Natural gas pipeline network intelligent operation scheduling method and system based on user priority

Abstract

The invention provides a natural gas pipeline network intelligent operation scheduling method and system based on user priorities, and relates to the technical field of natural gas, and the method comprises the steps: collecting edge calculation node data, constructing a sub-network, distributing calculation weights, executing distributed training to generate a calculation task distribution scheme, writing pipeline network state parameters into a block chain, and verifying the pipeline network state parameters. And generating and executing a pipe network optimization regulation and control instruction according to the parameters. According to the method, efficient distribution of pipe network resources, load balancing of calculation tasks and safe and reliable operation of the pipe network are realized, and the operation efficiency of the natural gas pipe network and the user service quality are improved.

Description

Technical Field

[0001] The present invention relates to natural gas technology, and in particular to a method and system for intelligent operation and scheduling of a natural gas pipeline network based on user priorities. Background Art

[0002] With the widespread adoption of natural gas as a clean energy source, the scale of natural gas pipeline networks continues to expand, and their operation and management are becoming increasingly complex. Intelligent operation and scheduling of natural gas pipeline networks is a key technology for ensuring energy supply security and improving network efficiency. Traditional natural gas pipeline scheduling relies primarily on manual experience and simple calculation models, making it difficult to adapt to the operational demands of modern, complex pipeline networks. In recent years, the development of emerging technologies such as the Internet of Things, edge computing, and blockchain has provided new technical means and solutions for intelligent scheduling of natural gas pipeline networks.

[0003] In existing technologies, natural gas pipeline dispatch systems typically employ a centralized computing architecture, transmitting collected pipeline data to a central server for processing and analysis before issuing control instructions. This approach presents several practical problems: First, centralized computing architectures face computing resource bottlenecks. As the pipeline network expands and data volume surges, the central server's computational burden becomes excessive, making it difficult to respond to network status changes in real time, resulting in inefficient dispatching. Second, traditional dispatching systems lack effective user priority management mechanisms, making it impossible to formulate differentiated gas supply strategies based on the importance of different users. Consequently, they are unable to guarantee the gas needs of key users during periods of tight gas supply. Finally, existing systems lack data security mechanisms, making pipeline operation data susceptible to tampering or loss. This makes it difficult to guarantee system reliability and security, and fails to meet the safe operation requirements of critical infrastructure. Summary of the Invention

[0004] The embodiments of the present invention provide a method and system for intelligent operation and scheduling of a natural gas pipeline network based on user priorities, which can solve the problems in the prior art.

[0005] A first aspect of an embodiment of the present invention provides a method for intelligent operation and scheduling of a natural gas pipeline network based on user priorities, comprising:

[0006] Collect pressure data, flow data, temperature data, and user gas usage data from multiple edge computing nodes in the natural gas pipeline network;

[0007] Dividing the edge computing nodes into a plurality of computing subnetworks according to the geographic location information and the network topology of the edge computing nodes; setting a computing weight coefficient for each node based on the computing power and communication delay of the nodes in each computing subnetwork; performing distributed training on the edge computing nodes in each subnetwork to obtain a computing task allocation coefficient for each node; and generating a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient;

[0008] According to the computing task allocation scheme, the pipe network optimization computing tasks are allocated to each edge computing node, and parallel computing is performed to obtain the pipe network status parameters;

[0009] The computing task allocation plan is written into the blockchain to construct a computing task transaction block; the pipeline network status parameters are signed and encrypted based on the timestamp, and the encrypted pipeline network status parameters are verified for consistency through a voting mechanism;

[0010] The pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline network node are calculated according to the pipeline network state parameters, and pipeline network optimization control instructions are generated. The pipeline network optimization control instructions are executed to realize intelligent control of the natural gas pipeline network.

[0011] Based on the computing power and communication delay of each node in the computing subnetwork, the computing weight coefficient of each node is set; the edge computing nodes in each subnetwork are distributedly trained to obtain the computing task allocation coefficient of each node, including:

[0012] Obtain computing resource usage data, communication performance data, and operating status data of edge computing nodes;

[0013] Based on the computing resource usage data, a basic weight value of the edge computing node is calculated according to a preset resource weight coefficient; based on the communication performance data, a communication weight value of the edge computing node is calculated using an exponential decay function; based on the operating status data, a health value of the edge computing node is calculated by time series feature analysis; the basic weight value, the communication weight value, and the health value are weighted and combined to obtain a comprehensive weight value;

[0014] Calculating a training task allocation ratio for the edge computing node based on the comprehensive weight value; allocating training data to the edge computing node based on the training task allocation ratio; performing model training on the edge computing node using the training data to obtain a local training model; and performing weighted aggregation on the local training model based on the comprehensive weight value and the health value to obtain a global training model;

[0015] The dynamic load balancing factor of the edge computing node is calculated based on the global training model, and the dynamic load balancing factor is proportional to the comprehensive weight value and the health value; the computing task allocation coefficient of the edge computing node is generated according to the training task allocation ratio, the dynamic load balancing factor and the change rate of the health value.

[0016] Allocating training data to the edge computing node based on the training task allocation ratio; and performing model training on the edge computing node using the training data to obtain a local training model includes:

[0017] Obtaining feature vectors and sample weights of a training sample set, calculating a global feature representation based on the feature vectors and the sample weights, and using the global feature representation to train a teacher model to obtain knowledge representation data of the teacher model;

[0018] Performing a preliminary division of the training data based on the training task allocation ratio, calculating a correlation coefficient between the global feature representation and the local features of each edge computing node, and optimizing the preliminary division result of the training data based on the correlation coefficient and the knowledge representation data to obtain an optimized training data allocation scheme;

[0019] allocating the training data to each edge computing node according to the optimized training data allocation scheme, calculating an adaptive distillation strength coefficient of each edge computing node based on the correlation coefficient, constructing a local training loss function using the adaptive distillation strength coefficient, and performing model training on each edge computing node based on the local training loss function and the knowledge representation data;

[0020] During the model training process, the training convergence data of each edge computing node is obtained, the training temperature parameter is dynamically adjusted according to the training convergence data, the knowledge migration degree between the teacher model and each edge computing node model is calculated, and the optimized training data allocation scheme is dynamically adjusted based on the knowledge migration degree to obtain a local training model of each edge computing node.

[0021] According to the computing task allocation scheme, the network optimization computing tasks are allocated to each edge computing node, and parallel computing is performed to obtain the network status parameters including:

[0022] Analyze the data dependency between related computing tasks based on the computing task allocation plan, collect the real-time communication delay between edge computing nodes that execute related computing tasks, construct a node communication cost matrix based on the data dependency and the real-time communication delay, and optimize and adjust the computing task allocation plan using the node communication cost matrix to generate a final computing task allocation plan;

[0023] The final computing task allocation plan is sent to each edge computing node, and each edge computing node performs parallel calculation of the pipeline network status parameters according to the computing task type and computing resource allocation ratio specified in the final computing task allocation plan, and integrates the parallel calculation results of each edge computing node to obtain complete pipeline network status parameters.

[0024] The computing task allocation plan is written into the blockchain to construct a computing task transaction block; the pipe network status parameters are signed and encrypted based on the timestamp, and the encrypted pipe network status parameters are verified for consistency through a voting mechanism, including:

[0025] Generate a task information sequence containing the task identifier, allocation scheme, timestamp, and previous block hash, perform a hash concatenation operation on the task information sequence with the block metadata and node signature to generate a computing task transaction block, use the public key to verify the link relationship between adjacent computing task transaction blocks, and generate a blockchain integrity verification result;

[0026] Confirming the validity of the computing task transaction block based on the blockchain integrity verification result, associating the state parameter with the timestamp and encrypting it using a symmetric key to obtain an encrypted state parameter, performing a hash calculation on the encrypted state parameter and generating multi-signature data using the private keys of multiple nodes, dynamically adjusting the encryption key length using an exponential decay method based on the communication time interval between nodes and the system security risk, and updating the encryption strength of the encrypted state parameter;

[0027] Obtain the historical reputation value and real-time computing power index of the participating verification nodes, calculate the node voting weight based on the historical reputation value and the real-time computing power index, use the node voting weight and the multi-signature data to perform consistency voting verification on the encryption state parameters, dynamically adjust the consensus threshold of the consistency verification based on the number of faulty nodes in the system and historical consensus data, and generate a verification result.

