56 results about "Time series dataset" patented technology

This invention discloses a wastewater treatment system based on artificial intelligence to predict water quality changes, belonging to the field of intelligent prediction technology for wastewater treatment. It includes modules for pretreatment, model training, trend prediction, strategy simulation, and decision delivery. The pretreatment module collects real-time water quality data from key nodes of wastewater treatment through a multi-parameter sensor array, forming a standardized water quality time-series dataset. The model training module constructs a water quality prediction model combining time-series decomposition and causal inference, learning the dynamic coupling relationship of water quality parameters through training with historical data. The trend prediction module outputs future water quality change trends over multiple time windows. The strategy simulation module simulates the effluent water quality response under different process control strategies. The decision delivery module selects and delivers combinations of compliance strategies to the execution unit. This system can accurately reconstruct the transmission logic of water quality changes and verify the matching degree between control strategies and emission standards in advance.

The application discloses a kind of unsupervised time series anomaly detection methods, the method includes obtaining historical time series data, according to the correlation of historical time series data and target screening characteristic variable, according to characteristic variable constructs instance anomaly detection task historical time series data;To instance anomaly detection task historical time series data relevant field adopts rolling statistical characteristic strategy, constructs instance anomaly detection task historical time series data model;Instance anomaly detection task historical time series data model is input into self-encoding framework after batch processing, after three times coding, three times decoding obtains reconstructed instance anomaly detection task historical time series data module;Instance anomaly detection task historical time series data module is compared and analyzed with reconstructed instance anomaly detection task historical time series data module, obtains loss time series dataset;Decision tree model is constructed, and each sample of loss time series dataset is scored, and final anomaly detection result is obtained according to score.The application can improve the ability and efficiency of time series data anomaly identification.

The present application relates to the technical field of aviation safety and artificial intelligence, in particular to an ADS-B time series data adversarial sample generation method based on improved WGAN-GP, comprising the following steps: obtaining ADS-B historical flight data, extracting the longitude, latitude, height, speed and heading five-dimensional features to construct a time series data set; building an adversarial sample generation model based on an improved Wasserstein distance generative adversarial network, and configuring a multi-objective composite loss function for the generator; the present application builds a generator and a discriminator based on LSTM, designs a composite loss function that integrates adversarial loss, concealment loss and time series smoothness loss, adopts gradient penalty and asymmetric update strategy to train the model, and the generated adversarial sample has high concealment and physical consistency, can effectively avoid various unsupervised anomaly detection models, the model training is stable, and technical support is provided for building a safer air traffic control defense mechanism.

The application provides a kind of adaptive tracking type nondestructive testing method and system for special equipment inspection, belongs to detection technical field, method includes gathering equipment historical operation and detection data to construct multidimensional time series dataset, trains intelligent model combined with deep reinforcement learning and convolutional neural network;The model is deployed to edge computing system, and the initial reference baseline of multidimension is established on site, and the current state is compared with baseline in continuous detection, through the two-way decision mechanism of intelligent model and preset logic threshold, dynamically drive mechanical arm to adjust probe pose and synchronously optimize ultrasonic excitation parameters, to actively compensate the change of coupling state, maintain the optimal detection condition, and accurately identify and evaluate defects in stable state;The system includes defect accurate identification and evaluation module, etc.;The application realizes the dynamic self-optimization of detection condition under non-stop complex working condition, improves the defect detection rate, quantitative accuracy and repeatability of detection results.

A computing device receives a user input directed to a dataset of time series data. The device converts the user input into a set of search terms, and executes a query against a search index for the dataset using the set of search terms to retrieve labeled trend events. Each labeled trend event corresponds to respective portion of a line chart representing the time series data and has a respective chart identifier. The device determines a composite score for each labeled trend event and assigns it to a group. The device ranks the one or more groups and retrieves, from the search index, data for a first subset of line charts having the respective chart identifiers of the ranked one or more groups. The device generates the first subset of line charts and displays one or more line charts of the first subset of line charts.

