Gas system management method and system based on data analysis

Abstract

The invention discloses a fuel gas system management method and system based on data analysis, and belongs to the technical field of fuel gas system management.The fuel gas system management method comprises the steps that a wax deposition kinetic model and a deep neural network are combined, and a hybrid prediction model architecture is designed; the wax deposition kinetic model is used for describing physical and chemical processes of wax deposition and providing basic prediction capability; the deep neural network is used for capturing a non-linear relationship and a complex mode in the data and improving the accuracy and generalization ability of prediction; based on an improved particle swarm optimization algorithm (PSO-ADE), a self-adaptive pigging decision engine is designed; initializing the algorithm to generate an initial particle swarm; operating an algorithm, and evaluating and updating the particles according to a target function; through iterative optimization, an optimal pigging strategy is found; and feedback data is collected, and the algorithm is further adjusted and optimized, so that the accuracy and effectiveness of decision making are improved. The pipeline cleaning strategy of the oil pipeline can be optimized.

Description

Technical Field

[0001] This application relates to the technical field of gas system management, and particularly to a gas system management method and system based on data analysis. Background Art

[0002] As an important part of urban infrastructure, the gas system not only plays a key role in residents' lives, industrial production, and commercial services, but also has a positive impact on environmental protection and economic development. Therefore, optimizing the management and operation efficiency of the gas system is of great significance for improving the quality of urban life, promoting industrial development, and driving commercial prosperity.

[0003] With the acceleration of urbanization and the popularization of gas use, the safe and efficient management of the gas system has become an increasingly important issue. In the gas system, wax deposition is a complex and difficult-to-accurately-predict physical and chemical process. Wax deposition not only affects the flow capacity of pipelines, but may also cause pipeline blockages, increased pressure losses, and safety hazards. Traditional wax deposition prediction methods are often based on simplified physical models, which are difficult to comprehensively capture the non-linear relationships and complex patterns in the wax deposition process, resulting in inaccurate prediction results.

[0004] At the same time, the formulation of pigging strategies is also a challenging issue. Pigging operations consume a large amount of resources and time, and frequent pigging may damage the pipeline. Therefore, how to formulate an effective pigging strategy based on real-time wax deposition prediction results to reduce unnecessary pigging operations, improve pigging efficiency, and reduce operating costs is an urgent problem to be solved. Summary of the Invention

[0005] To solve these problems, this application provides a gas system management method and system based on data analysis. This method combines a wax deposition kinetic model with a deep neural network, designs a hybrid prediction model architecture to improve the accuracy and generalization ability of wax deposition prediction. At the same time, based on an improved particle swarm optimization algorithm (PSO-ADE), an adaptive pigging decision engine is designed to find the optimal pigging strategy through iterative optimization.

[0006] The gas system management method and system based on data analysis provided by this application adopt the following technical solutions:

[0007] In a first aspect, this application provides a gas system management method based on data analysis, including the following steps:

[0008] Combine the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture; the wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture non-linear relationships and complex patterns in the data to improve the accuracy and generalization ability of the prediction;

[0009] Design an adaptive pigging decision engine based on the improved particle swarm optimization algorithm (PSO-ADE);

[0010] Initialize the algorithm to generate an initial particle swarm;

[0011] Run the algorithm to evaluate and update the particles according to the objective function;

[0012] Find the optimal pigging strategy through iterative optimization;

[0013] Collect feedback data to further adjust and optimize the algorithm to improve the accuracy and effectiveness of the decision-making.

[0014] Furthermore, in the step of combining the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture, it further includes:

[0015] Build a thermal-fluid-solidification four-field coupling model based on COMSOL Multiphysics, and the solution equations include:

[0016]

[0017] Among them, ρ is the fluid density, which represents the mass of the fluid per unit volume; u is the velocity vector, which represents the movement velocity of the fluid microelement; t represents time; p is the pressure, which represents the force acting vertically on the unit area inside the fluid; μ is the dynamic viscosity coefficient, which represents the magnitude of the internal friction force of the fluid and reflects the viscosity of the fluid; S wax is the wax phase change source term, and an improved molecular diffusion-gel deposition dual-mechanism model is adopted to represent the influence of the phase change of wax in the fluid on the flow;

[0018] Introduce dynamic grid adaptive technology to automatically refine the grid to a resolution of 0.5 mm in the area where the deposition rate changes suddenly;

[0019] Train the XGBoost-GRU hybrid model, and the input features include temperature gradient (ΔT), wall shear stress (τ_w), and wax crystal volume fraction (φ), and the output is the deposition rate correction factor α;

[0020] Use a generative adversarial network (GAN) to augment the limited experimental data and generate a synthetic data set covering the temperature range of -10 to 50 °C to solve the problem of insufficient extrapolation ability of traditional models.

[0021] Furthermore, in the step of initializing the algorithm, it further includes:

[0022] Establish an objective function:

[0023]

[0024] where C pigging is the cost of the pigging operation, representing the cost of each pigging operation; f i is the frequency of the i-th pigging, representing the number of pigging operations within a certain period of time; λ is the risk penalty coefficient, representing the degree of penalty for risk and reflecting the cost of risk; δ i is the predicted deposition thickness, representing the thickness of the sediment in the pipeline predicted during the i-th pigging operation; δ safe is the safe deposition thickness, representing the safe thickness of the sediment in the pipeline, and exceeding this thickness may bring risks;

[0025] The purpose of the objective function is to find a balance between the cost of the pigging operation and the risk brought by the sediment, so as to minimize the total cost;

[0026] Adopt an improved PSO-ADE algorithm and introduce an inertia weight adaptive mechanism to enhance the local search ability in the later stage of iteration. The formula is as follows:

[0027] w(t) = w min +(w max -w min )·e -20t / T

[0028] The formula is used to describe the change of w(t) with time t. The specific parameter meanings are as follows: w min represents the minimum value or stable value of w(t). When time t approaches infinity, w(t) will approach this value; w max represents the initial maximum value of w(t). When time t = 0, the value of w(t) is w max ; t represents time; T represents the time constant, with the same unit as time t, reflecting the speed at which w(t) decays to the stable value. The larger the time constant T, the slower the decay process.

