BIM-based trajectory analysis method and system, terminal and storage medium

Abstract

The invention relates to the technical field of smart park safety monitoring, and particularly provides a BIM-based trajectory analysis method and system, a terminal and a storage medium, and the method comprises the steps: S1, collecting real-time target information, obtaining a real-time target emotional state, obtaining a real-time target position and a real-time target movement trajectory based on monitoring equipment; dynamically dividing the building area into a plurality of sub-areas based on the BIM model, and obtaining boundary information of each sub-area; calculating the group intensity in each sub-region; weight values are configured for the target emotional state, the target moving track and the group intensity of the targets in each area; constructing a multi-input LSTM risk assessment model, and outputting a predicted target risk score; and analyzing and judging the target risk score based on a preset risk threshold, and when the target risk score is greater than the risk threshold, judging that the target has a risk. The identification and management capability of potential risks is effectively improved, and safety precaution measures are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of smart park security monitoring technology, and in particular to a BIM-based trajectory analysis method, system, terminal and storage medium. Background Art

[0002] With the rapid development of smart cities and smart parks, accurate analysis of the target person's trajectory has become increasingly important. This demand mainly stems from the continuous improvement of security management level, continuous optimization of resource allocation, and significant enhancement of emergency response capabilities. In the daily operation of smart cities and smart parks, being able to accurately track and analyze the movement trajectory of the target person is of great significance for preventing crime, optimizing crowd management, and improving resource utilization efficiency.

[0003] In smart cities and smart parks, the scenes are very complex, which is reflected in the intricate building structures, diverse personnel composition, rich activity patterns, etc. In such an environment, accurate, comprehensive and efficient trajectory analysis technology is needed to support the management and operation of cities and parks. Traditional trajectory analysis uses basic positioning technology.

[0004] Under complex environmental conditions, it is easy to make recognition errors or fail to accurately track the target, with low recognition efficiency and analysis accuracy. Moreover, it can often only process trajectory data in isolation, with poor information integration, resulting in poor risk assessment capabilities within the park. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a BIM-based trajectory analysis method, system, terminal and storage medium to solve the above-mentioned technical problems.

[0006] In a first aspect, the present invention provides a BIM-based trajectory analysis method, comprising the following steps: S1, collects real-time target information based on monitoring equipment, obtains real-time target emotional state based on target information combined with face recognition technology, and obtains real-time target position and real-time target movement trajectory based on target tracking algorithm; S2, dynamically divide the building area into multiple sub-areas based on the BIM model and obtain the boundary information of each sub-area; S3, calculating the group density in each sub-region based on the target location and the boundary information of each sub-region; S4, configures weight values ​​for the target emotional state, target movement trajectory, and group density of the targets in each area; S5, build a multi-input LSTM risk assessment model, input the target emotional state with weight values, target movement trajectory and group density, and output the predicted target risk score; S6, analyzing and judging the target risk score based on a preset risk threshold. When the target risk score is less than the risk threshold, the target is judged to be risk-free. When the target risk score is greater than the risk threshold, the target is judged to be risky.

[0007] In an optional implementation, in step S5, the historical target emotional state, historical target movement trajectory, historical group density and real historical risk score with weight values ​​are used as historical data sets to construct a multi-input LSTM risk assessment model. The specific steps include: S5-1, in the model input layer, input channels are set for the target emotional state, target movement trajectory, and group density, and the data in the historical data set are input into the model; S5-2, for each input channel, a separate LSTM layer is constructed for feature extraction to obtain the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point; S5-3, perform weighted summation of the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point to obtain a new feature vector, and capture the feature pattern through the ReLU activation function; S5-4, mapping the feature vector through the Sigmoid activation function to obtain the predicted historical risk score; S5-5, based on the predicted historical risk score and the actual historical risk score, the mean square error loss function is used to calculate the loss value. When the loss value is greater than the loss threshold, the model parameters are updated to obtain the final multi-input LSTM risk assessment model.

[0008] In an alternative embodiment, The sub-areas include public areas, functional areas and restricted areas; In step S4, when configuring the weight values, the sum of the weight values ​​of the target emotional state, target movement trajectory and group density is 1. In public areas, the weight value of group density is the largest; in functional areas, the weight value of the target's emotional state is the largest; in restricted areas, the weight value of the target's movement trajectory is the largest.

