A heterogeneous analytical integration method for polar observation network data based on edge computing
By constructing a dynamic knowledge graph and optimizing the quantum instruction flow closed loop through edge computing, the problems of topological mismatch and quantum instruction transmission reliability in the polar observation network were solved, improving the anti-interference ability and control accuracy of instruction transmission in extreme environments and ensuring information security.
Patent Information
- Application Number
- CN202611561200.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-10
- Publication Date
- 2026-11-17
AI Technical Summary
In polar observation networks, knowledge graph topology mismatch and quantum instruction transmission reliability issues caused by dynamic changes in equipment topology affect the accuracy and security of scheduling strategies.
By constructing a dynamic knowledge graph through edge computing, combining it with graph neural networks to fuse node features, generating scheduling instructions, and employing polarization-phase composite quantum state transmission for polarization decoding and decoy state detection, closed-loop optimization of the quantum instruction flow is achieved.
It improves the anti-interference capability and control precision of command transmission in extreme environments, ensuring information security and spatiotemporal consistency of decision-making.
Smart Images

Figure CN123115292A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of edge computing and quantum communication interdisciplinary technology, and in particular to a method for heterogeneous analytical integration of polar observation network data based on edge computing. Background Technology
[0002] Polar observation networks, as critical infrastructure for polar scientific research, require processing large amounts of multi-source spatiotemporal data generated by heterogeneous equipment. Current mainstream solutions employ a centralized data processing architecture based on a cloud platform, integrating equipment data through standardized communication protocols and utilizing static knowledge graphs to model equipment relationships. In the area of data security, traditional public-key encryption systems are typically combined with the TLS transport layer protocol to ensure secure command transmission, while equipment scheduling commands are generated based on a predefined policy rule base. This approach has formed a relatively mature technical paradigm in conventional environments and can meet the collaborative control needs of basic observation tasks.
[0003] However, existing methods have significant limitations. On the one hand, the topology of polar equipment changes in real time due to ice displacement, node failures, etc., and the static graph cannot dynamically update the node association weights, resulting in the accumulation of feature fusion deviations between adjacent equipment and affecting the accuracy of the scheduling strategy output by the graph neural network. On the other hand, traditional encryption protocols are difficult to cope with channel attenuation in the ultra-low temperature environment of the polar region, and long-distance transmission leads to an increase in the error rate of quantum key distribution, which cannot meet the security threshold of real-time instructions and exacerbates the risk of decoding failure of scheduling instructions. Summary of the Invention
[0004] In view of the aforementioned existing problems, the present invention is proposed.
[0005] Therefore, this invention provides a method for heterogeneous analytical integration of polar observation network data based on edge computing to solve the problems of knowledge graph topological mismatch and quantum instruction transmission reliability in dynamic polar environments.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, this invention provides a method for heterogeneous data analysis and integration of polar observation networks based on edge computing. The method includes: connecting to field equipment via a mobile terminal near-field interaction, collecting effective parameters and basic equipment data, creating corresponding digital twin entities, and parsing them using a pre-set data analysis instruction set to obtain a standardized dataset; using the digital twin entities as nodes and combining them with the standardized dataset to construct a dynamic knowledge graph; based on the dynamic knowledge graph, fusing features of adjacent nodes through a graph neural network to generate a graph fusion feature matrix; defining an action space based on the graph fusion feature matrix, calculating real-time reward values, dynamically optimizing strategy parameters, and generating scheduling instructions; encoding the scheduling instructions into polarization-phase composite quantum states using a dual-field protocol to generate a quantum instruction stream, and performing polarization decoding and decoy state detection to restore it to executable control commands; performing output adjustment operations on the polar observation equipment based on the executable control commands, obtaining the actual response parameters of the equipment, transmitting the actual response parameters back to the digital twin entity, and calculating the scheduling deviation rate; optimizing the quantum instruction stream based on the scheduling deviation rate, and generating a polar observation equipment analysis and integration report.
[0007] As a preferred embodiment of the edge computing-based heterogeneous data analysis and integration method for polar observation networks described in this invention, the steps include: connecting to field equipment via a mobile terminal near-field interaction, collecting effective parameters and basic equipment data, creating corresponding digital twin entities, and parsing them using a preset data parsing instruction set to obtain a standardized dataset. The effective parameters include equipment output value, location coordinates, environmental parameters, and equipment status code; The basic equipment data includes a text description of the equipment type, numerical coordinates of the installation location, and text of maintenance history records. Bind valid parameters to basic device data to create a digital twin entity; The valid parameters are parsed field by field using a preset data parsing instruction set to generate a numerical data matrix. The numerical data matrix is subjected to Box-Cox transformation and Z-score standardization to obtain a standardized dataset.
[0008] As a preferred embodiment of the edge computing-based heterogeneous data analysis and integration method for polar observation networks described in this invention, the steps for constructing a dynamic knowledge graph by using digital twin entities as nodes and combining them with a standardized dataset are as follows. Each digital twin entity is converted into a knowledge graph node, and the entity type name is labeled to generate a set of knowledge graph nodes; Based on the knowledge graph node set, weighted edges with neighbor relationship type names are created, generating a neighbor relationship edge set; Retrieve the observation fields of the knowledge graph node set, match node pairs with the same observation text, create undirected edges with the measurement relation type name, and generate a measurement relation edge set; The numerical values of the standardized dataset are mapped to the corresponding ID nodes of the knowledge graph node set, and a dynamic knowledge graph is constructed by combining the neighbor relationship edge set and the measurement relationship edge set.
[0009] As a preferred embodiment of the heterogeneous analytical integration method for polar observation network data based on edge computing described in this invention, the step of fusing features of adjacent nodes through a graph neural network to generate a graph fusion feature matrix includes the following steps: Read the node attribute field values from the dynamic knowledge graph and generate a node feature vector matrix; Based on the set of adjacent edges and the set of measured edges, an adjacency matrix is created, and a summation operation is performed on the rows of the adjacency matrix to generate a node degree matrix; Based on the node degree matrix and the adjacency matrix, a normalized adjacency matrix is generated by performing matrix concatenation multiplication through symmetric normalization. The normalized adjacency matrix and the node feature vector matrix are subjected to a linear transformation operation through a three-layer graph convolution, and negative values are reduced to zero to generate a graph fusion feature matrix.
