Marine low-pressure carbon dioxide tank pressure control method and system

Abstract

The invention discloses a marine low-pressure carbon dioxide tank pressure control method and system, and relates to the technical field of storage tank pressure control, and the method comprises the steps: obtaining real-time pressure data in a low-pressure carbon dioxide tank, and extracting pressure feature data; analyzing the pressure characteristic data by using a pressure diagnosis algorithm, and identifying key factors influencing the pressure stability in the low-pressure carbon dioxide tank; based on the key factors, a pressure regulation and control model is established, pressure parameters of the low-pressure carbon dioxide tank are adjusted in real time, and a control strategy is optimized. According to the method, key factors are analyzed by using a pressure diagnosis algorithm, so that accurate pressure regulation is realized, the stability of the internal pressure of the storage tank is improved by establishing an optimization model and adjusting pressure parameters in real time, the risk of carbon dioxide state change caused by pressure fluctuation is reduced, and the operation safety of a ship is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of storage tank pressure control, and more specifically, to a method and system for controlling the pressure of a low-pressure carbon dioxide tank on a ship. Background Art

[0002] In key application scenarios such as ship fire protection, inerting protection, and emergency cooling, the pressure control performance of low-pressure carbon dioxide storage tanks is crucial for the stability and safety of the system. Reasonable pressure regulation can ensure that the fire extinguishing system quickly releases carbon dioxide during a fire, improve the fire extinguishing response speed, and prevent the spread of fire. At the same time, in the inerting protection of cargo holds, stable pressure control can ensure the uniform release of carbon dioxide, effectively reduce the oxygen content in the hold, and reduce the risks of explosion and combustion. In addition, in emergency cooling applications, precise pressure regulation can ensure the stable release of carbon dioxide, provide continuous cooling effect, avoid overheating and damage of equipment, improve transportation convenience and application efficiency, and provide a safer and more economical guarantee for ship operation.

[0003] Existing technologies mostly rely on manual monitoring and simple pressure setting, lacking the ability of real-time data analysis and dynamic adjustment, which easily leads to excessive pressure fluctuations or inability to respond to sudden changes in a timely manner. Secondly, it is not convenient to fully consider the influence of various key factors on pressure stability, making it difficult to achieve precise pressure prediction and regulation, which may lead to abnormal gas states and affect the safety of ship operation. Moreover, it is not convenient to effectively optimize control strategies, and it is easy to cause an imbalance between cost and safety.

[0004] For the problems in the related technologies, no effective solutions have been proposed yet. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technologies, the present invention provides a method and system for controlling the pressure of a low-pressure carbon dioxide tank on a ship, which solves the problems in the above background art that existing technologies mostly rely on manual monitoring and simple pressure setting, lack the ability of real-time data analysis and dynamic adjustment, easily lead to excessive pressure fluctuations or inability to respond to sudden changes in a timely manner. Secondly, it is not convenient to fully consider the influence of various key factors on pressure stability, making it difficult to achieve precise pressure prediction and regulation, which may lead to abnormal gas states and affect transportation safety. Moreover, it is not convenient to effectively optimize control strategies, and it is easy to cause an imbalance between cost and safety.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0007] According to one aspect of the present invention, there is provided a method for controlling the pressure of a low-pressure carbon dioxide tank on a ship, including:

[0008] S1. Obtain the real-time pressure data in the low-pressure carbon dioxide tank and extract the pressure characteristic data;

[0009] S2. Analyze the pressure characteristic data using a pressure diagnosis algorithm to identify the key factors affecting the pressure stability in the low-pressure carbon dioxide tank;

[0010] S3. Based on the key factors, establish a pressure regulation model and adjust the pressure parameters of the low-pressure carbon dioxide tank in real time to optimize the control strategy;

[0011] Among them, S2 includes:

[0012] S23. Use the node anomaly algorithm to eliminate the abnormal node branches and backtrack to the nearest compliant node to regenerate the alternative paths;

[0013] S23 includes:

[0014] S232. Calculate the probability change of each node in the path in combination with the martingale theory and evaluate the difference of the nodes according to the anomaly score.

[0015] Furthermore, before S23, it also includes:

[0016] S21. Set the initial pressure as the root node and gradually expand it using the node estimated cost function to generate a pressure change path tree;

[0017] S22. Use the target detection function to verify each pressure path, screen the compliant paths that meet the threshold, mark the abnormal paths and generate conflict nodes;

[0018] After S23, it also includes:

[0019] S24. Generate an optimal pressure regulation scheme through the adjustment optimization algorithm and output the key factors affecting the pressure stability in the low-pressure carbon dioxide tank.

[0020] Furthermore, before S232, it also includes:

[0021] S231. Initialize the starting node in the path through the coin operator and the transformation operator, calculate the quantum walk probability distribution of each node, and generate the initial path;

[0022] After S232, it also includes:

[0023] S233. Identify the abnormal nodes according to the evaluation results, eliminate the abnormal node branches and backtrack to the nearest compliant node to generate alternative paths.

[0024] Furthermore, the formula for calculating the quantum walk probability distribution of each node is:

[0025] P(x,t) = |<x|(U C U T ) t |ψ0>| 2 ;

[0026] In the formula, P(x, t) represents the quantum walk probability of node x during the t-th step of quantum evolution; <x| represents the projection of the quantum state; |ψ0> represents the starting node of the quantum walk; U C represents the coin operator; U T represents the transformation operator.

[0027] Furthermore, by combining the martingale theory to calculate the probability change of each node in the path, the differences of nodes are evaluated according to the anomaly score, including:

[0028] S2321. Construct a path, set the initial state of all nodes in the path to the uncovered state, and set the overlap times of each node to the non-overlapped state;

[0029] S2322. Select a node in the path as the central node, calculate the ratio of the exclusion mass to the centrality, and divide the other nodes within a specified radius from the central node into the same group as the preliminary central node path;

[0030] S2323. Mark the current central node as a covered node, and among the remaining uncovered nodes, select the next central node according to the node with the smallest ratio as the new central node;

[0031] S2324. Repeat the operations of selecting the central node and dividing the path until all nodes in the path are divided and each node is marked as covered;

[0032] S2325. Combine the martingale theory to calculate the probability change of each node in the path, evaluate the difference between the state evolution of the node and other nodes, obtain the anomaly score, and evaluate the differences of nodes based on the anomaly score.

