Method for tracing dangerous chemical fire accidents based on RFID

Abstract

The application provides an RFID-based dangerous chemical fire accident tracing method, and belongs to the technical field of fire tracing. The method comprises the following steps: configuring an RFID tag for each dangerous chemical container; reading the information of each RFID tag based on a pre-deployed read-write network, and transmitting the information to a central data platform to establish a dynamic information database of dangerous chemicals; integrating residual tag information with the dynamic information database to trace the position sequence and state sequence of dangerous chemicals before the fire occurs; simultaneously, simulating the fire spreading path and the expansion trend of each path point on the spreading path according to the current fire situation and the residual tag information of dangerous chemicals; comprehensively determining the cause of the fire accident and generating a tracing accident report. The safety of dangerous chemical management is improved.

Description

Technical Field

[0001] This invention relates to the field of fire source tracing technology, and in particular to a method for tracing hazardous chemical fire accidents based on RFID. Background Technology

[0002] Hazardous chemicals pose a fire risk during production, storage, transportation, and use. In the event of a fire, accurately and quickly tracing the cause is crucial for subsequent liability determination, accident handling, and prevention of similar incidents. Traditional methods for tracing the cause of hazardous chemical fires primarily rely on on-site investigations, personnel interviews, and historical record reviews. These methods suffer from low efficiency, poor accuracy, and are highly susceptible to human factors, especially in complex fire scenes where it is difficult to quickly and accurately determine the source and progression of the accident.

[0003] With the development of information technology, RFID (Radio Frequency Identification) technology has been widely used in the field of item tracking and management due to its advantages such as non-contact identification, long-distance reading, simultaneous identification of multiple targets, and large information storage capacity. However, research and practice on applying RFID technology to the traceability of hazardous chemical fire accidents are still relatively limited, and a systematic and effective RFID-based method for tracing the source of hazardous chemical fire accidents is lacking.

[0004] Therefore, this invention proposes a method for tracing the source of hazardous chemical fire accidents based on RFID. Summary of the Invention

[0005] This invention provides an RFID-based method for tracing the source of hazardous chemical fire accidents, in order to solve the aforementioned technical problems.

[0006] This invention proposes an RFID-based method for tracing the source of hazardous chemical fire accidents, comprising:

[0007] Step 1: Configure each hazardous chemical container with an RFID tag, wherein the RFID tag contains basic information and the initial location;

[0008] Step 2: Based on the pre-deployed read / write network, read the information of each RFID tag and transmit it to the central data platform to establish a dynamic information database of hazardous chemicals, and record the external environmental factors, operation behavior and usage status of the hazardous chemicals at different time points;

[0009] Step 3: After a fire involving hazardous chemicals, use a portable RFID reader to read the RFID tags present at the fire scene and obtain information about the remaining tags;

[0010] Step 4: Integrate the residual label information with the dynamic information database to trace the location and state sequence of hazardous chemicals before the fire. At the same time, based on the current fire situation and the residual label information of hazardous chemicals, simulate the fire spread path and the expansion trend of each path point on the spread path.

[0011] Step 5: Based on the source tracing results, simulation results, and other investigation evidence from the fire scene, comprehensively determine the cause of the fire accident and generate a source tracing accident report.

[0012] Preferably, each hazardous chemical container is equipped with an RFID tag, including:

[0013] The set of uses for the hazardous chemicals is statistically analyzed from historical databases, wherein the set of uses includes the frequency of use for each use and the usage scenarios for each use;

[0014] A probability table is constructed based on the frequency of use and the usage scenario, and RFID tags that match the probability table are matched from the table-tag lookup library and configured for the hazardous chemical containers.

[0015] Preferably, prior to the pre-deployment of the read-write network, the following steps are included:

[0016] The chemical storage location before the fire occurred was obtained by scanning the chemical storage location using a combination of lidar, structured light camera and millimeter wave radar to obtain three-dimensional point cloud data. The point cloud data in each dimension was then reconstructed to obtain the corresponding first three-dimensional structure, second three-dimensional structure and third three-dimensional structure.

[0017] The first three-dimensional structure, the second three-dimensional structure, and the third three-dimensional structure are projected into a three-dimensional coordinate system, and anomalies are locked. The anomalies are those where the spatial arrangement at the same coordinate point is different.

[0018] Statistically determine the coordinates of points in three dimensions under the same spatial arrangement, and determine the normal vectors in the corresponding spatial structure dimensions based on the corresponding coordinates of each point.

[0019] The smoothing vector is obtained by calculating the smoothness of all edge segments in each projected structure projected into the three-dimensional coordinate system, and the smoothing variance is also obtained.

