Blockchain storage and tracing method and system for diving operation data

Abstract

The application provides a kind of diving operation data's blockchain storage and traceability method and system, the method comprises: the real-time data collected by multiple-source underwater sensor is divided into overlapping state unit, the dynamic directed topological atlas is constructed with the pre-sequence and post-sequence state characteristics as node in data center, and single-source diving time series state chain is reconstructed by de-looping and interpolation.The single-source state chain is converted into directed state snapshot branch, and an asynchronous state evolution network is constructed by merging at the logical intersection point through hash pointer.The state transition characteristics between adjacent nodes in the network are extracted before chaining, and cross-collision matching is carried out with the preset underwater physical constraint benchmark, and the abnormal nodes that fail to match are isolated to the independent branch with physical failure label.The complete network containing the isolated branch is packaged and chained.The application solves the technical problem that the original data does not have physical authenticity and logical consistency due to channel interference and data anomaly, so that the blockchain storage loses factual basis.

Description

Technical Field

[0001] This invention belongs to the field of blockchain technology, specifically relating to a blockchain-based method and system for storing and tracing diving operation data. Background Technology

[0002] During underwater diving operations, comprehensive data collection and recording are required using multi-source sensors deployed on divers, underwater robots, and the surrounding environment. This data is not only used for real-time monitoring but also serves as crucial evidence for accident tracing after the operation. To ensure the originality, integrity, and immutability of this critical operational data, blockchain technology is typically used for decentralized notarization. However, directly hashing and uploading raw underwater multi-source sensor data to the blockchain presents technical challenges in practice stemming from the physical processes of data acquisition and transmission. Underwater acoustic channels are typically time-varying, high-error-rate, and narrow-bandwidth channels. During data transmission from underwater sensors to the surface data center, data packets are highly susceptible to loss, out-of-order delivery, or corruption due to environmental noise interference. This results in the data center ultimately receiving time-series data streams from different sensors that are physically misaligned and may even contain a large amount of erroneous data. If this raw data, mixed with channel interference, is directly stored as notarization, blockchain technology cannot guarantee the authenticity and physical logic of the recorded operational process, nor can it trace the true results represented by the data. Summary of the Invention

[0003] This invention provides a blockchain-based method and system for storing and tracing diving operation data to solve the aforementioned technical problems.

[0004] In a first aspect, the present invention provides a blockchain-based method for storing and tracing diving operation data, the method comprising the following steps: Real-time operational data from multiple underwater sensors in diving operations are collected. A preset length of overlapping sliding window is set to divide the continuous real-time operational data into overlapping state units with nested context states. Multiple overlapping state units are then encapsulated and transmitted to the data center. By extracting the preceding and following state features reflecting the evolution of physical states within each overlapping state unit from the data center, a dynamic directed topological graph representing the succession relationship of underlying states is constructed in memory space using the preceding and following state features as nodes. In the dynamic directed topology graph, the path traversal algorithm is used to eliminate redundant loops caused by channel interference and perform breakpoint interpolation to extract the acyclic traversal path that can connect all valid nodes. Based on the acyclic traversal path, a single-source diving time-series state chain with physical time consistency is reconstructed. The single-source diving time-series state chain corresponding to each multi-source underwater sensor is transformed into a directed state snapshot branch containing hash pointer references. When different single-source diving time-series state chains logically intersect, the directed state snapshot branches are merged based on hash pointers to construct an asynchronous state evolution network that represents the evolution of integrated diving actions. When the asynchronous state evolution network is ready to submit evidence, the state transition feature parameters between adjacent nodes in the asynchronous state evolution network are extracted, and the state transition feature parameters are cross-collision matched with the preset underwater physical constraint benchmark. Abnormal nodes that fail to match are blocked and independently assigned to isolated evidence storage branches with physical failure markers. The complete asynchronous state evolution network, including isolated evidence storage branches, is hashed, packaged, and broadcast to the blockchain network for consensus on-chain. During the tracing phase, the topology of the asynchronous state evolution network is reverse-analyzed based on the root hash on the blockchain to trace back the physical evolution trajectory and abnormal state source of the diving operation data.

[0005] Optionally, the step of eliminating redundant loops caused by channel interference and performing breakpoint interpolation in the dynamic directed topology graph through a path traversal algorithm, extracting acyclic traversal paths that can connect all valid nodes, and reconstructing a single-source diving time-series state chain with physical time consistency based on the acyclic traversal paths includes the following steps: Identify the in-degree and out-degree of nodes in a dynamic directed topology graph, and use a path traversal algorithm to detect multipath redundant blocks in the dynamic directed topology graph based on the in-degree and out-degree of nodes. Evaluate the topological equivalence of multipath redundant tiles and eliminate multipath redundant tiles based on topological equivalence to generate a cleaned connected subgraph; Identify graph breakpoints caused by signal loss in connected subgraphs and perform breakpoint interpolation; Connect the connected subgraphs after breakpoint interpolation, extract the acyclic traversal path that can connect all valid nodes, and reconstruct the single-source diving time-series state chain with physical time consistency based on the acyclic traversal path.

[0006] Optionally, identifying graph breakpoints caused by signal loss in the connected subgraph and performing breakpoint interpolation includes the following steps: Extract the edge free nodes in the connected subgraph and calculate the state gradient difference between the edge free nodes and the target free nodes in the adjacent connected subgraphs; When the state gradient difference is less than the preset physical evolution extreme value, a virtual bridging state feature is generated. Virtual directed edges are constructed between edge free nodes and target free nodes by utilizing virtual bridging state characteristics; By connecting the corresponding connected subgraphs with virtual directed edges, the breakpoint interpolation of the graph breakpoints is completed.

[0007] Optionally, calculating the state gradient difference between an edge free node and a target free node in an adjacent connected subgraph includes the following steps: Read the first physical state vector contained in the edge free node, and read the second physical state vector contained in the target free node in the adjacent connected subgraph; Obtain the Euclidean distance between the first physical state vector and the second physical state vector in the data space; Divide the Euclidean distance by the theoretical interval steps between the edge free node and the target free node in the received sequence to obtain the unit evolution rate of change, and use the unit evolution rate of change as the state gradient difference between the edge free node and the target free node.

[0008] Optionally, the step of converting the single-source diving time-series state chain corresponding to each multi-source underwater sensor into a directed state snapshot branch containing hash pointer references, and merging the directed state snapshot branches based on hash pointers to construct an asynchronous state evolution network representing the evolution of integrated diving actions when different single-source diving time-series state chains logically intersect, includes the following steps: In each single-source diving time-series state chain, key nodes representing physical parameter jumps are selected and encapsulated into independent data snapshot blocks. The first hash value of the previous data snapshot block is calculated using a cryptographic hash function; The first hash value is embedded as a hash pointer into the current data snapshot block to form a directed state snapshot branch; Monitor the diving action tags between directed state snapshot branches corresponding to different multi-source underwater sensors. When multiple directed state snapshot branches with the same diving action tag are ready within the time window, generate a global merge node containing the hash values ​​of the ends of all ready branches. By using a global merging node to logically bind multiple directed state snapshot branches, an asynchronous state evolution network representing the evolution of integrated diving actions is constructed.

[0009] Optionally, when the asynchronous state evolution network is ready to submit evidence, the following steps are included: extracting state transition feature parameters between adjacent nodes in the asynchronous state evolution network, performing cross-collision matching between the state transition feature parameters and a preset underwater physical constraint benchmark, blocking abnormal nodes that fail to match, and independently assigning them to an isolated evidence storage branch with a physical failure marker: The data payload difference and time extrapolation step size of adjacent nodes in an asynchronous state evolution network are analyzed, and the state transition characteristic parameters are calculated based on the data payload difference and time extrapolation step size. Connect to the smart contract environment built into the data center blockchain system, and retrieve the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries. The state transition feature parameters are input into a matching engine containing static constraint boundaries for cross-collision matching; Identify anomalous nodes that fail to match beyond the static constraint boundary, sever the association between the anomalous nodes and the asynchronous state evolution network backbone, and transfer the anomalous nodes to an isolated evidence-keeping branch with attached physical failure markers.

[0010] Optionally, the connection to the smart contract environment built into the data center blockchain system, retrieving the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries, includes the following steps: Connect to the smart contract storage module built into the data center blockchain system and read the absolute environmental pressure threshold corresponding to the current diving depth from the smart contract storage module; Derivation of fluid statics formula parameters based on absolute environmental pressure threshold; The composition of the gas mixture in the diver's breath was read and the theoretical gas partial pressure limit was calculated. The theoretical gas partial pressure limit was then converted into thermodynamic formula parameters. The parameters of the hydrostatic formula and the thermodynamic formula obtained from the derivation are set together as the static constraint boundary for cross-collision matching.

