107results about How to "Solve forecasting problems" patented technology

The invention relates to a method and a system for predicting remaining service lives of different lithium ion batteries of the same type. The method comprises the following steps of 1) extracting health factors capable of reflecting the performance degradation of lithium ion batteries; 2) establishing a health factor prediction model, wherein the health factor prediction model is a relation modelof battery health factor service life early stage and service life later stage and constructed by utilizing a neural network; 3) constructing a battery capacity prediction model, wherein the batterycapacity prediction model is a relation model of the heath factors and the battery actual capacity and constructed by utilizing a neural network; and 4) taking the service life early stage health factors of different batteries of the same type to be predicted as input, obtaining a battery service life later stage battery capacity prediction value based on the health factor prediction model and thebattery capacity prediction model, and then working out the remaining service life value of each battery at the current moment. The method has the advantages that the accuracy and the adaptability are relatively high in the prediction of RUL of different batteries of the same type.

The invention relates to a pollution flashover index forecasting method for a regional power grid, belonging to the technical field of transmission and distribution monitoring. The method of the invention comprises the following steps: (1) forecasting the current value of the equivalent salt deposit density in real time by an insulator surface equivalent salt deposit density forecasting model; (2) forecasting pollution flashover critical voltage by an insulator pollution flashover critical voltage forecasting model; (3) forecasting the insulator pollution flashover index of the power grid by a pollution flashover graded forecasting and prewarning model; and (4) judging the state. The invention has the advantages of providing a multivariable equivalent salt deposit density time sequence forecasting model based on phase space reconstruction, solving by a supporting vector machine model, solving the forecasting problem under the conditions of small sample of equivalent salt deposit density data and noise, and improving the forecasting prediction.

The invention discloses a human movement mode speculation model, a training method and a speculation method based on variation track context perception. The method comprises the following steps: firstly, respectively obtaining a track semantic vector and a variation hidden variable through a circular track encoder and a variation track encoder; and obtaining an attention vector of the track basedon a variational attention mechanism, cascading the attention vector with the variational hidden variable so as to reconstruct input data of a decoder, and finally restoring a previous track and generating a prediction track according to an output semantic vector of the decoder. According to the invention, the frame of the encoder-decoder solves the problem of track context learning. Two sub-tasks, namely track recovery and track prediction, speculated by a human movement mode are completed; not only can the probability density be estimated and the lower limit of the data possibility be optimized, but also the sequence and time characteristics of human mobility can be captured, the problem of track speculation according to track context perception is effectively solved, and the effect is improved for speculation of a human movement mode.

The invention discloses a thermal power pollution factor control method of air fine particles. The method includes the steps that data related to production of thermal power plants are obtained from the power production department, meteorological and PM2.5 monitoring data are obtained from the environment protection bureau, influences of pollutants emitted by one single thermal power plant on PM2.5 are calculated quantitatively, and an analysis model of the influences of one single thermal power plant on surrounding environment pollution is built; model superposition is carried out on the single thermal power plants to form thermal power generation emission PM2.5 quantitative analysis model around a city, and thermal power PM2.5 pollution around important cities even larger geographic areas is analyzed quantitatively; prediction and quantitative calculation are carried out on pollution emission by means of power generation amount of the thermal power plants, the production process and environment parameters to obtain quantitative priority of thermal power plant government. By means of the method, the influences of the pollutants emitted by thermal power enterprises around the city on PM2.5 are quantitatively analyzed, prediction of a week level and a month level of the PM2.5 value in the important city area is carried out, bases are provided for the government to make policies, and environment protection is achieved while production is guaranteed.

The invention discloses a reservoir physical property parameter prediction method combined with deep learning, and the method comprises the steps: introducing the nonlinear correlation between an MICquantitative measurement physical property parameter and a logging curve, and selecting the logging curve which is obvious in response to the physical property parameter; introducing CEEMDAN to decompose the physical property parameter data sequence to obtain an IMF component and a residual RES component of an intrinsic mode function, and subjecting the physical property parameter data sequence tostationary processing; introducing SE to evaluate the complexity of each IMF component and RES margin, and recombining component sequences with similar entropy values to obtain a new intrinsic mode component; carrying out normalization processing on the new intrinsic mode component data and then dividing the new intrinsic mode component data into a training set and a test set; introducing an LSTMrecurrent neural network to establish a prediction model for the reconstructed new component, and obtaining a prediction value of each new intrinsic mode component; and carrying out inverse normalization on the prediction value of each new intrinsic mode component, and carrying out superposition reconstruction to obtain a physical parameter prediction result. According to the method, the modelingnumber of redundant information and prediction components is reduced, and the prediction precision and the prediction speed are improved.