[0029] Furthermore, it further includes:

[0030] Select typical accidents of the gas system as the analysis object and clarify the severity of the accident consequences;

[0031] Regard the typical accidents of the gas system as the top event and decompose the top event into multiple sub-top events, where the sub-top events are the possible causes of the top event;

[0032] Construct an accident tree logic model based on the secondary top event and the top event; use "AND gates" (all conditions must be met simultaneously) and "OR gates" (any one condition is met) to decompose the causes layer by layer until the basic events corresponding to the secondary top event are obtained;

[0033] Calculate the minimum cut sets and structure importance coefficients. The minimum cut sets are the combinations of the fewest basic events that lead to the top event and are used to identify key risk factors;

[0034] Structure importance coefficient

[0035] Among them, I i represents the value of the i-th index, which is a comprehensive value obtained by summation; ∑ represents the summation operation, which accumulates all terms that meet the conditions; n j represents the number of events in the j-th minimum cut set; 2 nj-1 : represents the power operation with base 2 and exponent n j -1, which is used to calculate the denominator of each term;

[0036] Calculate the occurrence probabilities of the basic events in combination with historical data and deduce the probability of the top event.

[0037] Furthermore, after the step of deducing the probability of the top event, it also includes:

[0038] Obtain the ranking of key risk factors from a preset database;

[0039] Generate a safety checklist according to the probability of the top event and the ranking of key risk factors;

[0040] Push the safety checklist to the user's intelligent terminal.

[0041] Furthermore, select the TOPSIS method to calculate the structural risk closeness of the basic events, specifically including:

[0042] Establish a matrix based on the structural importance and risk probability of the basic events and perform positive, standard, and normalization processing;

[0043] Select the entropy weight method to calculate the evaluation index weights to obtain a weighted normalized index matrix;

[0044] Calculate the structural risk closeness of the basic events, sort according to the closeness size, and judge the key degree of each basic event to the top event.

[0045] Furthermore, in the step of establishing a matrix based on the structural importance and risk probability of the basic events and performing positive, standard, and normalization processing, it also includes:

[0046] First, establish an initial matrix A=(b according to the number of items m and the number of indicators nij ) m×n , since the structural importance I and the risk probability index have been made positive, the Min-Max normalization method is then used to normalize the initial matrix A = (b ij ) m×n for the indicators, and the formula is as follows:

[0047] y = (b - A min ) / (A max - A min )

[0048] where y is the normalized value; b is the original value of a certain attribute; A min is the minimum value of a certain attribute; A max is the maximum value of a certain attribute;

[0049] After normalization, the indicator matrix B = (x ij ) m×n is obtained, and the normalization calculation formula is as follows:

[0050]

[0051] where x ij is the normalized value.

[0052] Furthermore, after the step of obtaining the indicator matrix B = (x ij ) m×n , it further includes:

[0053] Based on the obtained indicator matrix B, the weights Wj of each indicator are calculated to obtain the weighted normalized indicator matrix C. The expression of the weighted normalized indicator matrix C is as follows:

[0054]

[0055] The entropy weight method is selected to calculate the indicator weights to ensure the scientificity and rationality of the indicator weights. The specific calculation formula is as follows:

[0056]

[0057] where W j is the entropy weight of the calculated indicator; H j is the entropy value of the calculated indicator; k is the Boltzmann constant; x ij is the i-th normalized and standardized value under the j-th indicator;

[0058] When calculating the closeness, the maximum value of a certain indicator is taken as the positive ideal point v j + , and the minimum value is taken as the negative ideal point v j -; From this, calculate the relative distance S between a certain item and all positive and negative ideal points i + and S i - , and then calculate the closeness S i . The calculation formula is as follows:

[0059]

[0060] In the formula

[0061]

[0062] wherein, v ij is the value after weighted standardization of the j-th parameter of the i-th item; S i is the closeness, and the greater the closeness, the closer it is to the ideal state; use S i as the benchmark to evaluate the key degree of the basic event to the top event.

[0063] Second, the present application provides a gas system management system based on data analysis, including:

[0064] A model architecture design module, which is used to combine the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture; the wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture the non-linear relationships and complex patterns in the data and improve the accuracy and generalization ability of prediction;

[0065] A decision engine design module, which is used to design an adaptive pigging decision engine based on the improved particle swarm optimization algorithm (PSO-ADE);

[0066] An initialization module, which is used to initialize the algorithm and generate an initial particle swarm;

[0067] A running module, which is used to run the algorithm, evaluate and update the particles according to the objective function;

[0068] An iterative optimization module, which is used to find the optimal pigging strategy through iterative optimization;

[0069] A feedback data collection module, which is used to collect feedback data, further adjust and optimize the algorithm to improve the accuracy and effectiveness of decision-making.

[0070] Third, the present application provides an intelligent terminal, including a memory and a processor, and a computer program capable of being loaded and executed by the processor for the above-mentioned gas system management method based on data analysis is stored on the memory.

[0071] Fourthly, the present application provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor to implement the above-mentioned gas system management method based on data analysis.

[0072] In summary, compared with the prior art, the beneficial effects of the above technical solutions are as follows:

[0073] The gas system management method based on data analysis described in the present application combines a wax deposition kinetics model and a deep neural network, and can more accurately predict the wax deposition situation in gas pipelines. The wax deposition kinetics model provides a prediction basis based on physical and chemical processes, while the deep neural network can capture the non-linear relationships and complex patterns in the data, thereby improving the accuracy and generalization ability of the prediction.

[0074] Based on the improved particle swarm optimization algorithm (PSO-ADE), this solution designs an adaptive pigging decision-making engine. This engine can find the optimal pigging strategy through iterative optimization according to real-time data and prediction results, which helps to reduce unnecessary pigging operations, improve pigging efficiency, and reduce operating costs. By collecting feedback data, the algorithm can be further adjusted and optimized, which means that the algorithm can adapt to different pipeline conditions, operating environments, and wax deposition characteristics, thereby improving the accuracy and effectiveness of decision-making. Description of the Drawings

[0075] Figure 1 is a schematic flowchart of a gas system management method based on data analysis according to an embodiment of the present application. Detailed Embodiments

[0076] The following further describes the present application in detail with reference to all the drawings.

[0077] An embodiment of the present application discloses a gas system management method and system based on data analysis. Referring to Figure 1 a gas system management method based on data analysis includes:

[0078] S101. Combine a wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture.