[0011] Optionally, the step of collecting real-time operational data from multi-source underwater sensors in a diving operation scenario, setting a preset length of overlapping sliding window to divide continuous real-time operational data into overlapping state units with nested contextual states, and encapsulating and transmitting multiple overlapping state units to the data center includes the following steps: The multi-source underwater sensors carried by the diver are activated to collect analog signals and convert them into real-time operational data. Load a preset length of overlapping sliding window into the data buffer, and control the overlapping sliding window to translate and capture real-time job data in steps smaller than its own length. In each translation and truncation, contextual data features are extracted, and overlapping state units that retain the continuity between the preceding and following states are generated based on the contextual data features. Multiple overlapping state units are compressed, and all the compressed overlapping state units are encapsulated and transmitted to the data center using an underwater acoustic modem.

[0012] In a second aspect, the present invention also provides a blockchain-based evidence storage and traceability system for diving operation data, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the blockchain-based evidence storage and traceability method for diving operation data as described in any one of the first aspects.

[0013] Thirdly, the present invention also provides a computer-readable storage medium storing instructions, characterized in that, when executed by a processor, the instructions cause the processor to be configured to perform the blockchain-based evidence storage and traceability method for diving operation data according to any one of the first aspects.

[0014] The beneficial effects of this invention are: This invention constructs a dynamic directed topological graph representing the succession of underlying states, and performs loop removal and interpolation processing on it. This ensures the physical temporal consistency and logical integrity of each single-source temporal state chain before being uploaded to the blockchain, resolving the problem of out-of-order and missing original data caused by channel interference. Furthermore, by constructing an asynchronous state evolution network containing hash pointer references, it represents the complex asynchronous logical convergence and evolutionary relationships between different sensor data streams. Compared to traditional linear chain structures, the asynchronous graph data structure can more faithfully reproduce the real process of complex underwater operations. In the final stage before uploading to the blockchain, a state transition feature cross-collision matching mechanism based on underwater physical constraints is introduced. This adds a verification checkpoint to data storage, actively identifying and isolating abnormal nodes that, although the data format is correct, violate basic physical laws, ensuring the physical authenticity of the uploaded data. Ultimately, this solves the technical problem that channel interference and data anomalies cause the original data to lack physical authenticity and logical consistency, thus rendering blockchain evidence without factual basis. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a blockchain-based method for storing and tracing diving operation data in one embodiment of this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0017] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0018] Figure 1 This is a flowchart illustrating a blockchain-based method for storing and tracing diving operation data in one embodiment. It should be understood that, although... Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps. For example Figure 1 As shown, the blockchain-based evidence storage and traceability method for diving operation data disclosed in this invention specifically includes the following steps: S101. Collect real-time operational data from multiple underwater sensors in diving operation scenarios, set a preset length of overlapping sliding window to divide continuous real-time operational data into overlapping state units with nested context states, and encapsulate and transmit multiple overlapping state units to the data center.

[0019] The underwater multi-source sensor array deployed in the diving operation environment continuously monitors the physiological parameters, environmental parameters, and equipment status parameters of the operators. These sensors include, but are not limited to, pressure sensors, temperature sensors, dissolved oxygen sensors, depth gauges, inertial measurement units, and physiological monitoring equipment. Each sensor acquires analog signals at a fixed sampling frequency, which are then converted into a digital real-time operational data stream by an analog-to-digital converter. To ensure the temporal continuity and status correlation of the data, an overlapping sliding window mechanism is introduced in the data preprocessing stage to segment the continuous data stream. Specifically, the window length is set to... There are sampling points, and the sliding step size is . There are 1 sampling points, among which This allows for the existence of adjacent windows. The overlapping region of each sampling point. This overlapping design ensures that each state unit not only contains independent state information at the current time but also retains contextual features related to previous and subsequent time points, forming overlapping state units with state nesting properties. Each overlapping state unit It can be represented as a set of data vectors within a time window. ,in Indicates time Multidimensional sensor data. After multiple overlapping state units are processed by a data compression algorithm, they are transmitted in the form of data packets to a surface or shore-based data center via an underwater acoustic modem. Forward error correction coding is used during transmission to combat the high bit error rate of the underwater channel.

[0020] S102. Extract the preceding and following state features reflecting the evolution of physical states within each overlapping state unit through the data center, and construct a dynamic directed topological graph representing the succession relationship of the underlying states in the memory space using the preceding and following state features as nodes.

[0021] The data center receives multiple overlapping state units, first decompresses and verifies the integrity of each unit, and then extracts the feature vectors reflecting the physical state evolution within the unit. For each overlapping state unit... It is divided into a preorder part and a postorder part, and the preorder state features This was obtained by extracting statistical features from the data in the first half of the window, including the changing trends of physical quantities such as mean, variance, and gradient; subsequent state features... This corresponds to the feature representation of the data in the latter half of the window. Due to the overlapping window design, adjacent units... and The overlapping areas make Post-sequence features and The preceding features are physically related. Based on this relationship, a dynamic directed topological graph is constructed in memory space. , where the node set Includes all extracted pre-order and post-order state features, edge set This represents the succession relationship between states. The specific construction rule is: if the state features... and The similarity in the feature space exceeds a threshold. Then a directed edge is established between the two nodes. The weights of the edges reflect the confidence level of the state transition.

[0022] S103. In the dynamic directed topology graph, the redundant loops caused by channel interference are eliminated by the path traversal algorithm and the breakpoint interpolation is performed to extract the acyclic traversal path that can connect all valid nodes. Based on the acyclic traversal path, a single-source diving time-series state chain with physical time consistency is reconstructed.

[0023] During the construction of the dynamic directed topology graph, interference factors such as multipath effects of the underwater channel, latency jitter, and out-of-order data packets may lead to topological redundancy loops and breakpoints. To address the redundancy loop problem, a depth-first search combined with a strongly connected component detection algorithm is used to identify multipath redundant patches in the graph. For the detected redundant patches, their topological equivalence index is calculated. ,in A set of nodes for redundant paths to share. This is the union of all involved nodes. When When the threshold is exceeded, the path with the earliest timestamp is retained and redundant branches are deleted, generating a cleaned connected subgraph. To address the graph fragmentation problem, edge free nodes with zero out-degree and target free nodes with zero in-degree are identified, and the state gradient difference between them is calculated. ,in The edge node state vector, The target node state vector. This represents the theoretical interval number of steps. This represents the Euclidean distance. When... When the value is less than the physical evolution extreme value, a virtual bridging state is generated and a virtual directed edge is constructed to complete the breakpoint interpolation.

[0024] S104. The single-source diving time-series state chain corresponding to each multi-source underwater sensor is transformed into a directed state snapshot branch containing hash pointer references. When different single-source diving time-series state chains logically intersect, the directed state snapshot branches are merged based on hash pointers to construct an asynchronous state evolution network that represents the evolution of integrated diving actions.

[0025] Specifically, for each underwater sensor's corresponding single-source diving time-series state chain, key nodes representing significant jumps in physical parameters are first identified. The criterion for this is that the rate of change of state between adjacent nodes exceeds a dynamic threshold. These key nodes, along with several adjacent nodes before and after them, are then encapsulated into independent data snapshot blocks. Each snapshot block contains the physical state value matrix of a node and its corresponding local adjacency matrix. For the previous snapshot block... When performing hash calculations, first extract its local adjacency matrix. With physical state numerical matrix Perform matrix multiplication Generate a correlation tensor that integrates topological logic and physical features. Expand the tensor into a one-dimensional sequence. Input cryptographic hash function The first hash value is calculated. .Will Embedded as a hash pointer in the current snapshot block The metadata area forms directed state snapshot branches. When branches from different sensors are detected to have the same diving action label and are within the time window... When all nodes are ready, a global merge node is generated. This node contains the set of hash values ​​from the ends of all ready branches. By merging nodes and logically binding multiple directed state snapshot branches, an asynchronous state evolution network representing the evolution of integrated diving actions is constructed. This network records the collaborative evolution relationship of multi-source data in a directed acyclic graph structure.

[0026] S105. When the asynchronous state evolution network is ready to submit evidence, extract the state transition feature parameters between adjacent nodes in the asynchronous state evolution network, perform cross-collision matching between the state transition feature parameters and the preset underwater physical constraint benchmark, block abnormal nodes that fail to match and independently assign them to isolated evidence storage branches with physical failure markers.

