scaling; storing and preprocessing the hyperspectral data; and extracting and analyzing the spectral characteristic data. According to the fresco scaling damage assessment method, the visual display of the distribution and the orders of severity of the fresco scaling damage is realized by obtaining the near-infrared hyperspectral images of the fresco and extracting and analyzing the spectral characteristics. The operation process is simple and the reliability of the obtained data is high. Meanwhile, due to noncontact imaging, the fresco surface is not damaged.

The present invention relates to the field of traffic information release and traffic management and control and discloses a road traffic flow parameter prediction method based on granular computing. The method comprises the steps of (1) replacing a data point by information particulate to be the basic unit of data mining analysis, (2) with granular computing ideology throughout the whole prediction framework, taking granular processing as a data processing method with a unified structure, allowing a policy maker to clearly understand the positions of various forms of systems in mutual interaction, grasping the communication mode of the systems, and establishing an enhanced harmonious environment among different ways, (3) with a fuzzy time series and the Gath-Geva cluster theory as the basis, by focusing on the commonalities of existing formal methods, recognizing the orthogonality of an existing good frame ( such as the probability theory and the probability density functions of various variables), with variable granularity concept as a basis, establishing the interval length analysis model of a granularity range according to a numerical entity, and thus realizing pattern recognition and speculation on the above basis.

The invention discloses a regional flood early warning method which comprises the following steps: step 1, carrying out image aerial photography on a waterlogging region by adopting an unmanned aerialvehicle, and generating a ground DEM digital elevation model according to the image; 2, according to the DEM, the runoff producing areas of the points prone to waterlogging and the concave land are calculated, and the water blocking building is recognized; Step 3, establishing a relation curve of the river water level, the rainfall and the drainage capacity according to the historical data information to form a waterlogging model; Wherein the input data of the model are river water level and rainfall, and the output data are waterlogging water depth and flood inundation range; 4, dynamicallydisplaying the real-time flood inundation range in a three-dimensional real-scene model according to the three-dimensional real-scene model generated by the aerial image; In the prior art, only rainstorm is considered in regional flood early warning, factors such as poor drainage and the like are not considered, and therefore the problems that large deviation exists in regional flood early warning, and flood early warning precision is not high are solved.

The invention belongs to the field of Chinese language data processing, and particularly relates to a hierarchical evaluation modeling method for readability of a simplified Chinese text. The hierarchical evaluation modeling method for the readability of the simplified Chinese text comprises the following steps: creating a standard corpus; extracting text features; and constructing a readability formula and evaluating the effect of the formula. On the basis of an existing Chinese readability formula, text characteristics of three layers of Chinese characters, vocabularies and sentences are selected, and a Chinese text readability formula which is suitable for simplified Chinese mother language in the primary school stage and has a grade division function is constructed.

The invention discloses a traffic travel behavior regulation and control method based on a block chain technology and travel plan sharing. The method comprises the following steps: firstly, enabling atraveler to share a travel plan of himself / herself before traveling; secondly, predicting traffic network dynamic traffic demands in different time periods in the future according to a travel plan shared by travelers; thirdly, evaluating the operation level of the urban traffic network based on the prediction result of the dynamic traffic demand and the urban traffic supply information; optimizing a travel plan; randomly selecting travelers, and providing corresponding travel suggestions for the travelers; and enabling the traveler to select whether to accept the travel suggestion or not, andperform dynamic traffic demand prediction based on the selection result after completion, and repeating the cycle until the operation level of the urban traffic network reaches a satisfactory level.According to the method, a hash algorithm is adopted for encryption, a travel plan is shared through public key and private key technologies of a block chain, and the privacy of travelers is protected; and travelers are motivated to share the travel plan and accept travel suggestions.

The invention belongs to a nitrate concentration prediction method, and particularly relates to a nitrate concentration prediction method based on an iPLS-PA algorithm. The problems of high price, complex operation, long analysis time, reagent consumption, secondary pollution to a water body and the like of a traditional nitrate determination method are solved. Cross validation modeling is carriedout by utilizing the peak area of each sub-interval and the corresponding nitrate concentration, and an optimal peak area calculation interval is selected to obtain higher nitrate concentration prediction precision.

The invention provides a hybrid wind power prediction method and device based on statistics, a storage medium and computer equipment. The method comprises the steps: training a prediction model by using various historical numerical weather prediction source data and historical power data, wherein the training process only needs to use historical data within a short time, the requirement for training data is low, and the training period is short; after training is finished, testing the trained prediction model by using prediction meteorological data provided by various historical numerical weather prediction source data in a second preset time period, so that whether the relationship among various parameters in the training stage is established or not is verified, and the prediction accuracy of the prediction model is ensured; under the condition that the relationship among the parameters is established, predicting the wind power of various prediction meteorological data; and finally, combining various prediction results to obtain a final wind power prediction result, so the prediction model can obtain a more accurate prediction result under the condition of having less historical data.

