61 results about "Anomaly detection algorithm" patented technology

The application discloses a digital thread driven intelligent management system for a clinical laboratory, which comprises a system body, and the system body is operated through the following method, and the specific method comprises the following steps: obtaining patient historical records and same batch sample data from a pre-established database by collecting current test result data, and obtaining multimodal information containing original images, detection curves and trend deviations; determining potential abnormal patterns by processing the original images and the detection curves through an abnormality detection algorithm according to the multimodal information; if the abnormal patterns are inconsistent with historical comparison results, comparing previous results of the patient through a trend analysis algorithm to judge the trend deviation degree; obtaining critical project specific threshold values according to the trend deviation degree to obtain key clue recognition results; and the application aims to solve the problem that in the prior art, when facing critical value projects, comprehensive verification and rapid release cannot be balanced, resulting in poor quality control and efficiency of the clinical laboratory.

The application discloses an electronic medical record and a cataloging and classifying method, system and device thereof. The method comprises the following steps: obtaining standardized data by preprocessing multi-format medical record original data through a medical OCR model, a VAE anomaly detection algorithm, a medical knowledge graph word segmentation tool and a Transformer term standardization model; checking semantic rationality, extracting multi-dimensional features, and obtaining a comprehensive feature vector after strengthening and fusing; constructing a transfer learning classification model and training the model, inputting the feature vector to generate cataloging information; and finally, evaluating the classification result through an active learning mechanism, combining artificial labeling data and a causal forest dynamic updating framework, and incrementally learning and optimizing the model performance. The application solves the problems of low multi-format data extraction accuracy, insufficient term standardization and poor model generalization in the prior art.

The application discloses a landslide danger assessment method under rainfall conditions and relates to the technical field of landslide disaster early warning, and comprises the following steps: a multi-source sensing monitoring network is constructed, surface deformation data, deep state data and environmental parameter data of a landslide body are collected in real time, based on the data monitored by the multi-source sensing monitoring network, an abnormality detection algorithm is used to identify precursor events before landslides, and the precursor events are classified into micro-disturbance events, significant abnormality events, pre-landslide precursor events and landslide occurrence events. According to the application, a segmented state transition equation is constructed based on the three-stage theory of creep, the initial creep, constant-speed creep and accelerated creep stages are automatically identified by means of the stability coefficient and displacement acceleration, and evolution modes such as exponential decay, rainfall driving and exponential growth are respectively adopted, so that the whole process from rainfall infiltration to instability can be accurately described, and the early warning accuracy is improved.

This invention relates to the field of supply chain management technology, and in particular to a supply chain knowledge graph-driven method for diagnosing inventory anomalies. This method involves multi-dimensional data collection and preprocessing of a retail FMCG supply chain system to obtain an inventory-related dataset; constructing a multi-dimensional inventory knowledge graph based on this dataset; using the knowledge graph in conjunction with a pre-defined anomaly detection algorithm to detect inventory turnover anomalies and obtain anomaly event data; extracting anomaly event features and constructing a root cause analysis model based on the knowledge graph, performing root cause reasoning, and generating a candidate set of anomaly root causes; deeply verifying and locating key anomaly sources through the confidence propagation mechanism and path analysis technology of the knowledge graph; generating dynamic inventory adjustment strategies based on key anomaly sources and the knowledge graph, conducting simulations and effect evaluations, optimizing and implementing the adjustment plan, achieving accurate diagnosis and dynamic adjustment of inventory anomalies, and significantly improving supply chain operational efficiency and risk resistance.

