1151 results about "Industrial setting" patented technology

The invention provides an industrial environment monitoring and accident prediction method fusing multi-modal data, and relates to the technical field of data processing, and the method comprises the steps: carrying out the semantic collection and causal association preprocessing of multi-modal heterogeneous data collected in real time through constructing a dynamic industrial knowledge graph; a customized deep learning model is adopted to extract deep abstract features of each mode, and weak signals and potential risks are accurately represented and uncertainty is quantified; a high-fidelity digital twin model is utilized to drive a deep reinforcement learning algorithm, and dynamic optimization and verification are performed to generate a multi-level and multi-target preventive intervention strategy combination; an intervention strategy is executed through an edge-end-cloud three-layer collaborative intelligent architecture, and online learning and system sustainable evolution are realized by using a closed-loop data feedback mechanism. According to the method, the sensing and early warning capability of the early weak and complex abnormal state of the industrial environment can be remarkably improved, the accident evolution path is accurately predicted, and credible explanation is provided.

In example embodiments, an industrial technology stack for an industrial environment includes a set of computational resources and a set of layers executed by the set of computational resources, the set of layers including a governance layer, an enterprise layer, an offering layer, a transaction layer, an operations layer, a network layer, a data layer, and a resource layer. In example embodiments, the industrial technology stack may include one or more artificial intelligence models for implementing one or more components of one or more layers of the set of layers.

The invention discloses a multi-source heterogeneous data fusion system based on an RAG. According to the method, through deep knowledge fusion and a dynamic cognitive evolution mechanism, the decision-making intelligence level in a complex data environment is remarkably improved, and equipment operation parameters, environment indexes and a domain knowledge base are deeply associated to form a panoramic data view with space-time continuity. A generative enhancement mechanism endows original data with a self-evolution characteristic, industry empirical rules and real-time situation awareness are injected while the fidelity of the original characteristic is maintained, so that a decision model can capture micro data fluctuation and follow macroscopic business logic, accurate balance between risk early warning and resource scheduling is realized, and the risk early warning efficiency is improved. The dynamic adaptation characteristic enables the system to autonomously update a knowledge system and optimize a decision path in a complex and changeable industrial environment, and a post response mode of a traditional static analysis model is converted into an intelligent center with prospective pre-judgment and real-time regulation and control capabilities.

The invention relates to the technical field of the Internet of Things, in particular to an intelligent device communication method and system based on the Internet of Things. The method comprises the following steps: acquiring spectrum occupation data, electromagnetic interference mode data and terminal equipment energy state data of a plurality of communication channels in an industrial environment through a preset distributed spectrum sensing network; performing adaptive spectrum analysis processing on the spectrum occupancy data and the electromagnetic interference mode data to obtain channel quality evaluation data including interference source characteristics and channel attenuation characteristics; and according to the channel quality evaluation data and a pre-acquired historical channel performance memory bank, performing priority ranking on the available communication channels to obtain dynamic channel priority data based on time continuity. The spectrum resource allocation condition in the industrial environment is monitored in real time through the distributed spectrum sensing network, the current interference can be identified, and the interference can be pre-judged through an advanced prediction model.

The invention discloses an equipment operation evaluation method based on multi-source data fusion, and relates to the technical field of equipment operation evaluation, and the method comprises the following steps: S1, data collection: collecting multi-source data of equipment in a real-time operation state, the multi-source data comprising sensor data, operation logs, environment data and maintenance history of the equipment; according to the equipment operation evaluation method based on multi-source data fusion, accurate equipment state monitoring and fault prediction can be provided, and the performance and accuracy of the equipment can be continuously optimized according to actual equipment operation data and environmental conditions of the equipment, so that efficient and intelligent equipment management and maintenance strategies are realized; according to the method, specific steps and features of the equipment operation evaluation method based on multi-source data fusion are clearly defined, comprehensiveness and high efficiency of the method are ensured, and the method is suitable for various complex industrial environments so as to improve reliability and maintenance efficiency of equipment.

