A park building security state monitoring method and system based on an internet of things perception layer cooperation

Abstract

The application discloses a kind of garden building security state monitoring method and system based on internet of things perception layer cooperation, and the present application relates to building security monitoring technical field.It includes multiple perception cooperative clusters covering perception node and convergence subnode, cloud server, and local linkage control module, adopts space clustering and neighborhood mutual check, and removes abnormal data such as node failure and local interference;At the same time, preset absolute dangerous threshold for each physical quantity, when all similar node data in neighborhood exceed the threshold, directly adopt the data to prevent missing detection of real large-area anomaly.Multiple-dimension index is fused by trust allocation, and conflict trust is redistributed when evidence is highly conflicted, to solve the evaluation jump problem caused by index contradiction, so that the security confidence generated finally is more stable.

Description

Technical Field

[0001] This invention relates to the field of building safety monitoring technology, specifically to a method and system for monitoring the safety status of buildings in a park based on the Internet of Things (IoT) sensing layer collaboration. Background Technology

[0002] Existing building monitoring solutions often rely on single-sensor threshold alarms. Due to the lack of spatial redundancy verification between adjacent similar sensors, abnormal fluctuations in sensor readings can easily lead to false alarms or missed alarms, resulting in low data confidence. Furthermore, in real-world building monitoring scenarios, monitoring subsystems for different safety dimensions, such as electrical, fire protection, and structural monitoring, often operate independently. The monitoring information collected by these independent subsystems typically belongs to different spatial reference points, making it impossible to directly and effectively integrate this information from different sources under a single spatial coordinate framework for comprehensive fusion and overlay processing. These issues make it currently difficult to quantitatively calculate and assess the comprehensive risk level across multiple safety aspects for any specified local area within the entire building.

[0003] Therefore, a campus building security status monitoring solution is needed that can achieve collaborative verification at the perception layer, cross-dimensional conflict identification and fusion, and edge offline autonomy to overcome the shortcomings of the existing technologies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration, thus solving the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration, comprising the following steps: S1. Divide the physical space of the building into multiple sensing and coordination clusters, each sensing and coordination cluster corresponding to a unique number k. Deploy multiple sensing nodes and one aggregation sub-node for each sensing and coordination cluster, where the aggregation sub-node has edge computing capabilities. Among them, the sensing node is used to collect at least one physical quantity, and in the same sensing cooperative cluster, at least two sensing nodes used to collect the same physical quantity are deployed, which together form a sensor group of the same type. S2. Synchronize the clocks within the sensing cluster. The aggregation sub-node triggers all sensing nodes within the sensing cluster to synchronously collect physical quantities at the same time, thereby generating raw data with synchronization timestamps. For each sensor group of the same type, neighborhood mutual verification is performed using the raw data of multiple sensing nodes that are physically adjacent within the same sensor group at the same synchronization time stamp. The method is as follows: a statistical benchmark is determined based on the raw data of the neighboring sensing nodes within the same sensor group. If the deviation of the raw data of any sensing node from the statistical benchmark exceeds the preset deviation condition, the raw data is recorded as invalid data and discarded; otherwise, it is recorded as valid data. Among them, if the physical distance between sensing nodes is less than the preset installation radius, they are determined to be neighboring sensing nodes; S3: The aggregation sub-node extracts features from all valid data within the perception collaboration cluster after neighborhood mutual verification, obtains multiple feature indicators, and divides each feature indicator into the corresponding dimension group according to the preset security dimension. Using a trust-based multi-source fusion method, the feature indicators contained in each security dimension are fused to obtain the preliminary security confidence components of the perception collaboration cluster in each security dimension. S4. The cloud server obtains the preliminary security confidence components of each sensing and collaborative cluster in each security dimension, and performs cross-cluster association and fusion with the spatial grid of the building information model to generate the local security risk level of each cubic grid; based on the local security risk level of each cubic grid, it outputs linkage control commands to the corresponding building equipment.

[0006] As a further aspect of the present invention, the specific method of neighborhood mutual verification is as follows: For a sensing node to be verified in a group of sensors of the same type, a neighborhood set is formed by identifying sensing nodes of the same type within the same sensor group whose physical distance from it is less than a preset installation radius. The statistical baseline is determined by calculating the median of the raw data of the physical quantities of each sensing node in the neighborhood set as the statistical baseline, and calculating the median absolute deviation; the normalized deviation is determined based on the median and the median absolute deviation, and the normalized deviation is used as the quantification value of the degree of deviation. When the absolute value of the normalized deviation is greater than the preset deviation threshold, it means that the preset deviation condition has been exceeded.

[0007] As a further aspect of the present invention: before performing deviation elimination, the neighborhood mutual verification further includes: if the original data of the physical quantity of all sensing nodes in the neighborhood set exceeds the preset absolute danger threshold for the physical quantity type, then the original data of each node in the neighborhood set is directly retained.

[0008] As a further aspect of the present invention: the normalized deviation is achieved through... The calculation shows that, To normalize the deviation, The raw data of the physical quantities of the sensing nodes to be verified. The median of the neighborhood set. The median absolute deviation of the original data from all sensing nodes within the neighborhood set is used to characterize the dispersion of the neighborhood data. This is the distribution correction factor. It is a preset minimum positive number to prevent the denominator from being zero when the sum of all sensing node data in the neighborhood set is completely consistent.

