Blockchain-based low-altitude economy data management method and system

Abstract

The application discloses a low-altitude economy data management method and system based on a block chain, relates to the technical field of information management, and comprises the following steps: inputting a multidimensional low-altitude economy data set into a space-time fusion network model; performing topological relation analysis and neighborhood feature aggregation by a space relation analysis layer; performing cross-dimension feature interaction by a space-time interaction layer; outputting a low-altitude space compliance smart contract; performing symbolization analysis and Kriging space interpolation on the low-altitude space compliance smart contract; obtaining a space risk heat map; synchronously performing multi-layer pyramid convolution on the space risk heat map; forming a space risk value matrix; and performing conditional mask calculation on the space risk value matrix to obtain a violation probability distribution. The space-time fusion network model and the secure hash algorithm greatly enhance the security and privacy protection level of the low-altitude economy data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information management, in particular to a low-altitude economy data management method and system based on a blockchain. BACKGROUND

[0002] With the rapid development of unmanned aerial vehicles and low-altitude aircraft, low-altitude economic activities are becoming increasingly frequent, covering multiple application scenarios such as logistics transportation, agricultural plant protection, and emergency rescue. In order to ensure the safety and compliance of low-altitude flight, it is urgent to efficiently integrate and manage low-altitude economic data. Traditional low-altitude economic data management methods mainly rely on centralized database systems and rule-based expert systems to realize airspace scheduling and flight supervision through pre-set logical judgments and manual intervention, which can meet the data processing needs of early low-altitude flight activities to a certain extent.

[0003] However, the existing technology still has some deficiencies. On the one hand, the current data management method lacks effective adaptability to dynamically changing low-altitude environments and aircraft behavior, making it difficult to update flight safety-related decision support information in real time. On the other hand, in terms of data security and privacy protection, the existing data management method cannot provide sufficient safeguards, making low-altitude economic data vulnerable to unauthorized access or tampering. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a low-altitude economy data management method based on a blockchain to solve the problems of weak adaptability and insufficient security of existing data management methods.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a low-altitude economy data management method based on a blockchain, which includes inputting a multi-dimensional low-altitude economy data set into a spatio-temporal fusion network model, a spatial relationship analysis layer performing topological relationship analysis and neighborhood feature aggregation, a spatio-temporal interaction layer performing cross-dimensional feature interaction, and outputting a low-altitude airspace compliance smart contract.

[0008] Performing symbolic analysis and Kriging spatial interpolation on the low-altitude airspace compliance smart contract to obtain an airspace risk heat map, and simultaneously performing multi-layer pyramid convolution on the airspace risk heat map to form an airspace risk value matrix.

[0009] Performing conditional mask calculation on the airspace risk value matrix to obtain a violation probability distribution, and according to the violation probability distribution, performing dynamic key fragmentation management on the low-altitude airspace compliance smart contract to generate partitioned encrypted data streams.

[0010] The time-space marking coding is performed on the partition encrypted data stream by using a secure hash algorithm, a blockchain storage record is obtained, zero-knowledge audit proof is performed on the blockchain storage record, and a data management audit report is output.

[0011] As a preferred scheme of the low-altitude economy data management method based on the blockchain, the multi-dimensional low-altitude economy data set comprises aircraft dynamic data, low-altitude airspace environment data and low-altitude airspace compliance rule data.

[0012] As a preferred scheme of the low-altitude economy data management method based on the blockchain, the output low-altitude airspace compliance smart contract specifically comprises the following steps,

[0013] The airspace relationship analysis layer and the time-space interaction layer are built, and the multi-scale stacking is performed by applying the skip connection to construct the time-space fusion network model.

[0014] The multi-dimensional low-altitude economy data set is input into the time-space fusion network model, the airspace relationship analysis layer performs the topological relationship analysis and the neighborhood feature aggregation by using the graph convolution to generate the airspace node state vector.

[0015] The time-space interaction layer applies the multi-head attention mechanism to perform the cross-dimensional feature interaction to obtain the time-space correlation weight matrix.

[0016] The airspace node state vector and the weight matrix are subjected to the importance weighting and the cross-modal feature fusion by using the gating unit to form the airspace rule feature tensor.

[0017] The airspace rule feature tensor is subjected to the contract structure decoding and the parameter filling to output the low-altitude airspace compliance smart contract.

[0018] As a preferred scheme of the low-altitude economy data management method based on the blockchain, the airspace risk heat map is obtained by specifically comprising the following steps,

[0019] The low-altitude airspace compliance smart contract is subjected to the logical structure decomposition to form the structured rule matrix.

[0020] The structured rule matrix is subjected to the symbolic analysis and the weight re-allocation by using the entropy weight method to obtain the symbolic rule parameter distribution.

[0021] The symbolic rule parameter distribution is subjected to the Kriging space interpolation and the probability surface fitting to generate the airspace risk heat map.

[0022] As a preferred scheme of the low-altitude economy data management method based on the blockchain, the airspace risk value matrix is formed by specifically comprising the following steps,

[0023] The spatial domain risk heat map is subjected to wavelet transform and multi-scale decomposition to obtain a multi-resolution feature map sequence, and multi-layer pyramid convolution is performed on the resolution feature map sequence to obtain a multi-level convolution feature map;

[0024] Channel attention weighting and cross-scale feature fusion are performed on the multi-level convolution feature map to form a fusion feature tensor, and linear projection is performed on the fusion feature tensor to output a spatial domain risk value matrix.

