51results about How to "Taking into account accuracy" patented technology

The invention relates to the technical field of Internet of Things communication, discloses an Internet of Things multi-Android client state synchronization method, and solves the problems of poor state consistency, high back-end dependence, large delay and insufficient compatibility in the prior art. According to the scheme, the method comprises the steps that after a client completes initialization, MQTT connection with a rear end is established, an MQTT state message pushed by the rear end is monitored, after the message is received, equipment state information is obtained through analysis, local state cache and a UI interface are updated, meanwhile, the client serves as a distribution node, a UDP synchronization message containing the equipment state and the IP of the distribution node is established, and the UDP synchronization message is sent to the client. The UDP synchronization message is broadcasted to the local area network, other clients in the local area network receive the UDP synchronization message and then verify whether the distribution node IP is the client in the local area network, if the verification is passed, the message is analyzed, the state updating timestamp and the local cache timestamp are compared, and the local state cache and the UI interface are updated only when the state updating timestamp is larger.

This invention discloses a system and method for labeling automotive abnormal noise data, relating to the field of automotive NVH detection technology. The method includes: receiving raw signal data of automotive abnormal noise; performing time-frequency transformation on the raw signal to generate a time-frequency diagram; identifying key frequency ranges characterizing the abnormal noise features; then performing directional filtering to enhance the abnormal noise signal; identifying candidate time ranges for abnormal noise through an automatic detection algorithm; and outputting the final abnormal noise time range labels and key frequency ranges in a structured format after verification and correction. The system includes a data receiving module, a signal processing module, a display module, an intelligent labeling module, an audio playback module, a labeling output module, and an interaction module, achieving triple-assisted labeling through auditory, visual, and intelligent pre-selection methods. This solves the problems of low efficiency, low accuracy, and poor consistency in traditional manual listening methods, providing high-quality labeled data for deep learning classification of automotive abnormal noises.

This application discloses a question response method and related equipment, which balances the response speed and accuracy of question responses. The method includes: extracting the target question from the interaction request and inputting the target question into a top-level big model; the top-level big model then performs the following processing on the target question: determining the matching degree between the target question and entries in the target general knowledge base, and / or determining the scenario hit rate of the target question in the target standard business operation process knowledge base; if the matching degree reaches a preset threshold, and / or the scenario hit rate is not hit, then a response is provided to the target question based on the target general knowledge base; if the matching degree does not reach the preset threshold, and / or the scenario hit rate is hit, then a single underlying intelligent agent is triggered to match the corresponding target standard business operation process from the target standard business operation process knowledge base based on the target scenario identifier, and then a response is provided to the target question based on the target standard business operation process.

The application discloses an intrinsic voltage estimation and parameter self-setting method and system for low-voltage governance, and is applied to the technical field of smart grids, and the method comprises the following steps: acquiring electrical quantity data of a target governance device; correcting the current line equivalent resistance and line equivalent reactance according to the electrical quantity data; calculating the intrinsic power grid voltage according to the electrical quantity data, the corrected line equivalent resistance and the corrected line equivalent reactance; generating the proportional gain of a closed-loop controller through a preset inverse proportional relationship of the corrected line equivalent resistance, generating the integral gain and the differential gain according to the proportional gain according to a preset proportion; and updating the control parameters through the calculated proportional gain, integral gain and differential gain. According to the application, the line impedance does not need to be manually input, the line characteristics are self-learned through online active disturbance and passive observation, the deployment conditions of different transformer areas, different lengths and different loads are adapted, and the deployment cost and maintenance cost are significantly reduced.

The application provides a first-aid hemostasis robot and a control method thereof. The first-aid hemostasis robot comprises a robot main body capable of walking on the ground, a mechanical arm with multiple degrees of freedom arranged on the robot main body, a pressing head and a blocking piece arranged on the mechanical arm, the pressing head is used for pressing and hemostasis of a blood vessel of a bleeding port, the blocking piece is carried by a flexible continuum robot, a front end of the continuum robot is provided with a camera used for collecting information of a bleeding site, the blocking piece can be inserted into the inside of the bleeding port through the continuum robot and expanded after blocking the bleeding port to seal a large blood vessel of the bleeding port. The mechanical arm carries the pressing head and the blocking piece to perform hemostasis on different conditions of large bleeding respectively, if the bleeding is of a split type, the pressing head is used for pressing and hemostasis at a proximal end of an artery bleeding or a distal end of a vein bleeding, if the bleeding is of a broken type, the blocking piece is inserted into a wound through the continuum robot to seal a large blood vessel of the bleeding.