[0020] Based on the smoothed variance, weighting factors are assigned to the spatial structure in each dimension.

[0021] The normal vector corresponding to the largest factor is extracted from the weight factors and regarded as the first vector, while the remaining two normal vectors are regarded as the second vector.

[0022] Calculate the deviation angle between each second vector and the first vector, and at the same time, calculate the distance deviation between the coordinates of the point corresponding to each second vector and the coordinates of the point corresponding to the first vector.

[0023] Adjust the coordinates of the points under the corresponding second vector according to the deviation angle and deviation distance under the same second vector;

[0024] Based on the two adjusted point coordinates and combined with the point coordinates corresponding to the first vector, and with weighted calculation relying on weight factors, the final coordinates are obtained and assigned to the abnormal point for updating.

[0025] Preferably, the coordinates of the points under the corresponding second vector are adjusted according to the deviation angle and deviation distance under the same second vector, including:

[0026]

[0027]

[0028] in, These are the coordinates of the point before adjustment; These are the adjusted point coordinates; , This is the axial coefficient; For adjustment functions; Indicates the corresponding second vector With the first vector The included angle; Represents the point under the corresponding second vector Points under the first vector The Euclidean distance between the points is given, and the point is given. The coordinates of the point are ; For the corresponding second vector With the first vector The modulus length; identity matrix The length of the module.

[0029] Preferably, the pre-deployed read / write network includes:

[0030] The updated site structure is categorized, and the signal attenuation coefficient matching the corresponding category is obtained from the category-signal database.

[0031] In the updated site structure, preset candidate locations are selected, and environmental feature vectors of each preset candidate location are extracted. A facility avoidance operator is introduced to determine the final location. The environmental feature vectors are related to the distance from the metal facility, the theoretical coverage volume under preset power, and the signal attenuation coefficient of the categories of objects involved in the coverage area under preset power.

[0032] The deployment and simulated reading of RFID reading devices are carried out according to all final locations. The simulated signal strength between the RFID reading device and the corresponding standard tag at each final location is determined by randomly arranging standard tags at known locations. By obtaining the corresponding actual signal strength, a deviation correction table is established to calibrate the deployment locations and obtain the read / write network.

[0033] Preferably, a dynamic information database for hazardous chemicals is established, including:

[0034] Redis is used to store frequently updated trajectory data. The data is sharded by hash of tag ID and a sliding time window is set for each shard. Expired data is automatically migrated to HDFS. HDFS stores data according to tag type and date, combined with Parquet columnar storage format.

[0035] A combined index is established for the spatiotemporal trajectories of the corresponding hazardous chemical containers, and implemented based on the R-tree spatial index.

[0036] Preferably, the location sequence and state sequence of hazardous chemicals prior to the fire include:

[0037] The RFID reader network collects the three-dimensional coordinates of the tags, the warehouse shelf codes, and the movement timestamps in real time to form a position sequence W1 during the movement process.

[0038] Simultaneously, the actual dynamic information associated with the tag is collected to form a state sequence X1 during the movement process. The real-time dynamic information is related to external environmental factors, the operated behavior, and the usage status.

[0039] Preferably, simulating the fire spread path and the expansion trend of each path point along the spread path includes:

[0040] The current fire situation is input into the site analysis model to obtain the initial fire path;

[0041] Based on the locational correlation between hazardous chemicals and the initial path of the fire, and in conjunction with residual label information, determine the alteration factor and alteration time of each hazardous chemical on the initial path of the fire.

[0042] The change factor is applied to the initial fire path according to the change time to simulate the fire spread path and the expansion trend of each path point on the spread path.

[0043] Preferably, the process of comprehensively determining the cause of a fire accident and generating a source tracing accident report includes:

[0044] Align the position sequence W1 with the shape sequence X1 to create a synchronized dataset D={W2,X2}, where W2 is the aligned position sequence and X2 is the aligned shape sequence.

[0045] Constructing a joint feature space for W2 and X2 defines a coupling index. ;

[0046]

[0047] When the coupling index Ct is greater than the preset threshold, it is determined to be a position-state coupling anomaly. The abnormal trajectory segment is associated with the device operation record and environmental data to construct an abnormal event directed graph containing four-tuples, wherein the four-tuples are related to time, position, state and operation.

[0048] For missing data segments, a spatiotemporal interpolation algorithm based on RBF neural network is used to predict trend values ​​by using the position and state of two adjacent time points to fill in the difference.

[0049] Based on the structural causal model, the causal relationships between trajectory variables are established to construct a trajectory causal graph. The effect value of each causal relationship in the trajectory causal graph is calculated and sorted to obtain the abnormal key events, thus generating the causal chain before the fire.

