161 results about "Information consistency" patented technology

The invention provides a semantic vision SLAM method based on depth mask segmentation in a dynamic environment. The semantic vision SLAM method comprises the following steps: inputting an RGB image and a depth image into a tracking thread for preprocessing; inputting the preprocessed RGB image into a semantic segmentation thread for precise detection and segmentation of a scene target, obtaining a semantic tag of each pixel, and generating a segmentation mask; in combination with the depth information consistency of the preprocessed depth image, correcting and optimizing the mask to obtain a mask image; inputting each frame of corrected mask image into a dynamic feature screening thread, and determining dynamic feature points of potential dynamic objects; epipolar constraints are constructed through matching points in a background region of the RGB image, and dynamic feature points in a high dynamic object and a potential dynamic object in the RGB image are eliminated; and merging and splicing local point clouds corresponding to each group of key frames in the RGB image after the dynamic feature points are eliminated through a point cloud mapping thread, and constructing a global three-dimensional point cloud map without the dynamic objects. According to the method, the mapping precision is improved.

The invention provides an illusion detection method and device based on semantic embedding and structure entropy detection. The method comprises the following steps: aiming at an input question, generating a plurality of possible answer sets by utilizing a trained language model; converting each answer into a vector by using a pre-trained sentence embedding model, and calculating cosine similarity between the answers so as to construct a similarity graph; representing the similarity graph through a coding tree, and calculating the structure entropy of each node to identify nodes with relatively high semantic uncertainty; dynamically adjusting a threshold value by adopting a moving average method to ensure that high-entropy output is marked as illusion in time; and the answers marked as illusion are eliminated, so that the accuracy of the output answers and the information consistency are ensured. According to the method, the performance of the model on complex problems is effectively improved, and the generation of misleading information is reduced. Preliminary experiments show that according to the method, the illusion output proportion is remarkably reduced on multiple data sets, and the semantic consistency and accuracy of generated answers are enhanced.

The invention discloses an intelligent interaction system based on a large language model and a knowledge graph, and relates to the technical field of artificial intelligence and intelligent interaction. According to the system, data processed by a data acquisition and preprocessing module is stored in a local knowledge warehouse in a three-layer nested structure; the multi-modal semantic understanding module constructs a knowledge system in which a triple knowledge graph and a vector database are complementary by optimizing a BERT model and performing dual-channel analysis; the intelligent interaction module realizes natural language interaction and business task automatic triggering based on an RAG technology and a dialogue memory mechanism; the information change identification and analysis module generates a personnel change risk report and performs early warning; and the summary report automatic generation module outputs a structured report. According to the method, the problems of data splitting, complex interaction and slow response in traditional enterprise management are solved, the functions of system intelligent question answering, personnel change risk analysis, structured report automatic generation and the like are realized, and the management efficiency and the information consistency are improved.

The invention provides a bid invitation risk identification method and system, electronic equipment and a storage medium, and aims to solve the problems of single bid invitation form and low judgment accuracy in consideration of bid invitation risk, and the method comprises the steps: building a bidding relation matrix for bid invitation projects according to bid invitation activities, bid winners and participant information; historical bidding and tendering data is cleaned and combed, and project risk evaluation information is obtained based on the bidding relation matrix and comprises a bid rate, a bid winning rate, a quotation gap and an enterprise information consistency result; performing unsupervised learning on the project data, calculating a project risk value according to the risk evaluation information, and judging whether the project needs text analysis according to the project risk value; similarity analysis is carried out on the technical part and the price part of the bidding document for the project needing to be analyzed; and for the high-risk-value project, combining the technical part similarity and the price part similarity of the bidding file to judge whether the bidding and confusion bidding exist or not. According to the invention, the accuracy and comprehensiveness of bid enclosing and bid stringing analysis can be improved.

The invention provides a multilevel multi-model collaborative compliance detection system and method based on a large model, and the system comprises a content analysis and verification module which is used for extracting text information in to-be-detected content and correcting errors in the extracted content; the compliance detection module is used for detecting compliance; the consistency detection module is used for detecting the consistency; and the detection result rechecking module is used for rechecking and checking the detection result based on the compliance criterion. According to the technical scheme, compliance detection can be carried out on various types of materials, and compliance detection capability is flexibly constructed for compliance detection of specific professions and scene vertical fields; the information consistency of specific professional and scene contents can be flexibly detected.