This application discloses a character recognition method. The method includes: performing text detection on a target image to obtain at least one polygonal region; determining a target polygonal region from the at least one polygonal region, wherein the target polygonal region has an irregular shape; determining four target vertices from multiple vertices of the target polygonal region, and using these four target vertices as the four vertices of a rectangular region corresponding to the target polygonal region; mapping pixel information of each pixel in the target polygonal region to the rectangular region based on the positions of each vertex of the rectangular region in the target image to generate pixel information of each pixel in the rectangular region; and performing OCR processing on the rectangular region and the polygonal regions in the at least one polygonal region other than the target polygonal region to obtain the character recognition result of the target image. This application embodiment can simultaneously achieve both accuracy and efficiency in character recognition.

The embodiment of the application discloses a kind of multi-party collaborative automatic driving traffic responsibility determination method and system, wherein the method comprises: real-time acquisition of the original data of each data source;Each original data is calculated by hash digest;After the accident, the relevant original data of the data source involved in the accident within the accident time period is obtained, the actual hash digest is calculated and compared with the hash digest for integrity check;Identity verification is carried out according to identity certificate;According to the ciphertext accident characteristics of the relevant original data, a space-time consistency graph is constructed, the deviation data of each data source and the abnormal data source mark are calculated;According to the equipment state data, the identity verification result, the integrity check result and the deviation data, the trusted weight of the corresponding data source is calculated;According to the ciphertext accident characteristics and the trusted weight, the responsibility proportion of each responsible party is obtained;According to the responsibility proportion, the trusted weight, the space-time consistency graph and the abnormal data source mark, a responsibility determination report is generated and is chained to store evidence. The responsibility determination efficiency and the reliability are improved.

This invention provides a method and system for assembling a genome stack, a method and system for assembling a genome architecture, and a method and system for assembling a genome sequence, belonging to the field of bioinformatics. This invention optimizes a global loss function using robust regression techniques to determine the stacking relationships between sequencing sequences, eliminating interference from false positive alignments, thereby obtaining a more accurate assembled stack set and avoiding copy number loss or misassembly in the assembled genome. The estimation of stack arrangement based on regression matrix calculations has a higher tolerance for outliers; unless the number of ambiguous sample information exceeds the number of true samples, a "collapse" will not occur. By using resampling techniques to independently sample, assemble, improve, and evaluate sequencing data multiple times, and then integrating them into the final assembled genome, the uncertainty of the assembly results caused by noise in the sequencing data can be reduced, and the sensitivity of the assembly results to the selection of assembly parameters can be decreased.

The application discloses a kind of image shadow removal methods of balancing deterministic recovery and perceived quality, comprising:1, first stage global regression analysis is carried out to shadow image, global illumination and color consistency of image are repaired using global regression network based on no residual, and deterministic illumination anchor point is established;2, the output of global regression network is input into the diffusion model of second stage as conditional space prior, and local perceptual refinement is carried out using conditional diffusion model based on block, and fine texture details are recovered by generative prior;3, in the reverse denoising sampling process of second stage, introduce random block overlap mechanism, calculate the mean of all covering block prediction noise as updating basis for the purpose of eliminating the boundary artifact when full resolution inference.The application can significantly improve the visual fidelity of the de-shadowed image while ensuring pixel-level accuracy, and can efficiently process high-resolution images under limited video memory constraints.

Embodiments of the present specification disclose an application processing method and device, a storage medium and an electronic device, wherein the method comprises: obtaining a target file hash value of a target application file for first-level deduplication detection processing to obtain a first deduplication detection result; if the target application file is of the same type, performing deduplication filtering processing; if the target application file is of different types, determining file metadata features of the target application file, performing multi-level deduplication detection processing based on the file metadata features to obtain a multi-level deduplication detection result, and executing an application file detection process on the target application file based on the multi-level deduplication detection result.