This invention discloses a method for intelligent monitoring and precise positioning of dust collector bag breakage based on big data time series analysis, specifically relating to the field of industrial dust monitoring technology. It involves simultaneously collecting temperature, humidity, airflow velocity, and dust concentration at the flue gas inlet and outlet and writing them into a multimodal time series dataset. The method records the operating condition control sequence, constructs a coupling feature vector containing pairwise and three-source interaction terms, recursively calculates the coupling contribution parameter vector, applies projection operator constraints, and outputs a decoupled dust concentration sequence, a residual sequence, and a coupling dominant marker. Within the coupling dominant marker indication segment, the coupling contribution is gated and converged to generate an interaction influence map and path weights. Based on the path weights and residual sequence, an interference coupling map and a sequence of handling action instructions are generated. The handling actions are written into the handling event marker sequence and handling fragment records, and an index is established for reuse, thus solving the problems of misjudgment and difficulty in closing the loop due to multi-source interference coupling.

The present application belongs to the technical field of industrial process monitoring and data visualization management, and particularly relates to a kind of air separation process whole process data monitoring and visualization management method, comprising: collecting each process section monitoring signal is obtained by pre-processing uniform time series dataset;Segment calibration obtains segmented state anchor point set and operation envelope library;According to the inter-segment coupling relationship, the time lag calibration is carried out in the load band to obtain the process propagation time lag matrix;The upstream disturbance is deduced to the expected response and compared with the measured response to calculate the consistency, to obtain the abnormal identification result;Combined with real-time margin and change rate, the touch edge time is calculated to obtain the hierarchical hidden state risk window;The above results are superimposed to the process topology map to generate the verification priority jointly weighted, to obtain the hierarchical visualization management interface.The present application can realize the active identification of cross-section disturbance propagation, the effective distinction of flexible fluctuation and internal anomaly, and the early perception and linkage visualization output of hidden state risk.

Artificial intelligence for time-series data analytics is provided. A first time-series data set is provided to a pre-trained recurrent neural network trained based on a second time-series data set. A prediction of a ground truth state of the first time-series data set is received therefrom. The first time-series data set is provided to a dynamical recurrent neural network trained based on the second time-series data set and the pre-trained recurrent neural network. A noise-reduced prediction of a ground truth system state of the first time-series data set is received therefrom. An estimate of sensor noise is read. The estimate of sensor noise is generated based on the second time-series data set and the pre-trained recurrent neural network. A prediction of a state of the system is generated based on the pre-trained recurrent neural network, the dynamical recurrent neural network, and the estimate of sensor noise.

The application discloses a kind of based on multi-modal data fusion fermentation environment control method, comprising: collection multi-modal fermentation data and executes pre-processing, forms multivariate time series dataset;Improved MTAD-GAT model is input, calculates prediction error reconstruction error and consistency score, generates dynamic reliable weight;Coupling relationship matrix is constructed across mode, coupling residual is calculated and iterative correction data, reconstructs state vector and uncertainty;Coupling stability index is calculated, and state anchoring is executed, obtains stable reference state parameter;Based on state vector and reference state, construct metabolic feasible region, form control constraint interval;Self-scaling robust controller is constructed, dynamic scaling constraint and rolling optimization calculation, output fermentation control instruction.The application realizes stable identification to fermentation process state and adaptive control to fermentation environment parameter by multi-modal data fusion, coupling relationship analysis and self-scaling robust predictive control.

This invention provides a dataset distillation method suitable for fine-tuning a time series baseline model, comprising the following steps: First, prepare the original time series dataset and the baseline model, and initialize the multi-scale parameter set; then, construct a new time series model, and calculate a temporary synthetic dataset based on the multi-scale parameter set; input the original time series dataset and the temporary synthetic dataset into the new time series model, and output preliminary results; finally, calculate and optimize the multi-scale parameter set based on the preliminary output results, and calculate the final synthetic dataset based on the optimized multi-scale parameter set. This invention can distill a large original time series dataset to obtain a smaller synthetic dataset, and use the obtained synthetic dataset to replace the original dataset for fine-tuning the baseline model, which not only reduces the training overhead of the baseline model fine-tuning but also ensures the performance of the fine-tuned model.

This invention relates to the field of communication signal anti-interference evaluation technology, specifically to a training method for an evaluation model of the anti-interference capability level of intelligent fusion terminals. The method involves acquiring signal and bit error data and aligning them by scenario to construct a time-series dataset. Interference features and bit error segments are extracted to establish temporal correlations. Changing nodes are identified to divide scenario segments into intervals. The boundaries of multiple scenario intervals are adjusted to generate a unified segmentation relationship. Data is assigned to each segment interval, and the level division boundaries are iteratively adjusted based on the temporal correlation to generate the training results of the level evaluation model. This invention, by aligning signal and bit error data by time and classifying them by scenario, analyzing energy cycling frequencies to identify interference features, establishing a temporal correspondence between bit errors and interference, forming intervals based on changing nodes and aligning and verifying across scenarios, and iteratively adjusting level boundaries to ensure consistent assignment, achieves a fine correlation between interference and bit errors, improves evaluation accuracy and segmentation stability, and enhances the anti-interference evaluation effect of terminals.

