HPC curtain wall installation safety protection intelligent monitoring system

Abstract

The application discloses an HPC curtain wall installation safety protection intelligent monitoring system and relates to the technical field of intelligent monitoring.The system comprises a risk pre-performance module, a digital neural pre-performance model is constructed, installation posture data and historical accident characteristics are compared through a reinforcement learning algorithm, a safety risk heat map is generated, the safety risk heat map is spatially superimposed with a preset installation scene, and a safety risk index is generated; a risk early warning module, after a voice instruction signal is parsed through natural language, an instruction type is generated, a contradiction value is calculated for the instruction type and the safety risk index, when the contradiction value exceeds a preset safety threshold, a risk protection protocol of the digital neural pre-performance model is activated and an alarm signal is sent; and a communication module, which transmits the alarm signal and the safety risk index to a remote monitoring terminal.The application realizes intelligent quantitative analysis and spatial visual expression of safety risks through dynamic visual presentation of the digital neural pre-performance model and the safety risk heat map.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, and in particular to an intelligent monitoring system for safety protection of HPC curtain wall installation. Background Technology

[0002] With the continuous development of the construction industry and the increasing prevalence of high-rise and super high-rise buildings, HPC curtain walls have become widely used due to their excellent durability and decorative properties. Curtain wall installation, as a core aspect of high-altitude operations, carries extremely high safety risks, making intelligent monitoring for safety protection during HPC curtain wall installation particularly important. Currently, intelligent monitoring methods for safety protection during HPC curtain wall installation in the construction industry mainly rely on manual inspections, supplemented by simple sensor alarms. These methods can, to some extent, detect potential safety hazards and reduce the occurrence of accidents.

[0003] However, the core of conventional safety protection and management methods still relies on the timeliness and reliability of personnel observation. They still have inherent limitations in proactive prevention and risk warning. Currently, a prominent challenge facing the industry is the difficulty in achieving comprehensive monitoring of the safety status of personnel, equipment, and key connection points on the high-altitude curtain wall installation work surface. There are monitoring blind spots and response delays. Safety officers cannot cover all high-risk locations and dynamically changing work environments at all times, making it difficult to detect and warn in a timely manner. This results in a significant time lag between risk identification and emergency response. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent monitoring system for safety protection during HPC curtain wall installation to solve the problems of difficulty in monitoring high-altitude curtain wall work surfaces without blind spots and untimely risk handling.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an intelligent monitoring system for safety protection during HPC curtain wall installation, comprising: The data acquisition module collects real-time installation posture data of the HPC curtain wall and voice command signals from safety officers, and extracts historical accident features from the historical accident database. The risk simulation module constructs a digital neural simulation model, compares installation posture data with historical accident characteristics through reinforcement learning algorithms, generates a safety risk heat map, and spatially overlays the safety risk heat map with a preset installation scenario to generate a safety risk index. The risk warning module parses the voice command signal into natural language to generate command type, calculates the contradiction value between command type and security risk index, and activates the risk protection protocol of digital neural pre-simulation model and issues an alarm signal when the contradiction value exceeds the preset security threshold. The communication module transmits alarm signals and safety risk indices to the remote monitoring terminal.

[0007] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the installation posture data includes displacement, tilt angle and vibration frequency parameters; The safety officer's voice command signals include operation instructions, safety warnings, status confirmations, and emergency instructions.

[0008] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for extracting historical accident features are as follows: The association rule mining method is used to filter high-frequency features in the historical accident database to generate a basic accident association feature set. The causal chain verification and temporal interaction enhancement are performed on the basic accident association feature set to output historical accident features.

[0009] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for constructing the digital neural pre-model are as follows: Call and initialize the convolutional neural network and the long short-term memory network, and build the pose extraction layer and the accident matching layer; A digital neural pre-model is constructed by optimizing the topology search and interaction mechanism of the attitude extraction layer and the accident matching layer execution layer using the NAS algorithm.

[0010] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for generating the safety risk heat map are as follows: The installation attitude data and historical accident features are input into the digital neural pre-model. The attitude extraction layer extracts the fluctuation features through a gated recurrent network to generate a time-series feature vector. The accident matching layer uses a self-attention mechanism to mine causal patterns and generate an accident association feature matrix. The time-series feature vector and the accident-related feature matrix are integrated through feature channels to generate a safety risk heat map.

[0011] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for obtaining the preset installation scenario are as follows: Extract scene structure coordinates from equipment monitoring records and match them with the spatial coordinates of the safety risk heat map to generate a spatial mapping table; Spatial semantic segmentation and dynamic risk weight assignment are performed on the spatial mapping table to generate preset installation scenarios.