This invention belongs to the field of energy efficiency management technology. It discloses an integrated system for oil well power metering and energy efficiency analysis based on the Internet of Things (IoT). The system includes: real-time acquisition and fusion processing of parameter data through an IoT sensing layer to obtain a corresponding basic energy consumption dataset; generation of standardized power metering data using harmonic analysis and phase compensation algorithms; multi-scale pattern mining and load decomposition of the standardized power metering data to construct a power load characteristic model; analysis of the relationship between production status and energy consumption to establish an energy efficiency evaluation index system and energy consumption baseline; real-time evaluation and benchmarking analysis using group clustering and anomaly detection algorithms to form an energy efficiency optimization strategy knowledge base; output of intelligent control decision sequences through multi-objective optimization; execution of the intelligent control decision sequences and adoption of a closed-loop feedback mechanism to achieve adaptive and precise equipment control and collaborative energy consumption management, effectively solving the problems of traditional oil well power management.

This application provides a single-node fault operation and maintenance method and system based on multi-algorithm fusion. The method includes: collecting time-series indicators of a single-node Linux operating system; performing anomaly detection on the time-series indicators using multiple sets of anomaly detection algorithms; performing fusion processing on the output results of the multiple sets of anomaly detection algorithms to generate anomaly events; collecting kernel-state process data within the time window corresponding to the anomaly event through multiple eBPF subroutines based on graph construction logic decomposition to construct a kernel-state causal dependency graph; completing root cause localization of the anomaly event based on the causal dependency graph; determining a fault elimination strategy based on the root cause localization; and processing the anomaly event according to the elimination strategy. This invention can improve anomaly detection coverage and root cause localization accuracy, and efficiently realize single-node fault operation and maintenance.

This invention relates to the field of financial risk assessment technology, specifically to a case risk assessment system and method based on machine learning algorithms. The system mainly includes a pre-training module and a prediction module. This invention is designed with the application of artificial intelligence algorithms in the risk assessment of suspicious cases as its core concept. Guided by theories such as supervised learning and unsupervised learning, it analyzes the performance of various features of cases within a fine-grained feature system through classification algorithms, anomaly detection algorithms, and statistical analysis methods. This process uncovers hidden and anomalous information, assesses the risk of cases, and assigns scores and rankings them. Through the application of AI technology, financial institutions can more quickly and accurately identify and handle high-risk cases, rationally allocate the time and effort of business personnel in handling cases, and promptly reduce risks and losses.

The application discloses an order risk control method and device based on an anomaly detection algorithm, computing equipment and a storage medium. According to the technical scheme provided by the application, a historical order data set is acquired, data preprocessing is performed on the historical order data set to obtain a sample data set; for each sample data in the sample data set, a plurality of residual data corresponding to the sample data are calculated according to a plurality of moving average indexes; model training is performed on a plurality of residual data corresponding to a plurality of sample data to obtain a target anomaly detection model; a plurality of residual data corresponding to real-time data are calculated according to a plurality of moving average indexes, the plurality of residual data corresponding to the real-time data are input into the target anomaly detection model for risk prediction to obtain a corresponding risk prediction result. Through the preprocessing of data and different moving average indexes, the model is adapted to the periodicity and real-time nature of data, and potential abnormal information is fully mined.

This invention relates to vibration data detection technology and discloses a vibration data trend anomaly detection algorithm based on OAKR-MEWMA fusion, comprising: data preprocessing: the raw vibration data is first standardized to eliminate dimensional differences and ensure input consistency; data transmission: the preprocessed data is input in parallel to two core modules: an online adaptive kernel regression module and an improved multivariate exponential weighted moving average control chart module; anomaly score output: the online adaptive kernel regression module captures nonlinear anomaly features through kernel function mapping. The technical problem this invention aims to solve is to provide a vibration data trend anomaly detection algorithm based on OAKR-MEWMA fusion. By constructing an OAKR-MEWMA dual-channel fusion architecture and combining it with a quantile normalization strategy, the difficulty of fusing heterogeneous anomaly scores is successfully solved, achieving simultaneous capture of sudden anomalies and gradual degradation trends. The constructed multi-dimensional trend quantification index system realizes a leap from qualitative alarm to quantitative analysis.

