393results about How to "Fast prediction" patented technology

The invention discloses a power grid security situation predicting method based on an improved deep learning model and belongs to the technical field of power system safety. The power grid security situation predicting method includes: performing power grid security situation evaluation through power grid data collection and preprocessing; aiming at the characteristic that indicator data of power grid security situation evaluation are high in relevance and dimension, providing an improved self-coding network method to lower dimension of the indicator data, and utilizing a data sample after dimension reduction and a power grid security situation value corresponding to a next time monitoring point; adopting an improved deep belief network to build a deep learning situation predicting model with multi-input and multi-output for power grid security situation prediction. By the power grid security situation predicting method, speed and accuracy of power grid security situation prediction can be improved effectively.

The invention discloses a machine fault prediction method based on MFCC feature extraction, and belongs to machine fault prediction methods. The machine fault prediction method comprises the steps that the feature of a current acoustical signal of running of a machine is obtained through an acoustical sensor installed on the machine, the acoustical signal is preprocessed, and then Mel conversion is carried out on the preprocessed acoustical signal to obtain an MFCC feature vector of the acoustical signal; according to the obtained MFCC feature vector, prediction is carried out on the health condition of the machine, the specific clustering process is that a SVM conducts clustering on the MFCC feature extracted when the machine runs and stored sample data obtained when the machine runs normally, the clustering result is analyzed through a vote method, and then the machine fault is predicted. The machine fault prediction method based on the MFCC feature extraction has the advantages that the acoustical feature of the machine is extracted and converted into the Mel domain, then clustering analysis is carried out on the feature vector through the SVM, the health condition of the machine can be rapidly, accurately and easily predicted, operation is easy, prediction precision is high, the prediction speed is high, the anti-noise performance is good, and nonlinear, random and time-varying signals can be accurately predicted.

The invention discloses an on-line prediction method for residual life of rolling bearing, As that roll bearing move from a healthy state to a damaged state, The original signal samples and corresponding degeneration energy indexes are extracted from the running process of the bearing, and the original signal samples are used as the input of the five-layer convolution neural network model, and thedegeneration energy indexes are used as the output of the convolution neural network model, and the degeneration energy state model is obtained by training. Real-time acquisition of the original running signals of the rolling bearings to be tested; The original running signal of the rolling bearing to be tested is input into the degradation energy state model, and the degradation energy index isestimated. Then the estimated energy degradation index is used to predict the residual life of the rolling bearings to be tested. The prediction process of the invention only needs to collect the original operation signal of the bearing, and does not need to extract and screen the features, thus overcoming the technical problems that the prior art adopts the methods of feature extraction, featurescreening and regression prediction, which have the characteristics extraction difficulty and the precision is limited.

The invention discloses a target detection method, a target detection model and a target detection system based on a cascade detector. The target detection method comprises the following steps of S1,training the target detection model by adopting a training data set with a target label; S2, inputting a to-be-detected picture into the trained target detection model, and performing feature extraction, upsampling and feature fusion to obtain feature maps P2, P3,..., Pn with different sizes; S3, predicting according to the feature map Pn to obtain a target position Bn and a category Cn; extracting corresponding features from the corresponding feature map Pi-1 according to the target position Bi for target prediction to obtain a target position Bi-1 and a category Ci-1, i = 3-n; S4, taking thetarget position B2 as a final prediction result. According to the invention, through the multiple iterative regression prediction of the the target candidate box, the target position with more accurate prediction is obtained, the target detection precision is improved, and the detection accuracy is higher.

The invention discloses a short-term power load forecasting method based on a fuzzy clustering similar day. The method includes that firstly meteorological factors are divided into temperature, pressure, wind speed, rain and other occasions and then constitute influence factors of the similar day with week styles and date styles, a fuzzy coefficient characteristic mapping table is built through fuzzy rules, clarification is carried out by the method of fuzzy clustering based on the preceding steps, the similar day is chosen according to clustering levels, according to obtained load data of the similar day, load sequences are projected to different scales and low frequency components are obtained by means of wavelet transformation, a support vector machine is optimized by means of a particle swarm optimization algorithm to achieve forecasting for a short-term power load low frequency portion, and forecasting for a high frequency portion is achieved by the method of weighted average. Eventually, application researches are carried out by means of the load data of a power grid in Shanghai city, and good forecasting effects can be achieved in weekdays, at weekends and in holidays.