[0028] The task information sequence is hashed together with the block metadata and node signature to generate a computing task transaction block. The public key is used to verify the link relationship between adjacent computing task transaction blocks. The blockchain integrity verification result generated includes:

[0029] Extracting feature data including task identifier, block metadata, and node signature from the task information sequence, calculating computational complexity and security strength of the feature data, and generating a feature fitness evaluation value;

[0030] Selecting a basic hash function based on the characteristic fitness evaluation value, dynamically optimizing the basic hash function using an exponentially decaying mutation step size, and introducing a random mutation factor to construct an evolutionary hash algorithm; calculating the anti-attack capability and group diversity index of the blockchain node, and generating group immunity strength based on the anti-attack capability and the group diversity index;

[0031] Performing a hash join operation on the task information sequence, the block metadata, and the node signature using the evolutionary hash algorithm, adjusting hash join operation parameters according to the group immunity strength, and generating a computing task transaction block;

[0032] A block verification strategy is constructed based on the evolutionary hash algorithm and the group immunity strength. According to the block verification strategy, the link relationship between adjacent computing task transaction blocks is verified using the public key to generate a blockchain integrity verification result.

[0033] Calculating the pressure adjustment parameters, flow adjustment parameters, and valve opening parameters of each pipeline network node according to the pipeline network state parameters, and generating pipeline network optimization control instructions include:

[0034] Characteristically encoding the pressure parameters, flow parameters, and valve parameters of the pipeline network nodes to generate a node state vector, and constructing a multi-mode intelligent agent based on the node state vector, including a path optimization function, a flow balancing function, and a collaborative optimization function;

[0035] Extracting pressure gradient and pheromone intensity from the multimodal agent to calculate path optimization parameters, obtaining node flow deviation values ​​based on the path optimization parameters and generating flow balance parameters in combination with group behavior characteristics, and combining the flow balance parameters with information propagation function and exploration function to generate collaborative decision parameters;

[0036] The pressure regulation parameter, the flow regulation parameter and the valve opening parameter are calculated respectively according to the path optimization parameter, the flow balancing parameter and the collaborative decision parameter, and the pressure regulation parameter, the flow regulation parameter and the valve opening parameter are combined to generate a pipeline network optimization control instruction.

[0037] A second aspect of an embodiment of the present invention provides a natural gas pipeline network intelligent operation and scheduling system based on user priority, including:

[0038] The first unit is used to collect pressure data, flow data, temperature data and user gas usage data from multiple edge computing nodes in the natural gas pipeline network;

[0039] The second unit is configured to divide the edge computing nodes into a plurality of computing subnetworks according to the geographic location information and the network topology of the edge computing nodes; set a computing weight coefficient for each node based on the computing power and communication delay of the nodes in each computing subnetwork; perform distributed training on the edge computing nodes in each subnetwork to obtain a computing task allocation coefficient for each node; and generate a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient;

[0040] The third unit is used to allocate the pipe network optimization computing tasks to each edge computing node according to the computing task allocation scheme, and perform parallel computing to obtain the pipe network status parameters;

[0041] The fourth unit is configured to write the computing task allocation plan into the blockchain and construct a computing task transaction block; sign and encrypt the pipeline network status parameters based on the timestamp, and verify the consistency of the encrypted pipeline network status parameters through a voting mechanism;

[0042] The fifth unit is used to calculate the pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline node based on the pipeline state parameters, generate pipeline optimization control instructions, execute the pipeline optimization control instructions, and realize intelligent control of the natural gas pipeline network.

[0043] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0044] processor;

[0045] a memory for storing processor-executable instructions;

[0046] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0047] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0048] The beneficial effects of this application are as follows:

[0049] The user priority-based intelligent operation and scheduling method for natural gas pipeline networks provided by the present invention collects data from multiple edge computing nodes and performs distributed computing, thereby achieving efficient utilization of computing resources and improving the computing efficiency and response speed of pipeline network operation and scheduling.

[0050] This invention writes the computing task allocation plan into the blockchain and performs consistency verification through a voting mechanism, ensuring the security and reliability of the data, effectively preventing data tampering and network attacks, and improving the stability and security of pipeline network operation.

[0051] The present invention automatically calculates adjustment parameters and generates optimized control instructions based on pipeline network status parameters, realizing intelligent control of the natural gas pipeline network, reducing manual intervention, lowering operation and maintenance costs, while improving the reliability and gas utilization efficiency of pipeline network operation and ensuring the gas supply needs of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of a flow chart of a method for intelligent operation and scheduling of a natural gas pipeline network based on user priority according to an embodiment of the present invention;

[0053] Figure 2 This is a flow chart of the distributed training and task allocation system for edge computing nodes according to an embodiment of the present invention;

[0054] Figure 3 Schematic diagram of the relationship analysis between the adaptive distillation intensity coefficient and the correlation coefficient according to an embodiment of the present invention;

[0055] Figure 4 This is a flowchart of blockchain consistency verification based on multi-signatures in an embodiment of the present invention;

[0056] Figure 5 This is a bar chart comparing and analyzing the efficiency of the pipeline network optimization and control method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0058] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0059] Figure 1 FIG. 1 is a flow chart of a method for intelligent operation and scheduling of a natural gas pipeline network based on user priority according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0060] Collect pressure data, flow data, temperature data, and user gas usage data from multiple edge computing nodes in the natural gas pipeline network;

[0061] Dividing the edge computing nodes into a plurality of computing subnetworks according to the geographic location information and the network topology of the edge computing nodes; setting a computing weight coefficient for each node based on the computing power and communication delay of the nodes in each computing subnetwork; performing distributed training on the edge computing nodes in each subnetwork to obtain a computing task allocation coefficient for each node; and generating a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient;

[0062] According to the computing task allocation scheme, the pipe network optimization computing tasks are allocated to each edge computing node, and parallel computing is performed to obtain the pipe network status parameters;

[0063] The computing task allocation plan is written into the blockchain to construct a computing task transaction block; the pipeline network status parameters are signed and encrypted based on the timestamp, and the encrypted pipeline network status parameters are verified for consistency through a voting mechanism;

[0064] The pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline network node are calculated according to the pipeline network state parameters, and pipeline network optimization control instructions are generated. The pipeline network optimization control instructions are executed to realize intelligent control of the natural gas pipeline network.

[0065] In an optional embodiment, based on the computing power and communication delay of each node in the computing subnetwork, a computing weight coefficient of each node is set; and distributed training is performed on the edge computing nodes in each subnetwork to obtain the computing task allocation coefficient of each node, including:

[0066] Obtain computing resource usage data, communication performance data, and operating status data of edge computing nodes;

[0067] Based on the computing resource usage data, a basic weight value of the edge computing node is calculated according to a preset resource weight coefficient; based on the communication performance data, a communication weight value of the edge computing node is calculated using an exponential decay function; based on the operating status data, a health value of the edge computing node is calculated by time series feature analysis; the basic weight value, the communication weight value, and the health value are weighted and combined to obtain a comprehensive weight value;

[0068] Calculating a training task allocation ratio for the edge computing node based on the comprehensive weight value; allocating training data to the edge computing node based on the training task allocation ratio; performing model training on the edge computing node using the training data to obtain a local training model; and performing weighted aggregation on the local training model based on the comprehensive weight value and the health value to obtain a global training model;

[0069] The dynamic load balancing factor of the edge computing node is calculated based on the global training model, and the dynamic load balancing factor is proportional to the comprehensive weight value and the health value; the computing task allocation coefficient of the edge computing node is generated according to the training task allocation ratio, the dynamic load balancing factor and the change rate of the health value.