The application provides a single-node fault operation and maintenance method and system based on multi-algorithm fusion. The method comprises: collecting time sequence indexes of a single-node Linux operating system, performing abnormality detection on the time sequence indexes through multiple sets of abnormality detection algorithms, performing fusion processing on the output results of the multiple sets of abnormality detection algorithms to generate an abnormal event; collecting kernel state process data in a time window corresponding to the abnormal event through multiple eBPF subprograms based on graph construction logic splitting to construct a kernel state causal dependency graph; completing root cause positioning of the abnormal event based on the causal dependency graph, determining a fault elimination strategy based on the root cause positioning, and processing the abnormal event according to the elimination strategy. The application can improve abnormality detection coverage and root cause positioning accuracy, and efficiently realize single-node fault operation and maintenance.

This specification discloses a distributed screen inspection and repair control method, system, device, and medium. The method uses a monitoring module deployed on the screen terminal to collect real-time operating parameters and playback status logs of the screen device. The collected data is processed to extract key health data and generate a terminal status vector. This terminal status vector is then input into a pre-trained diagnostic model. Rule matching and anomaly detection algorithms are used to evaluate and calculate the terminal's health status. Based on the real-time health score of the terminal calculated by the diagnostic model, fault types are identified and fault levels are determined. Corresponding remote repair strategies are executed, and post-repair status data is collected to verify the repair effect and update the terminal's health status. This method effectively solves the problems of poor real-time performance, high labor costs, and slow fault recovery in existing screen maintenance, demonstrating significant technological advancement and broad engineering application value.

The application discloses a loan application process abnormality correction method based on a conditional adversarial generative network, and the method comprises the following steps: calling historical loan application process data of normal execution, firstly injecting abnormal behaviors in each normal execution track, and saving the original normal execution track; then performing word segmentation on the original normal execution track and the constructed abnormal execution track to carry out word embedding, and constructing an abnormal correction data set; then correcting the abnormal execution track by using the conditional adversarial generative network, constructing a generative model, pre-training a generator, alternately training a discriminator and the generator, and maximizing the correction ability of the generator; finally, using the generator after the adversarial training to correct the abnormal execution track. The application has the advantages that: the application can be used as downstream work of abnormality detection, and can make up for the deficiency that the abnormality detection algorithm can only detect abnormalities but cannot correct abnormalities.

This invention discloses a supplier cheating detection method based on encoding / decoding algorithms, comprising: collecting node data information; calculating correlation using mutual information, and selecting combinations based on correlation thresholds to obtain combined data features; constructing a graph structure, training a bandwidth prediction model using GCN on the graph structure, optimizing the model using the mean square error between the model output and the actual bandwidth traffic as a first loss function, and obtaining a graph embedding representation based on the optimized bandwidth prediction model; concatenating the graph embedding representation, data information, and combined data features to obtain abnormal prediction demand data; dividing the dataset based on business and corresponding loss rate thresholds, updating the overall loss function of the abnormal detection algorithm using the first loss function to obtain an abnormal detection model; using the abnormal detection model for detection, manually confirming the detection results for false positives, and processing suppliers or optimizing the abnormal detection model based on the false positive results. This invention can cope with diverse cheating behaviors and achieve efficient cheating detection.

This invention discloses an unsupervised domain adaptive anomaly detection algorithm, comprising the following steps: Step 1) Experimental verification on the UCSD dataset and the ShanghaiTech Campus dataset; Step 2) Constructing an unsupervised domain adaptive anomaly detection algorithm: the model includes a pre-training module and a domain adaptation module; Step 3) Iteratively training the unsupervised domain adaptive anomaly detection algorithm model; Step 4) Obtaining the detection results of the unsupervised domain adaptive anomaly detection model. This unsupervised domain adaptive anomaly detection algorithm performs supervised pre-training on the source domain dataset and proposes a perceptual contrast loss, which can promote the reconstruction of normal samples by the video frame reconstruction network and inhibit the reconstruction of abnormal samples by the video frame reconstruction network, clearly defining the discrimination boundary between normal and abnormal events in the source domain and introducing prior knowledge from the source domain that can well define normal and abnormal events into the target domain.