The invention discloses an analogue simulation method for discharged smoke dust pollution of a thermal power plant. The analogue simulation method comprises the following steps: firstly, obtaining relative operation data of power generation and electric dust removal of a power station, and monitoring data of smoke dust of a chimney air outlet; combining weather and environment parameters in the peripheral region of a concerned single thermal power plant to establish a smoke dust diffusion simulation mathematic model in the certain region of the periphery of the single thermal power plant; secondly, considering a polymerization effect of PM2.5 according to electric dust removal operation parameters and environment and weather conditions of the thermal power plant and establishing a diffusion and polymerization simulation mathematic model of influences caused by the environmental pollution of the thermal power plant in the peripheral analyzing region; thirdly, carrying out simulation predication and quantitative calculation on PM2.5 and PM10 pollution emissions respectively and establishing a simulation procedure of comprehensive effects of the two models; and fourthly, considering a plurality of power station comprehensive effects to form a quantitative and simulation analyzing model for the smoke dust emission of the thermal power plant in a researching range to obtain predicated quantitative data of the pollution of the thermal power plant.

The invention discloses a method and system for label-free data classification and predication. History data is utilized and rigid requirements on history data labels are removed. Also, experience information database accumulated in a service process is combined and an aim of automatically improving predication precision can be achieved through subsequence continuous optimization of real data. Themethod includes inputting service flow data and acquiring service scene data in a service flow; inputting service content data and grouping the service content data according to the service scene data; constructing characteristic indexes of service content data according to the service scene data and a corresponding service content database; cleaning the characteristic indexes of the service content data; clustering the service content data subjected to characteristic index cleaning and determining the class centers of different classes; calculating the sampling weights of the different classes; sampling the service content data according to the sampling weights and marking predication labels on the sampling results.

The invention discloses a method and system for predicting a wind machine state based on a Bayesian reasoning mode. The method comprises the steps that operation data of a wind machine in a certain time are obtained and are classified; the wind machine operation state type of each time period is subjected to probability statistics processing, probability values are generated, and overall operation state types in the time periods are determined; the overall operation state types are subjected to Bayesian network modeling, and the priori condition probability of the overall operation state types is generated; and the priori condition probabilities of the time periods and the probability values of the overall operation state types are subjected to probability processing in a Bayesian reasoning mode, a joint probability value is generated, the state distribution curve of the wind machine is established, and according to the state distribution curve, the wind machine is subjected to fault state prediction. The problem that the fault operation state of a wind power generator cannot be predicted easily is solved.

The invention discloses a tube explosion predicting method based on a grey neural network. The tube explosion predicting method based on the grey neural network comprises the steps that firstly predicting is conducted on tube explosion rate sequences through static grey modeling according to given tube explosion factors and tube explosion rate data sequences, and predicting results and the original tube explosion rate sequences are compared to obtain residual errors; then neural network approximate models are established among the residual errors and the tube explosion factors by using the neural network, and the neural network being repeatedly trained is a mapping relation among the residual errors and grey model data; in the final predicting process, predicted values of the grey models are compensated by compensation values of the neural network. According to the tube explosion predicting method based on the grey neural network, the grey neural network models are established by combining a grey modeling method and a neural network model, the defect that a traditional tube explosion model needs a large amount of data is overcome, problems of predicting small samples can be excellently solved, and predicting accuracy is improved.