[0012] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the generation of the safety risk index refers to extracting the structural type and dynamic risk weight in a preset installation scenario based on a safety risk heat map, and then quantifying and integrating the structural type and dynamic risk weight to generate the safety risk index.

[0013] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for obtaining the contradictory value are as follows: Extract the spectral features of the safety officer's voice commands, generate a text string, and perform lexical analysis and semantic association on the text string to generate the command type; Operational risk matching and difference calculation are performed on instruction types and security risk indices to obtain contradictory values.

[0014] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the specific steps for issuing the alarm signal are as follows: Extract the highest safety threshold value that led to the occurrence of an accident from the historical accident database, and perform correlation correction on the highest safety threshold value to obtain the safety threshold; The system monitors the conflict values ​​in the current operating area in real time. If the conflict value exceeds the safety threshold, it triggers the risk protection protocol of the digital neural pre-simulation model and issues an alarm signal.

[0015] As a preferred embodiment of the intelligent monitoring system for safety protection of HPC curtain wall installation described in this invention, the transmission of alarm signals and safety risk indices to the remote monitoring terminal refers to encapsulating the alarm signals and safety risk indices into data and transmitting them to the remote monitoring terminal using the MQTT protocol for encryption.

[0016] The beneficial effects of this invention are as follows: By comparing installation posture data with historical accident characteristics through a digital neural pre-simulation model, this invention achieves intelligent quantitative analysis and spatial visualization of safety risks. By integrating monitoring data and historical accident data, it reduces the potential risks to construction personnel. The dynamic visualization of the safety risk heat map provides on-site and remote monitoring personnel with an intuitive, global, and real-time situational awareness of the location, severity, and evolution trend of high-risk areas, enabling the focus of safety supervision to be accurately shifted to the prevention stage. The output of the safety risk index provides a quantitative assessment benchmark, enhances the scientific nature and timeliness of safety risk early warning, provides key decision-making basis for risk disposal, and significantly reduces the probability of major installation safety accidents. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Fig. 1 A schematic diagram of installing a safety protection intelligent monitoring system for HPC curtain walls.

[0019] Fig. 2 A flowchart for generating the safety risk index.

[0020] Fig. 3 A flowchart for generating alarm signals.

[0021] Fig. 4 This is a flowchart for secure data transmission. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figs. 1-4 As one embodiment of the present invention, this embodiment provides an intelligent monitoring system for safety protection of HPC curtain wall installation, comprising the following steps: The data acquisition module collects real-time installation posture data of the HPC curtain wall and voice command signals from safety officers, and extracts historical accident features from the historical accident database. Installation attitude data includes parameters such as displacement, tilt angle, and vibration frequency. It should be noted that the displacement is obtained by measuring the change in support displacement using a high-precision laser rangefinder; the tilt angle is obtained by monitoring the tilt angle offset of the panel relative to the horizontal plane using a MEMS tilt sensor; and the vibration frequency parameter is obtained by collecting vibration signals from a triaxial accelerometer and analyzing the vibration frequency parameter through a fast Fourier transform. The safety officer's voice commands include operating instructions, safety warnings, status confirmations, and emergency commands; It should be noted that operating instructions are naturally captured by an omnidirectional microphone array covering the work area; safety warnings are accurately picked up by noise-canceling directional microphones pointing towards the work area; status confirmations are made by using a near-field microphone focused on the safety officer's mouth to ensure clear instructions without reverberation; and emergency instructions are responded to quickly by a high-sensitivity gooseneck microphone.

[0026] Collect multi-source accident data and perform preprocessing; It should be noted that the multi-source accident data is collected as follows: accident reports are periodically archived from enterprise and institutional accident archives; safety logs are entered by organizing on-site operation records; equipment monitoring records are exported from the historical operating data of sensors, instruments, and other equipment management settings; and third-party investigation reports are obtained through public information platforms. The collected multi-source accident data, including accident reports, safety logs, equipment monitoring records, and third-party investigation reports, undergoes data cleaning to remove incomplete, erroneous, or irrelevant data records. Format conversion is performed to ensure all data sources adhere to a unified format standard for subsequent processing. Deduplication is performed to eliminate duplicate data entries and ensure the uniqueness of the dataset. Normalization is used to map data values ​​from different sources to the same scale, preventing certain features from unduly affecting the results due to excessively large differences in magnitude. Finally, outlier handling is used to identify and correct data points that deviate from the normal range, ensuring the accuracy and reliability of the dataset.

