228 results about "Concentration prediction" patented technology

The invention relates to a CNN and LSTM fused neural network-based air PM2.5 concentration prediction method. The method comprises the steps of S1: building a city PM2.5 concentration prediction modelbased on a deep learning principle, a CNN and LSTM; S2: for the built model, selecting training data and test data from environment monitoring data, and finishing initialization of the prediction model; S3: training the model by utilizing the training data; S4: by utilizing the trained model, obtaining a test prediction result according to test data; S5: judging the accuracy of the test prediction result, and if the accuracy exceeds a threshold, executing the step S6, or otherwise, returning to the step S2; and S6: performing prediction by utilizing the trained model. Compared with the priorart, the prediction accuracy of the prediction method is higher than that of a conventional prediction method; and under the same working duration and working conditions, a better result can be generated.

The invention discloses a control method for a selective catalytic reduction flue gas denitrification system. The system includes: an entrance NOx concentration predictor, an ammonia injection flow controller, an entrance NOx concentration measurer and an NH3 concentration measurement device. The method utilizes a neural network unit to carry out weighted calculation on an entrance NOx concentration prediction value obtained by the entrance NOx concentration predictor and an NOx concentration measurement value measured by the entrance NOx concentration measurer to obtain a corrected NOx concentration, and according to the corrected NOx concentration and a set ammonia nitrogen mole ratio, the ammonia injection quantity demand of the ammonia injection flow controller can be positioned, thus realizing automatic control of the denitrification and ammonia injection quantity. The method provided by the invention matches the denitrification and ammonia injection quantity with the actual NOx concentration in flue gas, effectively reduces the error caused by the time delay problem of NOx concentration monitoring by instrument, and improves the reliability and economical efficiency of the denitrification system.

The invention relates to an RBF-neural-network-based atmospheric pollutant concentration prediction method. The RBF-neural-network-based atmospheric pollutant concentration prediction method includesthe steps: dividing experimental data according to the actual situation of the predicted area, and pre-processing the atmospheric pollutant concentration data; using the MMOD improved K-means++ algorithm to solve the center of clustering, and calculating each kernel function width based on the variance; sampling the experimental data, wherein data subsets taking part in creation of RBF neural networks are IOB, and the remaining data that are not drawn are OOB data; evaluating learners to screen out the RBF neural network with the smallest generalization error, training an integrated RBFNN model; and by means of the weighted integrated RBFNN algorithm, based on weighted Euclidean distance, training single parameter through the center of clustering, the width and the weight to optimize RBFNN, and applying the single parameter to the integrated RBFNN to predict data. The RBF-neural-network-based atmospheric pollutant concentration prediction method is applied to atmospheric pollutant concentration prediction, and can greatly improve accuracy of atmospheric pollutant concentration prediction.

The invention discloses a PM2.5 concentration prediction method and device and a medium, and relates to the technical field of pollutant prediction, and the method comprises the steps: building a PM2.5 prediction model based on a CNN and a bidirectional GRU neural network and based on a one-dimensional convolutional neural network CNN and a bidirectional GRU neural network; the meteorological training data tensor is sent to a PM2.5 prediction model for training; the one-dimensional convolutional neural network CNN respectively performs local feature learning and dimension reduction on each input variable time sequence, and forms a low-dimensional feature sequence through convolution and pooling operation in sequence; inputting the feature sequence into a bidirectional GRU neural network, and learning the feature sequence from the time positive sequence and the time negative sequence by the bidirectional GRU neural network; the meteorological test data tensor is sent to a trained PM2.5prediction model for prediction, and a PM2.5 prediction concentration value is obtained. According to the model, the speed and lightweight characteristics of the convolutional neural network and the sequential sensitivity of the RNN are effectively utilized, more data volume is allowed to be checked during training, and the prediction accuracy is improved.

The invention provides a metal-oxide gas sensor array concentration detecting method based on drift compensation. According to the method, the independent component analysis is utilized, the influence of abnormal values caused by the environment temperature, the environment humidity and environment factors to the drifting regulation and the drifting quantity estimation is shielded, the law that a independent concentration component changes along with the time drifting under the condition of base line responses is found out, subsequently, and when the concentration detection treatment on a gas sample is subjected to the concentration detection treatment by a metal-oxide gas sensor array, the drifting quantity of the concentration independent component is estimated by using the law that the independent concentration component changes along with the time drifting under the condition of the base line responses, so as to carry out drifting compensation on the independent concentration component in concentration detection response data; and the prediction calculation is carried out through using a concentration prediction function by virtue of the independent concentration component subjected to the drifting compensation in the concentration detection response data, so as to obtain a sensitive gas concentration detection result. The precision in estimating the drifting law and the drifting amount can be effectively improved, and the accuracy of concentration detection of the metal-oxide gas sensor array is ensured.