The invention discloses an industrial environment information wireless monitoring system based on the Internet of Things and a control method thereof, and relates to the technical field of industrial management, and the system comprises an intelligent edge calculation module, a data visualization module, a self-healing network management module, a precise energy consumption management module, an environment anomaly detection module and a wireless network optimization module. According to the industrial environment information wireless monitoring system based on the Internet of Things and the control method thereof, real-time analysis and decision are realized at a data source by using an intelligent edge computing module, data transmission delay is reduced, response can be quickly made in an emergency, potential faults are predicted through a time sequence abnormity self-learning network model, and the monitoring accuracy of the industrial environment information wireless monitoring system based on the Internet of Things is improved. And network self-healing is actively carried out before a problem occurs, so that the network downtime is effectively reduced. In addition, due to the introduction of a low-power-consumption ad hoc network protocol, the energy consumption optimization management of the sensor nodes is realized, and the service life of equipment is prolonged.

The invention discloses an industrial data acquisition and management method based on the Internet of Things, and relates to the technical field of data management, and the method comprises the following steps: carrying out the demand analysis of an industrial environment and a production process, and determining to-be-monitored industrial equipment indexes in the industrial environment and the production process, namely an equipment performance index and a production performance index; according to the industrial data acquisition and management method based on the Internet of Things, equipment performance indexes and production performance indexes are monitored in real time, so that the reliability of equipment is remarkably improved, the accidental downtime is shortened, and through comprehensive analysis of an equipment state evaluation index, a fault analysis evaluation index and a production performance evaluation index, the reliability of the equipment is improved. The method can predict possible faults and performance reduction of the equipment, so as to take preventive maintenance measures in advance, not only can reduce production interruption caused by equipment faults, but also can improve the efficiency and effect of maintenance work by optimizing a maintenance plan.

The invention discloses an intelligent power distribution room safety monitoring management system, which comprises the following steps: checking states of a sensor, a gateway and a camera, and collecting environment data through the sensor; preprocessing the collected data, identifying a smoke area by using a YOLOv5 model, and starting a linkage safety mechanism according to abnormal environmental parameters; performing personnel detection by using a YOLOv5s model, capturing and analyzing a behavior track through a SimAM mechanism, and realizing multi-target tracking and abnormal behavior detection; grading the abnormal events based on a rule classifier, and triggering equipment linkage control; and storing the data in an NoSQL database, analyzing a trend to predict potential dangers, and optimizing a YOLOv5 model through historical data. The system can monitor the environment and personnel safety conditions in the power distribution room in real time, has efficient response capability and a trend prediction function, and can be widely applied to the field of safety monitoring of intelligent power grids and industrial environments.

The invention relates to the technical field of data processing, in particular to an industrial environment data acquisition device and method, and the device comprises a sensing layer, an edge calculation layer, a cloud platform layer, and an application layer. Compared with a traditional scheme in which a single sensor is adopted to work independently, software synchronization is relied on, and signal conditioning parameters are fixed, the technical scheme of the invention realizes significant breakthrough through collaborative design of a multi-mode sensor array and an intelligent signal conditioning system, microsecond-level clock synchronization is realized by adopting a hardware-level PTP protocol, and the signal conditioning precision is improved. The program control amplifier is combined with the self-adaptive anti-aliasing filter, so that the conditioning circuit can be self-adaptive to the working condition change of equipment; meanwhile, nanosecond-level real-time processing of signals is achieved through an FPGA + ARM heterogeneous processor architecture, abnormal high-speed sampling is dynamically triggered in cooperation with an ARM, compared with a traditional scheme, the real-time performance is improved by three orders of magnitude, meanwhile, invalid data transmission is reduced by 80%, and a high-timeliness and low-redundancy intelligent monitoring solution is provided for early failure early warning of industrial equipment.

The invention discloses a charging and moisture regaining equipment temperature prediction method based on a time sequence fusion network model, and relates to the field of production process control, and the method comprises the steps: collecting time sequence data in a charging and moisture regaining equipment production environment, and carrying out the preprocessing; dividing the data set into a training set, a verification set and a test set, and injecting Gaussian noise into the training set; a prediction model for predicting the outlet temperature is constructed, and the prediction model is a time sequence fusion network model and comprises a residual TCN time sequence convolutional network, an SK-Net multi-scale attention network and a BiLSTM bidirectional circulation network; pre-training the prediction model by using the training set; utilizing the trained prediction model to predict the outlet temperature of the feeding and moisture regaining equipment; and evaluating a prediction result, and if an evaluation index is greater than a threshold value, starting an incremental training process to re-train the prediction model. According to the invention, through a multi-module combined deep learning model, the prediction accuracy of the outlet temperature of the charging and moisture regaining equipment can be improved in a complex and changeable industrial environment.