[0027] In this process, the asynchronous state evolution network must undergo physical constraint consistency verification before submitting its data to the blockchain. This involves traversing all adjacent node pairs in the network and extracting their data payload differences. Step length of time progression Calculate the state transition characteristic parameters This parameter reflects the rate of change of a physical quantity per unit time. Connecting to the smart contract environment built into the data center blockchain system, it reads the absolute environmental pressure threshold corresponding to the current diving depth from the contract storage module. Deducing pressure gradient constraints based on the principles of hydrostatics ,in The density of seawater, It is the acceleration due to gravity. For depth. Simultaneously, the composition of the diver's breath mixture is read, and the theoretical gas partial pressure limit is calculated based on Dalton's law of partial pressures. ,in For the first mole fraction of the gas The total pressure is used. The hydrostatic and thermodynamic parameters are jointly set as static constraint boundaries to construct a matching engine. Perform cross-collision matching. Identify abnormal nodes that exceed the constraint boundaries, sever their association edges with the backbone network, and transfer the abnormal nodes and their associated paths to an isolated evidence branch with a physical failure marker to ensure the physical rationality of the backbone network.

[0028] S106. The complete asynchronous state evolution network, including the isolated evidence storage branch, is hashed, packaged, and broadcast to the blockchain network for consensus on-chain. During the tracing phase, the topology of the asynchronous state evolution network is reverse-analyzed based on the root hash on the blockchain to trace back the physical evolution trajectory and abnormal state source of the diving operation data.

[0029] The asynchronous state evolution network, after physical constraint verification, consists of a backbone network and isolated evidence storage branches. The complete network structure needs to be hashed and packaged to achieve blockchain evidence storage. First, the network is serialized and encoded, converting the node set, edge set, node attributes, and branch marker information into a standardized byte stream format. A Merkle tree structure is used to perform hierarchical hashing of the network data. Leaf nodes correspond to the hash values ​​of various data snapshot blocks, intermediate nodes are generated by recursively calculating the concatenated hashes of their child nodes, and finally, the root node yields the root hash. The root hash, serving as the unique digital fingerprint of the entire asynchronous state evolution network, is encapsulated along with metadata such as timestamps and diving mission identifiers into the blockchain transaction payload. The transaction is broadcast to all verification nodes of the blockchain network via a P2P network. Nodes execute a consensus algorithm to verify the transaction's validity, including hash integrity checks, digital signature verification, and smart contract rule checks. Once consensus is reached, the transaction is packaged into a new block and appended to the main blockchain. During the tracing phase, the root hash recorded at the time of notarization is used... By retrieving corresponding transactions from the blockchain and parsing the serialized data in the transaction payload, the complete topology of the asynchronous state evolution network can be reconstructed. By traversing the hash pointer chain in reverse, the physical evolution trajectory of each state node can be traced back, the source node of the abnormal state and its marking information in the isolated branch can be located, and the traceability and immutability of the diving operation data throughout its entire lifecycle can be guaranteed.

[0030] In one implementation, a path traversal algorithm is used to eliminate redundant loops caused by channel interference in the dynamic directed topology graph and breakpoint interpolation is performed to extract acyclic traversal paths that can connect all valid nodes. The reconstruction of a single-source diving time-series state chain with physical time consistency based on the acyclic traversal paths includes the following steps: Identify the in-degree and out-degree of nodes in a dynamic directed topology graph, and use a path traversal algorithm to detect multipath redundant blocks in the dynamic directed topology graph based on the in-degree and out-degree of nodes. Evaluate the topological equivalence of multipath redundant tiles and eliminate multipath redundant tiles based on topological equivalence to generate a cleaned connected subgraph; Identify graph breakpoints caused by signal loss in connected subgraphs and perform breakpoint interpolation; Connect the connected subgraphs after breakpoint interpolation, extract the acyclic traversal path that can connect all valid nodes, and reconstruct the single-source diving time-series state chain with physical time consistency based on the acyclic traversal path.

[0031] In this implementation, the in-degree of a node represents the number of directed edges pointing to that node, reflecting how many preceding states reference the state feature; the out-degree of a node represents the number of directed edges emanating from that node, reflecting how many subsequent states the state feature can transition to. Traverse all nodes in the graph and count the values ​​for each node. in-degree With out When a node has an in-degree greater than 1, it indicates that multiple paths converge to that node, potentially forming a redundant tile. A depth-first search algorithm is used to backtrack from the node with the abnormal in-degree, marking the set of all predecessor paths that can reach that node. Simultaneously, tracing back from that node, the common starting node of these paths is identified. If multiple paths form a parallel structure between the starting and converging nodes, and there are no intersecting connections between the paths, it is determined to be a multipath redundant tile. For the detected redundant tiles, the node sequences of each parallel path are first extracted, and the structural similarity between the paths is calculated. A topological equivalence coefficient is defined. ,in This represents a set of nodes shared by multiple paths. This represents the union of all nodes involved in the path. When A value close to 1 indicates a high degree of path overlap. Further calculations are made to determine the similarity of state features at corresponding nodes along the path, using cosine similarity as a metric. ,in and These are the state feature vectors of the corresponding nodes on the two paths. Combining the similarity between the topology and state features, redundant paths are deemed equivalent when both exceed a preset threshold. After redundancy elimination, a cleaned connected subgraph is generated. This subgraph retains the core logic of state evolution and eliminates topology noise caused by channel interference.

[0032] Next, the cleaned connected subgraph is traversed to identify edge detached nodes with zero degree and not terminating nodes, and target detached nodes with zero in-degree and not starting nodes. These detached nodes are typically generated due to signal attenuation or Doppler shift causing data packets to fail to be received. For each pair of potentially connectable detached nodes, the state gradient difference is calculated to evaluate the physical viability of the connection. The state gradient difference is defined as follows: ,in Let be the physical state vector of the edge free node. Let be the physical state vector of the target free node. This represents the interval between the two nodes in the theoretical received sequence. This represents the Euclidean distance. The formula calculates the rate of change of state within a unit time step, when... When the evolutionary extreme value is less than the value set according to the physical laws of diving operations, a reasonable state transition relationship is considered to exist between the two nodes. At this point, a virtual bridging state characteristic is generated, which is calculated using a linear interpolation method. Virtual directed edges are constructed between edge nodes and target nodes to complete breakpoint interpolation.

[0033] A topological sorting algorithm is used to sort the integrated graph structure. This algorithm determines the access order based on the in-degree information of the nodes. A queue is initialized, and all nodes with an in-degree of zero are added to the queue. During iteration, a node is removed from the queue each time, added to the sorted sequence, and the in-degree of all its successor nodes is decremented by 1. When the in-degree of a successor node becomes zero, it is added to the queue. The topological sorting process ensures that the access order of nodes conforms to the directional constraints of directed edges, and the generated sequence is an acyclic traversal path. This path connects all valid nodes, and the node order is consistent with the direction of physical time evolution. During the sorting process, if a cycle is found in the graph that prevents the sorting from being completed, a cycle detection and breaking mechanism is triggered to identify and break edges with abnormal timestamps in the cycle. When reconstructing the single-source diving time-series state chain based on the acyclic traversal path, the physical state data of each node is extracted according to the path order to construct a time-series data structure.

[0034] In one implementation, identifying graph breakpoints caused by signal loss in a connected subgraph and performing breakpoint interpolation includes the following steps: Extract the edge free nodes in the connected subgraph and calculate the state gradient difference between the edge free nodes and the target free nodes in the adjacent connected subgraphs; When the state gradient difference is less than the preset physical evolution extreme value, a virtual bridging state feature is generated. Virtual directed edges are constructed between edge free nodes and target free nodes by utilizing virtual bridging state characteristics; By connecting the corresponding connected subgraphs with virtual directed edges, the breakpoint interpolation of the graph breakpoints is completed.

[0035] In this implementation, all connected subgraphs after redundancy elimination are traversed, and node degree checks are performed for each subgraph. Edge-free nodes are defined as nodes that meet specific degree conditions: they have zero out-degree in the directed graph but are not the natural termination point of the entire temporal chain, or zero in-degree but are not the natural start point of the temporal chain. The occurrence of such nodes usually stems from shadowing effects in underwater acoustic channels, multipath fading, or packet loss caused by transient noise interference. During the extraction process, a node attribute table is maintained, recording the timestamp, sensor identifier, and theoretical sequence number of each node in the original data stream. The break points in the sequence are identified by comparing the actual sequence number with the theoretical sequence number. For two adjacent connected subgraphs, the tail edge-free nodes of the preceding subgraph and the head target edge-free nodes of the following subgraph are extracted, respectively. When calculating the state gradient difference between two nodes, the physical state vector contained in the edge-free node is first read. This vector contains multi-dimensional physical parameters such as pressure, temperature, depth, and acceleration. Similarly, the physical state vector of the target free node is read. Calculate the Euclidean distance between two vectors in a high-dimensional state space. And obtain the number of interval steps between the two nodes in the theoretical received sequence. The state gradient difference is defined as follows: This value characterizes the average rate of change of the physical state within a unit time step, reflecting the severity of the state transition.