[0009] As a further aspect of the present invention: for each security dimension, the fusion processing method is as follows: Define an identification framework consisting of security assumptions and danger assumptions, and determine a basic trust assignment value for each feature index within the security dimension. The basic trust assignment value includes a first trust value assigned to the security assumption, a second trust value assigned to the danger assumption, and a third trust value assigned to the entire set of the identification framework. The complete identification framework includes two core assumptions: safety assumption and danger assumption, representing comprehensive situations where the monitoring status cannot be accurately determined. The basic trust assignment values ​​of all feature indicators within the security dimension are successively fused using the orthogonal combination rule, and a conflict factor reflecting the degree of evidence conflict is calculated at each fusion. When the value of the conflict factor is greater than a preset conflict upper limit, the conflict redistribution rule is activated. That is, all the conflict trust values ​​generated by this fusion are assigned to the entire set of the identification framework. The result obtained after redistribution continues to participate in the next fusion process until a comprehensive trust allocation value for this security dimension is obtained.

[0010] Based on the second trust value assigned to the dangerous hypothesis in the comprehensive trust allocation value, the preliminary security confidence component of the perception collaboration cluster in the security dimension is generated.

[0011] As a further aspect of the present invention, the orthogonal combination rule and conflict redistribution rule are as follows: For two sets of basic trust assignment functions m1 and m2, find the conflict factor K between them; For any subset A under the recognition framework F, when A is a subset that is neither empty nor universal, by... Calculate its fusion trust value When A is the entire set of recognition frames F, through Calculate its fusion trust value ; in, Let A be the complete set, and let B and C be the subsets corresponding to the fusion results. and A subset of the corresponding recognition framework; and , For an empty set, the compensation coefficient δ is determined by comparing the conflict factor K with the preset conflict upper limit: when the conflict factor K is less than the preset conflict upper limit, δ=0; when the conflict factor K is not less than the preset conflict upper limit, δ=K.

[0012] As a further aspect of the present invention: cross-cluster association fusion is performed by combining the spatial grid of the Building Information Model, including: The building information model of the building is preloaded into the cloud server, and the building information model is divided into several cubic grids; The initial security confidence components of each sensing cooperative cluster in each security dimension are mapped to the covered cubic grid. For a sensing cooperative cluster, when the initial security confidence component in a security dimension is allocated to each relevant cubic grid, the weight of the sensing node of the corresponding type in the sensing cooperative cluster to each cubic grid is calculated. For the mapping values ​​received by the same cubic grid from different sensing cooperative clusters in the same security dimension, the maximum value is taken as the risk component of the grid in that security dimension. For each grid cell, the weighted sum of its mapping values ​​across each security dimension yields the comprehensive risk value of the cubic grid. The overall building security status level is generated based on the comprehensive risk value of all cubic grids. The security status level is divided according to a preset risk threshold range and is used for status display rather than for direct linkage control commands.

[0013] As a further aspect of the present invention: the weight of grid g on the security dimension d Calculate using the following formula: Among them, when At that time, ,when At that time, ; in, The preset effective sensing radius of the sensor. express arrive Euclidean distance, To determine the equivalent spatial location of the sensing node corresponding to dimension d in the sensing cooperative cluster k. Let g be the center coordinates of the grid and k be the index of the sensing cooperative cluster. The equivalent spatial location is the geometric center coordinates of all sensing nodes belonging to the security dimension d within the sensing cooperative cluster k. If there is only one sensing node in this dimension, then the installation coordinates of that node are taken.

[0014] As a further aspect of the present invention, for each cubic grid, the comprehensive risk value of the cubic grid is divided into three safety risk levels: low risk, medium risk, and high risk, according to a preset risk threshold range. Under normal communication conditions, for any target area covered by a cubic grid, when the local safety risk level of the building is low, an early warning message is pushed to the management terminal; when the local safety risk level of the building is medium, the edge side drives the corresponding zone's audible and visual alarms and / or smoke exhaust fans and / or power circuit breakers to operate; when the local safety risk level of the building is high, the cloud also coordinates to drive the building's emergency lighting, elevator emergency landing, and fire door release devices.

[0015] As a further aspect of the present invention: the aggregation sub-node monitors the communication status with the cloud server. When the duration of continuous communication interruption reaches a preset threshold, it enters the local autonomous mode. In the local autonomous mode, the aggregation sub-node calls the preliminary security confidence component that was generated latest before the communication interruption and has completed conflict resolution. When the preliminary security confidence component of at least one security dimension reaches the local high-risk trigger threshold, it directly drives the building equipment connected to the aggregation sub-node to perform security protection actions, that is, it directly drives the audible and visual alarm, smoke exhaust fan, power circuit breaker and fire door release device connected to the aggregation sub-node to act, and records it synchronously to the cloud.