The application discloses a neural network-based operator home broadband complaint prediction method and system, and relates to the technical field of big data mining and analysis. Since home broadband problems are difficult to be quickly solved, the scheme comprises the following steps: obtaining user complaint work orders, extracting device state information and alarm information at the time of user complaints, constructing feature engineering and generating a supervised learning time series dataset after mining and preprocessing; dividing the dataset, designing a neural network architecture based on LSTM, training the neural network by using the divided dataset, and finally outputting an LSTM multivariate time series prediction model; optimizing the training results of the prediction model, introducing a time series attention mechanism and a combined loss function, and improving the performance of the model in adapting to the complaint prediction task; and deploying the optimized prediction model to an actual business system, positioning potential complaint customers and predicting complaint reasons by accessing real-time device state information and alarm information. The application can improve the efficiency of on-site handling by installation and maintenance personnel.

The present application relates to the field of information technology, and discloses a method for evaluating and predicting the adaptability of soft capsules to the environment, which comprises the following steps: generating an environmental risk sequence according to the environmental data and the logistics and storage scene data, and obtaining product tolerance analysis results through simulation analysis; predicting the potential damage probability by using the product tolerance analysis results, and determining the storage adaptability of the target product in the proposed sales area; performing an accelerated aging experiment on a product sample according to the composite environmental stress spectrum, collecting aging-related data, and determining stability indicators; constructing a time series data set, analyzing the coupling relationship and lag effect between different indicators in the time series data set; identifying a key acceleration path according to the analysis results, and correcting a shelf life prediction model; and inputting the environmental stress spectrum of different market areas into the corrected shelf life prediction model to calculate the differentiated shelf life prediction value.

The application discloses a dynamic carbon emission prediction and optimization scheduling system and method based on feature importance weight guided optimization, constructs a prediction-weight-optimization closed-loop coupling mechanism, forms a standardized multi-source time series dataset through a data acquisition and preprocessing module; an intelligent prediction module outputs future carbon emission prediction values and a dynamic feature importance weight sequence based on a time series model of a multi-head self-attention mechanism; and an optimization decision scheduling module dynamically adjusts the search process of a multi-objective evolutionary algorithm guided by the weight sequence to generate production load scheduling instructions that optimize the total cost under the premise of guaranteeing emission standards. Through the above coupling mechanism, the application realizes fine optimization of key decision variables, and improves the efficiency, quality and system adaptive ability of the scheduling scheme.

This invention discloses a method for predicting the risk of complications after esophageal surgery. The method includes: acquiring multi-parameter monitoring data collected at fixed time intervals during postoperative monitoring of patients after esophageal surgery, generating a multi-parameter time series dataset; based on a database of historical patients with no complications, grouping and matching historical patients according to age group, surgical type, and surgical duration, calculating the mean and standard deviation of each parameter at each time point after surgery for each group, and generating a standard recovery trajectory reference template containing the expected value and normal fluctuation range of each parameter at each time point after surgery; calculating the difference between the measured values ​​of each parameter of the current patient at each time point and the expected values ​​at the corresponding time points in the standard recovery trajectory reference template, and standardizing the difference by dividing it by the corresponding standard deviation; this invention achieves dynamic postoperative assessment of the risk of complications after esophageal surgery and can capture early abnormal patterns of slight deviations in multiple parameters.

The application discloses a kind of identification and classification method for karst mountain city air pollution typical process, comprising: obtaining multi-source data to construct time series dataset;Pollutant concentration is identified based on pollution process start and end period;Extract time characteristics, meteorological statistical characteristics and diffusion condition characteristics to construct process feature vector;Time type principal component analysis is combined with clustering algorithm to carry out meteorological field objective typing;Based on Lagrange trajectory model, backward trajectory analysis is carried out, and transport path characteristics are extracted;Meteorological typing and transport path characteristics are integrated, and pollution process is divided into static stable cumulative type, regional transport type and composite influence type;Typical pollution process sample library is constructed.The application can accurately identify and classify karst mountain city air pollution typical process, provide data support for pollution mechanism analysis and prediction model training.