[0033] Furthermore, the formula for calculating the probability change of each node in the path by combining the martingale theory is:

[0034]

[0035] In the formula, ΔP(x, t) represents the change degree of the quantum walk probability of node x during the t-th step of quantum evolution; i represents the i-th node in the path; k represents the k-th iteration; represents the probability change factor of the i-th node at the k-th iteration.

[0036] Furthermore, by adjusting the optimization algorithm to generate the optimal pressure regulation scheme, the key factors affecting the pressure stability in the low-pressure carbon dioxide tank are output, including:

[0037] S241. Set the parameters of the adjustment optimization algorithm and randomly generate an initial scheme set;

[0038] S242. Form multiple niche groups based on the adjustment factor weights and scheme distributions, with each group focusing on a preset pressure adjustment region;

[0039] S243. Within each niche group, perform local pressure adjustment search and global pressure search respectively;

[0040] S244. Calculate the fitness values of each scheme, evaluate its contribution to the pressure stability of the low-pressure carbon dioxide tank, and obtain the optimal pressure adjustment scheme;

[0041] S245. Output the optimal pressure adjustment scheme and analyze to obtain the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank.

[0042] Furthermore, based on the key factors, establish a pressure regulation model and adjust the pressure parameters of the low-pressure carbon dioxide tank in real time. The optimization control strategy includes:

[0043] S31. Preset the parameters of the model algorithm, obtain the key factor set, and divide the key factor set into a test set and a training set to construct a pressure regulation model;

[0044] S32. Set the upper and lower limits of the pressure of the low-pressure carbon dioxide tank and randomly initialize several adjustment schemes, with each scheme corresponding to a pressure control setting;

[0045] S33. Calculate the fitness value of each adjustment scheme and evaluate its impact on pressure stability;

[0046] S34. Apply the explosion operator and mutation operator, and apply the mapping rule to the adjustment schemes that exceed the boundary;

[0047] S35. Use roulette wheel selection to select the best-performing individual from the current adjustment schemes and enter the next-generation optimization process;

[0048] S35 If the number of iterations reaches the maximum value, output the optimal adjustment scheme; otherwise, return to step S33 to continue optimization;

[0049] S36. Based on the training set, use the optimal adjustment scheme to train the pressure regulation model, use the test set to verify the accuracy of the model, and adjust the parameters of the pressure regulation model according to the verification results;

[0050] S37. Map the optimal adjustment scheme to the weight matrix of the pressure regulation model and adjust the pressure of the low-pressure carbon dioxide tank in real time.

[0051] Furthermore, using roulette wheel selection to select the best-performing individual from the current adjustment schemes and entering the next-generation optimization process includes:

[0052] S351. Calculate the fitness value of each individual and convert it into a probability;

[0053] S352. Construct a roulette wheel for roulette selection according to the fitness ratio. Each sector on the roulette wheel corresponds to an individual, and the size of the sector is proportional to the selection probability of the individual.

[0054] S353. Randomly select a starting point, simulate the rotation of the roulette wheel, and select the individual corresponding to the sector where the stopping point is located as a member of the next generation.

[0055] S354. Repeat the roulette selection process, select the best-performing individual, and expand it into a new population according to the optimization strategy to enter the next-generation optimization.

[0056] According to another aspect of the present invention, a low-pressure carbon dioxide tank pressure control system for ships is further provided. The system includes:

[0057] A data acquisition module for acquiring real-time pressure data inside the low-pressure carbon dioxide tank and extracting pressure characteristic data.

[0058] A diagnostic analysis module for analyzing the pressure characteristic data using a pressure diagnosis algorithm to identify the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank.

[0059] An adjustment control module for establishing a pressure regulation model based on the key factors and adjusting the pressure parameters of the low-pressure carbon dioxide tank in real time to optimize the control strategy.

[0060] The beneficial effects of the present invention are as follows:

[0061] 1. By acquiring real-time pressure data and analyzing the key factors using a pressure diagnosis algorithm, the present invention realizes precise pressure regulation. By establishing an optimization model and adjusting the pressure parameters in real time, the stability of the internal pressure of the storage tank is improved, the risk of carbon dioxide state change caused by pressure fluctuation is reduced, and the optimization control strategy effectively reduces the gas emission problem caused by excessive pressure, reduces environmental pollution, and at the same time avoids abnormal transformation of carbon dioxide caused by too low pressure, ensuring the safety of ship operation.

[0062] 2. The present invention analyzes the pressure characteristic data inside the carbon dioxide tank using a pressure diagnosis algorithm. Through path tree construction, anomaly detection, and optimization algorithms, the key factors affecting pressure stability are accurately identified. Combining the martingale theory and the quantum walk method, the path evaluation is optimized, and the accuracy of anomaly detection is improved, ensuring the scientific nature of the regulation plan. By local and global search, the pressure regulation strategy is optimized, the regulation accuracy is improved, the risk of pressure fluctuation is reduced, and the impact of abnormal phase change of carbon dioxide on ship safety is effectively avoided.

[0063] 3. The present invention realizes precise pressure control by establishing a pressure regulation model based on key factors and using roulette wheel selection to optimize the regulation scheme. Model training and verification ensure high-precision regulation, real-time adjustment of pressure parameters, improve system stability, explosion and mutation operators enhance the diversity of the scheme, prevent local optimum, the optimal scheme is mapped to the model weights, improve the real-time regulation efficiency, reduce the risk of pressure fluctuation, ensure the safety of ship transportation, and improve economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0065] Figure 1 is a flowchart of a method for controlling the pressure of a low-pressure carbon dioxide tank on a ship according to an embodiment of the present invention;

[0066] Figure 2 is a schematic block diagram of a system for controlling the pressure of a low-pressure carbon dioxide tank on a ship according to an embodiment of the present invention.