The application discloses a spacecraft intelligent anomaly detection method supporting multi-source mixed heterogeneous data types, first, for discrete data, the discrete data dictionary is used to detect data anomalies by using residual reconstruction; then, for continuous data, the analysis result of discrete data is used to prune the search space to detect anomalies; finally, for mixed type data, the above two methods are combined to adjust the algorithm optimization target, and fast and accurate detection of anomalies is realized. Compared with the traditional method, the application patent fully explores the relationship between data sparse representation and abnormal characteristics, and establishes an abnormal detection method supporting discrete signals, aiming at the discrete data processing needs of the autonomous health management and intelligent precise operation and maintenance scene of the spacecraft. In addition, based on the idea of discrete and continuous data fusion, the application designs an abnormal detection algorithm based on sparse representation, which supports the simultaneous detection of abnormal characteristics of two types of heterogeneous data.

The application provides a method for determining an anomaly detection algorithm, which is used for automatically determining an anomaly detection algorithm for performing anomaly detection on target time series data. The method comprises: determining a first target algorithm set, the first target algorithm set comprising a plurality of anomaly detection algorithms, and the anomaly detection algorithms in the first target algorithm set having the capability of performing anomaly detection on target time series data; determining a second target algorithm set according to a first target anomaly type, the second target algorithm set comprising a plurality of anomaly detection algorithms, and the anomaly detection algorithms in the second target algorithm set having the capability of detecting anomalies of the first target anomaly type; and determining a first target anomaly detection algorithm according to the first target algorithm set and the second target algorithm set. The application also provides corresponding apparatuses, computing device clusters, chips, computer readable storage media, and computer program products.

The application relates to a cabinet component intelligent assembly optimization method based on multi-modal data fusion in the technical field of intelligent manufacturing, and comprises the following steps: collecting initial position data of a cabinet component through a visual recognition system, adopting an image processing algorithm to perform feature extraction on the collected position data, and obtaining accurate coordinate information of the component; for the obtained connection state dynamic index, a data fusion algorithm is adopted to integrate multi-source data of a sensor network, whether the integrated data indicates a poor connection problem is judged, and a connection verification result is obtained; potential defect features are extracted from the obtained connection verification result, if the potential defect features show position deviation, production log data is analyzed through a backtracking algorithm, and specific process links corresponding to the defects are determined; for the obtained function verification data, an abnormality detection algorithm is adopted to identify real-time failure modes, if a failure mode is identified, a production database is updated through a data backtracking mechanism, and a final defect correction scheme is determined.

This invention relates to the field of computer technology, specifically to a real-time analysis method based on a hot-loadable standard specification package. The method includes the following steps: A layered architecture is used to design the real-time analysis module, breaking down its functions into independent modules for data preprocessing, feature extraction, and pattern matching. A standardized interface protocol is defined to unify input and output data formats. A version control mechanism manages template changes. The modular analysis framework is embedded into a stream processing task. Parallel computation is achieved using window sharding and a state backend. Task parallelism is dynamically adjusted to adapt to hot loading. Analysis latency is extracted. An anomaly detection algorithm identifies performance degradation. When a rollback mechanism is triggered, fault logs are automatically recorded, resulting in a real-time analysis module with self-healing capabilities. This invention solves the problems of traditional real-time analysis, such as difficult module coupling and updates, inflexible event rule management, high data processing latency and poor parallelism, and a lack of effective monitoring and self-healing capabilities.