The embodiment of the invention provides an ontology-based heterogeneous modeling language unified description method, device and equipment. According to the method, a heterogeneous modeling language used by a multi-architecture model is obtained; carrying out analysis and disassembly processing on the heterogeneous modeling language based on the element category of the multi-architecture model ontology to obtain a target element; the multi-architecture model ontology is a predefined standard semantic architecture. Mapping and storing the target element into the multi-architecture model body according to a mapping rule, and determining an element name of the target element; the mapping rule is used for indicating a corresponding relationship between the standard semantic architecture and the heterogeneous modeling language. And performing unified description on the heterogeneous modeling language according to the element names. According to the method, the interoperability and the information consistency among the models are realized by uniformly processing heterogeneous modeling languages in the multi-architecture model.

The embodiment of the invention provides a consistency directory access method, a directory controller and computer equipment, and the method comprises the steps: obtaining an access request which is used for accessing cache information of a target cache data block in a consistency directory; the consistency catalog comprises a coarse-grained catalog and a fine-grained catalog; judging whether the access request hits a directory cache or not, wherein the directory cache caches a part of items with a consistent directory; if the access request hits the directory cache, judging whether the access request causes the state change of the target cache data block in the consistent directory record or not; and if the access request does not cause the state change of the target cache data block recorded in the consistent directory, determining response information of the access request based on cache information of the target cache data block hit in the directory cache. According to the embodiment of the invention, the access overhead of the consistency directory can be reduced, and the performance of a computer system is improved.

The invention discloses a direct-current GIS (gas insulated switchgear) partial discharge detection method, which comprises the following steps: acquiring multi-modal sensing data in a direct-current GIS, the multi-modal sensing data comprising any two or more of current sensor data, optical sensor data, ultrasonic sensor data, infrared sensor data and ultrahigh-frequency sensor data; the collected multi-mode sensing data is preprocessed; performing feature extraction on the preprocessed multi-modal sensing data by using a deep learning model; and synthesizing the extracted features by using data fusion to obtain partial discharge information consistency explanation of the multi-modal sensing data. According to the method, multiple partial discharge detection methods can be comprehensively configured according to the requirements of the operation condition of the direct current GIS, so that the partial discharge condition of the direct current GIS can be more effectively evaluated.

The invention relates to the technical field of artificial intelligence intention recognition, and discloses a large model intention recognition method and system based on context awareness, and the method comprises the steps: obtaining a text sequence inputted by a user in multiple rounds of dialogues; calculating inter-sentence semantic distance and emotional tendency, extracting keyword co-occurrence frequency and theme continuity, and analyzing lengthy degree; if the lengthy score is relatively high, extracting a memory fragment and performing end-to-end modeling to generate a context semantic vector; dynamically optimizing an intention recognition strategy based on the behavior pattern and the questioning frequency, and generating a clarified question; obtaining a user response, calculating semantic similarity and information consistency between the user response and the recognition result, and confirming a final intention; and generating a personalized response in combination with the context features and the response time delay. According to the method, the dynamic interaction mode of the user can be adapted, and the complex intention depending on the context in multiple rounds of conversations can be processed.

The invention discloses a scientific and technological achievement value estimation system, and the system comprises a data collection module which is configured to obtain technical data, market data, legal data and social data through an API interface; the feature fusion module is configured to adopt a heterogeneous data converter to convert the multi-source heterogeneous data into a unified feature space, and the feature fusion module comprises a text data conversion unit, a numerical data conversion unit and a time sequence data conversion unit; the correlation analysis module is used for constructing a cross-domain knowledge graph; the dynamic evaluation module is configured to adopt an attention mechanism weight distribution model; by integrating the multi-source heterogeneous data acquisition module, the system can comprehensively collect information from technologies, markets, laws and society, limitation of a single data source is avoided, the cross-modal alignment unit realizes alignment of multi-source data spaces through comparative learning, information consistency among different data sources is ensured, and the accuracy of the multi-source data space alignment is improved. And the reliability of the evaluation result is further enhanced.

