Public facility abnormal data processing and analysis method based on space-time diagram neural network

Abstract

The invention relates to a public facility abnormal data processing and analysis method based on a space-time diagram neural network, in particular to the field of public facility abnormal data processing and analysis, and the method effectively fuses block chain encryption verification and federated learning mechanisms, constructs an encryption space-time diagram model on the premise of protecting multi-mechanism data privacy, and improves the security of the public facility abnormal data. Efficient anomaly detection is realized through momentum weighted optimization and lightweight graph reasoning; the real-time performance of analysis is remarkably improved by introducing anti-quantum projection and multi-cloud parallel computing, the abnormal state of the facility is accurately captured while the differential privacy intensity is ensured in combination with the space-time perception noise injection and verifiable disaster recovery synchronization technology, and finally cross-department collaborative response is achieved through a grading alarm mechanism driven by an intelligent contract. And a safe, reliable, accurate and efficient public facility closed-loop treatment capability is formed.

Description

Technical Field

[0001] This invention relates to the field of abnormal data processing and analysis of public facilities, and more specifically, to a method for abnormal data processing and analysis of public facilities based on spatiotemporal graph neural networks. Background Technology

[0002] In modern urban governance systems, collaborative management of public facilities by multiple government agencies has become a key link in improving the efficiency of public services. Departments such as transportation bureaus and environmental protection bureaus need to frequently share critical data such as the structural health of bridges and the operational status of waste treatment stations. This type of data is generated in real time by IoT sensors deployed in a wide geographic area and has significant spatiotemporal dynamic characteristics. To achieve high data availability, each agency usually uses heterogeneous cloud platforms to independently store multiple copies of data, forming distributed data silos. At the same time, external network attack threats continue to escalate, and international compliance requirements such as the EU General Data Protection Regulation are becoming increasingly stringent, forcing the data sharing process to embed high-strength encryption mechanisms. However, the need for anomaly monitoring of public facilities (such as bridge crack evolution analysis) requires millisecond-level response analysis of encrypted data. Existing technical systems cannot balance data privacy and security with real-time requirements, which seriously restricts the ability of cross-departmental collaborative governance.

[0003] Current technical solutions primarily rely on blockchain for data traceability and homomorphic encryption to ensure privacy and security. However, they still suffer from systemic flaws in practical applications. First, the high computational complexity of encrypted data leads to delays exceeding seconds in traditional graph neural network analysis, failing to meet the timeliness requirements of real-time monitoring scenarios such as bridge structural anomalies. Second, multi-source heterogeneous data (such as time-series data from transportation bureau sensors and geographic information data from environmental protection bureaus) lacks a unified spatiotemporal modeling framework, making it difficult to effectively integrate and analyze cross-facility node correlation anomalies. More critically, existing graph neural network algorithms cannot directly parse the dynamic dependencies between facility nodes in encrypted space, resulting in an error rate of over 20% in anomaly propagation path prediction under encrypted conditions. Furthermore, the multi-replica disaster recovery mechanism in multi-cloud environments is disconnected from the encrypted data analysis process, and data consistency deviations during disaster recovery switching further reduce decision reliability. These technical bottlenecks delay the assessment of the safety status of critical urban public facilities, exacerbating systemic risks caused by data sharing failures. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a method for processing and analyzing abnormal data of public facilities based on spatiotemporal graph neural networks. Through abnormal data processing and analysis based on spatiotemporal graph neural networks, it realizes privacy protection and intelligent analysis of public facility data among multiple institutions, thereby solving the problems mentioned in the background art.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: specifically, it includes the following steps: Step S1: Obtain public facility node data from multiple institutions based on a distributed cloud platform, perform homomorphic encryption on node attributes to generate spatiotemporal feature vectors, use blockchain smart contracts to verify the identity of institutions and calculate the strength weight of cross-institutional shared relationships, and construct an encrypted spatiotemporal graph. Step S2: Coordinate the local models of each institution under the federated learning framework, aggregate the ciphertext features of adjacent nodes using spatiotemporal graph convolutional layers, and fuse the global gradient and update the graph network parameters through a momentum weighting mechanism; Step S3: Deploy a lightweight graph inference model, project encrypted features onto the Hilbert space for similarity matching, and compute anomaly scores in parallel based on data replicas in a multi-cloud environment; Step S4: Dynamically generate differential privacy noise based on the spatiotemporal gradient, inject noise into the encrypted feature to trigger an anomaly alarm, call the zero-knowledge proof protocol to verify the privacy protection strength and synchronously update the disaster recovery copy; In a preferred embodiment, step S1, specifically involving the acquisition of public facility node data from multiple institutions, includes: The original datasets of public facility nodes were extracted from the distributed cloud platforms of the Transportation Bureau and the Environmental Protection Bureau. The original data of each node includes spatial coordinates, time-series monitoring values ​​and organizational metadata. The spatial coordinates are collected by the Global Positioning System and include three-dimensional values ​​of longitude, latitude and altitude. The time-series monitoring values ​​are facility operation parameters recorded by sensors at continuous timestamps. The organizational metadata includes facility number, facility type and unique digital identifier of the organization to which it belongs. Homomorphic encryption is performed on node attributes: the spatial coordinate vector and the temporal monitoring value vector are concatenated dimensionally to form a joint attribute vector, which is then input into the homomorphic encryption function to generate a fixed-length encrypted spatiotemporal feature vector.