[0079] Specifically, the wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture non-linear relationships and complex patterns in the data and improve the accuracy and generalization ability of the prediction. For the hybrid architecture that combines the wax deposition kinetics model (physical mechanism) and the deep neural network (data-driven), two-way coupling rather than simple superposition of the two needs to be realized.

[0080] Take the wax deposition kinetic equations (such as the molecular diffusion coefficient formula and the shear stripping effect formula in the Matzain model) as the physical constraint terms of the neural network, and implement hard constraints through residual connections or loss function design. For example, add a loss term based on Fick's diffusion law to the output layer of the neural network to ensure that the prediction results conform to the diffusion mechanism.

[0081] Then, adopt a parallel - series hybrid structure to dynamically and weightedly fuse the outputs of the kinetic model (such as wax deposition rate, temperature gradient) and the original sensor data (flow rate, pressure) through the attention mechanism, introduce multi-scale feature extraction, use a convolutional neural network (CNN) to process the pipeline cross-section temperature field image, and use a recurrent neural network (RNN) to capture the temporal dynamics, and splice them with the scalar features output by the kinetic model.

[0082] Construct a transfer learning framework for different oil product characteristics (such as wax content, asphaltene ratio), eliminate the influence of oil product differences on model generalization through domain - adversarial training, deploy an online learning module, and dynamically update the model parameters using the measured wax layer thickness data after pigging operations to form a closed - loop optimization system.

[0083] S102. Design an adaptive pigging decision engine based on the improved particle swarm optimization algorithm (PSO - ADE).

[0084] Specifically, the particle swarm optimization algorithm (PSO) is an optimization algorithm based on swarm intelligence, which simulates the foraging behavior of bird flocks and searches for the optimal solution through the cooperation and competition of individuals and groups. To improve the performance of PSO, an adaptive dynamic adjustment mechanism (ADE) is introduced to dynamically adjust parameters such as the inertia weight and learning factor to balance the global search ability and local search accuracy, avoid premature convergence and falling into local optima. The adaptive pigging decision engine takes PSO - ADE as the core, dynamically adjusts the pigging strategy by combining pipeline operation data and sediment monitoring data, and realizes the automation, intelligence, and high efficiency of pigging operations.

[0085] In this embodiment, the input of the decision-making engine is pipeline operation data (such as flow rate, pressure, temperature, etc.), sediment monitoring data (such as sediment thickness, composition, etc.), and pigging operation history records (such as pigging frequency, cost, effect, etc.); the output of the decision-making engine is pigging operation suggestions (such as whether to perform pigging, pigging method, pigging time, etc.). The objective function is to minimize the pigging cost and pipeline operation risk, and factors such as pigging cost, sediment risk, and operation complexity need to be considered; the constraints are the safety, effectiveness, and feasibility of pigging operations. For example, the pigging frequency cannot be too high to avoid affecting pipeline operation, and the pigging effect should meet the requirements. Each particle represents a pigging strategy, including parameters such as pigging frequency, pigging method, and pigging time. The management system calculates the fitness of each particle according to the objective function, evaluates the pros and cons of the pigging strategy, and uses the update rules of PSO-ADE to dynamically adjust the speed and position of the particles to find the optimal pigging strategy, and terminates when the maximum number of iterations is reached or the fitness meets the requirements.

[0086] Specifically, the management system collects transmission information from the pipeline operation system and the sediment monitoring system, including flow rate, pressure, temperature, sediment thickness, and composition data, removes outliers, fills in missing values, ensures the accuracy and integrity of the data, and normalizes the data to a certain range to improve the convergence speed and stability of the algorithm. Initialize the particle swarm, determine parameters such as the number of particles and dimensions, initialize the positions and speeds of the particles, then set the inertia weight and learning factors, and dynamically adjust the inertia weight and learning factors according to the ADE mechanism to balance the global search and local search capabilities. Combine the actual situation to define the objective function and constraints to ensure the feasibility and effectiveness of the algorithm.

[0087] Implement the PSO-ADE algorithm using programming languages (such as Python, MATLAB, etc.), including functions such as particle update, fitness calculation, and optimal solution selection. Test the algorithm on simulated data or historical data, adjust the parameters and optimize the algorithm according to the results to improve the performance and effect. Evaluate the performance of the decision-making engine according to the actual pigging effect, such as whether the pigging cost is reduced and whether the pipeline operation risk is reduced, and feedback the evaluation results into the algorithm to further optimize and improve the decision-making engine to form a closed-loop optimization.

[0088] S103. Initialize the algorithm to generate an initial particle swarm.

[0089] Specifically, the management system sets the maximum number of iterations or termination conditions of the algorithm to ensure that the algorithm converges within a reasonable time, and randomly assigns initial positions and speeds to each particle. The positions can be generated uniformly distributed between the upper and lower bounds of the solution space, and the speeds can be set to zero or small random numbers. For example, determine the number of particle swarms N, N = 50, and set the maximum number of iterations T of the algorithm, T = 100.

[0090] S104. Run the algorithm to evaluate and update the particles according to the objective function.

[0091] Specifically, the management system defines the objective function, which reflects the optimization objectives of the pigging strategy, such as minimizing the pigging cost, maximizing the pipeline flow capacity, etc. Then, according to the actual requirements and pipeline operating conditions, appropriate weights and constraint conditions are set for the objective function. Design a particle evaluation mechanism to evaluate the fitness of the particles according to the objective function, and adopt update strategies such as inertia weight, learning factor, etc. to update the position and velocity of the particles to guide the particles to move towards a better solution.

[0092] According to the actual requirements, define the objective function f(x), where x represents the position of the particle (i.e., the parameters of the pigging strategy). For example, the objective function can be the weighted sum of the pigging cost and the pipeline flow capacity, and the weights are set according to the actual requirements. In each iteration, the management system calculates the fitness value of each particle, that is, the value of the objective function f(x), and sorts the particles according to the fitness value to find the current optimal particle (i.e., the particle with the smallest fitness value) and the global optimal particle (i.e., the particle with the smallest fitness value in the historical iterations).

[0093] The management system updates the velocity and position of each particle according to parameters such as inertia weight and learning factor. Among them, the inertia weight can control the exploration and development ability of the particles, and the learning factor can affect the degree of learning of the particles towards the individual optimal and global optimal. Until the maximum number of iterations T is reached or other termination conditions are met (such as the fitness value converges). After each iteration, record the positions and their fitness values of the current optimal particle and the global optimal particle.