The invention relates to a PM2.5 hour concentration prediction method and system fusing SSAE deep feature learning and an LSTM network. The method comprises steps that an SSAE-LSTM (Spatial Spatial Absorption Emphasis-Long Short Term Memory) model is constructed; an air pollutant time sequence with a certain time step length is input into the model; an SSAE network is adopted to extract abstract features of input data through an unsupervised method, the extracted features serve as input features of an LSTM network, feature distribution of air pollutant information within a certain time step length is obtained, and finally the PM2.5 hour concentration is predicted in combination with a full-connection network. According to the invention, the prediction accuracy can be effectively improved.

The present invention discloses a solar energy collection power prediction method based on a grey neural network. The method comprises: selecting the original solar energy collection power at the sametime each day for a plurality of days and the factor data sequence affecting the solar energy collection, and using the grey predicting method to predict the solar energy collection power sequence toobtain preliminary prediction results; normalizing the grey prediction results and the original factor data sequence affecting the solar energy as the input of the neural network, taking the originalsolar energy data sequence as the output of the neural network, establishing a neural network model, and training the neural network until the neural network converges; and finally calling the trained neural network to perform the final solar energy collection power prediction. According to the method disclosed by the present invention, the grey modeling method and the neural network method are combined to establish the grey neural network model; and compared with the common neural network model, the grey prediction model is introduced to reduce the amount of calculation in the prediction andachieve relatively high accuracy in the case of a few samples, and the prediction accuracy is higher.

The invention discloses a ship collision accident level prediction method and system based on a Bayesian network. According to the method, by constructing a ship collision accident prediction model, grade prediction of ship collision accidents is achieved, from the perspective of data, the law between accident factors is found by deeply mining the relation between the accident factors, an expert knowledge method with too high subjectivity is replaced, and the Bayesian network of the ship collision accidents is established. From the perspective of the method, a ship collision accident prediction model is constructed based on the Bayesian network, the ship collision accident level prediction problem is solved, and the development of the maritime risk theory is promoted. From the perspectiveof application, research results can provide scientific basis for ship navigation command and emergency decision making.

The invention discloses a repairable spare part demand prediction method for a phased-mission system, which is used for solving the technical problem of poor spare part demand prediction effects of an existing method. According to the technical scheme, the method comprises the following steps: firstly, analyzing spare parts required by missions of different phases, and determining part and system states, comprising a running state and a failure state; then, performing failure mode modeling on each part (group), and calculating the availability of each part (group) in the whole phase of mission, wherein different parts and corresponding spare parts thereof in the system are considered as a whole part group when the spare parts exist and are repairable; next, generating a BDD model of the phased-mission system on the premise of a certain sequencing rule; finally calculating the reliability of the phased-mission system under the corresponding spare part number on the basis of the established BDD model and a Markov chain of the parts (groups), and comparing the reliability with required mission system reliability, thereby predicting a repairable spare part demand of the system. The repairable spare part demand can be accurately predicted.

The invention discloses a product performance degradation interval prediction method based on support vector machine (SVM) and fuzzy information granulation, comprising the following steps of: step 1, collecting product multi-parameter performance degradation data; step 2, analyzing principal components of the multi-parameter degradation data; step 3, executing fuzzy information granulation on the obtained principal component data; step 4, executing SVM modeling for the granulated data; step 5, executing interval prediction on product performance degradation trends. In the method, a fuzzy information granulation method and a SVM method are combined, the interval prediction method for the product performance degradation trends is presented for the first time, and the problem of prediction on the degradation trends of performance state and the changing space in the product running process is solved. The method solves the problem of evaluation and prediction under a condition that a plurality of output performance characteristic parameters of some products with complex structures concurrently degrade, based on the principal component analysis method.

The present invention discloses a multi-state equipment system multi-stage spare part demand prediction method facing a repairable spare part, used for solving the technical problem that the conventional spare part demand prediction method is poor in practicality. The technical scheme is that firstly the task demands of the stages and the states of the parts are determined respectively according to the equipment system part composition; secondly, the different parts and the corresponding spare parts in an equipment system are used as the repairable part sets respectively to consider overall, and an MMDD model of a task system is established; then a Markov chain is utilized to establish a failure model of each part set of a multi-state task system carrying the repairable spare parts in the multi-stage tasks, the paths from a root node to a final node 1 in the MMDD model are enumerated to form an non-intersect path set; finally, based on the established MMDD model and the Markov chain of the part sets, the reliability of the task system at a stage under a corresponding spare part number is calculated and is compared with the required reliability of the task system, and the repairable spare part demand amount of the multi-state equipment system is predicted.

