64results about How to "Accurate perception" patented technology

The invention provides a distributed computing resource smart evolution method and system based on digital twinning, and relates to the technical field of resource evolution.The distributed computing resource smart evolution method comprises the steps that a smart sensing probe is deployed at a computing node to collect operation data, a double-layer attention mechanism is used for extracting features, and a digital twinning model is constructed based on a graph neural network; a resource allocation strategy and a scheduling decision are optimized in combination with a hierarchical reinforcement learning framework, a prediction sequence is generated through an attention-enhanced sequence prediction network, and finally an optimal configuration scheme is solved by adopting a particle swarm algorithm. According to the method, the utilization rate of distributed computing resources can be improved, the energy consumption of the system is reduced, and intelligent scheduling and optimal configuration of the computing resources are realized.

The invention belongs to the technical field of the Internet of Things, and discloses a situation awareness method and system based on the Internet of Things, and the method comprises the steps: collecting environment awareness data, and extracting environment features and mineral features from the environment awareness data; storing the environment features and the mineral features in a block chain, and respectively judging whether the environment features and the mineral features meet preset standards or not; respectively predicting the environment situation and the mineral situation based on the judgment results of the environment characteristics and the mineral characteristics; and comprehensively evaluating the ore exploitability through a pre-established decision cluster on the basis of the prediction results of the environment situation and the mineral product situation. According to the system and the method, through a mechanism of sensing surrounding environment parameters and extracting specified mineral signs and other attributes, the environment parameters and the mineral signs can be accurately sensed and collected, and powerful data support is provided for accurate monitoring and resource management.

The invention belongs to the technical field of building robots, and particularly discloses a self-propelled reinforcing steel bar binding robot moving chassis and method, the moving chassis comprises a cross-layer inverted-T-shaped rack assembly, an environment sensing and automatic control assembly is arranged at the top end of the cross-layer inverted-T-shaped rack assembly, and a step-by-step lifting transverse propelling assembly is arranged at the bottom end of the cross-layer inverted-T-shaped rack assembly; self-adaptive high-position longitudinal moving assemblies connected with the cross-layer inverted-T-shaped rack assembly are arranged on the two sides of the environment sensing and automatic control assembly at intervals. Two pairs of driven double-hub wheel assemblies connected with the cross-layer inverted-T-shaped rack assembly are arranged on the two sides of the step-by-step lifting transverse propelling assembly in a clearance mode. The mobile chassis of the self-propelled reinforcing steel bar binding robot can automatically and stably walk on complex reinforcing steel bar mesh surfaces with different sizes and specifications, so that the construction efficiency of reinforcing steel bar binding operation is greatly improved, and the construction period is remarkably shortened.

The invention discloses a dynamic obstacle avoidance method and system for a substation inspection robot. The position of the robot is positioned based on laser point cloud data, and static and dynamic obstacle features are extracted at the same time. An environment association graph is constructed through perception information in the advancing process, and feature information of each node serves as an attribute of the environment association graph; calculating the distance, the relative speed and the relative angle between the edge weights of the nodes according to the nodes; on the basis of the environment association graph, aggregating all-graph obstacle node information to generate real-time feature representation for robot nodes, and correcting a calculation mode of attention weight according to edge weight between the nodes; and finally outputting a real-time feature vector of the robot node. And defining a reinforcement learning framework and executing path planning. According to the invention, through combination of GAT and dynamic planning, efficient obstacle avoidance and path planning of the robot in a substation environment are realized, and the decision and execution capability of the robot in a dynamic complex scene can be improved.

The invention discloses a driving area visualization method, device and equipment for automatic parking and a storage medium, and relates to the technical field of parking, and the driving area visualization method for automatic parking comprises the following steps: obtaining vehicle-mounted camera data, radar point cloud data, vehicle size and safety distance, a target obstacle recognition model based on an encoder-decoder and a self-attention module is used for analyzing camera data, and the initial position, size and type of an obstacle are determined. Then, the positions are adjusted through radar data to obtain accurate real-time obstacle positions, and a drivable area is constructed in combination with the vehicle size and the safe distance. And finally, the system sends camera data, radar data and drivable area information to a vehicle controller, the vehicle controller reconstructs an automatic parking scene by using a scene restoration model, and a result is transmitted to a vehicle display to present an automatic parking view. According to the invention, the automatic parking scene view can be displayed, the driving area can be displayed, and the collision risk is reduced.