[0016] A building safety status monitoring system based on IoT sensing layer collaboration includes: Multiple sensing clusters, each containing multiple sensing nodes and a convergence sub-node with edge computing capabilities; within the same sensing cluster, multiple sensing nodes that collect the same physical quantity are deployed to form a group of similar sensors. The aggregation sub-node is used for clock synchronization within the sensing cluster, triggering each sensing node to synchronously collect physical quantities and generate raw data with synchronization timestamps at the same time. For each sensor group of the same type, neighborhood cross-verification is performed using the raw data of physically adjacent nodes within the same sensor group. Raw data of the corresponding physical quantity type of sensing nodes whose raw data deviates from the statistical distribution characteristic benchmark by more than the corresponding preset deviation threshold are discarded as invalid data; otherwise, they are retained as valid data. Feature extraction is performed on the valid data, and they are grouped according to preset security dimensions. Using a trust-based multi-source fusion method, each feature index contained in each security dimension is fused to obtain the preliminary security confidence components of the perception collaboration cluster in each security dimension; in the fusion process, the conflict factor is calculated, and when the conflict factor exceeds the preset conflict limit, the conflict redistribution rule of assigning the conflict trust value to the entire set of the identification framework is adopted. The cloud server is used to obtain the preliminary security confidence components of each sensing and collaborative cluster in each security dimension, and to perform cross-cluster association and fusion with the building information model spatial grid to generate the local security risk level of each cubic grid, and generate linkage control commands accordingly; at the same time, it generates the security status level of the entire building for status display. The local linkage control module, located on the aggregation sub-node, is used to execute commands under normal communication conditions. When it detects that the duration of continuous communication interruption with the cloud server has reached a preset threshold, it enters the local autonomous mode. Based on the preliminary security confidence components of this sensing and collaborative cluster in each security dimension, when any preliminary security confidence component reaches the local high-risk trigger threshold, it directly drives the building equipment connected to this aggregation sub-node to perform security protection actions.

[0017] As a further aspect of the present invention: in the neighborhood mutual verification, the aggregation sub-node is further configured such that: if the original data of all sensing nodes in the neighborhood set of a similar sensor group exceeds the preset absolute danger threshold for the physical quantity type, then the original data of each node in the neighborhood set is directly retained.

[0018] This invention provides a method and system for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration. Compared with existing technologies, it has the following advantages: By using neighborhood mutual verification of similar sensor groups within a sensing collaborative cluster, and leveraging spatial redundancy to identify and eliminate abnormal data caused by node faults and local interference, the problem of false alarms / missed alarms in single-sensor threshold alarms is solved.

[0019] By employing a multi-source fusion method based on trust allocation and conflict redistribution rules, stable confidence assessment results can still be output even when evidence is highly conflicting, thus solving the problem of indicator contradictions and misjudgments caused by the independent operation of multi-dimensional subsystems.

[0020] By mapping the initial confidence components of each cluster to the BIM spatial grid and performing cross-cluster correlation and fusion, the scattered local assessment results are unified under the three-dimensional spatial coordinate framework, realizing the comprehensive risk assessment of any local area in the entire building.

[0021] By automatically switching to local autonomous mode when communication is interrupted, the aggregation sub-nodes continuously update the cluster confidence level and drive local device actions directly based on the latest confidence level, ensuring the continuity and reliability of safety response under extreme conditions. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration, according to the present invention.

[0023] Figure 2This is a system block diagram of a campus building safety status monitoring system based on IoT sensing layer collaboration, according to the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1 and Figure 2 As shown, the embodiments of the present invention provide the following technical solutions: As an embodiment of the present invention: like Figure 1 As shown, this invention is a method for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration, comprising the following steps: S1. Divide the physical space of the building into multiple sensing coordination clusters. A sensing coordination cluster is a group of devices consisting of sensing nodes that are physically adjacent and functionally coordinated, and a convergence sub-node. For example, divide each floor, power distribution room, underground garage, and core computer room into independent clusters. Each sensing coordination cluster is denoted as k, where k is the index number of the sensing coordination cluster, k=1,2,......,U, and U represents the number of sensing coordination clusters.

[0026] Deploy multiple sensing nodes and a convergence sub-node with edge computing capabilities for each sensing collaboration cluster; Sensing nodes are used to collect at least one physical quantity, and within the same sensing cluster, several sensing nodes that collect the same physical quantity are deployed to form a group of similar sensors.

[0027] For example, within the sensing cluster corresponding to the power distribution room, multiple current transformer nodes for current acquisition, multiple temperature sensor nodes for temperature acquisition, and one smoke concentration node are deployed. The current transformer nodes form a group of similar sensors corresponding to current, and the temperature nodes form a group of similar sensors corresponding to temperature.

[0028] In this embodiment, the aggregation sub-node is connected to each sensing node via a local communication link such as RS-485 or LoRa, and has a processor and memory to run edge computing programs.

[0029] S2. Synchronize the clocks within the sensing cluster. The aggregation sub-node obtains the reference time through the Network Time Protocol (NTP). Each sensing node calibrates its local clock according to the reference time. The aggregation sub-node triggers all sensing nodes within the sensing cluster to synchronously collect physical quantities at the same time and generate raw data with synchronization timestamps.

[0030] Let S be the raw data collected by the i-th sensing node of the k-th sensing cluster at synchronization time t for physical quantity type p. k,i，t，p .

[0031] Where i is the index of the sensing node in the k-th sensing cluster, t is the synchronization timestamp index after clock synchronization in the k-th sensing cluster, and p is the index of the physical quantity type corresponding to the physical quantity collected by the sensing node.

[0032] In this embodiment, if the sensing node is a composite sensor, it can simultaneously output raw data of multiple physical quantities of different types.