[0050] Preferably, a comprehensive assessment of the cause of a fire accident and the generation of a source tracing accident report includes:

[0051] Construct a fire spread chain based on simulation results and obtain positive factors for each critical point in the fire spread chain based on other survey evidence to obtain the maximum spread chain;

[0052] Based on the maximum diffusion chain and the causal chain before the fire, the cause of the fire accident is determined and a source tracing accident report is generated.

[0053] Compared with the prior art, the beneficial effects of this application are as follows:

[0054] By integrating residual label information and historical databases, the situation before the fire can be quickly and accurately reconstructed, the fire spread process can be simulated, and combined with other evidence at the scene, the cause of the fire can be accurately identified. This can prevent similar fire accidents from happening again and greatly improve the safety of hazardous chemical management.

[0055] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0056] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0057] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0058] Figure 1 This is a flowchart of the RFID-based hazardous chemical fire accident tracing method in an embodiment of the present invention. Detailed Implementation

[0059] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0060] This invention proposes an RFID-based method for tracing the source of hazardous chemical fire accidents, such as... Figure 1 As shown, it includes:

[0061] Step 1: Configure each hazardous chemical container with an RFID tag, wherein the RFID tag contains basic information and the initial location;

[0062] Step 2: Based on the pre-deployed read / write network, read the information of each RFID tag and transmit it to the central data platform to establish a dynamic information database of hazardous chemicals, and record the external environmental factors, operation behavior and usage status of the hazardous chemicals at different time points;

[0063] Step 3: After a fire involving hazardous chemicals, use a portable RFID reader to read the RFID tags present at the fire scene and obtain information about the remaining tags;

[0064] Step 4: Integrate the residual label information with the dynamic information database to trace the location and state sequence of hazardous chemicals before the fire. At the same time, based on the current fire situation and the residual label information of hazardous chemicals, simulate the fire spread path and the expansion trend of each path point on the spread path.

[0065] Step 5: Based on the source tracing results, simulation results, and other investigation evidence from the fire scene, comprehensively determine the cause of the fire accident and generate a source tracing accident report.

[0066] In this embodiment, the pre-deployed read / write network consists of RFID read / write devices pre-installed at locations where hazardous chemicals are stored and used, which read information from RFID tags.

[0067] The information in the dynamic information database is not static but constantly updated.

[0068] Portable RFID readers are used to find RFID tags that are still functional in the ruins after a fire and to read the information remaining on them.

[0069] Residual tag information refers to the information that can still be read from RFID tags after a fire. If no information can be read from the RFID tag, the residual tag information is empty. In this case, the location and status sequence of the hazardous chemicals before the fire are traced based on the last update status of the dynamic information database before the fire.

[0070] Location sequence is a chronological arrangement of all the places a hazardous chemical went from the beginning of its journey to the point before the fire.

[0071] A state sequence arranges the states of hazardous chemicals at different points in time in chronological order.

[0072] Fire spread path is a simulation of the route a fire takes from its initial ignition point to the surrounding area, such as burning from one corner of a warehouse to another.

[0073] The expansion trend is a prediction of how much the fire will grow in each place it passes through as it spreads, whether it will get bigger or smaller.

[0074] Other evidence found at the fire scene, besides RFID tag information, included other clues such as burnt-out equipment and residual chemical traces.

[0075] An incident investigation report is a detailed report that compiles all the investigation results and analysis process.

[0076] The beneficial effects of the above technical solution are: by integrating residual label information and historical databases, the situation before the fire can be quickly and accurately reconstructed, the fire spread process can be simulated, and combined with other evidence on site, the cause of the fire can be accurately identified, which can prevent similar fire accidents from happening again and greatly improve the safety of hazardous chemical management.

[0077] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires, which involves equipping each hazardous chemical container with an RFID tag, including:

[0078] The set of uses for the hazardous chemicals is statistically analyzed from historical databases, wherein the set of uses includes the frequency of use for each use and the usage scenarios for each use;

[0079] A probability table is constructed based on the frequency of use and the usage scenario, and RFID tags that match the probability table are matched from the table-tag lookup library and configured for the hazardous chemical containers.

[0080] In this embodiment, the historical database records all past information about hazardous chemicals, the usage set collects all the various purposes for which the hazardous chemicals have been used in the past, the usage frequency is the number of times each usage purpose has been used, and the usage scenario describes the specific environment in which each usage purpose occurs.

[0081] In this embodiment, the probability table is a table created based on the frequency and scenario of use to show the probability of each use occurring. The table-tag lookup library records RFID tags with different characteristics and their suitable use scenarios and conditions. Matching RFID tags is based on finding RFID tags in the table-tag lookup library from the probability table.