The invention provides an air conditioner control method and system based on target room load prediction. The air conditioner control method comprises the steps of 1, collecting the indoor temperatureof a target room and the outdoor temperature at a plurality of different positions of the target room before and after an air conditioner works, the starting time of the air conditioner, the workingtime of the air conditioner and the working state of the air conditioner during the corresponding working time, and establishing a target room load prediction model according to the collected data; 2,according to the indoor temperature and the outdoor temperature of the target room before the air conditioner works at the current moment, predicting through the target room load prediction model toobtain first instantaneous refrigerating / heating capacity of the air conditioner to the target room in the predicted working time in order to enable the indoor temperature of the target room to reachthe target temperature; and 3, according to the first instantaneous refrigerating / heating capacity to the target room, changing the total refrigerating / heating capacity provided by the air conditioner, and / or changing the refrigerating / heating capacity provided to the target room.

The invention discloses an agricultural drought diagnosis and evaluation method coupling a crop model and a machine language. The method comprises the steps of S1, obtaining related data in a researcharea; S2, localizing a crop model in the research area by utilizing the data obtained in the step S1; S3, setting different drought and rainfall scene data according to different growth stages of crops, and inputting the data into the localized crop model to obtain simulation data of each grid point in the research area under each drought scene; S4, calculating a characteristic variable based onthe drought situation data and the corresponding crop model simulation data in S3; S5, taking the yield loss rate data simulated by the corresponding crop model as a prediction variable based on the characteristic variable in the step S4, constructing a sample data set, and inputting the sample data set into a random forest model to construct a crop drought vulnerability model; S6, using the cropdrought vulnerability model for diagnosing and evaluating the crop drought condition in the research time period.

The invention discloses an expressway toll station traffic flow big data prediction method based on multi-target regression. The method comprises the following steps: 101, carrying out preprocessing operation on data; 102, marking the data; 103, performing feature engineering construction operation on the data; 104, constructing a multi-target regression model combining the specific characteristics of the target and the target correlation; and 105, predicting the traffic flow of the toll station every 20 minutes from 8 o'clock to 10 o'clock according to the historical traffic flow data of thetoll station, the weather data and other information through the established model. The method mainly preprocesses and analyzes historical traffic flow data, weather data and other information of a toll station to extract characteristics; a multi-target regression model combining target specific characteristics and target correlation is established, and the traffic flow from 8 o'clock to 10 o'clock per 20 minutes is predicted, so that a traffic management department can take measures in time by using big data to reduce the congestion of the toll station.

The invention discloses a construction method for a communication traffic threshold model of an intelligent substation based on FARIMA. The method comprises the following steps: analyzing the stability and autocorrelation of a traffic sequence; designing an FIR filter according to Hurst parameters of the sequence, and filtering data; performing order determination for an ARMA (p, q) model by usingAIC information criteria, and then performing residual test for a predicted value of an FARIMA (p, d, q) model; and simulating the FARIMA model for multiple times, performing short-term prediction, and designing an algorithm to generate traffic threshold models with different confidence levels. According to the construction method disclosed by the invention, a detailed algorithm description is provided by taking the traffic actually acquired by a station control layer of a certain substation in Tianjin as test data, and the communication traffic threshold model of the station control layer ofthe intelligent substation is also obtained through a series of experiments.

The invention belongs to the technical field of mobile flow prediction, and discloses a mobile flow prediction method based on a space-time attention convolutional network, and the method comprises the steps: enabling the space-time attention convolutional network to carry out the modeling of a mobile flow network in an hour period, a day period and a week period through three time parts, and obtaining three pieces of corresponding mobile flow prediction information; and fusing the three pieces of mobile traffic prediction information with external interference information to obtain a final mobile traffic prediction result. According to the invention, the prediction problem of the mobile flow is effectively solved.