The invention discloses a method for predicting the rheological parameters of fresh concrete through a slump test. The method comprises the following steps: (1) placing a base plate on a horizontal surface; (2) fixing a parallel-light-source flat plate and a graduated flat plate on two opposite sides of the base plate; (3) placing a slump cone at the center of the base plate and allowing the axisof the slump cone to be superposed with the center of the base plate; (4) turning on a parallel lattice source and cameras; (5) pouring fresh concrete into the slump cone; (6) separating two halves ofthe slump cone to two sides at the same time so as to allow the concrete to begin to slump; (7) recording the changes of the height and radius of the concrete with time by using the cameras, and thenreading the values of slump height and radius at each time in subframes; and (8) picking out points where a radial change rate is more than 5 times the change rate of height, substituting corresponding height, radius and time values to obtain a curve, and then subjecting the curve to fitting to obtain a straight line so as to calculate an intercept and a slope, wherein the intercept is yield stress and the slope is plastic viscosity. The method of the invention simplifies the testing process of rheological parameters and reduces testing cost.

The invention discloses a hybrid neural network training method, system and device and a storage medium, and belongs to the field of machine learning algorithms. In the training scheme provided by the invention, a gradient calculation method and an error transmission method of a gradient descent-based training method are used, and a self-adaptive variable-step-size Manhattan rule-based training method in the training process is adopted to update the network weight, so that the network prediction accuracy and the convergence speed of the network are considered; compared with a Manhattan-based neural network training algorithm, the method can achieve a higher convergence speed, and compared with a training algorithm based on stochastic gradient descent, the method can achieve a higher accuracy rate. A hybrid neural network training mode provided by the invention is low in complexity and high in convergence speed, can ensure the network prediction accuracy of a trained neural network, and can be more suitable for online training of a neural network calculation circuit based on a resistive random access memory ReRAM.

The invention discloses a method for predicting fall risk of elderly people. The method comprises the following steps: dividing a plantar pressure area into a hallux area, a second-fifth toe area, a forefoot area, a middle foot area and a heel area; dividing a support phase into an initial contact section, an initial metatarsal bone grounding section, an initial forefoot flat section, a heel off-ground section and a final contact section; then, based on the plantar pressure areas and the supporting phase, carrying out plantar pressure testing on a subject by using a Footscan plantar pressure flat plate testing system so as to obtain pressure change curves of the different plantar pressure areas in all supporting phase periods; constructing a deep neural network model by using a convolutional neural network and a recurrent neural network, training the prediction model, and selecting an optimal prediction model as a foot pressure prediction model; and finally, inputting the pressure change curves into the foot pressure prediction model to obtain a prediction value. The method has the characteristics of high data measurement precision, various characteristic indexes and good prediction accuracy.

The present application relates to data processing techniques and provides a method, apparatus and computer device for predicting potential users based on lost users. The method comprises the following steps of: acquiring data of a lost user in a specified time period; According to the acquired data, predicting by a plurality of potential user prediction models to obtain a prediction tag corresponding to the potential user prediction model; Determining a target prediction tag which satisfies a preset condition in each prediction tag; When the target prediction label is a specified target label, determining the lost user to be a potential user. The method can ensure the stability of the prediction results of potential users, so as to ensure the accuracy of the prediction.

The invention discloses a vehicle driving condition prediction method based on working condition characteristics. The vehicle driving condition prediction method comprises the steps of: performing characteristic parameter extraction of historical condition data of a vehicle, carrying out clustering analysis, and establishing a condition characteristic parameter database; according to the actual working conditions of the vehicle, constructing a relationship between the working condition characteristic data and road and traffic characteristic parameters, and establishing a road and traffic-basedworking condition characteristic parameter prediction model; acquiring road and traffic characteristic parameters of a determined driving route according to the determined driving route, and predicting the working condition characteristic parameters by using the prediction model; and comparing predicted working condition characteristic parameters with the working condition characteristic parameters in the database to obtain the working condition of the driving route to be driven. According to the vehicle driving condition prediction method, the universality and accuracy of the prediction model are ensured through model prediction according to the road and traffic characteristics on the planned route of the vehicle, and meanwhile, the influence of the road and traffic conditions on the vehicle driving state can be reflected.

The invention discloses a disk fault prediction method based on time sequence feature processing and model optimization. The disk fault prediction method is characterized in that the method comprisesthe following steps: obtaining SMART attribute data of a disk and a timestamp of the SMART attribute data; acquiring expansion data according to the acquired standard value and the original value of the SMART attribute data of the disk and the timestamp of the SMART attribute data; selecting a plurality of features from the extended data and the standard value and the original value of the SMART attribute data by using a principal component analysis method; and constructing a multi-dimensional matrix, inputting the obtained multi-dimensional matrix into the trained random forest model to obtain a fault prediction result of the disk, and updating the random forest model according to the obtained fault prediction result of the disk to obtain an updated random forest model. According to the method, time sequence feature processing and model optimization are utilized, so that the technical problem that the accuracy of disk fault prediction is relatively low due to the fact that the incidence relation between SMART attributes is not considered in the existing SMART technology is solved.

The invention discloses an air quality prediction method and system based on data mining. The system is used for achieving the air quality prediction method based on data mining. The method comprisesthe following steps of acquiring a wind speed level, a fine particulate matter pollution value and a local pollutant pollution value, and taking the three pieces of information as the input parameters; standardizing the input parameters; and inputting the standardized input parameters into a BP neural network model to predict the air quality, wherein the local pollutants are selected according tothe main pollutant type of the prediction region, the BP neural network model is of a three-layer design composed of an input layer, a hidden layer and an output layer, and the number of the nodes ofthe hidden layer is 5, 6 or 7. The network model provided by the invention is simple in structure and low in computing resource consumption, can realize the accurate prediction, and also has the characteristics of high convergence speed and strong network generalization capability.