[0082] The beneficial effects of the above technical solution are: by analyzing the past usage of hazardous chemicals, the most suitable RFID tags can be accurately matched to them, so that RFID tags with different characteristics can work stably in the corresponding usage scenarios, ensuring that the information on the tags is read and recorded accurately and completely.

[0083] This invention proposes an RFID-based method for tracing the source of hazardous chemical fire accidents. Before deploying the read / write network, the method includes:

[0084] The chemical storage location before the fire occurred was obtained by scanning the chemical storage location using a combination of lidar, structured light camera and millimeter wave radar to obtain three-dimensional point cloud data. The point cloud data in each dimension was then reconstructed to obtain the corresponding first three-dimensional structure, second three-dimensional structure and third three-dimensional structure.

[0085] The first three-dimensional structure, the second three-dimensional structure, and the third three-dimensional structure are projected into a three-dimensional coordinate system, and anomalies are locked. The anomalies are those where the spatial arrangement at the same coordinate point is different.

[0086] Statistically determine the coordinates of points in three dimensions under the same spatial arrangement, and determine the normal vectors in the corresponding spatial structure dimensions based on the corresponding coordinates of each point.

[0087] The smoothing vector is obtained by calculating the smoothness of all edge segments in each projected structure projected into the three-dimensional coordinate system, and the smoothing variance is also obtained.

[0088] Based on the smoothed variance, weighting factors are assigned to the spatial structure in each dimension.

[0089] The normal vector corresponding to the largest factor is extracted from the weight factors and regarded as the first vector, while the remaining two normal vectors are regarded as the second vector.

[0090] Calculate the deviation angle between each second vector and the first vector, and at the same time, calculate the distance deviation between the coordinates of the point corresponding to each second vector and the coordinates of the point corresponding to the first vector.

[0091] Adjust the coordinates of the points under the corresponding second vector according to the deviation angle and deviation distance under the same second vector;

[0092] Based on the two adjusted point coordinates and combined with the point coordinates corresponding to the first vector, and with weighted calculation relying on weight factors, the final coordinates are obtained and assigned to the abnormal point for updating.

[0093] Preferably, the coordinates of the points under the corresponding second vector are adjusted according to the deviation angle and deviation distance under the same second vector, including:

[0094]

[0095]

[0096] in, These are the coordinates of the point before adjustment; These are the adjusted point coordinates; , This is the axial coefficient; For adjustment functions; Indicates the corresponding second vector With the first vector The included angle; Represents the point under the corresponding second vector Points under the first vector The Euclidean distance between the points is given, and the point is given. The coordinates of the point are ; For the corresponding second vector With the first vector The modulus length; identity matrix The length of the module.

[0097] In this embodiment, the first three-dimensional structure, the second three-dimensional structure, and the third three-dimensional structure are point cloud data reorganized in different ways, and each model shows the appearance of the site from different angles.

[0098] In this embodiment, there are some locations where the three models display different things. For example, the first model shows a box, but the second model shows it as empty. This location is an anomaly.

[0099] In this embodiment, the edge segment is the outline segment of the model projected into the three-dimensional coordinate system using a wireframe.

[0100] The smoothness is determined by observing whether these edge segments are smooth and fluid.

[0101] The smoothing vector is composed of the smoothness of each edge segment in the corresponding dimension, and the smoothing variance is obtained by using the variance calculation formula.

[0102] The weighting factor is the variance of the corresponding smoothed vector divided by the sum of the variances of all smoothed vectors.

[0103] The distance deviation is the calculated Euclidean distance between the two points.

[0104] In this embodiment, arccos(di,d0) is used to strictly define the vector angle to avoid ambiguity in direction deviation; the cross product |di×d0| ensures that the correction amount is linked to the perpendicular relationship of the vector plane (e.g., in three-dimensional space, deviations in the vertical direction need special correction); the unit matrix modulus |I| is used as a reference quantity to ensure the normalization of the correction amplitude and avoid correction imbalance caused by differences in units (e.g., distance in meters, angle in radians).

[0105] The beneficial effects of the above technical solution are as follows: by comprehensively scanning the site using multiple radar technologies, a three-dimensional model is established, and anomalies are identified and reconstructed point-to-point to ensure the accuracy of spatial data. This is crucial for the subsequent deployment of RFID reading and writing equipment, ensuring that the equipment is installed in the optimal location for better reading of tag information. At the same time, after a fire accident, it also helps to more accurately analyze the location and status of hazardous chemicals, providing a reliable spatial basis for accident tracing and simulating fire spread paths, greatly improving the safety of hazardous chemical management and the accuracy of accident handling.