[0027] The preprocessed multi-source accident data is classified by keyword matching to construct a historical accident database; It should be noted that an accident classification framework is constructed based on the fundamental dimensions of accident analysis, with the first-level dimension being accident type, the second-level dimension being causal mechanism, and the third-level dimension being severity. Subsequently, an "information extraction" process is used for data annotation: for text data such as accident reports, preliminary annotation is completed through regular expression matching and named entity recognition; for ambiguous expressions (such as "support deformation"), type labels are completed by combining contextual relationships (such as the subsequent "overall collapse"), and direct causes (equipment malfunction) and indirect causes (maintenance deficiency) are labeled. The historical accident database uses a hierarchical storage structure: the main table stores accident fields such as unique accident identifier, timestamp, geographical location, type label, causal classification, and severity level; the extended tables, through foreign key relationships, store detailed data such as on-site image paths, involved equipment numbers, and casualty information, ensuring logical connections and query efficiency.

[0028] The association rule mining method is used to filter high-frequency features in the historical accident database to generate a basic accident association feature set. It should be noted that, with accident type, direct cause, indirect cause, and severity as the core analysis items, all accident records in the historical accident database are traversed to identify frequently occurring feature combinations (such as "collapse type" often appearing together with "operational deviation" and "general accident"), forming candidate frequent itemsets. Subsequently, pruning optimization is performed to remove low-frequency subsets: the support of all k-1 dimensional subsets of each candidate frequent itemset is checked to see if they are all higher than the minimum support; if the support of any subset in the candidate frequent itemset is lower than the minimum support, this candidate frequent itemset is removed; finally, only candidate frequent itemsets in which all subsets meet the support requirements are retained. Then, random associations are excluded (such as "equipment aging - object strike" being removed because there is no stable causal relationship); the filtered candidate frequent itemsets are transformed into basic accident association feature sets.

[0029] It should also be noted that support is an indicator used in association rule mining to measure the frequency of feature combinations in a dataset, specifically the ratio of the number of accident records containing feature combinations to the total number of accident records. Minimum support is a statistical threshold defined based on the proportion of the frequency of feature combinations in the dataset to the total number of records, with an example value range of 1% to 5%.

[0030] Perform causal chain verification and temporal interaction enhancement on the basic accident association feature set, and output historical accident features; It should be noted that the Granger causality test is used to verify whether the temporal sequence and statistical dependence among the basic accident association feature sets constitute causality (e.g., whether "operational deviation" frequently occurs before "collapse"). Accident cases are retrospectively analyzed to eliminate spurious associations and retain strong causal chains with genuine causal relationships. Subsequently, temporal interaction enhancement is performed: dynamic change information of the basic accident association feature sets in the time dimension is extracted, and the temporal dependence of the basic accident association feature sets is captured through sliding window analysis. The temporal interaction information between features is quantified, such as the 20% increase in "collapse risk" for each additional instance of "operational deviation," and this temporal interaction information is encoded as temporal interaction features, such as "operational deviation - vibration anomaly." Finally, the strong causal chains and temporal interaction features after causal verification are integrated to output historical accident features containing causal logic and dynamic temporal information.

[0031] The risk simulation module constructs a digital neural simulation model, compares installation posture data with historical accident characteristics through reinforcement learning algorithms, generates a safety risk heat map, and spatially overlays the safety risk heat map with a preset installation scenario to generate a safety risk index. Call and initialize the convolutional neural network and the long short-term memory network, and build the pose extraction layer and the accident matching layer; It should be noted that in the TensorFlow framework, the pose extraction layer and the incident matching layer are built by sequentially calling the convolutional neural network and the long short-term memory network through the Keras API: The pose extraction layer takes the installed pose data as input, configures 3×3×3 convolutional kernels, stacks 4 layers of convolutional networks, and then connects a BatchNormalization layer after each convolutional network for standardization processing, and then enhances the nonlinear expression through the ReLU activation function, and finally flattens out a 256-dimensional pose feature vector; The incident matching layer takes the pose feature vector as input, calls the bidirectional long short-term memory network, and connects a multi-head attention mechanism at the output of the long short-term memory network to capture the temporal dependency of the pose feature vector before the incident occurs, thus completing the initial construction of the pose extraction layer and the incident matching layer.

