75results about How to "Improve analytical ability" patented technology

The invention relates to the technical field of computer processing, in particular to a software architecture code generation method and system based on a large language model, and provides the following scheme: obtaining demand information input by a user, analyzing demands by using a natural language processing technology, and extracting functional elements; performing classification processing on the functional elements through a clustering algorithm to generate a functional module group; mapping the function module group to a system architecture layer, and generating an architecture diagram represented by UML (Unified Modeling Language); and generating a corresponding source code according to the UML graph, and optimizing the source code. The optimization strategy comprises static analysis of format problems and potential vulnerabilities in identification codes and performance optimization of complex functions. Through combination of the large language model and the automatic code generation and optimization technology, the software development efficiency is improved, manual intervention is reduced, and the correctness and maintainability of the generated code are ensured.

The invention relates to an intelligent semantic analysis knowledge graph system for multi-source heterogeneous data. The system comprises a data preprocessing module used for preprocessing to-be-processed multi-source heterogeneous data to obtain format-standardized text data; the semantic analysis module is used for performing natural language processing on the text data to obtain words, and then performing vocabulary layer, syntactic layer and semantic layer analysis on the text data and the words to obtain structured data; the structured data comprises a triple consisting of entities, relationships and events and a network diagram of the entities and the relationships; the model structure mapping module is used for mapping the structured data into a predefined structure of a knowledge graph to obtain structured graph data; and the knowledge graph module analyzes entities and relationships in the graph data, performs reasoning to obtain potential relationships in the graph data by utilizing a logic rule and a machine learning algorithm, realizes cross-data-source and cross-document deep semantic association, and improves the efficiency of large-scale data processing and knowledge management.

The invention relates to the technical field of wireless communication testing, and particularly discloses a Bluetooth headset anti-interference performance testing system which comprises an interference signal generation module, an interference signal dynamic simulation module, a signal stability monitoring and evaluation module and an anti-interference capability identification module. An interference signal comprising a modulation signal, a cross-band noise signal and a dynamic radio frequency signal is generated and injected into a test environment, and a real interference scene is simulated by using dynamic conversion of signal strength and frequency; the monitoring module obtains the intensity and frequency characteristics of the interference signal in real time, and comprehensively evaluates the signal stability based on an intensity abnormal coefficient and a frequency abnormal coefficient; and the anti-interference capability identification module calculates an anti-interference capability coefficient by using a support vector machine in combination with performance data of the Bluetooth earphone under different interference levels, and realizes graded evaluation of earphone performance.

The invention relates to the technical field of power distribution cabinet monitoring, in particular to a power distribution cabinet for realizing environment monitoring based on a sensor network, which realizes omnibearing real-time monitoring of the surrounding environment of the power distribution cabinet by arranging multi-point temperature and humidity sensors on the outer side of a power distribution cabinet body and utilizing a cloud monitoring module to uniformly manage data acquisition and storage. A dynamic weight distribution algorithm is adopted, a comprehensive weight is generated according to the historical abnormal frequency, the position and the environment change rate of the sensor, and a temperature and humidity characteristic matrix is given. The data relevance and weight change are learned through the graph attention network, the global feature vector of the environment state is generated, and the analysis capability of the complex environment data is greatly improved. Meanwhile, in combination with a Gaussian process regression prediction model, the future environment trend is accurately predicted. The alarm score and the alarm level are generated by comparing the deviation interval of the predicted value and the real-time value and combining the parameter change rate, and the real-time alarm information is provided, so that the manual inspection cost is effectively reduced, and the monitoring accuracy is improved.

The invention discloses a disease prevention management and monitoring system design method based on an adaptive algorithm. The method comprises the following steps: S1, obtaining health data of a user; s2, performing feature extraction and fusion on the health data by using a hypergraph neural network; s3, constructing a disease risk prediction model based on the generative reasoning network, and introducing an anti-fact generation module to generate a health change hypothesis scene; s4, adaptively adjusting a personalized intervention strategy through a nested game model, and adjusting the intervention intensity in the first layer; s5, in the second layer, generating an optimized intervention combination strategy based on a Nash equilibrium driven multi-target adaptive game; s6, risk weighted scoring is carried out on the health data through a structured clustering algorithm, and a feedback event is generated; and S7, dynamically optimizing a disease prevention management strategy through a chained feedback mechanism, and managing node permission. According to the method, the self-adaptive algorithm is combined with the multi-layer game and the distributed anomaly detection, so that personalized disease prevention management and monitoring are realized.