[0094] S105. Through iterative optimization, find the optimal pigging strategy.

[0095] Specifically, the management system outputs the position of the global optimal particle, that is, the parameters of the optimal pigging strategy, and outputs the fitness value of the global optimal particle, that is, the value of the objective function corresponding to the optimal pigging strategy. According to the changes in the actual requirements and pipeline operating conditions, adjust and optimize the algorithm parameters (such as the number of particle swarms, the maximum number of iterations, inertia weight, learning factor, etc.), and then effectively initialize the algorithm, generate the initial particle swarm, and run the algorithm to find the optimal pigging strategy.

[0096] S106. Collect feedback data to further adjust and optimize the algorithm.

[0097] Specifically, after the pigging strategy is implemented, the management system collects pipeline operation data, wax deposition monitoring data, and pigging operation data in real-time or regularly. Analyze the performance and existing problems of the current algorithm, such as slow convergence speed and insufficient search ability. According to the analysis results, adjust the algorithm parameters, such as increasing the number of particle swarms, increasing the number of iterations, adjusting the learning factor and inertia weight, etc., and revise and improve the objective function to more accurately reflect the optimization goal of the pigging strategy. Rerun the algorithm for a new round of iterative optimization. Furthermore, effectively find the optimal pigging strategy through iterative optimization, and collect feedback data to further adjust and optimize the algorithm to improve the accuracy and effectiveness of decision-making. At the same time, through continuous monitoring and optimization, we can ensure that the algorithm always adapts to the needs of actual gas system management.

[0098] In another embodiment, S101 specifically includes the following sub-steps:

[0099] Build a thermal-fluid-solidification four-field coupling model based on COMSOL Multiphysics, and the solved equations include:

[0100]

[0101] Among them, ρ is the fluid density, used to represent the mass of the fluid per unit volume; u is the velocity vector, used to represent the motion velocity of the fluid microelement; t represents time; p is the pressure, used to represent the force acting perpendicularly on the unit area inside the fluid; μ is the dynamic viscosity coefficient, used to represent the magnitude of the internal friction force of the fluid, reflecting the viscosity of the fluid; S wax is the wax phase change source term, and an improved molecular diffusion-gel deposition dual-mechanism model is adopted to represent the influence of the phase change of wax in the fluid on the flow;

[0102] Introduce dynamic grid adaptive technology to automatically refine the grid to a resolution of 0.5 mm in the region where the deposition rate changes suddenly;

[0103] Train the XGBoost-GRU hybrid model, and the input features include temperature gradient (ΔT), wall shear stress (τ_w), and wax crystal volume fraction (φ), and the output is the deposition rate correction factor α;

[0104] Use a generative adversarial network (GAN) to augment the limited experimental data and generate a synthetic data set covering the temperature range of -10 to 50 °C to solve the problem of insufficient extrapolation ability of traditional models.

[0105] Specifically, the management system creates a model through COMSOL Multiphysics software, defines the geometry and dimensions according to the actual problem, and ensures that the geometric part accurately describes the simulation area. Select and set the physical parameters related to heat conduction, fluid flow, solid mechanics, and chemical phase change. The specific content is as follows:

[0106] Heat conduction module: Set parameters such as thermal conductivity and heat capacity.

[0107] Fluid flow module: Set parameters such as fluid density ρ, velocity vector u, and dynamic viscosity coefficient μ.

[0108] Solid mechanics module: Set material properties such as elastic modulus and Poisson's ratio.

[0109] Chemical phase change module: Set the wax phase change source term S wax , and adopt an improved molecular diffusion-gel deposition dual-mechanism model.

[0110] Define the physical fields related to wax phase change, such as phase field variables and chemical potential.

[0111] The management system performs mesh generation on the geometric model to ensure that the mesh quality meets the simulation requirements. Introduce dynamic mesh adaptation technology to automatically refine the mesh to a resolution of 0.5 mm in the area where the deposition rate changes abruptly, in order to capture the detailed changes during the wax deposition process. Apply boundary conditions and initial conditions, including boundary conditions such as temperature, pressure, and velocity, as well as the initial conditions of wax phase change, to ensure that the settings of boundary conditions and initial conditions can accurately reflect the physical background of the actual problem.

[0112] Collect experimental data, including features such as temperature gradient (ΔT), wall shear stress (τ_w), and wax crystal volume fraction etc. Preprocess the data, such as normalization and missing value processing. Build an XGBoost-GRU hybrid model and input the feature data for training to output the deposition rate correction factor α, which is used to correct the deposition rate in the simulation results. Moreover, adopt a generative adversarial network (GAN) to augment the experimental data and generate a synthetic dataset covering the temperature range of -10 to 50 °C. Combine the synthetic dataset with the experimental data to improve the generalization ability and extrapolation ability of the model. Analyze the simulation results, including the changes in the temperature field, flow field, solid mechanics field, and chemical phase change field, verify the accuracy of the simulation results, such as comparing with experimental results and performing mesh independence verification, etc. Optimize and adjust the model according to the simulation results and the output of the machine learning model.

[0113] By introducing dynamic grid adaptation technology, the grid is automatically refined in the area of sudden change in deposition rate, improving the accuracy and resolution of the simulation results. By training a machine learning model and applying a deposition rate correction factor, the accuracy of the simulation results is further improved. The adversarial generative network (GAN) is used to augment experimental data, generating a synthetic dataset covering a wider temperature range, solving the problem of insufficient extrapolation ability of traditional models and enabling the model to better adapt to the simulation of wax deposition processes under different temperatures and conditions. Based on the simulation results and the output of the machine learning model, the thermal-fluid-solidification four-field coupling system is optimized and designed to provide decision-making support and solutions for wax deposition problems in industrial processes, such as optimizing operating parameters and improving equipment structure.