The invention discloses a gas concentration prediction method and device based on an extreme learning machine. The method comprises the steps that over time, historical gas concentration time sequences arranged sequentially are formed for each kind of gas; a sample set with the h-1 historical gas concentration time sequence as input and with the h historical gas concentration time sequence as output is built for each kind of gas; each sample set is divided into a training sample set and a test sample set; by means of the optimal input layer weight and hidden layer offset, training and learning are carried out on each training sample set through the extreme learning machine, and concentration prediction evolution extreme learning models corresponding to different kinds of gas respectively are obtained; the last historical gas concentration time sequence of each kind of gas is adopted as input, a future gas concentration time sequence of each kind of gas is output according to the corresponding concentration prediction evolution extreme learning model. According to the gas concentration prediction method and device based on the extreme learning machine, the implementation process is simple and efficient, and prediction accuracy is improved.

The invention discloses an air pollutant concentration prediction method based on a graph attention mechanism. The method comprises steps of constructing a spatial pollutant concentration prediction model based on a graph attention mechanism; and taking the meteorological data, the air monitoring data and the environmental factor data as model input data, constructing a graph adjacency matrix through a graph attention mechanism, extracting graph information characteristics by utilizing a graph convolutional neural network layer and a multi-layer perceptron network layer, and outputting a predicted air pollutant concentration value. According to the method, the air pollutant concentration prediction is more accurate, and the process is more efficient.

The invention relates to an atmospheric pollution factor concentration space-time distribution prediction method and system, and the method comprises the following steps: constructing a sparse feature vector based on the historical monitoring data of all stations in a monitoring region, and predicting the historical data of atmospheric pollution factor concentration through a factor decomposition machine; combining the historical data of the atmospheric pollution factor concentration with meteorological parameters to train a long-short-term memory neural network, and predicting future data of the atmospheric pollution factor concentration through the trained long-short-term memory neural network; and training the radial basis neural network by combining the future data of the atmospheric pollution factor concentration with the meteorological parameters and the geographic latitude and longitude of the station, and predicting the future data of the atmospheric pollution factor concentration of the target position point in the monitoring area through the trained radial basis neural network. The method can solve the problems that in the current atmospheric pollution factor monitoring technology, a large number of monitoring devices are needed, atmospheric pollution factor concentration prediction is inaccurate, and the future distribution trend cannot be dynamically analyzed.

The invention discloses an EMD and LSTM fused urban PM2.5 concentration prediction method, and relates to the field of air quality concentration prediction. The method comprises the following steps offirstly, acquiring the time sequence data per hour, and performing data cleaning on the acquired data; then, using the EMD (empirical mode decomposition) for carrying out stationary processing on thePM2.5 concentration data to obtain a plurality of components; then, determining a sliding time window T, carrying out data sequence segment segmentation processing on each component, and normalizinga unified dimension to obtain a plurality of data sets; dividing the data set into a training set and a test set, respectively constructing an LSTM network model for training, finally predicting eachcomponent by using the trained model, and carrying out reverse normalization processing on each component to obtain a final urban PM2.5 concentration prediction result; on the basis, constructing a long-term and short-term memory neural network LSTM model and training; finally, using the trained model for prediction, carrying out reverse normalization processing on the model, and obtaining the final urban PM2.5 concentration prediction result.

The invention discloses a PM2.5 concentration prediction method which combines stack-type self-coding and support vector regression. Steps are as follows: selecting influential factors with strong correlation as independent variables of the PM2.5 concentration prediction model, continuously collecting the selected independent variables and corresponding PM2.5 concentration data for N hours, and constructing a training set; a deep learning network model of stack self-encoder with K hidden layers being constructed; the training set being input into the model for training and the weights of eachvariable being obtained; the SVR model being trained by using the eigenvalues of the learning network model as input vectors, and the output of the SVR model being obtained; the test set being obtained, the test sample being predicted by the predictive model, and the precision of the predictive result being calculated. The invention has the advantages of SAE extracting data set features and SVR excellent prediction ability, and the prediction performance and precision are better than widely used ANN algorithm and SVR algorithm.