The invention relates to the technical field of laser radars, in particular to an unmanned vehicle road obstacle sensing system based on a laser radar, and the system comprises a laser radar data collection module which adjusts the scanning frequency of the laser radar, collects point cloud data from static and dynamic obstacles, and obtains preliminary scanning data; and filtering the preliminary scanning data, and removing noise data. According to the invention, by calculating the dynamic characteristics of the point cloud data, the key point and the moving track of each obstacle are extracted, and the ability of sensing the obstacles in the dynamic environment is optimized. By adopting a time sequence analysis method, the moving state and track of the obstacle can be directly deduced according to the point cloud change, so that the ability of tracking the moving obstacle in a complex dynamic scene is improved. In the spatial reconstruction process, geometric subdivision is utilized to reconstruct point cloud data, and dynamic information is combined with a static point cloud structure, so that the analysis capability of spatial data is enhanced.

The invention relates to the technical field of RFID chip manufacturing, in particular to an RFID chip image anomaly detection method and system based on intelligent learning. The method comprises the following steps: acquiring a real-time RFID chip sampling image set, performing feature alignment based on a registration template, generating an average grey-scale map by using an alignment image set, and obtaining a chip edge mask image set by using an edge search algorithm; segmenting the edge mask image set of the RFID chip to obtain an effective edge mask variance image set of the RFID chip; integrating the variance image set to obtain a pre-generated variance image; acquiring field environment data of the RFID chip and performing sensitivity coefficient estimation to obtain an edge sensitivity coefficient; and performing convolution filtering smoothing processing on the pre-generated variance image to obtain a chip edge variance image, and constructing an RFID chip image anomaly detection model based on the edge variance image and the average grey-scale image. According to the invention, the detection accuracy and stability are improved.

The invention belongs to the technical field of artificial intelligence computing power resource management, and particularly relates to an AI computing power resource monitoring and predicting system and method based on data driving. The acquired computing power resources are sent to an observability analysis module to perform distributed tracking on various application services, and a generated performance flame map is sent to a data integration and output module; meanwhile, protocol analysis is carried out on the obtained computing power resources, full-stack performance indexes are collected to serve as historical computing power resource use data to be sent to the computing power demand prediction module for modeling and prediction, and an obtained computing power resource demand prediction result is sent to the data integration and output module; and the data integration and output module gathers all the data together and injects a uniform label to finally provide a basis for resource scheduling. According to the invention, zero-intruding panoramic monitoring is realized, the coverage range is wide, and the use condition of computing power resources can be mastered more comprehensively by various application services.

The invention discloses a tool calling semantic analysis method and system for a low-resource data environment, and relates to the technical field of semantic analysis, and the method comprises the following steps: constructing a logic expression of each tool based on a context-independent grammar, and generating initial training data in combination with a cue word project and a large language model; performing iterative training on a pre-configured semantic analysis model according to the obtained training data in combination with a self-training method to obtain an optimized semantic analysis model; and performing similarity calculation on the tool candidate data and the training data through a recall model to obtain vector scores of the tool candidate data and the training data, and screening out tools which do not conform to user semantics to obtain a final semantic analysis model. According to the method, the initial analysis data set based on the context-independent grammar is constructed, and the analysis capability of the model is enhanced in combination with the self-training strategy, so that the performance of the model in a low-resource environment is improved, and the semantic analysis capability is effectively expanded.

The invention provides an intention recognition method based on unmanned aerial vehicle cluster three-dimensional trajectory analysis, and relates to the field of group behavior recognition of unmanned aerial vehicle clusters. Through preprocessing the three-dimensional trajectory, the consistency of the trajectory in time and space dimensions is ensured, the stability and quality of data are improved, the subsequent intention recognition process is more accurate and reliable, through the improved HDBSCAN algorithm, the direction information of the trajectory is effectively recognized, and the clustering structure is dynamically adjusted under different density levels, so that the accuracy of intention recognition is improved. And the intention identification of the unmanned aerial vehicle in different group scales and complex task scenes is more accurate. In combination with a deep learning network, the method can better capture the time dependency relationship of a three-dimensional track, improves the analysis capability of a complex behavior mode in an intention recognition process, improves the accuracy and reliability of a recognition result, and improves the recognition efficiency in complex scenes such as multi-unmanned aerial vehicle cooperative combat. More effective identification of the unmanned aerial vehicle cluster intention is realized, and the intention identification performance is greatly improved.