The invention discloses a path planning method and system based on multi-modal joint integration, and the method employs an image processing and deep learning technology to automatically recognize and eliminate the influence of a dynamic object, and guarantees that SLAM and VIO systems are only matched with a stable static background object. According to the technical scheme, firstly, feature points in an input image are preliminarily screened and matched through the image feature point matching technology; and then introducing a deep learning network to carry out refined segmentation and instance identification. The deep learning network can accurately distinguish a static object from a dynamic object, and accurately identify the existence of the dynamic object in a complex scene. After the dynamic objects are identified, the dynamic objects are eliminated from the feature matching process by adopting an intelligent elimination strategy, so that the SLAM and VIO systems are ensured to perform feature matching and positioning calculation only based on stable static background objects, the negative influence of the dynamic objects on the positioning precision is avoided, and the stability and applicability of the system in a complex dynamic environment are improved.

The invention relates to the technical field of robot autonomous following, in particular to a robot autonomous following method based on a single-line laser radar. Comprising the steps of collecting laser radar data, performing intelligent processing on the laser radar data to obtain accurate laser radar data, identifying contour features of a following target by using a simultaneous positioning and mapping technology, forming a reference point of the following target, and constructing a map; the constructed map is updated and optimized, and an optimized map is obtained; based on the optimized map and the accurate laser radar data, performing dynamic target tracking optimization and target motion prediction by using a dynamic adaptive path planning and target tracking algorithm to obtain a preliminary dynamic path plan; and based on the initial dynamic path planning, introducing a priority-based task scheduling mechanism to obtain the optimal path planning. The technical problems of low environmental adaptability and low tracing accuracy in the autonomous tracing implementation process of the robot in a closed environment are solved.

The invention discloses an aircraft anti-collision early warning method and device. The method comprises the steps that first flight information of a first aircraft and multiple pieces of second flight information corresponding to multiple second aircrafts adjacent to the first aircraft in a one-to-one mode are obtained in real time; analyzing a plurality of corresponding three-dimensional linear distances D between the first aircraft and the plurality of second aircrafts one by one according to the first flight information and the plurality of pieces of second flight information; allocating corresponding early warning levels to the plurality of three-dimensional linear distances according to a preset early warning strategy; according to the plurality of three-dimensional linear distances, distributing corresponding early warning levels to match an early warning mode so as to generate early warning information corresponding to the first airplane; and performing early warning prompt according to the sequence of the early warning information. The real-time receiving of the aircraft information in the preset range is realized, and the timeliness of the anti-collision early warning information is ensured. And real-time monitoring and accurate positioning of surrounding aircrafts are realized. Meanwhile, a multi-level early warning mechanism and a humanized interaction interface are adopted, and it is ensured that a pilot can timely and accurately perceive potential collision risks.

This invention relates to a navigation method for underwater unmanned vehicles (UUVs), comprising the following steps: acquiring data from multiple sensors and performing spatiotemporal synchronization, feature extraction, and data fusion preprocessing; defining the underwater operating area as a directed graph, with edge weights determined based on terrain complexity, obstacle risk, and water flow influence; based on the directed graph, employing an improved A* algorithm for global path planning and a fast exploratory random tree algorithm for local path planning and dynamic obstacle avoidance; assessing the environmental state considering aquatic environmental parameters, water flow characteristics, and topographic changes; and adjusting the navigation strategy using a deep reinforcement learning algorithm, with the navigation strategy adjustment objective function as the optimization goal, based on the environmental assessment results and the UUV's own state. Compared with existing technologies, this invention offers advantages such as improved navigation accuracy, reliability, and adaptability of underwater UUVs in complex underwater environments.

This invention discloses a multi-scale icing detection method and system for transmission line insulators under extreme weather conditions. The method includes: dividing a visible light image into pixel blocks and converting them into feature vectors with position encoding; inputting these vectors into an improved Swin Transformer integrating a Dynamic Deep Sample Generation (DDSGM) module; applying DDSGM to enhance features in a four-stage feature extraction phase; additionally fusing polarized light features in the first two stages; and outputting a multi-scale feature map. The multi-scale feature map is then input into a deformable attention unit containing DDSGM, undergoing secondary enhancement and adaptive sampling (fusing physical field information to correct for offset) to obtain enhanced features. A prediction head predicts the icing category, bounding box, and thickness; Hungarian matching is used to associate the target; and an early warning system outputs the results and a warning. This invention is the first to integrate DDSGM, optimize sampling, and combine a multi-physical quantity strategy to improve robustness to low-quality images and transparent ice layers, achieving accurate detection and early warning under extreme weather conditions.