The invention provides a rapid data acquisition method based on fusion of intelligent number asking and an industrial connection technology, and the method comprises the steps: obtaining industrial protocols of a plurality of devices and configuration parameters corresponding to the devices, and generating an acquisition model of each device; generating a structural fingerprint for each acquisition model; performing matching degree analysis on the plurality of acquisition models through multi-modal similarity calculation on the basis of the structure fingerprint, screening the plurality of acquisition models according to a matching degree analysis result to perform uniqueness management, and performing association mapping on the equipment and the corresponding acquisition model to obtain a de-duplicated acquisition model; edge gateway configuration accurate issuing of the collection model after duplicate removal is realized through a two-way transmission channel, and consistency is guaranteed by combining real-time structure fingerprint return and matching degree analysis. Through the AI-driven process arrangement and semantic understanding technology, the modeling efficiency, the information consistency and the issuing capability of the data acquisition system are remarkably improved.

The invention discloses a vehicle diagnosis method, device and equipment and a storage medium, and relates to the technical field of vehicle diagnos.The vehicle diagnosis method comprises the steps that when operation of diagnosis equipment is monitored, an equipment serial number corresponding to the diagnosis equipment and a computer MAC address of a target computer operating the diagnosis equipment are obtained, and whether the target computer is in an online state or not is judged; if the device is in the online state, judging whether the serial number of the device is registered or not; if the to-be-diagnosed vehicle is not registered, performing registration based on the user information, the equipment serial number and the computer MAC address, and performing vehicle diagnosis on the to-be-diagnosed vehicle based on the preset authentication file and the diagnosis equipment after the registration succeeds; if not, judging whether the diagnosis equipment has a preset authentication file or not; if yes, judging whether the preset authentication file meets a corresponding preset information consistency condition, a preset offline legal condition and a preset offline use condition or not; and if yes, performing vehicle diagnosis on the to-be-diagnosed vehicle based on a preset configuration code. And dependence on a network environment is avoided.

The invention relates to the field of Internet of Things data processing, and discloses an Internet of Things data quality screening method and system based on multistage intelligent evaluation, and the method comprises the following steps: (1) carrying out the primary screening of a sample based on data energy distribution: dividing windows for an input Internet of Things data flow, calculating the energy value of each window, dynamically adjusting a screening threshold value according to the energy distribution, and carrying out the initial screening of the sample; high-energy samples are reserved; and (2) sample screening based on time sequence information consistency: carrying out frequency domain analysis on the primarily screened samples, and determining main frequency and co-frequency bands in the samples through signal analysis. According to the method, data energy distribution is automatically analyzed through an iterative algorithm, the screening threshold value is dynamically determined, manual rule presetting is not needed, the method adapts to energy heterogeneity under different equipment models, sampling frequencies and transmission conditions, the threshold value is periodically updated in a real-time data stream scene, the method adapts to environment changes, and high-value data missing or noise misjudgment is avoided.

The invention discloses a method and a device for realizing configuration information consistency in a test environment, the method is suitable for a plurality of test environments, and each test environment is provided with a business database corresponding to a respective business system; the configuration information in any service database comes from the corresponding reference database in the operation environment; aiming at a first database operation statement triggered on any business system in any test environment, determining whether the first database operation statement is used for operating a configuration table in the business database or not; if yes, the first database operation statement is converted into a second database operation statement, and the second database operation statement is executed through the reference database; each piece of configuration information is recorded in the configuration table. By adopting the method, the change of the configuration information in each test environment can be updated in real time in the reference database, and the accuracy of the configuration table in the reference database is improved.

The invention discloses an AI-supported multi-modal desktop interactive projection method and system, and the method comprises the following steps: synchronously collecting visual signals, tactile data and voice input in a desktop projection environment, and obtaining a multi-modal information set after data fusion; calculating priority weight distribution of each mode according to the multi-mode information set, and obtaining a high-priority mode sequence; if the visual signals in the high-priority modal sequence are judged to be dominant, extracting a visual dominant feature group; performing cross validation on the visual dominant feature group and the tactile data, and judging information consistency after data fusion; according to an information consistency result, dynamically adjusting transparency control parameters of projection layering to obtain optimized hierarchical configuration; and updating the projection display content according to the optimized hierarchical configuration, and outputting the final projection content. According to the method, the interaction real-time performance and immersion of desktop projection are improved, and finally the technical effects of optimizing user experience and improving system intelligence are achieved.