[0114] In another embodiment, S103 specifically includes the following sub-steps:

[0115] Establish an objective function:

[0116]

[0117] Where C pigging is the cost of the pigging operation, representing the cost of each pigging operation; f i is the frequency of the i-th pigging, representing the number of pigging operations within a certain period of time; λ is the risk penalty coefficient, representing the degree of penalty for risk and reflecting the cost of risk; δ i is the predicted deposition thickness, representing the thickness of the sediment in the pipeline predicted during the i-th pigging operation; δ safe is the safe deposition thickness, representing the safe thickness of the sediment in the pipeline, and exceeding this thickness may pose risks;

[0118] The purpose of the objective function is to find a balance between the cost of the pigging operation and the risk brought by the sediment, minimizing the total cost;

[0119] An improved PSO-ADE algorithm is adopted, introducing an inertia weight adaptation mechanism to enhance the local search ability in the later stage of iteration. The formula is as follows:

[0120] w(t) = w min + (w max - w min ) · e -20t / T

[0121] The formula is used to describe the change of w(t) with time t. The specific parameter meanings are as follows: w min represents the minimum or stable value of w(t). When time t approaches infinity, w(t) will approach this value; w max represents the initial maximum value of w(t). When time t = 0, the value of w(t) is wmax ; t represents time; T represents the time constant, which has the same unit as time t and reflects the speed at which w(t) decays to the stable value. The larger the time constant T, the slower the decay process.

[0122] Specifically, the management system determines the form of the objective function, which should include two parts: the cost of pigging operation and the risk cost. The cost of pigging operation is represented by the product of the cost per pigging and the pigging frequency. The risk cost is represented by the functional relationship between the risk penalty coefficient and the difference between the predicted deposition thickness and the safe deposition thickness (if the difference is positive, it means the sediment thickness exceeds the safe value). The goal is to minimize the total cost (pigging operation cost + risk cost). The management system selects the improved PSO - ADE algorithm for optimization, sets the algorithm parameters, and adjusts the size of the particle swarm, the number of iterations, etc. according to the problem scale and complexity. The objective function is transformed into a fitness function to evaluate the quality of particles. The smaller the fitness function value, the better the solution corresponding to the particle.

[0123] The management system initializes the particle swarm, including the position and velocity of particles, evaluates the quality of particles according to the fitness function, updates the position and velocity of particles, introduces an inertia weight adaptive mechanism to enhance the local search ability, and iteratively executes the above steps until the preset number of iterations is reached or the convergence condition is satisfied. The management system analyzes the optimization results, including the optimal solution, sub - optimal solution, etc., and verifies the rationality and effectiveness of the optimization results to ensure that the actual requirements are met.

[0124] By optimizing the pigging frequency and sediment management strategy, the total cost is reduced, the pigging operation cost and risk cost are balanced, and the economic benefits are maximized. Predict the sediment thickness, take pigging measures in a timely manner to avoid the sediment thickness exceeding the safe value, and reduce the risks such as pipeline blockage and leakage caused by excessive sediment. Provide a scientific decision - making basis to help managers formulate reasonable pigging plans and risk management strategies, improve the decision - making efficiency and accuracy, and reduce the decision - making risk. Introduce an inertia weight adaptive mechanism to enhance the local search ability of the algorithm, improve the convergence speed and solution quality of the algorithm, and be applicable to the solution of complex optimization problems.

[0125] Furthermore, it also includes:

[0126] Select typical gas system accidents as the analysis object and clarify the severity of accident consequences;

[0127] Take the typical gas system accident as the top - event and decompose the top - event into multiple sub - top - events, where the sub - top - events are the possible causes of the top - event;

[0128] Construct an accident tree logic model based on the secondary top event and the top event; use "AND gates" (all conditions are satisfied simultaneously) and "OR gates" (any one condition is satisfied) to decompose the causes layer by layer until the basic events corresponding to the secondary top event are obtained;

[0129] Calculate the minimum cut sets and the structure importance coefficients. The minimum cut sets are the combinations of the fewest basic events that lead to the top event and are used to identify key risk factors;

[0130] Structure importance coefficient

[0131] Among them, I i represents the value of the i-th index, which is a comprehensive value obtained by summation; ∑ represents the summation operation, which accumulates all terms that meet the conditions; n j represents the number of events in the j-th minimum cut set; 2 nj-1 : represents the power operation with base 2 and exponent n j -1, which is used to calculate the denominator of each term;

[0132] Combine historical data to calculate the occurrence probabilities of basic events and deduce the probability of the top event.

[0133] Specifically, the management system selects typical accidents of the gas system as the analysis object, such as liquefied petroleum gas (LPG) or natural gas (NG) leakage and explosion accidents, and clarifies the severity of the accident consequences, including casualties, property losses, environmental impacts, etc. Take the typical accidents of the gas system as the top event, such as "gas system explosion accident", and decompose the top event into multiple secondary top events, that is, the possible causes leading to the top event, such as "equipment failure", "human operation error", "environmental factors", etc.

[0134] Based on the secondary top event and the top event, construct an accident tree logic model, use "AND gates" (all conditions are satisfied simultaneously) and "OR gates" (any one condition is satisfied) to connect the events to represent the logical relationship between the events, and then continue to use logical reasoning to decompose the secondary top event into more specific causes until the basic events are obtained. Among them, the basic events are the most direct causes leading to the secondary top event, usually specific factors such as equipment failures, human errors, and environmental factors.

[0135] The management system simplifies the fault tree using Boolean algebra to find the minimal cut sets. The minimal cut sets are the combinations of the fewest basic events that lead to the occurrence of the top event, which are used to identify key risk factors and calculate the structural importance coefficients to evaluate the impact degree of each basic event on the occurrence of the top event. The structural importance coefficients can be obtained through summation calculations, which involve the number of events in the minimal cut sets and the corresponding power operations. Collect relevant historical data, including the occurrence frequencies of basic events, accident records, etc. Calculate the occurrence probabilities of basic events based on the historical data, and then deduce the occurrence probability of the top event, which can be specifically achieved through methods such as statistical frequency, trend analysis, time series analysis, etc.

[0136] By constructing a fault tree logic model, the key factors and paths leading to typical accidents in the gas system can be clearly identified. The identification of minimal cut sets helps to determine key risk factors, so as to take targeted preventive measures to reduce the possibility of accidents. The calculation of structural importance coefficients provides decision-makers with information on which basic events have the greatest impact on the occurrence of the top event, which helps decision-makers prioritize which factors in order to achieve the greatest risk reduction effect with limited resources. Understanding the occurrence probabilities of basic events and the occurrence probability of the top event helps to formulate more effective emergency plans and response measures, which can ensure a rapid and effective response in case of an accident, thus reducing the losses caused by the accident.