[0032] A digital neural pre-model was constructed by optimizing the topology search and interaction mechanism of the attitude extraction layer and the accident matching layer using the NAS algorithm. It should be noted that the searchable layer topology space is defined as follows: For the pose extraction layer, the search space covers the optional types of convolutional kernel size, the optional range of the number of convolutional layers, and the optional types of cross-modal connection methods; for the incident matching layer, the search space covers the optional range of the number of long short-term memory networks, the optional range of the number of long short-term memory network layers, and the optional types of time-series interaction methods. Subsequently, a gradient-based NAS (Network Attached Storage) algorithm is adopted, with the incident matching accuracy and inference latency in historical incident features as joint optimization objectives. Probabilistic sampling is performed through a relaxation and continuous relaxation strategy, and the performance of different topologies is iteratively evaluated to generate a layer topology parameter set. During the optimization process, the inter-layer interaction mechanism is dynamically adjusted: Based on the layer topology parameter set, the high-dimensional pose feature vector is mapped to a low-dimensional space compatible with historical incident features. A gated interaction network is added before the incident matching layer, where the input is the concatenation of the low-dimensional pose feature vector and historical incident features, and the information fusion ratio is controlled by a gating function. Finally, when the incident matching accuracy and inference latency no longer improve, the search is terminated, the optimal layer topology parameter set is determined, and the digital neural pre-model is output.

[0033] Installation posture data and historical accident features are divided into a sample set, training set, and validation set in a 7:2:1 ratio. Linear interpolation is used to fill in missing time steps in the installation posture data of the sample set, and batch normalization is applied to standardize the data, forming enhanced standard samples. The Adam optimizer is used to dynamically adjust the parameters in the training set, cosine annealing is applied for learning rate scheduling, and early stopping is used to monitor the validation loss. Training is terminated if the validation loss does not decrease for 10 consecutive rounds. Focal Loss is used as the loss function to improve the accuracy of accident matching. Finally, when the validation set loss fluctuates less than the convergence threshold for 5 consecutive rounds, the digital neural pre-model is considered converged, and the trained digital neural pre-model is output.

[0034] It should also be noted that the accident matching accuracy is obtained by the degree of matching between the basic accident association feature set and the historical accident features; Inference latency is derived from the actual time taken by the digital neural pre-model to process historical accident features. The convergence threshold is defined based on the rate of change of the prediction error on the validation set, and its value ranges from 0.0001 to 0.01.

[0035] The installation attitude data and historical accident features are input into the digital neural pre-model. The attitude extraction layer extracts the fluctuation features through a gated recurrent network to generate a time-series feature vector. It should be noted that the attitude extraction layer performs time alignment and preprocessing on the installation attitude data and historical accident features: The installation attitude data is segmented into fixed time windows, and attitude numerical features such as displacement deviation, tilt angle shift, and average vibration frequency are extracted for each time step; simultaneously, discrete features such as operation instructions and safety warnings corresponding to the attitude window time in the historical accident features are encoded into accident discrete features; subsequently, the attitude numerical features and accident discrete features within the fixed time step are concatenated along the channel dimension to form an input matrix, which is then input into a gated recurrent network. The gated recurrent network dynamically captures the temporal correlation between attitude numerical features and accident discrete features through input gates, forget gates, and output gates. Finally, the hidden state of the last time step is extracted and mapped to a temporal feature vector.

[0036] It should also be noted that the hidden state is an internal information carrier maintained by the recurrent neural network when processing sequential data, integrating key information from historical steps with the current input; The specific steps of the gated recurrent network to capture the temporal correlation between attitude numerical features and accident discrete features are as follows: After receiving the concatenated input matrix, the input gate generates input gating coefficients through the sigmoid function to filter out key information important for correlation analysis (such as the combination of "sudden increase in tilt angle of 0.5° + emergency braking command" is retained, and stable low-frequency vibrations are suppressed); then, the forget gate generates forgetting gating coefficients based on the correlation patterns recorded in the previous time step, such as "no abnormal vibration of the equipment in the past 10 seconds" and the input matrix, and determines how much historical information to retain—if the correlation pattern is strongly correlated with the current input matrix (such as similar vibrations before past accidents), more correlation patterns are retained; otherwise, the interference of redundant historical information is reduced; next, the key information selected by the input gate and the historical information retained by the forget gate are weighted and fused to generate the current cell state, completing the fusion of historical information and current key information; finally, the output gate generates output gating coefficients through the sigmoid function, which are applied to the current cell state to generate the hidden state of the current time step, representing the temporal correlation between the current attitude numerical features and accident discrete features.