[0010] As a preferred embodiment of the edge computing-based heterogeneous analytical integration method for polar observation network data described in this invention, the steps of defining the action space based on the graph fusion feature matrix, calculating the real-time reward value, dynamically optimizing strategy parameters, and generating scheduling instructions are as follows: Extract the device output values from the graph fusion feature matrix, construct a state space matrix, and bind action names; Based on the equipment output value and the preset load demand value, calculate the power supply balance rate, and combine the equipment safety margin benchmark value and the line loss rate benchmark value to calculate the real-time bonus value; Based on the real-time reward value and the current state encoding of the state space matrix, the Bellman equation iteration is performed to obtain the updated policy parameter table; Based on the updated policy parameter table, the optimal action name is selected for each state code in the state space matrix using an ε-greedy policy to generate scheduling instructions.
[0011] As a preferred embodiment of the edge computing-based heterogeneous analytical integration method for polar observation network data described in this invention, the steps of encoding scheduling instructions into polarization-phase composite quantum states using a dual-field protocol to generate a quantum instruction stream, and performing polarization decoding and decoy state detection to restore them to executable control commands are as follows. The scheduling instructions are converted into a binary bit stream, quantum state encoding is performed using a dual-field protocol, a polarization-phase composite quantum state pulse sequence is output, and decoy state pulses are added to generate a quantum instruction stream; The decoy state pulses of the quantum instruction stream are separated, and the signal state pulses are polarized and split to obtain two optical intensity signals; The phase information of the two optical intensity signals is compared, and the binary bit stream is restored to obtain the decoded binary sequence; Based on the decoded binary sequence and the preset original decoy state sequence, the bit error rate of the decoy state pulse is calculated and verified, and an executable control command is output.
[0012] As a preferred embodiment of the edge computing-based heterogeneous analytical integration method for polar observation network data described in this invention, the steps of performing output adjustment operations on polar observation equipment, obtaining actual equipment response parameters, transmitting these parameters back to the digital twin entity, and calculating the scheduling deviation rate are as follows: Based on executable control commands, pulse width modulation signals are sent to polar observation equipment to perform power output adjustment operations; During the output adjustment operation, a cold-resistant sensor is used to acquire the equipment operating status in real time, generate a set of actual response parameters of the equipment, write them into the digital twin entity, and output the updated digital twin entity. Read the predicted output value and the actual output value from the updated digital twin entity, and calculate the scheduling deviation rate.
[0013] As a preferred embodiment of the edge computing-based heterogeneous analytical integration method for polar observation network data described in this invention, the steps for optimizing the quantum instruction stream and generating a polar observation equipment analytical integration report are as follows: The dual-field protocol parameters of the quantum instruction stream are adjusted according to the scheduling deviation rate to generate an optimized quantum instruction stream; By integrating the actual response parameters of the equipment, the scheduling deviation rate, and the optimized quantum command stream, an analytical integration report of polar observation equipment is generated.
[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the edge computing-based heterogeneous analytical integration method for polar observation network data as described in the first aspect of the present invention.
[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the edge computing-based heterogeneous analytical integration method for polar observation network data as described in the first aspect of the present invention.
[0016] The beneficial effects of this invention are as follows: By constructing a dynamic knowledge graph, spatiotemporal correlation modeling of multi-source heterogeneous data in the polar region is realized, enabling the knowledge graph to dynamically respond to environmental changes. Its beneficial effect lies in improving the completeness of state perception and the spatiotemporal consistency of decision-making basis. Through quantum instruction flow closed-loop optimization, a closed-loop link of quantum instruction generation → execution → feedback → optimization is established. Its beneficial effect lies in significantly improving the anti-interference capability and control accuracy of instruction transmission in extreme environments through adaptive adjustment of quantum state parameters, while ensuring information security throughout the entire process. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a flowchart of a heterogeneous analytical integration method for polar observation network data based on edge computing.
[0019] Figure 2 A flowchart for constructing a dynamic knowledge graph.
[0020] Figure 3 The flowchart shows the generation of the feature matrix for graph fusion.
[0021] Figure 4 A flowchart for optimizing the quantum instruction flow closed loop. Detailed Implementation
[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.
[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.