The invention provides a multi-magnetic-source initial positioning method, device and system based on NSS-VMGT fusion, and solves the problems of misjudgment of the number of magnetic sources, missing detection of edge magnets and low actuarial efficiency in the prior art. The method comprises the following steps of: acquiring 7 * 7 three-component sensor array data by an information processing module, and preprocessing by moving average filtering; deriving a magnetic field space change rate through an MGT module, and solving a related angle and an X / Y direction gradient in combination with a VMGT module; the NSS module detects a magnetic gradient tensor feature value and eliminates a coupling pseudo peak, the candidate point recognition module extracts internal candidate points and delimits an initial search area, and the candidate point fusion module detects edge candidate points and performs merging and duplicate removal. The device is based on an STM32H743 hardware platform and comprises a 7 * 7 array positioning plate, a double-MCU regional acquisition framework, a serial port data interaction link and a hardware signal filtering and voltage stabilizing circuit. According to the design, the recognition accuracy of the number of the magnetic sources is improved to 94%, and the scene requirements of multi-magnetic capsule targeted therapy and the like are met.

The invention discloses a mathematical question and answer method and device based on a chat big model and a computer program product, and the method comprises the steps: firstly inputting a sample mathematical question text into an initial chat big model to obtain an initial answer, then carrying out the named entity recognition of the initial answer through a pre-trained mathematical answer named entity recognition model, and carrying out the named entity recognition of the initial answer; obtaining an entity recognition result; inputting the entity recognition result and a reference answer corresponding to a pre-acquired sample mathematical question text into a pre-trained mathematical answer matching model to obtain a matching result, calculating a value of a GRPO loss function by using the matching result, and stopping updating the parameters of the initial chat large model until the value meets a preset condition; obtaining a trained smooth chat large model; therefore, when the trained chat model is used for answering the target mathematical question text proposed by the user, the accuracy and anthropomorphism of the answering result can be effectively considered, and the mathematical question and answer experience of the user is improved.

The application discloses a multi-target dynamic marking method and system of multi-dimensional trajectory fingerprint and intelligent arbitration, relates to the technical field of radar signal processing and target identification, and comprises the following steps: acquiring radar original trajectory data and preprocessing to obtain a multi-dimensional motion feature vector; inputting the multi-dimensional motion feature vector into a local sensitive hash module, performing trajectory fingerprint matching by using preset composite hash functions in L independent hash tables, and obtaining an association ID; monitoring distances between all targets in real time, and when the distance between two targets is less than a trajectory conflict determination distance threshold, determining whether the trajectories are independent based on trajectory decoupling verification, if yes, maintaining the association ID, and if not, starting a conflict arbitration mechanism; and decoupling the conflict targets by adopting a progressive three-level arbitration strategy, so as to maintain the association ID; the method improves the distinguishing ability of similar trajectories, takes into account real-time performance and accuracy, and reduces the frequency of manual intervention.

The application provides a traffic pavement crack dehazing and de-noising segmentation method based on FA-Diff-UNet, introduces a FA-Diff module in the encoding front end, fuses diffusion amplitude residual reconstruction and phase fine-tuning, and internally builds a super-resolution reconstruction head to recover high-frequency details suppressed by fog and noise without introducing pseudo texture; a double-branch down-sampling structure of a CDM module and a DBAT module is used in the encoder part to expand the effective receptive field and retain the characteristics of small cracks under the condition that the parameters are controllable; a DHLA module is introduced in the bottleneck layer to decouple and suppress noise for high-frequency cracks and low-frequency backgrounds; the decoder replaces the simple skip connection of the UNet module with a BFM module to realize bidirectional gating fusion and redundancy suppression of semantics and details, and a DBAT refinement block is connected in series after each level of up-sampling to further enhance cross-block connectivity and direction consistency. The crack edge definition and connectivity are significantly improved, the recall of small cracks is improved, the false detection is reduced, the accuracy and the calculation efficiency are considered, and the method is suitable for actual road scenes such as complex weather and weak light.

This invention discloses a cable production identification and monitoring system and method based on cable images, belonging to the field of image recognition technology. The system includes: acquiring a sequence of cable surface depth images to construct a grayscale surface map along the cable surface depth direction; using the central ridge of the grayscale surface map as a dynamic axis of symmetry, detecting candidate pixels through multi-scale morphological top-hat transformation to form a symmetrical residual distribution map; combining the eight-neighbor gradient direction field and Hessian matrix curvature to remove false defect points, obtaining an effective image, and inputting it into a pre-trained multi-task neural network to simultaneously output a defect segmentation mask and category label; generating a three-dimensional mesh model with dimensional accuracy through Poisson surface reconstruction; and finally, using the physical coordinates of the defect center to drive a laser marking machine to mark a QR code containing key information, which is then associated and stored in a blockchain database. This application improves the accuracy and efficiency of defect identification, realizing the quantitative characterization and full lifecycle management of defects.