The invention provides a comment text aspect-level sentiment classification method based on deep learning. The method comprises the following steps: preprocessing a comment text, including word segmentation and stop word removal, balancing aspect words and corresponding tags to generate a balanced sample, and vectorizing the balanced sample and Chinese words in an original sample to obtain word vectors in the balanced sample; inputting the word vectors into the model to predict a comment result, wherein the model is a deep learning model constructed according to a deep neural network, the similarity calculation is carried out on word vectors of aspect words and other words of sentences, and an aspect emotion semantic matrix of a balance sample is generated. According to the method, throughthe balance processing and construction of the Attn-Bi-LCNN model, the emotion semantic matrix can be effectively output, and the accuracy of the model and the prediction speed in practical application are improved, so the method is suitable for aspect-level fine-grained emotion classification of texts.

The invention provides an online advertisement recommending system and method for large-scale medium data and relates to the technical field of the calculation advertisement science. An advertisement dispatch engine module in the online advertisement recommending system is respectively connected with a user side, an advertisement management module and a flow analysis module. Parameter exchange is carried out between the flow analysis module and an advertisement searching module, a user behavior inquiry module and a webpage management module. A user behavior mining module is respectively connected with the advertisement management module and the user behavior inquiry module. The advertisement management module is connected with the advertisement searching module. According to the online advertisement recommending method, when a user finish accessing a webpage, the user is identified according to user information, user interests are inquired, user behaviors are learned, matched advertisements are searched for according to the predicted user behaviors, and finally the online advertisements are recommended to the user. The system has the good self-learning ability, can effectively improve the intelligent level of advertisement recommendation, and is suitable for online advertisement recommendation under the background with the large-scale data.

The invention discloses a novel method for discovering a drug lead compound through an image recognition system based on a convolutional neural network so as to solve the problems that an existing lead compound is low in virtual screening efficiency and not high in accuracy. The method comprises the steps that firstly, the structural formula of the compound is converted into planar pictures, black and white processing and color inverse processing are performed, all the pictures are classified according to liveness property of the compound, and digital labels are added according to categories and input to a system; some pictures are selected as a training set for deep learning of classification problems for the convolutional neural network, and the other pictures are adopted as a testing set to evaluate a model. After learning is completed, pictures, except for the testing set and the training set, obtained after the identical processing is performed are input to be calculated by the system, and the probability of the liveness property corresponding to the pictures is predicted.

The invention discloses a Hadoop framework-based short-term load prediction method for a distributed BP (Back Propagation) neural network. The method specifically comprises the steps of obtaining an initial load data set; dividing the load data set into small data sets and storing the small data sets in data nodes of a distributed file system; initializing BP neural network parameters and uploading a parameter set into the distributed file system; training the BP neural network according to a current load sample, and obtaining correction values of a weight and a threshold of the BP neural network in the current data set; performing statistics on sum of weight and threshold parameters of all layers and between the layers of the network according to a key value of a key value pair; judging whether the convergence precision or the maximum iterative frequency is reached or not in a current iterative task, and if yes, establishing a distributed BP neural network model, or otherwise, performing correction of the weight and threshold parameters of the network; and inputting prediction day data and obtaining load power data of a prediction day. According to the method, the load prediction speed is increased and the requirements of load prediction precision are met.

The invention discloses an IGBT module junction temperature online estimation circuit system and method, and the method comprises the following steps: 1, carrying out the online measurement of the IGBT on-state voltage drop VCE(ON) of an IGBT power module, and building an IGBT junction temperature Tj lookup table; 2, calculating the power loss of an IGBT and a diode in the IGBT power module; 3, periodically and timely updating the IGBT equivalent thermal network model by adopting a thermal network updating strategy considering the fatigue accumulation effect of the solder layer; 4, establishing a state space model by taking the junction temperature Tj and the IGBT and diode power loss as state variables according to the IGBT equivalent thermal network model, and updating the state space model; and 5, designing an adaptive Kalman filtering algorithm program according to the state space model to obtain an IGBT junction temperature Tj estimated value. The method integrates the advantagesof model estimation and online measurement, overcomes the defect that on-state voltage drop VCE (ON) measurement is inaccurate, keeps signal continuity, reduces noise interference, is high in estimation precision, and can estimate the IGBT module junction temperature in real time in a non-intervention mode.