[0025] Reference Figures 1-4This is one embodiment of the present invention, which provides a heterogeneous analytical integration method for polar observation network data based on edge computing, including the following steps: S1. Connect to field equipment via near-field interaction of mobile terminal, collect effective parameters and basic equipment data, create corresponding digital twin entities, and parse them through a preset data parsing instruction set to obtain standardized datasets; Valid parameters include equipment output value, location coordinates, environmental parameters, and equipment status code; It should be noted that the equipment output value is collected in real time by the cold-resistant power sensor built into the equipment; the position coordinates are generated by the Beidou-3 dual-frequency positioning chip on the field equipment, which outputs latitude and longitude coordinates once per second; the environmental parameters refer to temperature, humidity and wind speed. Temperature and humidity are collected by the anti-freeze sensor connected by I²C bus, and wind speed is collected by the pulse counting cup anemometer; the equipment status code is read directly from the 16-bit status register of the field equipment. Basic equipment data includes a text description of the equipment type, numerical coordinates of the installation location, and text records of maintenance history. It should be noted that the equipment type text description is generated by reading the hexadecimal encoding of the QR code on the equipment nameplate using a cold-resistant industrial-grade QR code scanner, and then decoding and converting it using ASCII (American Standard Code for Information Interchange); the installation location coordinate values are real-time latitude and longitude coordinate values obtained using the equipment's built-in Beidou-3 dual-frequency positioning chip at a sampling frequency of 1Hz; the maintenance history text is the operation and maintenance event record extracted according to the equipment ID and sorted by the latest 3 timestamps. Bind valid parameters to basic device data to create a digital twin entity; Furthermore, the basic equipment data is first formatted to generate a structured basic data table, and timestamps and device IDs are injected into the valid data, outputting a parameter stream with spatiotemporal labels. Then, the structured basic data table and the parameter stream with spatiotemporal labels are dynamically bound using a spatiotemporal association key (device ID + timestamp), outputting a spatiotemporally aligned fused dataset. Next, based on the spatiotemporally aligned fused dataset, a digital twin entity is instantiated, outputting an initial-state digital twin entity. Finally, polar environment metadata is injected into the initial-state digital twin entity, outputting an enhanced digital twin entity. It should be noted that polar environmental metadata is obtained by loading pre-stored polar observation databases and static environmental benchmark parameters (such as ice thickness thresholds and historical minimum temperature extremes). It should be noted that the pre-stored polar observation database is a static environmental benchmark dataset loaded from historical scientific expeditions and satellite remote sensing data. It includes physical environmental benchmark constants necessary for equipment operation, such as critical values of ice thickness, historical extreme temperatures, glacier displacement vectors, geomagnetic declination calibration parameters, and polar night cycle duration. The valid parameters are parsed field by field using a preset data parsing instruction set to generate a numerical data matrix. Furthermore, firstly, the set of valid parameter fields to be parsed is extracted from the digital twin entity. At the same time, the environmental constraint rules in the preset data parsing instruction set are called. Anomaly detection and filtering are performed on the set of valid parameter fields according to the environmental constraint rules. The cleaned set of valid fields is output and then converted into numerical types field by field to output the basic numerical field set. The basic numerical field set is then aligned according to the spatiotemporal dimension to generate the final numerical data matrix. It should be noted that the pre-configured data parsing instruction set is an executable package loaded during device initialization that includes environmental constraint rule definitions, data type conversion functions, anomaly detection threshold tables, and vectorized recombination algorithms; the environmental constraint rules are a set of verification logic for polar scenarios in the pre-configured data parsing instruction set, including three core verification conditions: reasonable range of sensor values (e.g., temperature must be within [-60℃, 5℃]), regular expression of coordinate format, and whitelist of device status enumeration values. The numerical data matrix is subjected to Box-Cox transformation and Z-score standardization to obtain a standardized dataset. Furthermore, the numerical data matrix is subjected to Box-Cox transformation (power transformation): the power transformation parameters are iteratively calculated through maximum likelihood estimation, the transformation formula with compensation parameters is executed, and the normalized sequence is output; then the normalized sequence is standardized by Z-score, and the arithmetic mean and standard deviation within the window are calculated using a 100-sample sliding window, finally generating a standardized dataset with dimensions aligned with the input. The expression for Z-score standardization of the normalized sequence is: ; in, It is the first Window mean; It is the first Window standard values; It is a window index; This is the result after Z-score standardization; It is a node index; It is the first of the double-precision floating-point numerical matrices. The values of each element before the Box-Cox transformation.
[0026] S2. Using digital twin entities as nodes and combining them with standardized datasets, construct a dynamic knowledge graph; Each digital twin entity is converted into a knowledge graph node, and the entity type name is labeled to generate a set of knowledge graph nodes; Furthermore, the device ID field of each entity in the digital twin entity is first parsed, and basic device data is extracted to generate node basic data records containing device ID, type description, and location coordinates. Then, the node basic data records are matched with entity types, and the type descriptions are converted into standardized entity type names through a predefined mapping table, outputting enhanced node records with type labels. Next, the real-time operating parameters in the digital twin entity are read, and the entity operating parameters are aligned with timestamps to generate attribute key-value pairs. Finally, the type labels and attribute key-value pairs are combined in RDF triple format to generate a knowledge graph node set. It should be noted that the predefined mapping table is a mapping relationship established based on the semantic features of the device type text description and the corresponding rules of the knowledge graph ontology class, and includes regular expression matching patterns, target type names, and attribute predicate triples; Based on the knowledge graph node set, weighted edges with neighbor relationship type names are created, generating a neighbor relationship edge set; Furthermore, firstly, the installation location coordinates of each node in the knowledge graph node set are extracted and stored as a node location coordinate set according to the node ID order. Then, the node location coordinate set is input into the distance calculation formula, and spherical distance calculation is performed on every two different nodes in the node location coordinate set: first, the latitude and longitude of every two different nodes in the node location coordinate set are converted into radian values, and then substituted into the Haversine (semi-sine) formula to calculate and output the spherical distance between the two different nodes, generating a distance matrix containing the distances of all node pairs. Then, all off-diagonal elements in the distance matrix are traversed, and node pairs are filtered (for example: node pairs with a distance value ≤ 5km are determined based on the maximum effective communication distance of polar observation equipment and the attenuation of sound wave signals in the ice medium, in order to ensure reliable data transmission between relay nodes), and substituted into the weight formula to calculate the edge weight, creating a weighted edge with a neighbor relationship type name for each pair of nodes, and finally generating a neighbor relationship edge set. It should be noted that the proximity relationship type name is set based on the functional coupling and geographical proximity between the devices. For example, if two devices are observation devices of the same type, they are marked as "isomorphic proximity", and if they are complementary devices, they are marked as "heterogeneous complementary proximity". Retrieve the observation fields of the knowledge graph node set, match node pairs with the same observation text, create undirected edges with the measurement relation type name, and generate a measurement relation edge set; Furthermore, firstly, each node ID in the knowledge graph node set is traversed, and the text value of the observation field for each node ID is extracted to generate a node-observation text mapping table. Next, the node-observation text mapping table is aggregated into a list by using a hash function to generate an inverted index of observations. Then, the list of node IDs corresponding to each observation text in the inverted index of observations is read, and starting from the first element of the list, it is paired with all subsequent elements in turn to generate a set of node pairs that are unique and undirected, ensuring that each pair of nodes appears only once and the order is irrelevant, and a set of candidate node pairs is output. Finally, each node pair in the candidate node pair set is sorted by a unique node identifier to create an undirected edge with the name of the measurement relationship type, generating a set of measurement relationship edges. The numerical values of the standardized dataset are mapped to the corresponding ID nodes of the knowledge graph node set, and a dynamic knowledge graph is constructed by combining the neighbor relationship edge set and the measurement relationship edge set. Furthermore, firstly, the key-value pairs of node IDs and standardized values from the standardized dataset are read. The target node in the knowledge graph node set is precisely matched using the node ID string, and the value from the standardized dataset is overwritten into the "standardized observation value" attribute field of the target node, generating a knowledge graph node set with standardized attributes. Next, the nodes in the neighbor relationship edge set and the measurement relationship edge set are matched, retaining neighbor relationship edges and removing duplicate measurement relationship edges, generating a unified relationship edge set. Then, using the knowledge graph node set with standardized attributes as the vertex data source and the unified relationship edge set as the edge data source, the adjacency list structure is initialized: a hash table stores vertices (the key is the node ID, and the value is the attribute dictionary), and a linked list stores edges (neighbor edges are appended with weight values, and measurement edges are marked as undirected), finally outputting a dynamic knowledge graph.