The invention belongs to the technical field of image detection, and discloses a packaging box printing quality detection method and system based on image analysis, and the method comprises the steps: obtaining an initial suspected defect region based on the alignment and comparison of a real shot image and a standard template image, and extracting the position and direction information of a physical crease in combination with cutting die file data to define a folding region; local feature displacement parameters are calculated for the suspected defect area which is spatially overlapped with the folding area, and the area with spatial displacement is identified; performing orthogonal decomposition on displacement parameters by utilizing the crease direction, extracting parallel and vertical displacement components, and judging that the difference attribute is printing defect or deformation deviation according to the numerical constraint relation of the two displacement components; and when the deformation deviation exists, executing one-way flattening recheck to generate a detection result. According to the method, classification judgment of physical deformation and printing defects is achieved through orthogonal decomposition, the defects covered by deformation can be detected through flattening and rechecking, false detection is restrained, and meanwhile missing detection is avoided.

The invention discloses a plate strip steel surface defect detection method based on partial multi-label learning, and belongs to the technical field of industrial quality detection and data mining. In order to solve the problem that in the prior art, due to the fact that labeling noise is difficult to eliminate, model robustness is insufficient, a semantic alignment mechanism is introduced, and collaborative modeling is carried out on a sample feature space and a label semantic space. The process comprises the following steps: in a data preparation stage, extracting plate strip steel image features and constructing candidate tags; in the label denoising stage, a partial multi-label learning framework is adopted, label false correlation is eliminated through orthogonal rotation, label reliability is improved through joint projection, a label relation is reconstructed through manifold learning, and finally a denoised discrimination label is obtained. In the classifier training stage, a depth perception classifier is trained by using a discriminant label and an original label; and in the detection stage, defects such as cracks and scratches are identified. According to the method, the detection precision and the anti-interference capability are improved, the dependence on manual labeling is reduced, and reliable support is provided for plate and strip steel quality control.

The invention discloses a detection unmanned vehicle target identification method based on computer vision, and relates to the technical field of artificial intelligence. During operation of the system, multi-source heterogeneous vision sensors such as wide-angle visible light, narrow field of view, high resolution and infrared thermal imaging are cooperatively deployed; a lightweight adaptive feature extraction network is introduced, a network structure can flexibly adjust the calculation scale according to the texture complexity of an input image through channel pruning and a dynamic sparse activation mechanism, and the calculation load is effectively reduced while the feature expression capability is ensured; fusing time sequence information through a bidirectional gating circulation unit to generate a target candidate area with motion prior; carrying out weighted fusion on the multi-modal features in the candidate region, and then classifying the multi-modal features; and an online domain adaptive correction mechanism is introduced, and network parameters are finely adjusted in an unsupervised manner according to environmental changes. In addition, occlusion recovery, resource awareness scheduling and offline-online collaborative learning strategies are integrated.

The application provides a multi-modal large model task-driven target detection method and system based on meta learning, relates to the technical field of target detection, and comprises the following steps: obtaining multi-modal input data and performing preprocessing to obtain unified semantic representation; a general candidate generation network and a large language model are used to generate a candidate region set and a visual attribute word group set; region feature extraction is performed on each candidate target frame in the candidate region set to obtain a candidate visual embedding sequence; the visual attribute word group set is subjected to vectorization processing to obtain an attribute text embedding sequence; the candidate visual embedding sequence and the attribute text embedding sequence are jointly calibrated to obtain a calibrated candidate visual embedding sequence and a calibrated attribute text embedding sequence; a task-related score is calculated through a trainable scoring function, and all candidate frames in the candidate region set are screened to obtain a target detection result. The application realizes target detection guided by task description and has the ability of cross-task rapid adaptation.

This invention relates to the fields of medical image processing and artificial intelligence-assisted diagnosis, specifically to a data-augmented fundus image recognition lesion segmentation and analysis system. The system includes: an image acquisition module for acquiring the target fundus image to be analyzed and extracting its texture distribution features; a drift measurement module for calling a preset synthetic training domain feature distribution and generating a feature domain offset index from the homology distance; a feature decoupling module for dynamically adjusting the activation weights of the enhancement-sensitive feature channel and the topological feature channel in the segmentation network based on the feature domain offset index, generating an adaptive feature tensor; and a lesion segmentation module for performing pixel-level classification of the target fundus image based on the adaptive feature tensor, outputting a lesion segmentation mask that conforms to anatomical constraints. This invention effectively solves the problems of model aging and environmental adaptability during long-term operation.