The present application relates to the technical field of power system monitoring, and more particularly to a power stealing identification method and system based on multi-source data, an electronic device and a medium; the method comprises: collecting operation data streams and local time stamps to obtain multiple sets of time series data sets; extracting electrical quantity mutation events or non-instruction switch action events in each set of time series data sets; based on the topological connection relationship of the power line and the Kirchhoff law and the law of conservation of energy, a multi-constraint model is constructed; the clock offset of each monitoring device relative to the system logical time is iteratively calculated; the node current imbalance and the loop power imbalance are solved by inversion to output the determination result of the power stealing event. In this way, the technical problem of circular dependence between event causal inference and time stamp calibration under the condition of multi-source asynchronous monitoring data in the prior art is solved, and the reliability, accuracy and automation level of power stealing behavior analysis are improved.

Various methods and processes, apparatuses / systems, and media for modeling and training a generative model for time series datasets are disclosed. A processor partitions a training data into a plurality of segments based on a sliding window approach; inputs the plurality of segments sequentially into a GAN discriminator that generates a sequence of scores corresponding to the plurality of segments; computes a Wasserstein distance between two consecutive segments; detects change points by specifying corresponding change point with reference to difference of scores between two consecutive segments at which the computed Wasserstein distance between the two consecutive segments is the largest; models the neural SDEs with the detected change points; and trains the neural SDEs by alternating between detecting the change points while holding model parameters of the neural SDEs fixed and optimizing parameters of the GAN while holding the detected change points fixed, thereby improving performance of the generative model.

The application discloses a kind of server limit power consumption prediction method and system, it is related to server energy consumption monitoring technical field, including obtaining multi-source data and preprocessing, obtain original time series dataset, construct lag feature matrix based on original time series dataset, analyze thermal coupling derived features and carry out supplementary coding to obtain the time series feature matrix of temperature lag effect, construct multi-branch fusion deep learning model, based on multi-branch fusion deep learning model, model fitting and parameter optimization training are carried out to time series feature matrix, obtain target limit power consumption prediction model, based on the real-time feature data of target server and target cluster, hardware configuration data and computer room cabinet topology data are predicted using target limit power consumption prediction model and are handled using multi-dimensional engineering margin correction to obtain the limit power consumption prediction value of engineering adaptation single machine and cluster two levels.The application realizes limit power consumption high-precision prediction using the above method, solves the problem of prediction deviation caused by temperature lag.

The application discloses a kind of flexible supply chain resilience evaluation full-link collaborative optimization method, it is related to supply chain intelligent management technical field, including obtaining supply chain multi-source data and pre-processing, obtain structured time series dataset;Supply chain heterogeneous graph is constructed and is fused with domain knowledge graph, and supply chain semantic graph structure is obtained;Through the way that rule calculation and large language model inference are combined, generate resilience index set and identify weak link of supply chain;Order data is input into time series prediction model to obtain future demand prediction value, calculate supply chain risk prediction result;Generation candidate scheme set, and resilience index set and candidate scheme set are input into multi-objective optimization model and are solved, and collaborative instruction set is generated;Collaborative instruction set is executed and feedback data is collected, and model parameters are updated according to feedback data.The application realizes the intelligent evaluation of supply chain resilience and the accurate early warning of risk, and significantly improves the anti-risk ability and response efficiency of supply chain through collaborative optimization.

This invention discloses a method, system, device, and storage medium for predicting electric vehicle charging load, belonging to the field of electric vehicle charging load prediction technology. The invention constructs a path generation model based on user equilibrium criteria, allocates traffic flow according to OD demand, and obtains vehicle arrival flows for each charging station; then, combining the time dimension and vehicle charging behavior, it models the charging station load, constructing a charging station load time series dataset; next, it constructs a spatiotemporal prediction model integrating graph neural networks and Transformers, and trains it using the load series; finally, it uses the trained model to predict the charging load at future times. This invention organically integrates the traffic user equilibrium mechanism with a deep spatiotemporal prediction model, realizing an end-to-end mapping from travel demand, road network parameters to charging load time series, ensuring the interpretability of the physical process while significantly improving prediction accuracy and generalization ability.