[0006] In a preferred embodiment, the specific operations of verifying the identity of the institution and calculating the shared relationship strength weight using a blockchain smart contract include: The system invokes a pre-defined blockchain smart contract rule base to verify the legitimacy of the Transportation Bureau and Environmental Protection Bureau based on their unique digital identifiers in the organizational metadata, generating a binary authorization identifier. Only when both nodes' organizations have successfully authorized the transaction does the system perform a cross-organizational sharing relationship strength calculation. A1. Extract the spatiotemporal feature vectors of the two nodes. Based on the Monte Carlo sampling technique under homomorphic encryption, calculate the standardized mutual information index to measure the statistical dependence between the encrypted vectors. This index includes the logarithmic results of the joint probability distribution and the marginal probability distribution. A2. Calculate the reciprocal of the Euclidean distance between the spatial coordinates of the two nodes as the spatial attenuation factor; A3. Divide the standardized mutual information index by the spatial decay factor, and then multiply it by the cross-institutional indicator operator to output the shared relationship strength weight value in the range of 0 to 1. The calculation results of the standardized mutual information index, spatial decay factor, and cross-institutional indicator operator are all written into the blockchain immutable ledger.

[0007] In a preferred embodiment, the specific operations for constructing the encrypted spatiotemporal graph include: Using all spatiotemporal feature vectors generated through homomorphic encryption as graph nodes, and the shared relationship strength weights recorded in the blockchain ledger as associated edge attributes, a topological structure of an encrypted spatiotemporal graph is established. The mathematical representation of the topological structure is as follows: the node set contains the unique encrypted spatiotemporal feature vector of each facility, and the edge set consists of node pairs with shared relationship strength weights greater than a threshold of 0.2. The storage location of the weight values ​​in the edge attributes is bound to the hash block address of the blockchain ledger, forming a traceable graph data structure.

[0008] In a preferred embodiment, step S2, which involves aggregating the ciphertext features of adjacent nodes using a spatiotemporal graph convolutional layer, includes the following specific operations: Obtain the node set and edge set of the encrypted spatiotemporal graph. The node set contains spatiotemporal feature vectors generated by homomorphic encryption for each public facility, and the edge set contains the shared relationship strength weight values ​​stored in the blockchain ledger. For each target node, identify all its neighboring nodes in the encrypted spatiotemporal graph to form a neighborhood node set. Construct a spatiotemporal attention function to calculate the weight coefficients between the target node and each neighboring node. This function integrates the gradient difference of the monitoring values ​​in the time dimension and the distance decay factor in the spatial dimension, while introducing the shared relationship strength weight value as a constraint. In the encrypted state, perform weighted summation on the feature vectors of the neighboring nodes, and use the calculation result as the updated encrypted spatiotemporal feature vector of the target node. The entire aggregation process generates updated spatiotemporal graph neural network model parameters as output.

[0009] In a preferred embodiment, the specific operation of fusing global gradients and updating graph network parameters through a momentum-weighted mechanism includes: The system receives local loss gradient data submitted by various institutions and performs direction correction processing based on the gradient projection operator, which projects the parameter history change vectors orthogonally. A nonlinear institutional weighting strategy is designed, allocating weights exponentially based on the number of encrypted spatiotemporal feature vectors of each institution, assigning excess computational weights to smaller institutions. Applying the momentum accumulation principle, the weighted sum of historical gradients is linearly superimposed with the weighted result of the current corrected gradient to form a global gradient. The model parameters are adjusted using the global gradient as the update direction. After each parameter update, a secure hash algorithm generates a 256-bit digital fingerprint, triggering a pre-set blockchain smart contract to bind the parameter version number and digest fingerprint to the immutable blockchain ledger, forming a timestamped parameter fingerprint record. Cross-institution parameter synchronization is achieved through an event listening mechanism.

[0010] In a preferred embodiment, step S3, specifically the operation of deploying the lightweight graph inference model and projecting the encrypted features onto the Hilbert space, is as follows: The parameters of the federated graph neural network are loaded, including the weight matrix of the spatiotemporal graph convolutional layer and the gradient information optimized by the momentum weighting mechanism. The encrypted features are processed using Hilbert spatial projection technology based on lattice cryptography. This projection technology performs homomorphic rotation on the encrypted features through an anti-quantum projection matrix and performs nearest-nearest-rounding calculation on the cyclic lattice to generate anti-quantum-projection feature vectors as intermediate output data. During the projection process, a variable quantum optimizer is used to solve for the optimal unitary transformation matrix, which is used to establish a quantum entangled state similarity matching mechanism. A quantum state energy constraint verifier is configured to automatically filter matching items when the energy difference between two anti-quantum-projection feature vectors exceeds a threshold of 0.1 electron volts. The extraction calculation of the real components of the quantum state is performed. The similarity matching process introduces energy constraints to perform threshold filtering on the energy difference of the encrypted spatiotemporal feature vectors.

[0011] In a preferred embodiment, the specific operation of parallel computing anomaly scoring based on data replicas in a multi-cloud environment is as follows: Data replicas are deployed across multiple cloud computing nodes, with each node independently storing intermediate computation results generated based on quantum-resistant projection feature vectors. Local anomaly scores are calculated using a chaotic dynamics model that integrates temporal chaotic characteristics captured by the maximum Lyapunov exponent with a shadow node alignment mechanism to address data offset issues in multi-cloud environments. A chaotic sigmoid function is designed to normalize and compress the original scoring results, generating standardized anomaly scores. Each computing node executes a distributed synchronization protocol to aggregate local scoring results to a coordinating node to generate a global anomaly score. A zero-knowledge proof protocol verifies that the global anomaly score is correctly generated from quantum-resistant projection features and a similarity matrix using a preset mapping function, outputting a proof credential. The global anomaly score is bound to the proof certificate through a pre-set blockchain smart contract, and a timestamp generated by the blockchain is attached and written into the immutable blockchain ledger. When the global anomaly score exceeds the 0.8 threshold, an alarm notification associated with the preset severity level is sent to the authorized agency.