[0027] S3. Based on dynamic knowledge graphs, a graph fusion feature matrix is generated by fusing features of adjacent nodes through graph neural networks. Read the node attribute field values from the dynamic knowledge graph and generate a node feature vector matrix; Furthermore, the hash table of the dynamic knowledge graph is traversed first, and the "standardized observation value" attribute field value of each vertex is extracted in lexicographical order by node ID to generate the original attribute value sequence. Then, the type validation (forced float64 type) and dimension padding are performed on each numerical array in the original attribute value sequence: zeros are padded at the end when the length is less than 128 dimensions, and the first 128 dimensions are truncated when the length is too long, and a 128-dimensional aligned attribute matrix is output. Then, the 128-dimensional vectors of each node in the 128-dimensional aligned attribute matrix are arranged in lexicographical order by node ID to generate the node feature vector matrix. Based on the set of adjacent edges and the set of measured edges, an adjacency matrix is created, and a summation operation is performed on the rows of the adjacency matrix to generate a node degree matrix; Furthermore, firstly, an N×N floating-point zero matrix is created based on the number of rows N of the node feature vector matrix. Then, each edge of the unified relation edge set generated by combining the neighbor relation edge set and the measurement relation edge set is traversed, and weight values are written into the corresponding positions of the zero matrix according to the start ID and end ID indices, generating an asymmetric adjacency relation matrix. Next, the asymmetric adjacency relation matrix is symmetricized: a deep copy of the asymmetric adjacency relation matrix (rows and columns are interchanged) is performed to generate a transpose matrix. The arithmetic mean of the elements at the same positions of the asymmetric adjacency relation matrix and the transpose matrix is taken, while the diagonal elements are kept in their original values, generating a symmetric adjacency relation matrix. Then, the symmetric adjacency relation matrix is accumulated row by row using floating-point summation to obtain the degree sequence. Finally, an N×N zero matrix is created, and the diagonal elements are written into the N×N zero matrix according to the degree sequence to generate a node degree matrix. Based on the node degree matrix and the adjacency matrix, a normalized adjacency matrix is generated by performing matrix concatenation multiplication through symmetric normalization. Furthermore, firstly, the diagonal elements of the node degree matrix are raised to the power of -0.5, while the off-diagonal elements remain at zero, generating a strictly diagonal node degree negative power matrix. Next, a row vector weighted propagation operation is performed on the node degree negative power matrix and the symmetric adjacency matrix: each row vector of the node degree negative power matrix is traversed and each column vector of the symmetric adjacency matrix is multiplied by a vector dot product to obtain the left-multiplied intermediate matrix. Finally, a column vector weighted diffusion operation is performed on the left-multiplied intermediate matrix and the node degree negative power matrix: each row vector of the left-multiplied intermediate matrix is traversed and each column vector of the node degree negative power matrix is multiplied by a vector dot product to generate a normalized adjacency matrix. It should be noted that symmetric normalization is a graph data processing method that generates an adjacency matrix with normalization properties by performing a cascaded multiplication operation on the adjacency relation matrix and the negative half power of the node degree matrix. The normalized adjacency matrix and the node feature vector matrix are subjected to a linear transformation operation through a three-layer graph convolution, and negative values are reduced to zero to generate a graph fusion feature matrix. Furthermore, firstly, the target node ID is located from the row / column index of the normalized adjacency matrix, and the node feature vector matrix is queried using the node ID as the key to extract the device output value as the key feature. The normalized adjacency matrix and the node feature vector matrix containing the device output value are then subjected to a first-layer convolution, performing neighborhood feature aggregation to generate a primary aggregated feature matrix. This primary aggregated feature matrix is then multiplied and accumulated with the pre-trained first-layer parameter set, and negative values are zeroed out to generate the first-layer activation feature matrix. Finally, the first-layer activation feature matrix is further processed with the normalized adjacency matrix... A second-layer convolution is performed to conduct second-order neighborhood aggregation, generating a secondary aggregated feature matrix. The secondary aggregated feature matrix is then multiplied and accumulated with the pre-trained second-layer parameter set, and negative values are zeroed out to output a second-layer activation feature matrix. Subsequently, the second-layer activation feature matrix is convolved with the normalized adjacency matrix to perform full-image feature fusion, generating a deep aggregated feature matrix. The deep aggregated feature matrix is then multiplied and accumulated with the pre-trained final-layer parameter set, and negative values are zeroed out to output a graph fusion feature matrix containing optimized features of device output values. It should be noted that the first-layer parameter set of the pre-training is the parameter tensor stored after convergence by optimizing the graph convolution loss function through gradient descent algorithm, using 128-dimensional node features and adjacency relationship data for iterative calculation; the second-layer parameter set of the pre-training is the parameter tensor generated after freezing it after 100,000 iterations based on the output features of the first layer and using the ADAM optimizer to learn high-order neighborhood relationships in cross-node collaboration; the final-layer parameter set of the pre-training is the parameter tensor dynamically optimized in the digital twin environment through a reward mechanism and distributed through quantum encryption.
