Method, system, device, and program for safety evaluation of a chemical storage site

Abstract

The application discloses a kind of safety evaluation method, system, equipment and procedure of chemical storage place, it is related to dangerous chemical management technical field, the method includes: to the space-time alignment of multi-source data;Wherein, the multi-source data includes storage chemical data, environmental data, equipment state data, management data and historical data;Extract the spatial feature of the multi-source data after space-time alignment, generate spatial risk weight chart;Pre-training risk evaluation model;The weight in the spatial risk weight chart is mapped to the risk evaluation model, and risk grade is obtained.Through the processing scheme of the present disclosure, potential risks of storage place can be found in time, early warning and emergency response are started, the possibility of accident occurrence is effectively reduced, and the safety of chemical storage place is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hazardous chemical management, and in particular to a safety evaluation method, system, equipment and program for chemical storage sites. Background Art

[0002] Explosive, radioactive, and highly toxic hazardous chemicals are extremely dangerous. If leaks or explosions occur during storage, they can cause immeasurable damage to human life and the ecological environment. Currently, safety assessments of chemical storage sites mostly rely on static evaluations, which primarily rely on regular manual inspections and simple data recording. These assessments are incomplete, lack real-time performance, and are difficult to effectively predict potential risks. Furthermore, the characteristics of different types of hazardous chemicals vary significantly, making it difficult for existing methods to comprehensively consider the risk factors of multiple chemicals and conduct accurate assessments.

[0003] Although the above static assessment method can perform security assessment to a certain extent, it is found that there are still some shortcomings in its actual use, which cannot achieve the best effect. Its shortcomings can be summarized as follows:

[0004] 1. Relying on manual inspections and static assessment models, it is impossible to capture dynamic risks (such as sudden temperature rise, radiation leakage, container corrosion, etc.) in real time.

[0005] 2. There is a lack of systematic analysis tools for the chain reactions caused by vibration / friction of explosive materials and the mixed storage of radioactive materials and highly toxic substances.

[0006] 3. Accident scenario simulation relies on historical data and is difficult to cope with complex environmental variables (such as extreme weather and human errors).

[0007] Therefore, the above-mentioned existing static assessment methods still have inconveniences and defects in use and are in urgent need of further improvement. How to create a new safety assessment method for chemical storage sites has become a goal that the industry urgently needs to improve. Summary of the Invention

[0008] In view of this, an embodiment of the present disclosure provides a safety assessment method for a chemical storage site, which at least partially solves the problems existing in the prior art.

[0009] In a first aspect, an embodiment of the present disclosure provides a method for safety assessment of a chemical storage site, the method comprising the following steps:

[0010] Performing spatiotemporal alignment on multi-source data, wherein the multi-source data includes stored chemical data, environmental data, equipment status data, management data, and historical data;

[0011] Extract spatial features of multi-source data after spatiotemporal alignment and generate spatial risk weight maps;

[0012] Pre-trained risk assessment model;

[0013] The weights in the spatial risk weight map are mapped to the risk assessment model to obtain the risk level.

[0014] According to a specific implementation of the embodiment of the present disclosure, the environmental data includes: internal environmental data and external environmental data; wherein the internal environmental data includes three-dimensional point cloud data of the warehouse building structure; the external environmental data includes wind speed, rainfall, earthquake waves, groundwater level, and surrounding traffic flow;

[0015] The device status data includes: temperature sensor data, humidity sensor data, gas concentration sensor data and vibration sensor data;

[0016] The management data includes: chemical in-and-out information, storage quantity, storage location, shelf life, management personnel operation records, and safety training records.

[0017] According to a specific implementation of an embodiment of the present disclosure, the spatiotemporal alignment of multi-source data includes:

[0018] Generate a BIM model of the building structure based on the 3D point cloud data of the warehouse building structure;

[0019] Generate sensor time series data based on device status data;

[0020] Performing spatiotemporal coordinate alignment, including: mapping sensor positions to the three-dimensional coordinates of the BIM model;

[0021] Perform time window alignment, including: windowing the sensor data series into preset time periods and aligning them with the static data in the BIM model; and updating the external environment data at the minute level and interpolating them to the sensor timestamp.

[0022] According to a specific implementation of the embodiment of the present disclosure, extracting spatial features of the multi-source data after time-space alignment to generate a spatial risk weight map includes:

[0023] Extract spatial features of BIM models based on GCN;

[0024] Extract time series features from temperature sensor data, humidity sensor data, and gas concentration sensor data based on long short-term memory networks;

[0025] The environment features of the extracted environment data include: calculating a radial component of wind speed; feature extraction on vibration sensor data based on wavelet transform; Fourier transform on seismic wave signals to extract main frequency energy spectrum;

[0026] The BIM spatial features, sensor time sequence features and environment features are mapped to a shared hidden space, and are compressed to unified dimension features through a full connection layer;

[0027] The weight coefficients of the spatial features, the time sequence features and the environment features are generated based on a gating mechanism;

[0028] The spatial risk weight map is output through the deep learning model.