The invention discloses an internet of things-oriented federated learning anti-poisoning attack method. The method comprises the following steps of S1, initializing an internet of things federated learning system; s2, performing mask processing on the privacy data of the Internet of Things participants; s3, the cluster server collects mask data provided by all the Internet of Things participants and aggregates the mask data; and S4, identifying malicious poisoning participants of the Internet of Things through robust clustering. According to the method, an advanced machine learning algorithm and a privacy protection technology are introduced, a more effective anti-poisoning protection measure is provided for the Internet of Things system, the security and robustness of the Internet of Things system are remarkably improved through successful application of the method, and an important guarantee is provided for coping with potential hostile attacks and data tampering.

The invention discloses an unmanned vehicle formation automatic control method and system based on a large language model. The method comprises the steps that an obtained task instruction is analyzed, task semantics are obtained, and the task semantics comprise the number of unmanned vehicles, task targets and font styles; extracting a task formation in the task target, and determining a task level according to the task formation; determining a target external tool based on the task level; acquiring a corresponding control strategy based on the target external tool; and automatically controlling the unmanned vehicle formation according to the control strategy. According to the method, key information such as the number of unmanned vehicles and task targets can be accurately extracted, and a control strategy can be dynamically adjusted according to the complexity and importance of tasks, so that formation control can flexibly adapt to various different types of task requirements; an intelligent strategy generation mechanism ensures the collaboration and efficiency of the unmanned vehicle formation during task execution, and the operation efficiency, reliability and formation accuracy are effectively improved.

The invention relates to the technical field of optical microscopic imaging, and discloses a scanning structured light microscopic imaging system and method based on a multi-focus diffraction-free light beam. Exciting light emitted by a laser light source sequentially passes through an electro-optical modulator and a lens group I and then enters a spatial light modulator; the spatial light modulator is used for synchronously regulating and controlling the multi-focus array and the non-diffraction light beam to generate the multi-focus array with the non-diffraction light beam characteristic; afterwards, the light sequentially passes through a lens group II, a two-dimensional scanning galvanometer, a scanning lens, a tube lens, an emission optical filter, a double-color piece and an objective lens and then irradiates a sample to be imaged; fluorescent light generated by a to-be-imaged sample passes through the objective lens and the double-color sheet and then is transmitted to the area-array camera, and fluorescent light signals at a plurality of focal points are collected to form a fluorescent light image. According to the invention, a diffraction-free optical phase modulation mechanism is introduced, simultaneous imaging of different depth layers is realized, the imaging speed is improved, sample information missed due to limited focusing depth is avoided, and the optical power utilization rate and the imaging quality are greatly improved.

The invention provides an identifier analysis performance improvement method based on a block chain. The method comprises the following steps: A, identifier registration: S1, establishing an object identity; s2, establishing a sub-chain division mechanism; s3, constructing identification metadata and a data template; s4, building an access control mechanism based on the data template and the signature mechanism: 1) constructing a DID document data structure; 2) a DID-based identification registration process and a signature scheme; b, identification analysis: S1, constructing an identification analysis mechanism, and developing an identification analysis interface; according to the invention, on the basis of the object DID identifier region molecular chain, cross-enterprise unified coding is established, and one object has one code. It is ensured that the identification data of the same object are stored on the same sub-chain, the number of times of block chain access in the identification analysis process is effectively reduced, and the identification analysis performance is improved. An identification metadata management mechanism is introduced, an identification data template is built through identification metadata, and data standardization during identification registration is achieved through the data template.

The invention discloses an attention mechanism-guided electric power field ground oil and water stain identification method, which comprises the following steps of: 1, constructing a feature extraction residual block for spatial attention guidance for guiding a module to pay attention to a key feature region; 2, constructing a multi-scale feature extractor according to the feature extraction residual block, wherein the multi-scale feature extractor is used for obtaining rich semantic features from the input image; 3, constructing a feature fusion module for channel attention guidance, wherein the feature fusion module is used for fusing features from a coder and a decoder; 4, constructing a model decoder according to the feature fusion module; step 5, training the established detection model, storing model parameters and generating a final oil-water stain area detection result; according to the method, the problems that in the electric power field ground oil and water stain identification process, the key area attention capability is weak, and the fusion feature key channel information identification capability is insufficient are solved.

