123results about How to "Improve the speed of prediction" patented technology

The invention discloses a multi-basin real-time intelligent water quality predication method and system. According to the multi-basin real-time intelligent water quality predication method and system, an NARX model is optimized through a genetic algorithm and a problem that earlier stage parameters of the NARX model are uncertain; a GA-NARX model is stored or called through a model storage module to predicate the multi-basin water quality condition in real time; essential data, recent data and history emergency data are performed on three-section training and accordingly an optimized GA-NARX model can basically comprise river pollution characteristics and the predication accuracy is improved; weather data replaces hydrological data and is performed on obfuscation processing and accordingly influences to the model from flow data missing are effectively solved; matching of similar pollution time sequence templates is performed through an improved DTW algorithm, a similar pollution process in the basin history is rapidly found out, experience is referred and learned, and sudden emergency accidents are accurately predicted. The multi-basin real-time intelligent water quality predication method and system can be widely applied to the water quality predication field.

The invention discloses an empirical mode decomposition and Elman neural network combined wind power forecasting method, comprising the following steps: screening samples which are used for forecasting a wind power field, and selecting wind power outputs of forecasting periods within fluctuation months to implement forecasting; implementing empirical mode decomposition on multiple groups of output time sequence sample data of the wind power field, and ensuring that each group can obtain multiple intrinsic mode functions (IMFs) and trend components Res according to decomposition termination conditions; implementing fluctuation degree classification on decomposed IMFs according to a run distinguishing method, and reconstructing the IMFs according to a similar fluctuation frequency principle to obtain total high-frequency components and total low-frequency components; establishing an Elman neural network model, and implementing data normalization on the total high-frequency components, the total low-frequency components and the trend components to obtain training and test data of a neural network; and implementing day-ahead power forecasting for 72h by adopting an Elman improved learning algorithm to obtain a day-ahead forecasting power value of 72h of target wind power outputs. By adopting the empirical mode decomposition and Elman neural network combined wind power forecasting method disclosed by the invention, the number of forecasting components can be reduced, and the forecasting accuracy and forecasting speed can be increased.

The invention discloses a method for predicting a line loss rate of a power distribution network. The method comprises the steps of determining multiple electrical characteristic parameters, influencing the line loss rate, of the power distribution network, performing standardization processing on parameter values of the electrical characteristic parameters, and performing normalization processingon the line loss rate; by taking the parameter values of the electrical characteristic parameters as inputs of an input layer, and taking the value of the line loss rate as an output of an output layer, building an initial neural network model, wherein the initial neural network model comprises at least one hidden layer; determining the number of hidden layer nodes; by dynamically adjusting an inertial factor and a learning factor of a particle swarm algorithm, improving the particle swarm algorithm; by utilizing the improved particle swarm algorithm, optimizing a weight value and a thresholdvalue of the initial neural network model to determine an optimized neural network model; and inputting the electrical characteristic parameters of a power distribution network line to the optimizedneural network model, and predicting the line loss rate corresponding to the electrical characteristic parameters by utilizing the optimized neural network model.

The embodiment of the invention provides an echo state network based prediction method and a prediction device, and relates to the field of computers. The prediction method and prediction device can effectively improve the prediction performance of the ESN (Echo State Network), and increase the speed and the accuracy of prediction. The prediction method comprises the following steps: firstly, establishing an echo state network prediction model, wherein nerve cells in a dynamic pool of the echo state network prediction model serves as wavelet nerve cells; acquiring a training set according to the input data, utilizing the training set to carry out prediction training on the echo state network prediction model, and obtaining the trained prediction model; predicting according to the trained prediction model, so as to obtain output data. The echo state network based prediction method and the prediction device disclosed by the embodiment of the invention can be applied to prediction of nonlinear chaotic time series data.

The invention discloses an adenosine triphosphate binding site predicting method for protein. The adenosine triphosphate binding site predicting method for the protein comprises the following steps: firstly, acquiring evolution information and secondary structural information of the protein by using an IPSI-BLAST and PSIPRED program, and extracting characteristics of each amino acid residue by a sliding window technology; secondly, performing random downsampling on non-binding site samples for multiple times by a random downsampling technology, sample of several; thirdly, training an SVM (support vector machine) based on a non-binding site sample subset obtained in each random downsampling and a binding site sample set, and performing random downsampling on all the sample sets to obtain a plurality of SVMs; and finally, integrating the trained SVMs through Dempster-Shafer theoretic evidence. The adenosine triphosphate binding site predicting method for the protein has the advantages as follows: by a random downsampling technology, the scale of a training set can be effectively reduced and the model training speed can be effectively increased; and by an SVM integrating technology, information loss caused by downsampling can be effectively reduced and the model predicting precision can be effectively improved.