[0025] As a preferred scheme of the low-altitude economic data management method based on the blockchain, the method comprises the following steps of:

[0026] The spatial domain risk value matrix is subjected to risk quantization and conditional mask calculation to generate a binary mask, and the binary mask is subjected to spatial sliding window statistics to obtain a violation probability distribution.

[0027] According to the violation probability distribution, the low-altitude airspace compliance smart contract is subjected to dynamic permission adjustment to form an encryption control parameter, and the encryption control parameter is subjected to dynamic key fragmentation management to generate a partitioned encrypted data stream.

[0028] As a preferred scheme of the low-altitude economic data management method based on the blockchain, the method comprises the following steps of:

[0029] The partitioned encrypted data stream is subjected to coordinate binding and space-time marker coding by using a secure hash algorithm to obtain a space-time marker hash sequence, and the space-time marker hash sequence is subjected to hierarchical aggregation to generate a blockchain storage record.

[0030] The blockchain storage record is subjected to space-time feature extraction and reorganization to form an audit evidence set, and the audit evidence set is subjected to zero-knowledge audit proof to obtain a verifiable statement.

[0031] The verifiable statement is subjected to structured integration to output a data management audit report.

[0032] In a second aspect, the application provides a low-altitude economic data management system based on a blockchain, comprising:

[0033] The contract generation module is configured to input a multi-dimensional low-altitude economic data set into a space-time fusion network model, a spatial domain relationship analysis layer is configured to perform topological relationship analysis and neighborhood feature aggregation, and a space-time interaction layer is configured to perform cross-dimensional feature interaction to output a low-altitude airspace compliance smart contract.

[0034] The risk assessment module is configured to perform symbolic analysis and Kriging spatial interpolation on the low-altitude airspace compliance smart contract to obtain a spatial domain risk heat map, and perform multi-layer pyramid convolution on the spatial domain risk heat map to form a spatial domain risk value matrix.

[0035] A secure encryption module is configured to perform conditional mask calculation on the airspace risk value matrix, obtain a violation probability distribution, and perform dynamic key fragmentation management on the low-altitude airspace compliance smart contract according to the violation probability distribution, thereby generating partitioned encrypted data streams.

[0036] An audit evidence storage module is configured to perform space-time marker coding on the partitioned encrypted data streams by using a secure hash algorithm, obtain a blockchain evidence record, perform zero-knowledge audit proof on the blockchain evidence record, and output a data management audit report.

[0037] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, any step of the blockchain-based low-altitude economic data management method according to the first aspect of the present application is implemented.

[0038] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, any step of the blockchain-based low-altitude economic data management method according to the first aspect of the present application is implemented.

[0039] The present application has the following beneficial effects: the spatio-temporal fusion network model is used to perform real-time processing and analysis on multi-dimensional low-altitude economic data, thereby significantly improving the adaptability to dynamic changes in the environment and the behavior of aircraft. The secure hash algorithm is used to perform space-time marker coding on the partitioned encrypted data streams, thereby ensuring that the data is not subject to unauthorized access and tampering, and greatly enhancing the security and privacy protection level of the low-altitude economic data. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0041] Fig. 1 The flowchart of the blockchain-based low-altitude economic data management method.

[0042] Fig. 2 The schematic diagram of the blockchain-based low-altitude economic data management system.

[0043] Fig. 3 The flowchart of the spatio-temporal fusion network model processing process.

[0044] Fig. 4 The flowchart of the airspace risk assessment and encryption. DETAILED DESCRIPTION

[0045] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0046] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. The present application, however, can be practiced in a variety of ways other than those specifically described herein, and the present application can be practiced in embodiments other than those explicitly described herein, without departing from the spirit or scope of the present application. It should be recognized, therefore, that the specific embodiments described herein are not intended as being limiting, but rather as being illustrative.

[0047] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is independent or alternative to other embodiments.

[0048] Reference Figs. 1-4 For one embodiment of the present application, the embodiment provides a low-altitude economy data management method based on a blockchain, comprising the following steps:

[0049] S1, inputting a multi-dimensional low-altitude economy data set into a space-time fusion network model, a spatial relationship analysis layer performing topological relationship analysis and neighborhood feature aggregation, a space-time interaction layer performing cross-dimension feature interaction, and outputting a low-altitude airspace compliance smart contract.

[0050] Specifically, the method comprises the following steps:

[0051] S1.1, collecting a multi-dimensional low-altitude economy data set, the multi-dimensional low-altitude economy data set including aircraft dynamic data, low-altitude airspace environment data, and low-altitude airspace compliance rule data.

[0052] The aircraft dynamic data includes position information, speed parameters, and flight states; the position information is collected by an onboard GPS, the speed parameters are collected by an inertial measurement unit, and the flight states are collected by a flight control unit;

[0053] The low-altitude airspace environment data includes weather conditions, terrain and topography, and electromagnetic interference intensity; the weather conditions are captured by a weather radar, the terrain and topography are collected by a laser radar scanning device, and the electromagnetic interference intensity is collected by a spectrum analyzer;

[0054] The low-altitude airspace compliance rule data includes airspace division rules, flight permission conditions, and no-fly area restrictions; the airspace division rules are collected from an air traffic control database, the flight permission conditions are extracted from aviation regulations documents, and the no-fly area restrictions are collected by a geographic fence unit.