The invention discloses a panoramic activity recognition method based on adaptive two-way propagation prototype learning, and the method achieves the combined recognition of multi-granularity activities in a panoramic scene through an adaptive focusing detector and a two-way propagation prototype learning device. In a self-adaptive focusing detector, individuals are effectively detected through a panoramic adaptive focuser in a coarse and fine combination mode, and changes of individual sizes and shielding in a crowded panoramic scene are dealt with. In a two-way propagation prototype learning device, closed-loop interaction and information consistency among different granularities are promoted through the two-way propagation prototype learning device, and information loss caused by inaccurate positioning is reduced. The method comprises a detection branch and an identification branch, the detection branch outputs an original detection frame and inputs the original detection frame into a panoramic adaptive focuser to obtain a final size adaptive detection result, and the identification branch encodes a video and obtains different levels of feature representations for identification through a bidirectional propagation prototype device. And the overall framework is subjected to joint optimization through detection loss and multi-granularity identification loss.

The embodiment of the invention provides a network security information consistency representation method and device based on ontology modeling, and the method comprises the steps: classifying various entities related to network space security, and the relationships and attributes of the entities, and constructing an ontology model of a hierarchical structure; concept conflicts and ambiguity of multi-source data in the ontology model are eliminated through context binding, multi-view reconstruction and fuzzy reasoning; data is injected in real time through a heterogeneous data source to trigger ontology updating, and an ontology model is updated through a collaborative reasoning engine and an atomization updating operation set; and performing cross-level consistency verification on the static structure and the dynamic behavior in the ontology model, and verifying the robustness of the ontology model through a antagonism test. According to the scheme, the heterogeneity, conflict and dynamic variability of multi-source data in the field of network security are effectively solved, so that more accurate and robust security management and decision are realized.

The invention discloses a rolling bearing fault diagnosis method based on time-frequency fusion and a double-branch deep network, and relates to the technical field of mechanical equipment state monitoring and fault diagnosis, and the method comprises the steps: carrying out the time-frequency joint representation of VMD and Multi-taper PSD on a single-channel vibration signal, rearranging the signal into a fixed-size gray matrix according to a time sequence, and carrying out the fault diagnosis of the single-channel vibration signal; a time sequence is reserved, and stable spectrum energy estimation is obtained; parallel double branches are constructed on network modeling; hierarchical window self-attention modeling is carried out on a frequency domain branch, and long-time-history dependence and key moment attention are enhanced by a time domain branch through TCN-CBAM; an XPAF module is introduced, pairwise cross attention is taken as a core, priori bias B and a directional mixed structure are combined, feature alignment and a residual path are matched, and time-frequency information consistency and efficient information flow are ensured. According to the method, the separability of weak faults and the robustness under noise / variable working conditions are remarkably improved, and the method is suitable for online monitoring and intelligent diagnosis of rotating machines such as bearings.

The invention provides a deep forged video detection method and device based on identity and space-time inconsistency. The method comprises the following steps: acquiring a to-be-detected video and a forged video detection model; the forged video detection model comprises an identity comparison network, a space comparison network, a frame comparison network and a feature fusion network; performing identity information consistency detection on the to-be-detected video based on the identity comparison network to obtain identity information features; performing spatial inconsistency detection on the to-be-detected video based on the spatial comparison network to obtain spatial inconsistency features; extracting time inconsistency characteristics of the video to be detected based on the frame comparison network; and fusing the identity information feature, the space inconsistency feature and the time inconsistency feature by using a feature fusion network to obtain a forged video detection result. By combining identity information consistency detection, space inconsistency detection and time inconsistency detection, the authenticity of the video can be comprehensively analyzed, and various counterfeiting techniques and novel counterfeiting technologies can be coped with.

The invention relates to the technical field of wind power plant AGC (Automatic Gain Control) systems, and discloses a video decoding and model fusion-based AGC system reliability early warning system, which is characterized in that monitoring equipment is accessed through a bypass of a video acquisition security access module, and an interface video stream is captured in real time; a video analysis feature extraction module synchronously decodes and extracts visual state features and multi-source data features; the intelligent diagnosis reasoning module independently diagnoses a system dead state based on visual features, and verifies information consistency and a locking state through a logic knowledge base based on multi-source data. And finally, the dual effects of realizing reliable external detection on a dead halt state with complete internal monitoring failure and realizing deep intelligent diagnosis on multi-source information logic conflicts are achieved, and the problems of blind area and shallow alarm of traditional monitoring are solved.