[0033] For each sensor group of the same type, neighborhood cross-verification is performed using the raw data from multiple sensing nodes that are physically adjacent within the same sensor group at the same synchronization timestamp, as follows: The statistical distribution characteristics of the raw data of adjacent sensing nodes within the same type of sensor group are calculated as a benchmark. If the raw data of the physical quantity of the sensing node within the same type of sensor group deviates from the benchmark by more than the preset deviation threshold for the physical quantity type, the raw data of the sensing node corresponding to the physical quantity type is marked as invalid data and removed; otherwise, it is retained as valid data. Taking a temperature sensor group as an example, the set of sensing node numbers within the temperature sensor group is denoted as W. temp For the sensing node i to be verified, where i∈W temp Determine its neighborhood set N i The physical installation distance between the temperature sensor group and sensing node i is less than the preset installation radius d. max Similar temperature nodes.

[0034] Among them, the preset installation radius d max The value is determined based on the sensor deployment density and effective monitoring range; for example, for a temperature sensor, d max The setting can be 8-12 meters; in this embodiment, d max The value is 10 meters.

[0035] Collect the raw temperature physical quantity S(W) of each sensing node i in the neighborhood set at the same synchronization timestamp t. k,i，t，p And calculate the median B(W) and median absolute deviation BP(W) of the data; where the median and median absolute deviation are the statistical distribution characteristics; Among them, the median absolute deviation is obtained through Calculations show that median() represents taking this group. the median; pass Calculate the normalized deviation. ,in, The raw data for the temperature physical quantity of the sensing node to be verified. The median of the neighborhood set. The median absolute deviation of the neighborhood set. This is the distribution correction factor. In this embodiment, the denominator is set to a preset minimum positive number to prevent it from being zero. The value is 1.483. The value is 0.0001.

[0036] in, =0 and = When Z = 0, it means that it is completely consistent with the neighborhood set.

[0037] Will Compared with the preset deviation threshold ZY(W) for temperature type: If |Z|>ZY(W), then S(W) k,i，t，p Mark as invalid and remove; otherwise, keep.

[0038] In this embodiment, if the sensing node is also a composite sensor and provides physical quantities of other physical quantity types, then the physical quantities of other physical quantity types are not affected.

[0039] If the neighborhood set N i If the original data of the temperature physical quantity of all sensing nodes exceeds the preset absolute danger threshold for temperature, then the original data of each sensing node in the neighborhood set corresponding to the physical quantity type will be directly retained, and deviation elimination will no longer be performed.

[0040] In this embodiment, the absolute danger threshold for temperature is set to 90°C; the smoke concentration can be set to 0.5%obs / m, etc., depending on the actual sensor specifications.

[0041] S3. After completing the neighborhood mutual verification of all similar sensor groups, the aggregation sub-node performs feature extraction on all valid data within the sensing collaboration cluster to obtain multiple feature indicators. That is, from the original data that has been confirmed as valid data through neighborhood mutual verification, feature indicators representing the status of the device or environment are calculated according to preset extraction rules.

[0042] In this embodiment, for example, the temperature rise rate is extracted from the effective temperature data, and the average smoke concentration is extracted from the effective smoke concentration data; For valid temperature data, through and Extract the average temperature separately and temperature rise rate The corresponding feature indicators, among which, To obtain valid data from the j-th temperature sensor within the cooperative cluster at time t, and They are the current time and Valid temperature data up to the current time. The preset time window length is 60 seconds in this embodiment; j=1, 2, ... , To perceive the amount of effective data from this type of sensor within the collaborative cluster. For effective data on smoke concentration, through and Extract the average smoke concentration separately and smoke concentration change rate The corresponding feature indicators; among them, To obtain the effective data of the j-th smoke concentration sensor within the cooperative cluster at time t; For the RMS current data, through and Extract the mean current separately and peak current The corresponding feature indicators; among them, To obtain the valid data of the j-th current sensor within the cooperative cluster at time t, max() represents the operation of taking the maximum value.

[0043] The feature indicators are pre-divided into multiple safety dimensions based on the attributes of the monitored objects. In this embodiment, the multiple safety dimensions include at least electrical safety dimension, fire safety dimension, structural safety dimension, civil defense safety dimension, and electromechanical safety dimension.

[0044] In this embodiment, the characteristic indicators corresponding to the electrical safety dimension include the effective value of current, voltage deviation, and harmonic distortion rate; the characteristic indicators corresponding to the fire safety dimension include ambient temperature, ambient smoke concentration, and CO concentration; the characteristic indicators corresponding to the structural safety dimension include vibration frequency, tilt angle, and crack width; the characteristic indicators corresponding to the civil defense safety dimension include intrusion alarm signals and unauthorized access control signals; and the characteristic indicators corresponding to the electromechanical safety dimension include elevator operating speed and water pump outlet pressure. When a characteristic indicator can be classified into multiple safety dimensions based on its physical meaning, for example, an increase in equipment surface temperature can be used to assess both electrical safety and fire safety, a copy of that characteristic indicator will be generated and independently participate in the subsequent trust fusion process of each safety dimension to which it is classified, with the calculations between the dimensions not interfering with each other.

[0045] Using a trust-based multi-source fusion method, the feature indicators contained in each security dimension are fused to obtain the preliminary security confidence components of the perception collaboration cluster in each security dimension. The preliminary security confidence components refer to the quantitative indicators of risk status under a single security dimension.