[0036] The physical evolution extrema are pre-set based on theories of human physiology, fluid mechanics, and diving medicine, including constraint parameters across multiple dimensions such as upper limits for pressure change rate, temperature gradient, and depth change rate. Taking pressure change as an example, according to Henry's Law and Boyle's Law, the rate of pressure change during ascent or descent should not exceed a specific threshold to avoid the risk of decompression sickness. The physical evolution extrema for the pressure dimension are set as follows: The extreme values ​​of the temperature dimension are The extreme value of the depth dimension is The calculated state gradient difference Component decomposition is performed, and the component gradients of each physical dimension are extracted and compared with the extreme values ​​of the corresponding dimensions. When the component gradients of all dimensions are less than their respective physical evolution extreme values, a reasonable physical state transition relationship is determined to exist between the edge free node and the target free node, satisfying the preconditions for breakpoint interpolation. At this point, the virtual bridging state feature generation process is initiated. The virtual bridging state features are constructed using linear interpolation or spline interpolation methods, the specific choice depending on the number of interval steps. When the value is small, linear interpolation is used, and the calculation formula is: ,in This is the generated virtual bridging state feature vector.

[0037] Based on the generated virtual bridging state features Create new virtual nodes in a graph data structure. The node's attribute fields include a virtual flag, an interpolated timestamp, and a state feature vector. The virtual timestamp is calculated by linearly interpolating the timestamps of the edge nodes and the target node, ensuring the monotonicity of the time series. In edge-free nodes... With virtual nodes Establish the first virtual directed edge between them The edge weights are set as the state transition confidence level, which is inversely proportional to the state gradient difference. Meanwhile, in the virtual nodes... With target free node Establish a second virtual directed edge between them. The attribute fields of both virtual directed edges are marked as virtual edge type to facilitate the differentiation between original data edges and interpolated reconstructed edges during subsequent tracing. (When the interval number...) When the value is greater than 2, multiple virtual bridging nodes need to be generated to form an interpolation chain. The state characteristics of each virtual node are calculated according to the principle of equal-interval interpolation. The construction of virtual directed edges follows the topological rules of directed graphs, ensuring that the direction of the edges is consistent with the direction of time evolution and that no new loop structures are introduced. The adjacency list and adjacency matrix representation of the graph are updated, incorporating virtual nodes and virtual edges into the topological structure of the graph, so that the originally broken connected subgraphs are logically connected through virtual bridging.

[0038] The connection operation of connected subgraphs using virtual directed edges achieves breakpoint interpolation repair of graph breaks. During the connection process, the topological validity of the repaired graph structure needs to be verified, and loop detection is performed to ensure no circular dependencies are introduced. A depth-first search algorithm is used to traverse from the repaired starting node, marking all reachable nodes and verifying whether all originally isolated target nodes can be reached. If a visited node is found during the traversal, it indicates the existence of a loop, requiring backtracking and adjustment of the virtual edge connection strategy. After breakpoint interpolation is completed, the previously separate connected subgraphs are merged into a single connected graph structure, which maintains the characteristics of a directed acyclic graph. The global attribute statistics of the graph are updated, and topological parameters such as the total number of nodes, the total number of edges, and the graph diameter are recalculated. Detailed information on interpolation repair is recorded in the graph's metadata layer, including the number of virtual nodes, the location of virtual edges, and the physical constraint parameters on which the interpolation is based. This metadata will be encapsulated into data snapshot blocks in the subsequent blockchain notarization stage to achieve traceability of the interpolation process.

[0039] In one implementation, calculating the state gradient difference between an edge free node and a target free node in an adjacent connected subgraph includes the following steps: Read the first physical state vector contained in the edge free node, and read the second physical state vector contained in the target free node in the adjacent connected subgraph; Obtain the Euclidean distance between the first physical state vector and the second physical state vector in the data space; Divide the Euclidean distance by the theoretical interval steps between the edge free node and the target free node in the received sequence to obtain the unit evolution rate of change, and use the unit evolution rate of change as the state gradient difference between the edge free node and the target free node.

[0040] In this implementation, each node is assigned a state vector containing multi-dimensional physical parameters upon creation. This vector encapsulates the complete physical state information of the node at that given moment. Edge-free nodes, as the tail nodes of the preceding connected subgraph, store the first physical state vector internally. It includes physical quantities such as pressure, temperature, depth, triaxial acceleration, triaxial angular velocity, and dissolved oxygen concentration. Specifically, the reading process first locates the node's memory address using its identifier, accesses the node's attribute field area, and parses the binary encoding format of the state vector. The state vector is stored using the IEEE 754 floating-point standard, with each physical component occupying 4 or 8 bytes of space. The read operation converts the binary data into a floating-point array, forming a vector structure suitable for numerical calculations. For the target free node, the same read process is performed to obtain the second physical state vector. First physical state vector With the second physical state vector In a data space, the Euclidean distance between two points is defined as the square root of the sum of the squares of the differences between their dimensional components. Because different physical quantities have significantly different dimensions and numerical ranges, direct calculation may result in certain dimensions dominating the distance value. Therefore, the vectors need to be normalized before calculation to map the values ​​of each dimension to the same scale range.

[0041] Using the min-max normalization method, for the th The dimensional component, the normalization formula is as follows: ,in and These represent the minimum and maximum values ​​of this dimension in historical data. The normalized vector components all range from 0 to 1. The Euclidean distance calculation formula is as follows: ,in The vector dimension is used. The calculation process employs vectorized operations to improve efficiency. First, the difference vectors for each dimension are calculated. Then, the square of each element of the difference vector is calculated, and the square root of all squared values ​​is taken to obtain the final distance. The calculation of the unit evolution rate transforms spatial distance into the rate of change in the time dimension, establishing a quantitative relationship between state differences and temporal evolution. Theoretical interval steps. This represents the number of data packets that should separate an edge detached node from a target detached node under ideal, lossless conditions. This number is calculated using the node's timestamp and the data sampling period. ,in For the target node timestamp, For edge node timestamps, The sensor sampling period, This indicates rounding down. Euclidean distance. Divide by the theoretical interval number of steps The unit evolution rate was obtained. This rate of change characterizes the magnitude of change in the physical state per average time step, expressed as normalized distance per step. Physically, a smaller unit rate of change indicates a smooth state transition, consistent with the gradual nature of diving operations; an excessively large rate of change suggests potential anomalous jumps or measurement errors.

[0042] In one implementation, the single-source diving time-series state chain corresponding to each multi-source underwater sensor is transformed into a directed state snapshot branch containing hash pointer references. When different single-source diving time-series state chains logically intersect, the directed state snapshot branches are merged based on hash pointers to construct an asynchronous state evolution network representing the evolution of integrated diving actions, including the following steps: In each single-source diving time-series state chain, key nodes representing physical parameter jumps are selected and encapsulated into independent data snapshot blocks. The first hash value of the previous data snapshot block is calculated using a cryptographic hash function; The first hash value is embedded as a hash pointer into the current data snapshot block to form a directed state snapshot branch; Monitor the diving action tags between directed state snapshot branches corresponding to different multi-source underwater sensors. When multiple directed state snapshot branches with the same diving action tag are ready within the time window, generate a global merge node containing the hash values ​​of the ends of all ready branches. By using a global merging node to logically bind multiple directed state snapshot branches, an asynchronous state evolution network representing the evolution of integrated diving actions is constructed.

[0043] In this implementation, a physical parameter jump is defined as the phenomenon where the change in a certain physical quantity between adjacent nodes exceeds a dynamic threshold, reflecting critical actions or sudden environmental events during diving operations. Each node in the state chain is traversed, and the rate of change between the current node and its predecessor node in each physical dimension is calculated. For the pressure dimension, the relative rate of change is calculated. ,in This represents the current node's pressure value. This represents the pressure value at the precursor node. When... When a preset pressure jump threshold is exceeded, the node is marked as a critical pressure jump node. Similarly, the same jump detection is performed on dimensions such as temperature, depth, and acceleration. A multi-dimensional joint judgment strategy is adopted: when a node experiences a jump in any physical dimension, or when multiple dimensions experience moderate changes simultaneously and the overall score exceeds the threshold, the node is marked as a critical node. Critical nodes typically correspond to important action nodes of the diver, such as descent, ascent, stay, and operational procedures. The selected critical nodes, along with several adjacent nodes before and after them, are encapsulated into independent data snapshot blocks.

[0044] A collision-resistant hash algorithm such as SHA-256 or SHA-3 is selected as the cryptographic hash function. Before computation, the preceding data snapshot blocks need to be standardized and serialized, converting the structured data into a byte stream. The serialization order strictly follows predefined field arrangement rules to ensure that snapshot blocks with the same content generate the same byte stream. The byte stream contains all the key information of the snapshot block: the physical state value matrix of the node set, the local adjacency matrix, the timestamp sequence, and the action tag encoding. Specifically, the local adjacency matrix... With physical state numerical matrix First, perform matrix multiplication. This process generates a correlation tensor that integrates topological structure and physical features. This tensor is expanded into a one-dimensional vector using either row-major or column-major order, and then concatenated with other field data to form a complete input sequence. The input sequence is then fed into a hash function. After multiple rounds of iterative compression operations, a first hash value of 256 bits is output. The first hash value not only serves as a unique identifier for the previous snapshot block, but also implicitly binds to the topological origin information of that snapshot block, because the adjacency matrix participates in the hash calculation, and any change in the topology will be reflected in the hash value.