The invention relates to the technical field of government affair services, in particular to an NLP entity recognition method based on semantic extension, which comprises the following steps: collecting multi-source time sequence data and generating an entrepreneurship ecological data graph through a dynamic time warping algorithm; dynamic evolution of a term library is realized by using a generative adversarial network, and a term semantic association graph is constructed in combination with a graph convolutional network; a double-path attention mechanism is deployed in a Transform encoder, and global semantics and local attention guided by terms are fused through a gating loop unit; modeling fuzzy query analysis into a Markov decision process, and optimizing a semantic extension strategy by adopting a reinforcement learning agent; multi-granularity features are extracted through a feature pyramid network, and model parameters are optimized in combination with a bidirectional long-short-term memory network and an online learning feedback loop; and finally, calculating entity service association strength by using a graph attention network, and generating an interpretable report based on a Shapley value attribution algorithm. A closed-loop learning system is formed, and the business starting policy matching accuracy is improved.

The invention discloses a deep learning algorithm for predicting a monomer and compound structure of cyclic peptide and linear peptide containing non-standard amino acid. The deep learning algorithm specifically comprises the following steps: constructing a training data set containing head and tail cyclopeptide monomers and compounds, disulfide bond cyclopeptide monomers and compounds, and linear peptide monomers and compounds of non-standard amino acid; training and optimizing the peptide structure prediction model containing the basic protein knowledge; constructing a test data set of a linear peptide monomer containing non-standard amino acid, a linear peptide compound, a head-tail cyclopeptide monomer and a disulfide bond cyclopeptide monomer; and testing on the test data set by using the trained peptide structure prediction model. Compared with an existing peptide structure prediction model, high-precision structure prediction of monomers and compounds of cyclic peptide and linear peptide containing non-standard amino acid is proposed for the first time, structural information or pocket information of any target protein is not needed, and only one-dimensional sequences of the protein and side chain groups of the non-standard amino acid need to be input, so that the non-standard amino acid containing cyclic peptide and linear peptide is obtained. And the three-dimensional structure of the polypeptide can be accurately predicted.

The invention relates to the technical field of multi-signal processing, in particular to a multi-dimensional electrical signal processing method and system, and the method comprises the following steps: carrying out the synchronous collection of time, frequency and spatial features of electrical signals of differentiated target sources, and generating a preliminary time-frequency-space feature set through the measurement and comparison of features and the screening of key time-frequency-space data. According to the method, the information extraction efficiency is greatly improved in a complex signal environment through hierarchical analysis in combination with feature correlation revelation and optimization of the contrast and definition of data, voltage fluctuation and frequency anomaly can be rapidly identified based on matching and classification of known abnormal modes, efficient classification and processing of abnormal signals are achieved, and the method is suitable for popularization and application. The real-time performance and precision of signal monitoring and the dynamic adjustment and optimization processing of the signals are enhanced, the reliability and adaptability of the whole signal processing process are improved, challenges in a complex signal environment are effectively handled, the analysis capability and processing efficiency of the signals are remarkably improved, and the signal quality in practical application is enhanced.

The invention provides a collaborative development process data synchronization and notification method. According to the method, an automatic process is triggered by monitoring a test-submitted event message queue, development data is positioned based on module identification, key operation features (such as database change and task start and stop) are analyzed and extracted by using natural language, and feature extraction precision is improved by combining a dynamic keyword library and cross-modal fusion; the timeliness and reliability of risk reminding are ensured through template message generation and multi-channel delivery; a risk propagation range is further predicted by means of a dependency graph model, and a closed-loop early warning mechanism is formed. According to the method, the problem of delay caused by information lagging and insufficient collaboration in the development process is solved, the risk of manual operation is reduced, and the efficiency and controllability of project management are improved.