[0029] According to a specific implementation manner of the embodiment of the present disclosure, the weight coefficients of the spatial features, the time sequence features and the environment features are generated based on a gating mechanism, and the method comprises the following steps of:

[0030] The gating mechanism dynamically allocates the weight coefficients based on the following formula:

[0031] ;

[0032] Wherein, is a weighted sum of the spatial features, the time sequence features and the environment features; is the weight coefficient generated by the gating, and satisfies ; is the spatial feature; is the time sequence feature; is the environment feature;

[0033] The spatial feature weight is obtained based on the following formula :

[0034] ;

[0035] Wherein, is a learnable weight matrix of the spatial feature weight ; is the spatial feature; is an environment parameter;

[0036] The time sequence feature weight is obtained based on the following formula :

[0037] ;

[0038] Wherein, is a learnable weight matrix of the time sequence feature weight ; is the time sequence feature; is an environment parameter encoder; is a feature splicing operation;

[0039] The environmental feature weight is obtained based on the following formula :

[0040] ;

[0041] wherein, is the environmental feature weight is a learnable weight matrix of the environmental feature weight is a tensor product operation; is a graph neural network.

[0042] According to a specific implementation manner of the embodiments of the present disclosure, the risk evaluation model is a CNN model;

[0043] The pre-training risk evaluation model comprises the following steps:

[0044] obtaining historical data; the historical data comprises historical storage chemical data, historical environmental data, historical equipment state data, historical management data and historical safety evaluation data;

[0045] normalizing the historical data, so that the input value is scaled to the interval [-1, 1] or [0, 1];

[0046] dividing the normalized historical data into a training set, a validation set and a test set according to a ratio of 7:2:1;

[0047] training the risk evaluation model based on the training set;

[0048] verifying the trained risk evaluation model based on the validation set, and terminating the training when the validation set loss does not decrease for 5 consecutive rounds;

[0049] testing the verified risk evaluation model based on the test set; wherein, when the test result meets a preset threshold, the training is stopped; when the test result does not meet the preset threshold, the hyperparameters of the risk evaluation model are optimized based on Bayesian optimization, and the optimized risk evaluation model is retrained, verified and tested until the test result meets the preset threshold.

[0050] In a second aspect, the embodiments of the present disclosure provide a safety evaluation system for a chemical storage site, the system comprising:

[0051] a data processing module configured to perform spatio-temporal alignment on multi-source data; wherein the multi-source data comprises storage chemical data, environmental data, equipment state data, management data and historical data;

[0052] a weight module configured to extract spatial features of the spatio-temporally aligned multi-source data to generate a spatial risk weight map;

[0053] A model training module is configured to pre-train the risk assessment model;

[0054] The risk prediction module is configured to map the weights in the spatial risk weight map to the risk assessment model to obtain a risk level.

[0055] According to a specific implementation of the embodiment of the present disclosure, the system further includes:

[0056] The spatiotemporal alignment module is configured to generate a BIM model of the building structure based on the three-dimensional point cloud data of the warehouse building structure; generate a sensor time series data sequence based on the equipment status data; perform spatiotemporal coordinate alignment, including: mapping the sensor position to the three-dimensional coordinates of the BIM model; perform time window alignment, including: windowing the sensor data sequence according to a preset time period and aligning it with the static data in the BIM model; and update the external environment data at the minute level and interpolate it to the sensor timestamp.

[0057] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0058] at least one processor; and,

[0059] a memory communicatively connected to the at least one processor; wherein,

[0060] The memory stores instructions that can be executed by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor implements the safety assessment method for chemical storage sites as described in any one of the first aspect or any one of the implementations of the first aspect.

[0061] In a fourth aspect, an embodiment of the present disclosure further provides a non-transitory computer-readable storage medium, which stores computer instructions, and when the computer instructions are executed by at least one processor, the at least one processor executes the safety assessment method for chemical storage sites in the aforementioned first aspect or any implementation of the first aspect.

[0062] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, comprising a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program comprises program instructions, which, when executed by a computer, cause the computer to execute the safety assessment method for chemical storage sites in the aforementioned first aspect or any implementation of the first aspect.