[0070] like Figure 2 As shown, the method includes:

[0071] The computational weight coefficients of the nodes in each computing subnetwork are set, and then distributed training is performed on the edge computing nodes in each subnetwork to finally obtain the computational task allocation coefficients of each node.

[0072] During implementation, the system collects computing resource usage data, communication performance data, and operational status data from each edge computing node through the network management module. Computing resource usage data includes CPU utilization, memory usage, and storage space usage; communication performance data includes network bandwidth, communication latency, and packet loss rate; and operational status data includes node online time, failure rate, and response time.

[0073] Based on the collected computing resource usage data, the system calculates the basic weight value of the edge computing node according to the preset resource weight coefficients. For example, if a node has a CPU utilization rate of 60%, a memory utilization rate of 40%, and a storage space utilization rate of 30%, and the weight coefficients of these three resources are 0.5, 0.3, and 0.2 respectively, the basic weight value of the node is calculated as: (1-0.6)×0.5+(1-0.4)×0.3+(1-0.3)×0.2=0.48, indicating that the node still has 48% of its resources available for new tasks.

[0074] Based on communication performance data, the system uses an exponential decay function to calculate the communication weight of edge computing nodes. Lower communication latency results in a higher weight, and higher latency results in an exponentially decreasing weight. For example, assuming a node's communication latency is 50 milliseconds, the system's baseline latency is 20 milliseconds, and the decay factor is 0.05, the node's communication weight is calculated as 0.8 after exponential decay, representing the coefficient of influence of communication performance on task allocation.

[0075] For operational status data, the system calculates the health of edge computing nodes through time series feature analysis. This analysis involves observing node status changes over the past 24 hours and calculating metrics such as failure rate and response time fluctuation. For example, if a node experienced two brief offline events over the past 24 hours, and the average response time fluctuation was within 15%, the calculated health value would be 0.85, indicating the node's stability and reliability.

[0076] The basic weight, communication weight, and health value are weighted together to obtain a comprehensive weight. Assuming the combined weights of the three are 0.4, 0.3, and 0.3, respectively, the node's comprehensive weight is: 0.48 × 0.4 + 0.8 × 0.3 + 0.85 × 0.3 = 0.687, which represents the node's overall strength in the entire distributed system.

[0077] Based on the calculated combined weights, the system calculates the training task allocation ratio for edge computing nodes. For example, in a subnetwork containing five nodes, if the combined weights of these nodes are 0.687, 0.725, 0.634, 0.712, and 0.690, the system allocates 19.9%, 21.0%, 18.4%, 20.6%, and 20.0% of the training tasks to these nodes, respectively.

[0078] Based on the training task allocation ratio, the system distributes training data to edge computing nodes. Assuming there are 10,000 training samples, the five nodes mentioned above are allocated 1,990, 2,100, 1,840, 2,060, and 2,000 training samples, respectively. Each node uses the allocated training data to perform model training, generating its own local training model.

[0079] After training is complete, the system performs a weighted aggregation of the local training models based on the combined weight and health value of each node to form a global training model. The aggregation weight is determined by multiplying the combined weight and health value, and after normalization, it is used to calculate the weighted average of the model parameters. For example, if a node has a combined weight of 0.687 and a health value of 0.85, its aggregation weight is 0.687 × 0.85 = 0.584, which is then normalized and used to calculate the global model aggregation.

[0080] Based on the global training model, the system calculates the dynamic load balancing factor for edge computing nodes. This factor is proportional to the combined weight and health value. For example, if the combined weight of a node is 0.687 and its health value is 0.85, the dynamic load balancing factor can be set to 0.687 × 0.85 × 1.2 = 0.700, where 1.2 is the adjustment coefficient used to control the elasticity of load distribution.

[0081] The computing task allocation coefficient for edge computing nodes is generated based on the training task allocation ratio, dynamic load balancing factor, and the rate of change of the health value. For example, if the training task allocation ratio of a node is 19.9%, the dynamic load balancing factor is 0.700, and the health value has increased by 5% compared to the previous period, the computing task allocation coefficient can be calculated as: 19.9% ​​× 0.700 × 1.05 = 14.6%. This coefficient will be used to allocate subsequent actual computing tasks, ensuring that system resources are fully utilized while maintaining a relatively balanced load across nodes.

[0082] Through the above method, the system can dynamically adjust the computing task allocation according to the actual capabilities and status of the edge computing nodes, improve the efficiency of distributed computing, reduce communication overhead, and enhance the stability and reliability of the system.

[0083] In an optional embodiment, allocating training data to the edge computing node based on the training task allocation ratio; and performing model training on the edge computing node using the training data to obtain a local training model includes:

[0084] Obtaining feature vectors and sample weights of a training sample set, calculating a global feature representation based on the feature vectors and the sample weights, and using the global feature representation to train a teacher model to obtain knowledge representation data of the teacher model;

[0085] Performing a preliminary division of the training data based on the training task allocation ratio, calculating a correlation coefficient between the global feature representation and the local features of each edge computing node, and optimizing the preliminary division result of the training data based on the correlation coefficient and the knowledge representation data to obtain an optimized training data allocation scheme;

[0086] allocating the training data to each edge computing node according to the optimized training data allocation scheme, calculating an adaptive distillation strength coefficient of each edge computing node based on the correlation coefficient, constructing a local training loss function using the adaptive distillation strength coefficient, and performing model training on each edge computing node based on the local training loss function and the knowledge representation data;

[0087] During the model training process, the training convergence data of each edge computing node is obtained, the training temperature parameter is dynamically adjusted according to the training convergence data, the knowledge migration degree between the teacher model and each edge computing node model is calculated, and the optimized training data allocation scheme is dynamically adjusted based on the knowledge migration degree to obtain a local training model of each edge computing node.

[0088] A training sample set containing multiple samples is obtained, each with a corresponding feature vector and sample weight. The system then calculates a global feature representation for this training sample set. This is calculated by taking a weighted average of the feature vectors of all samples, where the weights are the sample weights of each sample. For example, if there are 1000 samples, each with a 128-dimensional feature vector and a sample weight between 0 and 1, a global feature representation vector is obtained by taking a weighted average of these feature vectors according to their weights.

[0089] The teacher model is trained using this global feature representation. The training process follows the standard deep learning model training process, including forward propagation, loss calculation, and backpropagation. After training, knowledge representation data, including model parameters and feature representations, is extracted from the teacher model. This knowledge representation data is used to guide model training on edge nodes. For example, the teacher model can be a deep neural network with 5 million parameters. The trained model parameters and the output features of each layer constitute the knowledge representation data.

[0090] The system performs a preliminary division of the training data based on a predetermined training task allocation ratio. For example, if there are three edge computing nodes with a distribution ratio of 3:4:3, then out of a total of 10,000 training samples, 3,000 will be allocated to node A, 4,000 to node B, and 3,000 to node C.

[0091] Calculate the correlation coefficient between the global feature representation and the local features of each edge computing node. The specific calculation method is: for each edge node, calculate the local feature representation of its assigned data (calculated in the same way as the global feature representation), then calculate the cosine similarity between the local feature representation and the global feature representation to obtain the correlation coefficient. For example, the correlation coefficient of node A is 0.85, that of node B is 0.78, and that of node C is 0.92.

[0092] Based on the correlation coefficient and knowledge representation data, the system optimizes the initial partitioning of the training data. The optimization rule is: nodes with higher correlation coefficients are assigned more samples with similar global characteristics; nodes with lower correlation coefficients are assigned more samples related to the teacher model's knowledge representation to promote knowledge transfer. This results in the initial partitioning being adjusted to 3200 samples allocated to node A, 3600 samples to node B, and 3200 samples to node C.

[0093] Training data is distributed to each edge computing node based on the optimized training data distribution scheme. Simultaneously, an adaptive distillation strength coefficient is calculated for each edge computing node based on the correlation coefficient. The calculation principle is: the lower the correlation coefficient, the greater the distillation strength coefficient, thereby enhancing knowledge transfer. For example, the distillation strength coefficient for node A is 0.3, for node B it is 0.4, and for node C it is 0.2.