The invention provides a subspace fusion-based protein-vitamin binding location point predicting method. The method comprises the steps of feature extracting and feature combining: extracting evolution information, secondary-level structural information the binding tendency information of a protein by respectively using a PSI-BLAST, a PSIPRED and a protein-vitamin binding location point tendency table, and converting amino acid residues in a protein sequence into a vector presentation mode by a sliding window and serial combination; using a multi-feature selection algorithm to perform feature selection on an original feature space for multiple times; forming a feature subspace by feature subsets obtained from feature selection every time and establishing multiple feature subspaces; training one SVM (support vector machine) classifier for each obtained feature subspace; fusing multiple SVM classifiers which are trained by a weighted average classifier fusing mode; performing protein-vitamin binding location prediction on the protein to be predicted based on a fused SVM predictor. The prediction method is fast in prediction sped and high in prediction precision.

The invention discloses a flight delay real-time probability prediction method based on a Bayesian network algorithm, and the method comprises the steps: formulating a flight delay judgment standard,analyzing a delay wave and the impact on flight delay, and determining the release fairness of departure flights; analyzing the delay characteristics, determining flight delay factors, and creating aflight delay dynamic prediction model based on a Bayesian network; adopting a dynamic prediction technology based on a time sequence to predict the present transverse wave and measurement index to obtain a final flight delay prediction value, and generating a prediction set; and carrying out probability prediction on prediction set data by utilizing the flight delay dynamic prediction model obtained by training, and obtaining a prediction value of each flight delay level by adopting a probability maximum principle. According to the invention, real-time probability prediction can be carried outon the departure delay level of a single flight of an airport every day, the flight delay prediction precision is improved, a delay early warning notice is issued to passengers in time, an operationstrategy is adjusted in time, and various adverse effects caused by flight delay are reduced.

The invention discloses a C-RAN carrier migration resource demand prediction method based on improved PSO. The method comprises the following steps: acquiring and preprocessing data, and generating a training sample set of a prediction model; determining the structure of a GRNN prediction model according to the generated training sample set; training the GRNN prediction model by using an improved PSO algorithm; and prediction the resource demand of a carrier virtual machine at a next moment by employing the trained GRNN prediction model. The problem in the existing carrier migration technology that resource allocation is carried out only depending on the load condition of the carrier virtual machine at the current moment, resulting in resource waste or resource shortage to cause short-term secondary migration is effectively solved; and in view of the defect that the existing prediction method cannot be applicable to the C-RAN scene with high instantaneity due to an overlong prediction time, the improved PSO algorithm is proposed to train the prediction model, so as to effectively improve the prediction accuracy and the prediction speed in resource demand prediction.

The invention discloses an oil and gas dessert prediction method and a storage medium, and relates to the oil gas exploration field. The method comprises the following steps: obtaining logging data and various earthquake attributes of an oil gas dessert area; obtaining an oil and gas dessert area indication curve according to the logging data, and using the indication curve and various earthquakeattributes as machine learning samples; optimizing features of the learning samples; carrying out machine learning according to optimized learning samples, thus obtaining a prediction model; predicting oil gas desserts in a random target area according to the prediction model. The invention provides the oil and gas dessert prediction method and the storage medium, aims to improve the prediction accuracy, and can improve the prediction resolution, thus quantitatively predicting oil gas desserts, preventing artificial factor influences, and simplifying a calculation process; the method can obtain high resolution and high precision prediction results without making complex redundant calculations.

The invention provides an optimization method of a convolution neural network. The optimization method comprises steps of setting a shortcut connection on an added layer of the convolution neural network and acquiring residual error mapping corresponding to the shortcut connection; according to the residual error mapping, determining expectation mapping corresponding to the shortcut connection; and replacing a layer corresponding to the shortcut connection with the expectation mapping, and carrying out convolution neural network model prediction. According to the invention, parameters of the added layer can be effectively reduced; data circulation between networks is allowed to be quite smooth; and improvement of prediction precision and prediction speed of the model is facilitated.

The embodiment of the invention provides a prediction method and a prediction device of an echo state network (ESN) and relates to the technical field of a communication service network so that the prediction performance of the ESN can be improved effectively and the prediction speed and the prediction precision can be improved. The method is as follows: through establishment of a prediction model of the ESN, wherein a dynamic pool of the prediction model is a mixed neuron dynamic pool which includes wavelet neurons, a training set is obtained from input data and prediction training is carried out on the prediction model according to the training set and then the prediction model which completes training is obtained; and through prediction input data, the prediction model which completes training is predicted and then prediction output data in a prediction process is obtained. The prediction method and the prediction device are used for prediction of nonlinear chaotic time sequence data.