This disclosure relates to the technical field of gas pipeline condition monitoring devices. It provides a collision alarm device for overhead natural gas pipelines, used to monitor the columns supporting overhead natural gas pipeline sections. The device includes a swing arm, a sliding member, a pressure sensor, and an elastic member. The swing arm is oscillatingly mounted on the column. The sliding member is slidably mounted on the column and contacts the swing arm, configured such that after the swing arm oscillates, the swing arm pushes the sliding member to slide. The pressure sensor is mounted on the column and is used to transmit data to an external controller. The two ends of the elastic member act on the sliding member and the pressure sensor, respectively, to absorb the force between the swing arm and the sliding member and transmit the force to the pressure sensor. This technical solution solves the problem in related technologies where collisions or scrapes between external objects cause pipeline leaks in overhead sections of natural gas pipelines or columns, leading to untimely detection, natural gas loss, and increased on-site safety hazards.

This invention discloses a target recognition and localization method based on self-switching of a multi-view vision system. The method uses a target recognition algorithm to identify and perceive the target object, and acquires local depth information based on the binocular imaging principle. Based on this local depth information, three indicators are calculated to evaluate localization accuracy: single-pixel deviation influence factor σ, depth information error rate ε, and depth dispersion ξ. Two evaluation indicators, Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity-to-Simple (SSIM), are obtained based on the local image information of the target. A target localization evaluation model is established to quantify the localization accuracy. According to the autonomous switching logic, the target recognition result and the localization accuracy evaluation score are compared sequentially to achieve autonomous switching between multiple binocular cameras with different focal lengths. This satisfies the requirements for both near-range and far-range target recognition and localization, overcoming the problems of low localization accuracy of low-focal-length binocular cameras for distant targets and small field of view of high-focal-length binocular cameras.

The invention discloses a self-adaptive multi-path redundant data transmission method and device based on multi-strategy configuration. The method comprises the following steps of: pre-configuring a logic table for defining a service type and transmission logic; obtaining service data, querying the logic table according to the service type identifier to determine the data copy number and the optimal port combination, and copying the data according to the data copy number and the optimal port combination; a multi-port connection state is maintained by a double-frequency dynamic mechanism, and the mechanism is switched to high-frequency keep-alive when service data exists and is recovered to low-frequency keep-alive when no service exists; and selecting an available port for data replication and sending the available port based on a preferred port combination and a real-time port connection state in the logic. According to the invention, differentiated redundancy transmission can be carried out according to service importance, the real-time performance and signaling overhead of link state detection are balanced through a double-frequency keep-alive mechanism, and the data transmission reliability and the resource utilization rate in a heterogeneous network environment are improved.

The embodiment of the invention provides a mode memory enhanced dynamic graph convolution traffic flow prediction method and related equipment, and belongs to the field of intelligent traffic and spatio-temporal data prediction. According to the method, a two-state driven dynamic graph generation mechanism is constructed, and the method comprises the following steps: extracting dynamic mode embedding of traffic data through a mode memory network to capture unsteady disturbance characteristics; time period embedding is extracted through an independent period embedding unit so as to encode a steady state rule; and fusing the two to generate a dynamic graph structure at each moment. According to the method, a dynamic graph generation module is further integrated at a decoding end of an encoder-decoder architecture, so that an adaptive dynamic graph can be regenerated according to time information of a target moment when each future moment is predicted, and graph structure dynamic evolution of a whole training and prediction process is realized. According to the invention, the traffic flow prediction precision, especially the accuracy of long-term prediction, is effectively improved, and the generalization ability of the model to unexperienced scenes is enhanced.

The invention relates to a rapid positioning detection method for a high-precision pipeline route. The method comprises the following steps: S1, firstly, putting a detection device into a pipeline along the flowing direction of a fluid medium; s2, starting a measuring instrument of the head mechanism, performing data fusion by a Kalman filter, and transmitting the data to a control center of the head mechanism in real time; s3, opening a fan plate, and then starting propellers on the head mechanism and the propelling mechanism: the two mechanisms work together to realize the swimming and steering movement of the detection device in the pipeline, and detecting the condition in the pipeline by using a sensor; s4, when the head mechanism detects that the flow velocity in the pipeline changes suddenly, the sector plate is withdrawn, the position and range of the flow velocity change are recorded, a warning signal is sent out through a control center of the head mechanism, and then other parts in the pipeline continue to be detected; and if the signal cannot be sent, ending detection and immediately returning. According to the invention, the pipeline can be automatically, efficiently and automatically detected, the accuracy is high, the real-time performance is good, the labor cost is reduced, and the manual operation risk is reduced.