[0028] S4. Define the action space based on the graph fusion feature matrix, calculate the real-time reward value, dynamically optimize the strategy parameters, and generate scheduling instructions; Extract the device output values from the graph fusion feature matrix, construct a state space matrix, and bind action names; Furthermore, firstly, a fixed dimension index number for the device output value is located from the 128-dimensional features of the graph fusion feature matrix using a pre-defined dimension mapping table, generating a device output value dimension index number. Next, the device output value dimension index number is used to traverse all node rows of the graph fusion feature matrix, extracting the device output value at the specified index position and recording the node ID, thus generating a node-output value mapping table. Then, the node-output value mapping table is sorted lexicographically by node ID, and a millisecond-level UTC timestamp and a SHA-256 hash value of the location coordinates are added to each node to construct an N×3 state space matrix. Finally, based on the device type mapping table, the device type is matched according to the node ID, an action name is bound to each node row of the state space matrix, and an action name label is added to the last column, generating a state space matrix with action names. It should be noted that the pre-defined dimension mapping table is a table that corresponds device features to the input dimensions of the graph neural network, including device type ID, feature name, corresponding dimension range and feature weight coefficient; the device type mapping table is a lookup table that stores the correspondence between node ID and device type and action permissions, including node ID, device type and list of allowed actions; Based on the equipment output value and the preset load demand value, calculate the power supply balance rate, and combine the equipment safety margin benchmark value and the line loss rate benchmark value to calculate the real-time bonus value; The expression for calculating the power supply balance rate is: ; in, It is the power supply balance rate, representing the first... Power supply status of each device; It is the first Load demand values for each device; It is the first The output value of each device; The expression for calculating the real-time reward value is: ; in, It is the safety margin benchmark value; This is the baseline value for line loss rate; It is a real-time reward value; Furthermore, firstly, the device output values are extracted from the "Output Value" column of the state space matrix with action names to generate a device output value sequence. Simultaneously, the preset load demand value at the same index position is accessed to generate a load demand value sequence. Next, a protected division operation is performed between the device output value sequence and the load demand value sequence (for example: if the load demand value ≤ 0, output 1.0; if the load demand value > 0, calculate device output value ÷ demand value) to obtain the power supply balance rate and generate a power supply balance rate sequence. Then, based on the node ID of the state space matrix, the static benchmark value of the device safety margin is read from the device safety configuration library, and combined with real-time ice thickness sensor data, the dynamic safety margin benchmark value is calculated using a linear compensation formula to generate a safety margin sequence. The theoretical parameters of the line loss rate are read from the power grid topology database, and based on real-time wind speed and temperature data, the dynamic line loss rate benchmark value is calculated using an environmental compensation formula to generate a line loss rate sequence. Finally, based on the power supply balance rate sequence, the safety margin sequence, and the line loss rate sequence, a weighted calculation method is used to calculate element-by-element using a fixed coefficient formula to obtain the real-time reward value. It should be noted that the preset load demand value is generated based on historical load curves for the same period and real-time weather forecast data through a time series prediction algorithm; the equipment safety configuration library is a dedicated database that stores the boundary parameters for safe operation of polar observation equipment. Safety parameters are obtained through type tests before the equipment leaves the factory and are then verified and solidified into radiation-resistant flash memory chips; the power grid topology database is a dynamic graph database that records the connection relationships of equipment and the electrical characteristics of lines in the polar observation network. The initial data is obtained through on-site surveys by the scientific expedition team, and the node topology is subsequently updated in real time through ice layer displacement sensors. Based on the real-time reward value and the current state encoding of the state space matrix, the Bellman equation iteration is performed to obtain the updated policy parameter table; Furthermore, the process first iterates through each row of the state space matrix, concatenates them into a string according to a fixed format, and performs 64 rounds of compression using an MD5 hasher to generate a 128-bit state code sequence. Then, using this 128-bit state code sequence as the key, it queries the policy parameter table to obtain the current action value vector. It then retrieves the action value vector of the next state code from the original policy parameter table and takes the maximum value as the future maximum action value. Next, iterative calculations are performed using the Bellman equation: based on the real-time reward value and the future maximum action value, a weighted summation method is used to obtain the action value target value; based on the action value target value and the current action value, a difference calculation method is used to obtain the time difference scalar; based on the current action value and the time difference scalar, a weighted accumulation calculation method is used to obtain the updated action value; finally, the updated action value is written to the incremental update log in JSON format, a memory copy of the policy parameter table is created, and the corresponding key-value pairs are updated one by one according to the log, outputting the updated policy parameter table. It should be noted that the original strategy parameter table is a global action value mapping table generated by merging the action value vector sets uploaded by each node using a security aggregation algorithm. Based on the updated policy parameter table, the optimal action name is selected for each state code in the state space matrix using an ε-greedy policy to generate scheduling instructions. Furthermore, the 128-bit state code sequence is used as the key to retrieve the updated policy parameter table, obtain the corresponding current state action value matrix, and generate random numbers for the action value vector of each state code in the corresponding current state action value matrix. At the same time, it is compared with the preset action selection threshold ε (exemplary: ε=0.1, verified by historical data from the polar observation network, ε=0.1 has the lowest failure rate and the fastest convergence speed): when ε>0.1, the maximum value action is selected, and when ε≤0.1, random actions are uniformly sampled, and the selected action name is output. The selected action name is combined with the device ID field of the state space matrix into the format "device ID: action name" to generate the scheduling instruction.