[0094] The adaptive distillation strength coefficient is used to construct a local training loss function. This loss function consists of two parts: the standard task loss and the knowledge distillation loss. The weight of the knowledge distillation loss is determined by the adaptive distillation strength coefficient. For example, if the standard task loss is a cross-entropy loss of 1.2, the knowledge distillation loss is 0.8, and the distillation strength coefficient of node A is 0.3, then the total loss of node A is 1.2 + 0.3 × 0.8 = 1.44.

[0095] Based on the local training loss function and knowledge representation data, the system performs model training on each edge computing node. During the training process, the model parameters of each node are updated based on the local data and loss function, while also learning from the knowledge representation of the teacher model.

[0096] During model training, the system continuously collects training convergence data from each edge computing node, including changes in loss and model performance metrics during each training round. Based on this convergence data, the system dynamically adjusts the training temperature parameter. The temperature parameter adjustment rule is as follows: for nodes with slower convergence, increase the temperature parameter to soften the knowledge distillation; for nodes with faster convergence, decrease the temperature parameter. For example, if node A has a slower convergence rate, the temperature parameter is adjusted from the initial value of 2.0 to 2.5; if node B has a moderate convergence rate, the temperature parameter remains at 2.0; and if node C has a faster convergence rate, the temperature parameter is adjusted to 1.8.

[0097] The system calculates the degree of knowledge transfer between the teacher model and each edge computing node model. This calculation is performed by comparing the consistency between the edge node model's output and the teacher model's output for the same input. For example, for the same 100 test samples, if the output consistency between the teacher model and Node A's model is 85%, the knowledge transfer degree is 0.85.

[0098] Based on the degree of knowledge transfer, the system dynamically adjusts the optimized training data allocation scheme. The adjustment rule is: nodes with low knowledge transfer are allocated more samples with knowledge representations similar to the teacher model; nodes with high knowledge transfer are allocated fewer such samples. Through this dynamic adjustment, the system ensures that each node fully learns the knowledge of the teacher model. Ultimately, the system obtains local training models on each edge computing node, all of which have good performance and knowledge representations similar to the teacher model.

[0099] Figure 3 Schematic diagram of the relationship analysis between the adaptive distillation intensity coefficient and the correlation coefficient according to an embodiment of the present invention:

[0100] The figure shows the performance curves of three different distillation strength strategies (adaptive distillation strength, fixed distillation strength -0.5, and fixed distillation strength -0.8) as the feature correlation coefficient changes. The adaptive distillation strength strategy shows a steady upward trend, starting from 0.35 when the feature correlation coefficient is 0, passing through 0.67 (coefficient 0.3), 0.89 (coefficient 0.6), and finally reaching the optimal performance of 1.00 at a coefficient of 0.9. The fixed distillation strength -0.5 strategy shows a clear downward trend, gradually decreasing from the initial 0.90 to 0.50, especially reaching the lowest point of approximately 0.1 at the correlation coefficient of 0.7. The fixed distillation strength -0.8 strategy shows a slow growth trend, starting from 0.00 when the coefficient is 0, passing through 0.30 (coefficient 0.3), 0.60 (coefficient 0.6), and finally reaching 0.85 at a coefficient of 0.9. The data clearly shows that the adaptive distillation strength strategy exhibits significant advantages in the high feature correlation range (0.6-0.9), while being relatively weak in the low correlation range (0-0.3). However, its overall performance curve is the most stable, demonstrating good adaptability and robustness. The chart, in the form of a line graph, clearly distinguishes the three strategies with different markers, intuitively illustrating the performance trends across the entire feature correlation range.

[0101] In an optional embodiment, according to the computing task allocation scheme, the pipe network optimization computing task is allocated to each edge computing node, and the parallel computing is performed to obtain the pipe network status parameters, including:

[0102] Analyze the data dependency between related computing tasks based on the computing task allocation plan, collect the real-time communication delay between edge computing nodes that execute related computing tasks, construct a node communication cost matrix based on the data dependency and the real-time communication delay, and optimize and adjust the computing task allocation plan using the node communication cost matrix to generate a final computing task allocation plan;

[0103] The final computing task allocation plan is sent to each edge computing node, and each edge computing node performs parallel calculation of the pipeline network status parameters according to the computing task type and computing resource allocation ratio specified in the final computing task allocation plan, and integrates the parallel calculation results of each edge computing node to obtain complete pipeline network status parameters.

[0104] The process of optimizing computational task allocation begins with analyzing the data dependencies between computational tasks. The system first decomposes the pipe network optimization computational tasks and identifies a set of interdependent computational tasks. For example, in hydraulic calculations, node pressure calculations depend on flow distribution calculations, while flow balance calculations, in turn, depend on node pressure results. The system represents these dependencies by constructing a directed acyclic graph, where vertices represent computational tasks and edges represent data flows. For a specific pipe network system, assume there are five computational tasks: flow initialization (T1), node pressure calculation (T2), flow balance calculation (T3), water quality analysis (T4), and energy consumption calculation (T5). This analysis reveals the data dependencies: T1 → T2 → T3, T2 → T4, and T3 → T5.

[0105] Collect real-time communication latency between edge computing nodes performing related computing tasks. In a distributed edge computing environment, communication latency between edge nodes significantly impacts computing efficiency. The system deploys a network probe module that measures the round-trip time between each edge computing node by sending probe packets. The system performs multiple measurements during different time periods (such as peak and off-peak periods) to obtain more accurate latency data. For example, in a system consisting of four edge nodes (E1, E2, E3, and E4), the measured communication latency data (in milliseconds) is: E1-E2: 15ms, E1-E3: 25ms, E1-E4: 20ms, E2-E3: 18ms, E2-E4: 22ms, and E3-E4: 16ms.

[0106] Based on data dependencies and real-time communication latency, the system constructs a node communication cost matrix. This matrix is ​​an n×n matrix (n is the number of edge computing nodes), and each element in the matrix represents the communication cost required to transmit a unit of data between two nodes. The cost calculation comprehensively considers factors such as communication latency, bandwidth resource utilization, and network stability. For the four edge nodes mentioned above, the node communication cost matrix C can be expressed as: C[i,j] represents the cost of transmitting data from node i to node j. Based on the measured latency data and taking into account network bandwidth, the system generates a communication cost matrix where C[1,2] = 1.5, C[1,3] = 2.5, C[1,4] = 2.0, C[2,3] = 1.8, C[2,4] = 2.2, and C[3,4] = 1.6 (diagonal elements are 0, indicating that node-to-node communication is free).

[0107] The node communication cost matrix is ​​used to optimize the allocation of computing tasks. The system uses a heuristic algorithm to optimize task scheduling, which considers data dependencies between tasks, communication costs, and the computing power of each edge node. The goal of the algorithm is to minimize the overall completion time, that is, the time from the start of the first task to the completion of the last task.

[0108] For the aforementioned pipe network computing tasks and edge nodes, the initial allocation scheme is: T1 to E1, T2 to E2, T3 to E3, T4 to E4, and T5 to E1. By considering the node communication cost matrix, the system discovers that data is transmitted between T2 and T4, while the communication cost between E2 and E4 is higher. Therefore, the system adjusts the allocation scheme, assigning T4 to E2, thereby reducing communication overhead. The adjusted scheme is: T1 to E1, T2 to E2, T3 to E3, T4 to E2, and T5 to E1.

[0109] The final computing task allocation plan is sent to each edge computing node. The plan distribution process uses a reliable communication protocol to ensure that task allocation instructions are accurately conveyed. The instructions received by each edge computing node include information such as the type of computing task it needs to perform, the source of input data, the proportion of computing resources allocated, and the output target of the results. For example, the instructions received by node E2 include: execute tasks T2 and T4, the input data for T2 comes from the calculation results of T1 calculated by E1, 60% of the computing resources are allocated to T2, and 40% to T4, the results of T2 need to be transmitted to E3 for T3 calculation, and the results of T4 need to be uploaded to the central server.