[0063] The safety evaluation method for chemical storage sites in the disclosed embodiment realizes a comprehensive, real-time and accurate evaluation of the safety status of storage sites for explosive-making hazardous chemicals, radioactive hazardous chemicals and highly toxic hazardous chemicals by integrating multi-source data and a weighted dynamic intelligent algorithm, discovers potential risks in advance and issues timely warnings, and effectively reduces the probability of accidents. The multi-source data integrated by the present invention covers multiple dimensions such as environment, equipment, and management, and comprehensively considers various risk factors in the storage process of explosive-making, radioactive sources and highly toxic hazardous chemicals. Compared with traditional static evaluation methods, the evaluation is more comprehensive and accurate. Through real-time data collection and dynamic evaluation, it is possible to timely discover potential risks in storage sites, issue early warnings and initiate emergency responses, effectively reduce the possibility of accidents, and improve the safety of chemical storage sites. The weight coefficients of spatial characteristics, temporal characteristics and environmental characteristics are generated based on the gating mechanism, so that the weight distribution is more scientific and reasonable, and the reliability of the evaluation results is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic flow chart of a safety assessment method for a chemical storage site provided in an embodiment of the present disclosure;

[0065] Figure 2 A flowchart of a safety assessment method for a chemical storage site provided in an embodiment of the present disclosure;

[0066] Figure 3 A flowchart of a method for performing spatiotemporal alignment of multi-source data provided in an embodiment of the present disclosure;

[0067] Figure 4 A schematic diagram of a safety assessment system for chemical storage sites provided in an embodiment of the present disclosure;

[0068] Figure 5 A schematic diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0069] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0070] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present disclosure.

[0071] It should be noted that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or functionality described herein is illustrative only. Based on this disclosure, those skilled in the art will appreciate that one aspect described herein may be implemented independently of any other aspect, and that two or more of these aspects may be combined in various ways. In addition, other structures and / or functionality other than one or more of the aspects described herein may be used to implement this apparatus and / or practice this method.

[0072] Additionally, in the following description, specific details are provided to provide a thorough understanding of the examples. However, one skilled in the art will appreciate that the aspects described may be practiced without these specific details.

[0073] Figure 1 A schematic diagram of the process of a safety assessment method for a chemical storage site provided in an embodiment of the present disclosure.

[0074] Figure 2 For Figure 1 Corresponding flow chart of the safety assessment method for chemical storage sites.

[0075] like Figure 1 As shown, at step S110, the multi-source data is temporally and spatially aligned; wherein the multi-source data includes stored chemical data, environmental data, equipment status data, management data and historical data.

[0076] More specifically, required data is acquired, including stored chemical data, environmental data, equipment status data, management data, and historical data.

[0077] In an embodiment of the present invention, the environmental data includes: internal environmental data and external environmental data; wherein the internal environmental data includes three-dimensional point cloud data of the warehouse building structure; the external environmental data includes wind speed, rainfall, earthquake waves, groundwater level, and surrounding traffic flow;

[0078] The device state data includes temperature sensor data, humidity sensor data, gas concentration sensor data, and vibration sensor data; and the device state data further includes material data of the storage container, storage container structure data, storage container sealing data, building structure data, fire-fighting equipment data, and leakage detection data.

[0079] The management data includes warehouse-in and warehouse-out information of the chemicals, storage quantity, storage location, shelf life, management personnel operation record, safety training record, safety inspection data, and emergency disposal plan data.

[0080] More specifically, in the external environment data, meteorological data is accessed through an API interface to an authoritative meteorological platform (such as the China Meteorological Administration) to obtain real-time meteorological data (temperature and humidity, wind speed, rainfall, thunderstorm probability, etc.) and disaster warnings (typhoon, earthquake, flood); geological data is collected in real time by deploying vibration sensors and geological monitoring equipment to collect warehouse surrounding seismic wave, soil settlement, and underground water level change data.

[0081] The internal environment data includes real-time temperature and humidity, air pressure, ventilation conditions, illumination data, gas concentration data, radiation data, and surrounding building structure (BIM model, Building Information Modeling), etc. The internal sensor network (temperature and humidity, air pressure, vibration) in the warehouse is linked and calibrated with external data to eliminate environmental interference (such as sensor errors caused by temperature and humidity gradient).

[0082] Through the input storage chemical type information, the chemical properties and storage condition data are obtained, for example: friction sensitivity of primary explosive (such as ammonium nitrate), half-life of radioactive source (such as iridium-192), volatility threshold of toxic substance (such as sodium cyanide), and factors that affect the safety and stability of the storage chemicals, such as the hygroscopicity of ammonium nitrate, which affects the safety and stability of ammonium nitrate, including humidity, temperature, thunderstorm weather, and the sealing property of the storage container.

[0083] The device state data includes material data of the storage container, storage container structure data, storage container sealing data, building structure data, fire-fighting equipment data, and leakage detection data.

[0084] The historical accident data includes diffusion range, influence time, and data at the time of the accident of similar chemical leakage accidents.

[0085] More specifically, the above data is preprocessed, including interpolation processing of missing values; wavelet transform (Daubechies wavelet) is used to remove high-frequency noise and retain key risk signals.

[0086] For example, the historical accident data includes diffusion range, influence time, and data at the time of the accident of similar chemical leakage accidents. Figure 3As shown, in the embodiment of the present application, the spatio-temporal alignment of the multi-source data comprises: generating a BIM model of the building structure based on the three-dimensional point cloud data of the warehouse building structure; generating a sensor time series data sequence based on the equipment state data; performing spatio-temporal coordinate alignment, comprising: mapping the sensor position into the three-dimensional coordinates of the BIM model; performing time window alignment, comprising: window aggregation of the sensor data sequence according to a preset time period, aligning with the static data in the BIM model; and updating the external environment data at a minute level, and interpolating to the sensor timestamp.