[0029] S5. The scheduling instructions are encoded into polarization-phase composite quantum states using a dual-field protocol to generate a quantum instruction stream. Polarization decoding and decoy state detection are then performed to restore the executable control commands. The scheduling instructions are converted into a binary bit stream, quantum state encoding is performed using a dual-field protocol, a polarization-phase composite quantum state pulse sequence is output, and decoy state pulses are added to generate a quantum instruction stream; Furthermore, the scheduling instructions are first converted into an 8-bit binary byte block sequence, and horizontally polarized or vertically polarized pulses are generated according to bit values 0 / 1 respectively, outputting a basic polarized state pulse sequence. Then, based on the basic polarized state pulse sequence, a quantum random number controller generates true random numbers to drive phase offset loading, randomly selecting 0 or π phase offsets to generate a polarization-phase composite quantum state pulse sequence. Then, a fixed proportion (e.g., 30%) of the pulse positions in the polarization-phase composite quantum state pulse sequence are randomly selected and replaced with decoy state pulses of random polarization and random phase to generate a hybrid quantum state pulse sequence. Finally, the hybrid quantum state pulse sequence is grouped into 16 pulses / frames, and a "10" frame header pulse is added to generate a quantum instruction stream. It should be noted that the dual-field protocol is a quantum key distribution protocol that achieves secure information transmission at the single-photon level by simultaneously utilizing the polarization and phase degrees of freedom of the light field for encoding. The decoy state pulses of the quantum instruction stream are separated, and the signal state pulses are polarized and split to obtain two optical intensity signals; Furthermore, the quantum instruction stream is first analyzed to identify the frame header "10", locate the 16-pulse frame boundary, and extract the decoy state position markers, outputting frame structure analysis data. Then, based on the frame structure analysis data, a pulse separation operation is performed to remove the decoy state pulses, generating a decoy state pulse set and simultaneously outputting a signal state pulse set. Next, the signal state pulse set is calibrated by rotating a half-wave plate to calibrate the polarization reference axis. The calibrated signal state is then split by an α-BBO (low-temperature phase barium borate) crystal: the horizontally polarized state is refracted to the 0° channel, and the vertically polarized state is refracted to the 90° channel, obtaining two light intensity signals. The phase information of the two optical intensity signals is compared, and the binary bit stream is restored to obtain the decoded binary sequence; Furthermore, a phase-stable local reference light is first generated by a 1550nm laser source, and a reference phase signal is output. Then, based on the two light intensity signals and the reference phase signal, phase difference detection is performed through zero-difference interferometry, and a high / low level signal sequence is output according to the interference light intensity. Then, the high / low level signal sequence is restored to the original bit sequence through a dual-field protocol. Finally, the original bit sequence is reassembled into a byte stream through an 8-bit grouping operation, with each consecutive 8 bits forming a byte unit, to generate a decoded binary sequence. Based on the decoded binary sequence and the preset original decoy state sequence, the bit error rate of the decoy state pulse is calculated and verified, and an executable control command is output. Furthermore, firstly, the decoy state position index field is read from the preset original decoy state sequence to generate a decoy state position index set. Then, using the decoy state position index set as the index key, the corresponding bit values are extracted from the decoded binary sequence to generate a received decoy state bit sequence that is in the same order as the index set. Next, the received decoy state bit sequence is XORed bit by bit with the standard value field in the original decoy state sequence. The number of bits with a result of 1 (based on the classic error detection theory of binary communication, ensuring that each bit error is counted independently and without missed detection) is used to generate the bit error count. Finally, the ratio of the bit error count to the total length of the decoy state position index set is calculated to obtain the bit error rate. The floating-point value of the bit error rate is compared with the security bit error rate threshold (exemplary: 0.02, determined by combining the theoretical security boundary of quantum key distribution and the actual channel noise characteristics). If the floating-point value of the bit error rate is > 0.02, the comparison with the security bit error rate threshold is terminated, and a "channel abnormality" flag is output. If the floating-point value of the bit error rate is ≤ 0.02, the difference between the total position range and the decoy state position index set is calculated to obtain the signal state position set. Based on the signal state position set, the corresponding bits are extracted from the decoded binary sequence to generate the signal state binary sequence. Finally, the signal state binary sequence is decoded, and 7 consecutive bits are divided into a group and converted into characters. The characters are then concatenated to generate an executable control command. It should be noted that the preset original decoy state sequence is a record table of decoy state positions and standard values randomly generated during the quantum state encoding stage. The random combination of polarization state and phase is generated in real time by a quantum random number generator. The total position range refers to the set of temporal positions of all pulses in a single frame of the quantum instruction stream, and its value is equal to the position index 0-15 corresponding to the frame length (fixed 16 pulses / frame).