The application discloses a kind of light storage fills integrated equipment charging protection device, including charging pile body, the charging pile body both sides are connected with plug-in mouth, two The plug-in mouth inside is inserted with charging head, the charging pile body both sides correspond two charging head below It is connected with wire collecting box, two The wire collecting box lower end is respectively connected with the upper end both sides of base, the charging pile body both sides lower part It is connected with electric cable, two The electric cable is respectively penetrated corresponding wire collecting box and extends to corresponding wire collecting box side, the present application is realized active anti-collision early warning by sensing positioning strip and light alarm equipment, with the automatic regularity of wire collecting mechanism It is realized that cable is protected, and relies on broken line mechanism to execute physical cut-off in fire time, so as to realize early warning, protection and isolation integration's charging safety protection.

This utility model provides an air-to-ground unmanned early warning and monitoring system, comprising: an unmanned aerial vehicle (UAV) swarm: including multiple UAVs, each UAV further comprising a UAV transmission module, a UAV controller, an UAV onboard processor, a gimbal camera, and a positioning module. The UAV controller is mounted on the UAV, and the UAV onboard processor is an onboard computer (Orin) mounted on the UAV. A GPS antenna is mounted on the top of each UAV. This utility model uses a heterogeneous collaborative architecture as the core of the air-to-ground unmanned early warning and monitoring system. By complementing the wide-area field of view of UAVs with the local details of unmanned vehicles, it eliminates monitoring blind spots. It adopts a lightweight vision solution to achieve accurate perception and avoids high-cost load. Relying on a cross-platform control framework to dynamically schedule the linkage of air-to-ground equipment, it significantly improves the efficiency and real-time performance of early warning.

The utility model provides a trajectory environment collection mechanism, including shell, reading board part, collection module group and detection circuit group, reading board part collection module group along the axial direction of shell from top to bottom sets gradually, reading board part collection module group and detection circuit group are fixedly connected through the terminal strip and screw, when reading board part collection module group and detection circuit group assemble in place, will make the turn and screw cap and install in the shell, and use electronic pouring sealant to pour seal the space between reading board part collection module group and detection circuit group. The utility model discloses a trajectory environment collection mechanism, through the integrated structure design, waterproof performance is excellent, and the anti -interference ability is strong, safe, and the reliability is high.

The application provides a kind of multi-dimensional behavior situation awareness and intelligent decision-making method, system and electronic equipment, belongs to behavior situation awareness decision-making technical field.Method includes: constructing dynamic behavior relationship diagram, collecting behavior data and mapping therein;Generate behavior semantic vector, mine causal dependence and construct behavior causal graph;Establish probability benchmark model based on historical normal behavior, calculate current behavior deviation and locate abnormal source, generate situation assessment results;Risk evolution path and propagation range are deduced using causal time series prediction model;Through hierarchical decision engine, generate differentiated disposal instructions, and iteratively optimize the model according to the execution effect.The application improves the accuracy of anomaly detection and risk prediction ability through deep fusion of multi-source data and accurate positioning of abnormal source, solves the problems of insufficient data fusion, low detection accuracy and delayed decision response of traditional methods, realizes real-time perception of system behavior situation, and provides intelligent decision support for network security and system operation.

The application relates to the technical field of intelligent interaction systems and automobile customer service marketing, and discloses a traffic automobile customer service marketing method and system based on a large language model, wherein the traffic automobile customer service marketing method based on the large language model comprises the following steps: collecting driver fixation point data by using an eye movement tracking algorithm to generate eye movement trajectory data sets; constructing a visual attention heat map to form a user attention distribution model; calculating multi-medium characteristic parameters to establish a medium characteristic model; using the large language model to generate and adapt personalized marketing content and output multi-medium compatible marketing information; realizing intelligent attention guidance according to the user attention distribution model, the medium characteristic model and the multi-medium compatible marketing information; and realizing accurate perception and quantitative analysis of the visual attention of the driver through the eye movement tracking technology, so that the marketing system can master the user focus points in real time, and the technical problem that a traditional interface cannot perceive actual user focus points is solved.