[0045] Current data snapshot block A dedicated hash pointer field is allocated during creation, which stores the first hash value of the previous snapshot block. Write this field. Unlike traditional memory address pointers, hash pointers do not directly point to the physical storage location of data; instead, they establish logical associations through hash values. This ensures that any modification to the content of a previous snapshot block will cause a change in the hash value, which will then be detected by the current snapshot block, achieving data tamper-proofing. Multiple snapshot blocks are sequentially connected through hash pointers, forming directed state snapshot branches. This branch has the topological characteristics of a singly linked list; each node can only trace forward and cannot be referenced backward, guaranteeing the irreversibility of the time sequence. Each directed state snapshot branch is assigned an action tag sequence during its generation. The tags are extracted from physical state change characteristics using a temporal pattern recognition algorithm. Action tags include discrete states such as start of descent, end of descent, hovering, horizontal movement, start of ascent, and end of ascent. Although different sensors measure different physical quantities, they will exhibit related state change patterns when the same diving action occurs. For example, when a diver begins to descend, the pressure sensor shows an increase in pressure, the depth gauge shows an increase in depth, and the inertial measurement unit shows downward acceleration. The monitoring program continuously scans the directed state snapshot branches corresponding to all sensors, extracting the action tag for each branch at the current moment. Set time window The window length is set based on the typical duration of the action, usually ranging from several seconds to tens of seconds. Within the time window, the number of branches with the same action label is counted. When multiple branches are found to be simultaneously labeled with the same action and the timestamp differences are within the allowable range, these branches are deemed ready to be merged. A global merge node is generated. The data structure of this node includes a list of identifiers for ready branches and a set of hash values ​​for snapshot blocks at the end of each branch. Merge timestamps and action tags.

[0046] Each global merge node acts as a convergence node in the network, receiving input edges from multiple directed state snapshot branches. During the connection process, a directed edge is established between the final snapshot block of each ready branch and the global merge node. The edge direction points from the snapshot block to the merge node, representing the convergence direction of the state flow. Edge attributes include sensor type, data confidence, and time synchronization deviation. The global merge node can serve as the starting point for new branches, extending further to form merged state evolution paths. When new action label matches, the network can branch again after the merge node, forming a tree-like or mesh-like topology. The asynchronous nature of the asynchronous state evolution network is reflected in the asynchronous arrival times of data from different sensors; each branch grows at its own rate, merging synchronously only when action label matches. The network topology evolves dynamically, continuously expanding the scale of nodes and edges as new data arrives. In the global view of the network, the independent evolution paths of each sensor and the collaborative relationships of multi-source data can be clearly identified. The network structure is stored in the form of a graph database or adjacency list, supporting efficient topology queries and path tracing.

[0047] In one implementation, calculating the first hash value of the previous data snapshot block using a cryptographic hash function includes the following steps: The set of constituent sub-state nodes corresponding to the previous data snapshot block in the single-source diving time-series state chain; Extract the local adjacency matrix of directed edges that form the internal connections between sets of sub-state nodes in a dynamic directed topological graph. Extract the physical state numerical matrix from the previous data snapshot block, perform matrix multiplication between the local adjacency matrix and the physical state numerical matrix, and generate a fusion correlation tensor that binds the topology source logic and physical state features. The dimensionality reduction expansion of the fusion correlation tensor generates a one-dimensional state topological logic sequence. The one-dimensional state topological logic sequence is input into the cryptographic hash function for one-way hash calculation, and the output is the first hash value containing the self-certifying property of the underlying reconstructed topology.

[0048] In this embodiment, each data snapshot block records key node identifiers and the identifier sequence of its preceding and following adjacent nodes during encapsulation. These identifiers constitute the node index table of the snapshot block. The backtracking process first reads the metadata area of ​​the previous data snapshot block and extracts all node identifiers recorded in the node index table. The node identifiers adopt a globally unique identifier format and contain a combination of information such as sensor number, timestamp, and sequence number. Based on the node identifiers, the global node registry of the single-source diving time-series state chain is searched and located to obtain the position index of each node in the state chain. Since the snapshot block may contain non-contiguous nodes, a node set needs to be constructed. ,in This represents the total number of nodes contained in the snapshot block. For each node, it is also necessary to trace back the complete attribute information of that node in the original dynamic directed topology graph, including the node's physical state vector, in-degree and out-degree information, and connection relationships.

[0049] For including A subgraph with nodes, and its adjacency matrix. for A square matrix. Matrix elements. Represents a node To the node Does a directed edge exist? If so, then ,otherwise In the case of a weighted graph, matrix elements can store edge weights, which reflect the confidence level of a state transition or the strength of a physical association. The extraction process traverses all node pairs in the set of child state nodes, querying the global adjacency list or adjacency matrix of the dynamic directed topology graph to check if a directed edge exists between the two nodes. Since the dynamic graph may contain thousands or even tens of thousands of nodes, directly accessing the global adjacency matrix is ​​inefficient; therefore, a hash index or B-tree index is used to accelerate the edge query operation. For each node pair... The existence of edges is quickly determined through an index structure, and the attribute information of the edges is extracted. The local adjacency matrix only contains the connection relationships within the child state node set, and does not include the connections between the child set and external nodes, thus realizing the localized extraction of topological information.

[0050] Physical state numerical matrix The dimension is ,in This represents the number of child state nodes. Let be the dimension number of the physical state vector. The _th _th_ ... Row corresponding node The physical state vector contains multi-dimensional physical parameters such as pressure, temperature, depth, and acceleration. The extraction process iterates through the set of sub-state nodes, reads the physical state vector stored in each node, and fills it row-by-row into the physical state value matrix. Matrix multiplication operations are then performed. Multiplying the local adjacency matrix by the physical state matrix generates a fused correlation tensor. The physical meaning of this operation is that each row of the adjacency matrix represents the outgoing edge connection pattern of a node; after multiplication with the physical state matrix, each row of the resulting tensor aggregates the physical state information of all successor nodes of that node. Specifically, the fused correlation tensor... The Line number Column elements This value reflects the node The impact of topological connections on the first The cumulative effect of physical quantities. The generated fusion correlation tensor still has a dimension of 1. However, the values ​​in the tensor are no longer simply physical state values, but rather incorporate the correlation characteristics of topological propagation effects.

[0051] The dimensionality reduction expansion operation of fused correlation tensors transforms two-dimensional matrix structures into one-dimensional sequences, facilitating processing by cryptographic hash functions. The dimensionality reduction expansion employs a row-first or column-first traversal strategy to expand the tensor... All elements are arranged in a fixed order to form a one-dimensional vector. The row-major expansion proceeds from the first row, the second row, and so on up to the [missing row]. Read elements sequentially in row order to generate a length of One-dimensional state topological logic sequence This sequence preserves the relative positions of the elements in the tensor; elements at different positions correspond to different physical dimensions of different nodes. The first part of the sequence... Each element corresponds to the fusion feature of the first node, and then... Each element corresponds to the second node, and so on. The generation of a one-dimensional sequence requires handling the precision of floating-point numbers. A fixed-precision numerical representation method is used, typically retaining 6 significant digits to avoid the impact of floating-point errors on hash calculations. The values ​​in the sequence are further converted into a byte stream format, with each floating-point number encoded into a 4-byte or 8-byte binary representation according to the IEEE 754 standard.

[0052] The SHA-256 algorithm was chosen as the cryptographic hash function. This algorithm possesses strong collision resistance, avalanche effect, and one-way cryptographic security properties. It can be used to process one-dimensional sequences. The byte stream is used as input to the hash function. After steps such as message padding, block processing, and multiple rounds of iterative compression, the final output is a 256-bit first hash value. During the hash calculation process, the input sequence is divided into 512-bit message blocks. Each message block undergoes 64 rounds of nonlinear transformations and modular addition operations, gradually obfuscating the bit distribution of the input data. Since the input sequence incorporates information from the local adjacency matrix and the physical state numerical matrix, the generated hash value implicitly binds to the dual attributes of topological structure and physical characteristics. This hash value possesses a self-verifying characteristic of underlying topological reconstruction, meaning that any tampering with the original topological structure or physical state will lead to unpredictable changes in the hash value. In the source verification phase, the integrity and authenticity of the data can be verified by recalculating the hash value and comparing it with the stored hash value. The first hash value is represented in hexadecimal string form, with a length of 64 characters, facilitating storage and transmission.