[0030] S6. Based on executable control commands, perform output adjustment operations on polar observation equipment, obtain actual response parameters of the equipment, transmit the actual response parameters of the equipment back to the digital twin entity, calculate the scheduling deviation rate, optimize the quantum instruction stream based on the scheduling deviation rate, and generate a polar observation equipment analytical integration report. Based on executable control commands, pulse width modulation signals are sent to polar observation equipment to perform power output adjustment operations; Furthermore, the device ID field and action parameters in the executable control command are first parsed, and a square wave signal is generated by querying the preset pulse width mapping table according to the device output value. Then, the square wave signal is input into the cold-resistant transmission interface, converted into differential voltage fluctuations, and transmitted to the target device through a double-shielded cable. The power driver of the target device receives the differential signal, restores it to a 3.3V square wave through a level conversion circuit (matching the CMOS logic level specification of the power driver, and also compatible with the IGBT gate drive threshold voltage). The H-bridge (full-bridge circuit) alternately turns on the insulated gate bipolar transistors, and controls the on and off time of the power semiconductor devices in the energy storage unit at a 65% duty cycle (determined by linear interpolation through the pulse width mapping table, corresponding to the optimized control parameters for a 12MW output value), thereby increasing the output. It should be noted that the preset pulse width mapping table is a lookup table that stores the relationship between the power output value of the storage device and the PWM (pulse width modulation) duty cycle. The power-duty cycle linearity of the polar device at -55℃ is calibrated through experiments. During the output adjustment operation, a cold-resistant sensor is used to acquire the equipment operating status in real time, generate a set of actual response parameters of the equipment, write them into the digital twin entity, and output the updated digital twin entity. Furthermore, firstly, the electrical signals of the power adjustment parts of the equipment are collected in real time through a cold-resistant Hall sensor array, and sampling and range calibration are performed to generate a digital sequence of power values with millisecond-level timestamps. Then, the digital sequence of power values is combined with the equipment ID, ambient temperature and wind speed fields and encapsulated into a set of actual response parameters of the equipment in JSON format. Then, the digital twin registry is queried with the equipment ID as the key to obtain the access handle of the target entity, and the entity is locked through the access handle of the target entity. The set of actual response parameters of the equipment is appended to the historical status queue and the latest status field is updated to generate the updated digital twin entity. It should be noted that the digital twin registry is a key-value store table generated through pre-registration during the device deployment phase, containing device ID index, entity memory mapping address, and version control metadata; Read the predicted output value and the actual output value from the updated digital twin entity, and calculate the scheduling deviation rate; The expression for calculating the scheduling deviation rate is: ; in, It is the scheduling deviation rate; This is the actual output value; It is the predicted output value; Furthermore, firstly, the starting address of the historical state queue of the updated digital twin entity is located, the latest record at index 0 is extracted, and the predicted output value field and the actual output value field are read to generate a predicted-actual output value pair; then, the difference operation is performed based on the predicted-actual output value pair to output the absolute deviation value, and the predicted output value is detected simultaneously: if the predicted output value ≤ 0, the output is switched to constant 0; if the predicted output value > 0, the absolute deviation value and the predicted value are divided by a floating-point operation to output the scheduling deviation rate. It should be noted that the predicted output value refers to the expected output value of the equipment generated by the dynamic knowledge graph reasoning engine of the digital twin based on the graph fusion feature matrix. Its essence is the theoretical output prediction of the polar observation equipment within the next 5-second time window. The dual-field protocol parameters of the quantum instruction stream are adjusted according to the scheduling deviation rate to generate an optimized quantum instruction stream; Furthermore, a floating-point comparison is first performed between the scheduling deviation rate and the safety threshold (exemplary: 0.05): if the scheduling deviation rate > 0.05, the optimization trigger flag is output as true (correct); if the scheduling deviation rate ≤ 0.05, the optimization trigger flag is output as false (incorrect). When the optimization trigger flag is true, the polarization basis rotation angle increment is calculated to generate a new polarization basis angle. Simultaneously, the phase offset scaling factor is calculated, and the phase compensation parameter is output. At the same time, the decoy state density increment is calculated, and a new decoy state density value is output. Then, the new polarization basis angle, phase compensation parameter, and new decoy state density value are integrated to optimize the dual-field protocol parameter set. Finally, the optimized dual-field protocol parameter set is input into the quantum instruction encoder to re-execute polarization modulation, phase loading, and decoy state injection on the original scheduling instruction to generate an optimized quantum instruction stream. The expression for calculating the new polarization basis angle is: ; in, It is the new polarization basis angle; It is the scheduling deviation rate; It is the original polarization basis angle; The expression for calculating the phase offset scaling factor is: ; in, It is the phase offset scaling factor; The expression for calculating the new decoy state density is: ; in, It is a new deceptive state density; It is the original decoy state density; It should be noted that the security threshold is determined based on the theoretical security boundary of quantum key distribution (proof security with a bit error rate ≤11%) and the measured baseline noise in the polar environment (average bit error rate of 0.018 at -55℃), combined with 100,000 attack test data, to ensure that while retaining a security margin, it covers equipment noise and environmental disturbances. By integrating the actual response parameter set of the equipment, the scheduling deviation rate, and the optimized quantum command stream, an analytical integration report of polar observation equipment is generated. Furthermore, firstly, using the timestamps of the actual response parameter set of the equipment as a benchmark, data with the same timestamps are matched in the scheduling deviation rate record and the optimized quantum instruction stream to generate time-aligned data triples. Next, the quantum instruction stream segments in the data triples are subjected to polarization beam splitting to separate the horizontal / vertical polarization states, and a zero-difference interferometry phase restoration operation is performed simultaneously to extract phase information. Finally, the separated polarization states and the extracted phase information are recombined and output as a binary byte stream. Then, the binary byte stream is decoded and converted into readable scheduling instruction text. The actual output value of the actual response parameter set of the equipment, the scheduling deviation rate, and the decoded scheduling instruction text are combined and encapsulated into an original report dataset, and encrypted with an RSA-2048 private key to generate a digital signature. Finally, the original report dataset, digital signature, and data triples are assembled to generate a polar observation equipment analysis and integration report.
[0031] This embodiment also provides a computer device applicable to the heterogeneous analysis and integration method of polar observation network data based on edge computing, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the heterogeneous analysis and integration method of polar observation network data based on edge computing as proposed in the above embodiment.
[0032] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0033] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the heterogeneous analytical integration method for polar observation network data based on edge computing, as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0034] In summary, this invention achieves spatiotemporal correlation modeling of multi-source heterogeneous data in polar regions by constructing a dynamic knowledge graph, enabling the knowledge graph to dynamically respond to environmental changes. Its beneficial effects lie in improving the completeness of state perception and the spatiotemporal consistency of decision-making basis. Through quantum instruction flow closed-loop optimization, a closed-loop link of quantum instruction generation → execution → feedback → optimization is established. Its beneficial effects lie in significantly improving the anti-interference capability and control accuracy of instruction transmission in extreme environments through adaptive adjustment of quantum state parameters, while ensuring information security throughout the entire process.