[0055] S1.2, preprocessing the multi-dimensional low-altitude economic data set, in specific operation, for aircraft dynamic data, mean filtering is applied to filter out noise, and linear interpolation method is used to fill in missing values to ensure data continuity, and time alignment is performed through PTP time stamp protocol to enhance the time sequence consistency of the data; for low-altitude airspace environment data, Min-Max standardization is applied to unify the dimension and eliminate unit differences, and Kriging interpolation is used for spatial sampling to eliminate abnormal data and improve the reliability of low-altitude airspace environment data; for low-altitude airspace compliance rule data, structure standardization and semantic normalization are performed to eliminate ambiguity, and principal component analysis is used for dimension compression to ensure storage efficiency, and the preprocessed multi-dimensional low-altitude economic data set is output.

[0056] S1.3, constructing a space-time fusion network model and training, in specific operation, in the PyTorch framework, the graph convolution network architecture is called through nn.Module base class, the node feature dimension of the graph convolution network architecture is set to 128, the number of attention heads is set to 4, and the activation function is set to LeakyReLU; layer normalization is connected after the graph convolution network architecture for gradient stabilization to suppress node feature scale drift, and feature reservation is performed through residual connection to complete the construction of the airspace relationship analysis layer; the Transformer architecture is called through nn.keras parameters, and the Transformer architecture is added with multi-head attention mechanism to realize cross-dimensional feature interaction; the Key dimension of the Transformer architecture is set to 64, the Value dimension is set to 64, and the Dropout rate is set to 0.1; gate linear unit is connected after the Transformer architecture for feature screening to enhance effective information flow, and bidirectional time convolution is used for time sequence feature enhancement to complete the construction of the space-time interaction layer;

[0057] The airspace relationship analysis layer and the space-time interaction layer are connected through a skip connection to perform cross-scale feature splicing, obtain a 256-dimensional fusion feature tensor, perform 1x1 convolution dimension reduction on the 256-dimensional fusion feature tensor, and generate a 128-dimensional rule feature vector; the importance of the 128-dimensional rule feature vector is weighted using the Softmax function to obtain the contribution degree distribution of each scale feature; the contribution degree distribution of each scale feature is dynamically weighted to obtain a multi-scale fusion weight; according to the multi-scale fusion weight, the airspace relationship analysis layer and the space-time interaction layer are stacked in multiple scales, and cross-scale feature aggregation is performed through three-dimensional convolution to complete the construction of the space-time fusion network model.

[0058] Next, the spatio-temporal fusion network model is trained. Further, the historical multi-dimensional low-altitude economic data set is divided into a sample set, a training set, and a validation set in a ratio of 7:2:1. On the sample set, data augmentation is performed using random affine transformation to form augmented training samples. On the training set, the augmented training samples are subjected to gradient backpropagation using AdamW, and a cosine annealing learning rate scheduler is applied simultaneously to dynamically adjust the learning rate, obtaining the intermediate parameters of the spatio-temporal fusion network model. On the validation set, the intermediate parameters of the spatio-temporal fusion network model are monitored for loss using an early stopping mechanism, obtaining the average absolute error of the validation set. When the average absolute error of the validation set exceeds the convergence threshold for 15 consecutive rounds, the training is terminated, and the trained spatio-temporal fusion network model is output simultaneously using ModelSnapshot.

[0059] It should be noted that the convergence threshold is defined based on the three times sliding standard deviation of the average absolute error of the validation set, with a value range of [0.03, 0.07].

[0060] S1.4, using the spatio-temporal fusion network model to generate a low-altitude airspace compliance intelligent contract. In specific operations, the multi-dimensional low-altitude economic data set is input into the trained spatio-temporal fusion network model through a standardized data interface. The airspace relationship analysis layer uses three-dimensional graph convolution to perform spatio-temporal feature separation on the multi-dimensional low-altitude economic data set, obtaining flight position features and environmental correlation features. The flight position features are used as node attributes, and the environmental correlation features are used as edge weights to construct a structured airspace topology graph.

[0061] A double-layer graph convolution network architecture is applied simultaneously to perform topological relationship analysis on the structured airspace topology graph. Further, the first layer applies graph convolution to perform local neighborhood feature propagation, obtaining primary node features. The primary node features are subjected to layer normalization and residual connection to generate stable node representations. The second layer applies high-order graph convolution to the stable node representations to perform cross-hop neighborhood aggregation and nonlinear transformation, obtaining high-order topological features. The high-order topological features are subjected to feature concatenation and linear projection to obtain topological relationship representations.

[0062] A gating attention unit is applied to weight the importance of the topological relationship representations, obtaining refined topological features. Neighborhood feature aggregation is performed on the refined topological features to generate airspace enhanced features. A multilayer perceptron is applied to the airspace enhanced features to perform nonlinear mapping and dimension compression, obtaining a 128-dimensional airspace node state vector.