The invention provides a thermal power plant condenser vacuum degree prediction method based on multilayer LSTM. Firstly, a sample data set is constructed, input data and output data in a sample are subjected to standardization processing respectively, then the input data sequentially enter a multi-layer long-short-term memory neural network, and model training is performed by using an adaptive moment estimation algorithm; and finally, to-be-predicted data is input into the trained prediction model, and the vacuum degree of the condenser is predicted. According to the method, the long-term andshort-term memory neural network structure is applied to condenser vacuum degree analysis, deep mining of big data is achieved, the accuracy and speed of data prediction are improved, and use of system resources is optimized. According to the two-layer LSTM structure designed by the invention, the model depth can be increased, the model capacity can be increased, the prediction precision can be improved, and meanwhile, the adaptive moment estimation algorithm is used for optimizing the traditional gradient descent algorithm, so that the problem of gradient explosion is avoided.

The invention discloses an intelligent working face coal-rock interface recognition method based on geological data. The method includes the first step of coal seam three-dimensional modeling based on geological data, the second step of coal seam primary mining and coal-rock interface recognition, and the third step of coal seam subsequent mining and coal-rock interface recognition. The first step includes the substeps that 101, geological data is obtained through actual measurement, wherein actual measurement geological data of to-be-mined coal seam is obtained and includes coal seam geological data, mining roadway geological data and cut hole geological data; 102, a coal seam model is constructed, wherein data storage, interpolation operation, three-dimensional coordinate database generation and model construction are conducted. The third step includes the substeps that the to-be-mined coal seam continues to be mined by means of a coal mining machine from back to front, and a coal-rock interface of the working face is recognized before each working face is mined. The method is simple in step, reasonable in design, convenient to implement and good in use effect, and the coal-rock interface is conveniently, quickly and accurately recognized according to the actual measurement geological data and cylinder height adjustment track prediction data.

The invention discloses a method for predicting elastic cloud computing resources based on an SARIMA-WNN model. The method comprises the steps that complementary advantages are achieved by using a seasonal autoregressive integrated moving average (SARIMA) model combining with a wavelet neural network (WNN) prediction model to improve the prediction accuracy; according to the SARIMA, seasonal periodic factors are added on the basis of an ARIMA model, periodic cloud resource demand data of a past section is input to a SARIMA (q, d, q)(P, D, Q) s model to obtain d, p, q, D, P and Q respectively;prediction is conducted on tranquilized and sequenced codes by using the SARIMA model, and a prediction result is marked as and an L residual value is marked as rt, wherein the prediction result andthe residual value can be obtained through the prediction; a model which conforms to elastic cloud source prediction is obtained through conducting training on the WNN network by using training samples, prediction is conducted aiming at the residual sequence rt, and the prediction result is marked as ; finally the prediction result of the SARIMA-WNN combined model is obtained. By means of the method, the problems of inaccuracy of a single model, poor effects of other combined models and the like are solved.

The invention discloses a method for discriminating corn haploid kernels based on a near-infrared spectrum technology. The method comprises the following steps of: reading into an original spectrum file and acquiring spectrum intensity data; dividing data sets; carrying out feature normalization on spectrum data; adopting a partial least-squares regression method to reduce the dimensionality; adjusting parameters of a neural network classifier; carrying out fine adjustment and performance evaluation on the neural network classifier; and storing the parameters of the neutral network. The method disclosed by the invention has the advantages that the corn haploid kernels can be identified from heterozygote kernels, and simultaneously higher correct discrimination rate and model stability are ensured.

The invention relates to the technical field of wind power, and particularly relates to a short-term wind power combined prediction method. The short-term wind power combined prediction method is based on a grey system and a least square support vector machine. The short-term wind power combined prediction method is characterized by comprising the following steps that original wind power data are preprocessed and grey prediction is performed so as to obtain a residual error sequence, and then the residual error sequence is predicted by using a least square support vector machine model so as to obtain a new residual error value; a kernel function is selected, and the regression parameters of the least square support vector machine are determined by using a cross validation method; after a data set is obtained, a radial basis function is selected to act as the kernel function, including width parameters and optimization parameters of quadratic programming; a combined prediction model is constructed; the data set is inputted and a prediction function is generated; and prediction error evaluation analysis is performed. Prediction accuracy and speed can be enhanced by the method.

The invention relates to a heat exchanger filth blockage detection method and device. The method comprises the steps of collecting the actual operation data of a heat exchanger during operation, and obtaining the theoretical operation state data through calculation according to the actual operation data and a predetermined operation state model, wherein the operation state model is determined according to operation data and operation state data when the heat exchanger is not blocked by dirt; acquiring actual operation state data corresponding to the actual operation data when the heat exchanger operates; and according to the actual operation state data and the theoretical operation state data, determining whether filth blockage occurs to the heat exchanger. According to the method, the filth blockage degree of the unit heat exchanger can be judged based on historical data, the judgment effect and the detection speed are improved, the problems of too large algorithm error and insufficient stability caused by few monitored operating parameters are solved, the accuracy of filth blockage judgment of the heat exchanger is improved, and the development trend can be summarized through historical operation data and real-time operation data, so that the filth blockage condition of the heat exchanger in the future is accurately predicted.