[0106] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires, which includes a pre-deployed read / write network:

[0107] The updated site structure is categorized, and the signal attenuation coefficient matching the corresponding category is obtained from the category-signal database.

[0108] In the updated site structure, preset candidate locations are selected, and environmental feature vectors of each preset candidate location are extracted. A facility avoidance operator is introduced to determine the final location. The environmental feature vectors are related to the distance from the metal facility, the theoretical coverage volume under preset power, and the signal attenuation coefficient of the categories of objects involved in the coverage area under preset power.

[0109] The deployment and simulated reading of RFID reading devices are carried out according to all final locations. The simulated signal strength between the RFID reading device and the corresponding standard tag at each final location is determined by randomly arranging standard tags at known locations. By obtaining the corresponding actual signal strength, a deviation correction table is established to calibrate the deployment locations and obtain the read / write network.

[0110] In this embodiment, category segmentation divides the site structure into different areas according to different functions or item types, and the category-signal database is a dictionary that records the influence of different category areas on signals.

[0111] In this embodiment, the preset candidate locations are some places in the site that may be suitable for installing RFID reading devices. The environmental feature vector records the environmental information around the preset candidate locations.

[0112] In this embodiment, the final location is the determined installation location of the RFID reading device.

[0113] In this embodiment, the deployment simulation and simulated reading are performed virtually on a computer after the installation location is determined. The standard tag is an RFID tag with a known accurate location and characteristics.

[0114] Simulated signal strength is the signal strength value displayed when the device reads a standard tag during a simulated test. Actual signal strength is the real signal strength value obtained when the RFID reader is actually installed in a predetermined location and reads a standard tag.

[0115] The deviation correction table compares the simulated signal strength with the actual signal strength, identifies the differences between the two, and creates a table.

[0116] Calibration involves fine-tuning the installation location or parameters of the equipment according to a deviation correction table to make the signal readings of the equipment more accurate.

[0117] The beneficial effects of the above technical solution are as follows: by classifying and analyzing the chemical storage locations, signal attenuation patterns are identified; the optimal installation location is determined by combining environmental characteristics and avoidance rules; and errors are corrected through simulation and actual testing. The final read / write network ensures that RFID tag information is read stably and accurately, providing reliable data support for tracking the location and status of chemicals in daily management, as well as obtaining residual tag information after a fire. This greatly improves the efficiency and safety of hazardous chemical management, making management work more intelligent and precise.

[0118] This invention proposes an RFID-based method for tracing the source of hazardous chemical fire accidents, establishing a dynamic information database for hazardous chemicals, including:

[0119] Redis is used to store frequently updated trajectory data. The data is sharded by hash of tag ID and a sliding time window is set for each shard. Expired data is automatically migrated to HDFS. HDFS stores data according to tag type and date, combined with Parquet columnar storage format.

[0120] A combined index is established for the spatiotemporal trajectories of the corresponding hazardous chemical containers, and implemented based on the R-tree spatial index.

[0121] The beneficial effects of the above technical solution are: Redis is used to quickly record frequently changing data, ensuring real-time performance; hash sharding and sliding time windows are used for efficient data management; expired data is migrated to HDFS for long-term storage, saving high-speed storage resources; and Parquet columnar storage and R-tree spatial indexing make data querying and analysis faster and more efficient. In this way, whether it's routine monitoring of the movement of hazardous chemicals or quickly tracing the historical location and status of chemicals after a fire, the required information can be obtained rapidly and accurately, providing strong support for the safety management of hazardous chemicals and accident tracing.

[0122] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires, tracing the location and state sequence of hazardous chemicals before the fire, including:

[0123] The RFID reader network collects the three-dimensional coordinates of the tags, the warehouse shelf codes, and the movement timestamps in real time to form a position sequence W1 during the movement process.

[0124] Simultaneously, the actual dynamic information associated with the tag is collected to form a state sequence X1 during the movement process. The real-time dynamic information is related to external environmental factors, the operated behavior, and the usage status.

[0125] The beneficial effect of the above technical solution is that by determining the sequences in two cases, it provides a basis for subsequent anomaly analysis.

[0126] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires, simulating the fire spread path and the expansion trend of each path point, including:

[0127] The current fire situation is input into the site analysis model to obtain the initial fire path;

[0128] Based on the locational correlation between hazardous chemicals and the initial path of the fire, and in conjunction with residual label information, determine the alteration factor and alteration time of each hazardous chemical on the initial path of the fire.

[0129] The change factor is applied to the initial fire path according to the change time to simulate the fire spread path and the expansion trend of each path point on the spread path.