[0063] The space-time interaction layer performs space-time feature separation on the multi-dimensional low-altitude economic data set to obtain space-time decoupling features; performs dimension expansion and nonlinear activation on the space-time decoupling features to obtain enhanced trajectory representations; performs cross-dimensional feature interaction on the enhanced trajectory representations by using a multi-head attention mechanism (8 heads); further performs channel splitting and reorganization on the enhanced trajectory representations to generate multi-head attention subspaces; performs parallel attention calculation on the multi-head attention subspaces to obtain trajectory vectors, environment vectors and rule vectors; defines the trajectory vectors as query vectors, the environment vectors as key vectors, and the rule vectors as value vectors; performs scaling dot product similarity operation on the query vectors, the key vectors and the value vectors to obtain attention weight values of different dimensions, and the specific mathematical formula is as follows,

[0064] ;

[0065] wherein, represents the attention weight value, represents the query vector, represents the key vector, represents the transposition operation, represents the scaling factor, represents the value vector,

[0066] It should be explained that the scaling factor is defined based on the feature dimension of the key vector, and the value range is [32-512].

[0067] performing Softmax normalization on the attention weight values of different dimensions to obtain an attention distribution matrix; performing weighted and multi-head splicing on the attention distribution matrix to obtain a fusion feature matrix; performing batch normalization (BatchNorm) on the fusion feature matrix to obtain a space-time enhanced feature; performing linear projection and dimension compression on the space-time enhanced feature to output a 256-dimensional space-time correlation weight matrix.

[0068] The spatial node state vector and the weight matrix are subjected to feature splicing to obtain preliminary fusion features; the preliminary fusion features are subjected to channel dimension separation and GELU nonlinear activation to obtain multi-channel feature representation; the multi-channel feature representation is subjected to directional feature processing by using a bidirectional gate unit, further, an importance weighting of the multi-channel feature representation is performed by a feature screening gate of a forward gate unit to obtain forward refined features, and a dimension transformation and feature scaling of the forward refined features are performed by a feature projection gate to generate forward projection features; a context fusion gate of a backward gate unit performs environment feature injection on the generated forward projection features to form bidirectional interaction features, and a scale adjustment and distribution alignment of the bidirectional interaction features are performed by a feature normalization gate to obtain stable feature output; the stable feature output is subjected to spatio-temporal feature fusion to obtain unified feature representation, and the unified feature representation is subjected to tensor reorganization to obtain a four-order tensor structure containing spatial grid, time step, feature channel and rule dimension; the four-order tensor structure is subjected to inter-channel coupling to obtain a structured feature tensor, and the structured feature tensor is subjected to multi-scale projection by using a separable convolution to generate a spatial rule feature tensor.

[0069] Next, in the contract structure decoding phase, the spatial rule feature tensor is subjected to hierarchical decomposition by using Tucker decomposition, further, the spatial rule feature tensor is subjected to tensor unfolding and matrix reorganization to obtain an unfolded multi-dimensional matrix, the unfolded multi-dimensional matrix is subjected to least square projection to obtain a core tensor and a factor matrix, wherein the core tensor retains the potential structure of the spatial rule, and the factor matrix encodes the associated weights of each dimension (space / time / rule); the core tensor and the factor matrix are subjected to feature reweighting by using an attention mechanism to obtain a weighted feature representation, and the weighted feature representation is subjected to cross-dimensional feature fusion to obtain a contract decoding intermediate representation.

[0070] In the parameter filling phase, the contract decoding intermediate representation is subjected to dynamic feature mapping to convert the contract decoding intermediate representation to a parameter space to generate a parameterized contract field, the parameterized contract field is subjected to field alignment and parameter filling to obtain an initial contract clause; the initial contract clause is subjected to structured conversion and field optimization reorganization to generate a structured contract representation, and the structured contract representation is subjected to format serialization to output a low-altitude airspace compliance smart contract.

[0071] S2, the low-altitude airspace compliance smart contract is subjected to symbolic analysis and Kriging spatial interpolation to obtain an airspace risk heat map, and the airspace risk heat map is simultaneously subjected to multi-layer pyramid convolution to form an airspace risk value matrix.

[0072] Specifically includes the following steps,

[0073] S2.1, adopt non-negative matrix decomposition method to low altitude airspace compliance intelligent contract for logical structure decomposition, form structured rule matrix, in the specific operation, the characteristic decoupling of low altitude airspace compliance intelligent contract is carried out, the core semantic feature is obtained, and the logical relationship mapping and multilayer stacking are carried out on the core semantic feature, and the logical feature unit is generated; Next, the logical feature unit is dynamically projected and nonlinearly activated, and the logical structure representation is generated, and the iteration optimization and non-negative decomposition are carried out on the logical structure representation, and the initial basic matrix and weight coefficient matrix are extracted; The initial basic matrix is sparsified and orthogonally projected to form the rule base matrix, and the weight coefficient matrix is executed outlier rejection and normalized scaling to obtain the standardized weight matrix; The semantic clustering and rule merging are carried out on the standardized weight matrix and the rule base matrix, and the structured rule matrix is output.