The application discloses a training method and device of an anomaly detection model, equipment and a storage medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring a historical time series data set; extracting a plurality of training samples from the historical time series data set; performing mixed sample generation processing on a target time series variable to obtain a processed target time series variable and sample label information; processing a reference time series variable and the processed target time series variable respectively through two detection networks contained in an anomaly detection model to obtain a plurality of vectors corresponding to the reference time series variable and a plurality of vectors corresponding to the processed target time series variable, and determining an anomaly credibility; and adjusting network parameters of the anomaly detection model according to the anomaly credibility and the sample label. The application reduces the calculation complexity of the model and achieves more accurate and more fine-grained anomaly detection effects.

A system, method and non-transitory computer readable medium for generating validated remote sensing change images that includes a user input device for selecting high-resolution satellite images, and processing circuitry to generate a depth map and a semantic map from a static image. A change simulator determines candidate areas for change simulation and generates a change depth map and change mask focusing on objects removed from the static image. An image diffusion neural network applies a control network and stable diffusion to generate pre-change and post-change image tiles. Validation processing circuitry iterates through a validation process to validate the pair of change tiles to obtain a validated pair of change tiles and a validated change mask.

The application discloses a kind of based on data-driven super charging pile charging process segmentation regression fitting method, it is related to electrical engineering technical field, including the data of super charging pile charging process is collected and is standardized processing, generates multidimensional time series data set, based on multidimensional time series data set using equivalent capacitance-resistance model establishes the segmentation model including rising stage, constant current stage, constant voltage stage and trickle stage;Using segmentation model calculates the fitting value of each time point, obtains power sequence, constructs bayesian hierarchical model to carry out change point detection to power sequence, and uses Markov chain monte carlo method to sample change point after distribution, obtains stage boundary result.The application establishes bayesian hierarchical model by combining change point detection technology, accurately optimizes the stage division and fitting accuracy of charging process, improves the prediction accuracy, efficiency and stability of charging process.

The application discloses a low-carbon construction scheduling optimization method, comprising the following steps: collecting multi-source data through a multi-source heterogeneous sensor network, and constructing a standardized time series dataset; identifying the running state of construction equipment based on the time series dataset, and generating energy consumption data of each construction equipment in each running state; obtaining corresponding dynamic carbon emission factors according to the energy type of the construction equipment, combining the dynamic carbon emission factors and the energy consumption data, and calculating the device-level carbon emission power; based on a preset mapping relationship, mapping the device-level carbon emission power to the process level and the component level, and outputting multi-scale carbon emission data; based on the time series dataset and the multi-scale carbon emission data, constructing a Bayesian network, and identifying key factors affecting carbon emission; constructing a multi-objective optimization model, and generating a low-carbon construction scheduling strategy in combination with the key factors. The application can realize full-link tracing and scheduling optimization of construction carbon emission, and provides reliable technical support for green and low-carbon development of building construction.

This invention discloses an engine bearing fault diagnosis method based on causal graphs and feature fusion networks. The steps are as follows: S1: Collect signals from engine bearings under different operating conditions to construct a multivariate time series dataset; S2: Use the PCMCI causal discovery algorithm to perform causal inference on the multivariate time series, construct a causal graph of dynamic causal dependencies between sensors, and generate a corresponding adjacency matrix; S3: Construct a dual-branch feature extraction network: the local branch takes the causal graph and node features as input and extracts sensor topological dependency features through a graph attention network; the global branch reconstructs the time series into a spatiotemporal grid and extracts long-distance spatiotemporal dependency features through a Swin Transformer; S4: Fuse the output features of the dual branches through a convolutional block attention module to generate fused features; S5: Input the fused features into a classification network and output the bearing fault type diagnosis result. This invention can effectively improve the accuracy, stability, and robustness of engine bearing fault diagnosis under complex operating conditions.

The present application belongs to the technical field of lithium battery safety, and particularly relates to a lithium battery thermal runaway early warning method, system, device and storage medium based on an improved Transformer, the method comprising: experiment design and data collection: building an experiment platform and performing lithium battery overcharge experiments, collecting voltage data and micro-strain data during lithium battery operation; data preprocessing: constructing a time series dataset in a time series manner from the obtained data, and dividing into a training set, a test set and a validation set; constructing a prediction model: globally optimizing the hyperparameters of the WHO Transformer model, and constructing a WHO-Transformer prediction model; model training and verification: training the WHO-Transformer prediction model using the training set; thermal runaway warning: real-time prediction based on the trained WHO-Transformer prediction model, and thermal runaway warning through a prediction error threshold judgment mechanism; the method shortens the early warning time of lithium battery thermal runaway, and makes the model widely applicable to different application scenarios or different lithium batteries.