[0046] The fusion process is as follows: a basic trust assignment value is determined for each feature indicator. The basic trust assignment value includes a first trust value assigned to the security hypothesis, a second trust value assigned to the danger hypothesis, and a third trust value assigned to the entire set of the identification framework. After obtaining the basic trust assignment values ​​of the two feature indicators, they are fused in sequence to obtain the comprehensive trust assignment value. The specific method is as follows: For each security dimension, an identification framework consisting of security assumptions and danger assumptions is defined, and a trust assignment function is constructed for each feature index within that security dimension. The trust assignment function will output three basic trust assignment values ​​based on the feature index values, which are respectively assigned to the security assumption, the danger assumption, and the security or danger assumption that cannot be determined. The trust assignment functions of all feature indicators under this security dimension are fused sequentially using the orthogonal combination rule to obtain the comprehensive trust assignment value; During the fusion process, a conflict factor reflecting the degree of conflict between evidence is calculated simultaneously. When the conflict factor exceeds the preset conflict limit, a conflict redistribution rule is adopted, that is, all the trust values ​​that cannot be redistributed due to conflict are assigned to the "safe or dangerous" hypothesis representing ignorance. The result of the conflict redistribution continues to participate in the next fusion until the initial security confidence component of the security dimension is generated.

[0047] Taking fire safety as an example, the characteristic indicators under this fire safety dimension include the rate of temperature rise and the average smoke concentration; the preliminary safety confidence components of this fire safety dimension are obtained as follows: Define the universal set of identification frameworks F={H1,H2}, where H1 represents the safety assumption corresponding to "normal" and H2 represents the danger assumption corresponding to "fire". According to the preset membership mapping rules, the value of this feature index is converted into a basic trust assignment value, i.e., the trust assignment function m1, and the assignment method is as follows: The rate of temperature increase is denoted as r (W); When the temperature rise rate is lower than the preset safety limit T safe At that time, the first trust value is generated. =0.9, indicating a trust level of 0.9 to support "normal"; third trust value =0.1 indicates that the remaining 0.1 level of trust cannot be determined and is assigned to the entire set of the recognition framework.

[0048] When the rate of temperature rise exceeds the preset danger limit T danger At that time, a second trust value is generated. =0.85, third trust value =0.15; When the rate of temperature rise is between the safe limit and the hazardous limit, i.e., T safe ≤ ≤T danger Then, linear interpolation will be used for allocation. The formula for linear interpolation is as follows: Assigned to security assumptions The first trust value is Assigned to dangerous assumptions The second trust value ; Assigned to the entire set of recognition frames The third trust value ; In this embodiment, the temperature rise rate T safe Set to 2℃ / min, T danger Set to 8℃ / min.

[0049] Similarly, for the characteristic index corresponding to the average smoke concentration, the same principle is used to construct its trust assignment function m2, and the trust values ​​assigned to {H2}, {H1} and the supporting set are obtained respectively.

[0050] After obtaining the trust assignment functions for the two feature indicators, they are fused sequentially.

[0051] pass Calculate the conflict factor K, which reflects the degree of conflict between the two. In the formula, Represented as assigned to the security assumption First level of trust Represented as assigned to the dangerous hypothesis The second trust value, Represented as assigned to the security assumption First level of trust Represented as assigned to the dangerous hypothesis The second trust value; Let the basic trust assignment function correspond to the first characteristic indicator. In this embodiment, the first characteristic indicator is the temperature rise rate. The basic trust assignment function corresponding to the second feature indicator is the average smoke concentration; Extract the pre-set conflict limit K max In this embodiment, K max The value is 0.8, if K < K max This indicates that the conflict between pieces of evidence is acceptable. Following the orthogonal combination rule, the evidence is fused, and the resulting comprehensive trust allocation value m is calculated as follows: , , ; If K≥K max This indicates a high degree of conflict in the evidence. In this case, a conflict redistribution rule is adopted to transfer all conflicting parts to the entire set representing the unknown, thereby improving the robustness of the fusion result. The fusion rule is adjusted as follows: , , During the fusion process, the trust value of the empty set is always kept at 0, that is... After fusion, the comprehensive trust value corresponding to the hazard hypothesis H2 is taken as the preliminary safety confidence component of this perception collaboration cluster in the fire safety dimension, denoted as C. k,fire = .

[0052] Similarly, the above process is performed on the dimensions of electrical safety, structural safety, civil defense safety, and electromechanical safety respectively. Finally, the aggregation sub-node generates the preliminary safety confidence component vector of the sensing cooperative cluster. ;in, This represents the initial safety confidence component in the electrical safety dimension. This represents the initial safety confidence score component in the fire safety dimension. This represents the initial safety confidence component in the structural safety dimension. As a preliminary safety confidence component in the civil defense security dimension This represents the initial safety confidence component in the electromechanical safety dimension.

[0053] S4. The cloud server obtains the preliminary security confidence components of each sensing and collaborative cluster in each security dimension, and performs cross-cluster association and fusion in combination with the spatial grid of the building information model to generate the security risk level of the entire building. In this embodiment, the evaluation results of perception collaboration clusters from different spatial locations are spatially interpolated and superimposed using unified three-dimensional spatial coordinates provided by the Building Information Model. The specific method of cross-cluster association fusion is as follows: The cloud server preloads the building information model of the building and divides its three-dimensional space into a cubic grid with a side length of L, and the grid numbers are g=1,2,......,G.