[0037] The accident matching layer uses a self-attention mechanism to mine causal patterns and generate an accident association feature matrix. It should be noted that the temporal feature vector output by the attitude extraction layer is linearly transformed to generate the Q (query) matrix, K (key) matrix, and V (value) matrix: The temporal feature vector output by the attitude extraction layer is input into three independent linear layers, and Q-space mapping, K-space mapping, and V-space mapping are performed respectively to generate the corresponding Q matrix, K matrix, and V matrix. The Q matrix is ​​used to capture the "query requirement" of the current time step (such as "whether the current vibration anomaly is related to historical accidents"), the K matrix is ​​used to encode the "key information" of each time step (such as "vibration pattern characteristics before historical accidents"), and the V matrix is ​​used to store the "value content" of each time step (such as "tilt offset associated with the accident"). Next, the correlation scores between each time step are obtained by the dot product of the Q matrix and the K matrix. Then, the correlation scores are normalized by the softmax function to generate the attention weight matrix. Then, the V matrix is ​​weighted and aggregated using causal attention weights: the attention weight matrix and the V matrix are summed element-wise to obtain the context-aware features of each time step. Finally, the context-aware features are mapped to the target dimension through a non-linear activation function and a fully connected layer to generate the accident correlation feature matrix.

[0038] The time-series feature vector and the accident correlation feature matrix are integrated through feature channels to generate a safety risk heat map; It should be noted that in the feature channel, a timestamp interpolation algorithm is used to align the time-series feature vector with the accident-related feature matrix in the time dimension to eliminate differences in time steps. Then, feature fusion is performed through a multi-head attention gating layer: the aligned time-series feature vector uses a one-dimensional convolutional kernel with a sliding window covering the current and adjacent time steps to capture vibration amplitude fluctuations in continuous time steps and output local perceptual features. Simultaneously, the sliding window method is used to segment the time-series data contained in the accident-related feature matrix at fixed time intervals, forming a continuous sequence of feature segments. Subsequently, dynamic change features of displacement, tilt angle, and vibration frequency parameters are extracted within each feature segment sequence. The similarity scores of the dynamic change features at different time intervals are obtained and sorted. Based on the similarity score sorting results, the dynamic change features at different time intervals are aggregated to extract the long-range dependency of the accident-related feature matrix. Then, the local perceptual features of the time-series feature vector and the long-range dependency of the accident-related feature matrix are weighted and fused through a feature splicing channel to generate a multi-dimensional feature tensor. Finally, the multi-dimensional feature tensor is used for feature enhancement and spatial mapping using a non-linear activation function to generate a safety risk heatmap.

[0039] Extract scene structure coordinates from equipment monitoring records and match them with the spatial coordinates of the safety risk heat map to generate a spatial mapping table; It should be noted that the spatial coordinates of the scene structure are parsed from the spatial coordinates of the installation logs, 3D scan point clouds, and sensor position records recorded by the equipment monitoring system. The 3D coordinates, dimensions, orientation, and other geometric features of equipment components such as transmission devices and sensor interfaces are extracted. At the same time, noise data is filtered and the spatial coordinates are calibrated to a spatial reference system consistent with the scene structure coordinates. Subsequently, the safety risk heatmap is spatially parsed: the spatial coordinate information of the risk distribution in the safety risk heatmap is extracted and mapped to the spatial reference system through affine transformation to ensure a unified spatial benchmark. Next, spatial coordinate matching is performed: the coordinate points of the scene structure are matched with the coordinate points of the high-risk areas in the safety risk heatmap. The risk intensity value of the safety risk heatmap is mapped to the coordinate points of each scene structure using an interpolation algorithm to obtain the matching structure coordinates. Finally, the matching structure coordinates and the corresponding risk intensity value, risk type, and other information are organized into a structured table to generate a spatial mapping table containing the fields "structure coordinates-risk value-risk type".

[0040] It should also be noted that the risk intensity value of the safety risk heatmap is directly determined by the output value of the multidimensional feature tensor after processing by a nonlinear activation function; The high-risk areas in the safety risk heatmap are derived from continuous areas in the safety risk heatmap where the risk intensity value is continuously set with a risk threshold. The risk threshold is defined based on the accident matching accuracy of accident association features in the historical accident database, with an exemplary value range of 0.7 to 0.9.