[0009] In an optional embodiment, the risk threshold includes a low risk threshold, a medium risk threshold and a severe risk threshold; When the predicted risk score is less than the low-level risk threshold, there is no risk; When the predicted risk score is not less than the low risk threshold and less than the medium risk threshold, the target is in a low risk state and is marked in the BIM model; When the predicted risk score is not less than the medium risk threshold and less than the severe risk threshold, the target is in a medium risk state and an alarm is issued; When the predicted risk score is not less than the heavy risk threshold, the target is in a heavy risk state, an alarm is issued and security personnel are notified.

[0010] In an optional implementation, in step S1, obtaining the real-time target emotional state based on target information combined with face recognition technology specifically includes: Acquire the face area in the target information based on the face detection algorithm; Matching the face region with pre-stored face information based on a deep learning model to obtain the target source of the face region; Obtain the target emotional state in the face area based on convolutional neural network.

[0011] In an optional implementation, obtaining the real-time target position and the real-time target movement trajectory based on the target tracking algorithm specifically includes: In the initial frame of the target information, the initial target position and target identity feature information of the target are obtained based on the automatic detection algorithm; In subsequent frames of target information, the new target position is determined based on a feature matching algorithm combined with target identity feature information; Connect all target positions in consecutive frames to form a complete target movement trajectory.

[0012] In an optional implementation, step S1 further includes mapping the target movement trajectory data to the BIM model, specifically including: Preprocessing the target movement trajectory data; Map the preprocessed target movement trajectory data into the coordinate system of the BIM model; Draw the moving trajectory of the target using line and arrow elements according to the mapped trajectory data; Update the movement trajectory of the target in the BIM model in real time.

[0013] In a second aspect, the present invention provides a BIM-based trajectory analysis system, wherein the system implements the steps of the BIM-based trajectory analysis method, including: Data acquisition module, which collects real-time target information based on monitoring equipment; BIM area division module, which dynamically divides the building area into multiple sub-areas based on the BIM model and obtains the boundary information of each sub-area; The data processing module obtains the real-time target emotional state based on the target information combined with face recognition technology, obtains the real-time target position and real-time target movement trajectory based on the target tracking algorithm, and calculates the group density in each sub-area based on the target position and the boundary information of each sub-area; The model building module configures weight values ​​for the target emotional state, target movement trajectory, and group density of the target in each area; builds a multi-input LSTM risk assessment model, inputs the target emotional state, target movement trajectory, and group density with weight values, and outputs the predicted target risk score; The risk judgment module analyzes and judges the target risk score based on the preset risk threshold. When the target risk score is less than the risk threshold, the target is judged to be risk-free. When the target risk score is greater than the risk threshold, the target is judged to be risky.

[0014] In a third aspect, a terminal is provided, including: A memory for storing a BIM-based trajectory analysis program; A processor is used to implement the steps of the above-mentioned BIM-based trajectory analysis method when executing the BIM-based trajectory analysis program.

[0015] In a fourth aspect, a computer-readable storage medium storing a computer program is provided, wherein a BIM-based trajectory analysis program is stored on the readable storage medium, and when the BIM-based trajectory analysis program is executed by a processor, the steps of the above-mentioned BIM-based trajectory analysis method are implemented.

[0016] The beneficial effect of the present invention is that the BIM-based trajectory analysis method, system, terminal and storage medium provided by the present invention realize dynamic monitoring and analysis of the target's emotional state, movement trajectory and group density by integrating face recognition technology, target tracking, BIM model and LSTM model. By configuring weights for different factors, the system can accurately calculate the risk score of each sub-area and make risk judgments based on preset thresholds, thereby effectively improving the ability to identify and manage potential risks and enhancing safety precautions. Under complex environmental conditions, it can correctly track the situation of the target, with high recognition efficiency and analysis accuracy, and because the trajectory data is fused with other data, the information integration is high, resulting in a high risk assessment capability within the park.

[0017] In addition, the invention has a reliable design principle, a simple structure and a very broad application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 It is a schematic flow chart of a BIM-based trajectory analysis method according to an embodiment of the present invention.