[0110] Each edge computing node performs parallel computations of network status parameters according to the computing task type and computing resource allocation ratio specified in the final computing task allocation plan. Each node dynamically adjusts computing resources, such as the number of CPU cores and memory usage, based on the characteristics of the assigned task. For compute-intensive tasks, the system allocates more computing resources; for data-intensive tasks, the system optimizes data storage and access strategies. For example, node E3 executing task T3 allocates six CPU cores for traffic balancing calculations and establishes an efficient data caching mechanism to accelerate the processing of frequently accessed data.

[0111] The parallel computation results from each edge computing node are integrated to obtain complete network status parameters. This integration process utilizes a data aggregation method, combining the computation results distributed across various nodes into a central processing unit according to a predetermined data format and protocol. The system performs consistency checks on the received results to ensure data integrity and accuracy. In the event of data loss or inconsistency, the system triggers appropriate error handling mechanisms, such as requesting recalculation or using an alternate data source.

[0112] The final integrated pipeline network status parameters include the flow rate of each pipe section (such as the flow rate of a certain pipe section is 0.8 cubic meters per second), node pressure (such as the pressure of a certain node is 0.6 MPa), water quality indicators (such as the residual chlorine content is 0.5 mg / L) and system energy consumption data (such as the energy consumption of a certain pump station is 85 kWh), etc. These data together constitute a complete description of the pipeline network operation status and provide a decision-making basis for subsequent pipeline network optimization control.

[0113] In an optional embodiment, the computing task allocation plan is written into the blockchain to construct a computing task transaction block; the pipe network status parameter is signed and encrypted based on the timestamp, and the encrypted pipe network status parameter is verified for consistency through a voting mechanism, including:

[0114] Generate a task information sequence containing the task identifier, allocation scheme, timestamp, and previous block hash, perform a hash concatenation operation on the task information sequence with the block metadata and node signature to generate a computing task transaction block, use the public key to verify the link relationship between adjacent computing task transaction blocks, and generate a blockchain integrity verification result;

[0115] Confirming the validity of the computing task transaction block based on the blockchain integrity verification result, associating the state parameter with the timestamp and encrypting it using a symmetric key to obtain an encrypted state parameter, performing a hash calculation on the encrypted state parameter and generating multi-signature data using the private keys of multiple nodes, dynamically adjusting the encryption key length using an exponential decay method based on the communication time interval between nodes and the system security risk, and updating the encryption strength of the encrypted state parameter;

[0116] Obtain the historical reputation value and real-time computing power index of the participating verification nodes, calculate the node voting weight based on the historical reputation value and the real-time computing power index, use the node voting weight and the multi-signature data to perform consistency voting verification on the encryption state parameters, dynamically adjust the consensus threshold of the consistency verification based on the number of faulty nodes in the system and historical consensus data, and generate a verification result.

[0117] like Figure 4 As shown, the method includes:

[0118] The computing task allocation plan is written into the blockchain, and the computing task transaction block is constructed. Then, the pipeline network status parameters are signed and encrypted based on the timestamp, and the encrypted pipeline network status parameters are verified for consistency through a voting mechanism.

[0119] When constructing a computing task transaction block, the system generates a task information sequence consisting of a task identifier, an allocation scheme, a timestamp, and a previous block hash. The task identifier is a unique string called "Task-GWJSCN20230612001." The allocation scheme contains the node ID and task assignment relationship, such as "{Node1:TaskA,Node2:TaskB,Node3:TaskC}." The timestamp is in the UTC format "2023-06-12T10:15:30Z." The previous block hash is a 64-bit hexadecimal string called "8f7d6e5c4b3a2910."

[0120] The task information sequence is hashed together with the block metadata and node signature to generate the computational task transaction block. Block metadata includes fields such as version number "1.0", difficulty value "3", and transaction count "15". The node signature uses a digital signature generated using the ECDSA algorithm. The concatenation operation uses the SHA-256 hash algorithm to generate a hash value such as "9e8d7c6b5a4f3e2d1c0b9a8f7e6d5c4b" as the block identifier.

[0121] When using public keys to verify the links between adjacent computing task transaction blocks, the system compares the hash of the previous block in the current block with the hash of the previous block and verifies the node signature in the previous block. If the comparison results are consistent and the signature is valid, a blockchain integrity verification result with the value "Valid" is generated; otherwise, the value "Invalid" is generated and the cause of the exception is recorded.

[0122] The validity of the computing task transaction block is confirmed based on the blockchain integrity verification result. When the verification result is "Valid", the system associates the pipe network status parameters with the timestamp and encrypts them using a symmetric key. The pipe network status parameters include the flow parameter "120.5m 3 / h", pressure parameter "0.65MPa", temperature parameter "25.3℃", etc. The system uses the AES-256 algorithm and the key "k7dJq9sM3pR5tX2z" to encrypt the associated parameters and generate encrypted state parameters. SHA-256 hash calculation is performed on the encrypted state parameters to obtain a hash value, which is then signed using the private keys of at least three nodes to generate multi-signature data.

[0123] The encryption key length is dynamically adjusted using an exponential decay method based on the inter-node communication interval and the system security risk. When the inter-node communication interval exceeds a threshold of 500 milliseconds, or the system security risk exceeds 0.6, the key length increases from the original 256 bits to 512 bits. When the inter-node communication interval falls below a threshold of 200 milliseconds and the system security risk is less than 0.3, the key length reverts to 256 bits. Real-time security risk calculation is based on a comprehensive assessment of indicators such as network congestion rate, external access request frequency, and the proportion of abnormal nodes.

[0124] To obtain the historical reputation and real-time computing power metrics of participating validation nodes, the system extracts data from the blockchain, including verification accuracy, response time, and resource contribution, over the past 100 block periods. Node A's historical reputation is 0.95 (out of a maximum score of 1.0), Node B's is 0.87, and Node C's is 0.92. The real-time computing power metric, calculated by measuring the node's current CPU utilization, memory usage, and network bandwidth, is 85 (out of a maximum score of 100) for Node A, 78 for Node B, and 92 for Node C.

[0125] Node voting weights are calculated based on historical reputation and real-time computing power. Weights are calculated using a weighted approach where historical reputation accounts for 60% and real-time computing power accounts for 40%. Node A's voting weight is 0.95 × 0.6 + 85 / 100 × 0.4 = 0.91, Node B's voting weight is 0.87 × 0.6 + 78 / 100 × 0.4 = 0.834, and Node C's voting weight is 0.92 × 0.6 + 92 / 100 × 0.4 = 0.92. Node voting weights and multi-signature data are used to perform consensus voting verification on the cryptographic state parameters. Each node verifies the integrity and authenticity of the cryptographic state parameters and votes based on the verification results.

[0126] The consensus threshold for consistency verification is dynamically adjusted based on the number of faulty nodes in the current operating environment and historical consensus data. When the faulty node ratio is less than 5%, the consensus threshold is set to 0.85; when the faulty node ratio is between 5% and 15%, the consensus threshold is lowered to 0.75; and when the faulty node ratio exceeds 15%, the consensus threshold is further lowered to 0.67. Historical consensus data shows that the system average consensus over the past 24 hours was 0.92, and the faulty node ratio was 3%. Therefore, the current consensus threshold is set to 0.85.

[0127] During the consensus voting verification process, each node verifies the encrypted state parameters and submits its voting results. The system aggregates the weighted voting results of all nodes to calculate an overall consistency score. For example, in a certain round of verification, if nodes A and C vote "Valid" and node B votes "Invalid", the overall consistency score is 0.91 + 0.92 = 1.83, the total node weight is 0.91 + 0.834 + 0.92 = 2.664, and the consistency ratio is 1.83 / 2.664 = 0.687. Because the consistency ratio of 0.687 is less than the current consensus threshold of 0.85, the verification result for this round is "Failed", and the system will trigger a re-verification process.

[0128] If the consistency ratio is greater than or equal to the consensus threshold, the verification result is "Passed". The system confirms that the current encrypted state parameters are valid and updates them to the blockchain network, completing the consistency verification process of the entire pipeline network state parameters.