[0137] Furthermore, after the steps of deducing the probability of the top event, as another implementation manner, the embodiments of the present application may further include the following steps:

[0138] S201. Obtain the sorting of key risk factors from a preset database.

[0139] Specifically, the management system retrieves the key risk factors and their sorting from a preset database. The sorting of key risk factors is pre-set and generated by the user in combination with factors such as historical accident analysis, expert evaluation, and industry standards, ensuring that the risk factor sorting is up-to-date and matches the specific situation of the current gas system.

[0140] S202. Generate a safety checklist according to the probability of the top event and the sorting of key risk factors.

[0141] Specifically, the management system combines the probability of the top event derived with the ranking of key risk factors obtained from the database, associates the probability values with the ranking of risk factors, so that when generating a safety checklist, it can clearly reflect which factors have a higher risk level. According to the integrated information, a safety checklist is designed, which includes information such as key risk factors, corresponding risk levels (based on probability and ranking), recommended inspection frequencies, inspection methods, preventive measures, etc. Ensure that the checklist is easy to understand and convenient for users to execute in actual operations. By combining the ranking of key risk factors with the probability of the top event, users can more clearly understand which factors need to be prioritized for attention and management. The generation and push of the safety checklist make the risk management process more systematic and standardized, improving management efficiency.

[0142] S203. Push the safety checklist to the intelligent terminal of the user.

[0143] Specifically, the management system pushes the generated safety checklist to the intelligent terminal of the user through relevant channels (such as email, text message, mobile application, etc.), ensures the safety and reliability of the push process, and considers user privacy protection, providing necessary instructions and guidance to help users understand and use the safety checklist.

[0144] In this embodiment, the TOPSIS method is selected to calculate the structural risk closeness degree of basic events, which specifically includes the following steps:

[0145] S301. Establish a matrix and perform positive, standard, and normalized processing.

[0146] Specifically, the management system establishes a matrix and performs positive, standard, and normalized processing according to the structural importance degree and risk probability of basic events; collects data such as the structural importance degree and risk probability of basic events, cleans and preprocesses the data to ensure the accuracy and consistency of the data. According to the structural importance degree and risk probability of basic events, an initial matrix is established, and the matrix is positively processed to uniformly convert all index types into extremely large type indexes (if there are extremely small type, intermediate type, or interval type indexes, corresponding conversions are required). The positively processed matrix is standardized to eliminate the influence of the dimension of each index, and the standardized matrix is normalized to obtain a normalized matrix.

[0147] S302. Select the entropy weight method to calculate the evaluation index weights and obtain a weighted normalized index matrix.

[0148] Specifically, the management system calculates the information entropy of each index according to the normalized matrix, calculates the difference coefficient according to the information entropy to reflect the dispersion degree of the data of each index, obtains the weights of each index by normalizing the difference coefficient, and multiplies the normalized matrix by the weight vector to obtain a weighted normalized index matrix.

[0149] S303. Calculate the structural risk proximity degree of basic events.

[0150] Specifically, the management system calculates the structural risk proximity degree of basic events, sorts them according to the proximity degree, and judges the critical degree of each basic event to the top event. In the weighted normalized index matrix, the optimal solution (the maximum value of each index) and the worst solution (the minimum value of each index) are determined, the Euclidean distances between each basic event and the optimal solution and the worst solution are calculated, and the relative proximity degree (i.e., the structural risk proximity degree) of each basic event is calculated according to the Euclidean distance. The basic events are sorted according to the size of the structural risk proximity degree, and the critical degree of each basic event to the top event is judged. The greater the proximity degree, the greater the impact of the basic event on the top event.

[0151] In another embodiment, S301 specifically includes the following sub-steps:

[0152] Furthermore, in the step of establishing a matrix and performing positive, standard, and normalized processing according to the structural importance and risk probability of basic events, it also includes:

[0153] First, an initial matrix A=(b ij ) m×n is established according to the number of projects m and the number of indicators n. Since the structural importance I and the risk probability index are already positive, the Min-Max standardization method is then used to standardize the initial matrix A=(b ij ) m×n , and the formula is as follows:

[0154] y=(b - A min ) / (A max - A min )

[0155] where y is the standardized value; b is the original value of a certain attribute; A min is the minimum value of a certain attribute; A max is the maximum value of a certain attribute;

[0156] After normalization, the index matrix B=(x ij ) m×n is obtained, and the normalization calculation formula is as follows:

[0157]

[0158] where x ij is the normalized value.

[0159] Specifically, the management system constructs an m×n initial matrix A=(b ij ) m×n according to the number of projects m and the number of indicators n. Among them, bij Denote the original value of the \(i\)-th item on the \(j\)-th index, calculate the minimum and maximum values of each index. Use the Min-Max normalization formula to normalize each index, and fill the normalized values into a new matrix to obtain a normalized matrix. Based on the normalized matrix, perform normalization processing. The purpose of normalization processing is to convert each element in the matrix into a value within the range of \([0, 1]\) to eliminate the dimensional differences between different indexes.

[0160] It should be noted that the specific formula for normalization processing needs to be adjusted according to the actual situation. For example, if it is desired to convert each element in the matrix into a unit vector (i.e., the norm of the vector is 1), then the L2 normalization method can be adopted. This embodiment adopts a method similar to Min-Max normalization, that is, maintaining the proportional relationship between elements and mapping them to the range of \([0, 1]\).

[0161] After Min-Max normalization processing, each element in the matrix is converted into a proportional value between the minimum and maximum values of its corresponding index. This helps to eliminate the dimensional differences between different indexes, making subsequent analysis and comparison more accurate and fair. The normalized matrix has better comparability and interpretability, providing strong support for subsequent data analysis and decision-making.

[0162] Normalization processing further converts each element in the matrix into a value within the range of \([0, 1]\). This helps to simplify the subsequent calculation and analysis process. The normalized matrix is more intuitive and easy to understand, and can clearly show the performance of each item on different indexes. Normalization processing also helps to improve the stability and accuracy of the model.