[0020] Figure 2 It is a schematic block diagram of a BIM-based trajectory analysis system according to an embodiment of the present invention.

[0021] Figure 3 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention.

[0022] Among them, 100, system; 110, data acquisition module; 120, BIM area division module; 130, data processing module; 140, model building module; 150, risk judgment module; 200, terminal; 210, processor; 220, memory; 230, communication unit. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0025] The BIM-based trajectory analysis method provided in the embodiment of the present invention is executed by a computer device, and accordingly, the BIM-based trajectory analysis system runs in the computer device.

[0026] Figure 1 is a schematic flow chart of a method according to an embodiment of the present invention. Figure 1 The execution subject may be a trajectory analysis system based on BIM. According to different requirements, the order of the steps in the flowchart may be changed, and some may be omitted.

[0027] like Figure 1 As shown, the method includes: S1, collects real-time target information based on monitoring equipment, obtains real-time target emotional state based on target information combined with face recognition technology, and obtains real-time target position and real-time target movement trajectory based on target tracking algorithm; By combining monitoring equipment with face recognition technology to obtain the real-time emotional state of the target, it is possible to detect emotional fluctuations of personnel in a timely manner. If negative emotions of personnel, such as anger or anxiety, can be detected in a timely manner, managers can quickly take measures to appease them and avoid possible conflicts or adverse events. In addition, for some key monitored personnel, changes in their emotional state may indicate the occurrence of certain dangerous behaviors. Knowing this in advance can effectively prevent safety accidents.

[0028] Based on the target tracking algorithm, the target position and movement trajectory are obtained, which can achieve accurate tracking and positioning of personnel.

[0029] S2, dynamically divide the building area into multiple sub-areas based on the BIM model and obtain the boundary information of each sub-area; Dynamically dividing building areas and obtaining boundary information can more accurately monitor the flow of people and provide a basis for subsequent density calculations and risk assessments.

[0030] S3, calculating the group density in each sub-region based on the target location and the boundary information of each sub-region; Real-time calculation of the crowd density within each sub-area helps identify crowded areas and take timely measures to ensure personnel safety and prevent potential stampedes.

[0031] S4, configures weight values ​​for the target emotional state, target movement trajectory, and group density of the targets in each area; Configuring weight values ​​for the target's emotional state, movement trajectory, and group density can more comprehensively reflect the target's actual risk status and improve the accuracy of risk assessment.

[0032] S5, build a multi-input LSTM risk assessment model, input the target emotional state with weight values, target movement trajectory and group density, and output the predicted target risk score; Using the LSTM model to process multiple input factors can effectively predict the risk score of the target, help managers identify and deal with potential risks in advance, and improve site safety.

[0033] S6, analyzing and judging the target risk score based on a preset risk threshold. When the target risk score is less than the risk threshold, the target is judged to be risk-free. When the target risk score is greater than the risk threshold, the target is judged to be risky.

[0034] By analyzing the relationship between the risk score and the preset threshold, it is possible to quickly determine whether the target is at risk, ensure the timeliness of safety measures, and reduce the possibility of accidents.

[0035] Optionally, as an embodiment of the present invention, in step S1, obtaining the real-time target emotional state based on the target information combined with face recognition technology specifically includes: Acquire the face area in the target information based on the face detection algorithm; The Adaboost algorithm based on Haar-like features uses a large number of Haar-like features to describe local areas in the image. These features can be simply understood as the light and dark contrast of rectangular areas of different directions and sizes. For example, the eye area is usually darker than the cheek area. By combining these features, a classifier that can distinguish between face and non-face areas can be constructed.

[0036] Slide a fixed-size window across the entire image, calculate the Haar-like features within the window, and input them into the classifier for judgment. If it is judged to be a face, the area corresponding to the window is recorded, which is the detected face area.

[0037] Matching the face region with pre-stored face information based on a deep learning model to obtain the target source of the face region; Deep learning models (such as convolutional neural networks, CNN) are used here to match the detected face regions with pre-stored face information. First, the pre-stored face information needs to be processed. Usually, these face images are input into the deep learning model for feature extraction.

[0038] During the training phase, the model learns how to map a face image into a low-dimensional feature space. For example, for a trained face recognition model, it can convert a face image into a 128-dimensional feature vector.