[0041] Spatial semantic segmentation and dynamic risk weight assignment are performed on the spatial mapping table to generate preset installation scenarios; It should be noted that the DBSCAN clustering algorithm is used to group all coordinate points in the spatial mapping table according to risk type (e.g., collision, tipping) and spatial proximity. Coordinate points of the same risk type with low spatial proximity (e.g., coordinate difference ≤ 0.5 meters) are aggregated into independent semantic regions, and the region type and the included coordinate range are labeled. Subsequently, a dynamic risk weight is assigned to each independent semantic region: the dynamic risk weight is determined by a combination of three data sources, including the number of accidents in the independent semantic region in the historical accident database (the more accidents, the higher the dynamic risk weight); the daily maintenance frequency of structural components in the installation log (the higher the maintenance frequency, the higher the dynamic risk weight of the independent semantic region); and the risk level of the independent semantic region in the spatial mapping table (the higher the risk level, the higher the dynamic risk weight). This ensures that the dynamic risk weight intuitively reflects the potential risks and protection priorities of the independent semantic region. Finally, the location information, type, and dynamic risk weight of the independent semantic regions are integrated to generate a preset installation scenario.

[0042] Based on the safety risk heat map, the structural type and dynamic risk weight of the corresponding location in the preset installation scenario are extracted, and the structural type and dynamic risk weight are quantified and integrated to generate a safety risk index. It should be noted that the positions of each coordinate point in the safety risk heatmap are mapped to a preset installation scenario, and the structural types and dynamic risk weights of the transmission devices, sensor interfaces, etc., in the area where each coordinate point is located are extracted. Subsequently, the structural types are quantified and assigned values: based on the functional attributes of the components in the installation log (e.g., the transmission device is a "core component," and the sensor interface is an "auxiliary component"), a basic value is assigned (e.g., a basic value of 5 for core components and 3 for auxiliary components); the product of the basic value and the dynamic risk weight is used as the initial risk value of the coordinate point; finally, the initial risk values ​​of all coordinate points are linearly scaled and normalized to generate a safety risk index for each coordinate point, which is then integrated into a safety risk index covering the entire scenario.

[0043] The risk warning module parses the voice command signal into natural language to generate command type, calculates the contradiction value between command type and security risk index, and activates the risk protection protocol of digital neural pre-simulation model and issues an alarm signal when the contradiction value exceeds the preset security threshold. Extract the spectral features of the safety officer's voice commands, generate a text string, and perform lexical analysis and semantic association on the text string to generate the command type; It should be noted that the safety officer's voice commands are processed in frames. Each frame is smoothed at the edges using a Hamming window and then converted into a frequency domain representation using a short-time Fourier transform. The Mel frequency cepstral coefficients, which reflect the characteristics of human hearing perception, are extracted as spectral features. Subsequently, the spectral features are used for speech recognition and converted into corresponding text strings through pattern matching. After obtaining the text strings, word segmentation is performed to divide the continuous text in the text strings into independent words. Then, part-of-speech tagging is performed to identify the grammatical roles of each independent word, such as verbs, nouns, adverbs, etc. Semantic association analysis is then performed to extract key action words (such as "lift," "stop," and "inspect"), safety status words (such as "loose," "in place," and "abnormal"), and adverbs of urgency (such as "immediately," "right away," and "attention"). Finally, the instruction type is determined based on the extracted keywords and their corresponding grammatical roles: if they contain strong warning words such as "stop," "evacuate," and "danger," they are determined to be emergency instructions; if they contain reminder words such as "attention," "careful," and "observe," they are determined to be safety warnings; if they begin with verbs such as "execute," "start," and "move," and are associated with specific operational actions, they are determined to be operational instructions; if they contain status descriptive words such as "confirm," "already," and "completed," they are determined to be status confirmation. Finally, a clear instruction type is generated based on the judgment results.

[0044] Perform operational risk matching and difference calculation on instruction type and security risk index to obtain contradictory values; The expression for calculating the contradiction value is: ; in, Indicates the contradictory value. Indicates the safety risk index. This represents the expected risk function for instruction type. Indicates the weighting coefficient; It should be noted that the weighting coefficient is based on the definition of risk sensitivity and is used to adjust the sensitivity of the contradictory value to the deviation of the actual risk. The exemplary value range is 0.5 to 2.

[0045] The instruction type risk expectation function is defined based on the correlation between instruction type and potential risk in historical experience.