This invention discloses a green and ecological planting method for daylilies, belonging to the field of agricultural planting technology. It addresses the technical problems of excessive reliance on chemical fertilizers and traditional mulching in current daylily cultivation, leading to soil ecological imbalance and low fertilizer utilization. The method includes: applying organic fertilizer made from cow and sheep manure and Asteraceae plant residues through aerobic composting, using a layered strip application method. First, fully decomposed organic fertilizer is applied as a base fertilizer layer in the planting furrow, followed by semi-decomposed organic fertilizer that has completed the main fermentation stage but not the later aging stage as a middle layer fertilizer; after land preparation and ridging, wood ash is spread on the ridge surface, followed by the laying of black fully biodegradable mulch film; finally, holes are made in the mulch film for daylily seedling planting. This method is mainly used for sustainable ecological planting of daylilies, improving soil fertility and structure, regulating soil pH and providing potassium, conserving moisture and suppressing weeds while avoiding residual pollution, thereby increasing daylily yield and quality while reducing environmental impact.

The present application belongs to the technical field of networked converter control, and discloses a networked converter control method, system, device and medium based on network side information feedback to solve the problem of subsynchronous oscillation. The method comprises: collecting voltage signals of a grid side point of common connection and voltage signals and current signals of a converter side; calculating steady-state active power and steady-state reactive power based on the voltage signals of the grid side point of common connection and line impedance parameters; calculating transient active power and transient reactive power based on the voltage signals and current signals of the converter side; introducing a composite factor to fuse the steady-state active power and the transient active power to obtain composite active and reactive power; and inputting the composite active and reactive power to a power outer loop controller of the networked converter to generate a reference signal for inner loop control. The present application improves system damping and the ability to suppress subsynchronous oscillation.

The application belongs to the technical field of robot control, and particularly relates to a master-slave operating arm with electromagnetic force feedback for precision assembly and a bidirectional transparent control method, which comprises a master-slave robot unit, an information transmission unit, a master-slave control unit, an electromagnetic force feedback unit, a bidirectional transparent control unit, a singular point processing unit, an adaptive path planning unit and a master-slave proportion matching unit. The electromagnetic force feedback unit obtains environmental force information through a magnetic coupling component and transmits the information to the master control unit; the bidirectional transparent control unit constructs a master-slave control closed loop, balances the system transparency and stability under communication delay, detects and avoids joint singular points through a Jacobian matrix, and constructs a potential field function to generate an optimal motion path. The application realizes high-precision control and force feedback under a communication delay environment, and improves the accuracy and safety of remote operation.

This invention discloses a method, apparatus, device, and medium for morphological operations based on local density maps, relating to the field of image processing technology. The method includes: calculating a local density map based on a target image; using the local density map to characterize the local grayscale distribution features of each pixel; assigning structural element parameters of different sizes to different regions based on the local density map; performing morphological operations on the target image using structural element parameters of different sizes to obtain multiple morphological operation results; and fusing the multiple morphological operation results based on the local density map to obtain the final morphological operation result. This invention accurately perceives local image features through local density maps, adaptively assigns structural element parameters of different sizes to different regions, and organically combines the processing advantages of different parameters through a fusion strategy. While maintaining high computational efficiency, it effectively balances detail preservation and noise suppression, significantly improving the processing quality of morphological operations.

The invention relates to the technical field of intelligent park digital operation and maintenance, in particular to a multi-node collaborative intelligent park digital operation and maintenance system and a control method, and the method comprises the following steps: collecting multi-source operation data, and executing protocol conversion, time synchronization and local jitter removal processing at a gateway end; sending the processed collected data to a central access layer, carrying out authentication and flow limiting on the data, and writing the data into message middleware; mapping data accessed by the message-oriented middleware to a unified asset model and performing unit and semantic standardization; writing the time sequence data into a time sequence database, and writing the event and the audit information into an event and audit library; and performing missing interpolation, quality labeling and preliminary filtering on the data. According to the invention, through a hierarchical alarm and strategy scheduling mode, the accuracy of abnormity identification and the timeliness of operation and maintenance processing are improved, the frequency of manual intervention is reduced, and the stability and reliability of system operation and the overall efficiency of intelligent park digital operation and maintenance are improved.