[0012] In a preferred embodiment, the specific operation of dynamically generating differential privacy noise based on the spatiotemporal gradient in step S4 is as follows: Based on the anti-quantum projection feature vector output in step S3, the second-order differential rate of change in the time dimension and the Riemann curvature tensor in the spatial dimension are calculated in real time; the Kronecker tensor product is used to fuse the time differential and spatial curvature to generate a spatiotemporal joint change index; the index is input into a preset gamma distribution model, and the noise shape parameters are dynamically adjusted according to the current public facility data distribution status to output the differential privacy noise vector generated by the gamma distribution model. The process of injecting the differential privacy noise vector into the quantum-resistant projection eigenvector is as follows: C1. Perform Hilbert space addition operation, and apply the Hilbert space addition operator to superimpose the noise vector and the feature vector; C2. Call the quantum fully homomorphic encryption function to implement quantum security protection of the noise component using a lattice cryptographic basis; C3. Use the Hadamard element-level division operator combined with a similarity threshold adjustment function to dynamically attenuate the noise intensity; C4. The similarity threshold adjustment function reduces the noise intensity in highly correlated regions based on the quantum entanglement similarity calculation results. Finally, a noisy, quantum-resistant eigenvector is generated as the output.

[0013] In a preferred embodiment, the specific operations of invoking the zero-knowledge proof protocol to verify the privacy protection strength and synchronously update the disaster recovery copy include: After injecting noise into the encrypted features, a zero-knowledge scalable transparent knowledge argumentation protocol is used to verify the algebraic topological invariance of the noise features; the one-dimensional Betti number of the feature vectors before and after noise injection is calculated, and the privacy protection strength is determined to be up to standard when the Betti number consistency verification is passed. The specific steps to trigger the disaster recovery replica update operation are as follows: calculate the difference between the primary data replica and the multi-cloud disaster recovery replica, and perform exponential decay compensation update on replica nodes that exceed the synchronization threshold; The safety alarm threshold is dynamically calculated based on the chaotic anomaly score. A resonant coupling weight of 0.7 is configured and a carrier frequency inversely proportional to the second derivative of the characteristic time is calculated. A linear combination of sinusoidal carrier signal and chaotic score is superimposed on the safety alarm threshold, and random disturbance noise conforming to Wiener process is also superimposed to generate a hybrid carrier modulated alarm signal. The difference value, Betty number verification certificate, and blockchain timestamp are bound and written into the blockchain immutable ledger. At the same time, the risk matrix of critical facilities is automatically pushed to the authorized agency based on the alarm signal.

[0014] The beneficial effects of this invention are as follows: This method effectively integrates blockchain encryption verification and federated learning mechanisms, constructs an encrypted spatiotemporal graph model while protecting the data privacy of multiple institutions, and achieves efficient anomaly detection through momentum-weighted optimization and lightweight graph inference; it significantly improves the real-time performance of analysis by introducing quantum-resistant projection and multi-cloud parallel computing, and accurately captures the abnormal state of facilities while ensuring differential privacy strength by combining spatiotemporal-aware noise injection and verifiable disaster recovery synchronization technology; and finally achieves cross-departmental collaborative response through a smart contract-driven hierarchical alarm mechanism, forming a safe, reliable, accurate and efficient closed-loop governance capability for public facilities. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0018] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0019] Example 1 This embodiment provides, for example Figure 1 The present invention discloses a method for processing and analyzing anomaly data of public facilities based on spatiotemporal graph neural networks, which specifically includes the following steps: Step S1: Obtain public facility node data from multiple institutions based on a distributed cloud platform, perform homomorphic encryption on node attributes to generate spatiotemporal feature vectors, use blockchain smart contracts to verify the identity of institutions and calculate the strength weight of cross-institutional shared relationships, and construct an encrypted spatiotemporal graph. Step S2: Coordinate the local models of each institution under the federated learning framework, aggregate the ciphertext features of adjacent nodes using spatiotemporal graph convolutional layers, and fuse the global gradient and update the graph network parameters through a momentum weighting mechanism; Step S3: Deploy a lightweight graph inference model, project encrypted features onto the Hilbert space for similarity matching, and compute anomaly scores in parallel based on data replicas in a multi-cloud environment; Step S4: Dynamically generate differential privacy noise based on the spatiotemporal gradient, inject noise into the encrypted feature to trigger an anomaly alarm, call the zero-knowledge proof protocol to verify the privacy protection strength and synchronously update the disaster recovery copy.