[0054] For each security dimension d, based on the physical deployment location of the sensing nodes, the confidence component C of the sensing collaborative cluster k is... k,d The diffusion map is applied to the surrounding cubic mesh.

[0055] The weights are calculated using a perceptual diffusion function during mapping. This function simulates the attenuation of sensor sensing capability with spatial distance. The weights of grid g on the safety dimension d are also considered. Calculate using the following formula: Where the spatial distance between sensing node k and grid g is less than the effective sensing radius of the sensor. At that time, that is At that time, When the spatial distance between sensing node k and grid g is greater than or equal to the effective sensing radius of the sensor At that time, that is At that time, ; in, To determine the equivalent spatial location of the sensing node corresponding to dimension d in the sensing cooperative cluster k. Let g be the center coordinates of grid g. express arrive Euclidean distance, In this embodiment, the preset effective sensing radius of the sensor is defined. The value is set to 15 meters, which is considered the effective sensing radius of a single sensing node's sensor. The value can be adjusted according to the actual sensor power and building space structure.

[0056] Subsequently, the preliminary security confidence components of each relevant cluster on dimension d are... With corresponding spatial weights Multiply the products, and take the maximum value of the product from the set of all clusters that have effective perception capabilities in dimension d, as the risk component of grid g in dimension d. ; The risk components of grid g in each security dimension are weighted and fused, that is, by... To obtain its comprehensive risk value In the formula, d = 1, 2, ..., D, where D is the number of security dimensions. Based on the preset weighting coefficients of each security dimension; In this implementation, the weighting coefficient for fire safety is 0.35, the weighting coefficient for electrical safety is 0.25, the weighting coefficient for structural safety is 0.20, the weighting coefficient for civil defense safety is 0.10, and the weighting coefficient for electromechanical safety is 0.10.

[0057] The sensing diffusion function also obtains information about the physical components located between the sensing node and the grid g based on the building information model, and in response to the physical components being marked as blocking components, adjusts the weight of the sensing node to the grid g to a preset blocking value, or multiplies it by a decay factor less than 1.

[0058] For example, temperature sensing node T1 is located in a corridor, and grid cell g5 is located in an adjacent room, with a spatial distance of 3 meters, less than the effective radius of 15 meters. However, the cloud server retrieves component attributes from the BIM and finds that the straight path between T1 and g5 passes through two interior brick wall components, which are marked as barrier components in the BIM. In this case, for this temperature node, the weight of grid g5 is directly set to 0, instead of being calculated as 0.8 based on distance. Conversely, for grid g6, which has no barriers in the same corridor, even at a distance of 6 meters, the weight is still calculated as 0.6 based on distance decay. The resulting risk diffusion map strictly reflects the actual barrier situation of the building's physical space.

[0059] As a second embodiment of the present invention: In its specific implementation, compared to Embodiment 1, the technical solution of this embodiment differs from that of Embodiment 1 only in that, in this embodiment, the orthogonal combination rule and the conflict redistribution rule are implemented in the following manner: First, solve for the conflict factor K of the two sets of basic trust allocation functions m1 and m2; For the entire set F of the recognition framework, A⊂F and A= Any subset A of A=F, through Calculate its fusion trust value ; For the entire set F of the recognition framework, through Calculate its fusion trust value ; in, Let A be the complete set, and let B and C be the subsets corresponding to the fusion results. and A subset of the corresponding recognition framework; and , For an empty set, the compensation coefficient δ is determined by comparing the conflict factor K with the preset conflict upper limit: when the conflict factor K is less than the preset conflict upper limit, δ=0; when the conflict factor K is not less than the preset conflict upper limit, δ=K.

[0060] As an embodiment of the present invention: In specific implementation, compared with Embodiment 1 and Embodiment 2, the technical solution of this embodiment is to combine the solutions of Embodiment 1 and Embodiment 2. The only difference between the technical solution of this embodiment and Embodiment 1 and Embodiment 2 is that the comprehensive risk score of each cubic grid is divided into three local security risk levels of low risk, medium risk and high risk according to the preset risk threshold range. During normal communication, the linkage control commands are divided into three levels: when the local safety risk level is low, only information is pushed; when the local safety risk level is medium, local audible and visual alarms, smoke extraction, and partial power outages are triggered; when the local safety risk level is high, emergency lighting throughout the building, elevator emergency descent, and fire door release throughout the building are added.

[0061] Meanwhile, the cloud server takes the maximum value of the comprehensive risk value of all cube grids, generates the building-wide security status level, and pushes it to the management terminal. This building-wide security status level is not directly used for automatic device linkage.

[0062] Meanwhile, the aggregation sub-node monitors the communication status with the cloud server. When the duration of continuous communication interruption reaches a preset threshold, it enters the local autonomous mode. In the local autonomous mode, the aggregation sub-node, based on the preliminary security confidence components of the local sensing and collaborative cluster in each security dimension, directly drives the building equipment connected to the aggregation sub-node to perform security protection actions when the preliminary security confidence component of at least one security dimension reaches a preset local high-risk trigger threshold.