The invention relates to an intelligent safety protection management method and system. The method comprises the following steps: acquiring behavior data and position data of personnel in an industrial site; analyzing the behavior data and the position data by using a pre-trained personnel behavior model to obtain a behavior recognition result; according to the behavior recognition result and the current operation state of the target equipment, judging a risk level corresponding to the personnel behavior; based on the risk level, a corresponding safety response strategy is matched in the edge computing node, a control instruction is generated according to the safety response strategy, and the control instruction is used for driving the target device to execute a corresponding response action; and the voice interaction module is used for acquiring voice input of an on-site operator, performing semantic recognition on the voice input to obtain a voice recognition result, and executing corresponding emergency control operation to cover a current control instruction when the voice recognition result meets a preset emergency instruction condition. The method has the effect of improving the accuracy of intelligent safety management in the industrial environment.

The invention discloses a dynamic calibration type laser surface real-time detection method and system, and the method comprises the following steps: obtaining initial calibration parameters of a laser projection module and a high-speed imaging module through a standard calibration plate, building an initial measurement coordinate system through a camera calibration method, and building a dynamic calibration model; through a dynamic calibration mechanism, calibration parameters are updated in real time, and a sub-pixel-level laser light strip center coordinate extraction method is combined, so that the detection precision is effectively improved, and tiny surface defects can be detected; by responding to changes of environmental parameters such as vibration, temperature and illumination intensity in real time and automatically triggering a calibration process, it is ensured that a high-precision detection result can still be kept in a complex industrial environment; by adopting the mode of combining the standard calibration plate and the workpiece reference area, the calibration time is shortened, and for workpieces of different models, quick adaptation can be achieved through quick migration of pre-training model parameters and a federal learning mode, and the production efficiency is improved.

The invention discloses a current sensor data calibration method for an industrial environment, and the method comprises the steps: collecting the original current sensor data of target equipment in the industrial environment in real time, and carrying out the dynamic preprocessing of the original current sensor data, and generating a first calibration data set; based on the first calibration data set, time domain features and frequency domain features are extracted, and a second feature vector set is constructed; iteratively calculating an optimal calibration parameter set through a self-adaptive optimization algorithm in combination with historical calibration data and the second feature vector set; and dynamically correcting the current data of the current sensor according to the optimal calibration parameter set, and outputting and updating the calibrated current value. According to the invention, through dynamic preprocessing, comprehensive time domain and frequency domain feature extraction and combination of historical calibration data, parameter adaptive optimization updating is carried out, and the problem that current sensor data calibration in an industrial environment is difficult to accurately adapt to dynamic change environment conditions in real time is effectively solved.

In some embodiments, the system includes a virtual reality environment that includes a digital twin of at least a portion of a physical industrial environment. In some embodiments, the system is configured to change the virtual reality environment based on changes in the physical industrial environment. In some embodiments, the changes are received by the system through sensors such as temperature and / or pressure sensors, as non-limiting examples. In some embodiments, the changes in the physical environment are detected through the analysis of images. In some embodiments, the system is configured to use artificial intelligence to detect the changes. In some embodiments, the system is configured to predict the effect of changes in the virtual and / or physical environment on other system components through variable changes in virtual models. In some embodiments, the system is configured to display the effects in the virtual environment before they are implemented in the physical environment.

The invention discloses an industrial zero sample anomaly detection method and system based on cross-modal prompt learning, and relates to the technical field of computer vision and deep learning. The method comprises the following steps: acquiring RGB images and point cloud data containing product defects and normal states in an industrial scene, and preprocessing the point cloud data; constructing an anomaly detection model, preliminarily extracting information representation of the RGB image and the point cloud data by using the anomaly detection model, and performing further feature extraction and semantic alignment operation through prompt information; training the anomaly detection model based on a cross-modal collaborative mechanism, and testing the trained anomaly detection model by adopting a collaborative modulation strategy; and performing anomaly detection on a to-be-detected product by using the trained anomaly detection model. According to the method, the real-time defect detection precision and robustness can be remarkably improved in an industrial environment, especially under the conditions of data scarcity and complex modality.