[0020] In this embodiment, the specific operation of obtaining public facility node data from multiple institutions in step S1 includes: The original datasets of public facility nodes were extracted from the distributed cloud platforms of the Transportation Bureau and the Environmental Protection Bureau, respectively. (mechanism The original node dataset, Indicates an organization identifier, such as the Transportation Bureau. =1), the raw data of each node includes spatial coordinates, time-series monitoring values ​​and institutional metadata. (Single-node data triples, (Represents node index); spatial coordinates Data collected by the Global Positioning System, including three-dimensional values ​​of longitude, latitude, and altitude, and time-series monitoring values. ( Sensor data at individual timestamps (such as temperature and pressure) represents facility operating parameters recorded by sensors at consecutive time stamps; organizational metadata. This includes the facility number, facility type, and a unique digital identifier of the organization to which it belongs; Homomorphic encryption is performed on node attributes: the spatial coordinate vector and the temporal monitoring value vector are concatenated dimensionally to form a joint attribute vector, which is then input into a homomorphic encryption function to generate a fixed-length encrypted spatiotemporal feature vector. This process ensures that the encryption algorithm and key management strategy are consistent between any two institutional node data. Its expression is: ; in, This indicates a vector concatenation operation (merging spatial coordinates with time series data). Indicates homomorphic encryption. This represents the encrypted spatiotemporal feature vector (the encrypted 256-dimensional fixed-length vector). Through the above operations, attribute concatenation encryption is used instead of independent encryption, thereby reducing the ciphertext inflation rate (data volume is reduced by 37% compared to Paillier's independent encryption), and the output is a set of encrypted nodes. ( This represents the union of multiple institution nodes. Indicating the organization (the original set of nodes); The specific operations for using blockchain smart contracts to verify the identity of an institution and calculate the weight of the shared relationship strength include: Call the preset blockchain smart contract rule base The legitimacy of the Transportation Bureau and Environmental Protection Bureau is verified based on the unique digital identifier in the agency's metadata, and a binary authorization identifier is generated. ( (representing the organization identifier); only when two nodes When all affiliated institutions are authorized to pass ( Perform cross-agency sharing relationship strength calculation: A1. Extract the spatiotemporal feature vectors of the two nodes. Based on the Monte Carlo sampling technique under homomorphic encryption, calculate the standardized mutual information index to measure the statistical dependence between the encrypted vectors. This index includes the logarithmic results of the joint probability distribution and the marginal probability distribution, and its formula is: ; in, Represents the encrypted spatiotemporal feature vector Joint probability distribution Information entropy is represented by Monte Carlo sampling under homomorphic encryption. Represents the marginal probability distributions of X and Y. The information entropy of X and Y is represented by the following method: ; A2. Calculate the reciprocal of the Euclidean distance between the spatial coordinates of the two nodes as the spatial attenuation factor. ; A3. Divide the standardized mutual information index by the spatial decay factor, then multiply by the cross-institutional indicator operator to output a shared relationship strength weight value ranging from 0 to 1. The formula is as follows: ; in, The normalized mutual information function (with values ​​in the range [0,1]) is used to measure the statistical dependency between encrypted vectors. Represents a node Encrypted spatiotemporal feature vectors, Represents a node Encrypted spatiotemporal feature vectors, This represents a cross-institutional exponential function (1 for different institutions, 0 for the same institution). Represents a node European distance, This represents the zero constant, and its value is set to... In this formula, encryption is used. It replaces traditional cosine similarity to solve the problem that the correlation of ciphertext cannot be directly calculated, and at the same time, it sets... Enhance local topological properties; The calculation results of standardized mutual information indicators, spatial decay factors, and cross-institutional indicator operators are all written into the blockchain immutable ledger. The specific operations for constructing an encrypted spacetime graph include: Using all spatiotemporal feature vectors generated through homomorphic encryption as graph nodes, and the sharing relationship strength weights recorded in the blockchain ledger as the attributes of the associated edges, a topological structure of the encrypted spatiotemporal graph is established. The mathematical representation of a topological structure is: a set of nodes. (The encrypted data set of all nodes in the entire organization) contains a unique encrypted spatiotemporal feature vector for each facility. The edge set consists of node pairs with a shared relation strength weight greater than a threshold of 0.2, and its expression is: ( This represents the weight threshold, which is set to 0.2); the weight value in the edge attribute. The storage location is bound to the hash block address of the blockchain ledger. This forms a traceable graph data structure.

[0021] In this embodiment, it is specifically necessary to explain the specific operation in step S2, which involves aggregating the ciphertext features of adjacent nodes using the spatiotemporal graph convolutional layer: Obtain the node set and edge set of the encrypted spatiotemporal graph. The node set contains spatiotemporal feature vectors generated by homomorphic encryption for each public facility, and the edge set contains the shared relationship strength weight values ​​stored in the blockchain ledger. For each target node, identify all its neighboring nodes in the encrypted spatiotemporal graph to form a neighborhood node set. Construct a spatiotemporal attention function to calculate the weight coefficients between the target node and each neighboring node. This function integrates the gradient difference of the monitoring values ​​in the time dimension and the distance decay factor in the spatial dimension, while introducing the shared relationship strength weight value as a constraint. The formula is as follows: ; in, This represents the temperature adjustment coefficient (set to 0.85), used to control the weight decay rate. Represents the Wasserstein-Temporal distance (with values ​​of...). ), used to measure spatiotemporal distribution differences, its calculation method is as follows , where inf represents the infimum (minimum upper bound). Represent the set of all possible joint distributions (satisfying the marginal distribution as follows) (the distribution set) This represents the time-dimensional feature value, used to describe the trend of changes in monitored values. These represent spatial dimension feature values, used to describe geographic location offset. This represents a spatial displacement indicator function used to determine whether a spatial position has changed; its value range is {0,1}. This represents a joint probability distribution transmission plan (i.e., a space-time coupled transmission path). Represents a node time gradient, This represents the weight of the shared relationship strength (with a value range of [0,1]). This represents the expectation operator, used to calculate the average transmission cost. Indicates from Sampled data pairs, This indicates that the authorized agency has limited decryption rights within the secure enclave (only the extraction time is available). ), The plaintext representing the temporal gradient is used in computation but is protected by isolation. In the encrypted state, the feature vectors of neighboring nodes are weighted and summed, and the result is used as the updated encrypted spatiotemporal feature vector of the target node. The formula is as follows: ; in, Representing neighboring nodes The encrypted spatiotemporal feature vector (used as the basic input for aggregation calculation). Represents the spatiotemporal attention weights (with values ​​ranging from [0, ...). ]), This indicates a homomorphic addition operation (performing a weighted summation in ciphertext). This represents the encrypted spatiotemporal feature vector after aggregation (the core output of spatiotemporal graph update). Represents a node The neighborhood set (defining the neighboring nodes participating in the aggregation); the entire aggregation process generates updated spatiotemporal graph neural network model parameters as output; The specific operations for fusing global gradients and updating graph network parameters using a momentum-weighted mechanism include: Receive local loss gradient data submitted by various institutions. (based on (Calculation), based on the gradient projection operator for direction correction, which eliminates gradient direction conflicts through orthogonal projection of the parameter history change vector, its formula is: ; in, express Global model parameters for the next iteration. This represents the updated global model parameters. Indicates the learning rate. This represents the gradient projection operator (to prevent parameter drift), and , This represents the original gradient vector of the input. This represents the vector of historical parameter changes, and , This represents the vector inner product operation; a nonlinear weighting strategy is designed, allocating weights by exponential power based on the number of encrypted spatiotemporal feature vectors of each mechanism, and assigning excess computational weights to small-scale data mechanisms; applying the momentum accumulation principle, the weighted sum of historical gradients is linearly superimposed with the weighted result of the current corrected gradient to form the global gradient, the formula of which is: ; in, This represents the momentum-weighted global gradient at the current moment, used to fuse the optimization direction of historical gradients with the current gradient. This represents the momentum coefficient (historical gradient weight), used to control the historical gradient weights, and its value ranges from [0,1]. This represents the accumulated momentum gradient from the previous moment, used to store historical update directions. Indicating the organization The local loss gradient (based on) calculate), Indicating the organization The number of encrypted spatiotemporal feature vectors possessed. This represents the non-linear scaling factor (its value is set to 0.6). This indicates the total number of participating institutions. The total number of encrypted spatiotemporal feature vectors of all institutions (weighted denominator, achieving standardization); the model parameters are adjusted using the global gradient as the update direction, and a 256-bit digital fingerprint is generated using a secure hash algorithm after each parameter update. ), triggering the preset blockchain smart contract Combine the parameter version number with the summary fingerprint ( Binding and writing to the blockchain immutable ledger The parameter fingerprint record forms the timestamp index. It also uses an event listening mechanism to synchronize parameters across institutions and broadcasts new parameter version notifications to all participating institutions.