[0067] In the figure:

[0068] 1. Data acquisition module; 2. Diagnostic analysis module; 3. Adjustment control module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments.

[0070] In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0071] According to an embodiment of the present invention, a method and system for controlling the pressure of a low-pressure carbon dioxide tank on a ship are provided.

[0072] Now, the present invention will be further described in conjunction with the drawings and specific embodiments. As Figure 1 shown, the method for controlling the pressure of a low-pressure carbon dioxide tank on a ship according to an embodiment of the present invention includes:

[0073] S1. Obtain the real-time pressure data in the low-pressure carbon dioxide tank and extract the pressure characteristic data;

[0074] Specifically, the methods for obtaining real-time pressure data in the low-pressure carbon dioxide tank in the cabin include:

[0075] 1) Sensors: Install pressure sensors inside the carbon dioxide tank to ensure real-time pressure monitoring.

[0076] 2) Wireless communication module: Since the ship may operate in waters far from the base station, wide area network (WAN) communication technologies (such as LoRa or NB-IoT) can be selected to ensure the stability of data transmission. In addition, Bluetooth or Wi-Fi can be used for short-distance data transmission to facilitate docking with on-board devices or mobile terminals.

[0077] 3) PLC or embedded system: The signals collected by the sensors are processed by a PLC (programmable logic controller) or an embedded system and uploaded to the central data acquisition system.

[0078] Specifically, the real-time pressure data includes:

[0079] 1) Instantaneous pressure value: The absolute or gauge pressure value inside the tank at the current time point.

[0080] 2) Pressure change rate: The change of pressure per unit time, reflecting the speed of pressure fluctuation.

[0081] 3) Pressure fluctuation amplitude: Refers to the fluctuation range of pressure within a certain time window, which can provide an indication of pressure instability.

[0082] Specifically, the methods for extracting pressure characteristic data include:

[0083] 1) Time-domain analysis: Conduct statistical analysis on the real-time pressure data to calculate the average value, maximum value, minimum value, standard deviation, etc. of the pressure.

[0084] 2) Frequency-domain analysis: Through methods such as Fourier transform, convert the pressure data into frequency-domain data, analyze the frequency components of pressure fluctuation, and identify periodic changes.

[0085] 3) Adaptive filtering: Filter the noise of the pressure data, extract effective pressure characteristics, and remove high-frequency noise.

[0086] 4) Trend analysis: Through trend analysis of the pressure data, identify the regularity of pressure change and abnormal change trends.

[0087] Specifically, the pressure characteristic data includes:

[0088] 1) Average pressure: The average pressure within a certain time period, reflecting the stability of the system.

[0089] 2) Maximum pressure and minimum pressure: Give the limit fluctuation range of the system pressure to help identify potential overpressure or underpressure risks.

[0090] 3) Pressure volatility: The standard deviation or amplitude of pressure change within a specific time window, which can be used as an indicator of system stability.

[0091] 4) Pressure change rate: Measures the speed of pressure change and plays an important role in analyzing the response speed and dynamic characteristics of the system.

[0092] 5) Pressure deviation: The degree to which the pressure value deviates from the preset target value, used to detect whether the system is within the normal operating range.

[0093] 6) Periodic analysis data: Analyze the periodic changes in pressure to help identify anomalies related to the periodic behavior of the system.

[0094] S2. Analyze the pressure characteristic data using a pressure diagnosis algorithm to identify the key factors affecting the pressure stability in the low-pressure carbon dioxide tank;

[0095] Specifically, the key factors affecting the pressure stability in the low-pressure carbon dioxide tank include:

[0096] 1) External environmental factors:

[0097] Ambient temperature: Changes in the external temperature will cause changes in the gasification or liquefaction rate of carbon dioxide, affecting pressure stability.

[0098] Atmospheric pressure change: Changes in the external atmospheric pressure may affect the pressure balance inside the storage tank, especially during ship transportation in different sea areas or climatic conditions.

[0099] Ship motion state: The acceleration, tilt angle, vibration, etc. of the ship during navigation will all affect the gas-liquid phase balance inside the storage tank, resulting in pressure fluctuations.

[0100] Humidity and seawater environment: High humidity or seawater environment may affect the sealing performance of the storage tank and the accuracy of sensors, thereby affecting pressure measurement and control.

[0101] 2) Equipment performance factors:

[0102] Storage tank material and design: The heat insulation performance, strength, and sealing performance of the storage tank will affect its ability to maintain internal pressure.

[0103] Response speed of the pressure regulating valve: The adjustment sensitivity and response time of the valve determine its adaptability to pressure changes.

[0104] Gas replenishment system: The rate and uniformity of carbon dioxide gas replenishment affect the recovery speed of internal pressure.

[0105] Pressure relief system: Whether the safety valve or exhaust device can reasonably regulate the pressure to avoid overpressure or negative pressure situations.

[0106] Sensor accuracy and latency: The measurement accuracy, data update frequency, and response latency of the pressure sensor affect the accuracy of pressure control.

[0107] 3) Operating control factors:

[0108] Filling and discharging rates: Too fast filling or discharging rates of carbon dioxide may cause large fluctuations in instantaneous pressure.

[0109] Maintenance and fault detection: Whether the equipment is regularly maintained, and whether the sensors and valves are regularly calibrated to ensure the reliability of the control system.