The invention provides a multi-point photovoltaic output probability prediction method based on a Pair-copula theory. The method comprises the steps of: performing cluster analysis on a target photovoltaic power station group to determine the optimal number for classification; performing statistics of the photovoltaic output values of various power stations and a photovoltaic power station total output value to obtain a probability density function and a cumulative distribution function respectively; performing parameter identification of a two-dimensional Copula function for the output of various power stations and the total output to obtain a series of partial correlation coefficients; solving a related coefficient of a Pair-copula function according to a formula; and employing the determined Pair-copula model and the prediction value of the photovoltaic output of each power station to obtain multi-point distributed photovoltaic joint output condition probability distribution. The clustering analysis theory is introduced to classify the photovoltaic power stations. Based on the Pair-copula theory, a conditional probability prediction model of the joint output is established to solve the problem of multi-point distributed photovoltaic combined output prediction.

The invention relates to the technical field of website data statistic analysis and particularly discloses a website page view prediction method based on historical tendency weights. The method comprises the steps of data preprocessing and prediction result processing, wherein in the data preprocessing step, the logarithm of page views is taken, the variance of all time points in the historical tendency is calculated, the regression coefficient of the ith historical tendency to the current tendency is calculated, the variance of correlation coefficients of a current value estimated according to the ith historical tendency is calculated, deviation is estimated through an optimization minimization method to solve the weights, and the value appearing after the current tendency is predicted according to the weights; in the prediction result processing step, a prediction result is subjected to exponential transformation. According to the website page view prediction method, the known current tendency is compared with the historical tendencies for calculation of the correlation coefficients, the deviation of the current tendency is estimated according to all the historical tendencies, the weight of each historical tendency is selected through the optimization method, estimations at all dates are superposed according to the weights, and the current tendency and a follow-up tendency can be predicted according to a superposition result, so that the deviation of the estimations of the current tendency by synthesizing the historical tendencies is minimum, and reliable prediction is achieved.

The invention discloses a power transmission line icing thickness prediction method based on CEEMDAN-QFOA-LSTM, and relates to the field of combination of power transmission line state evaluation anddeep learning. The method comprises the following steps o: (1) carrying out data acquisition and preprocessing; (2) carrying out CEEMDAN decomposition on an icing thickness historical data sequence (12); (3) optimizing hyper-parameters of the LSTM by a quantum drosophila melanogaster algorithm; (4) carrying out LSTM model training (14); and (5) predicting the icing thickness of a power transmission line and analyzing a result (15). According to the method, the CEEMDAN decomposition algorithm is used, a sequence which is difficult to directly predict is converted into a plurality of predictablecomponent sequences; a neural network can more accurately grasp the law of the sequence according to multi-dimensional feature information obtained through decomposition; a QFOA optimization algorithm is used for obtaining the hyper-parameters, a complex manual parameter adjustment process is avoided, and a network model is trained more effectively; the used LSTM neural network does not have theproblem of gradient disappearance of a general network, so that optimal convergence of the model is ensured, and the problem of short-term and long-term time sequence prediction is effectively solved.

The invention relates to an extreme value gradient lifting logical regression classification prediction method, belonging to the field of big data analysis and intelligent classification prediction. After the extreme value gradient lifting model is used to learn the samples, each sample falls on each classification and the leaf node position of the regression tree for unique thermal coding to combine into a new feature, and then combines with the previous features to form a combination feature, so that the features of the samples are increased and new samples are formed. Logistic regression isused to classify and forecast the new samples. Fusion of Extreme Value Gradient Lifting and Logistic Regression; Using extreme value gradient lifting to select features, choosing cart tree as the base classifier, utilizing kini impurity to form a series of uncorrelated features, enlarging the dimension of features, and training new features into logistic regression model will have a better prediction effect. The invention has the advantages that the feature selection and the feature expansion functions of the extreme value gradient lifting are respectively utilized, and the problem of low prediction accuracy of a single model logical regression model is solved.

The invention relates to a gene network analysis-based disease subtype prediction system. The system is divided into the following three modules: a data acquisition and preprocessing module; a patientclassification and disease subtype prediction module; and a result display module; the patient classification and disease subtype prediction module includes a gene network gene node vectorization sub-module, a disease patient vectorization sub-module, and a disease subtype prediction sub-module.