[0053] In one implementation, when the asynchronous state evolution network is ready to submit evidence, the following steps are taken: extracting state transition feature parameters between adjacent nodes in the asynchronous state evolution network, performing cross-collision matching between the state transition feature parameters and a preset underwater physical constraint benchmark, blocking abnormal nodes that fail to match, and independently assigning them to an isolated evidence storage branch with a physical failure marker: The data payload difference and time extrapolation step size of adjacent nodes in an asynchronous state evolution network are analyzed, and the state transition characteristic parameters are calculated based on the data payload difference and time extrapolation step size. Connect to the smart contract environment built into the data center blockchain system, and retrieve the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries. The state transition feature parameters are input into a matching engine containing static constraint boundaries for cross-collision matching; Identify anomalous nodes that fail to match beyond the static constraint boundary, sever the association between the anomalous nodes and the asynchronous state evolution network backbone, and transfer the anomalous nodes to an isolated evidence-keeping branch with attached physical failure markers.

[0054] In this implementation, all pairs of nodes with direct connections in the network are traversed, and for adjacent nodes... and First, the data payloads of the two nodes are extracted. The data payload contains the complete physical state vector of the node, covering multi-dimensional physical parameters such as pressure, temperature, depth, dissolved oxygen concentration, triaxial acceleration, and triaxial angular velocity. The difference in data payloads is then calculated. ,in and They are nodes and The data payload vector. The payload difference is a multi-dimensional vector, with each dimension corresponding to the change in a physical parameter. Time extrapolation step size. The difference between the timestamps of the two nodes represents the actual time interval during the state transition. Based on the load difference and the time step, the characteristic parameters of the state transition are calculated. This parameter is a multidimensional vector, where each component represents the rate of change of the corresponding physical quantity. For example, the pressure component. This represents the rate of change of pressure per unit time, with temperature components. This represents the rate of temperature change.

[0055] The smart contract environment built into the data center blockchain system uses the Ethereum Virtual Machine or a similar contract execution engine, supporting the execution of Turing-complete contract code. Smart contracts are written in languages ​​such as Solidity or Vyper, pre-deployed in the blockchain network, and the contract code is compiled and stored as bytecode on the blockchain ledger. Connecting to the smart contract environment requires establishing communication with blockchain nodes via JSON-RPC or Web3 interfaces to send contract call requests. Underwater physical constraint benchmarks are stored in the smart contract's state variables, including two main categories: hydrostatic formula parameters and thermodynamic formula parameters. The hydrostatic parameters are set based on Pascal's law and the principle of hydrostatic pressure, with the core formula being... ,in For depth Absolute pressure at the location, Atmospheric pressure at sea level The density of seawater, The acceleration due to gravity is used. The contract stores the upper and lower limits of the pressure threshold corresponding to different depths, as well as the limit values ​​of the rate of pressure change. Thermodynamic parameters are set based on Dalton's law of partial pressures and Henry's law, involving the calculation of the partial pressures of breathing gases. For a nitrogen-oxygen mixture, the partial pressure of nitrogen is... oxygen partial pressure ,in and mole fraction This represents the total pressure. The contract stores the safe upper and lower limits for the partial pressures of each gas; exceeding these limits may result in nitrogen anesthesia or oxygen poisoning. The retrieval operation is performed through the contract's publicly available read function, passing in parameters such as the current diving depth and gas composition. The contract returns the corresponding static constraint boundary values.

[0056] The matching engine adopts a rule-based engine architecture, internally maintaining a constraint rule base and matching logic. It transfers state transition feature parameters. Each component of the pressure dimension is compared one by one with its corresponding static constraint boundary. For the pressure dimension, the rate of change of pressure is checked. Is it within the permitted range? Within this range, determined according to diving decompression theory, the rate of pressure change during ascent should generally not exceed the pressure change corresponding to a depth of 18 meters per minute. For the temperature dimension, the rate of temperature change is verified. Whether it conforms to the natural laws of water temperature gradients is crucial; the temperature gradient of a temperature stratum in the ocean is typically between 0.1 and 1 degree Celsius per meter. For the depth dimension, it checks whether the rate of depth change conforms to the physiological limits of divers; professional divers typically descend at speeds not exceeding 30 meters per minute and ascend at speeds not exceeding 18 meters per minute. Cross-collision matching employs a parallel verification strategy, simultaneously checking the constraints of all physical dimensions. Matching results are categorized into three types: complete match (all dimensions satisfy constraints), partial match (some dimensions exceed constraints but are within tolerance), and failed match (critical dimensions severely violate physical constraints).

[0057] The matching reports of all node pairs are traversed, and node pairs that fail to match are filtered out. Successor nodes are marked as anomalous nodes. Anomalous nodes typically correspond to sensor failures, data transmission errors, or abnormal physical events. For each anomalous node, the number of incoming edges is first checked. If a node has multiple incoming edges and only some of these edges correspond to failed state transition matches, the source of the anomaly is further analyzed. If all incoming edges of a node fail to match, the node is determined to be an isolated anomalous node. The connection between the anomalous node and the network backbone is severed by deleting or disabling related edges. Specifically, directed edges pointing to the anomalous node are marked as invalid, and outgoing edges of the anomalous node are also marked as invalid to prevent the anomalous state from propagating. The anomalous node and its associated edges are transferred to an isolated evidence-keeping branch, which exists independently in the network topology and is not connected to the backbone network. Each node in the isolated branch is attached with a physical failure flag. The flag field records the failure type, failure cause, detection time, and related physical parameter exceedance information. The physical failure flag uses structured coding, including quantitative indicators such as failure level, failure dimension, and exceedance multiple. The isolated evidence-preserving branch retains complete information about the abnormal data, facilitating subsequent fault diagnosis and incident analysis. After the abnormal node is removed, the network backbone re-performs connectivity checks to ensure that the backbone network still maintains a complete path from the starting node to the ending node. If the removal of the abnormal node causes a backbone break, a path repair mechanism is activated to find an alternative path that bypasses the abnormal node or to reconstruct the missing state transitions through interpolation.

[0058] In one implementation, connecting to the smart contract environment built into the data center blockchain system and retrieving the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries includes the following steps: Connect to the smart contract storage module built into the data center blockchain system and read the absolute environmental pressure threshold corresponding to the current diving depth from the smart contract storage module; Derivation of fluid statics formula parameters based on absolute environmental pressure threshold; The composition of the gas mixture in the diver's breath was read and the theoretical gas partial pressure limit was calculated. The theoretical gas partial pressure limit was then converted into thermodynamic formula parameters. The parameters of the hydrostatic formula and the thermodynamic formula obtained from the derivation are set together as the static constraint boundary for cross-collision matching.

[0059] In this embodiment, the smart contract storage module maintains a depth-pressure mapping table, which is pre-calculated based on the principles of hydrostatics and stored in the contract's state variables. The mapping table employs a piecewise linear interpolation structure, dividing the diving depth range from 0 to 200 meters into several intervals, each corresponding to a set of pressure threshold parameters. Read operations are performed by calling the contract's public query function, passing the current diving depth as the query parameter. The contract function first determines the interval to which the depth belongs, and then performs linear interpolation within that interval to calculate the precise absolute environmental pressure threshold. The absolute environmental pressure threshold includes three values: a standard pressure value, an upper allowable deviation limit, and a lower allowable deviation limit. The standard pressure value is determined according to the formula... Calculation, where The standard atmospheric pressure is 101.325 kPa. The density of seawater is 1025 kg per cubic meter. The acceleration due to gravity is 9.8 meters per second squared. This represents the diving depth. The allowable deviation is set based on measurement error and environmental fluctuations, typically ±5% of the standard value. Based on the read absolute environmental pressure threshold, the pressure gradient parameter is first calculated. This parameter characterizes the pressure change per unit depth. The pressure gradient is obtained by differentiating the fundamental equations of hydrostatics and is numerically equal to... In a standard seawater environment, the pressure is approximately 10.045 kPa per meter. The pressure gradient parameter is used to verify whether the rate of pressure change during vertical movement conforms to physical laws. Further, the time constraint of the pressure change rate is extrapolated, and combined with decompression theory from diving medicine, the maximum permissible rate of pressure change is calculated. The rate of pressure change during ascent should not exceed 1.8 kPa per minute, corresponding to a depth change rate of approximately 18 meters per minute. The constraints during descent are relatively relaxed, with an upper limit of 3 kPa per minute. The extrapolation process also needs to consider the second derivative constraint of pressure, i.e., the limitation of pressure acceleration, to avoid abrupt shocks during pressure changes. The pressure acceleration threshold is set based on the tolerance of the human vestibular system, typically not exceeding 0.5 kPa per second.