[0129] In an optional embodiment, performing a hash connection operation on the task information sequence, the block metadata, and the node signature to generate a computing task transaction block, and using a public key to verify the link relationship between adjacent computing task transaction blocks to generate a blockchain integrity verification result includes:

[0130] Extracting feature data including task identifier, block metadata, and node signature from the task information sequence, calculating computational complexity and security strength of the feature data, and generating a feature fitness evaluation value;

[0131] Selecting a basic hash function based on the characteristic fitness evaluation value, dynamically optimizing the basic hash function using an exponentially decaying mutation step size, and introducing a random mutation factor to construct an evolutionary hash algorithm; calculating the anti-attack capability and group diversity index of the blockchain node, and generating group immunity strength based on the anti-attack capability and the group diversity index;

[0132] Performing a hash join operation on the task information sequence, the block metadata, and the node signature using the evolutionary hash algorithm, adjusting hash join operation parameters according to the group immunity strength, and generating a computing task transaction block;

[0133] A block verification strategy is constructed based on the evolutionary hash algorithm and the group immunity strength. According to the block verification strategy, the link relationship between adjacent computing task transaction blocks is verified using the public key to generate a blockchain integrity verification result.

[0134] The generation and verification process of computing task transaction blocks begins by extracting feature data from the task information sequence. The system receives a task information sequence containing detailed information about multiple computing tasks. For each task, feature data is extracted, including the task identifier (e.g., task ID: "TASK-20230712-0001"), block metadata (e.g., timestamp: "1689129600", previous block hash: "7a8b9c0d1 e2f3g4h5 i 6j"), and node signature (e.g., digital signature: "5d6e7f8g9h0 i 1j2k3 l4m").

[0135] The computational complexity of these feature data was calculated by analyzing the length of the task identifier, the information entropy of the metadata, and the number of bits in the signature, resulting in a computational complexity value of 87.5. Furthermore, the security strength was evaluated by considering the strength of the signature algorithm and the key length, resulting in a security strength value of 92.3. Taking a weighted average of the computational complexity and security strength (with weights of 0.4 and 0.6, respectively), the feature fitness evaluation value was 90.3.

[0136] The system selects a base hash function based on the feature fitness evaluation value. When the feature fitness evaluation value is between 85 and 95, the system selects SHA-256 as the base hash function. If the evaluation value is below 85, the system selects MD5, which has lower computational overhead; if it is above 95, the system selects SHA-384, which is more secure. For the evaluation value of 90.3 in this example, the system selects SHA-256 as the base hash function.

[0137] The basic hash function is dynamically optimized using an exponentially decaying mutation step size. The initial mutation step size is set to 0.1, and as the number of iterations increases, the step size decreases according to the exponential decay formula. For example, the step size is 0.1 after the first iteration, decreases to 0.06 after the fifth iteration, and further decreases to 0.036 after the tenth iteration. The system introduces a random mutation factor in each iteration, with an initial value set to 0.15 and randomly fluctuating between 0.05 and 0.25 in each iteration. By combining the mutation step size with the random mutation factor, the parameters of the hash function (such as the initial vector value and the number of rounds) are fine-tuned to construct an evolutionary hash algorithm.

[0138] The system calculates the attack resistance and population diversity indicators of blockchain nodes. Attack resistance is measured by simulating common attack scenarios (such as denial of service and 51% attacks) to test the node's response and generate a score. For example, in a network with 10 nodes, node 1 has a DoS attack resistance of 0.85 and a 51% attack resistance of 0.92, resulting in an overall attack resistance of 0.88. The population diversity indicator is calculated by assessing the heterogeneity of nodes in the network (such as operating system type, hardware configuration, and geographic distribution). In the test network, the operating system diversity was 0.78, the hardware diversity was 0.65, the geographic distribution diversity was 0.82, and the overall diversity index was 0.75. The weighted sum of the attack resistance and population diversity indicators (weights 0.6 and 0.4, respectively) yields a herd immunity strength of 0.826.

[0139] The system uses an evolutionary hashing algorithm to perform a hash concatenation operation on the task information sequence, block metadata, and node signature. The system first concatenates the task information sequence (e.g., "TASK-20230712-0001|Computational Image Processing|Priority: High"), block metadata (e.g., "1689129600|7a8b9c0d1 e2f3g4h5 i 6j|Difficulty: 4"), and node signature (e.g., "5d6e7f8g9h0i 1j2k3 l4m") into a single string.

[0140] Based on the herd immunity strength of 0.826, the hash join parameters are adjusted. For example, the number of hash iterations is increased from the standard 64 to 73, with the increase approximately equal to the herd immunity strength. The adjusted evolutionary hash algorithm is executed to generate the hash value of the computation task transaction block: "a1 b2c3d4e5f6g7h8 i9j0k1 l 2m3n4o5p6". This hash value is combined with the block header and other information to form a complete computation task transaction block.

[0141] The block verification strategy is built based on the evolutionary hash algorithm and herd immunity strength. The system sets the verification threshold to 0.85 based on the herd immunity strength of 0.826, requiring at least 85% of the verification nodes to reach a consensus to confirm the validity of the block.

[0142] The node signature is verified using a public key (e.g., "pub_key:8f9g0h1 i2j3k4 l5m6n") to ensure it comes from the claimed node. The system then verifies the links between adjacent computational task transaction blocks, checking whether the hash of the previous block contained in the current block matches the hash of the actual previous block. For example, if the hash value of the previous block in block #1025 is "7a8b9c0d1 e2f3g4h5 i 6j," the system verifies that this value matches the actual hash value of block #1024.

[0143] Through the above verification steps, the system generates a blockchain integrity verification result, which includes the verification status ("Verification Passed" or "Verification Failed"), the consistency level (such as "Node Consistency Rate: 92%), and a timestamp (such as "Verification Time: 1689130600"). The verification result is recorded and broadcast to other nodes in the network, ensuring that the entire network has a consistent understanding of the blockchain status.

[0144] In an optional embodiment, calculating the pressure adjustment parameter, flow adjustment parameter, and valve opening parameter of each pipeline network node according to the pipeline network state parameter and generating the pipeline network optimization control instruction includes:

[0145] Characteristically encoding the pressure parameters, flow parameters, and valve parameters of the pipeline network nodes to generate a node state vector, and constructing a multi-mode intelligent agent based on the node state vector, including a path optimization function, a flow balancing function, and a collaborative optimization function;

[0146] Extracting pressure gradient and pheromone intensity from the multimodal agent to calculate path optimization parameters, obtaining node flow deviation values ​​based on the path optimization parameters and generating flow balance parameters in combination with group behavior characteristics, and combining the flow balance parameters with information propagation function and exploration function to generate collaborative decision parameters;

[0147] The pressure regulation parameter, the flow regulation parameter and the valve opening parameter are calculated respectively according to the path optimization parameter, the flow balancing parameter and the collaborative decision parameter, and the pressure regulation parameter, the flow regulation parameter and the valve opening parameter are combined to generate a pipeline network optimization control instruction.

[0148] When the pressure parameters, flow parameters and valve parameters of the network nodes are encoded to generate the node state vector, the real-time pressure value such as 0.45MPa and flow value such as 25m are collected for each node. 3 / h and valve opening such as 65%, these heterogeneous parameters are mapped to the range of [0,1] through normalization. For example, for a certain node, its pressure parameter 0.45MPa is normalized to 0.417 in the range of [0.2MPa, 0.8MPa] set by the system, and the flow parameter 25m 3 / h in [0,100m 3 / h] is normalized to 0.25, and the valve opening of 65% is normalized to 0.65.

[0149] These normalized values ​​are combined using vector concatenation to form the node's state vector [0.417, 0.25, 0.65]. This process is repeated for each node in the network to generate a complete set of node state vectors. Based on these state vectors, a multimodal agent is constructed, which includes path optimization, flow balancing, and collaborative optimization functions. The multimodal agent adopts a three-layer architecture: the bottom layer implements path optimization logic, the middle layer implements flow balancing calculations, and the top layer is responsible for collaborative decision-making. A parameter sharing mechanism is implemented within the agent, and the three functional modules achieve coordinated optimization through information exchange.