A system can include a backend system and a sensor kit configured to monitor an industrial setting. The sensor kit can include an edge device and a plurality of sensors that capture sensor data and transmit the sensor data via a self-configuring sensor kit network. At least one sensor can capture sensor measurements and output instances of sensor data, generate and output reporting packets, and transmit the reporting packets to the edge device via the self-configuring sensor kit network in accordance with a first communication protocol. The edge device receives reporting packets from the plurality of sensors via the self-configuring sensor kit network and transmits sensor kit packets to the backend system via a public network. The backend system can include a processing system and a storage system, where the processing system performs backend operations on the sensor data and the storage systems stores the sensor data.

Methods and apparatus for Time-Sensitive Networking Coordinated Transfer Learning in industrial settings are disclosed. An example apparatus includes at least one memory, instructions in the apparatus, and processor circuitry to execute the instructions to cause performance of an operation by a first machine according to a first configuration, process a performance metric of the performance of the operation by the first machine to determine whether the performance metric is within a threshold range, and in response to a determination that the performance metric is not within the threshold range, cause performance of the operation by a second machine according to a second configuration different from the first configuration.

The invention discloses an equipment anomaly detection method based on audio analysis, and the method comprises the steps: collecting a sound signal from equipment, and carrying out the preprocessing of the sound signal, so as to generate a logarithmic Mel spectrogram; processing the logarithmic Mel spectrogram through a time feature enhancement network TANet and a time domain hybrid enhancement module TMixup so as to enhance the time sequence features of the logarithmic Mel spectrogram and generate feature representation; a multi-scale time sequence feature extraction module MSTNet is utilized to extract multi-scale time sequence features covering short-term, medium-term and long-term dependency relationships from the original audio signals; the method comprises the following steps: performing diffusion and antagonism training in a potential space through a potential diffusion model LDM and a generative adversarial network GAN in combination with enhanced logarithmic Mel spectrogram features and multi-scale time sequence features extracted by MSTNet, generating high-quality features, analyzing the features through a discriminator of the generative adversarial network, and judging whether equipment is in an abnormal state or not. According to the method, efficient anomaly detection performance can be kept in a complex and changeable industrial environment, and a reliable and economical anomaly detection solution is provided for industrial equipment.

The invention belongs to the technical field of inspection robot systems, and particularly relates to an inspection robot path planning method and system.The visual semantic perception module collects an industrial environment image through a top industrial camera, after graying and Gaussian filtering preprocessing, feature points are detected and matched through an ORB algorithm, and a path planning result is obtained; in combination with an illumination self-adaptive threshold screening mechanism, mismatching points are eliminated, semantics are marked, and a semantic feature map is constructed; the initial path planning module generates an initial path through a semantic cost-containing A * algorithm based on the map; the dynamic obstacle avoidance module captures a moving obstacle by using a visual sensor, and predicts a trajectory through Kalman filtering; the path optimization module combines an initial path and an obstacle track, optimizes the path by using quadratic programming of a fusion curvature constraint and an energy consumption model, and corrects positioning by fusing vision and IMU data through a dynamic weight fusion algorithm; and the execution feedback module generates an instruction according to the optimized path, re-triggers path optimization, forms a closed loop, and ensures the inspection stability.

The invention relates to a production workshop carbon flow twinborn mapping method, and belongs to the technical field of carbon emission optimization. The method comprises the steps of collecting production workshop data in real time to perform multi-modal data fusion; constructing a carbon flow dynamic accounting mechanism model, and calculating the real-time carbon emission intensity of the process; constructing a carbon flow intensity tensor model, extracting multi-granularity carbon flow features based on the carbon flow intensity tensor model, and realizing real-time digital twinborn deduction of carbon flow propagation by using an intelligent prediction algorithm; establishing a differential equation to describe double-flow real-time interaction of the carbon flow and the value flow, constructing a carbon value incidence matrix, and performing carbon flow value analysis; and on the basis of the carbon value incidence matrix, a space-time carbon chain-oriented cooperative adjustment strategy is dynamically generated through a multi-objective optimization algorithm, a double-layer topological optimization model is constructed, and a closed-loop feedback mechanism is introduced to drive the economic sustainability of the control strategy in an industrial environment. The workshop carbon footprint can be displayed in real time, the workshop carbon emission condition can be truly reflected, and the carbon emission data can be analyzed and optimized in time.