[0110] S3. Based on the key factors, establish a pressure regulation model and adjust the pressure parameters of the low-pressure carbon dioxide tank in real time to optimize the control strategy;

[0111] Among them, S2 includes:

[0112] S21. Set the initial pressure as the root node, and use the node estimated cost function to expand step by step to generate a pressure change path tree;

[0113] S22. Use the target detection function to verify each pressure path, screen the compliant paths that meet the threshold, mark the abnormal paths and generate conflict nodes;

[0114] S23. Use the node anomaly algorithm to eliminate the abnormal node branches, and backtrack to the nearest compliant node to regenerate the alternative paths;

[0115] S24. Generate the optimal pressure regulation scheme through the adjustment and optimization algorithm, and output the key factors affecting the pressure stability in the low-pressure carbon dioxide tank.

[0116] Specifically, the pressure diagnosis algorithm is the A* algorithm. The A* algorithm is a heuristic search method that can find the optimal pressure regulation path based on the known pressure characteristic data to reduce abnormal fluctuations.

[0117] Specifically, the node anomaly algorithm is the quantum walk algorithm. Quantum Walk (QW) is a random process in quantum computing, which is the quantum version of the classical random walk (Random Walk). Quantum Walk uses the characteristics of quantum superposition, interference, and entanglement, making the search and calculation faster than classical algorithms in some cases and having a more efficient information propagation ability.

[0118] To facilitate the understanding of the above technical solutions of the present invention, the following will detail the analysis of the pressure characteristic data using the pressure diagnosis algorithm in the actual process of the present invention to identify the key factors affecting the pressure stability in the low-pressure carbon dioxide tank.

[0119] Step 1. Establish a pressure change path tree:

[0120] 1) Set initial parameters:

[0121] Initial pressure: 50 psi.

[0122] Ambient temperature: 25 °C.

[0123] Initial gas flow rate: 0.50 m 3 / min.

[0124] Experiment duration: 5 minutes, record the pressure change every minute.

[0125] 2) Record the pressure change data:

[0126] During the experiment, record the change of the pressure in the tank over time, as shown in Table 1:

[0127] Table 1. Pressure change data

[0128]

[0129] It can be seen from the data that at t = 2 min, the pressure reached 55 psi, which is significantly 5 psi higher than the initial pressure of 50 psi, and the reason needs to be further analyzed.

[0130] Step 2. The target detection function verifies the pressure path:

[0131] 1) Set the pressure fluctuation threshold:

[0132] Define the allowable pressure fluctuation range as ±3 psi, that is:

[0133] Below 47 psi or above 53 psi is regarded as abnormal.

[0134] 2) Calculate the pressure deviation at each time point:

[0135] t = 1 min: 52 psi - 50 psi = +2 psi (normal).

[0136] t = 2 min: 55 psi - 50 psi = +5 psi (exceeding the threshold, marked as abnormal).

[0137] t = 3 min: 53 psi - 50 psi = +3 psi (normal).

[0138] t = 4 min: 50 psi - 50 psi = 0 psi (normal).

[0139] t = 5 min: 49 psi - 50 psi = -1 psi (normal).

[0140] Conclusion: At t = 2 min, the pressure of 55 psi exceeded the upper limit of 53 psi and was marked as an abnormal path.

[0141] Step 3. Eliminate the abnormal path and trace back to the nearest compliant node:

[0142] 1) Identify the influencing factors of the abnormal point:

[0143] At t = 2 min, the gas flow rate decreased to 0.45 m 3 / min, while the temperature increased to 27 °C, which might cause the CO2 volume to expand, resulting in too high pressure.

[0144] It is necessary to trace back to the nearest normal node, i.e., t = 1 min (52 psi), and find an optimization solution.

[0145] 2) Regenerate a feasible path:

[0146] Eliminate the path at t = 2 min (55 psi) and its subsequent paths, and re-plan the possible pressure change paths, as shown in Table 2:

[0147] Table 2. Optimized pressure change paths

[0148]

[0149]

[0150] Optimization results:

[0151] Adjust the gas flow rate to 0.48 m 3 / min before t = 2 min (to avoid a pressure surge caused by too rapid a decrease in the flow rate).

[0152] Lower the ambient temperature by 1 °C before t = 2 min to inhibit the gas expansion effect.

[0153] It is expected that the pressure at t = 2 min will drop from 55 psi to 53 psi, returning to the compliant range.

[0154] Step 4. Optimize the pressure regulation scheme:

[0155] 1) Key factors affecting pressure stability:

[0156] According to the analysis, the key factors affecting the pressure stability of the low-pressure CO2 tank include:

[0157] Change in gas flow rate:

[0158] Too rapid a decrease in the gas flow rate (at t = 2 min, 0.45 m 3 / min) will cause the pressure to rise.

[0159] Optimization plan: Maintain a stable flow rate of 0.48 - 0.50 m 3 / min.

[0160] Effect of ambient temperature:

[0161] When t = 2 min, the temperature rises to 27 °C, causing the volume of CO2 gas to expand and the pressure to increase.

[0162] Optimization plan: Maintain the temperature at 25 - 26 °C to avoid excessive gas expansion.

[0163] 2) Calculate the optimized pressure curve:

[0164] Using the adjusted flow rate and temperature parameters, the predicted optimized pressure changes are shown in Table 3:

[0165] Table 3. Predicted results of pressure changes after optimization

[0166]

[0167] After optimization, the pressure is stable between 50 ± 2 psi, meeting the requirements.

[0168] 3) Final optimization plan:

[0169] Adjust the gas flow rate:

[0170] Before t = 2 min, adjust the flow rate to 0.48 m 3 / min to prevent abnormal pressure increase.

[0171] Gradually increase it to 0.50 m at t = 3 min 3 / min to ensure that the pressure drops to the stable value.

[0172] Ambient temperature control:

[0173] Maintain the temperature at 25 - 26 °C to avoid large fluctuations in pressure caused by gas expansion.

[0174] Target pressure control range:

[0175] Target pressure: 50 ± 2 psi to ensure stable pressure in the tank.