[0163] Further, after the step of obtaining the index matrix \(B=(x ij ) m×n As another implementation manner, the embodiment of the present application may further include the following steps:

[0164] Based on the obtained index matrix \(B\), calculate the weights \(W_j\) of each index to obtain a weighted normalized index matrix \(C\). The expression of the weighted normalized index matrix \(C\) is as follows:

[0165]

[0166] Select the entropy weight method to calculate the index weights to ensure the scientificity and rationality of the index weights. The specific calculation formula is as follows:

[0167]

[0168] where \(W j is the entropy weight of the calculated index; \(Hj is for calculating the entropy value of the index; k is the Boltzmann constant; x ij is the i-th normalized and standardized value under the j-th index;

[0169] When calculating the closeness degree, take the maximum value of a certain index as the positive ideal point v j + , and the minimum value as the negative ideal point v j - ; Thus, calculate the relative distances S i + and S i - of a certain item from all positive and negative ideal points, and then calculate the closeness degree S i . The calculation formula is as follows:

[0170]

[0171] In the formula

[0172]

[0173] where, v ij is the value of the j-th parameter of the i-th item after weighted standardization; S i is the closeness degree, and the larger the closeness degree, the closer it is to the ideal state; use S i as the benchmark to evaluate the critical degree of the basic event to the top event.

[0174] Specifically, after normalization, an index matrix is obtained, and the entropy weight method is used to calculate the entropy value of each index. Among them, p ij =x ij / ∑x ij , representing the probability that the i-th item (or basic event) appears under the j-th index (i.e., the proportion of the normalized value in all normalized values); k is a proportionality coefficient of the Boltzmann constant, used to adjust the range of the entropy value. Usually, k = 1 / (lnm), where m is the number of items (or basic events), and the value of k here can be adjusted according to the actual situation.

[0175] Calculate the entropy weight of each index according to the entropy value H j . The entropy weight W j reflects the importance degree of each index in the evaluation. The smaller the entropy value, the more certain the information, and the greater the weight of this index in the evaluation. According to the entropy weight W j of each index and the index matrix B, construct a weighted normalized index matrix, which reflects the comprehensive performance of each item (or basic event) after considering the index weight.

[0176] Determine the positive ideal point and the negative ideal point. The positive ideal point takes the maximum value of each index, and the negative ideal point takes the minimum value of each index. Calculate the relative distances of a certain project (or basic event) from all positive and negative ideal points. The relative distances can be calculated using distance metrics such as Euclidean distance and Manhattan distance. Calculate the closeness degree Si according to the relative distances. The larger Si is, the closer the project (or basic event) is to the ideal state.

[0177] Calculate the index weights through the entropy weight method, which avoids the arbitrariness of subjective weight assignment, improves the scientificity and rationality of the weights. The entropy weight method can objectively reflect the information uncertainty of each index, thereby determining the weights and making the evaluation results more objective and accurate. The construction of the weighted normalized index matrix and the calculation of the closeness degree are both based on objective data, improving the accuracy and objectivity of the evaluation. The calculation of the closeness degree considers the relative distances of the project (or basic event) from the positive and negative ideal points, and can more accurately reflect the actual situation of the project and its closeness to the ideal state. The technical solution has clear steps, is easy to understand, and is convenient to operate and apply in actual evaluations. By calculating the closeness degree, the criticality of the basic event to the top event can be intuitively evaluated, providing strong support for decision-making. At the same time, the technical solution can be applied to the evaluations of different fields and different projects, and has strong generality and practicability.

[0178] Based on the above method, the embodiments of the present application also disclose a gas system management system based on data analysis. A gas system management system based on data analysis includes:

[0179] A model architecture design module, which is used to combine the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture; the wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture the non-linear relationships and complex patterns in the data and improve the accuracy and generalization ability of the prediction;

[0180] A decision engine design module, which is used to design an adaptive pigging decision engine based on the improved particle swarm optimization algorithm (PSO-ADE);

[0181] An initialization module, which is used to initialize the algorithm and generate an initial particle swarm;

[0182] An operation module, which is used to run the algorithm, evaluate and update the particles according to the objective function;

[0183] An iterative optimization module, which is used to find the optimal pigging strategy through iterative optimization;

[0184] A feedback data collection module, which is used to collect feedback data and further adjust and optimize the algorithm to improve the accuracy and effectiveness of the decision-making.

[0185] The embodiment of the present application also discloses an intelligent terminal, which includes a memory and a processor. Among them, a computer program capable of being loaded and executed by the processor, such as a gas system management method based on data analysis as described above, is stored on the memory.

[0186] The embodiment of the present application also discloses a computer-readable storage medium. A computer program capable of being loaded and executed by the processor, such as a gas system management method based on data analysis as described above, is stored in the computer-readable storage medium. The computer-readable storage medium includes, for example: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0187] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting the protection scope of the invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all embodiments. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art can still, without conflict and without creative efforts, combine, add, delete, or make other adjustments to the features in the various embodiments of the present invention according to the situation, so as to obtain different technical solutions that essentially do not deviate from the concept of the present invention. These technical solutions also belong to the scope of protection of the present invention.

Claims

1. A gas system management method based on data analysis, characterized in that, It includes the following steps: Combine the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture; the wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture the non-linear relationships and complex patterns in the data to improve the accuracy and generalization ability of prediction; Design an adaptive pigging decision-making engine based on the improved particle swarm optimization algorithm (PSO-ADE); Initialize the algorithm to generate an initial particle swarm; Run the algorithm to evaluate and update the particles according to the objective function; Find the optimal pigging strategy through iterative optimization.

2. The gas system management method based on data analysis according to claim 1, wherein, Among the steps of combining the wax deposition kinetics model with a deep neural network to design a hybrid prediction model architecture, it also includes: Construct a thermal-fluid-solidification four-field coupling model based on COMSOL Multiphysics. The solved equations include: Among them, ρ is the fluid density, which is used to represent the mass of the fluid per unit volume; u is the velocity vector, which is used to represent the movement speed of the fluid particles; t represents time; p is the pressure, which is used to represent the force acting vertically on the unit area inside the fluid; μ is the dynamic viscosity coefficient, which is used to represent the internal friction of the fluid and reflects the viscosity of the fluid; S wax As the wax phase change source term, an improved molecular diffusion-gel deposition dual mechanism model is used to represent the effect of wax phase change in fluid on flow. Introduce the dynamic mesh adaptation technology to automatically refine the mesh to a resolution of 0.5 mm in the area where the deposition rate mutates; Train the XGBoost-GRU hybrid model. The input features include the temperature gradient (ΔT), wall shear stress (τ_w), and wax crystal volume fraction (φ), and the output is the deposition rate correction factor α; Use the generative adversarial network (GAN) to augment the limited experimental data to generate a synthetic data set covering the temperature range of -10 to 50 °C to solve the problem of insufficient extrapolation ability of traditional models.