[0150] The specific steps for extracting pressure gradients and pheromone intensities from a multimodal agent to calculate path optimization parameters are as follows: The agent first collects pressure differences between adjacent nodes in the network, such as the pressure difference between nodes A and B is 0.05 MPa, and converts these differences into pressure gradient vectors. Simultaneously, the agent records the pheromone intensities along historically optimized paths. The pheromone intensities for optimal paths are set to a higher value, such as 0.8, for suboptimal paths to 0.5, and for suboptimal paths to 0.2. The path optimization parameters are calculated by weightedly combining the pressure gradients and pheromone intensities. For example, for the path [node 1 → node 2 → node 3], if the pressure gradient component is [0.07, 0.05] and the pheromone intensities are [0.7, 0.6], and the weight coefficients are set to 0.6 and 0.4, the path optimization parameters are calculated as [0.07 × 0.6 + 0.7 × 0.4, 0.05 × 0.6 + 0.6 × 0.4] = [0.322, 0.27].

[0151] The specific implementation method of obtaining the node flow deviation value based on the path optimization parameters and generating the flow balancing parameters in combination with the group behavior characteristics is as follows: the system compares the actual flow of each node with the expected flow and calculates the flow deviation value. For example, the actual flow of a node is 25m 3 / h, the expected flow rate is 30m 3 / h, the deviation value is -5m 3 / h, which is -0.167 after normalization.

[0152] Analyze the flow distribution characteristics of node groups across the entire pipe network and extract characteristic values ​​such as flow uniformity and aggregation. For example, if flow uniformity is 0.75 and aggregation is 0.35, then combine the node flow deviation value with group behavior characteristics to generate flow balancing parameters. For example, a node has a flow deviation value of -0.167. Combined with flow uniformity of 0.75 and aggregation of 0.35, a weighted average calculation yields a flow balancing parameter of -0.167 × 0.5 + 0.75 × 0.3 + 0.35 × 0.2 = -0.0835 + 0.225 + 0.07 = 0.2115.

[0153] The method for combining traffic balancing parameters with information propagation and exploration functions to generate collaborative decision-making parameters is as follows: the system establishes an information propagation network based on the connections between nodes, setting the information decay rate to 0.15. Starting with a node's traffic balancing parameter of 0.2115, the information propagation function calculates its impact on adjacent nodes, for example, 0.2115 × (1 - 0.15) = 0.1798. Simultaneously, the system runs the exploration function, adding a certain degree of randomness to each decision, such as a random perturbation value of ±0.05. The information propagation results are combined with the output of the exploration function to generate collaborative decision-making parameters, such as 0.1798 + 0.03 = 0.2098.

[0154] The process of calculating the pressure adjustment parameter, flow adjustment parameter and valve opening parameter according to the path optimization parameter, flow balancing parameter and collaborative decision parameter is as follows: For the pressure adjustment parameter, the system converts the path optimization parameter 0.322 into the actual pressure adjustment value 0.027MPa through the mapping function, that is, the current pressure value needs to be increased by 0.027MPa. For the flow adjustment parameter, the system converts the flow balancing parameter 0.2115 into the flow adjustment amount 6.345m 3 / h, indicating the need to increase 6.345m 3 / h flow rate.

[0155] For the valve opening parameter, the system converts the collaborative decision parameter 0.2098 into a valve opening adjustment of 10.49%, indicating that the valve opening needs to be increased by 10.49%. The system combines these three types of parameters into a control instruction package according to the preset format. The instruction package structure is [node identifier, pressure adjustment value, flow adjustment value, valve adjustment value], such as [Node_A, +0.027MPa, +6.345m 3 / h, +10.49%]. These optimization and control instructions are sent to each execution unit through the communication network. After receiving the instructions, the execution unit performs the corresponding adjustment operations to achieve optimized operation of the pipeline network.

[0156] Figure 5 This is a bar chart comparing and analyzing the efficiency of the pipe network optimization and control method according to an embodiment of the present invention:

[0157] The figure shows comparative data on five core performance indicators for three different control methods (traditional control method, general intelligent optimization, and multi-mode intelligent agent method). In terms of pressure regulation optimization, the traditional method achieved 72.5%, while general intelligent optimization improved to 85.6%, and the multi-mode intelligent agent method achieved the highest 94.3%. In terms of flow balancing efficiency, the three methods achieved 68.3%, 82.4%, and 91.7%, respectively. Valve collaborative control performance reached 65.7%, 79.8%, and 93.2%, respectively. System response time optimization was 78.2%, 84.5%, and 92.5%, respectively. In terms of energy consumption reduction, the multi-mode intelligent agent method performed best, reaching 88.6%, far exceeding the 59.4% of the traditional method and the 76.2% of the general intelligent optimization. The data clearly shows that the multimodal agent-based approach significantly outperforms the other two methods across all evaluation metrics, with the most notable improvements in pressure regulation optimization and valve coordinated control, exceeding traditional methods by 21.8 and 27.5 percentage points, respectively. This demonstrates the method's outstanding advantages in the field of intelligent pipe network control. The chart uses a bar graph to visually illustrate the performance differences between the three methods, using grayscale to distinguish between the different methods. A horizontal comparison highlights the comprehensive advantages of the multimodal agent-based approach.

[0158] A second aspect of an embodiment of the present invention provides a natural gas pipeline network intelligent operation and scheduling system based on user priority, including:

[0159] The first unit is used to collect pressure data, flow data, temperature data and user gas usage data from multiple edge computing nodes in the natural gas pipeline network;

[0160] The second unit is configured to divide the edge computing nodes into a plurality of computing subnetworks according to the geographic location information and the network topology of the edge computing nodes; set a computing weight coefficient for each node based on the computing power and communication delay of the nodes in each computing subnetwork; perform distributed training on the edge computing nodes in each subnetwork to obtain a computing task allocation coefficient for each node; and generate a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient;

[0161] The third unit is used to allocate the pipe network optimization computing tasks to each edge computing node according to the computing task allocation scheme, and perform parallel computing to obtain the pipe network status parameters;

[0162] The fourth unit is configured to write the computing task allocation plan into the blockchain and construct a computing task transaction block; sign and encrypt the pipeline network status parameters based on the timestamp, and verify the consistency of the encrypted pipeline network status parameters through a voting mechanism;

[0163] The fifth unit is used to calculate the pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline node based on the pipeline state parameters, generate pipeline optimization control instructions, execute the pipeline optimization control instructions, and realize intelligent control of the natural gas pipeline network.

[0164] According to a third aspect of an embodiment of the present invention, an electronic device is provided, including:

[0165] processor;

[0166] a memory for storing processor-executable instructions;

[0167] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0168] According to a fourth aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0169] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A natural gas pipeline network intelligent operation and scheduling method based on user priority, characterized in that: include: Collect pressure data, flow data, temperature data, and user gas usage data from multiple edge computing nodes in the natural gas pipeline network; Divide the edge computing nodes into multiple computing subnetworks based on the geographic location information and network topology of the edge computing nodes; set a computing weight coefficient for each node based on the computing power and communication delay of each node in the computing subnetwork; Perform distributed training on the edge computing nodes in each sub-network to obtain the computing task allocation coefficient of each node; Generate a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient; According to the computing task allocation scheme, the pipe network optimization computing tasks are allocated to each edge computing node, and parallel computing is performed to obtain the pipe network status parameters; Writing the computing task allocation plan into the blockchain to construct a computing task transaction block; Signing and encrypting the pipeline network status parameters based on the timestamp, and verifying the consistency of the encrypted pipeline network status parameters through a voting mechanism; The pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline network node are calculated according to the pipeline network state parameters, and pipeline network optimization control instructions are generated. The pipeline network optimization control instructions are executed to realize intelligent control of the natural gas pipeline network.