[0087] More specifically, the spatio-temporal coordinate alignment by mapping the sensor position into the three-dimensional coordinates of the BIM model comprises:

[0088] Firstly, the physical coordinates of the sensor and the type of the sensor are obtained, and then the sensor timestamp is synchronized with the BIM model time axis based on the NTP protocol (Network Time Protocol), and is converted into the BIM model coordinate system through a coordinate conversion tool (such as Proj4, GDAL).

[0089] Then, spatial alignment is performed, specifically including the following steps:

[0090] 1. Point cloud matching:

[0091] The sensor coordinates are matched with the point cloud of the BIM model through the ICP algorithm (Iterative Closest Point).

[0092] 2. Feature matching:

[0093] The nearest neighbor matching is performed between the key geometric features (such as wall edges, equipment contours) in the BIM model and the sensor position.

[0094] 3. Manual calibration:

[0095] The sensor icon is manually placed in the BIM software (such as Revit), and the position accuracy is verified through view sectioning.

[0096] 4. Error analysis:

[0097] The Euclidean distance between the sensor coordinates and the corresponding points of the model is calculated, the average error is counted, when the error is greater than a preset threshold, the key points (such as beam column nodes) are selected for field measurement, the position deviation between the model and the sensor is compared, and the space is matched again.

[0098] More specifically, next turn to step S120.

[0099] At step S120, the spatial features of the spatio-temporally aligned multi-source data are extracted, and a spatial risk weight map is generated.

[0100] In the embodiments of the present application, the space feature of the extracted spatio-temporal aligned multi-source data is generated to generate a space risk weight map, including: performing space feature extraction on the BIM model based on GCN (Graph Convolutional Network); performing time series feature extraction on the temperature sensor data, humidity sensor data and gas concentration sensor data based on the long short-term memory network; extracting environmental features of environmental data, including: calculating the radial component of wind speed; performing feature extraction on the vibration sensor data based on wavelet transform; performing Fourier transform on the seismic wave signal to extract the main frequency energy spectrum; mapping the BIM space features, sensor time series features and environmental features to a shared hidden space, and compressing them to a unified dimension feature through a fully connected layer; generating weight coefficients of the space features, time series features and environmental features based on a gating mechanism; and outputting a space risk weight map through a deep learning model.

[0101] More specifically, the space feature extraction on the BIM model based on GCN includes the following steps:

[0102] Based on the BIM model, relevant data including geometric information (such as coordinates, dimensions, shapes, etc.), topological information (such as connection relationship between components, adjacency relationship, etc.) and semantic information (such as type and function of components, etc.) of building components are extracted, and these data are organized and preprocessed to ensure that the data are on the same scale, facilitating subsequent calculation.

[0103] Each building component in the BIM model is regarded as a node in the graph. The edges in the graph are defined according to the actual connection relationship or spatial adjacency relationship between components. If two components are directly connected in space or meet certain adjacency conditions (such as distance less than a certain threshold), an edge is established between their corresponding nodes, thus constructing a graph that can reflect the spatial structure of the BIM model.

[0104] The graph reflecting the spatial structure of the BIM model is input into the trained GCN model for space feature extraction. The GCN model outputs the spatial feature representation of each node (building component) and analyzes and visualizes the extracted spatial features.

[0105] The time series feature extraction on the temperature sensor data, humidity sensor data and gas concentration sensor data based on the long short-term memory network includes:

[0106] Data are collected from temperature sensors, humidity sensors and gas concentration sensors. These sensors usually sample data at a predetermined time interval, for example, once every minute or hour.

[0107] Check if there are missing values in the dataset; if so, choose to fill them with mean, median or time series-based interpolation method according to specific circumstances.

[0108] Normalize the data collected by different sensors to map them to a specific interval such as [0, 1] or [-1, 1], to eliminate the dimensional differences between different data dimensions and speed up the convergence of the model.

[0109] Input the preprocessed data into the trained LSTM model (Long Short-Term Memory network) for feature extraction, input the time series characteristics of the data, and analyze and visualize the extracted features. For example, you can use principal component analysis (PCA) to map high-dimensional features to two or three-dimensional space, observe the distribution of different sensor data in the feature space, and their trend over time. You can also use clustering analysis to cluster data points with similar features to discover potential patterns and rules in the data. In addition, the extracted features can be used for subsequent tasks such as anomaly detection, prediction analysis, etc., to further verify the effectiveness of the features by evaluating the performance of the model on these tasks.