The invention provides a crack depth prediction method based on an infrared thermal imager and deep learning, and relates to the technical field of crack prediction.The method comprises the steps that a cooling infrared thermogram sequence and a visible light image of a crack area under thermal excitation are synchronously collected, and multi-modal features are extracted after registration fusion; based on consistency verification of heat conduction inversion, a width-depth empirical relation and a traditional detection means, a crack depth label with high confidence is automatically generated, and a data set with the label is constructed; and finally, designing a deep learning model containing attention enhancement and multi-scale fusion, learning a mapping relation between an input image and depth, and realizing accurate prediction of the crack depth. According to the method, a multi-source information consistency verification mechanism is introduced, physical model inversion, experience statistics and measured data are mutually verified, the bottleneck problem that high-quality deep labels are difficult to obtain in engineering practice is effectively solved, and a reliable training data basis is provided for a deep learning model.

The invention discloses a medical signal generation method and system based on a time-frequency double-domain constraint generative adversarial network. According to the method, a multi-category electrocardiosignal generation model is designed on the basis of a conditional generative adversarial network, global attention interaction is carried out by utilizing a learnable potential feature prototype and position-coded signal features through an adaptive neural representation module, and long-distance time sequence dependence of electrocardiosignals is modeled; through the adaptive spectrum sensing module and the dynamic mask mechanism of energy sensing, the discriminator intensifies the discrimination of the consistency of signal frequency domain information while paying attention to the time domain form, thereby effectively inhibiting the non-natural high-frequency artifacts in the generation process, and ensuring the high fitting of the generated signal and the real signal. Experiments show that the fidelity and diversity of generation of various types of electrocardiosignals can be remarkably improved.

The application discloses a planning calibration method and system for an information analysis intelligent assistant, and comprises the following steps: in each planning round, a candidate strategy set is generated by a planning agent, the candidate strategy set contains an information retrieval target, a retrieval keyword combination and an information processing flow; the candidate strategy set is evaluated by multiple agents with heterogeneous private memories, a strategy selection module based on the agents is constructed for information consistency, and a strategy with the most stable evaluation result is selected as an optimal strategy of the round; a difference between a self-selected strategy of the planning agent and the selected optimal strategy of the round is recorded, a calibration constraint module guided by consistency is constructed, and the difference is converted into a cognitive calibration constraint and stored in the memory of the planning agent; strategy selection optimization is realized in single-round planning by relying on the strategy selection module based on information consistency; and the cognitive calibration constraint generated by the calibration constraint module guided by consistency is combined, so that the constraint is transmitted between different rounds and gradually accumulated, and a constraint is formed on strategy generation of the planning agent; and the trained agent is applied to an open domain webpage retrieval task to generate accurate retrieval results.

The present application belongs to the technical field of model-based system engineering, and relates to a laser fusion superfast imaging instrument logic design method and system based on MBSE, which comprises: based on the demand information of the superfast imaging instrument, system indexes of the superfast imaging instrument are obtained through operation analysis; based on the system index identification, components required to achieve the system indexes are obtained, and component indexes are configured through decomposition of the system indexes; based on the component index identification, modules required to achieve the component indexes are obtained, and module indexes are configured through decomposition of the component indexes. The method establishes a multi-level two-way traceable relationship matrix in the collaborative development of the superfast imaging instrument, establishes a two-way traceable link among user demand, service use case and indexes at each level, ensures that any modification can be subjected to impact analysis, and improves information consistency; through standardized and visualized models such as service use case and activity diagram, a common understanding basis without ambiguity is provided for different teams, and team collaboration is promoted.

This disclosure relates to systems and methods for management of information, including environmental information, obtained by a variety of sensors associated with one or more distributed mobile sensor platforms. In certain embodiments, the geographically transitory nature of a mobile sensor platform may be leveraged to facilitate collection of environmental information over a larger geographic area than that of a fixed sensor platform. Embodiments disclosed herein provide for information consistency and / or quality checking of information obtained by mobile sensor platforms. Further embodiments may be used to incentivize the collection and / or acquisition of certain data via point and / or credit-based compensation.