The invention discloses a quantitative evaluation method for beer packaging production line operation efficiency based on information entropy, and belongs to an intelligent information processing technology assisting enterprise production management decision technology. The method mainly comprises five steps: step 1, determining a reference variable of efficiency evaluation; step 2, establishing amatter element evaluation model of a beer packaging production line; step 3, using an AHP and an entropy evaluation method to calculate composite weight of the evaluation model; step 4, quantitatively calculating health degree of the beer packaging production line; step 5, establishing a health degree prediction model of the beer packaging production line.

The invention provides a group target advancing trend prediction method based on an LSTM neural network, and the method comprises the steps: carrying out the modeling of interactive group targets in an actual scene through the relative spatial relation between group targets, and establishing the LSTM long-short-term memory neural network for the prediction and calculation of a track position. Dependence prediction is carried out according to contact communication between single targets in a group, a cooperative relationship, single target model characteristics and group target activity properties, unreasonable factors are eliminated, a minimum prediction space of a group target advancing track trend is obtained, and a final prediction result is visually expressed. Compared with the traditional single target trajectory prediction, according to the method, the related problem of predicting the advancing trend of the complex group target with certain correlation and interactivity is solved; information such as a target historical track, a target association relationship, a target model category, a target mass characteristic and a target activity property can be fully utilized to comprehensively calculate and predict, and a prediction result of a group target advancing trend is accurately provided.

The invention discloses a hydrocarbon source rock total organic carbon content prediction method considering a density factor. The method comprises the following steps: drawing the organic carbon content of a rock core of each well and a logging curve of the corresponding well on the same graph through software; segmenting the hydrocarbon source rock according to the maturity or age stratigraphictable of the hydrocarbon source rock; manually picking up the baseline value of the RD and the baseline value of the DT of each section; and according to the baseline value of the RD of each section,the baseline value of the DT of each section, the RD logging curve value corresponding to each depth and the DT logging curve value corresponding to each depth, solving the amplitude difference deltalogR of reverse superposition of the DT curve corresponding to the measurement point of the rock core of the multiple wells and the RD curve, and then predicting the total organic carbon content of the hydrocarbon source rock. According to the method, the tedious process that a traditional delta logR method needs to correspond to a maturity parameter chart is avoided, the influence of the compaction effect on the hydrocarbon source rock is considered, the application range of the traditional method is expanded, and the method has a good effect on the continental facies deep hydrocarbon sourcerock in China.

The invention discloses a multi-label data stream classification method based on incremental learning, and the method comprises the steps: step 1, an initial training stage: carrying out the modelingof a multi-label data stream into data blocks with a fixed instance number, carrying out the naive Bayes model training of each data block according to the initial data block, and obtaining a clustercenter set through employing a KMeans algorithm, wherein the trained naive Bayes classification model and the cluster center set jointly serve as a base classifier; step 2, a concept drift detection stage: when the number of the base classifiers in the naive Bayes integration model reaches a certain number in the initial learning stage, carrying out concept drift detection from the data level andthe model level respectively; step 3, an increment updating stage: when a latest data block Dt comes, updating the base classifier by using information carried by each sample in the Dt for each base classifier in the integration model, and performing instance information updating. The concept drift in the data flow can be detected in time, the situation that the algorithm performance encounters large downslide when the concept drift occurs is avoided, the latest data can be subjected to incremental learning, and the performance of the model is guaranteed.

The invention discloses a ship and boat shafting service life predicating method on the basis of vibration fatigue coupling analysis. The method comprises the steps of selecting a measuring time point ti, measuring the inherent frequency omega i (for n times of measurements, i=1, 2, 3, ..., and n and i are positive integers) of a ship and boat shafting located in the measuring time point ti, then, calculating the expected fatigue service life Ni when the ship and boat shafting is located in the measuring time ti, and finally calculating the remaining service life Mi when the ship and boat shafting is located in the measuring time ti. The method overcomes the defects of an existing method, starts from a ship and boat shafting model, calculates the service life of the ship and boat shafting on the basis of vibration and fatigue coupling environments, solves the problem that calculation and predication of the ship and boat shafting service life are not involved in the prior art, is easy to understand and easy to realize, and is more suitable for calculation and prediction of the remaining service life of the ship and boat shafting.