The invention relates to a prediction method for urban PM10 concentration based on a characteristic expansion fusion neural network, which comprises the following steps of: step S1: constructing a model of the concentration prediction of the urban PM10 based on the feature-expanded stack self-encoder and LSTM network; step S2: selecting training data and test data from monitoring data of pollutants and meteorology; step S3: carrying out the training on the feature-expanded stack self-encoder by using the training data; step S4: processing the feature vectors of the output of the stack self-encoder based on the Gaussian function, calculating the corresponding influence weights for the feature vectors of different cities, and weighting and summing to obtain a new feature vector; step S5: inputting the new feature vector into the LSTM to carry out integral training of the model; step S6: inputting test data into a trained model, and measuring the error of the prediction result generated by the test data; step S7: adopting the training and fine-tuning models for air pollutant concentration prediction. Compared with the prior art, the invention has the advantages of accurate predictionand the like.

The invention discloses an aneuploidy biological information analysis method and system, wherein the method comprises: 1) constructing a reference database; 2) calculating UR ratio; 3) constructing statistical parameters of the reference database; 4) calculating Z value; 5) decreasing fetal chromosomal aneuploidy false positives due to microdeletion or micro-duplication present in maternal chromosomes according to the above process of decreasing fetal chromosomal aneuploidy false positives due to microdeletion or micro-duplication present in maternal chromosomes; 6) constructing a fetal DNA concentration prediction model to predict fetal DNA concentration according to the above process of constructing a fetal DNA concentration prediction model; 7) calculating percentage of DNA of each fetal chromosome to total DNA; 8) judging autosomal aneuploidy; 9) judging allosomal anomalies. By using the method and system according to the technical scheme, analytical accuracy is improved greatly.

The invention relates to a method for controlling flue gas desulfurization based on neural network prediction. The method is characterized by comprising the following steps: S1, collecting sample datarelated to time variation in a wet flue gas desulfurization system, and determining neurons of an input layer and an output layer of a dynamic neural network according to the collected sample data; S2, modeling the wet flue gas desulfurization system by using the dynamic neural network in the S1 to establish a prediction model of the wet flue gas desulfurization system; and S3, calculating a sulfur dioxide concentration prediction value at a flue gas outlet of the wet flue gas desulfurization system by using the prediction model of the wet flue gas desulfurization system established in the S2, and controlling a slurry spray amount of the wet flue gas desulfurization system by using the sulfur dioxide concentration prediction value.

The invention provides a thickener underflow concentration prediction method based on integrated learning, and belongs to the technical field of mining. The method comprises the following steps: obtaining actual production historical record data, storing the actual production historical record data in an enterprise database, then preprocessing the obtained data set, and constructing a training setand a test set by using the preprocessed data; and an integrated learning method is adopted, the constructed training set and test set are utilized to establish a model, accurate prediction of the underflow concentration of the deep cone thickener is realized, and finally, a prediction result is displayed through a visual platform. According to the method, most factors influencing the underflow concentration can be comprehensively considered, so that the bottleneck problem of insufficient one-sided consideration when an existing underflow concentration prediction model considers the influencefactors is solved. And an integrated learning model is used, so that the problems that a single machine learning model is limited in learning capability and large-scale data cannot be processed are solved, and more effective and accurate reference is provided for control of the thickener.

The invention discloses a mobile pollution source emission concentration prediction method based on a fuzzy weighting ELM. According to the method, on the basis of spatial and temporal distribution features of pollutants from a mobile pollution source, a self-adaptive fuzzy weighting extreme learning machine model is provided; according to the features of a fuzzy system, partial input data of a mobile pollution source emission concentration data set is utilized to complete initialization of a self-adaptive fuzzy weighting extreme learning machine, namely fuzzification processing. Due to the self-learning and self-adjustment features of the fuzzy system, input weight values and implicit strata offset values of the extreme learning machine can be optimized, and then regularization processingis carried out by the weighting extreme learning machine to obtain the mobile pollution source emission concentration prediction method.

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 invention belongs to the application field of mutual cross combination of multiple subjects and technologies, and specifically relates to a spatial-temporal data model based city-level PM2.5 concentration prediction method. The method comprises the steps of firstly acquiring city-level PM2.5 concentration data by using a developed online data acquisition system; determining an optimal model prediction factor by comprehensively considering the model interpretability, the data accessibility and the model complexity, and thus determining the model type and the model structure; then determining an interpolation algorithm for the missing PM2.5 concentration data according to missing conditions of the actually acquired PM2.5 concentration data and the determined model structure; and finally determining an algorithm capable of performing real-time online prediction according to the model prediction precision. According to the invention, the model structure can be determined by making full use of readily available data, and thus an effective city-level PM2.5 concentration prediction model is built.