This invention relates to the field of ship cleaning technology, and more particularly to an integrated speedboat cleaning and self-floating device. The support frame of this invention has a U-shaped frame structure; a pontoon is installed on each of its lower two sides, bearing the weight of the equipment and providing buoyancy when floating; an inwardly inclined support frame is installed on each of the lower two sides of the support frame to support the bottom of the speedboat; at least three vertical electric actuators are evenly distributed on each of the upper two sides of the support frame; the electric actuators are connected to an electrical box on the shore; a horizontal rail is installed above each of the electric actuators on both sides; multiple reciprocating brush plate mechanisms are installed in the middle of the circular rail; the two ends of the circular rail are slidably connected to the horizontal rails on both sides, allowing it to slide back and forth along the horizontal rails; fixed brush plate mechanisms are installed on the horizontal rails. The technical solution of this invention solves the problems of low efficiency and high labor consumption in manual cleaning, difficulty in guaranteeing the cleaning effect of machine cleaning, and the limited cleaning range of robot cleaning in a single operation in the prior art.

This invention discloses a hierarchical multi-label attribution method and system that integrates atomic rule-driven credible features and knowledge distillation, belonging to the field of natural language processing technology. First, this invention constructs an atomic rule base for weakly supervised text annotation. Then, it uses a large language model as a teacher model to correct and supplement the weak annotation results, extracting the probability distribution of soft labels and intermediate layer feature representations on each level of labels. Next, it evaluates the credibility of the teacher model's output, selecting a subset of credible soft labels and credible feature dimensions. Then, it constructs a student model with a hierarchical output structure, designs a joint loss function, and distills the student model for training. Finally, it deploys only the student model for inference, outputting hierarchical multi-label attribution results and key evidence fragments. This invention, through the combination of atomic rules and credible knowledge distillation, significantly reduces inference costs while improving the accuracy, stability, and interpretability of hierarchical multi-label attribution.

The application discloses a laser radar and train body calibration method and device, relates to the technical field of train radar external parameter calibration, and comprises the following steps: determining a calibration track according to train straight track data, controlling a calibration device to move along the calibration track according to a preset calibration speed and a preset calibration section, and acquiring laser radar collection data; based on the position constraint of the calibration device on the calibration track, the constraint coordinates of the calibration area of the calibration device in the train coordinate system are determined; the radar coordinates are transformed by using a conversion matrix of the radar coordinate system and the train coordinate system, and estimated coordinates are acquired; taking the minimum difference between the constraint coordinates and the estimated coordinates as an optimization target, an optimal conversion matrix is solved, and the calibration result of the laser radar and the train body is output by using the optimal conversion matrix. The application has the beneficial effects that the existing straight track in rail transit is used as a calibration field, and the convenience and accuracy are considered.

The invention relates to a piping shaft overflow detection method and device, and belongs to the field of urban public safety monitoring. Collecting images covering well lid and water body characteristics; preprocessing and labeling; a YOLO26s target detection model is adopted for recognition training, after a real-time video stream is framed according to the frequency f, obtained images are sequentially input into the trained YOLO26s model, detection results of various features in each frame are output, space-time joint verification is carried out through the feature overlapping degree and multi-frame statistics, and vehicle splashing and real overflow are distinguished; and setting a sliding frame number window, counting the number of frames appearing in each feature in the window, and judging a triggering early warning level according to a priority sequence. According to the invention, the false alarm rate of overflow detection is greatly reduced; and multi-stage early warning in the whole process from occurrence to explosion can be realized, so that the early warning speed is increased.

The present application relates to the technical field of data processing, and provides a label level evolution document retrieval method, device, equipment and medium, the method comprises: based on the mixed architecture of regular matching and large language model, extracting candidate labels from text content and verifying and cleaning to obtain compliance labels; according to the hierarchical structure of the label dictionary, performing bidirectional hierarchical expansion on the compliance labels to obtain a complete label set and binding storage; extracting query labels from the retrieval request and performing bidirectional hierarchical expansion to obtain a retrieval label set, and deducing a retrieval opposite label set semantically exclusive to the retrieval label set; based on the retrieval label set and the retrieval opposite label set, performing double-mode filtering retrieval, and outputting the retrieval result. The present application makes the label labeling process of the ship field document get rid of the limitation of manual operation, the label system can be continuously improved, the document retrieval can be based on the forward matching and the reverse exclusion double logic at the same time, and the fullness rate and the precision rate of the ship document retrieval are realized.