[0074] S2.2, the structured rule matrix is symbolized and analyzed and the weight is redistributed by using entropy weight method, and the symbolized rule parameter distribution is obtained, in the symbolized analysis stage, the dimension feature in the structured rule matrix is extracted, and the range calculation is carried out on the dimension feature, the feature value dispersion is generated, the importance of the structured rule matrix is sorted according to the feature value dispersion, the feature sorting sequence is obtained, and the information entropy calculation is carried out on the feature sorting sequence, the information entropy value is obtained, and the specific mathematical formula is as follows,

[0075] ;

[0076] Among them, Information entropy value, The total number of dimensions of dimension feature, Dimension index, The probability weight of dimension ;

[0077] The information entropy value is dynamically scaled by using Min-Max standardization to obtain the initial allocation weight, and the semantic mapping is carried out on the initial allocation weight to obtain the symbolized weight; In the weight redistribution stage, the symbolized weight is hierarchically aggregated to obtain the optimization weight set, and the weight redistribution and dimension compression are carried out on the structured rule matrix according to the optimization weight set to form the symbolized rule parameter, and the multi-dimensional space projection and density clustering are carried out on the symbolized rule parameter to obtain the symbolized rule parameter distribution.

[0078] S2.3, the symbolization rule parameter distribution is kriging spatial interpolation and probability surface fitting, generate space risk heat map, in the specific operation, in the spatial interpolation stage, the symbolization rule parameter distribution is decoupled, the spatial correlation characteristics are extracted, the principal component decomposition is performed on the spatial correlation characteristics, the first principal component vector and the second principal component vector are generated, and the first principal component vector is used as the x axis in the three-dimensional space, the second principal component vector is used as the y axis in the three-dimensional space, a three-dimensional space interpolation network is constructed, then the three-dimensional space interpolation network is executed kriging spatial interpolation and neighborhood aggregation, and the initial interpolation result is obtained;

[0079] In the surface fitting stage, the probability density fitting is performed on the optimized interpolation result by using kernel density estimation, further, the neighborhood weighted average is performed on the optimized interpolation result, the initial probability distribution is generated, and the median filtering and surface fitting are performed on the initial probability distribution to obtain the probability density surface. At the same time, the probability density surface is adjusted in scale and converted in coordinates to form a standard heat map layer; the layer superposition is performed on the standard heat map layer to obtain the space risk heat map.

[0080] S2.4, the space risk heat map is wavelet transformed and multi-scale decomposed to obtain a multi-resolution feature map sequence, and multi-layer pyramid convolution is performed on the resolution feature map sequence to obtain a multi-level convolution feature map. In the specific operation, the space risk heat map is wavelet transformed, further, the time-frequency decomposition is performed on the space risk heat map to obtain a wavelet coefficient matrix, the coefficient normalization and feature reorganization are performed on the wavelet coefficient matrix to obtain a reorganized feature representation; next, the multi-scale decomposition is performed on the reorganized feature representation, further, the smoothing filtering and edge enhancement are respectively performed through low-pass and high-pass filtering to extract the approximate component and the detail component in the reorganized feature representation; the approximate component and the detail component are feature weighted and fused to obtain a fused feature map, and the bilinear interpolation and resolution alignment are performed on the fused feature map to generate a multi-resolution feature map sequence.

[0081] Then, a three-layer pyramid structure is applied to perform multi-layer pyramid convolution on the multi-resolution feature map sequence, further, the first layer pyramid structure performs 5x5 separable convolution on the multi-resolution feature map sequence, and gradually expands the receptive field through three-layer down-sampling to output a primary convolution feature map; the second layer pyramid structure uses the empty convolution to further expand the receptive field of the primary convolution feature map to obtain an enhanced feature map, and performs feature normalization on the enhanced feature map to obtain an intermediate convolution feature map; the third layer pyramid structure performs 1x1 point convolution on the intermediate convolution feature map to obtain a high-level convolution feature map, and performs cross-level feature fusion on the high-level convolution feature map to output a multi-level convolution feature map.

[0082] S2.5, perform channel attention weighting and cross-scale feature fusion on the multi-level convolution feature map to form a fusion feature tensor, perform linear projection on the fusion feature tensor to output a spatial risk value matrix, in the channel attention weighting stage, use global average pooling to perform channel feature compression and linear transformation on the multi-level convolution feature map to obtain channel attention weight; according to the channel attention weight, perform Sigmoid activation and channel weighting multiplication on the multi-level convolution feature map to generate a channel enhanced feature map; next, perform bilinear upsampling and resolution alignment on the channel enhanced feature map to obtain a standardized feature map, and perform cross-level feature splicing on the standardized feature map to generate a weighted multi-scale feature tensor;

[0083] In the cross-scale feature fusion stage, perform cross-scale fusion and residual connection on the weighted multi-scale feature tensor to obtain preliminary fusion features, and perform batch normalization and LeakyReLU activation on the preliminary fusion features to generate optimized fusion features; perform channel dimension reorganization and feature dimension reduction on the optimized fusion features to obtain a final fusion feature tensor;

[0084] In the linear projection stage, apply 1x1 convolution to the final fusion feature tensor for channel dimension reduction and probability mapping to output a normalized risk probability, perform feature space projection on the normalized risk probability to generate an initial risk matrix, perform edge smoothing and noise suppression on the initial risk matrix to obtain an optimized risk matrix, and perform probability normalization and physical dimension conversion on the optimized risk matrix to form a spatial risk value matrix.