The application discloses a polarization mode dispersion measurement method and device, equipment, storage medium and product thereof, and relates to the technical field of polarization measurement. The method comprises the following steps: acquiring interference spectrum data of a to-be-measured optical fiber based on a fixed analyzer; performing dimension reduction processing on the interference spectrum data to obtain initial feature data containing a feature vector of the to-be-measured optical fiber, and determining a credibility score representing spectral physical consistency based on the interference spectrum data; adaptively reweighting the initial feature data based on the credibility score to generate spectral feature data; inputting the spectral feature data into a preset back propagation neural network to obtain a polarization mode dispersion value of the to-be-measured optical fiber, wherein the back propagation neural network is obtained based on a nonlinear mapping relationship between the spectral feature and the polarization mode dispersion value. While retaining the real physical signal features, the influence of the pseudo structure caused by the link noise is automatically weakened, so that the measurement precision of the polarization mode dispersion of the optical fiber is improved.

The invention relates to the technical field of historical building structure health monitoring, and discloses a historical building safety performance evaluation method and system. The method comprises the steps of obtaining a structure response signal through monitoring of a multi-position sensor, and generating a fusion feature map through a multi-scale feature sensing network. And performing three-dimensional reconstruction based on the atlas, adaptively dividing difference precision entity units according to feature density, and constructing the high-fidelity digital twinborn body. And driving the twin to perform time-history evolution under simulation of long-term environment excitation, and extracting damage evolution behavior fingerprints. And performing multi-round iterative comparison and reverse traceability analysis on the behavior fingerprint and a safety baseline containing material time-varying degradation and component interaction influence, identifying a failure mode and an evolution path, calculating a residual bearing capacity margin of a key component, and generating a structural life map in combination with a component interaction network. According to the method, high-fidelity modeling based on monitoring data and reverse intelligent diagnosis of dynamic damage evolution are realized.

The application relates to a self-adaptive self-powered flexible sensing device based on triboelectricity and a preparation method thereof, which comprises a flexible device, a charging module and a single-chip circuit board; the flexible device comprises, from top to bottom, a triboelectric layer, an insulating layer, a potential layer and a battery layer; the triboelectric layer is connected with the charging module, the charging module is connected with the battery layer, and the battery layer is connected with the single-chip circuit board. The system collects signals of multiple electrode ports through a single-chip microcomputer and transmits the signals to a mobile phone or a computer terminal through Bluetooth wireless transmission, so that the convenience and expansibility of data acquisition are improved, and real-time monitoring and remote interaction application can be constructed.

The invention relates to the technical field of biology, in particular to a DNA tag library construction method based on droplet sorting and a magnetic microsphere technology, according to the method, the concentration of DNA tag template sequences is adjusted through Poisson distribution, it is ensured that each droplet contains at most one DNA tag template sequence, the problem that multiple templates are mixed into the droplets can be solved, and the DNA tag library construction efficiency is improved. The method has the advantages that the method is simple and convenient to operate, label repetition and collision are avoided, uniqueness of DNA labels in liquid drops is guaranteed, DNA label sequences after PCR amplification are further captured through magnetic microspheres, it is guaranteed that the amplified DNA labels can be efficiently recycled and purified, interference of non-specific reaction in PCR amplification is avoided, and the method is suitable for large-scale popularization and application. The combination of the technical points ensures the application advantages of the technical scheme in high-throughput sequencing, including diversity, accuracy and high efficiency of tags.

The invention belongs to the technical field of intelligent prediction, and relates to a key gene identification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining original sample data; preprocessing the original sample data to obtain preprocessed sample data; performing word segmentation operation on the preprocessed sample data according to the HFkme r to obtain word segmentation data; performing vector conversion operation on the word segmentation data to obtain word segmentation vector data; constructing an initial BGAC model, and performing model training operation on the initial BGAC model according to the word segmentation vector data to obtain a target BGAC model; obtaining a to-be-recognized genome protein sequence; and inputting the genome protein sequence into the target BGAC model for key gene recognition operation to obtain a key gene recognition result. The method not only solves the problems of high time and high cost in key gene identification in traditional molecular biology, but also provides a precious microscopic view angle for deeply exploring the origin and evolution of insect wings, and the complex biological macro-evolution process is better understood.