The application discloses an intrinsic voltage estimation and parameter self-setting method and system for low-voltage governance, and is applied to the technical field of smart grids, and the method comprises the following steps: acquiring electrical quantity data of a target governance device; correcting the current line equivalent resistance and line equivalent reactance according to the electrical quantity data; calculating the intrinsic power grid voltage according to the electrical quantity data, the corrected line equivalent resistance and the corrected line equivalent reactance; generating the proportional gain of a closed-loop controller through a preset inverse proportional relationship of the corrected line equivalent resistance, generating the integral gain and the differential gain according to the proportional gain according to a preset proportion; and updating the control parameters through the calculated proportional gain, integral gain and differential gain. According to the application, the line impedance does not need to be manually input, the line characteristics are self-learned through online active disturbance and passive observation, the deployment conditions of different transformer areas, different lengths and different loads are adapted, and the deployment cost and maintenance cost are significantly reduced.

This application discloses a method, device, system, and electronic device for intelligent meeting minutes and focus retracing based on physical space negative entropy. The method includes: acquiring multi-source non-semantic physical signals from the meeting, including at least voice activity status, desktop vibration status, and screen sharing dwell status, with the option to add meeting space device trigger status; performing timestamp alignment and dynamic baseline modeling on the multi-source signals, calculating the abnormal changes of each signal relative to historical windows; normalizing multi-path features, and fusing them with weight coefficients and confidence coefficients to obtain a meeting space information density index; generating a meeting value heatmap based on the index, dividing it into low-value discussion areas, high-value decision areas, and meeting restart areas; and implementing differentiated transcription, summary generation, to-do item extraction, and playback bookmark creation for different partitions, and associating and storing the data. This application can identify key decision windows in the meeting, improving the accuracy of meeting minutes focus positioning.

The application discloses a bid qualification performance automatic reuse method based on a state space model and relates to the technical field of artificial intelligence, including the following steps: step 1: performing text analysis on a bid document, performing entity boundary marking on a qualification performance reference paragraph to obtain a qualification performance entity set; step 2: performing associated assembly on the entities based on the geodesic distance between hyperbolic embedding vectors in the same paragraph to obtain a qualification performance associated group; step 3: calculating a non-consistency score for each qualification performance associated group and determining a calibration critical value, and automatically filling the entity information in the reusable associated group into a qualification performance reporting form of this bid. The application realizes long document unified modeling, hierarchical perception entity association matching and automatic reuse with statistical reliability guarantee, and effectively improves the efficiency and accuracy of bid qualification performance reporting.

The application discloses an active thermal management method for a motor driver of an aircraft. The method adopts temperature-sensitive electrical parameters to construct a thermal network and comprehensively considers measured temperature of a temperature sensor, and performs active thermal management control from three aspects of driving parameter, modulation strategy and working cycle to inhibit overall temperature fluctuation of the driver, balance temperature of the same bridge arm and balance temperature of the same phase and different bridge arms. The application has the beneficial effects that: the working state of the driving device is actively and overall adjusted, the temperature of the aircraft driving unit is balanced and stabilized, and the operation and maintenance cost of the driver is reduced; the real-time and accuracy of temperature collection are considered, the method has strong practicability and wide application prospect; and the method has unique advantages in the fields of inhibiting temperature fluctuation of the power device and prolonging the service life of the driver.

The invention discloses a multi-device concurrent emotion screening intelligent chip and terminal, and relates to the technical field of emotion screening, the multi-device concurrent emotion screening intelligent chip comprises a physical sign parameter acquisition module used for acquiring multi-source physical sign parameters; the data validity analysis module is used for performing consistency analysis on the multi-source physical sign parameters to obtain data validity parameters; the validity threshold acquisition module is used for acquiring a validity threshold; the acquisition frequency correction module is used for obtaining a corrected acquisition frequency to perform data acquisition to obtain an updated physical sign parameter and perform emotion recognition when the data validity parameter is smaller than a validity threshold value; and the acquisition frequency optimization module is used for performing emotion recognition on the multi-source physical sign parameters when the data validity parameters are greater than or equal to the validity threshold value, optimizing the acquisition frequency, acquiring the efficiency acquisition frequency and acquiring the multi-source physical sign parameters. The technical problem that an emotion screening data acquisition strategy in the prior art cannot adapt to an actual scene is solved.