3. A gas system management method based on data analysis according to claim 1, characterized in that, Among the steps of initializing the algorithm, it also includes: Establish an objective function: Among them, C pigging is the cost of pigging operation, representing the cost of each pigging operation; f i is the frequency of the i-th pigging, representing the number of pigging operations within a certain period of time; λ is the risk penalty coefficient, representing the degree of penalty for risk and reflecting the cost of risk; δ i is the predicted deposition thickness, representing the thickness of the sediment in the pipeline predicted during the i-th pigging operation; δ safe is the safe deposition thickness, representing the safe thickness of the sediment in the pipeline. Exceeding this thickness may pose risks; The purpose of the objective function is to find a balance between the cost of pigging operations and the risks brought by sediments to minimize the total cost; Adopt the improved PSO-ADE algorithm and introduce an inertia weight adaptation mechanism to enhance the local search ability in the later stage of iteration. The formula is as follows: w(t) = w min +(w max -w min )·e -20t / T The formula is used to describe the variation of w(t) with time t. The specific meanings of the parameters are as follows: w min represents the minimum value or stable value of w(t). As time t approaches infinity, w(t) approaches this value; w max represents the initial maximum value of w(t). When time t = 0, the value of w(t) is w max ; t represents time; T represents the time constant, which has the same unit as time t and reflects the speed at which w(t) decays to the stable value. The larger the time constant T, the slower the decay process.

4. The management method of a gas system based on data analysis according to claim 1, characterized in that, It also includes: Select typical accidents in the gas system as the analysis object and clarify the severity of the accident consequences; Take the typical accident in the gas system as the top event and decompose the top event into multiple sub-top events. The sub-top events are the possible causes of the top event; According to the sub-top events and the top event, construct an accident tree logic model; use "AND gates" (all conditions are met simultaneously) and "OR gates" (any one condition is met) to decompose the causes layer by layer until the basic events corresponding to the sub-top events are obtained; Calculate the minimum cut sets and structure importance coefficients. The minimum cut sets are the combinations of the fewest basic events that lead to the top event and are used to identify key risk factors; Structure importance coefficient Among them, I i represents the value of the i-th index, which is a comprehensive value obtained through summation calculation; ∑ represents the summation operation, accumulating all terms that meet the conditions; n j represents the number of events in the j-th minimal cut set; 2 nj-1 : represents the power operation with base 2 and exponent n j -1, used to calculate the denominator of each term; Calculate the occurrence probability of the basic events based on historical data and deduce the probability of the top event.

5. A gas system management method based on data analysis according to claim 4, characterized in that, After the step of deducing the probability of the top event, it also includes: Obtain the ranking of key risk factors from a preset database; Generate a safety checklist according to the probability of the top event and the ranking of key risk factors; Push the safety checklist to the user's intelligent terminal.

6. The gas system management method based on data analysis according to claim 4, wherein, Select the TOPSIS method to calculate the structural risk closeness of the basic events, specifically including: Establish a matrix based on the structural importance and risk probability of the basic events and perform positive, standard, and normalization processing; Select the entropy weight method to calculate the evaluation index weights to obtain the weighted normalized index matrix; Calculate the structural risk closeness of basic events, sort them according to the closeness, and judge the criticality of each basic event to the top event.

7. A gas system management method based on data analysis according to claim 6, characterized in that, In the step of establishing a matrix and performing positive, standard, and normalization processing according to the structural importance and risk probability of basic events, it also includes: First, an initial matrix A=(b is established according to the number of items m and the number of indicators n ij ) m×n . Since the structure importance I and the risk probability index itself have been made positive, the Min-Max normalization method is then used to normalize the initial matrix A=(b ij ) m×n . The formula is as follows: y=(b - A min ) / (A max - A min ) Among them, y is the normalized value; b is the original value of a certain attribute; A min is the minimum value of a certain attribute; A max is the maximum value of a certain attribute; After normalization, the index matrix B=(x ij ) m×n , and the normalization calculation formula is as follows: where x ij is the normalized value.

8. A gas system management method based on data analysis according to claim 6, characterized in that, After the step of obtaining the index matrix B=(x ij ) m×n through normalization, the method further includes: Based on the obtained index matrix B, calculate the weight Wj of each index, so as to obtain the weighted normalized index matrix C. The expression of the weighted normalized index matrix C is as follows: Select the entropy weight method to calculate the index weight to ensure the scientificity and rationality of the index weight. The specific calculation formula is as follows: Among them, W j is the entropy weight of the calculation index; H j is the entropy value of the calculation index; k is the Boltzmann constant; x ij is the i-th normalized value under the j-th index; When calculating the closeness degree, take the maximum value of a certain index as the positive ideal point v j + , and the minimum value as the negative ideal point v j - ; Then calculate the relative distances S i + and S i - of a certain item from all positive and negative ideal points, and further calculate the closeness degree S i . The calculation formula is as follows: In the formula Among them, v ij is the value after weighted standardization of the j-th parameter of the i-th project; S i is the closeness degree. The larger the closeness degree, the closer it is to the ideal state; taking S i as the benchmark to evaluate the critical degree of the basic event to the top event.

9. A gas system management system based on data analysis, characterized in that, including: The model architecture design module is used to combine the wax deposition kinetics model with the deep neural network to design the hybrid prediction model architecture; The wax deposition kinetics model is used to describe the physical and chemical processes of wax deposition and provide basic prediction capabilities; the deep neural network is used to capture the non-linear relationships and complex patterns in the data and improve the accuracy and generalization ability of prediction; The decision engine design module is used to design an adaptive pigging decision engine based on the improved particle swarm optimization algorithm (PSO-ADE); The initialization module is used to initialize the algorithm and generate the initial particle swarm; The operation module is used to run the algorithm, evaluate and update the particles according to the objective function; The iterative optimization module is used to find the optimal pigging strategy through iterative optimization; The feedback data collection module is used to collect feedback data, further adjust and optimize the algorithm to improve the accuracy and effectiveness of the decision.

Images