The application discloses a blockchain-based ticket settlement verification method and system, relates to the technical field of financial transaction safety, and comprises the following steps: S1, receiving a cross-border order submitted by a buyer, verifying the order integrity and the authenticity of the buyer's identity by using a digital signature algorithm, analyzing and extracting the buyer's information, the HS code of the commodity, the payment amount and the currency, and completing the consistency comparison of the ticket basis information in combination with the corresponding tax payment price interval of the HS code; S2, calling an off-chain risk control engine to carry out multi-dimensional compliance verification in parallel, generating a comprehensive risk score and dividing a risk level; after the verification, a preset aggregation template is matched based on the risk level, a unique feature identifier of the order is generated by using a hash algorithm; the risk level is adaptively matched with batch aggregation parameters and on-chain priority, and differentiated dynamic scheduling of off-chain batch evidence is realized. The application realizes full-dimensional pre-checking to avoid false transaction risks, and builds a risk level adaptive adjustment mechanism to balance business processing efficiency and risk control safety.

According to the CTC system multi-terminal multi-language display method, the data transmission display method and the terminals, a standardized data exchange format between a station autonomous machine and each event generation terminal is defined, so that the autonomous machine only processes standardized codes irrelevant to languages, and thorough decoupling of core business logic and language presentation is achieved. And meanwhile, a broadcast mode of centralized processing and distributed rendering is adopted, and an autonomous machine is used as a single information source to broadcast an operation result to all terminals at one time, so that the consistency and the real-time performance of information are ensured from the source. Besides, multi-language adaptive responsibilities are sunk to the terminals, and the terminals dynamically load language packages and complete conversion from codes to texts, so that dynamic pluggable localization support is realized. In addition, multi-language display and dynamic switching are achieved at low cost by matching with translation setting of an operating system and a small amount of code conversion, and the architecture advancement, information consistency, maintainability and internationalization adaptability of the CTC system are remarkably improved.

The specification discloses a task execution method and device, a storage medium and an electronic device, specifically comprising: generating original task execution instructions corresponding to a target task according to obtained task configuration information and sending the original task execution instructions to a target terminal device for execution. When it is determined that the task configuration information corresponding to the target task is changed, the task execution instructions can be updated according to the changed task configuration information, so that the target terminal device can execute the target task according to the updated task execution instructions. Through the method in the specification, the information consistency between the server side scheduling the task and the terminal device executing the task can be effectively ensured, and the terminal device executing the task according to the old configuration information after the configuration information is changed can be avoided, time consumption and execution cost are greatly saved, the execution efficiency of the target task is effectively improved, and the overall promotion speed of the business to which the target task belongs is also improved.

This invention discloses an image semantic segmentation processing method with adaptive context information extraction, belonging to the field of image processing technology. The specific steps of the method are as follows: First, basic features in three domains—spatial, frequency, and semantic—are collected and regularized into an original set. Then, cross-domain mapping is used to generate globally interconnected and fused features. Next, based on regional attributes, the features are decomposed into low-frequency structures and high-frequency detail components, and their proportions are adjusted to extract complete contextual features. Finally, after consistency verification and optimization of the three-domain information, pixel category determination is completed, and a high-precision semantic segmentation result is output. This invention solves the core technical problems of insufficient information coordination, difficulty in balancing global structure and local details, and inability to correct information conflicts between dimensions in traditional methods through a full-chain adaptive information processing mechanism. This improves the accuracy and regional clarity of semantic segmentation, making the segmentation results more consistent with actual scene requirements.

The application relates to the technical field of real-time monitoring, in particular to a data background real-time monitoring processing method and system based on plug-ins, which comprises the following steps: obtaining an activated plug-in thread identifier, analyzing field output content to extract a name path and a write state to establish a mapping structure, pairing path fields to identify name coincident data groups, judging field write time window overlap and buffer channel consistency to extract conflict path numbers, processing field order and delayed writing and updating the write state, and obtaining a monitoring field processing state list; in the application, plug-in field mapping structure is constructed to bind fields and output paths, conflict paths are identified, write time windows and buffer channel consistency are compared, a field order comparison mechanism is established to control repeated field delayed writing, a field control state table is generated, a monitoring processing state list is output according to field attribution and a function block, field behavior and control information consistency is guaranteed, and monitoring data processing integrity and logical reliability are improved.