[0035] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A heterogeneous analytical integration method for polar observation network data based on edge computing, characterized in that: include, By connecting to field equipment through near-field interaction via mobile terminals, valid parameters and basic equipment data are collected, corresponding digital twin entities are created, and standardized datasets are obtained by parsing them through a pre-set data parsing instruction set. By using digital twin entities as nodes and combining them with standardized datasets, a dynamic knowledge graph can be constructed. Based on dynamic knowledge graphs, a graph fusion feature matrix is generated by fusing features of adjacent nodes through graph neural networks. The action space is defined based on the graph fusion feature matrix, the real-time reward value is calculated, the strategy parameters are dynamically optimized, and the scheduling instructions are generated. The scheduling instructions are encoded into polarization-phase composite quantum states using a dual-field protocol to generate a quantum instruction stream. Polarization decoding and decoy state detection are then performed to restore the executable control commands. The power output of the polar observation equipment is adjusted based on executable control commands, the actual response parameters of the equipment are obtained, the actual response parameters of the equipment are sent back to the digital twin entity, the scheduling deviation rate is calculated, the quantum instruction stream is optimized based on the scheduling deviation rate, and an analytical integration report of the polar observation equipment is generated.
2. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The process involves connecting to field devices via a mobile terminal's near-field interaction, collecting valid parameters and basic device data, creating corresponding digital twin entities, and then parsing the data using a pre-set data parsing instruction set to obtain a standardized dataset. The steps are as follows: The effective parameters include equipment output value, location coordinates, environmental parameters, and equipment status code; The basic equipment data includes a text description of the equipment type, numerical coordinates of the installation location, and text of maintenance history records. Bind valid parameters to basic device data to create a digital twin entity; The valid parameters are parsed field by field using a preset data parsing instruction set to generate a numerical data matrix. The numerical data matrix is subjected to Box-Cox transformation and Z-score standardization to obtain a standardized dataset.
3. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The steps for constructing a dynamic knowledge graph by using digital twin entities as nodes and combining them with a standardized dataset are as follows. Each digital twin entity is converted into a knowledge graph node, and the entity type name is labeled to generate a set of knowledge graph nodes; Based on the knowledge graph node set, weighted edges with neighbor relationship type names are created, generating a neighbor relationship edge set; Retrieve the observation fields of the knowledge graph node set, match node pairs with the same observation text, create undirected edges with the measurement relation type name, and generate a measurement relation edge set; The numerical values of the standardized dataset are mapped to the corresponding ID nodes of the knowledge graph node set, and a dynamic knowledge graph is constructed by combining the neighbor relationship edge set and the measurement relationship edge set.
4. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The steps for fusing features from adjacent nodes using a graph neural network to generate a graph fusion feature matrix are as follows: Read the node attribute field values from the dynamic knowledge graph and generate a node feature vector matrix; Based on the set of adjacent edges and the set of measured edges, an adjacency matrix is created, and a summation operation is performed on the rows of the adjacency matrix to generate a node degree matrix; Based on the node degree matrix and the adjacency matrix, a normalized adjacency matrix is generated by performing matrix concatenation multiplication through symmetric normalization. The normalized adjacency matrix and the node feature vector matrix are subjected to a linear transformation operation through a three-layer graph convolution, and negative values are reduced to zero to generate a graph fusion feature matrix.
5. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The steps for defining the action space based on the graph fusion feature matrix, calculating the real-time reward value, dynamically optimizing strategy parameters, and generating scheduling instructions are as follows: Extract the device output values from the graph fusion feature matrix, construct a state space matrix, and bind action names; Based on the equipment output value and the preset load demand value, calculate the power supply balance rate, and combine the equipment safety margin benchmark value and the line loss rate benchmark value to calculate the real-time bonus value; Based on the real-time reward value and the current state encoding of the state space matrix, the Bellman equation iteration is performed to obtain the updated policy parameter table; Based on the updated policy parameter table, the optimal action name is selected for each state code in the state space matrix using an ε-greedy policy to generate scheduling instructions.
6. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The process involves encoding scheduling instructions into polarization-phase composite quantum states using a dual-field protocol to generate a quantum instruction stream, followed by polarization decoding and decoy state detection to reconstruct executable control commands. The steps are as follows: The scheduling instructions are converted into a binary bit stream, quantum state encoding is performed using a dual-field protocol, a polarization-phase composite quantum state pulse sequence is output, and decoy state pulses are added to generate a quantum instruction stream; The decoy state pulses of the quantum instruction stream are separated, and the signal state pulses are polarized and split to obtain two optical intensity signals; The phase information of the two optical intensity signals is compared, and the binary bit stream is restored to obtain the decoded binary sequence; Based on the decoded binary sequence and the preset original decoy state sequence, the bit error rate of the decoy state pulse is calculated and verified, and an executable control command is output.
7. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The steps for adjusting the output of the polar observation equipment, obtaining the actual response parameters of the equipment, transmitting these parameters back to the digital twin entity, and calculating the scheduling deviation rate are as follows: Based on executable control commands, pulse width modulation signals are sent to polar observation equipment to perform power output adjustment operations; During the output adjustment operation, the equipment operating status is acquired in real time, a set of actual response parameters of the equipment is generated, written into the digital twin entity, and the updated digital twin entity is output. Read the predicted output value and the actual output value from the updated digital twin entity, and calculate the scheduling deviation rate.
8. The method for heterogeneous analytical integration of polar observation network data based on edge computing as described in claim 1, characterized in that: The steps for optimizing the quantum instruction stream and generating the polar observation equipment analysis and integration report are as follows. The dual-field protocol parameters of the quantum instruction stream are adjusted according to the scheduling deviation rate to generate an optimized quantum instruction stream; By integrating the actual response parameters of the equipment, the scheduling deviation rate, and the optimized quantum command stream, an analytical integration report of polar observation equipment is generated.
9. 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 heterogeneous analytical integration method for polar observation network data based on edge computing as described in any one of claims 1 to 8.
10. 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 heterogeneous analytical integration method for polar observation network data based on edge computing as described in any one of claims 1 to 8.