[0046] Extract the highest safety threshold value that led to the occurrence of an accident from the historical accident database, and perform correlation correction on the highest safety threshold value to obtain the safety threshold; It should be noted that, firstly, monitoring parameters directly causing accidents (such as equipment temperature, pressure, displacement, etc.) are screened from the historical accident database. Valid data from monitoring parameters before and immediately after the accident is extracted through accident reports, equipment monitoring records, etc., while invalid and interfering data are filtered out to form an accident dataset. Secondly, based on association rule mining, the key causative parameters with the greatest impact on the occurrence of accidents are identified (e.g., "when the panel tilt angle exceeds 30°, 90% of accidents occur within 10 minutes"). Then, the maximum measured value of the identified key causative parameters in all accidents is statistically analyzed (e.g., the highest recorded panel tilt angle in historical accidents is 32°). Combined with the severity of the accident consequences, the highest safety threshold value for the key causative parameters causing the accident is determined. Subsequently, correlation correction is performed: based on the actual situation of the current scenario, such as the degree of equipment aging, changes in environmental conditions, maintenance status, etc., the correction factors affecting the highest safety threshold value are correlated through correlation analysis (e.g., if equipment aging reduces material strength by 15%, the threshold value needs to be reduced by 15%). Combined with correction experience from similar scenarios in historical accidents (e.g., the tilt angle threshold value needs to be reduced by an additional 3° in humid environments), the highest threshold value is adjusted to obtain a safety threshold suitable for the current scenario.

[0047] Real-time monitoring of the conflict value in the current operating area; if the conflict value exceeds the safety threshold, the risk protection protocol of the digital neural pre-simulation model is triggered and an alarm signal is issued. It should be noted that the steps for real-time monitoring of conflicting values ​​in the current operating area and triggering the protection process are as follows: First, the system collects the security risk index and the type of instruction being executed in the current operating area in real time, and calculates the contradiction value of the current operating area. Then, it compares the calculated contradiction value with the preset security threshold. If the contradiction value exceeds the security threshold, the risk protection protocol of the digital neural pre-simulation model is immediately triggered. After triggering, the digital neural pre-simulation model quickly executes protective actions: dynamically adjusting the current operating parameters, such as reducing the equipment operating speed, activating associated protective equipment, such as automatically reinforcing the support structure, and generating specific protection scheme suggestions (such as "the current contradiction value exceeds the security threshold, the panel adjustment must be stopped immediately and the support stability checked") and simultaneously issuing an alarm signal through the audible and visual alarm device.

[0048] The communication module transmits alarm signals and safety risk indices to the remote monitoring terminal.

[0049] The alarm signals and safety risk index are encapsulated and transmitted to the remote monitoring terminal using the MQTT protocol with encryption.

[0050] It should be noted that the transmission data is organized as follows: alarm signals and security risk indices are extracted from the current operating area and encapsulated in a structured manner, ensuring that the data fields are consistent with the parsing rules of the remote monitoring terminal. Subsequently, the encapsulated transmission data is encrypted using the TLS 1.3 protocol, generating encrypted binary data packets to ensure that the data content cannot be tampered with or stolen during transmission. Next, an MQTT connection is established: local MQTT client parameters are configured, a heartbeat mechanism is used to maintain a long connection, and an automatic reconnection policy is set to ensure the stability of the communication link. Finally, the encrypted data is published to the remote terminal: the encrypted binary data packets are sent to a predefined topic via MQTT PUBLISH messages, and a quality of service level is set to ensure that the message is delivered at least once. After receiving the topic, the remote monitoring terminal decrypts and parses the encrypted binary data packets, completing the remote transmission and reception of alarm signals and security risk indices.

[0051] It should also be noted that the steps for configuring local MQTT client parameters and maintaining a stable communication link are as follows: By configuring the connection location and port of the MQTT broker, a unique client ID and authentication information are generated, completing the basic configuration of the client parameters and ensuring that the client can legally connect to the broker. Next, a heartbeat mechanism is enabled: by setting connection parameters, the local MQTT client periodically sends heartbeat packets to the broker, and the broker returns a response upon receiving them. If the local MQTT client does not receive a response within the response time specified in the connection parameters, a connection error is determined, triggering disconnection. Finally, an automatic reconnection strategy is set: configuring the initial reconnection wait time (e.g., 1 second), the maximum wait time (e.g., 30 seconds), and the incrementing rule (e.g., doubling the wait time with each reconnection), and automatically restoring the subscribed topics after a successful reconnection, ensuring that the communication link can be quickly rebuilt after an interruption and maintaining the continuity of data transmission. The predefined topics are based on security risk classification, alarm type, and security risk index to distinguish different types of secure data transmission needs; The risk protection protocol of the digital neural pre-simulation model is an emergency response mechanism triggered when a conflict value exceeds a safety threshold in real time. It dynamically adjusts operating parameters, activates associated protection devices, and generates specific protection recommendations.