[0039] When a new face region needs to be matched, it is input into the same deep learning model to obtain its feature vector. Then the distance between the new face feature vector and the pre-stored face feature vector (such as Euclidean distance, cosine distance, etc.) is calculated to determine whether it matches.

[0040] If the distance is less than a certain threshold, the match is considered successful, thereby obtaining the target source of the face area (for example, this person is a known employee, visitor, etc.).

[0041] Obtain the target emotional state in the face area based on convolutional neural network.

[0042] In order to obtain the target emotional state in the face region, a convolutional neural network specially trained for emotion recognition is used. The convolutional neural network model is trained based on a large number of face image datasets with labeled emotional states (such as happy, sad, angry, calm, etc.).

[0043] The input of the model is a face region image, which is processed by multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer is used to extract local features in the image, such as the shape of the eyes, the curvature of the corners of the mouth, and other features that may be related to emotions. The pooling layer is used to reduce the amount of data while retaining the main features. The fully connected layer synthesizes the previously extracted features and finally outputs a vector representing the emotional state. For example, a vector of length 5 may be output, corresponding to the probabilities of the five emotions of happiness, sadness, anger, calmness, and surprise. The emotional category with the highest probability is the predicted target emotional state.

[0044] Optionally, as an embodiment of the present invention, in step S1, obtaining the real-time target position and the real-time target movement trajectory based on the target tracking algorithm specifically includes: In the initial frame of the target information, the initial target position and target identity feature information of the target are obtained based on the automatic detection algorithm; In the initial frame of the target information, the YOLO automatic detection algorithm scans the image or video frame, and it divides the input image into multiple grid cells, each of which is responsible for predicting several bounding boxes and the category probabilities of these bounding boxes. For each bounding box, it predicts the center coordinates (x, y), width and height (w, h) of the bounding box, and a confidence score.

[0045] The confidence score reflects the possibility that the predicted bounding box contains the target and the accuracy of the prediction. By setting a certain confidence threshold, the bounding boxes that meet the requirements are screened out, and the positions framed by these bounding boxes are the initial target positions of the targets.

[0046] At the same time, during the detection process, the algorithm also extracts the target's identity feature information. This feature information may include the target's appearance features, such as color histograms, texture features, etc., or high-dimensional feature vectors extracted based on deep learning models.

[0047] In subsequent frames of target information, the new target position is determined based on a feature matching algorithm combined with target identity feature information; Assume that the color histogram features of the target are extracted in the initial frame. In subsequent frames, the entire image or a specific area is searched to calculate the similarity between the color histogram of each potential position and the color histogram of the target in the initial frame. In this embodiment, the Bhattacharyya distance is used to calculate the similarity.

[0048] The new position of the target in the subsequent frame is determined by finding the area in the subsequent frame that best matches the target identity feature information in the initial frame. At the same time, in order to improve the accuracy and stability of the match, multiple features are usually combined for matching, and the Kalman filter is used to predict and correct the target position.

[0049] Connect all target positions in consecutive frames to form a complete target movement trajectory.

[0050] For example, the target position in the first frame is (x1, y1), in the second frame it is (x2, y2), in the third frame it is (x3, y3), and so on. The target movement trajectory is the broken line formed by connecting these points in sequence.

[0051] In practical applications, these trajectories are smoothed using methods such as moving average to reduce the impact of noise.

[0052] Optionally, as an embodiment of the present invention, step S1 further includes mapping the target movement trajectory data to the BIM model, specifically including: Preprocessing the target movement trajectory data; By setting the distance threshold, the distance relationship between each point and the adjacent points before and after is determined. If a point is too far away from the adjacent point (exceeding the threshold), it is determined as a noise point and removed. Sometimes, due to equipment failure or occlusion, the target position data in some frames is missing. For a small amount of missing data, linear interpolation can be used to fill it. Since the target movement trajectory data may be in different scale ranges, the target movement trajectory data is normalized to facilitate subsequent mapping to the BIM model coordinate system.

[0053] Map the preprocessed target movement trajectory data into the coordinate system of the BIM model; Determine the conversion relationship between the coordinate system of the target movement trajectory data and the BIM model coordinate system, establish a coordinate transformation matrix through at least 3 known corresponding points (for example, the coordinates of specific landmark points in the building in both the monitoring device coordinate system and the BIM model coordinate system are known), and use the affine transformation formula to achieve coordinate transformation.