The embodiment of the invention provides a method, device and apparatus for estimating the point cloud object attitude based on deep learning. The method comprises the steps of obtaining data requiredto be learnt; designing a network model; and performing model training and prediction. The design of the network model comprises the steps of modeling a point cloud object attitude estimation probleminto a non-distinctive multi-classification problem; designing a residual block structure to extract features; obtaining global feature through a maximum pooling layer according to extracted features; respectively sending the global features to three parallel multi-layer perceptrons to perform predicted category scoring on coordinate axes; performing final category prediction on features after predicted category scoring on the coordinate axes by using a classifier; carrying out equal-weight summation on loss values obtained after processing of the classifier, and enabling the sum to serve asan overall multi-classification loss function; and optimizing the multi-classification loss function by using adaptive moment estimation. The method provided by the embodiment of the invention can accurate estimate the point cloud object attitude so as to improve the accuracy of object attitude positioning and prediction.

The invention discloses a dredging operation yield prediction model analysis method based on a BP neural network. The dredging operation yield prediction model analysis method based on the BP neural network comprises the following steps that (1) data information influencing dredging operation yield factor variables is collected, p influence factors are determined, and a sample matrix is listed, wherein p is a positive integer; (2) pretreatment is conducted on sample data; (3) a network is established, and a training sample and a testing sample are determined; (4) the established network is trained according to the training sample; (5) according to the testing sample, the established network is tested; (6) the performance of the network is estimated by computing the offset condition between a predicated value and a true value. By the adoption of the dredging operation yield prediction model analysis method based on the BP neural network, a nonlinear mapping function from input to output is achieved, a nonlinear relationship is established between input and output, and an established model is high in fault-tolerant capacity and high in prediction speed; theoretical basis can be laid for optimization study of dredging operation yield, and the purposes of high efficiency, high yield and low energy consumption can be achieved.

The invention belongs to the technical field of agricultural product quality analysis, and particularly relates to a near-infrared detection method for the peanut quality and application. The method comprises the following steps that peanut samples are collected, physical and chemical testing is performed on the peanut samples, near-infrared scanning is performed on the peanut samples, denoising processing and preprocessing are performed on obtained light absorption values, obtained preprocessed light absorption values are analyzed, near-infrared spectrum characteristic wavelengths are obtained through screening, and a prediction model of the peanut quality is built through a stepwise regression method. According to the near-infrared detection method for the peanut quality and the application, the obtained information is intuitive and reliable, the characteristic wavelengths of the peanut quality are determined and are few in number, and the analytical method that the model is built through the characteristic wavelengths is applied, so that the model precision is improved; on the condition of the same prediction precision, the prediction speed is high; meanwhile, through the built near-infrared prediction model method for the moisture, protein, fat, total sugar and ash content of peanuts, the peanut quality can be analyzed more comprehensively, and usage and popularization are easy.

The invention discloses a construction method of a tilapia fillet quality evaluation model based on proteins and enzymes in muscle. According to the method, indexes of quality evaluation are determined according to factors causing quality changes of tilapia; the weight of each quality evaluation index is acquired by utilizing a fuzzy analytic hierarchy process, so that a relationship between subjective evaluation standard and objective evaluation standard can be established; and a quality evaluation model is established according to the weight, and an unknown sample can be subjected to quality evaluation. The method can effectively and subjectively evaluate the quality of tilapia fillet. The system can be used by manufactures and import and export companies in the industry to complete quality evaluation of the tilapia fillet.

The invention discloses a sequence characteristic analysis method for forecasting a miRNA target gene. The method comprises the steps of constructing related characteristics of 27 miRNA-target point pairing sequences on the basis of a CLASH experiment data set, and forming a characteristic set comprising 84 characteristic values by combining traditional characteristic; and performing machine learning by using a random forest model, and constructing a miRNA target gene forecast model to perform miRNA target gene recognition. The model constructed according to the method has the advantages of high accuracy, sensitivity, specificity and precision, and the miRNA target gene can be relatively and accurately forecasted.