[0110] Map BIM spatial features, sensor time series features and environmental features to a shared hidden space, compress them to a unified dimension feature through a fully connected layer, including the following steps:

[0111] Map the reduced BIM features to the shared hidden space using a fully connected layer:

[0112]

[0113] where, is the shared hidden space feature vector after mapping the BIM model; is the weight matrix applied to the BIM learnable; is the spatial feature of the BIM model; is the bias term to enhance the expression ability of the model.

[0114] Output dimension where, represents a d-dimensional real vector space, where d can be 128.

[0115] In the sensor feature embedding process, the time series feature Apply a multi-layer perceptron (MLP):

[0116]

[0117] wherein, is the mapped shared hidden space feature vector of the sensor; MLP is a multi-layer perceptron used for non-linear mapping of the sensor time-series feature; is the sensor time-series feature vector

[0118] An embedding layer or a shallow network is applied to the normalized environment feature.

[0119] Multi-Head Attention is used to calculate the inter-modal correlation weight, dynamically adjusting the feature importance.

[0120] Query: the mapped shared hidden space feature vector of the BIM .

[0121] Key: the mapped shared hidden space feature vector of the sensor .

[0122] Value: the mapped shared hidden space feature vector of the environment .

[0123] Output: the weighted fused feature .

[0124] The fused feature is compressed in dimension step by step (such as d→64→32), and each layer is followed by ReLU activation and Dropout (such as 0.3).

[0125] Output the unified dimension feature

[0126] .

[0127] In the embodiments of the present application, the weight coefficients of the spatial feature, the time-series feature and the environment feature generated based on the gating mechanism include:

[0128] The gating mechanism dynamically allocates the weight coefficients based on the following formula:

[0129] ;

[0130] wherein, is the weighted sum of the spatial feature, the time-series feature and the environment feature; is the weight coefficient generated by the gating, satisfying ; is the spatial feature; is the time-series feature; is the environment feature;

[0131] ​The spatial feature weight is obtained based on the following formula :

[0132] ;

[0133] wherein, is a learnable weight matrix of the spatial feature weight ; is a spatial feature; is an environment parameter, including a set of internal and external environment parameters;

[0134] The temporal feature weight is obtained based on the following formula :

[0135] ;

[0136] wherein, is a learnable weight matrix of the temporal feature weight ; is a temporal feature; is an environment parameter encoder; is a feature splicing operation;

[0137] The environment feature weight is obtained based on the following formula :

[0138] ;

[0139] wherein, is a learnable weight matrix of the environment feature weight ; is an environment feature; is a tensor product operation; is a Graph Neural Network.

[0140] Next, go to step S130.

[0141] At step S130, the pre-training risk evaluation model.

[0142] In the embodiment of the present application, the risk assessment model is a CNN model; the pre-training risk assessment model comprises the following steps: obtaining historical data; the historical data comprises historical storage chemical data, historical environmental data, historical equipment state data, historical management data and historical safety assessment data; normalizing the historical data to scale the input value to the interval [-1, 1] or [0, 1]; dividing the normalized historical data into a training set, a validation set and a test set according to a 7:2:1 ratio; training the risk assessment model based on the training set; verifying the trained risk assessment model based on the validation set, and terminating the training when the validation set loss does not decrease for 5 consecutive epochs; testing the verified risk assessment model based on the test set; wherein, when the test result meets the preset threshold, the training is stopped; when the test result does not meet the preset threshold, the hyperparameters of the risk assessment model are optimized based on Bayesian optimization, and the optimized risk assessment model is retrained, verified and tested until the test result meets the preset threshold.

[0143] Next, go to step S140.

[0144] At step S140, map the weights in the spatial risk weight map to the risk assessment model to obtain a risk level.

[0145] In the embodiment of the present application, the method further comprises: presetting a linkage response strategy; the linkage response strategy can divide the linkage strategy according to the risk level or the application scenario.

[0146] For example:

[0147] I. Linkage strategy divided according to risk level

[0148] 1. When low risk is prompted, and only temperature and humidity have fluctuations in time sequence characteristics, and the fluctuations are less than 5%, the system records risk data, generates a daily report, and notifies the safety officer to conduct routine inspection.

[0149] When low risk is prompted, and the activity concentration of radioactive substances is less than 0.1 μSv / h, or the leakage amount of toxic gas is less than 10 g / min (or the threshold is set according to the type of toxic chemicals), the local exhaust system is started to dilute the concentration of toxic gas. A drone is called to patrol and verify the location of the leakage point, and the on-site personnel are notified to wear protective devices and increase the monitoring frequency to once every 15 minutes.

[0150] 2. When medium risk is prompted, and the gas concentration rises at a rate greater than 10 ppm / h, start the local exhaust system (such as increasing the opening degree of the ventilation port by 30%), and simulate the risk evolution path, and notify the emergency team to be on standby.