The invention relates to the field of environment monitoring, and discloses an industrial source VOCs distribution analysis system based on a laser radar, which comprises a data acquisition module, a spectrum extraction module, a preliminary modeling module, an inversion optimization module, an intelligent analysis module, a visualization module and a dynamic monitoring and early warning module. The data acquisition module acquires VOCs data of an industrial source area and processes the VOCs data; the spectrum extraction module is used for extracting spectrum characteristics of VOCs from the processed laser signals; the preliminary modeling module estimates the VOCs concentration; the inversion optimization module globally optimizes the preliminary model in combination with an atmospheric transmission model, a nonlinear inversion algorithm and a genetic algorithm; the intelligent analysis module further improves the concentration prediction precision; the visual module is used for displaying the spatial distribution of the VOCs; and the dynamic monitoring and early warning module gives an early warning for abnormal concentration change. According to the invention, high-sensitivity, real-time monitoring and accurate distribution analysis of industrial source VOCs are realized, and a scientific basis is provided for pollution control and risk management of an industrial environment.

The invention relates to the technical field of industrial motion control of robots, and discloses an industrial motion control method and system based on an embedded mode. The speed of an end effector of the mechanical arm is described and comprises the relation between the linear speed and the angular speed and the joint speed, and the relation is the joint angle of the mechanical arm; the condition number is calculated, singular points are detected, specifically, the singular approaching degree is evaluated by calculating the condition number of the matrix, and the singular points are processed through pseudo-inverse; dynamic path planning is carried out, and a model of a current working space is constructed through monitored mechanical arm sensor data; and calculating an optimal path, selecting an execution path, and executing path planning. The performance of the industrial motion control system is remarkably improved, so that the industrial motion control system can operate efficiently, safely and reliably in a complex industrial environment, and the performance, reliability and adaptability of the system are improved.

The invention relates to the technical field of metal material heat treatment quality detection, in particular to a metal material rapid annealing effect detection method based on acoustic response, which comprises the following steps: step 1, acoustic excitation and synchronous acquisition; step 2, multi-physics field coupling compensation: calculating the phase compensation amount of the vibration signal based on the temperature distribution data and the acoustic characteristic temperature dependence of the material; the method comprises the following steps: acquiring an environment vibration frequency spectrum through a production line mechanical vibration monitoring device, and separating an excitation response component from an original vibration signal by adopting adaptive filtering; 3, annealing sensitive feature extraction; 4, dynamic quality mapping and control: inputting the real-time feature vector into a pre-trained machine learning model, and outputting a recrystallization proportion and a grain size; and when the output parameter exceeds the process threshold value, the control parameter adjustment of the annealing furnace is triggered. Through a non-destructive and accurate acoustic monitoring technology, the annealing effect of the metal material can be efficiently and accurately monitored in a high-speed, high-noise and strong-interference industrial environment.

Methods and systems for a monitoring system for data collection in an industrial environment including a data collector communicatively coupled to a plurality of input channels connected to data collection points operationally coupled to at least one industrial component in at least one of an industrial boiler system or industrial pipeline system; a data storage structured to store a library of stored noise patterns associated with operation of the at least one industrial component; a data acquisition circuit structured to interpret a plurality of detection values from the collected data; and a data analysis circuit structured to: analyze the collected data, determine a measured noise pattern for the at least one industrial component, and compare the measured noise pattern to the library of stored noise patterns to identify a changed condition of the at least one industrial component.

The invention relates to the technical field of intelligent monitoring of petroleum equipment, and particularly provides an industrial equipment state diagnosis method based on multi-modal fusion deep learning. The method comprises the following steps: synchronously acquiring load, vibration and current data through an explosion-proof RTU (Remote Terminal Unit), and transmitting the data to an edge node through an industrial LoRaWAN (Load RaWide Area Network); aligning a multi-source signal time sequence by adopting Kalman filtering, and fusing to generate data streams with consistent time domains; features are extracted by using wavelet packet energy entropy, and a lightweight ResNet18 model is input to realize high-precision diagnosis; based on the diagnosis result, a dynamic optimization strategy is generated through an LSTM-Transform parallel model; and finally, issuing a control instruction to the equipment for execution, and collecting feedback data to update the cloud model. Through the multi-rate signal time sequence alignment and feature-decision two-stage fusion technology, the problem of diagnosis errors caused by data asynchronization in a traditional method is solved, the fault recognition precision and the equipment energy efficiency are remarkably improved, and meanwhile the requirements for explosion prevention and real-time performance of the industrial environment are met.