[0063] like Figure 2 As shown, this embodiment also provides a campus building safety status monitoring system based on IoT sensing layer collaboration. This system is used to execute a campus building safety status monitoring method based on IoT sensing layer collaboration. The system includes: Multiple sensing clusters, each containing multiple sensing nodes and a convergence sub-node with edge computing capabilities; within the same sensing cluster, multiple sensing nodes that collect the same physical quantity are deployed to form a group of similar sensors. The aggregation sub-node is used for clock synchronization within the sensing cluster, triggering each sensing node to synchronously collect physical quantities and generate raw data with synchronization timestamps at the same time. For each sensor group of the same type, neighborhood cross-verification is performed using the raw data of physically adjacent nodes within the same sensor group. Raw data of the corresponding physical quantity type of sensing nodes whose raw data deviates from the statistical distribution characteristic benchmark by more than the corresponding preset deviation threshold are discarded as invalid data; otherwise, they are retained as valid data. Feature extraction is performed on the valid data, and they are grouped according to preset security dimensions. Using a trust-based multi-source fusion method, each feature index contained in each security dimension is fused to obtain the preliminary security confidence components of the perception collaboration cluster in each security dimension; in the fusion process, the conflict factor is calculated, and when the conflict factor exceeds the preset conflict limit, the conflict redistribution rule of assigning the conflict trust value to the entire set of the identification framework is adopted. The cloud server is used to obtain the preliminary security confidence components of each sensing and collaborative cluster in each security dimension, and to perform cross-cluster association and fusion with the building information model spatial grid to generate the local security risk level of each cubic grid, and generate linkage control commands accordingly; at the same time, it generates the security status level of the entire building for status display. The local linkage control module, located on the aggregation sub-node, is used to execute commands under normal communication conditions. When it detects that the duration of continuous communication interruption with the cloud server has reached a preset threshold, it enters the local autonomous mode. Based on the preliminary security confidence components of this sensing and collaborative cluster in each security dimension, when any preliminary security confidence component reaches the local high-risk trigger threshold, it directly drives the building equipment connected to this aggregation sub-node to perform security protection actions.

[0064] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0065] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0066] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0067] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration, characterized in that, Includes the following steps: S1. Divide the physical space of the building into multiple sensing collaboration clusters, and deploy multiple sensing nodes and a convergence sub-node with edge computing capabilities in each sensing collaboration cluster. Sensing nodes are used to collect at least one physical quantity, and within the same sensing cluster, several sensing nodes that collect the same physical quantity are deployed to form a group of similar sensors. S2. Synchronize the clocks within the sensing cluster, and trigger the child nodes to collect physical quantities synchronously at the same time, generating raw data with synchronization timestamps. For each sensor group of the same type, the raw data of multiple sensing nodes that are physically adjacent within the same sensor group at the same synchronization time stamp are used to perform neighborhood mutual verification. The neighborhood mutual verification method is as follows: a statistical benchmark is determined based on the raw data of the neighboring sensing nodes within the same sensor group. If the deviation of the raw data of any sensing node from the statistical benchmark exceeds the preset deviation condition, the raw data is recorded as invalid data and discarded; otherwise, it is recorded as valid data. S3. The aggregation sub-node extracts features from all valid data within the perception collaboration cluster after neighborhood mutual verification, obtains multiple feature indicators, and divides each feature indicator into the corresponding dimension group according to the preset security dimension; using the trust-based multi-source fusion method, it performs fusion processing on each feature indicator contained in each security dimension to obtain the preliminary security confidence components of the perception collaboration cluster in each security dimension. S4. The cloud server obtains the preliminary security confidence components of each perception collaboration cluster in each security dimension, and performs cross-cluster association and fusion in combination with the spatial grid of the building information model to generate the local security risk level of each cubic grid. Based on the local safety risk level of each cubic grid, linkage control commands are output to the corresponding building equipment.

2. The method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration as described in claim 1, characterized in that, The specific method of neighborhood mutual verification is as follows: For the sensing node to be verified in a group of similar sensors, determine the neighborhood set of similar sensing nodes in the same group whose physical distance from it is less than the preset installation radius. The statistical baseline is determined by calculating the median of the raw data of the physical quantities of each sensing node in the neighborhood set as the statistical baseline, and calculating the median absolute deviation; the normalized deviation is determined based on the median and the median absolute deviation, and the normalized deviation is used as the quantification value of the degree of deviation. When the absolute value of the normalized deviation is greater than the preset deviation threshold, it means that the preset deviation condition has been exceeded. The neighborhood mutual verification method also includes: if the original data of the physical quantity of all sensing nodes in the neighborhood set exceeds the preset absolute danger threshold for that physical quantity type, then the original data of each node in the neighborhood set is directly retained.

3. The method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration as described in claim 2, characterized in that, Normalized deviation through The calculation shows that, To normalize the deviation, The raw data of the physical quantities of the sensing nodes to be verified. The median of the neighborhood set. The median absolute deviation of the neighborhood set. This is the distribution correction factor. It is a preset minimum positive number.