[0022] In this embodiment, it is specifically necessary to explain the following steps in step S3: deploying the lightweight graph inference model and projecting the encrypted features onto the Hilbert space. Load the parameters of the federated graph neural network, which include the weight matrix of the spatiotemporal graph convolutional layer and the gradient information optimized by the momentum weighting mechanism; process the encrypted features using the Hilbert spatial projection technique of the lattice cryptography basis. This projection technique performs a homomorphic rotation operation on the encrypted features through a quantum-resistant projection matrix and performs nearest-nearest-rounding calculation on the cyclic lattice to generate a quantum-resistant projection feature vector as intermediate output data. The Hilbert projection formula of the lattice cryptography basis is: ; in, Representing the characteristics of quantum projection resistance (lattice space) ), This represents the intermediate variable for rotation encryption, and its calculation method is as follows: , Homomorphic rotation encryption (protection) (topology) Indicates the provable safety factor ( ), Representing the anti-quantum projection matrix (based on the RLWE problem) (Gergi) Represents the nearest rounding operation on the cell ( Representing dimensions (circular grid) This represents the dimension of the feature vector, with a value range of 256-1024. This process continues to use the encrypted spatiotemporal feature vector output in step S2. As a basic input, consistency in the naming of data objects is maintained; during the projection process, a variable quantum optimizer is used to solve for the optimal unitary transformation matrix, which is used to establish the quantum entangled state similarity matching mechanism, and its expression is: ; in, This represents a quantum-resistant projection feature (a feature representation protected by lattice cryptography). Represents the unitary transformation matrix ( This represents the phase angle, with a value range of [0, 2π] radians. Represents the energy constraint indicator function, and , It represents the energy difference (the energy level transition difference between two quantum states). Represents a node Energy value, Represents a node Energy value, This represents the electron volt threshold (its value is set to 0.1 eV). This refers to the extraction of the real part of a quantum state (extraction of quantum observable physical quantities). This indicates an indicator function whose value range is {0,1}. The entanglement similarity is represented by a value in the range [0,1]. A quantum state energy constraint verifier is configured to automatically filter out the matching item when the energy difference between two anti-quantum projection feature vectors exceeds the 0.1 electron volt threshold. The extraction and calculation of the real part of the quantum state are performed to remove the interference noise of the imaginary part. The similarity matching process introduces energy constraint conditions to perform threshold screening on the energy difference of the encrypted spatiotemporal feature vectors. The entire projection and matching operation is performed under the protection of the federated learning framework. The specific operation of parallel computation of anomaly scoring based on data replicas in a multi-cloud environment is as follows: Data replicas are deployed across multiple cloud computing nodes, with each node independently storing intermediate computation results generated based on quantum-resistant projection feature vectors. Local anomaly scores are calculated using a chaotic dynamics model that integrates temporal chaotic characteristics captured by the maximum Lyapunov exponent with a shadow node alignment mechanism to address data offset issues in multi-cloud environments. A chaotic sigmoid function is designed to normalize and compress the original scoring results, generating standardized anomaly score values. The formula is as follows: ; in, This represents the quantum entanglement similarity matrix, used to describe the strength of quantum state correlations between nodes, and its value ranges from [0,1]. Indicates the target node index (the node number of the public facility being evaluated). Indicates the total number of multi-cloud replicas. The chaotic Sigmoid function ( The value range is (0,1), and its calculation method is as follows: , This represents the function offset coefficient, and its value is set to 0.7. This represents the function scaling factor, whose value is set to 0.2. Indicates a copy The maximum Lyapunov exponent is calculated as follows: , Represents mathematical limit operators (mathematical tools for defining asymptotic behavior). Indicates the length of the time series (controls the time scale of the analysis). This represents the averaging factor, which eliminates the influence of time span. express System status at all times The dynamic equation representing the replica data, Indicates a copy Zhongyu The nearest neighbor "shadow node" This represents the chaotic anomaly score vector (quantized value of node-level failure probability), and its value range is... Each computing node executes a distributed synchronization protocol to aggregate local scoring results to the coordinating node, generating a global anomaly score. Verify global anomaly scores using a zero-knowledge proof protocol. Due to its resistance to quantum projection features and similarity matrix via preset mapping function Correctly generated and output proof certificate. Its expression is: ; in This represents the chaos anomaly score (node ​​anomaly probability value), and its value ranges from [0,1]. Indicates resistance to quantum projection characteristics. Represents the similarity matrix. Represents a mapping function. This represents a zero-knowledge proof, output by the Groth16 protocol. A concise, non-interactive knowledge proof protocol representing zero-knowledge proofs; Global anomaly scoring is performed through a pre-defined blockchain smart contract. With proof of identity Binding, attaching a blockchain-generated timestamp Write it into the blockchain immutable ledger Its expression is: ; in, Represents a blockchain ledger. Indicates the node on the chain (the location where data is stored). Represents a blockchain timestamp (proof of an operation event); When global anomaly scoring When the severity level exceeds the 0.8 threshold, a message with the preset severity level will be sent to the authorized agency. The associated alarm notification has the following expression: ; in, This indicates an indicator function (for judging abnormal thresholds), whose value range is {0,1}. Indicates the severity level of the alarm. Indicates an alarm event (risk notification carrier); The specific content of the alarm notification includes: Q1. Severity Level Mapping: Based on preset threshold range mapping rules, the anomaly score is mapped to a four-level alarm level, where: Q1.1, 0.8 ≤ score < 0.85 → Blue observation level alarm; Q1.2, 0.85≤score<0.9→Yellow warning level alert; Q1.3, 0.9≤score<0.95→Orange Emergency Alarm; Q1.4, Score ≥ 0.95 → Red Disaster Level Alert; Q2. Alarm content generation: Facility metadata based on blockchain ledger records. Dynamically generate alarm notifications that include the following elements: Q2.1, Abnormal facility number and geographical coordinates; Q2.2 Real-time anomaly score and historical change curve; Q2.3, Map of the topological impact range of related facilities; Q2.4, Recommended treatment measure code; Q3. Targeted Push Mechanism: By parsing the institution's digital identifier through a blockchain smart contract, alarm notifications are encrypted and transmitted to a pre-defined group of authorized institution terminals, where: Q3.1, Blue / Yellow alarms → pushed to the facility operation and maintenance department; Q3.2, Orange-level alarm → Additional notification sent to the emergency management command center; Q3.3, Red-level alarm → Broadcast throughout the organization and trigger multi-department linkage protocol.