[0054] Draw the moving trajectory of the target using line and arrow elements according to the mapped trajectory data; For the target moving trajectory data points mapped to the BIM model coordinate system, the adjacent points are connected in chronological order to form a trajectory line. The straight line drawing algorithm in computer graphics can be used, such as the Bresenham algorithm. This algorithm determines the actual drawn pixel point by judging the distance between the pixel point and the ideal straight line, thereby efficiently drawing a straight line segment. At the same time, an arrow is added to the trajectory line to indicate the moving direction of the target.

[0055] Update the movement trajectory of the target in the BIM model in real time.

[0056] Continuously acquire new target movement trajectory data and repeat the above data preprocessing, coordinate system mapping and trajectory drawing steps.

[0057] Optionally, as an embodiment of the present invention, the sub-areas include a public area, a functional area and a restricted area; In step S4, when configuring the weight values, the sum of the weight values ​​of the target emotional state, target movement trajectory and group density is 1. In public areas, the weight value of group density is the largest; in functional areas, the weight value of the target's emotional state is the largest; in restricted areas, the weight value of the target's movement trajectory is the largest.

[0058] Public areas usually have a large flow of people, and crowd gathering has the most significant impact on safety and order. For example, the atrium of a shopping mall, the waiting hall of a station, etc., where there are dense crowds, it is easy to cause incidents such as trampling and disputes. Functional areas have specific uses. People carry out specific activities in the area, and their emotional state has a great impact on the normal development of the activities and the safety of the area. For example, in offices and classrooms, the emotions of employees or students may affect work or study efficiency, and even cause conflicts and other incidents. Restricted areas are areas where unauthorized persons are not allowed to enter. Once a person enters and has an abnormal movement trajectory, there may be a major safety hazard, such as the important material storage area of ​​the warehouse, the core computer room of the data center, etc. Therefore, according to the different characteristics of each area, corresponding weight configuration is performed to make the target risk score of each area more reasonable.

[0059] Optionally, as an embodiment of the present invention, in step S5, the historical target emotional state, historical target movement trajectory, historical group density and real historical risk score with weight values ​​are used as historical data sets to construct a multi-input LSTM risk assessment model. The specific steps include: S5-1, in the model input layer, input channels are set for the target emotional state, target movement trajectory, and group density, and the data in the historical data set are input into the model; independent input channels are set for different types of data, so that the model can process and learn the characteristics of the target emotional state, target movement trajectory, and group density respectively. This multi-input structure helps capture the unique information and dynamic changes of each input feature.

[0060] The historical dataset is:

[0061] in, Indicates The target emotional state history data of samples, Indicates The target movement trajectory history data of samples, Indicates The historical data of population density of samples, Indicates The real historical risk scores of samples.

[0062] S5-2, for each input channel, a separate LSTM layer is constructed for feature extraction to obtain the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point; The LSTM layer includes the calculation of input gate, forget gate and output gate. The LSTM layer of each input channel extracts the output of the last time step, which represents the information accumulated in the entire time series, which is crucial for the subsequent risk score prediction.

[0063] S5-3, perform weighted summation of the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point to obtain a new feature vector, and capture the feature pattern through the ReLU activation function; By weighted summation, the features extracted from different input channels are fused, so that the model can comprehensively consider the contribution of each feature to the risk score. The way of setting weights enables the model to learn the importance of different features, thereby enhancing the prediction ability. The ReLU activation function introduces nonlinear factors to help the model capture complex feature patterns and interactions, and avoid information loss caused by linear combinations.

[0064] S5-4, mapping the feature vector through the Sigmoid activation function to obtain the predicted historical risk score; The activated feature vector is converted into a value between 0 and 1 through the Sigmoid function, which represents the predicted historical risk score. This mapping makes the output of the model have clear probabilistic meaning, which is convenient for decision-making or classification.