The present application relates to express data processing technical field, especially in a kind of late return express order intelligent settlement method, device and equipment and storage medium, the method gets rid of the subjective limitations of traditional late return examination depending on fixed rule and artificial processing, standardization is handled to the late return data of multi-source heterogeneous by preset knowledge graph and abnormal detection algorithm, the dynamic prediction and late return day intelligent calculation of return time limit are carried out in combination with long short-term memory network, realize the objective quantification and active prevention and control of late return examination;With the settlement risk predication mechanism of ladder pricing rule and classification model, ensure the accurate accounting of the amount to be deducted and the differentiated adaptation of settlement execution;The method solves the problems of data governance loss, settlement failure frequently occurs and other problems in the existing late return examination, reduces the labor cost and time cost of late return examination, improves the accuracy and fairness of the examination result, provides systematic technical support for the late return management of express industry.

The application discloses a kind of prediction and abnormality detection algorithm and application for time series data, belong to time series data abnormality detection technical field, comprising: the original time series data of input is preprocessed, and the standardized sample segment is generated;Multi-scale feature fusion time series prediction model is constructed and trained;Using the trained time series prediction model, the standardized sample segment is multi-step forward prediction, and the predicted sequence is obtained;Based on the predicted sequence and the corresponding actual observation sequence, dynamic residual threshold is calculated, and real-time abnormality detection and marking are carried out according to the threshold.The application solves the problem that the existing complex space-time correlation feature is not sufficient, cannot dynamically adapt to data change and model cross-scene migration and online learning ability is weak.The application fully captures complex nonlinear space-time correlation features, improves the adaptability of abnormality detection to data nonstationarity and concept drift, realizes online self-adaptation and threshold drift tracking, and improves the practicability and long-term robustness of the algorithm.

The application provides a kind of aircraft surface frost detection method based on laser thickness measurement and visual image analysis, which comprises the following steps: S1, build aircraft surface frost detection system;S2, train and build image feature reference library under ice-free state;S3, initialize setting to detection system;S4, calculate frost layer thickness;S5, collect aircraft surface image and calculate abnormal value;S6, collect image of ground area and calculate abnormal value;S7, normalize and weighted fusion calculation are carried out to multi-source data;S8, judge whether the fusion value is greater than the preset threshold, if the fusion value is greater than the preset threshold, there is ice and frost in the current detection area of aircraft surface, and the detection system will trigger defrosting prompt alarm and defrosting. The application constructs a multi-modal fusion scheme based on laser scattering model and visual image anomaly detection algorithm, considers the complex mechanism and optical properties of frost layer formation, this method has higher robustness to environmental interference, and can trigger alarm and deicing operation in time, effectively reduces the burden of aircraft maintenance personnel, enhances the safety and efficiency of aircraft ground operation.

This invention relates to the field of network security technology and discloses a method for dynamic mapping of cyberspace based on a large model. The method includes: Step 1, using a BERT model as the core feature extractor to capture deep semantic features of asset banner information and identify asset changes; Step 2, classifying asset types to mark the asset change process; Step 3, extracting newly added assets and using an unsupervised anomaly detection algorithm to identify newly added assets significantly different from normal patterns. Using this invention, the accuracy and efficiency of identifying dynamic changes in asset addresses can be improved, newly added assets can be accurately discovered, the situation of newly added assets in cyberspace can be clearly depicted, and the discovery process of new devices can be accelerated.

The application belongs to the technical field of resource scheduling, and discloses a mine safety and control linkage cloud and fog collaborative resource scheduling method, which classifies and disassembles mine safety and control linkage business and models, extracts safety and resource demand characteristics of various businesses, combines an LSTM deep learning model to construct a resource demand prediction system, completes safety event level division through a mine safety risk assessment model, and then uses a reinforcement learning algorithm to construct a dynamic priority allocation system, realizes real-time dynamic adjustment of task priority, generates a preemptive execution queue by matching a preemption threshold mechanism and a buffer period, deeply integrates safety attribute priority into the core logic of business and resource supply and demand matching, so that high-priority core safety tasks can preempt resources; the main scheduling node is set in each underground fog domain to complete high-frequency perception of local resources, and after invalid data is removed through an anomaly detection algorithm, effective resource data is subjected to cleaning standardization and pooling aggregation processing.