[0176] In this alternative embodiment, S23 includes:

[0177] S231. Initialize the starting node in the path through the coin operator and the transformation operator, calculate the quantum walk probability distribution of each node, and generate the initial path;

[0178] S232. Calculate the probability change of each node in the path in combination with the martingale theory, and evaluate the difference of the nodes according to the anomaly score;

[0179] S233. Identify abnormal nodes based on the evaluation results, remove the branches of the abnormal nodes, and trace back to the nearest compliant node to generate alternative paths.

[0180] It should be noted that the pressure change path is represented as a directed graph, and the pressure value at each time point corresponds to a path node. The initial node is the pressure state at zero, and the subsequent nodes represent the pressure changes at different time points. The Coin Operator is used to simulate the directionality of pressure changes (increase / decrease / remain unchanged). The Shift Operator is used to expand the path, calculate the probability distribution of each node, and generate possible pressure change paths. Through quantum walk, the association between different pressure states is established, providing a basis for subsequent analysis.

[0181] Combined with the Martingale Theory, analyze the randomness of pressure changes. Calculate the probability increase and decrease of each node in the path, and determine which nodes have significant deviations. Establish a mathematical model of normal pressure changes based on historical pressure data, and calculate the deviation degree of the current path from the historical model. Calculate the Anomaly Score for each node, and the nodes with higher scores may be abnormal pressure points. The scoring criteria can be based on: whether the pressure value exceeds the set safety range, whether the pressure change trend deviates from the normal change pattern, and whether the quantum walk distribution shows abnormal skewness. Through the anomaly score, find the nodes that exceed the threshold (i.e., abnormal pressure points).

[0182] Determine the type of abnormal nodes (such as sudden pressure increase, too low pressure, etc.). If a path contains abnormal nodes, mark the path as invalid and remove it from the path tree. Starting from the previous normal node of the abnormal node, recalculate the pressure change path. Generate new alternative paths and ensure that their pressure change trends are within the normal range.

[0183] In this alternative embodiment, the formula for calculating the quantum walk probability distribution of each node is:

[0184] P(x, t) = |<x|(U C U T ) t |ψ0>| 2 ;

[0185] In the formula, P(x, t) represents the quantum walk probability of node x in the t - step quantum evolution process; <x| represents the projection of the quantum state; |ψ0> represents the starting node of the quantum walk; U C represents the Coin Operator; U T represents the Shift Operator.

[0186] In this alternative embodiment, the probability change of each node in the combined martingale theory calculation path is calculated, and the evaluation of the differences between nodes based on the anomaly score includes:

[0187] S2321. Construct a path, set the initial state of all nodes in the path to the uncovered state, and set the overlap count of each node to the non-overlapped state;

[0188] S2322. Select a node (i.e., the box in the overlapping box covering algorithm) in the path as the central node, calculate the ratio of the exclusion mass to the centrality, and divide other nodes within a specified radius from this central node into the same group as the preliminary central node path;

[0189] S2323. Mark the current central node as a covered node, and among the remaining uncovered nodes, select the next central node based on the node with the smallest ratio as the new central node;

[0190] S2324. Repeat the operations of selecting the central node and dividing the path until all nodes in the path are divided and each node is marked as covered;

[0191] S2325. Calculate the probability change of each node in the path in combination with the martingale theory, evaluate the difference between the state evolution of the node and other nodes to obtain the anomaly score, and evaluate the differences between nodes based on the anomaly score.

[0192] Specifically, the overlapping box covering algorithm is used to evaluate the differences between nodes. The overlapping box covering algorithm is an algorithm commonly used to solve region covering or spatial search problems. The core idea of the overlapping box covering algorithm is to divide the target region into multiple covering regions and optimize the coverage of the target region by adjusting the position, size, and overlapping method of the boxes.

[0193] It should be explained that the pressure change path of the low-pressure carbon dioxide tank is represented as a set of nodes, and the pressure value at each time point corresponds to a node. The initial node is the starting state of the pressure in the tank. The quantum walk algorithm is used to calculate the expansion of the pressure path to determine the relationship between different time points and generate the initial path. According to experimental data or real-time monitoring data, the pressure change path in the low-pressure carbon dioxide tank is constructed. The initial state of all nodes in the path is set to the uncovered state, indicating that these nodes have not been classified or processed. An overlap count is set for each node, and initially, the overlap count of all nodes is set to the non-overlapped state. Select a node in the path as the central node. This node can be the node where the initial pressure value is located, or a node with a relatively large pressure change can be selected as a reference. Calculate the ratio of the exclusion mass to the centrality of this central node.

[0194] The exclusion quality represents the degree to which the pressure change around a node deviates from the normal trend, while the centrality represents the influence of a node in the entire path. Other nodes within a specified radius from the central node are grouped together, and these nodes form a preliminary central node path. Through this grouping, the pressure path is initially divided into multiple sub-paths. The current central node is marked as covered, indicating that the path division for this node has been completed. Among the remaining uncovered nodes, the node with the smallest ratio of exclusion quality to centrality is selected as the next central node. Update the path: Re-divide the path using the new central node and mark it as a covered node. Repeat the operations of selecting the central node and dividing the path until all nodes in the path are divided and each node is marked as covered.

[0195] Through multiple rounds of division, the entire pressure path can be decomposed into several sub-paths, and the nodes within each sub-path have similar pressure change patterns. Use martingale theory to analyze each node in the path. Martingale theory is a mathematical tool for dealing with stochastic processes, which can help us understand whether the pressure change of a node follows a certain stable trend. Model the pressure change of each node and calculate the probability change of the node, that is, the change trend of this node relative to other nodes. By calculating the probability change of each node and combining historical data or preset criteria, evaluate the anomaly score of the node.

[0196] The anomaly score reflects the difference between a node and other nodes. A node with a higher score indicates that there is a significant anomaly in its pressure change, which may be a potential pressure fluctuation problem. Combine martingale theory to evaluate the evolution process of the node state. If the state change of a node is large and the probability change is significantly different from other nodes, then this node may be considered an abnormal node.