[0085] S3, perform conditional mask calculation on the spatial risk value matrix to obtain a violation probability distribution, and according to the violation probability distribution, perform dynamic key fragmentation management on the low-altitude spatial compliance smart contract to generate partitioned encrypted data stream.

[0086] Specifically, the following steps are included,

[0087] S3.1, perform risk quantization and conditional mask calculation on the spatial risk value matrix to generate a binary mask; perform spatial sliding window statistics on the binary mask to obtain a violation probability distribution, in the risk quantization stage, perform numerical normalization on the spatial risk value matrix to obtain a risk value; use a segmentation threshold to perform region division on the spatial risk value matrix, for example, when the risk value is lower than the segmentation threshold, it is divided into a safe region (set to 0), and when the risk value is higher than the dynamic segmentation threshold, it is divided into a violation region (set to 1), perform matrix padding and edge mirror expansion on the division result to output a preliminary binary mask; then perform conditional mask calculation on the preliminary binary mask, further, use a 3x3 circular structure element matrix to perform close operation on the preliminary binary mask to output a close operation result value; then perform open operation on the close operation result value by using a 5x5 rectangular kernel matrix to eliminate isolated noise points to obtain a binary mask, the specific mathematical formula is as follows,

[0088] ;

[0089] wherein, denotes a binary mask, denotes a preliminary binary mask, denotes a circular structuring element matrix, denotes a rectangular kernel matrix, denotes a closing operation, denotes an opening operation;

[0090] It should be noted that the circular structuring element matrix refers to a 3x3 discretized disc kernel matrix used to eliminate the filling holes of the closing operation; the rectangular kernel matrix refers to a 5x5 all-1 matrix used to eliminate noise in the opening operation; the segmentation threshold is defined based on the statistical distribution characteristics of the risk value, and the value range is [0.15, 0.85].

[0091] Then, spatial sliding window statistics are performed on the binary mask, further, a 50x50 pixel square window is set, and the binary mask is slid with a step of 25 pixels, and a window statistical record of the binary mask is output; the window statistical record of the binary mask is spatially smoothed and aggregated by using mean filtering, and an initial probability distribution is obtained; probability normalization and region clustering are performed on the initial probability distribution, a region probability map is generated, and bilinear interpolation is performed on the region probability map to obtain a violation probability distribution.

[0092] S3.2, according to the violation probability distribution, the low-altitude airspace compliance smart contract is dynamically adjusted, the encryption control parameter is formed, the dynamic key fragmentation management is carried out on the encryption control parameter, and the partition encryption data stream is generated, in the specific operation, in the permission adjustment stage, the numerical value mapping is performed on the violation probability distribution, and the violation probability distribution is mapped to three numerical value intervals of [0, 0.3), [0.3, 0.6) and [0.6, 1.0]; wherein the interval [0, 0.3) corresponds to the 0-level permission level (fully open), the interval [0.3, 0.6) corresponds to the 1-level permission level (need to be approved), and the interval [0.6, 1.0] corresponds to the 2-level permission level (fully prohibited); according to the permission level, the low-altitude airspace compliance smart contract is dynamically adjusted, for example, in the 0-level permission level, the low-altitude airspace compliance smart contract is adjusted to be authorized to pass, in the 1-level permission level, the low-altitude airspace compliance smart contract is adjusted to be manually reviewed and approved, and in the 2-level permission level, the low-altitude airspace compliance smart contract is adjusted to be intercepted and forbidden to pass, in the adjustment process, the contract state variable, the timestamp and the space coordinate are recorded as the dynamic control parameter, and the dynamic control parameter is executed asymmetrically encrypted to obtain the encryption control parameter;

[0093] In the key fragment management stage, the encryption control parameters are randomly sampled and parameter aggregated to obtain a key encryption fragment set; the key encryption fragment set is distributed stored to obtain a fragment storage voucher; the fragment storage voucher is executed on-chain storage and terminal synchronization broadcast on the blockchain to obtain a distributed encryption record; and the distributed encryption record is flow data encapsulated to generate partitioned encryption data flow.

[0094] S4, the partitioned encryption data flow is executed space-time marker coding using a secure hash algorithm to obtain a blockchain storage record, and the blockchain storage record is executed zero-knowledge audit proof to output a data management audit report.

[0095] Specifically includes the following steps,

[0096] S4.1, the partitioned encryption data flow is executed coordinate binding and space-time marker coding using a secure hash algorithm to obtain a space-time marker hash sequence, and the space-time marker hash sequence is hierarchically aggregated to generate a blockchain storage record. In the coordinate binding stage, the partitioned encryption data flow is executed space feature extraction to obtain a space feature vector; the space feature vector is executed space-time alignment using a PTP time stamp synchronization protocol to obtain space-time aligned data, and the space-time aligned data is executed coordinate projection transformation to obtain high-precision coordinate data, and the high-precision coordinate data is embedded in the partitioned encryption data flow to obtain coordinate binding data; in the space-time marker coding stage, the coordinate binding data is executed space-time dimension division to obtain space-time fragment data; the space-time fragment data is executed hash field mapping and sequence normalization to generate a primary hash sequence; the primary hash sequence is executed time stamp embedding to output a space-time marker hash sequence; in the blockchain storage stage, the space-time marker hash sequence is executed hierarchical aggregation processing: the first layer is executed period bucketing and hash aggregation on the space-time marker hash sequence according to time dimension to obtain a daily time aggregation root; the second layer is executed block division and space fusion on the daily time aggregation root according to space dimension to obtain a space-time aggregation proof; the third layer writes the space-time aggregation proof and the daily time aggregation root into an Ethereum blockchain to generate a package blockchain storage record.