[0065] S5-5, based on the predicted historical risk score and the real historical risk score, the mean square error loss function is used to calculate the loss value. When the loss value is greater than the loss threshold, the model parameters are updated to obtain the final multi-input LSTM risk assessment model. The mean square error loss function measures the gap between the model's predicted value and the real value, helping the model evaluate its performance. The calculation of the loss value is a key indicator in the training process and guides the parameter update. The model parameters are optimized through the back propagation algorithm in order to reduce the loss and improve the accuracy of the model. Using a suitable optimization algorithm (such as Adam or SGD) can accelerate convergence and improve the performance of the model in practical applications.

[0066] Optionally, as an embodiment of the present invention, the risk threshold includes a low risk threshold, a medium risk threshold and a severe risk threshold; When the predicted risk score is less than the low-level risk threshold, there is no risk; When the predicted risk score is not less than the low risk threshold and less than the medium risk threshold, the target is in a low risk state and is marked in the BIM model; When the predicted risk score is not less than the medium risk threshold and less than the severe risk threshold, the target is in a medium risk state and an alarm is issued; When the predicted risk score is not less than the heavy risk threshold, the target is in a heavy risk state, an alarm is issued and security personnel are notified.

[0067] Suppose we are monitoring a large public event (such as a music festival) and using a multi-input LSTM model to predict the risk scores of the participants.

[0068] Risk threshold setting: low risk threshold (LRT): 0.3, medium risk threshold (MRT): 0.6, heavy risk threshold (HRT): 0.9.

[0069] Assume that the following predicted risk scores are obtained at different time points: Time point 1: predicted risk score = 0.2; Time point 2: predicted risk score = 0.4; Time point 3: predicted risk score = 0.7; Time point 4: Predicted risk score = 0.95.

[0070] Risk status and treatment: At time point 1, the predicted risk score obtained by the LSTM risk assessment model is 0.2, which means there is no risk. We continue to monitor the target and record the risk-free status.

[0071] At time point 2, the predicted risk score obtained by the LSTM risk assessment model is 0.4, which is low risk. The target is marked as low risk in the BIM model, and the status is recorded for subsequent analysis.

[0072] At time point 3, the predicted risk score obtained by the LSTM risk assessment model is 0.7, which is medium risk. An alarm is issued to remind security personnel to be vigilant and strengthen monitoring of the target.

[0073] At time point 4, the predicted risk score obtained by the LSTM risk assessment model is 0.95, which indicates a serious risk. An emergency alarm is issued, security personnel are notified immediately, and the emergency plan is activated to quickly assess the on-site situation and take corresponding measures.

[0074] Through this layered risk assessment and response mechanism, the event's security management team can quickly identify and respond to different levels of risks to ensure the safety of participants and the smooth running of the event.

[0075] In some embodiments, the BIM-based trajectory analysis system may include a plurality of functional modules composed of computer program segments. The computer programs of the various program segments in the BIM-based trajectory analysis system may be stored in a memory of a computer device and executed by at least one processor to perform (see Figure 1 Description) Functions of trajectory analysis based on BIM.

[0076] In this embodiment, the BIM-based trajectory analysis system can be divided into multiple functional modules according to the functions it performs, such as Figure 2 As shown. The functional modules of the system 100 may include: module 110, data acquisition module, 120, BIM area division module, 130, data processing module, 140, model building module, 150, risk judgment module. The module referred to in the present invention refers to a series of computer program segments that can be executed by at least one processor and can complete fixed functions, which are stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0077] Optionally, as an embodiment of the present invention, the data acquisition module collects real-time target information based on the monitoring equipment; a high-resolution, low-light-sensitivity camera array is used to ensure that clear facial images can be captured under various lighting conditions. The camera network covers key entrances, passages, public areas, etc. of the building, forming a monitoring network with no blind spots.

[0078] Optionally, as an embodiment of the present invention, the BIM area division module dynamically divides the building area into multiple sub-areas based on the BIM model, and obtains boundary information of each sub-area; Optionally, as an embodiment of the present invention, the data processing module obtains the real-time target emotional state based on target information combined with face recognition technology, obtains the real-time target position and real-time target movement trajectory based on the target tracking algorithm, calculates the group density in each sub-area based on the target position and the boundary information of each sub-area; and uses advanced deep learning algorithms (such as convolutional neural network CNN) to deeply extract and compare facial features. The algorithm has been optimized with a large amount of training data and can accurately identify faces of different ages, genders, races and expressions, while effectively resisting interference factors such as makeup, wearing glasses or masks.