[0151] When the risk is prompted, the radioactive activity is greater than or equal to 0.1 μSv / h, and less than 1 mSv / h, or the toxic gas leakage rate is greater than or equal to 10 g / min, less than 100 g / min, the adjacent ventilation duct is closed, the iodine piece distribution device is started (for radioactive iodine leakage), the robot is called to enter the contaminated area to collect samples, the warning area radius is determined according to the real-time wind speed, irrelevant personnel are prohibited from entering, and the emergency team wears anti-chemical clothing to enter the scene.

[0152] 3. When the high risk is prompted, and the environmental parameter mutates (such as wind speed greater than 10 m / s, rainfall greater than 20 mm / h, etc.), the whole area spraying system (such as ammonium nitrate storage area) is triggered, the power supply of the adjacent shelf is cut off, the evacuation path is generated, the emergency broadcast is started, and the personnel are evacuated to the safety area.

[0153] When the high risk is prompted, the radioactive activity is greater than 1 mSv / h, or the toxic substance diffuses to the population-dense area, the concentration of volatile toxic substances (such as hydrogen cyanide) reaches the IDLH (immediate threat to life and health) threshold, the whole storage place spraying dilution (for acidic toxic substances) or inert gas covering (such as nitrogen to inhibit cyanide combustion) is triggered, the emergency broadcast is started, the personnel are guided to evacuate along the wind direction, the fire brigade wears the positive pressure respirator (SCBA) to block the scene, the medical team prepares the decontamination and cleaning station, and the environmental protection department starts the regional radiation monitoring network.

[0154] 4. When the extreme risk is prompted, and there is a risk of chain reaction (such as dust explosion, toxic gas diffusion), the highest level of emergency protocol is started, the fire fighting system is linked, and the government emergency department is notified, the personnel of the whole factory are forced to evacuate, and the surrounding area is blocked.

[0155] II. Linkage strategy according to application scene

[0156] 1. When the dangerous chemical leakage is detected, the leakage source is located according to the spatial weight map, and the related pipeline valve is automatically closed; when the wind speed is greater than 5 m / s, the mobile fog cannon car is started to dilute the gas. When there are few personnel at night, the sound and light alarm is preferentially triggered to avoid panic.

[0157] 2. When the fire is detected, the fire spreading path (such as ventilation duct combustion support) is analyzed based on the BIM model, the opening and closing state of the fireproof roller shutter door is dynamically adjusted, and the fire source is isolated; the fire water cannon is automatically started and aimed at the high temperature area.

[0158] 3. When natural disasters are detected, such as earthquake wave intensity greater than 6, emergency lighting and evacuation passage indication are triggered; when the underground water level rises to the warning line, the underground warehouse waterproof gate is automatically closed; when thunderstorm warning is given, the warehouse ventilation system is automatically closed, and the nitrogen inerting device (to prevent oxidation of explosive materials) is started; when earthquake warning is given, the shelf locking mechanism is triggered, and the AGV transportation robot operation is stopped.

[0159] The above examples are only embodiments of the present application and do not limit the present application, and different linkage strategies can be preset according to different storage chemicals.

[0160] The safety evaluation method of the chemical storage site provided by the present application integrates multi-source data, covers multiple dimensions such as environment, equipment, and management, comprehensively considers various risk factors in the storage process of explosive, radioactive source, and toxic dangerous chemicals, and is more comprehensive and accurate compared with the traditional static evaluation method. Moreover, through real-time data collection and dynamic evaluation, potential risks of the storage site can be found in time, early warning and emergency response can be started, the possibility of accidents can be effectively reduced, the safety of the chemical storage site can be improved, and the reliability of the evaluation result can be improved.

[0161] Figure 4 The safety evaluation system 400 of the chemical storage site provided by the present application is shown, which includes a data processing module 410, a weight module 420, a model training module 430, and a risk prediction module 440.

[0162] The data processing module 410 is configured to perform spatio-temporal alignment on the multi-source data; wherein the multi-source data includes storage chemical data, environment data, equipment state data, management data, and historical data.

[0163] The weight module 420 is configured to extract spatial features of the spatio-temporally aligned multi-source data to generate a spatial risk weight map.

[0164] The model training module 430 is configured to pre-train a risk evaluation model.

[0165] The risk prediction module 440 is configured to map the weights in the spatial risk weight map to the risk evaluation model to obtain a risk level.

[0166] Referring to Figure 5 The present disclosure also provides an electronic device 50, which includes:

[0167] at least one processor; and

[0168] a memory connected in communication with the at least one processor; wherein

[0169] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the safety evaluation method of the chemical storage site in the foregoing method embodiments.

[0170] The present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the safety evaluation method of the chemical storage site in the foregoing method embodiments.

[0171] An embodiment of the present disclosure also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer, enable the computer to execute the safety assessment method for chemical storage sites in the aforementioned method embodiment.

[0172] Reference below Figure 5 , which shows a schematic structural diagram of an electronic device 50 suitable for implementing an embodiment of the present disclosure. The electronic devices in the embodiments of the present disclosure may include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0173] like Figure 5 As shown, electronic device 50 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 502 or programs loaded from a storage device 508 into a random access memory (RAM) 503. RAM 503 also stores various programs and data required for the operation of electronic device 50. Processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to bus 504.