[0060] The diving equipment management system or sensor network reads the current breathing gas formula. Common gas mixtures include air, nitrox, and helium-oxygen mixtures. Gas composition is expressed as mole fraction or volume fraction, such as the mole fraction of oxygen in standard air. Nitrogen mole fraction The calculation of the theoretical partial pressure limit of gases is based on Dalton's law of partial pressures, which states that the total pressure of a gas mixture is equal to the sum of the partial pressures of its components. For oxygen, the formula for calculating the partial pressure is: ,in This represents the absolute environmental pressure at the current depth. The safe range for oxygen partial pressure is determined based on the risk of oxygen toxicity; the upper limit of working partial pressure is typically 140 kPa, the upper limit of limiting partial pressure is 160 kPa, and the lower limit is 16 kPa to avoid hypoxia. Nitrogen partial pressure affects the nitrogen anesthetic effect; divers may experience impaired judgment when the partial pressure exceeds 320 kPa, and the anesthetic effect is significant when it exceeds 400 kPa. For helium, due to its inertness and low anesthetic effect, partial pressure constraints mainly consider hyperbaric neurosis, with an upper limit of approximately 1000 kPa. When converting the calculated partial pressure limits for each gas into thermodynamic formula parameters, it is necessary to establish the relationship between partial pressure and thermodynamic quantities such as temperature and density. This is based on the ideal gas law. At a given temperature, the density of a gas is directly proportional to its partial pressure.

[0061] Fluid statics equation parameters primarily constrain the variation of macroscopic physical quantities such as pressure and depth, while thermodynamic equation parameters primarily constrain the safe range of microscopic thermodynamic quantities such as gas partial pressure and temperature. These two types of parameters are physically interrelated; for example, ambient pressure directly determines gas partial pressure, thus requiring the establishment of coupling relationships between parameters. A hyperrectangular constraint domain model is used to construct the constraint boundaries, defining the allowable state region in the multidimensional parameter space. For the pressure dimension, the constraint boundary is... For the oxygen partial pressure dimension, the constraint boundary is: Other dimensions are defined similarly. The boundary of the constraint domain is determined by the upper and lower limits of each dimension. A state point falling within the constraint domain is considered physically reasonable, while falling outside the domain is considered abnormal. In addition to constraints in independent dimensions, cross-dimensional coupling constraints also need to be set, such as the constraint on the relationship between pressure and oxygen partial pressure. ,in This is the tolerance term. Coupled constraints are expressed as a system of inequalities, forming the internal constraint surface of the constraint domain.

[0062] In one embodiment, real-time operational data from multiple underwater sensors in a diving operation scenario is collected. A preset length of overlapping sliding window is used to divide the continuous real-time operational data into overlapping state units with nested contextual states. Multiple overlapping state units are then encapsulated and transmitted to a data center, comprising the following steps: The multi-source underwater sensors carried by the diver are activated to collect analog signals and convert them into real-time operational data. Load a preset length of overlapping sliding window into the data buffer, and control the overlapping sliding window to translate and capture real-time job data in steps smaller than its own length. In each translation and truncation, contextual data features are extracted, and overlapping state units that retain the continuity between the preceding and following states are generated based on the contextual data features. Multiple overlapping state units are compressed, and all the compressed overlapping state units are encapsulated and transmitted to the data center using an underwater acoustic modem.

[0063] In this embodiment, the sensor array carried by the diver includes various types of equipment such as pressure sensors, temperature sensors, depth gauges, inertial measurement units (IMUs), heart rate monitors, dissolved oxygen sensors, and underwater positioning beacons. Sensor activation is achieved by sending a wake-up command from the main control unit, causing each sensor to switch from low-power standby mode to operating mode and begin measuring physical quantities according to a preset sampling frequency. The pressure sensor uses piezoresistive or piezoelectric principles to convert water pressure into changes in resistance or charge, outputting an analog voltage signal. The temperature sensor is based on thermistor or thermocouple principles, outputting an analog voltage proportional to temperature. The IMU integrates a three-axis accelerometer and a three-axis gyroscope, outputting a six-degree-of-freedom motion state analog signal. After the analog signal is acquired, it undergoes signal conditioning circuitry for amplification, filtering, and anti-aliasing processing to eliminate high-frequency noise and low-frequency drift. The analog-to-digital conversion uses a successive approximation or Σ-Δ analog-to-digital converter to convert the continuous analog signal into a discrete digital signal. The conversion accuracy is typically 12 to 16 bits, and the sampling frequency is set according to the sensor type. Pressure and temperature sensors use sampling frequencies of 1 to 10 Hz, while inertial sensors can reach sampling frequencies of over 100 Hz. The converted digital signal is represented by a fixed-length binary number, and each sampling point includes fields such as timestamp, sensor identifier, physical quantity type, and numerical value. Data streams from multiple sensors are synchronized and aligned in time within the main control unit, using network time protocols or hardware clock synchronization mechanisms to ensure consistency of time bases across different data sources.

[0064] A preset length of overlapping sliding window is defined as containing A time window for consecutive sampling points, with the window length set according to the typical timescale of the diving operation, typically ranging from 5 to 30 seconds for the number of sampling points. The window data structure is loaded into a dedicated storage area of ​​the data buffer, which employs a double-buffered or circular-buffered architecture to support rapid window movement and data updates. Sliding step size. The step size is set to a positive integer less than the window length, typically 1 / 2 to 2 / 3 of the window length. For example, when the window length is 100 sampling points, the step size is set to 50 or 60 sampling points. The ratio of the step size to the window length determines the overlap between adjacent windows. The overlap calculation formula is... Higher overlap results in more complete preservation of context information but also greater computational overhead. The window panning and truncation process begins at the start of the real-time job data stream, with the initial truncation including the first to last frames. The data segment consists of sampling points. After the initial truncation is completed, the window moves forward. The sampling point, the second cut includes the first sampling point. To the The window continuously shifts in fixed increments, triggering a data truncation operation after each shift until all available data has been processed. The data window captured during each shift contains... First, statistical features are extracted from the data within the window using multiple sampling points. These statistical features include descriptive indicators such as mean, variance, maximum value, minimum value, peak-to-peak value, skewness, and kurtosis. These indicators reflect the distribution characteristics and trends of physical quantities within the window.

[0065] For time-series features, the first and second differences of the data within the window are calculated to extract the rate of change and acceleration information. Frequency domain features are extracted using Fast Fourier Transform (FFT) to calculate parameters such as the dominant frequency component, spectral energy distribution, and bandwidth power, used to identify periodic motion patterns. Contextual data features specifically refer to the correlation characteristics between the data in the first and second halves of the window. The window is divided into a first half and a second half, and their respective statistical features are calculated. Then, the correlation coefficient, mutual information, or dynamic time warping distance between the two halves are calculated to quantify the continuity between the preceding and following states. Based on the extracted contextual features, a data structure for overlapping state units is generated. Overlapping state units include fields such as the original data window, statistical feature vector, contextual correlation metric, and timestamp range. The overlap attribute of the units is reflected in the fact that adjacent units share some original data points, and the length of the shared region is... The overlapping design ensures state continuity between units, with the initial features of the later unit being highly similar to the later features of the previous unit, resulting in a smooth transition in state evolution.

[0066] Multiple overlapping state units first undergo data compression to reduce transmission load. The selection of the compression algorithm needs to balance compression ratio and computational complexity. Lossless compression algorithms such as LZ77, LZ78, or Huffman coding are used to ensure data integrity. Predictive coding is performed using temporal and spatial correlations, taking into account the characteristics of sensor data. Temporal correlation compression is achieved through differential coding, storing the difference between adjacent sampling points rather than their absolute values. The dynamic range of the difference is usually much smaller than the original data, allowing it to be represented with fewer bits. Spatial correlation compression utilizes redundancy between different sensors to jointly encode highly correlated sensor data. The compressed data is encapsulated into transmission data packets. The data packet structure includes a header, payload, and checksum. The header records the data packet sequence number, timestamp, data type, and compression algorithm identifier. The payload stores the compressed overlapping state unit data, and the checksum is calculated using a cyclic redundancy check (CRC) code for error detection at the receiver. The underwater acoustic modulator / demodulator converts the digital data packets into acoustic signals for transmission. Modulation methods typically employ anti-multipath fading techniques such as frequency shift keying (FPS), phase shift keying (PPS), or orthogonal frequency division multiplexing (OFDM). The carrier frequency of the acoustic signal is selected in the range of 10 to 30 kHz, which exhibits minimal attenuation and controllable Doppler shift during underwater propagation. The transmission power is dynamically adjusted according to the transmission distance; lower power is used for short-distance transmission to conserve energy, while higher power is used for long-distance transmission to counteract channel attenuation. An automatic retransmission request mechanism is employed for data packet transmission; when the receiver detects a checksum error, it sends a retransmission request to the transmitter to ensure reliable data delivery.

[0067] The present invention also discloses a blockchain-based evidence storage and traceability system for diving operation data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the blockchain-based evidence storage and traceability method for diving operation data as described in any of the above.