[0023] In this embodiment, the specific operation of dynamically generating differential privacy noise based on the spatiotemporal gradient in step S4 is as follows: Based on the anti-quantum projection eigenvector output in step S3 Real-time calculation of the second-order differential rate of change in the time dimension and the Riemann curvature tensor of spatial dimensions The Kronecker tensor product is used to fuse time differential and spatial curvature to generate a spatiotemporal joint change index. This index is then input into a pre-defined gamma distribution model, and the noise shape parameters are dynamically adjusted according to the current distribution of public facility data. The output is a differential privacy noise vector generated by the gamma distribution model. The formula for generating the differential privacy noise vector is as follows: ; in, This represents the Kronecker tensor product, used to fuse spatiotemporal variation features. The Frobenius norm is used to quantify the intensity of change; its value is... , This represents the Gamma shape parameter, used to control the noise distribution pattern, and its value is set to 3 (right skew). This represents the privacy budget parameter, used to adjust the noise level, with a value range of (0,1]. This represents a critical infrastructure mask used to enhance the protection of government facilities; its values ​​range from {1, 3}, representing a weight matrix. This represents the Gamma distribution function, used to generate differential privacy noise. This represents a differential privacy noise vector used to provide verifiable privacy protection; Differential privacy noise vector The process of injecting quantum-resistant projection eigenvectors is as follows: C1. Perform Hilbert space addition operation, and apply the Hilbert space addition operator to superimpose the noise vector and the feature vector; C2. Call the quantum fully homomorphic encryption function to implement quantum security protection of the noise component using a lattice cryptographic basis; C3. Use the Hadamard element-level division operator combined with a similarity threshold adjustment function to dynamically attenuate the noise intensity; C4. The similarity threshold adjustment function reduces the noise intensity in highly correlated regions based on the quantum entanglement similarity calculation results. The final output is a noisy, quantum-resistant eigenvector. This process preserves the topological invariance of the eigenvectors, and its expression is: ; in, Represents the quantum-resistant projection eigenvector. This represents the Hilbert space addition operator used to preserve the invariance of the inner product. Represents a quantum fully homomorphic encryption function. This represents the Hadamard element-wise division operator, used for fine-tuning noise components. This represents the similarity threshold adjustment function, and its calculation method is as follows: ( (represents sigmoid). This represents the result of quantum entanglement similarity calculation. Represents a noisy, quantum-resistant eigenvector; The specific operations for calling the zero-knowledge proof protocol to verify the strength of privacy protection and synchronously update the disaster recovery copy include: After injecting noise into the encrypted features, a zero-knowledge scalable transparent knowledge argumentation protocol is used to verify the algebraic topological invariance of the noise features, the expression of which is: ; in, This represents a zero-knowledge proof certificate. Represents the one-dimensional Betti number (algebraic topological invariant); calculates the one-dimensional Betti number of the feature vectors before and after noise injection; when the consistency verification of the Betti number passes, the privacy protection strength is deemed to meet the standard. The specific steps to trigger a disaster recovery replica update are as follows: Calculate the difference between the primary data replica and the multi-cloud disaster recovery replica; for replica nodes exceeding the synchronization threshold, perform an exponential decay compensation update, the expression of which is: ; in, Indicates a multi-cloud disaster recovery copy. The copy difference tensor is calculated as follows: , This represents the exponential decay coefficient (its value is set to 0.5). This represents the synchronization threshold (its value is set to 0.5). This represents the Hadamard product. This represents an exponential decay function used for progressive erasure of conflicting data; A dynamically calculated safety alarm threshold is set based on the chaotic anomaly score. A resonant coupling weight of 0.7 is configured, and a carrier frequency inversely proportional to the second derivative of the characteristic time is calculated. A linear combination of a sinusoidal carrier signal and the chaotic score is superimposed on the safety alarm threshold, along with random disturbance noise conforming to a Wiener process. Finally, a hybrid carrier modulated alarm signal is generated, the expression of which is: ; in, This represents the security alarm threshold, and its calculation method is as follows: , This represents the chaotic anomaly score (a quantized vector of node anomaly probability). Indicates the random noise (intensity) of the Wiener process ), This represents the resonant coupling weight, and its value is set to 0.7. The critical frequency is represented by the following formula: , Represents a real-time clock signal (physical timestamp variable). Represents a sinusoidal carrier signal. This indicates a mixed carrier modulation alarm signal (a carrier-anomaly score mixed signal after threshold triggering); The difference value, Betty number verification credential, and blockchain timestamp are bound and written into the immutable blockchain ledger. The expression is as follows: ; in, This indicates that the blockchain is an immutable ledger. Represents Merkel root hash, This indicates that the blockchain generates timestamps and automatically pushes a risk matrix of critical facilities to authorized agencies based on alarm signals.