4. The method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration as described in claim 1, characterized in that, For each security dimension, the fusion processing method is as follows: Define an identification framework consisting of safety assumptions and danger assumptions, and determine a basic trust assignment value for each feature indicator within the safety dimension. The basic trust assignment value includes a first trust value assigned to the safety assumption, a second trust value assigned to the danger assumption, and a third trust value assigned to the entire set of the identification framework. The entire set of the identification framework represents the safety or danger state that cannot be determined. The basic trust assignment values ​​of all feature indicators within the security dimension are successively fused using the orthogonal combination rule, and a conflict factor reflecting the degree of evidence conflict is calculated at each fusion. When the conflict factor exceeds the preset conflict limit, the conflict redistribution rule is adopted to assign all the conflict trust values ​​generated by the current fusion to the entire set of the identification framework, and the result of the conflict redistribution continues to participate in the next fusion until the comprehensive trust allocation value of this security dimension is obtained. Based on the second trust value assigned to the dangerous hypothesis in the comprehensive trust allocation value, the preliminary security confidence component of the perception collaboration cluster in the security dimension is generated.

5. A method for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration, as described in claim 4, is characterized in that... The orthogonal combination rule and conflict redistribution rule are as follows: For two sets of basic trust assignment functions m1 and m2, find the conflict factor K between them; For any subset A under the recognition framework F, when A is a subset that is neither empty nor universal, by... Calculate its fusion trust value When A is the entire set of recognition frames F, through Calculate its fusion trust value ;in, Let A be the complete set, and let B and C be the subsets corresponding to the fusion results. and A subset of the corresponding recognition framework; and , For an empty set, the compensation coefficient δ is determined by comparing the conflict factor K with the preset conflict upper limit: when the conflict factor K is less than the preset conflict upper limit, δ=0; when the conflict factor K is not less than the preset conflict upper limit, δ=K.

6. The method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration as described in claim 1, characterized in that, Cross-cluster association and fusion based on the spatial grid of Building Information Modeling (BIM) includes: The cloud server preloads the building information model of the building and divides the building information model into several cubic grids; The initial security confidence components of each sensing cooperative cluster in each security dimension are mapped to the covered cubic grid. For a sensing cooperative cluster, when the initial security confidence component in a security dimension is allocated to each relevant cubic grid, the weight of the sensing node of the corresponding type in the sensing cooperative cluster to each cubic grid is calculated. For the mapping values ​​received by the same cubic grid from different sensing cooperative clusters in the same security dimension, the maximum value is taken as the risk component of the grid in that security dimension. For each grid cell, its mapping values ​​in each security dimension are weighted and summed to obtain the comprehensive risk value of the cubic grid. The overall building security status level is generated based on the comprehensive risk value of all cube grids. The security status level is divided according to a preset risk threshold range and is used for status display.

7. A method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration as described in claim 6, characterized in that, Weight of grid g on the security dimension d Calculate using the following formula: Among them, when At that time, ,when At that time, ; in, The preset effective sensing radius of the sensor. express arrive Euclidean distance, To determine the equivalent spatial location of the sensing node corresponding to dimension d in the sensing cooperative cluster k. Let g be the center coordinates of grid g, and k be the index of the sensing cooperative cluster.

8. A method for monitoring the safety status of buildings in a campus based on IoT sensing layer collaboration as described in claim 6, characterized in that, Based on the safety risk level, the method for outputting linkage control commands to the corresponding building equipment is as follows: for each cube grid, the comprehensive risk value of the cube grid is divided into three local safety risk levels: low risk, medium risk, and high risk according to the preset risk threshold range. Under normal communication conditions, for any target area covered by a cubic grid, when the local safety risk level of the building is low, an early warning message is pushed to the management terminal; when the local safety risk level of the building is medium, the edge side drives the corresponding zone's audible and visual alarms and / or smoke exhaust fans and / or power circuit breakers to operate; when the local safety risk level of the building is high, the cloud also coordinates to drive the building's emergency lighting, elevator emergency landing, and fire door release devices.

9. A method for monitoring the safety status of buildings in a park based on IoT sensing layer collaboration, as described in claim 1, is characterized in that, The aggregation sub-node monitors the communication status with the cloud server. When the duration of continuous communication interruption reaches a preset threshold, it enters the local autonomous mode. In the local autonomous mode, the aggregation sub-node, based on the preliminary security confidence components of this sensing and collaborative cluster in each security dimension, directly drives the building equipment connected to the aggregation sub-node to perform security protection actions when any preliminary security confidence component reaches the preset local high-risk trigger threshold.

10. A campus building safety status monitoring system based on IoT sensing layer collaboration, the system being used to execute the campus building safety status monitoring method based on IoT sensing layer collaboration as described in any one of claims 1-9, characterized in that, The system includes: Multiple sensing collaborative clusters, each containing multiple sensing nodes and a convergence sub-node with edge computing capabilities; The aggregation sub-node is used to synchronize the clock within the sensing cluster, triggering each sensing node to synchronously collect data and perform neighborhood mutual verification, feature extraction, and fusion processing to obtain the initial security confidence component. The cloud server is used to obtain the preliminary security confidence components of each sensing and coordination cluster, combine them with the building information model spatial grid to perform cross-cluster association and fusion, generate the local security risk level of each cubic grid, and generate linkage control commands accordingly; at the same time, it generates the security status level of the entire building for status display. The local linkage control module, integrated into the aggregation sub-node, is used to execute direct drives in local autonomous mode when communication is interrupted.

Images