[0130] The current fire situation includes information such as the specific location of the fire, the size of the fire, and the direction of the flames. The site analysis model is a computer program that has been trained with a large number of fire cases and data, and knows how a fire is likely to develop in different environments.

[0131] The initial fire path is the predicted route along which a fire is most likely to spread initially. For example, it might predict that the fire will first spread from the third-floor warehouse to the adjacent corridor, and then towards the stairwell. In this embodiment, the locational correlation between hazardous chemicals and the initial fire path is determined by examining the distance and relative position of the hazardous chemicals' storage location to the initial fire path. For instance, if there is a warehouse storing gasoline on the initial fire path, gasoline is highly flammable and could accelerate the spread of the fire, while less flammable chemicals are stored further away from the initial fire path, having a smaller impact on the fire's spread.

[0132] The change factor is a numerical value that indicates the degree to which a hazardous chemical affects the path of fire spread. A large change factor indicates a greater impact on the spread of fire if the chemical is flammable, explosive, and located close to the initial path of the fire; conversely, a small change factor indicates a smaller impact. For example, gasoline might have a change factor of 0.8 (significant impact), while a non-flammable buffer might have a change factor of 0.1 (relatively small impact).

[0133] The change-of-path time is the point at which hazardous chemicals begin to significantly affect the initial path of a fire. For example, when flames reach a warehouse storing gasoline, the gasoline begins to burn and changes the direction of the fire's spread; this moment is the change-of-path time. For instance, at the change-of-path time, the initial path of the fire, which was originally spreading towards the corridor, may suddenly change direction due to the burning gasoline, spreading towards an area storing more flammable materials.

[0134] Simulating the fire spread path involves continuously adjusting and updating the fire spread route by adding changes to factors and time, ultimately resulting in a fire spread process that more closely resembles reality.

[0135] In this embodiment, the expansion trend of each path point not only knows where the fire will spread, but also predicts the changes in the size of the fire at each location along the spread path. For example, it can predict that the fire in a certain room will double in size after 10 minutes, or that the flames in a certain corridor will increase and the temperature will rise.

[0136] The beneficial effects of the above technical solution are: it allows for the estimation of a general fire spread route based on the initial situation at the fire scene, and by combining this with the location and characteristics of hazardous chemicals, the route can be precisely adjusted and refined. Furthermore, it can predict the changing trends of the fire at each stage and location. Such simulations provide a clear understanding of how the fire develops and which hazardous chemicals exacerbate it, enabling the development of more effective strategies during firefighting and rescue operations, more accurate analysis of the causes of the fire during accident investigations, and contributing to the improvement of subsequent hazardous chemical storage and management measures to prevent similar accidents from recurring.

[0137] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires. The process of comprehensively determining the cause of a fire and generating a source tracing report includes:

[0138] Align the position sequence W1 with the shape sequence X1 to create a synchronized dataset D={W2,X2}, where W2 is the aligned position sequence and X2 is the aligned shape sequence.

[0139] Constructing a joint feature space for W2 and X2 defines a coupling index. ;

[0140]

[0141] When the coupling index Ct is greater than the preset threshold, it is determined to be a position-state coupling anomaly. The abnormal trajectory segment is associated with the device operation record and environmental data to construct an abnormal event directed graph containing four-tuples, wherein the four-tuples are related to time, position, state and operation.

[0142] For missing data segments, a spatiotemporal interpolation algorithm based on RBF neural network is used to predict trend values ​​by using the position and state of two adjacent time points to fill in the difference.

[0143] Based on the structural causal model, the causal relationships between trajectory variables are established to construct a trajectory causal graph. The effect value of each causal relationship in the trajectory causal graph is calculated and sorted to obtain the abnormal key events, thus generating the causal chain before the fire.

[0144] In this embodiment, the joint feature space is to put the location and morphology data into a table, with each row containing "time, location coordinates, and morphology description".

[0145] The preset threshold is a pre-defined value of 0.7.

[0146] A four-tuple describes an anomaly using "time, location, state, and action".

[0147] A directed graph is a flowchart that connects abnormal events in causal order.

[0148] Structural causal models are obtained by training neural network models with samples composed of different causes and effects.

[0149] The effect value is a value that quantifies the influence of each causal relationship, and it ranges from 0 to 1.

[0150] The beneficial effects of the above technical solutions are: accurately locating the root cause of the problem, restoring the complete process of the event, and making post-event tracing and pre-event prevention more reliable.