The present application relates to the technical field of data analysis, in particular to a data analysis processing method and system for processing a chromatographic system, comprising the following steps: collecting a chromatographic matrix, calculating curvature variation, constructing a segmented function, mapping to a manifold space, screening a drift retention trunk, calculating a difference and a variation rate, cutting a high variation area, constructing a three-dimensional tensor, performing sparse decomposition to obtain an independent peak component, locally fitting to extract a peak shape moment feature, reconstructing a peak shape according to a fitting weight to generate an identification identifier, in the present application, a manifold space is constructed based on curvature mutation point division, a signal trunk is identified using a gradient variation rate and drift noise is screened out, a high gradient section of a signal response surface is constructed in combination with a retention time difference and a variation rate, potential overlapping peak areas in a complex structure are effectively located, a local subsection is fitted by a least square method and a peak shape moment feature is extracted, weighted reconstruction is performed in combination with global information, and the refinement expression of peak shape information and the enhancement of recognition accuracy are realized.

The application belongs to but is not limited to the technical field of intelligent education, education data mining and artificial intelligence, and discloses a cognitive dynamic tracking method and system based on asymmetric hierarchical perception and ordinal constraint, which carries out structured mapping of a target level of a learning task, a knowledge type and a question concept set, constructs a multi-dimensional heterogeneous feature group containing sequence constraint education semantics and behavior characteristics, and fuses basic interaction features and education classification method features through a gating mechanism to generate a deep interaction feature sequence; and based on a directed cognitive level difference, an asymmetric hierarchical bias matrix is constructed, which is introduced into a multi-head self-attention mechanism to realize adaptive dynamic perception and updating of a learner's cognitive level; further, the knowledge state and the cognitive level are continuously evolved and tracked in the whole learning process, and an interpretable cognitive level evolution track, a phased transition point and a future answer prediction result are output, which can be used for ability diagnosis, personalized recommendation and learning path optimization.

The application relates to a light source detection model processing method and device, a readable storage medium and a computer program product. The method comprises the following steps: extracting a sample face feature map from a sample face image, the sample face image being obtained by rendering sample illumination spherical harmonic coefficients; performing illumination prediction based on the sample face feature map to obtain first illumination spherical harmonic coefficients; obtaining reference illumination spherical harmonic coefficients, performing image reconstruction based on the sample face feature map and the reference illumination spherical harmonic coefficients to obtain a reconstructed face image; performing illumination prediction on the reconstructed face image to obtain second illumination spherical harmonic coefficients; determining a spherical harmonic coefficient prediction loss between the first illumination spherical harmonic coefficients and the sample illumination spherical harmonic coefficients, and determining a spherical harmonic coefficient discrimination loss between the second illumination spherical harmonic coefficients and the reference illumination spherical harmonic coefficients; and performing model training based on the spherical harmonic coefficient prediction loss and the spherical harmonic coefficient discrimination loss to obtain a light source detection model. The method can improve the accuracy of light source detection of an image.

The utility model discloses a high-temperature desorption oven for producing a solid lithium battery, which comprises oven equipment, a plurality of drying chambers arranged in the oven equipment, gas guide boxes arranged in the plurality of drying chambers, element desorption boxes arranged in the gas guide boxes, nitrogen introduction pipes arranged on the element desorption boxes and gas outlets arranged on the element desorption boxes. A temperature sensor is arranged at the bottom of the element analysis box; an operation interface, a flow switch and an oxygen analyzer are arranged at one end of the oven equipment; the temperature control device comprises a fan, a heater and a cooling pipe, the heater is located at the rear end of the fan, and the cooling pipe is located at the front end of the fan. According to the utility model, the oven equipment is internally provided with the plurality of mutually insulated drying chambers, the gas guide boxes are arranged in the drying chambers, and the element analysis box is placed in the gas guide boxes, so that not only is the temperature diffusion area reduced, but also the temperature is uniformly transmitted to the element analysis box by utilizing the plurality of ventilation holes arranged in the matrix on the gas guide boxes; and the temperature uniformity is improved.

The present application relates to the technical field of smart city, especially relates to a smart city bus stop board dynamic information accurate pushing method and system, wherein a query semantic representation structure is constructed by collecting query requests and scene state information, a joint understanding representation is obtained by modeling the scene state information multi-level semantic association, scene constraint multi-hop reasoning is executed in a bus knowledge association network, a candidate answer sequence is generated by filtering, an optimal answer is selected by calculating a scene adaptation score, pushing content is generated, interactive feedback is collected, feature fusion weights are adaptively adjusted to form a closed loop optimization, and accurate dynamic pushing of bus information and user experience improvement are realized.