[0068] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0069] The memory can be an internal storage unit of a computer device, such as a hard disk or RAM, or an external storage device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) provided on the computer device. Furthermore, the memory can be a combination of internal storage units and external storage devices of a computer device. The memory is used to store computer programs and other programs and data required by the computer device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0070] The present invention also discloses a computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to be configured to perform the blockchain-based evidence storage and traceability method for diving operation data described in any of the above embodiments.

[0071] The computer program can be stored in a machine-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The machine-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the machine-readable medium includes, but is not limited to, the above-mentioned components.

[0072] The blockchain-based evidence storage and traceability method for diving operation data described in the above embodiments is stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the above method.

[0073] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0074] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A blockchain-based method for storing and tracing diving operation data, characterized in that, Includes the following steps: Real-time operational data from multiple underwater sensors in diving operations are collected. A preset length of overlapping sliding window is set to divide the continuous real-time operational data into overlapping state units with nested context states. Multiple overlapping state units are then encapsulated and transmitted to the data center. By extracting the preceding and following state features reflecting the evolution of physical states within each overlapping state unit from the data center, a dynamic directed topological graph representing the succession relationship of underlying states is constructed in memory space using the preceding and following state features as nodes. In the dynamic directed topology graph, the path traversal algorithm is used to eliminate redundant loops caused by channel interference and perform breakpoint interpolation to extract the acyclic traversal path that can connect all valid nodes. Based on the acyclic traversal path, a single-source diving time-series state chain with physical time consistency is reconstructed. The single-source diving time-series state chain corresponding to each multi-source underwater sensor is transformed into a directed state snapshot branch containing hash pointer references. When different single-source diving time-series state chains logically intersect, the directed state snapshot branches are merged based on hash pointers to construct an asynchronous state evolution network that represents the evolution of integrated diving actions. When the asynchronous state evolution network is ready to submit evidence, the state transition feature parameters between adjacent nodes in the asynchronous state evolution network are extracted, and the state transition feature parameters are cross-collision matched with the preset underwater physical constraint benchmark. Abnormal nodes that fail to match are blocked and independently assigned to isolated evidence storage branches with physical failure markers. The complete asynchronous state evolution network, including isolated evidence storage branches, is hashed, packaged, and broadcast to the blockchain network for consensus on-chain. During the tracing phase, the topology of the asynchronous state evolution network is reverse-analyzed based on the root hash on the blockchain to trace back the physical evolution trajectory and abnormal state source of the diving operation data.

2. The blockchain-based evidence storage and traceability method for diving operation data according to claim 1, characterized in that, The process of eliminating redundant loops caused by channel interference and performing breakpoint interpolation in a dynamic directed topology graph using a path traversal algorithm to extract acyclic traversal paths that connect all valid nodes, and reconstructing a single-source diving time-series state chain with physical time consistency based on the acyclic traversal paths, includes the following steps: Identify the in-degree and out-degree of nodes in a dynamic directed topology graph, and use a path traversal algorithm to detect multipath redundant blocks in the dynamic directed topology graph based on the in-degree and out-degree of nodes. Evaluate the topological equivalence of multipath redundant tiles and eliminate multipath redundant tiles based on topological equivalence to generate a cleaned connected subgraph; Identify graph breakpoints caused by signal loss in connected subgraphs and perform breakpoint interpolation; Connect the connected subgraphs after breakpoint interpolation, extract the acyclic traversal path that can connect all valid nodes, and reconstruct the single-source diving time-series state chain with physical time consistency based on the acyclic traversal path.

3. The blockchain-based evidence storage and traceability method for diving operation data according to claim 2, characterized in that, The process of identifying graph breakpoints caused by signal loss in a connected subgraph and performing breakpoint interpolation includes the following steps: Extract the edge free nodes in the connected subgraph and calculate the state gradient difference between the edge free nodes and the target free nodes in the adjacent connected subgraphs; When the state gradient difference is less than the preset physical evolution extreme value, a virtual bridging state feature is generated. Virtual directed edges are constructed between edge free nodes and target free nodes by utilizing virtual bridging state characteristics; By connecting the corresponding connected subgraphs with virtual directed edges, the breakpoint interpolation of the graph breakpoints is completed.

4. The blockchain-based evidence storage and traceability method for diving operation data according to claim 3, characterized in that, The calculation of the state gradient difference between the edge free node and the target free node in the adjacent connected subgraph includes the following steps: Read the first physical state vector contained in the edge free node, and read the second physical state vector contained in the target free node in the adjacent connected subgraph; Obtain the Euclidean distance between the first physical state vector and the second physical state vector in the data space; Divide the Euclidean distance by the theoretical interval steps between the edge free node and the target free node in the received sequence to obtain the unit evolution rate of change, and use the unit evolution rate of change as the state gradient difference between the edge free node and the target free node.

5. The blockchain-based evidence storage and traceability method for diving operation data according to claim 1, characterized in that, The process of transforming the single-source diving time-series state chains corresponding to each multi-source underwater sensor into directed state snapshot branches containing hash pointer references, and merging the directed state snapshot branches based on hash pointers to construct an asynchronous state evolution network representing the evolution of integrated diving actions when different single-source diving time-series state chains logically intersect, includes the following steps: In each single-source diving time-series state chain, key nodes representing physical parameter jumps are selected and encapsulated into independent data snapshot blocks. The first hash value of the previous data snapshot block is calculated using a cryptographic hash function; The first hash value is embedded as a hash pointer into the current data snapshot block to form a directed state snapshot branch; Monitor the diving action tags between directed state snapshot branches corresponding to different multi-source underwater sensors. When multiple directed state snapshot branches with the same diving action tag are ready within the time window, generate a global merge node containing the hash values ​​of the ends of all ready branches. By using a global merging node to logically bind multiple directed state snapshot branches, an asynchronous state evolution network representing the evolution of integrated diving actions is constructed.

6. The blockchain-based evidence storage and traceability method for diving operation data according to claim 1, characterized in that, When the asynchronous state evolution network is ready to submit evidence, the process of extracting state transition feature parameters between adjacent nodes in the asynchronous state evolution network, performing cross-collision matching between the state transition feature parameters and a preset underwater physical constraint benchmark, and blocking abnormal nodes that fail to match and independently assigning them to an isolated evidence storage branch with a physical failure marker includes the following steps: The data payload difference and time extrapolation step size of adjacent nodes in an asynchronous state evolution network are analyzed, and the state transition characteristic parameters are calculated based on the data payload difference and time extrapolation step size. Connect to the smart contract environment built into the data center blockchain system, and retrieve the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries. The state transition feature parameters are input into a matching engine containing static constraint boundaries for cross-collision matching; Identify anomalous nodes that fail to match beyond the static constraint boundary, sever the association between the anomalous nodes and the asynchronous state evolution network backbone, and transfer the anomalous nodes to an isolated evidence-keeping branch with attached physical failure markers.

7. The blockchain-based evidence storage and traceability method for diving operation data according to claim 6, characterized in that, The smart contract environment built into the data center blockchain system, which retrieves the hydrostatic and thermodynamic formula parameters from the preset underwater physical constraint benchmark as static constraint boundaries, includes the following steps: Connect to the smart contract storage module built into the data center blockchain system and read the absolute environmental pressure threshold corresponding to the current diving depth from the smart contract storage module; Derivation of fluid statics formula parameters based on absolute environmental pressure threshold; The composition of the gas mixture in the diver's breath was read and the theoretical gas partial pressure limit was calculated. The theoretical gas partial pressure limit was then converted into thermodynamic formula parameters. The parameters of the hydrostatic formula and the thermodynamic formula obtained from the derivation are set together as the static constraint boundary for cross-collision matching.

8. The blockchain-based evidence storage and traceability method for diving operation data according to claim 1, characterized in that, The process of collecting real-time operational data from multiple underwater sensors in a diving operation scenario, dividing continuous real-time operational data into overlapping state units with nested contextual states using a preset length of overlapping sliding window, and encapsulating and transmitting multiple overlapping state units to the data center includes the following steps: The multi-source underwater sensors carried by the diver are activated to collect analog signals and convert them into real-time operational data. Load a preset length of overlapping sliding window into the data buffer, and control the overlapping sliding window to translate and capture real-time job data in steps smaller than its own length. In each translation and truncation, contextual data features are extracted, and overlapping state units that retain the continuity between the preceding and following states are generated based on the contextual data features. Multiple overlapping state units are compressed, and all the compressed overlapping state units are encapsulated and transmitted to the data center using an underwater acoustic modem.

9. A blockchain-based evidence storage and traceability system for diving operation data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the blockchain-based evidence storage and traceability method for diving operation data as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, this instruction causes the processor to be configured to perform the blockchain-based evidence storage and traceability method for diving operation data according to any one of claims 1 to 8.

Images