[0174] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 508 including, for example, a magnetic tape, hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device 50 to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows the electronic device 50 with various devices, it should be understood that not all of the devices shown are required to be implemented or present. More or fewer devices may alternatively be implemented or present.

[0175] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via the communication device 509, or installed from the storage device 508, or installed from the ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are performed.

[0176] It should be noted that the computer-readable medium described above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable storage media may include, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wire, optical cable, RF (radio frequency), or any suitable combination thereof.

[0177] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0178] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains at least two Internet Protocol addresses; sends a node evaluation request including the at least two Internet Protocol addresses to a node evaluation device, wherein the node evaluation device selects an Internet Protocol address from the at least two Internet Protocol addresses and returns it; receives the Internet Protocol address returned by the node evaluation device; wherein the obtained Internet Protocol address indicates an edge node in a content distribution network.

[0179] Alternatively, the computer-readable medium carries one or more programs, which, when executed by the electronic device, causes the electronic device to: receive a node evaluation request including at least two Internet Protocol addresses; select an Internet Protocol address from the at least two Internet Protocol addresses; and return the selected Internet Protocol address; wherein the received Internet Protocol address indicates an edge node in a content distribution network.

[0180] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0181] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0182] The units involved in the embodiments described in this disclosure may be implemented in software or hardware. In some cases, the name of a unit does not limit the unit itself. For example, the first acquisition unit may also be described as a "unit for acquiring at least two Internet Protocol addresses."

[0183] It should be understood that various parts of the present disclosure can be implemented in hardware, software, firmware, or a combination thereof.

[0184] The above description is only a specific embodiment of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with this technical field within the technical scope disclosed in this disclosure should be covered by the protection scope of the present disclosure.

Claims

1. A safety assessment method for chemical storage sites, characterized in that: The method comprises the following steps: Performing spatiotemporal alignment on multi-source data, wherein the multi-source data includes stored chemical data, environmental data, equipment status data, management data, and historical data; Extract spatial features of multi-source data after spatiotemporal alignment and generate spatial risk weight maps; Pre-trained risk assessment model; Mapping the weights in the spatial risk weight map to the risk assessment model to obtain a risk level; The environmental data includes: internal environmental data and external environmental data; wherein the internal environmental data includes three-dimensional point cloud data of the warehouse building structure; the external environmental data includes wind speed, rainfall, earthquake waves, groundwater level, and surrounding traffic flow; The device status data includes: temperature sensor data, humidity sensor data, gas concentration sensor data and vibration sensor data; The management data includes: chemical in-and-out information, storage quantity, storage location, shelf life, management personnel operation records, and safety training records; The step of extracting spatial features of the multi-source data after time-space alignment and generating a spatial risk weight map includes: Extract spatial features of BIM models based on GCN; Extract time series features from temperature sensor data, humidity sensor data, and gas concentration sensor data based on long short-term memory networks; Extract environmental features of environmental data, including: calculating the radial component of wind speed; extracting features from vibration sensor data based on wavelet transform; and extracting the main frequency energy spectrum from seismic wave signals through Fourier transform. Map BIM spatial features, sensor temporal features, and environmental features to a shared latent space and compress them into unified dimensional features through a fully connected layer; Generate weight coefficients of spatial features, temporal features, and environmental features based on the gating mechanism; Output spatial risk weight map through deep learning model; The weight coefficients of spatial features, temporal features, and environmental features generated based on the gating mechanism include: The gating mechanism dynamically assigns weight coefficients based on the following formula: Among them, F fused is the weighted sum of spatial features, temporal features, and environmental features; α, β, and γ are the weight coefficients generated by the gate, satisfying α+β+γ=1; F spatial is the spatial feature; F temporal is the time series feature; F env For environmental characteristics; The spatial feature weight α is obtained based on the following formula: α=Softmax(W α ×[GAT(F spatial ,E)]; Among them, W α is the learnable weight matrix of spatial feature weight α; F spatial is the spatial characteristic; E is the environmental parameter; The time series feature weight β is obtained based on the following formula: β=Sigmoid(W β ×[GRU(F temporal )||φ(E)]); Among them, W β is the learnable weight matrix of the temporal feature weight β; F temporal is the temporal feature; φ(E) is the environmental parameter encoder; || is the feature concatenation operation; The environmental feature weight γ is obtained based on the following formula: Among them, W γ is the learnable weight matrix of the environmental feature weight γ; F env For environmental characteristics; is a tensor product operation; GNN(·) is a graph neural network.