The application discloses an automatic material distribution control method and system based on radar and visual three-dimensional reconstruction and a pump truck, and belongs to the field of intelligent manufacturing of pump trucks. In order to solve the problems of high risk of manual operation, inaccurate single sensor sensing and inability to automatically distribute materials, the application adopts a radar and visual fusion scheme: sensors are arranged on the boom, images and point clouds are synchronously collected, a three-dimensional model of a construction scene is reconstructed after being processed and fused, and is mapped to a boom coordinate system; obstacle detection and real-time collision warning are carried out based on the model, a collision-free motion trajectory is planned through inverse kinematics, and adaptive closed-loop material distribution control is realized according to drop point information fed back by vision. The system integrates data acquisition, processing and reconstruction, scene recognition and control modules. The pump truck applying the application has high-precision environment sensing and real-time obstacle avoidance capability, can realize full-automatic and accurate material distribution, and significantly improves operation safety and efficiency.

The invention discloses a power distribution network disaster damage situation sensing method and system based on emergency communication under rainfall flood. The method comprises the following steps: establishing a power distribution network disaster damage situation sensing model under urban rainstorm waterlogging; constructing a communication working point site selection model, and determining possible parking points of the mobile unmanned aerial vehicle station and the emergency communication vehicle; based on possible stop points of the mobile unmanned aerial vehicle station and the emergency communication vehicle, considering path constraint and time constraint, determining a scheduling scheme and a transfer path of the mobile unmanned aerial vehicle station and the emergency communication vehicle, and constructing a vehicle-aircraft cooperative communication recovery model; and based on the constructed power distribution network disaster damage situation perception model under the urban rainstorm waterlogging and the constructed vehicle-machine cooperative communication recovery model, constructing a power distribution network disaster damage situation perception improvement model under the urban rainstorm waterlogging, and solving the power distribution network disaster damage situation perception improvement model under the urban rainstorm waterlogging to obtain a situation perception improvement strategy. The disaster damage situation sensing capability of the power distribution network is improved, and rapid recovery of the power distribution network after the disaster is realized.

The application discloses a kind of training map element mining network model method, device and readable storage medium.Map element mining network model is used to perceive map element in photo, method includes obtaining initial photo for marking high-precision map element;Initial photo is segmented into multiple picture blocks;According to preset proportion, multiple picture blocks are randomly occluded to obtain occlusion area and visible area;According to visible area, occlusion area is completed to obtain completed picture;When the pixel value of the same position of completed picture and initial photo exists difference value, difference value is used to adjust map element mining network model, and the difference value of the pixel value of the same position of the picture completed by the map element mining network model after adjustment and initial photo is less than set value, can save the cost of manual labeling map element, and make the map element mining network model after training can accurately perceive the new and unusual map element in photo.

The application relates to the technical field of automatic driving and provides an automatic driving assistance method and system, which comprises the following steps: acquiring point cloud data, wherein the point cloud data is data information acquired by a detection signal emitted by a laser radar installed on a first vehicle; generating a two-dimensional image according to the point cloud data, wherein the two-dimensional image is a two-dimensional image of the surrounding environment of the first vehicle perceived by the laser radar installed on the first vehicle; performing feature extraction on the two-dimensional image by using a neural network model to obtain feature information, wherein the feature information comprises feature information of each foreground point in the two-dimensional image; performing clustering on each foreground point in the two-dimensional image based on the feature information to obtain an instance segmented two-dimensional image; and assisting a user in taking over the first vehicle according to the instance segmented two-dimensional image, so that the automatic driving vehicle can accurately and quickly perceive the surrounding environment, the driving personnel can take over the vehicle in the automatic driving according to the perception result, and the safety of the automatic driving is improved.

The application relates to the technical field of audio processing, and discloses a vibration compensation audio processing system, which comprises a vibration sensing module connected with an audio signal processor, is used for collecting vibration signals between a vehicle body and a sound equipment in real time and transmitting the vibration signals to the audio signal processor; the audio signal processor is connected with the vibration sensing module, a loudspeaker and a control unit respectively, is used for receiving vibration-related signals and original audio signals, generating vibration compensation signals, superimposing the vibration compensation signals on the original audio signals and transmitting the superimposed signals to the loudspeaker; and the loudspeaker is used for converting electric signals into sound waves by receiving the superimposed audio signals and outputting the superimposed audio signals. The vibration sensing module is used for collecting multi-modal signals and extracting vibration features, so that the problem that audio signals are prone to distortion and sound quality is reduced in vehicle driving caused by the fact that most traditional vehicle-mounted audio systems adopt hardware isolation and software filtering to reduce the influence of vibration on audio signals is solved.