The present disclosure provides a pipeline anomaly detection device, which is a device for detecting pipeline anomalies, including: a sensing part including a plurality of sensors that sense a distributed signal through an optical fiber attached to one side of a pipeline to obtain sensing information; and a control part that determines one anomaly detection algorithm from among a plurality of anomaly detection algorithms based on a training completion degree, and determines whether the pipeline is abnormal based on an output value of the determined anomaly detection algorithm derived based on application of the sensing information as an input value.

The application provides a glass deep processing process analysis system and method based on AI cloud data, the method comprising: performing semantic analysis and structured packaging on the obtained process query text in combination with a glass industry term library to obtain a structured query instruction; performing database table association, list matching and context enhancement on the structured query instruction based on a bidirectional mode link algorithm, and performing sentence optimization on the matching result after context enhancement based on a binary selection strategy to obtain a primary query sentence; performing logical self-correction on the primary query sentence based on a multi-round syntax tree traversal method to obtain a database query sentence; querying the original data set from the glass processing database by using the database query sentence, and performing multi-dimensional data analysis on the original data set based on a dynamic threshold anomaly detection algorithm and a mutual information algorithm to obtain an analysis result set; and generating an interactive report according to the analysis result set.

The application provides a commodity master data anomaly checking method and system, wherein the method comprises the following steps: collecting commodity master data of in-store inventory on a daily schedule, and classifying the commodity master data according to commodity categories; based on an unsupervised learning anomaly detection algorithm, dividing the commodity master data of each commodity category into normal data and pending abnormal data; according to the statistical distribution characteristics of the normal data, obtaining the normal value interval of the commodity master data of the corresponding commodity category; marking the pending abnormal data falling outside the normal value interval as suspected abnormal data; and sending the normal value interval of the commodity master data of all commodity categories and the suspected abnormal data to manual review. The application solves the problem of low efficiency of manual review of commodity master data.

This invention relates to the field of anomaly detection technology, specifically to a multi-industry-adaptive technical foundation detection method and system. It is used to perform basic state detection on multivariate operational data generated by equipment, components, process units, or production line links in multiple industry scenarios. The method involves: offline, acquiring historical datasets with state annotations, establishing a candidate detection algorithm library, and training, validating, and evaluating the performance of each candidate algorithm to form associated metadata; extracting univariate statistical features, multivariate correlation features, anomaly distribution features, and overall structural features from each historical dataset, and obtaining a unified meta-feature representation through screening, aggregation, and embedding transformation; training a multi-output performance prediction model using the unified meta-feature representation as input and the candidate algorithm performance vector as output; online, extracting the meta-features to be detected from the industry object to be detected and inputting them into the model to obtain the predictive detection performance of each candidate algorithm, selecting a single target detection component or determining multiple algorithms and their integrated weights, and outputting at least one of anomaly score and state label to obtain the anomaly state detection result. This invention can solve the problem that anomaly detection algorithms are difficult to adapt quickly and accurately based on human experience when there are diverse types of industry objects, large differences in operational data structures, and frequent changes in operating conditions in multi-industry scenarios.

This invention specifically relates to a method and system for generating abnormal safety production images based on text and images. The method involves multimodal data acquisition and preprocessing, constructing a dataset containing normal reference images, real abnormal images, and corresponding text descriptions. It employs multimodal feature association modeling based on hypergraphs, extracting multimodal features and constructing a hypergraph structure, outputting fused feature vectors through hypergraph convolution. A Transformer architecture denoising network with a cross-attention module is built. A conditional diffusion model is established, defining forward denoising and reverse denoising processes, using fused features as strong conditions to guide denoising. The model is optimized using a mean squared error loss function, with normal images and abnormal description text input during inference, iteratively denoising to generate the target abnormal image. This method can generate realistic abnormal images that conform to the text descriptions while preserving the original equipment structural features, effectively supplementing the negative samples required for industrial safety detection and improving the robustness and generalization ability of downstream abnormal detection algorithms.