[0197] In this alternative embodiment, the formula for calculating the probability change of each node in the path by combining martingale theory is:

[0198]

[0199] In the formula, ΔP(x, t) represents the degree of change in the quantum walk probability of node x during the t-step quantum evolution process; i represents the i-th node in the path; k represents the k-th iteration; represents the probability change factor of the i-th node at the k-th iteration.

[0200] In this alternative embodiment, an optimal pressure regulation scheme is generated by adjusting the optimization algorithm, and the key factors affecting the pressure stability in the low-pressure carbon dioxide tank output include:

[0201] S241. Set the parameters of the adjustment optimization algorithm and randomly generate an initial scheme set;

[0202] S242. Based on the adjustment factor weights and solution distributions, multiple niche groups are formed, and each group focuses on a preset pressure adjustment region;

[0203] S243. Within each niche group, local pressure adjustment search and global pressure search are respectively performed;

[0204] S244. Calculate the fitness values of each solution, evaluate its contribution to the pressure stability of the low-pressure carbon dioxide tank, and obtain the optimal pressure adjustment solution;

[0205] S245. Output the optimal pressure adjustment solution and analyze to obtain the key factors affecting the pressure stability in the low-pressure carbon dioxide tank.

[0206] Specifically, the adjustment and optimization algorithm is based on the niche fish-catching algorithm. The basic idea of the fish-catching algorithm comes from the behavioral habits of fishermen. It is a search method proposed by modeling this behavioral habit. Its core part is: constructing a cube centered on the selected individual, and the individual moves and shrinks independently according to its own environment, and finally finds the global optimal solution. Then, the individuals are clustered into niche populations according to the mutual distance between them, and the fish-catching algorithm is used to optimize within the generated different niche populations. The sharing mechanism is used within the niche populations to change the individual fitness to further improve the global optimization ability of the group.

[0207] It should be explained that a suitable optimization algorithm (such as particle swarm optimization, genetic algorithm, simulated annealing, fish-catching algorithm, etc.) is selected to adjust the pressure. Set relevant parameters, for example:

[0208] Maximum number of iterations: Controls the number of loops executed by the algorithm.

[0209] Initial population size: Selects the size of the initial solution set.

[0210] Adjustment factor range: Defines the range of the pressure adjustment factor, such as flow rate, temperature, valve opening, etc.

[0211] Target pressure range: Set a stable target range for the pressure inside the tank (e.g., 50 ± 2 psi). Based on the set parameters, generate multiple random initial pressure regulation schemes. These schemes are the starting points of the optimization algorithm, and each scheme represents a possible pressure regulation configuration (such as flow regulation, temperature control, valve regulation, etc.). Each scheme can consist of multiple parameters, such as the initial gas flow rate, temperature adjustment value, pressure set value, etc. Based on the set parameters, generate multiple random initial pressure regulation schemes. These schemes are the starting points of the optimization algorithm, and each scheme represents a possible pressure regulation configuration (such as flow regulation, temperature control, valve regulation, etc.). Each scheme can consist of multiple parameters, such as the initial gas flow rate, temperature adjustment value, pressure set value, etc. Assign different weights according to the influence of each adjustment factor on pressure control.

[0212] For example, the gas flow rate may have a greater impact on pressure, so its weight is higher, while the impact of temperature change on pressure is relatively small, and the weight is lower. The setting of the weights is based on historical data or expert experience and can reflect the actual impact of different factors on pressure fluctuations. Divide the initial scheme set into multiple small groups, and each group focuses on a preset pressure regulation area.

[0213] For example, one group can focus on a low-pressure area (such as 45 - 50 psi), while another group focuses on a higher-pressure area (such as 50 - 55 psi). The goal of each small group is to focus within a certain pressure range so as to find the best regulation scheme suitable for that range. Within each niche group, perform local search. Local search means only making adjustments within the pressure range of the group to explore the regulation configuration that can minimize pressure fluctuations.

[0214] For example, local search can further optimize the pressure by adjusting the gas flow rate, adjusting the valve opening, etc. to ensure that the pressure fluctuates within the target range. Between all small groups, perform global search to discover possible global optimal solutions. Global search means not only focusing on the optimization of local areas but also spanning multiple groups to find strategies that can provide optimal regulation schemes across regions.

[0215] For example, the regulation schemes found by some groups may be applicable within a wider pressure range. Through global search, the limitations of local optimization can be effectively avoided. The fitness value of each optimization scheme is used to measure the contribution of the scheme to pressure stability. The fitness value is usually evaluated by calculating factors such as the pressure fluctuation range and the deviation between the pressure and the target value. A scheme with a high fitness value indicates that the scheme can better maintain the pressure within the stable range, while a scheme with a low fitness value may lead to larger fluctuations or instability. When evaluating the effect of each scheme, based on the pressure regulation target (such as 50 ± 2 psi), calculate the stability of pressure fluctuations under the scheme.

[0216] If a certain solution causes excessive pressure fluctuations (for example, the pressure exceeds the target range), then the fitness value of this solution is low; conversely, if the solution keeps the pressure stable within the target range, then the fitness value is high. After multiple iterative optimizations, the solution with the highest fitness value is selected as the optimal pressure regulation solution. The optimal solution may include specific measures such as gas flow adjustment, temperature control strategy, valve opening adjustment, etc. The final solution ensures that the pressure in the tank is stable within the set target range. After outputting the optimal solution, analyze and summarize the key factors affecting the pressure stability in the low-pressure carbon dioxide tank. These factors may include:

[0217] Gas flow regulation: The gas flow is highly sensitive to pressure changes and is a key factor in regulation.

[0218] Temperature control: Temperature changes will cause the gas to expand or contract, thereby affecting pressure stability. Therefore, temperature regulation is also crucial.