[0097] S4.2, the spatio-temporal feature extraction and reorganization of the blockchain storage record is performed to form an audit evidence set, and zero-knowledge audit proof is performed on the audit evidence set to obtain a verifiable statement. In the feature extraction and reorganization stage, random stratified sampling is performed on the blockchain storage record to obtain original storage data, time stamp extraction and coordinate decoding are performed on the original storage data to obtain a spatio-temporal data tuple, multi-dimensional decomposition is performed on the spatio-temporal data tuple to obtain a spatio-temporal feature vector, feature reorganization is performed on the spatio-temporal feature vector to generate a standardized feature matrix, feature weight distribution and weighted aggregation are performed on the standardized feature matrix to output the audit evidence set; in the zero-knowledge proof stage, the zk-SNARKs protocol is used to perform key mapping on the audit evidence set to obtain a proof key, the proof key is subjected to key pair separation to generate a verification key, and zero-knowledge proof and statement encapsulation are performed on the proof key and the verification key to output the verifiable statement.

[0098] Data desensitization and format conversion are performed on the verifiable statement to output a standard audit statement, the standard audit statement is subjected to security encapsulation to generate an audit evidence package, and the audit evidence package is subjected to structured integration to output a data management audit report.

[0099] The embodiment also provides a low-altitude economic data management system based on a blockchain, comprising: a contract generation module configured to input a multi-dimensional low-altitude economic data set into a spatio-temporal fusion network model, a spatial relationship analysis layer configured to perform topological relationship analysis and neighborhood feature aggregation, and a spatio-temporal interaction layer configured to perform cross-dimensional feature interaction, and output a low-altitude airspace compliance smart contract;

[0100] A risk assessment module is configured to perform symbolic analysis and Kriging spatial interpolation on the low-altitude airspace compliance smart contract to obtain an airspace risk heat map, and simultaneously perform multi-layer pyramid convolution on the airspace risk heat map to form an airspace risk value matrix.

[0101] A security encryption module is configured to perform conditional mask calculation on the airspace risk value matrix to obtain a violation probability distribution, and perform dynamic key fragmentation management on the low-altitude airspace compliance smart contract according to the violation probability distribution to generate partitioned encrypted data streams.

[0102] An audit storage module is configured to perform spatio-temporal marker coding on the partitioned encrypted data streams by using a secure hash algorithm to obtain a blockchain storage record, perform zero-knowledge audit proof on the blockchain storage record, and output a data management audit report.

[0103] The embodiment also provides a computer device suitable for the low-altitude economic data management method based on a blockchain, comprising: a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions to implement the low-altitude economic data management method based on a blockchain as described in the above embodiment.

[0104] The computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved by WIFI, operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device. In addition, the input device can also be an external keyboard, touchpad or mouse, etc.

[0105] The embodiment also provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the method for managing low-altitude economic data based on a block chain as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0106] In summary, the present application significantly improves the adaptability to dynamic changing environment and aircraft behavior by using a space-time fusion network model to perform real-time processing and analysis on multi-dimensional low-altitude economic data. The space-time marker coding of partitioned encrypted data stream is performed by using a secure hash algorithm to ensure that the data is free from unauthorized access and tampering, greatly enhancing the security and privacy protection level of low-altitude economic data.