[0151] This invention proposes an RFID-based method for tracing the source of hazardous chemical fires, comprehensively determining the cause of the fire and generating a source tracing accident report, including:

[0152] Construct a fire spread chain based on simulation results and obtain positive factors for each critical point in the fire spread chain based on other survey evidence to obtain the maximum spread chain;

[0153] Based on the maximum diffusion chain and the causal chain before the fire, the cause of the fire accident is determined and a source tracing accident report is generated.

[0154] In this embodiment, the on-site fire spread chain is a fire spread route map compiled based on simulation results, which connects each place the fire passes through in sequence to form a chain.

[0155] Critical points of fire spread are the locations or moments in the fire spread chain that play a crucial role in the fire's propagation. For example, the doorway from one room to another is a critical point; or the moment a flammable material is ignited is also a critical point. Positive factors are numerical values ​​used to measure the extent to which other investigative evidence contributes to the fire's spread at a critical point. For instance, if analysis suggests that traces of a short circuit found at the scene greatly contributed to the fire's expansion, its positive factor is high; conversely, if a piece of evidence has a relatively small impact on the fire's spread, its positive factor is low.

[0156] The maximum diffusion chain is the most likely fire spread path determined after comprehensively considering the positive factors of simulation results and other survey evidence.

[0157] Determining the cause of a fire accident involves combining the analysis of the maximum spread chain (how the fire spreads) and the pre-fire causal chain (why the fire occurred) to find the true cause of the fire. For example, if it is found that the fire was caused by a short circuit (pre-fire causal chain) and spread along a specific path (maximum spread chain), it can be ultimately determined that the fire was caused by a short circuit, coupled with improper storage of flammable materials at the scene, leading to the rapid spread of the fire.

[0158] The beneficial effects of the above technical solution are: by determining the most realistic fire spread path through simulation and evidence analysis, and then combining various causal relationships before the fire occurs, the root cause of the fire can be accurately identified, and the resulting source tracing accident report is easy to view.

[0159] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for tracing a hazardous chemical fire accident based on RFID, characterized in that, The method comprises the following steps: Step 1: configuring an RFID tag for each dangerous chemical container, wherein the RFID tag is written with basic information and an initial position; Step 2: reading the information of each RFID tag based on a pre-deployed read-write network and transmitting to a central data platform to establish a dynamic information database of dangerous chemicals, recording the external environmental factors, operating behaviors and use states of the dangerous chemicals at different time points; Step 3: when a fire accident of the dangerous chemicals occurs, reading the RFID tags existing in the fire scene based on a portable RFID read-write device to obtain residual tag information; Step 4: integrating the residual tag information with the dynamic information database to trace the position sequence and state sequence of the dangerous chemicals before the fire occurs, and simultaneously simulating a fire spreading path and an expansion trend of each path point on the spreading path according to the current fire situation and the residual tag information of the dangerous chemicals; Step 5: comprehensively determining the cause of the fire accident and generating a trace accident report according to the tracing result, the simulation result and other investigation evidence of the fire scene; Tracing the position sequence and state sequence of the dangerous chemicals before the fire occurs comprises: Collecting the three-dimensional coordinates of the tag, the storage shelf code and the moving time stamp in real time through the RFID read-write network to form a position sequence W1 in the moving process; Meanwhile, synchronously collecting actual dynamic information associated with the tag to form a state sequence X1 in the moving process, wherein the actual dynamic information is related to the external environmental factors, the operating behaviors and the use states; The process of comprehensively determining the cause of the fire accident and generating a trace accident report comprises: Aligning the position sequence W1 with the state sequence X1 to establish a synchronous data set D={W2, X2}, wherein W2 is the aligned position sequence and X2 is the aligned state sequence; Constructing a joint feature space of W2 and X2 defines the coupling degree indicator ; When the coupling degree index Ct is greater than a preset threshold, it is determined that the position-state coupling is abnormal, and the abnormal segment of the trajectory is associated with the device operation record and the environmental data to construct an abnormal event directed graph containing four tuples, wherein the four tuples are related to time, position, state and operation; For the data missing segment, a spatio-temporal interpolation algorithm based on RBF neural network is adopted to predict the trend value of the position and state of the adjacent two time points for difference supplement; A causal relationship between the trajectory variables is established based on a structural causal model to construct a trajectory causal graph, an effect value of each cause and effect in the trajectory causal graph is calculated, the effect values are sorted to obtain an abnormal key event, and a cause-effect chain before the fire is generated.

2. The RFID-based hazardous chemical fire incident traceback method of claim 1, wherein, Configuring an RFID tag for each dangerous chemical container comprises: Statistically obtaining a use purpose set of the dangerous chemicals from a historical database, wherein the use purpose set contains a use frequency under each use purpose and a use scene of each use purpose; Constructing a probability table according to the use frequency and the use scene, and matching the RFID tag matched with the probability table from a table-tag matching library to configure the dangerous chemical container.