[0079] Optionally, as an embodiment of the present invention, a model building module configures a weight value for a target emotional state, a target movement trajectory, and a group density of a target in each region; a multi-input LSTM risk assessment model is constructed, the target emotional state, the target movement trajectory, and the group density with weight values ​​are input, and a predicted target risk score is output; Optionally, as an embodiment of the present invention, the risk judgment module analyzes and judges the target risk score based on a preset risk threshold. When the target risk score is less than the risk threshold, it is judged that the target is risk-free; when the target risk score is greater than the risk threshold, it is judged that the target is risky.

[0080] Figure 3 A schematic diagram of the structure of a terminal 200 provided in an embodiment of the present invention. The terminal 200 can be used to execute the BIM-based trajectory analysis method provided in an embodiment of the present invention.

[0081] The terminal 200 may include: a processor 210, a memory 220 and a communication unit 230. These components communicate via one or more buses. Those skilled in the art will appreciate that the server structure shown in the figure does not limit the present invention, and it may be a bus structure or a star structure, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0082] The memory 220 can be used to store the execution instructions of the processor 210, and the memory 220 can be implemented by any type of volatile or non-volatile storage terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. When the execution instructions in the memory 220 are executed by the processor 210, the terminal 200 can perform some or all of the steps in the following method embodiments.

[0083] The processor 210 is the control center of the storage terminal, and uses various interfaces and lines to connect various parts of the entire electronic terminal. It runs or executes software programs and / or modules stored in the memory 220, and calls data stored in the memory to perform various functions of the electronic terminal and / or process data. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of multiple packaged ICs with the same or different functions. For example, the processor 210 can only include a central processing unit (CPU). In the embodiment of the present invention, the CPU can be a single computing core or multiple computing cores.

[0084] The communication unit 230 is used to establish a communication channel so that the storage terminal can communicate with other terminals, receive user data sent by other terminals or send user data to other terminals.

[0085] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, and when the program is executed, the program may include some or all of the steps in each embodiment provided by the present invention. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0086] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution in the embodiments of the present invention, in essence or in other words, the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, and other media that can store program codes, including several instructions for enabling a computer terminal (which can be a personal computer, a server, or a second terminal, a network terminal, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention.

[0087] In this specification, the same or similar parts between the various embodiments can be referred to each other. In particular, for the terminal embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.

[0088] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or modules, which can be electrical, mechanical or other forms.

[0089] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0090] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0091] Although the present invention has been described in detail with reference to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention. Any person of ordinary skill in the art may easily think of changes or substitutions within the technical scope disclosed by the present invention, and these shall be within the scope of protection of the present invention.

Claims

1. A trajectory analysis method based on BIM, characterized in that: The following steps are involved: S1, collects real-time target information based on monitoring equipment, obtains real-time target emotional state based on target information combined with face recognition technology, and obtains real-time target position and real-time target movement trajectory based on target tracking algorithm; S2, dynamically divide the building area into multiple sub-areas based on the BIM model and obtain the boundary information of each sub-area; S3, calculating the group density in each sub-region based on the target location and the boundary information of each sub-region; S4, configures weight values ​​for the target emotional state, target movement trajectory, and group density of the targets in each area; S5, build a multi-input LSTM risk assessment model, input the target emotional state with weight values, target movement trajectory and group density, and output the predicted target risk score; S6, analyzing and judging the target risk score based on a preset risk threshold. When the target risk score is less than the risk threshold, the target is judged to be risk-free. When the target risk score is greater than the risk threshold, the target is judged to be risky.

2. The BIM-based trajectory analysis method according to claim 1, characterized in that: In step S5, the historical target emotional state with weight values, the historical target movement trajectory, the historical group density and the real historical risk score are used as the historical data set to construct a multi-input LSTM risk assessment model. The specific steps include: S5-1, in the model input layer, input channels are set for the target emotional state, target movement trajectory, and group density, and the data in the historical data set are input into the model; S5-2, for each input channel, a separate LSTM layer is constructed for feature extraction to obtain the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point; S5-3, perform weighted summation of the historical target emotional state features, historical target movement trajectory features, and historical group density features at the last time point to obtain a new feature vector, and capture the feature pattern through the ReLU activation function; S5-4, mapping the feature vector through the Sigmoid activation function to obtain the predicted historical risk score; S5-5, based on the predicted historical risk score and the actual historical risk score, the mean square error loss function is used to calculate the loss value. When the loss value is greater than the loss threshold, the model parameters are updated to obtain the final multi-input LSTM risk assessment model.