[Problem] To provide a data concentration prediction device accurately predicting data concentration by analytically processing additional learning data extracted from within a necessary-sufficient range; a method thereof; and a program thereof. [Solution] A data concentration prediction means (31) analyzing, using a data storage means (21), a data structure of time-series data received by a data input means (11) to predict subsequent data concentration includes a learning data extraction processing unit (41) continuously extract-processes, as additional learning data necessary for predicting the subsequent data concentration, the time-series data deviating from a fluctuation permission range preset on the basis of time-series data within a past fixed period based on a time point immediately preceding an input time point of each time-series data. A prediction processing unit (71) calculates a prediction value concerning future data concentration on the basis of processed information resulting from subjecting the additional learning data to various calculation processes.

The invention provides a transformer fault prediction method and device, a terminal and a readable storage medium. The transformer fault prediction method comprises the steps of building a concentration prediction model of a characteristic gas according to collected historical concentration and historical electrical parameters of the characteristic gas dissolved in the transformer oil; processingthe collected current concentration and current electrical parameters of the characteristic gas by using the concentration prediction model to obtain the concentration of the characteristic gas at thenext moment; and performing fault prediction according to the concentration of the characteristic gas at the next moment to obtain the predicted fault type. According to the fault prediction method,the association relations between oil-soluble gases and between an oil-soluble gas and other electrical parameters are firstly analyzed, then the concentration prediction model of each oil-soluble gasbased on the other gases and the electrical parameters is built, the oil-soluble gas concentration of a transformer at any moment in the future is predicted according to the concentration predictionmodel, fault prediction is performed according to the oil-soluble gas concentration, and the accuracy of transformer fault prediction is improved.

The invention relates to an underground water pollution stratification evaluation method based on a specific polluted site. The method comprises steps of dividing evaluation into three levels through a pollution concept model and according to a relative position of a pollution source and an environment sensitive point, and starting with an area where a pollution source is furthest distant from the environment sensitive point, determining a prediction model of each level, collecting model parameters and determining a model parameter acquiring method, starting with the first level evaluation as a program evaluation point to evaluate, comparing target pollutant prediction concentration of each level evaluation terminal point and an evaluation standard, entering a higher level evaluation if the prediction concentration exceeds the evaluation standard, and otherwise ceasing the evaluation. Evaluation is divided into three levels; an evaluation terminal, a target pollutant concentration prediction model and evaluation standard of each level are clarified, so evaluation work for corresponding programs can be well guided and developed; the evaluation method is not only suitable for underground water pollution evaluation, but also suitable for evaluation of pollution to underground water within a site by unsaturation soil pollution.

The invention discloses a soft measuring method for a denitration control system. The method is characterized in that an auxiliary variable relevant to the outlet NOx concentrate value is analyzed; a field instrument records data; received data is uploaded into a DCS; then, the DCS inputs the data into a computer, and receives an outlet NOx concentration prediction value returned by the computer to form a closed loop; a support vector machine model analysis model is arranged in the computer; the modeling is performed through a support vector machine; the outlet NOx concentration value combining with a support vector machine regression algorithm is estimated to obtain the current moment prediction value; the current moment prediction value is returned into the DCS. The method provided by the invention has the advantages that the advanced support vector machine algorithm is used for regressing the SCR outlet NOx concentration value; the current value can be fast and effectively measured; in addition, the NOx change in a period of time in future can be obtained. Sample data used by the support vector machine are few; the calculation speed is high; the generalization capability is high; the SCR outlet NOx concentration value can be fast and effectively predicted.

The invention provides a method for predicting human health risk caused by groundwater pollution, relates to a groundwater pollution concentration prediction stage and a human health risk evaluation stage, and aims at solving the problem that the human cost investment is large, the sampling evaluation time is long and the evaluation result is incomprehensive. According to the method, simulation software is adopted to solve the migration and transformation process of groundwater in saturated zones so as to simulate and predict the concentration distribution conditions of pollutants in groundwater; and the human health risk values of different point positions through the software, so that support is provided for the field management and risk pre-warning.