3. The RFID-based hazardous chemical fire incident traceback method of claim 1, wherein, Before the pre-deployed read-write network, the method comprises the following steps: Acquire the chemical storage place before the fire occurs, scan the chemical storage place based on the combination of laser radar, structured light camera and millimeter wave radar, acquire three-dimensional point cloud data, and reconstruct the point cloud data in each dimension to obtain the corresponding first three-dimensional structure, second three-dimensional structure and third three-dimensional structure; Project the first three-dimensional structure, the second three-dimensional structure and the third three-dimensional structure into a three-dimensional coordinate system respectively, and lock the abnormal points, wherein the abnormal points are the differences in the spatial arrangement of the same coordinate point; Statistically count the point coordinates of the three dimensions under the same spatial arrangement, and determine the normal vector under the corresponding spatial structure dimension based on the corresponding point coordinate points; Obtain a smoothing vector and a smoothing variance according to the smoothness of all edge line segments in each projected structure projected into the three-dimensional coordinate system; According to the smoothing variance, assign a weight factor to the spatial structure in each dimension; Extract the normal vector corresponding to the maximum factor from the weight factor as the first vector, and regard the remaining two normal vectors as the second vectors; Calculate the deviation angle of each second vector with respect to the first vector, and calculate the distance deviation of the point coordinates corresponding to each second vector with respect to the point coordinates corresponding to the first vector. Adjust the point coordinates under the corresponding second vector according to the deviation angle and the deviation distance under the same second vector. Based on the two adjusted point coordinates and combined with the point coordinates corresponding to the first vector, and relying on the weight factor for weighted calculation, obtain the final coordinates and update the abnormal points.

4. The RFID-based hazardous chemical fire incident traceback method of claim 3, wherein, Adjusting the point coordinates under the corresponding second vector according to the deviation angle and the deviation distance under the same second vector includes: wherein is the point coordinate before adjustment; is the point coordinate after adjustment; , is the axial coefficient; is the adjustment function; denotes the angle between the second vector and the first vector ; denotes the Euclidean distance between the point under the second vector and the point under the first vector, and the point coordinate of the point is ; is the length of the second vector corresponding to the angle to the first vector ; is the length of the identity matrix 5. The RFID-based hazardous chemical fire incident traceback method of claim 3, wherein, Pre-deploying a read-write network includes: Classifying the updated site structure, and obtaining the signal attenuation coefficient matching the corresponding category from the category-signal database; Selecting a preset candidate position in the updated site structure, and extracting an environmental feature vector of each preset candidate position, and introducing a facility avoidance factor to determine a final position, wherein the environmental feature vector is related to the distance from the metal facility, the theoretical coverage volume under the preset power, and the signal attenuation coefficient of the category object involved in the covered range under the preset power. According to all the final positions, perform RFID read device deployment simulation and simulated reading, determine the simulation signal strength between the RFID read device and the corresponding standard tag in each final position by randomly arranging the standard tags of known positions, and establish a deviation correction table by obtaining the corresponding actual signal strength to calibrate the deployment position and obtain the read-write network.

6. The RFID-based hazardous chemical fire incident traceback method of claim 1, wherein, Establish a dynamic information database of hazardous chemicals, including: Use Redis to store high-frequency updated trajectory data, set a sliding time window for each shard according to the label ID hash, and automatically migrate expired data to HDFS, wherein HDFS stores according to label type and date, combined with Parquet columnar storage format for storage; Establish a combined index for the space-time trajectory of the corresponding hazardous chemical container, and realize it based on R-tree spatial index.

7. The RFID-based hazardous chemical fire incident traceback method of claim 1, wherein, simulate a fire spreading path and an expansion trend of each path point on the spreading path, comprising: inputting the current fire situation into a site analysis model to obtain a fire initial path; determining an alteration factor and an alteration time of each hazardous chemical to the fire initial path according to a position correlation between the hazardous chemical and the fire initial path and in combination with residual label information; acting the alteration factor on the fire initial path according to the alteration time to simulate a fire spreading path and an expansion trend of each path point on the spreading path.

8. The RFID-based hazardous chemical fire incident traceback method of claim 1, wherein, comprehensively determine a cause of the fire accident and generate a traceable accident report, comprising: constructing a site fire diffusion chain based on the simulation result and obtaining a positive factor of each diffusion key point in the site fire diffusion chain based on other survey evidences to obtain a maximum diffusion chain; determining a cause of the fire accident and generating a traceable accident report based on the maximum diffusion chain and a fire pre-cause chain.

Images