The invention relates to the technical field of artificial intelligence and equipment maintenance, and discloses an equipment maintenance strategy intelligent recommendation method and system based on artificial intelligence, and the method comprises the steps: carrying out the multi-view data enhancement and self-supervision pre-training; carrying out comparative learning on physical consistency constraints; performing representation learning of physical data double-track fusion; meta-learning-driven cross-device knowledge migration is carried out; personalized maintenance strategy recommendation is realized through few-sample fine tuning; according to the method, the problems that in the industrial environment, annotated data is scarce, a pure data driving method lacks physical understanding, the interpretability of the maintenance strategy is insufficient, a physical model is difficult to process a complex non-linear relation, and generalization ability is caused by equipment isomerism are solved, and intelligent maintenance strategy recommendation which is high in precision, interpretable and easy to migrate is achieved; and the equipment reliability and the maintenance efficiency are improved.

The invention discloses an explosion-proof and intrinsic safety type combined feeder switch state monitoring method based on big data, and relates to the technical field of industrial electrical equipment state monitoring and fault diagnosis, and the method is realized based on a three-level architecture constructed by an edge computing node, a region computing node and a cloud platform. Comprising the following steps: S1, collecting multi-source sensing data of a feeder switch in real time by an edge computing node, and performing feature extraction and fusion calculation on the multi-source sensing data; according to the method for monitoring the state of the explosion-proof and intrinsic safety type combined feeder switch based on the big data, the contradiction between the real-time performance and the accuracy in multi-source heterogeneous data fusion processing is effectively solved, abnormity can be found in the first time, early warning can be triggered, and the reliability of a diagnosis conclusion is improved; and the accuracy and foresight of fault diagnosis are ensured. According to the grading processing mechanism, computing resources are reasonably distributed, the network transmission load is greatly reduced, and the system can stably operate in a complex industrial environment.

The present invention relates to a three-phase methodology to obtain a highly concentrated, purified chemical component from plant material in an industrial setting. The three phases of the methodology of the present invention include Phase 1 : Preparing and solubilization, Phase 2: Separation and Salting and Phase 3: crystallization. The methodology results in dried composition purified from a plant material having at least 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, or 95% of the chemical component by weight.

The present invention relates to a collaborative robot system used in an industrial environment where workers collaborate. The collaborative robot system comprises: a collaborative robot body having a multi-joint structure; a control unit for controlling each joint or auxiliary shaft; one or more sensors for detecting the operational state of the robot in real time; and a status diagnosis and response module for diagnosing the state of the robot on the basis of detected data and controlling the operation of the robot accordingly. Specifically, the system is configured to enable the robot to autonomously perform actions such as deceleration, stopping, and recovery in response to various state changes occurring during work, and to perform self-learning and trajectory optimization on the basis of data accumulated through repetitive tasks. Furthermore, the system includes a graphical user interface (GUI) for intuitive user interaction, which is linked to a digital twin-based virtual simulation environment, enabling presetting and modification of work paths. When multiple collaborative robots are operated together, the system enables task synchronization, path collision avoidance, and sharing of status information among the robots, and may be connected to an external control server or a cloud-based control system to allow integrated management of the entire workflow.

The invention discloses a dynamic feature fusion and anomaly detection method for multi-modal data in an industrial environment, and belongs to the technical field of computers, and the method comprises the steps: carrying out the preprocessing, background interference removal, noise cleaning and data format adjustment of collected industrial data, and generating cleaner and normative point clouds and images; secondly, sampling and querying of different scales are carried out on the preprocessed point cloud data, the calculation complexity is reduced, local geometric structure information is reserved, and the robustness of the data is enhanced; thirdly, inputting the processed point cloud data into a feature extractor, extracting point cloud features and storing the point cloud features; meanwhile, image feature extraction and storage are carried out on the RGB image data; secondly, inputting the point cloud features and the RGB image features into a multi-modal feature dynamic fusion module, performing fusion by fully utilizing information of two modals, and storing fusion features; and finally, performing anomaly detection by combining the point cloud features, the image features and the fusion features, and outputting a feature map of an abnormal region.