[0219] Equipment operating status: Control factors such as valve opening also play an important role in regulating pressure changes.

[0220] Dynamic adjustment of regulation parameters: The dynamic optimization of adjusting factors based on real-time pressure data also plays a key role in maintaining stability.

[0221] In this alternative embodiment, based on the key factors, a pressure regulation model is established and the pressure parameters of the low-pressure carbon dioxide tank are adjusted in real time. The optimization control strategies include:

[0222] S31. Preset the parameters of the model algorithm, obtain the key factor set, and divide the key factor set into a test set and a training set to construct a pressure regulation model;

[0223] S32. Set the upper and lower limits of the pressure of the low-pressure carbon dioxide tank, and randomly initialize several adjustment solutions, each solution corresponding to a pressure control setting;

[0224] S33. Calculate the fitness value of each adjustment solution and evaluate its impact on pressure stability;

[0225] S34. Apply the explosion operator and the mutation operator, and apply the mapping rule to the adjustment solutions that exceed the boundary;

[0226] S35. Use roulette wheel selection to select the best individual from the current adjustment solutions to enter the next generation of optimization process;

[0227] S35 If the number of iterations reaches the maximum value, output the optimal adjustment solution; otherwise, return to step S33 to continue optimization;

[0228] S36. Based on the training set, train the pressure regulation model using the optimal adjustment scheme, verify the accuracy of the model using the test set, and adjust the parameters of the pressure regulation model according to the verification results;

[0229] S37. Map the optimal adjustment scheme to the weight matrix of the pressure regulation model and adjust the pressure of the low-pressure carbon dioxide tank in real time.

[0230] Specifically, a pressure regulation model is established by improving the fireworks algorithm and the pressure parameters of the low-pressure carbon dioxide tank are adjusted in real time to optimize the control strategy. Aiming at the problem that the explosion radius calculation method of the traditional fireworks algorithm may have a zero optimal fireworks explosion radius in the later stage of algorithm operation, an adaptive dynamic explosion radius is introduced to balance the global and local search capabilities of the algorithm in the solution space. In order to select the best-performing individuals, a roulette wheel selection mechanism is introduced. According to the fitness values of the individuals, the probability of each individual being selected is calculated. Individuals with higher fitness values have a greater probability of being selected and are used to generate the next generation of fireworks. This selection mechanism helps to retain high-quality solutions and guides the algorithm to evolve towards the optimal solution.

[0231] It should be explained that a suitable model algorithm (such as neural network, support vector machine, regression tree, etc.) is selected to construct the pressure regulation model. Set the relevant parameters of the model algorithm, such as learning rate, number of iterations, number of hidden layer nodes, etc. According to the previous analysis, determine the key factors affecting pressure stability, such as gas flow rate, temperature, valve opening, etc. Collect historical data or real-time monitoring data of these key factors. Divide the key factor set into a training set and a test set. The training set is used for model training, and the test set is used for model verification. Ensure that the division of the data set is representative so that the model can be generalized to practical applications.

[0232] Using the training set data, construct a preliminary pressure regulation model based on the selected algorithm. The input of the model is the key factors, and the output is the expected pressure value or pressure control signal. According to safety standards and operating requirements, set the upper and lower limits of the pressure of the low-pressure carbon dioxide tank (for example, 50±2 psi). Generate several random initial adjustment schemes, and each scheme contains a set of pressure control settings (such as different gas flow rates, temperature settings, etc.). These schemes serve as the initial population of the optimization algorithm for subsequent optimization processes.

[0233] For each adjustment scheme, calculate its fitness value. The fitness value reflects the impact of this scheme on pressure stability. The fitness value can be evaluated by indicators such as the pressure fluctuation range and the deviation between the pressure and the target value. Evaluate the effect of each scheme in maintaining pressure stability. A scheme with a higher fitness value indicates better control effect.

[0234] Mutate the adjustment scheme using the explosion operator and the mutation operator to explore more potential optimization spaces. The explosion operator is used to generate new schemes in the search space, and the mutation operator is used to fine-tune the existing schemes. For adjustment schemes that exceed the preset upper and lower pressure limits, apply the mapping rule to adjust them back to the legal range. The mapping rule can ensure that all schemes are within the safe operating range. Use the roulette wheel selection method to select the best-performing individuals from the current adjustment schemes to enter the next-generation optimization process.

[0235] Roulette wheel selection is based on fitness values for probabilistic selection. Schemes with higher fitness values have a higher probability of being selected. If the number of iterations has not reached the maximum value, return to step S33 to continue optimization. Repeat the steps of fitness value calculation, mutation, selection, etc. until the preset number of iterations is reached. Based on the training set, use the optimal adjustment scheme to train the pressure regulation model. Through model training, adjust the model parameters so that it can better fit the training data. Use the test set to verify the accuracy of the model and evaluate the performance of the model on unseen data. According to the verification results, adjust the model parameters to improve the generalization ability of the model. Map the optimal adjustment scheme to the weight matrix or other parameter forms of the pressure regulation model. Ensure that the model can control the pressure according to the optimal scheme.

[0236] In actual operation, use the trained pressure regulation model to adjust the pressure parameters of the low-pressure carbon dioxide tank in real time. According to the key factor data monitored in real time, the model outputs corresponding pressure control signals to maintain the pressure within the target range.

[0237] In this alternative embodiment, using roulette wheel selection to select the best-performing individuals from the current adjustment schemes to enter the next-generation optimization process includes:

[0238] S351. Calculate the fitness value of each individual and convert it into a probability;

[0239] S352. Construct a roulette wheel for roulette wheel selection according to the fitness ratio. Each sector on the roulette wheel corresponds to an individual, and the sector size is proportional to the selection probability of the individual;

[0240] S353. Randomly select a starting point, simulate the rotation of the roulette wheel, and select the individual corresponding to the sector where the stopping point is located as a member of the next generation;

[0241] S354. Repeat the roulette wheel selection process, select the best-performing individuals, and expand them into a new population according to the optimization strategy to enter the next-generation optimization.