3. The BIM-based trajectory analysis method according to claim 1, characterized in that: The sub-areas include public areas, functional areas and restricted areas; In step S4, when configuring the weight values, the sum of the weight values ​​of the target emotional state, target movement trajectory and group density is 1. In public areas, the weight value of group density is the largest; in functional areas, the weight value of the target's emotional state is the largest; in restricted areas, the weight value of the target's movement trajectory is the largest.

4. The BIM-based trajectory analysis method according to claim 1, characterized in that: The risk thresholds include a low-level risk threshold, a medium-level risk threshold, and a severe-level risk threshold; When the predicted risk score is less than the low-level risk threshold, there is no risk; When the predicted risk score is not less than the low risk threshold and less than the medium risk threshold, the target is in a low risk state and is marked in the BIM model; When the predicted risk score is not less than the medium risk threshold and less than the severe risk threshold, the target is in a medium risk state and an alarm is issued; When the predicted risk score is not less than the heavy risk threshold, the target is in a heavy risk state, an alarm is issued and security personnel are notified.

5. The BIM-based trajectory analysis method according to claim 1, characterized in that: In step S1, obtaining the real-time target emotional state based on target information combined with face recognition technology specifically includes: Acquire the face area in the target information based on the face detection algorithm; Matching the face region with pre-stored face information based on a deep learning model to obtain the target source of the face region; Obtain the target emotional state in the face area based on convolutional neural network.

6. The BIM-based trajectory analysis method according to claim 1, characterized in that: Acquiring real-time target position and real-time target movement trajectory based on target tracking algorithm specifically includes: In the initial frame of the target information, the initial target position and target identity feature information of the target are obtained based on the automatic detection algorithm; In subsequent frames of target information, the new target position is determined based on a feature matching algorithm combined with target identity feature information; Connect all target positions in consecutive frames to form a complete target movement trajectory.

7. The BIM-based trajectory analysis method according to claim 1, characterized in that: Step S1 also includes mapping the target movement trajectory data to the BIM model, specifically including: Preprocessing the target movement trajectory data; Map the preprocessed target movement trajectory data into the coordinate system of the BIM model; Draw the moving trajectory of the target using line and arrow elements according to the mapped trajectory data; Update the movement trajectory of the target in the BIM model in real time.

8. A trajectory analysis system based on BIM, characterized in that: The system implements the steps of the BIM-based trajectory analysis method according to any one of claims 1 to 7, including: Data acquisition module, which collects real-time target information based on monitoring equipment; BIM area division module, which dynamically divides the building area into multiple sub-areas based on the BIM model and obtains the boundary information of each sub-area; The data processing module obtains the real-time target emotional state based on the target information combined with face recognition technology, obtains the real-time target position and real-time target movement trajectory based on the target tracking algorithm, and calculates the group density in each sub-area based on the target position and the boundary information of each sub-area; The model building module configures weight values ​​for the target emotional state, target movement trajectory, and group density of the target in each area; builds a multi-input LSTM risk assessment model, inputs the target emotional state, target movement trajectory, and group density with weight values, and outputs the predicted target risk score; The risk judgment module analyzes and judges the target risk score based on the preset risk threshold. When the target risk score is less than the risk threshold, the target is judged to be risk-free. When the target risk score is greater than the risk threshold, the target is judged to be risky.

9. A terminal, characterized in that: include: A memory for storing a BIM-based trajectory analysis program; A processor, configured to implement the steps of the BIM-based trajectory analysis method according to any one of claims 1 to 7 when executing the BIM-based trajectory analysis program.

10. A computer-readable storage medium storing a computer program, characterized in that: The readable storage medium stores a BIM-based trajectory analysis program, and when the BIM-based trajectory analysis program is executed by a processor, the steps of the BIM-based trajectory analysis method according to any one of claims 1 to 7 are implemented.

Images