[0107] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A blockchain-based method for managing low-altitude economic data, characterized in that: include, The multidimensional low-altitude economic dataset is input into the spatiotemporal fusion network model. The airspace relationship parsing layer performs topological relationship analysis and neighborhood feature aggregation, while the spatiotemporal interaction layer performs cross-dimensional feature interaction, outputting a low-altitude airspace compliant smart contract. Specifically, the steps include the following: The multidimensional low-altitude economic dataset includes aircraft dynamic data, low-altitude airspace environmental data, and low-altitude airspace compliance rule data. A spatial relationship resolution layer and a spatiotemporal interaction layer are constructed, and skip connections are applied to perform multi-scale stacking to build a spatiotemporal fusion network model. The multidimensional low-altitude economic dataset is input into the spatiotemporal fusion network model. The spatial domain relationship parsing layer uses graph convolution to perform topological relationship analysis and neighborhood feature aggregation to generate spatial domain node state vectors. The spatiotemporal interaction layer uses a multi-head attention mechanism to perform cross-dimensional feature interaction and obtain the spatiotemporal correlation weight matrix. A gating unit is used to perform importance weighting and cross-modal feature fusion on the spatial node state vector and weight matrix to form a spatial regular feature tensor; Decode the contract structure and fill in the parameters of the airspace rule feature tensor to output a low-altitude airspace compliant smart contract. In the contract structure decoding stage, Tucker decomposition is used to perform hierarchical decomposition on the spatial domain rule feature tensor. Furthermore, tensor expansion and matrix recombination are performed on the spatial domain rule feature tensor to obtain the expanded multidimensional matrix. Least square projection is performed on the expanded multidimensional matrix to obtain the core tensor and factor matrix. The core tensor retains the latent structure of the spatial domain rules, and the factor matrix encodes the correlation weights of each dimension. Feature reweighting is performed on the core tensor and factor matrix through an attention mechanism to obtain a weighted feature representation. Cross-dimensional feature fusion is performed on the weighted feature representation to obtain the intermediate representation for contract decoding. During the parameter filling stage, dynamic feature mapping is performed on the intermediate representation of the contract decoding to transform it into the parameter space, generating parameterized contract fields. Field alignment and parameter filling are performed on the parameterized contract fields to obtain the initial contract terms. The initial contract terms are then subjected to structured transformation and field optimization and reorganization to generate a structured contract representation. The structured contract representation is then serialized to output a low-altitude airspace compliant smart contract. The process involves performing symbolic parsing and kriging space interpolation on low-altitude airspace compliance smart contracts to obtain an airspace risk heatmap. Simultaneously, multi-layer pyramid convolution is performed on the airspace risk heatmap to form an airspace risk value matrix. The specific steps include the following: The logical structure of low-altitude airspace compliant smart contracts is decomposed to form a structured rule matrix; The entropy weight method is used to perform symbolic parsing and weight redistribution on the structured rule matrix to obtain the distribution of symbolic rule parameters; Kriging space interpolation and probability surface fitting are performed on the distribution of symbolic rule parameters to generate a spatial risk heat map. The airspace risk value matrix is ​​subjected to condition mask calculation to obtain the violation probability distribution. Based on the violation probability distribution, dynamic key sharding management is performed on the low-altitude airspace compliant smart contract to generate a partitioned encrypted data stream. The specific steps include the following: Risk quantification and conditional mask calculation are performed on the spatial risk value matrix to generate a binary mask; The probability distribution of violations is obtained by performing spatial sliding window statistics on the binary mask. Based on the probability distribution of violations, the permissions of compliant smart contracts in low-altitude airspace are dynamically adjusted to form encryption control parameters. Dynamic key fragmentation management is performed on the encryption control parameters to generate partitioned encrypted data streams. The secure hash algorithm is used to perform spatiotemporal tagging encoding on the partitioned encrypted data stream to obtain blockchain evidence records. Zero-knowledge audit proofs are then performed on the blockchain evidence records to output a data management audit report.

2. The blockchain-based low-altitude economic data management method as described in claim 1, characterized in that: The formation of the spatial risk value matrix specifically includes the following steps. The spatial risk heatmap is subjected to wavelet transform and multi-scale decomposition to obtain a multi-resolution feature map sequence, and multi-level pyramid convolution is performed on the resolution feature map sequence to obtain multi-level convolutional feature maps. Channel attention weighting and cross-scale feature fusion are performed on the multi-level convolutional feature maps to form a fused feature tensor. The fused feature tensor is then linearly projected to output a spatial risk value matrix.

3. The blockchain-based low-altitude economic data management method as described in claim 1, characterized in that: The output data management audit report specifically includes the following steps. The secure hash algorithm is used to perform coordinate binding and spatiotemporal tag encoding on the partitioned encrypted data stream to obtain the spatiotemporal tag hash sequence. The spatiotemporal tag hash sequence is then aggregated in a hierarchical manner to generate a blockchain evidence record. Spatiotemporal features are extracted and recombined from blockchain-stored records to form an audit evidence set. Zero-knowledge audit proofs are then performed on the audit evidence set to obtain verifiable statements. Verifiable claims are structurally integrated to output data management audit reports.

4. A blockchain-based low-altitude economic data management system, based on the blockchain-based low-altitude economic data management method according to any one of claims 1 to 3, characterized in that: include, The contract generation module is used to input multidimensional low-altitude economic datasets into the spatiotemporal fusion network model. The airspace relationship parsing layer performs topological relationship analysis and neighborhood feature aggregation, and the spatiotemporal interaction layer performs cross-dimensional feature interaction to output low-altitude airspace compliant smart contracts. The risk assessment module is used to perform symbolic parsing and kriging space interpolation on low-altitude airspace compliance smart contracts to obtain an airspace risk heat map. Simultaneously, multi-layer pyramid convolution is performed on the airspace risk heat map to form an airspace risk value matrix. The security encryption module is used to perform condition mask calculation on the airspace risk value matrix, obtain the violation probability distribution, and perform dynamic key fragmentation management on low-altitude airspace compliant smart contracts based on the violation probability distribution to generate partitioned encrypted data streams. The audit and evidence storage module is used to perform spatiotemporal tagging encoding on the partitioned encrypted data stream using a secure hash algorithm, obtain blockchain evidence storage records, perform zero-knowledge audit proofs on the blockchain evidence storage records, and output a data management audit report.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the blockchain-based low-altitude economic data management method according to any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the blockchain-based low-altitude economic data management method according to any one of claims 1 to 3.

Images