2. The safety assessment method for chemical storage sites according to claim 1, characterized in that: The spatiotemporal alignment of multi-source data includes: Generate a BIM model of the building structure based on the 3D point cloud data of the warehouse building structure; Generate sensor time series data based on device status data; Performing spatiotemporal coordinate alignment, including: mapping sensor positions to the three-dimensional coordinates of the BIM model; Perform time window alignment, including: windowing the sensor data series into preset time periods and aligning them with the static data in the BIM model; and updating the external environment data at the minute level and interpolating them to the sensor timestamp.

3. The safety assessment method for chemical storage sites according to claim 1, characterized in that: The risk assessment model is a CNN model; The pre-trained risk assessment model includes the following steps: Acquiring historical data; the historical data includes historical stored chemical data, historical environmental data, historical equipment status data, historical management data, and historical safety evaluation data; Normalize historical data so that the input value is scaled to the range of [-1, 1] or [0, 1]; The normalized historical data is divided into training set, validation set and test set in a ratio of 7:2:1; Training the risk assessment model based on the training set; The trained risk assessment model is verified based on the validation set. When the validation set loss does not decrease for five consecutive rounds, the training is terminated. The verified risk assessment model is tested based on the test set; when the test result meets the preset threshold, the training is stopped; when the test result does not meet the preset threshold, the hyperparameters of the risk assessment model are tuned based on Bayesian optimization, and the tuned risk assessment model is re-trained, verified, and tested until the test result meets the preset threshold.

4. A safety assessment system for chemical storage sites, characterized in that: The system comprises: a data processing module configured to perform spatiotemporal alignment on multi-source data, wherein the multi-source data includes stored chemical data, environmental data, equipment status data, management data, and historical data; The environmental data includes: internal environmental data and external environmental data; wherein the internal environmental data includes three-dimensional point cloud data of the warehouse building structure; the external environmental data includes wind speed, rainfall, earthquake waves, groundwater level, and surrounding traffic flow; The device status data includes: temperature sensor data, humidity sensor data, gas concentration sensor data and vibration sensor data; The management data includes: chemical in-and-out information, storage quantity, storage location, shelf life, management personnel operation records, and safety training records; The weight module is configured to extract spatial features of multi-source data after spatiotemporal alignment and generate a spatial risk weight map; The step of extracting spatial features of the multi-source data after time-space alignment and generating a spatial risk weight map includes: Extract spatial features of BIM models based on GCN; Extract time series features from temperature sensor data, humidity sensor data, and gas concentration sensor data based on long short-term memory networks; Extract environmental features of environmental data, including: calculating the radial component of wind speed; extracting features from vibration sensor data based on wavelet transform; and extracting the main frequency energy spectrum from seismic wave signals through Fourier transform. Map BIM spatial features, sensor temporal features, and environmental features to a shared latent space and compress them into unified dimensional features through a fully connected layer; Generate weight coefficients of spatial features, temporal features, and environmental features based on the gating mechanism; Output spatial risk weight map through deep learning model; The weight coefficients of spatial features, temporal features, and environmental features generated based on the gating mechanism include: The gating mechanism dynamically assigns weight coefficients based on the following formula: Among them, F fused is the weighted sum of spatial features, temporal features, and environmental features; α, β, and γ are the weight coefficients generated by the gate, satisfying α+β+γ=1; F spatial is the spatial feature; F temporal is the time series feature; F env For environmental characteristics; The spatial feature weight α is obtained based on the following formula: α=Softmax(W α ×[GAT(F spatial ,E)]; Among them, W α is the learnable weight matrix of spatial feature weight α; F spatial is the spatial characteristic; E is the environmental parameter; The time series feature weight β is obtained based on the following formula: β=Sigmoid(W β ×[GRU(F temporal )||φ(E)]); Among them, W β is the learnable weight matrix of the temporal feature weight β; F temporal is the temporal feature; φ(E) is the environmental parameter encoder; || is the feature concatenation operation; The environmental feature weight γ is obtained based on the following formula: Among them, W γ is the learnable weight matrix of the environmental feature weight γ; F env For environmental characteristics; is a tensor product operation; GNN(·) is a graph neural network; A model training module is configured to pre-train the risk assessment model; The risk prediction module is configured to map the weights in the spatial risk weight map to the risk assessment model to obtain a risk level.

5. The safety evaluation system for chemical storage sites according to claim 4, characterized in that: The system further comprises: The spatiotemporal alignment module is configured to generate a BIM model of the building structure based on the three-dimensional point cloud data of the warehouse building structure; generate a sensor time series data sequence based on the equipment status data; perform spatiotemporal coordinate alignment, including: mapping the sensor position to the three-dimensional coordinates of the BIM model; perform time window alignment, including: windowing the sensor data sequence according to a preset time period and aligning it with the static data in the BIM model; and update the external environment data at the minute level and interpolate it to the sensor timestamp.

6. An electronic device, characterized in that: The electronic device includes: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor is caused to perform the safety assessment method for chemical storage sites according to any one of claims 1 to 3.

7. A computer program product, characterized in that The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer is caused to execute the safety assessment method for chemical storage sites as described in any one of claims 1 to 3.

Images