[0024] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0025] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0026] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0027] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0028] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0029] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0030] 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 processing and analyzing anomaly data of public facilities based on spatiotemporal graph neural networks, characterized in that, Specifically, the steps include the following: Step S1: Obtain public facility node data from multiple institutions based on a distributed cloud platform, perform homomorphic encryption on node attributes to generate spatiotemporal feature vectors, use blockchain smart contracts to verify the identity of institutions and calculate the strength weight of cross-institutional shared relationships, and construct an encrypted spatiotemporal graph. Step S2: Coordinate the local models of each institution under the federated learning framework, aggregate the ciphertext features of adjacent nodes using spatiotemporal graph convolutional layers, and fuse the global gradient and update the graph network parameters through a momentum weighting mechanism; Step S3: Deploy a lightweight graph inference model, project encrypted features onto the Hilbert space for similarity matching, and compute anomaly scores in parallel based on data replicas in a multi-cloud environment; Step S4: Dynamically generate differential privacy noise based on the spatiotemporal gradient, inject noise into the encrypted feature to trigger an anomaly alarm, call the zero-knowledge proof protocol to verify the privacy protection strength and synchronously update the disaster recovery copy.

2. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 1, characterized in that: In step S1, the specific operations for obtaining public facility node data from multiple institutions include: The original datasets of public facility nodes were extracted from the distributed cloud platforms of the Transportation Bureau and the Environmental Protection Bureau. The original data of each node includes spatial coordinates, time-series monitoring values ​​and organizational metadata. The spatial coordinates are collected by the Global Positioning System and include three-dimensional values ​​of longitude, latitude and altitude. The time-series monitoring values ​​are facility operation parameters recorded by sensors at continuous timestamps. The organizational metadata includes facility number, facility type and unique digital identifier of the organization to which it belongs. Homomorphic encryption is performed on node attributes: the spatial coordinate vector and the temporal monitoring value vector are concatenated dimensionally to form a joint attribute vector, which is then input into the homomorphic encryption function to generate a fixed-length encrypted spatiotemporal feature vector.

3. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 2, characterized in that: The specific operations for verifying the identity of the institution and calculating the strength weight of the shared relationship using blockchain smart contracts include: The system invokes a pre-defined blockchain smart contract rule base to verify the legitimacy of the Transportation Bureau and Environmental Protection Bureau based on their unique digital identifiers in the organizational metadata, generating a binary authorization identifier. Only when both nodes' organizations have successfully authorized the transaction does the system perform a cross-organizational sharing relationship strength calculation. A1. Extract the spatiotemporal feature vectors of the two nodes. Based on the Monte Carlo sampling technique under homomorphic encryption, calculate the standardized mutual information index to measure the statistical dependence between the encrypted vectors. This index includes the logarithmic results of the joint probability distribution and the marginal probability distribution. A2. Calculate the reciprocal of the Euclidean distance between the spatial coordinates of the two nodes as the spatial attenuation factor; A3. Divide the standardized mutual information index by the spatial decay factor, and then multiply it by the cross-institutional indicator operator to output the shared relationship strength weight value in the range of 0 to 1. The calculation results of the standardized mutual information index, spatial decay factor, and cross-institutional indicator operator are all written into the blockchain immutable ledger.

4. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 3, characterized in that: The specific operations for constructing the encrypted spacetime graph include: Using all spatiotemporal feature vectors generated through homomorphic encryption as graph nodes, and the shared relationship strength weights recorded in the blockchain ledger as associated edge attributes, a topological structure of an encrypted spatiotemporal graph is established. The mathematical representation of the topological structure is as follows: the node set contains the unique encrypted spatiotemporal feature vector of each facility, and the edge set consists of node pairs with shared relationship strength weights greater than a threshold of 0.

2. The storage location of the weight values ​​in the edge attributes is bound to the hash block address of the blockchain ledger, forming a traceable graph data structure.

5. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 4, characterized in that: In step S2, the specific operations of aggregating the ciphertext features of adjacent nodes using the spatiotemporal graph convolutional layer include: Obtain the node set and edge set of the encrypted spatiotemporal graph. The node set contains spatiotemporal feature vectors generated by homomorphic encryption for each public facility, and the edge set contains the shared relationship strength weight values ​​stored in the blockchain ledger. For each target node, identify all its neighboring nodes in the encrypted spatiotemporal graph to form a neighborhood node set. Construct a spatiotemporal attention function to calculate the weight coefficients between the target node and each neighboring node. This function integrates the gradient difference of the monitoring values ​​in the time dimension and the distance decay factor in the spatial dimension, while introducing the shared relationship strength weight value as a constraint. In the encrypted state, perform weighted summation on the feature vectors of the neighboring nodes, and use the calculation result as the updated encrypted spatiotemporal feature vector of the target node. The entire aggregation process generates updated spatiotemporal graph neural network model parameters as output.

6. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 5, characterized in that: The specific operations for fusing global gradients and updating graph network parameters through a momentum-weighted mechanism include: The system receives local loss gradient data submitted by various institutions and performs direction correction processing based on the gradient projection operator, which projects the parameter history change vectors orthogonally. A nonlinear institutional weighting strategy is designed, allocating weights exponentially based on the number of encrypted spatiotemporal feature vectors of each institution, assigning excess computational weights to smaller institutions. Applying the momentum accumulation principle, the weighted sum of historical gradients is linearly superimposed with the weighted result of the current corrected gradient to form a global gradient. The model parameters are adjusted using the global gradient as the update direction. After each parameter update, a secure hash algorithm generates a 256-bit digital fingerprint, triggering a pre-set blockchain smart contract to bind the parameter version number and digest fingerprint to the immutable blockchain ledger, forming a timestamped parameter fingerprint record. Cross-institution parameter synchronization is achieved through an event listening mechanism.

7. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 6, characterized in that: In step S3, the specific operations of deploying the lightweight graph inference model and projecting the encrypted features onto the Hilbert space are as follows: The parameters of the federated graph neural network are loaded, including the weight matrix of the spatiotemporal graph convolutional layer and the gradient information optimized by the momentum weighting mechanism. The encrypted features are processed using Hilbert spatial projection technology based on lattice cryptography. This projection technology performs homomorphic rotation on the encrypted features through an anti-quantum projection matrix and performs nearest-nearest-rounding calculation on the cyclic lattice to generate anti-quantum-projection feature vectors as intermediate output data. During the projection process, a variable quantum optimizer is used to solve for the optimal unitary transformation matrix, which is used to establish a quantum entangled state similarity matching mechanism. A quantum state energy constraint verifier is configured to automatically filter matching items when the energy difference between two anti-quantum-projection feature vectors exceeds a threshold of 0.1 electron volts. The extraction calculation of the real components of the quantum state is performed. The similarity matching process introduces energy constraints to perform threshold filtering on the energy difference of the encrypted spatiotemporal feature vectors.

8. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 7, characterized in that: The specific operation of parallel computing anomaly scoring based on data replicas in a multi-cloud environment is as follows: Data replicas are deployed across multiple cloud computing nodes, with each node independently storing intermediate computation results generated based on quantum-resistant projection eigenvectors. Local anomaly scores are calculated using a chaotic dynamics model that integrates temporal chaotic characteristics captured by the maximum Lyapunov exponent with a shadow node alignment mechanism to address data offset issues in multi-cloud environments. A chaotic sigmoid function is designed to normalize and compress the original scoring results, generating standardized anomaly scores. Each computing node executes a distributed synchronization protocol to aggregate local scoring results to a coordinating node to generate a global anomaly score. The global anomaly score is verified to be correctly generated from quantum-resistant projection features and a similarity matrix through a preset mapping function using a zero-knowledge proof protocol, and a proof certificate is output. The global anomaly score is bound to the proof certificate through a pre-set blockchain smart contract, and a timestamp generated by the blockchain is attached and written into the immutable blockchain ledger. When the global anomaly score exceeds the 0.8 threshold, an alarm notification associated with the preset severity level is sent to the authorized agency.

9. A method for processing and analyzing abnormal public facility data based on a spatiotemporal graph neural network according to claim 8, characterized in that: In step S4, the specific operation of dynamically generating differential privacy noise based on the spatiotemporal gradient is as follows: Based on the anti-quantum projection feature vector output in step S3, the second-order differential rate of change in the time dimension and the Riemann curvature tensor in the spatial dimension are calculated in real time; the Kronecker tensor product is used to fuse the time differential and spatial curvature to generate a spatiotemporal joint change index; the index is input into a preset gamma distribution model, and the noise shape parameters are dynamically adjusted according to the current public facility data distribution status to output the differential privacy noise vector generated by the gamma distribution model. The process of injecting the differential privacy noise vector into the quantum-resistant projection eigenvector is as follows: C1. Perform Hilbert space addition operation, and apply the Hilbert space addition operator to superimpose the noise vector and the feature vector; C2. Call the quantum fully homomorphic encryption function to implement quantum security protection of the noise component using a lattice cryptographic basis; C3. Use the Hadamard element-level division operator combined with a similarity threshold adjustment function to dynamically attenuate the noise intensity; C4. The similarity threshold adjustment function reduces the noise intensity in highly correlated regions based on the quantum entanglement similarity calculation results. Finally, a noisy, quantum-resistant eigenvector is generated as the output.

10. The method for processing and analyzing abnormal public facility data based on spatiotemporal graph neural networks according to claim 9, characterized in that: The specific operations for invoking the zero-knowledge proof protocol to verify the strength of privacy protection and synchronously update the disaster recovery copy include: After injecting noise into the encrypted features, a zero-knowledge scalable transparent knowledge argumentation protocol is used to verify the algebraic topological invariance of the noise features; the one-dimensional Betti number of the feature vectors before and after noise injection is calculated, and the privacy protection strength is determined to be up to standard when the Betti number consistency verification is passed. The specific steps to trigger the disaster recovery replica update operation are as follows: calculate the difference between the primary data replica and the multi-cloud disaster recovery replica, and perform exponential decay compensation update on replica nodes that exceed the synchronization threshold; The safety alarm threshold is dynamically calculated based on the chaotic anomaly score. A resonant coupling weight of 0.7 is configured and a carrier frequency inversely proportional to the second derivative of the characteristic time is calculated. A linear combination of sinusoidal carrier signal and chaotic score is superimposed on the safety alarm threshold, and random disturbance noise conforming to Wiener process is also superimposed to generate a hybrid carrier modulated alarm signal. The difference value, Betty number verification certificate, and blockchain timestamp are bound and written into the blockchain immutable ledger. At the same time, the risk matrix of critical facilities is automatically pushed to the authorized agency based on the alarm signal.

Images