The invention discloses an instant milling force prediction method in the variable-curvature curved surface side milling process. The method includes the steps of calculating position points of all tools along tracks of the tools at equal parameter intervals according to the variable curvature characteristic of a curved surface in the side milling process, calculating instant tool angular positions, feed directions, corresponding machining times and actual radial cutting depths, calculating the instant cut-in/cut-out angles according to the actual radial cutting depths, using the nominal feed engagement of each tooth as the actual feed engagement of each tooth, calculating the instant non-deformation cutting thickness of each cutting micro-unit through the combination with the instant tool angular positions, establishing a milling force model under a local coordinate system according to the feed directions, projecting the milling force model into a whole coordinate system, and then obtaining the instant milling force. By means of the method, the milling force prediction efficiency is high in the variable-curvature curved surface side milling process, and the milling force prediction is accurate.

The invention discloses a networked control system fault detection method based on neural network prediction, which comprises four steps of RBF neural network system building, system fault detection function building, system stability judgment and operation and system fault judgment and operation function building. The system building and operation process is simple, the operation efficiency and the operation precision are relatively high, an improved RBF neural network prediction controller is adopted to effectively predict system output data information, and thus, bad influences on the system by packet loss can be effectively cancelled, errors are smaller and training times are reduced through adjusting learning efficiency on the basis of adopting feedback correction on the obtained predicted output value for correction, and better convergence and quicker prediction speed can be obtained. Meanwhile, when fault happens to the system, happening of the fault can be quickly detected according to a designed fault observer and a judgment criterion.

The invention discloses a sewage energy saving processing optimization control method based on an improved firefly algorithm and a least squares support vector machine, and belongs to the field of intelligent control. The method comprises steps of using a multicore least squares support vector machine to model energy consumption and water quality of discharged water of a sewage processing factory; using the improved firefly algorithm to optimize established model parameters; and using the improved firefly algorithm to optimize a set value of the controller. According to the invention, the least squares support vector machine is used for modeling energy consumption and water quality of discharged water of a sewage processing factory; a multi-core idea is introduced; the improved firefly algorithm is used for optimizing model parameters, so accuracy of an energy consumption model and a discharged water quality model is greatly improved; the improved firefly algorithm is used for carrying out online optimization on set values of dissolved oxygen concentration and nitrate nitrogen concentration of the controller, so under the premise of meeting the discharged water quality, the energy consumption of the sewage processing factor is reduced; an objective of saving energy and carrying out optimization in the sewage processing process is obtained; and compared with other algorithms, the method is characterized by simple algorithm, few used parameters and high convergence accuracy.

The invention provides a two-stage ligand specificity protein-ligand binding area forecasting method. The method includes the following steps of firstly, forecasting protein-ligand binding residues in a protein sequence through a ligand specificity forecasting model and based on inputted protein sequence information and extraction and combination of multi-view characteristics; secondly, conducting spatial clustering on the binding residues obtained in the first step, conducting the spatial clustering through a spatial clustering algorithm, and obtaining one or more binding areas. The ligand specificity protein-ligand binding area forecasting method has the advantages that forecasting accuracy can be effectively improved by using the modularized ligand specificity forecasting model, the binding areas can be further obtained from the forecasted binding residues by using the spatial clustering algorithm, and therefore the protein-ligand binding area forecasting can be really achieved.

The invention discloses an audio scene recognition method based on a feature pyramid network. The method includes the steps of establishing a feature pyramid network model for audio scene recognition;training an audio scene recognition feature pyramid network model by using a training set containing the audio files of different scene categories and the corresponding scene categories; reading an audio file to be identified and cutting off the audio file; conducting extraction of Mel features, obtaining a two-dimensional Mel spectrogram of each audio frame, normalizing the two-dimensional Mel spectrogram, training the normalized two-dimensional Mel spectrogram for forward propagation of the audio scene recognition feature pyramid network model to obtain prediction probabilities for different audio scene categories, and taking the scene category with the maximum prediction probability as prediction output of the audio frame corresponding to the two-dimensional Mel spectrogram; and predicting the whole audio file which needs to be identified. According to the method, underlying feature information is fully utilized, and model performance is improved. Information brought by more and more data provided under the current big data trend can be fully utilized, and the prediction speed is high.