[0052] In summary, this invention, through a digital neural pre-simulation model, compares installation posture data with historical accident characteristics, achieving intelligent quantitative analysis and spatial visualization of safety risks. By integrating monitoring data and historical accident data, it reduces potential risks to construction personnel. The dynamic visualization of the safety risk heat map provides on-site and remote monitoring personnel with intuitive, global, and real-time situational awareness of the location, severity, and evolution trend of high-risk areas, enabling precise pre-emptive shifts in safety supervision. The output of the safety risk index provides a quantitative assessment benchmark, enhancing the scientific rigor and timeliness of safety risk warnings, providing crucial decision-making support for risk management, and significantly reducing the probability of major installation safety accidents.

[0053] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent monitoring system for safety protection during HPC curtain wall installation, characterized by: include, The data acquisition module collects real-time installation posture data of the HPC curtain wall and voice command signals from safety officers, and extracts historical accident features from the historical accident database. The installation posture data includes displacement, tilt angle, and vibration frequency parameters. The safety officer's voice command signals include operation instructions, safety warnings, status confirmations, and emergency instructions; The specific steps for extracting historical accident features are as follows: The association rule mining method is used to filter high-frequency features in the historical accident database to generate a basic accident association feature set. Perform causal chain verification and temporal interaction enhancement on the basic accident association feature set, and output historical accident features; The risk simulation module constructs a digital neural simulation model, compares installation posture data with historical accident characteristics through reinforcement learning algorithms, generates a safety risk heat map, and spatially overlays the safety risk heat map with a preset installation scenario to generate a safety risk index. The specific steps for constructing the digital neural pre-model are as follows. Call and initialize the convolutional neural network and the long short-term memory network, and build the pose extraction layer and the accident matching layer; A digital neural pre-model was constructed by optimizing the topology search and interaction mechanism of the attitude extraction layer and the accident matching layer using the NAS algorithm. The specific steps for generating the security risk heatmap are as follows: The installation attitude data and historical accident features are input into the digital neural pre-model. The attitude extraction layer extracts the fluctuation features through a gated recurrent network to generate a time-series feature vector. The accident matching layer uses a self-attention mechanism to mine causal patterns and generate an accident association feature matrix. The time-series feature vector and the accident correlation feature matrix are integrated through feature channels to generate a safety risk heat map; The risk warning module parses the voice command signal into natural language to generate command type, calculates the contradiction value between command type and security risk index, and activates the risk protection protocol of digital neural pre-simulation model and issues an alarm signal when the contradiction value exceeds the preset security threshold. The communication module transmits alarm signals and safety risk indices to the remote monitoring terminal.

2. The intelligent monitoring system for safety protection of HPC curtain wall installation as described in claim 1, characterized in that: The specific steps for obtaining the preset installation scenario are as follows. Extract scene structure coordinates from equipment monitoring records and match them with the spatial coordinates of the safety risk heat map to generate a spatial mapping table; Spatial semantic segmentation and dynamic risk weight assignment are performed on the spatial mapping table to generate preset installation scenarios.

3. The intelligent monitoring system for safety protection of HPC curtain wall installation as described in claim 2, characterized in that: The generation of the safety risk index refers to the process of extracting the structural types and dynamic risk weights in a preset installation scenario based on a safety risk heat map, and then quantifying and integrating the structural types and dynamic risk weights to generate the safety risk index.

4. The intelligent monitoring system for safety protection of HPC curtain wall installation as described in claim 3, characterized in that: The specific steps for obtaining the contradictory value are as follows. Extract the spectral features of the safety officer's voice commands, generate a text string, and perform lexical analysis and semantic association on the text string to generate the command type; Operational risk matching and difference calculation are performed on instruction types and security risk indices to obtain contradictory values.

5. The intelligent monitoring system for safety protection of HPC curtain wall installation as described in claim 4, characterized in that: The specific steps for issuing the alarm signal are as follows: Extract the highest safety threshold value that led to the occurrence of an accident from the historical accident database, and perform correlation correction on the highest safety threshold value to obtain the safety threshold; The system monitors the conflict values ​​in the current operating area in real time. If the conflict value exceeds the safety threshold, it triggers the risk protection protocol of the digital neural pre-simulation model and issues an alarm signal.

6. The intelligent monitoring system for safety protection of HPC curtain wall installation as described in claim 5, characterized in that: The transmission of alarm signals and security risk indices to remote monitoring terminals refers to encapsulating the alarm signals and security risk indices into data and transmitting them to remote monitoring terminals using the MQTT protocol for encryption.

Images