2. The method according to claim 1, characterized in that Based on the computing power and communication delay of each node in the computing subnetwork, setting a computing weight coefficient for each node; Distributed training is performed on the edge computing nodes in each sub-network, and the computing task allocation coefficients of each node are obtained, including: Obtain computing resource usage data, communication performance data, and operating status data of edge computing nodes; Based on the computing resource usage data, a basic weight value of the edge computing node is calculated according to a preset resource weight coefficient; based on the communication performance data, a communication weight value of the edge computing node is calculated using an exponential decay function; based on the operating status data, a health value of the edge computing node is calculated by time series feature analysis; the basic weight value, the communication weight value, and the health value are weighted and combined to obtain a comprehensive weight value; Calculating a training task allocation ratio for the edge computing node based on the comprehensive weight value; allocating training data to the edge computing node based on the training task allocation ratio; performing model training on the edge computing node using the training data to obtain a local training model; and performing weighted aggregation on the local training model based on the comprehensive weight value and the health value to obtain a global training model; The dynamic load balancing factor of the edge computing node is calculated based on the global training model, and the dynamic load balancing factor is proportional to the comprehensive weight value and the health value; the computing task allocation coefficient of the edge computing node is generated according to the training task allocation ratio, the dynamic load balancing factor and the change rate of the health value.

3. The method according to claim 2, characterized in that Allocate training data to the edge computing nodes based on the training task allocation ratio; Performing model training on the edge computing node using the training data to obtain a local training model includes: Obtaining feature vectors and sample weights of a training sample set, calculating a global feature representation based on the feature vectors and the sample weights, and using the global feature representation to train a teacher model to obtain knowledge representation data of the teacher model; Performing a preliminary division of the training data based on the training task allocation ratio, calculating a correlation coefficient between the global feature representation and the local features of each edge computing node, and optimizing the preliminary division result of the training data based on the correlation coefficient and the knowledge representation data to obtain an optimized training data allocation scheme; allocating the training data to each edge computing node according to the optimized training data allocation scheme, calculating an adaptive distillation strength coefficient of each edge computing node based on the correlation coefficient, constructing a local training loss function using the adaptive distillation strength coefficient, and performing model training on each edge computing node based on the local training loss function and the knowledge representation data; During the model training process, the training convergence data of each edge computing node is obtained, the training temperature parameter is dynamically adjusted according to the training convergence data, the knowledge migration degree between the teacher model and each edge computing node model is calculated, and the optimized training data allocation scheme is dynamically adjusted based on the knowledge migration degree to obtain a local training model of each edge computing node.

4. The method according to claim 1, wherein According to the computing task allocation scheme, the network optimization computing tasks are allocated to each edge computing node, and parallel computing is performed to obtain the network status parameters including: Analyze the data dependency between related computing tasks based on the computing task allocation plan, collect the real-time communication delay between edge computing nodes that execute related computing tasks, construct a node communication cost matrix based on the data dependency and the real-time communication delay, and optimize and adjust the computing task allocation plan using the node communication cost matrix to generate a final computing task allocation plan; The final computing task allocation plan is sent to each edge computing node, and each edge computing node performs parallel calculation of the pipeline network status parameters according to the computing task type and computing resource allocation ratio specified in the final computing task allocation plan, and integrates the parallel calculation results of each edge computing node to obtain complete pipeline network status parameters.

5. The method according to claim 1, wherein Writing the computing task allocation plan into the blockchain to construct a computing task transaction block; Signing and encrypting the pipe network status parameters based on the timestamp, and verifying the consistency of the encrypted pipe network status parameters through a voting mechanism include: Generate a task information sequence containing the task identifier, allocation scheme, timestamp, and previous block hash, perform a hash concatenation operation on the task information sequence with the block metadata and node signature to generate a computing task transaction block, use the public key to verify the link relationship between adjacent computing task transaction blocks, and generate a blockchain integrity verification result; Confirming the validity of the computing task transaction block based on the blockchain integrity verification result, associating the state parameter with the timestamp and encrypting it using a symmetric key to obtain an encrypted state parameter, performing a hash calculation on the encrypted state parameter and generating multi-signature data using the private keys of multiple nodes, dynamically adjusting the encryption key length using an exponential decay method based on the communication time interval between nodes and the system security risk, and updating the encryption strength of the encrypted state parameter; Obtain the historical reputation value and real-time computing power index of the participating verification nodes, calculate the node voting weight based on the historical reputation value and the real-time computing power index, use the node voting weight and the multi-signature data to perform consistency voting verification on the encryption state parameters, dynamically adjust the consensus threshold of the consistency verification based on the number of faulty nodes in the system and historical consensus data, and generate a verification result.

6. The method according to claim 5, characterized in that The task information sequence is hashed together with the block metadata and node signature to generate a computing task transaction block. The public key is used to verify the link relationship between adjacent computing task transaction blocks. The blockchain integrity verification result generated includes: Extracting feature data including task identifier, block metadata, and node signature from the task information sequence, calculating computational complexity and security strength of the feature data, and generating a feature fitness evaluation value; Selecting a basic hash function based on the characteristic fitness evaluation value, dynamically optimizing the basic hash function using an exponentially decaying mutation step size, and introducing a random mutation factor to construct an evolutionary hash algorithm; calculating the anti-attack capability and group diversity index of the blockchain node, and generating group immunity strength based on the anti-attack capability and the group diversity index; Performing a hash join operation on the task information sequence, the block metadata, and the node signature using the evolutionary hash algorithm, adjusting hash join operation parameters according to the group immunity strength, and generating a computing task transaction block; A block verification strategy is constructed based on the evolutionary hash algorithm and the group immunity strength. According to the block verification strategy, the link relationship between adjacent computing task transaction blocks is verified using the public key to generate a blockchain integrity verification result.

7. The method according to claim 1, characterized in that Calculating the pressure adjustment parameters, flow adjustment parameters, and valve opening parameters of each pipeline network node according to the pipeline network state parameters, and generating pipeline network optimization control instructions include: Characteristically encoding the pressure parameters, flow parameters, and valve parameters of the pipeline network nodes to generate a node state vector, and constructing a multi-mode intelligent agent based on the node state vector, including a path optimization function, a flow balancing function, and a collaborative optimization function; Extracting pressure gradient and pheromone intensity from the multimodal agent to calculate path optimization parameters, obtaining node flow deviation values ​​based on the path optimization parameters and generating flow balance parameters in combination with group behavior characteristics, and combining the flow balance parameters with information propagation function and exploration function to generate collaborative decision parameters; The pressure regulation parameter, the flow regulation parameter and the valve opening parameter are calculated respectively according to the path optimization parameter, the flow balancing parameter and the collaborative decision parameter, and the pressure regulation parameter, the flow regulation parameter and the valve opening parameter are combined to generate a pipeline network optimization control instruction.

8. A natural gas pipeline network intelligent operation and scheduling system based on user priority, used to implement the method according to any one of claims 1 to 7, characterized in that: include: The first unit is used to collect pressure data, flow data, temperature data and user gas usage data from multiple edge computing nodes in the natural gas pipeline network; A second unit is configured to divide the edge computing node into a plurality of computing subnetworks according to the geographic location information and the network topology of the edge computing node; Based on the computing power and communication delay of each node in the computing subnetwork, setting a computing weight coefficient for each node; Perform distributed training on the edge computing nodes in each sub-network to obtain the computing task allocation coefficient of each node; Generate a computing task allocation plan according to the computing task allocation coefficient and the computing weight coefficient; The third unit is used to allocate the pipe network optimization computing tasks to each edge computing node according to the computing task allocation scheme, and perform parallel computing to obtain the pipe network status parameters; The fourth unit is used to write the computing task allocation plan into the blockchain and construct a computing task transaction block; Signing and encrypting the pipeline network status parameters based on the timestamp, and verifying the consistency of the encrypted pipeline network status parameters through a voting mechanism; The fifth unit is used to calculate the pressure regulation parameters, flow regulation parameters and valve opening parameters of each pipeline node based on the pipeline state parameters, generate pipeline optimization and control instructions, execute the pipeline optimization and control instructions, and realize intelligent control of the natural gas pipeline network.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images