[0242] It should be noted that for each adjustment scheme (individual) in the current population, its fitness value is calculated. The fitness value reflects the performance of the scheme in maintaining pressure stability. The calculation of the fitness value can be based on multiple factors, such as the range of pressure fluctuations, the deviation between the pressure and the target value, the feasibility of the adjustment scheme, etc. The fitness value of each individual is converted into a selection probability. The selection probability is proportional to the fitness value, that is, the higher the fitness value, the greater the probability of being selected.

[0243] For example, if an individual has a high fitness value, its probability in the roulette wheel selection will be correspondingly high. According to the selection probability of each individual, a roulette wheel for roulette wheel selection is constructed. Each sector on the roulette wheel corresponds to an individual, and the size of the sector is proportional to the selection probability of the individual.

[0244] For example, if the selection probability of an individual A is 30%, then the sector corresponding to individual A on the roulette wheel will occupy 30% of the roulette wheel. The roulette wheel is divided into multiple sectors, and the angular size of each sector is consistent with the selection probability of the corresponding individual. In this way, individuals with high fitness values will occupy larger sectors, increasing the chance of being selected. Randomly select a starting point on the roulette wheel and simulate the rotation process of the roulette wheel. When the rotation stops, record the sector where the stopping point is located. Select the individual corresponding to this sector as a member of the next-generation optimization process, and repeat the roulette wheel selection process until a sufficient number of individuals are selected to form the next-generation population.

[0245] Each selection is based on the fitness value and selection probability of the individual, ensuring that individuals with better performance have a higher chance of being selected. The selected individuals are expanded according to the optimization strategy to form a new population. The expansion strategy can include genetic algorithm operations such as crossover and mutation to increase the diversity of the population. For example, the selected individuals can generate new adjustment schemes through crossover operations, or fine-tune certain parameters through mutation operations. The newly generated population will be used as the initial population of the next generation and continue to perform optimization iterations. Through multiple generations of iteration, the optimal pressure adjustment scheme is gradually approached.

[0246] According to another embodiment of the present invention, as Figure 2 shown, there is also provided a marine low-pressure carbon dioxide tank pressure control system, which includes:

[0247] A data acquisition module 1 for acquiring real-time pressure data in the low-pressure carbon dioxide tank and extracting pressure characteristic data;

[0248] A diagnostic analysis module 2 for analyzing the pressure characteristic data using a pressure diagnostic algorithm to identify key factors affecting the pressure stability in the low-pressure carbon dioxide tank;

[0249] The adjustment control module 3 is used to establish a pressure regulation model based on key factors, adjust the pressure parameters of the low-pressure carbon dioxide tank in real time, and optimize the control strategy.

[0250] The data acquisition module 1 is connected through the diagnostic analysis module 2 and the adjustment control module 3.

[0251] In summary, with the above technical solutions of the present invention, the present invention uses a pressure diagnosis algorithm to analyze the pressure characteristic data in the carbon dioxide tank. Through path tree construction, anomaly detection and optimization algorithms, the key factors affecting pressure stability are accurately identified. Combining the martingale theory and the quantum walk method, the path evaluation is optimized to improve the accuracy of anomaly detection and ensure the scientific nature of the regulation scheme. The pressure adjustment strategy is optimized through local and global searches, improving the adjustment accuracy, reducing the risk of pressure fluctuations, and effectively avoiding the impact of abnormal carbon dioxide phase change on ship safety. The present invention establishes a pressure regulation model based on key factors, and uses roulette wheel selection to optimize the adjustment scheme to achieve precise pressure control. Model training and verification ensure high-precision regulation, adjust the pressure parameters in real time, and improve system stability. The explosion and mutation operators enhance the diversity of the scheme, prevent local optima, map the optimal scheme to the model weights, improve the real-time regulation efficiency, reduce the risk of pressure fluctuations, ensure ship transportation safety, and improve economy.

[0252] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for controlling the pressure of a low-pressure carbon dioxide tank for marine use, characterized in that, Including: S1. Obtain the real-time pressure data inside the low-pressure carbon dioxide tank and extract the pressure characteristic data; S2. Analyze the pressure characteristic data using a pressure diagnosis algorithm to identify the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank; S3. Based on the key factors, establish a pressure regulation model and adjust the pressure parameters of the low-pressure carbon dioxide tank in real time to optimize the control strategy; Among them, the said S2 includes: S23. Use the node anomaly algorithm to eliminate the abnormal node branches and trace back to the nearest compliant node to regenerate the alternative paths; The said S23 includes: S232. Combine the martingale theory to calculate the probability change of each node in the path and evaluate the difference of the nodes according to the anomaly score.

2. The method for controlling the pressure of a low-pressure carbon dioxide tank for a ship according to claim 1, wherein Before the said S23, there is also included: S21. Set the initial pressure as the root node and use the node estimated cost function to expand step by step to generate a pressure change path tree; S22. Use the target detection function to verify each pressure path, screen out the compliant paths that meet the threshold, mark the abnormal paths and generate conflict nodes; After the said S23, there is also included: S24. Generate an optimal pressure regulation scheme through the adjustment optimization algorithm and output the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank.

3. A method for controlling the pressure of a low-pressure carbon dioxide tank for a ship according to claim 1, characterized in that, Before the said S232, there is also included: S231. Initialize the starting node in the path through the coin operator and the transformation operator, calculate the quantum walk probability distribution of each node, and generate the initial path; After the said S232, there is also included: S233. Identify the abnormal nodes according to the evaluation results, eliminate the abnormal node branches and trace back to the nearest compliant node to generate alternative paths.

4. A method for controlling the pressure of a marine low-pressure carbon dioxide tank according to claim 3, characterized in that, The formula for calculating the quantum walk probability distribution of each node is: P(x, t) = |<x|(U C U T ) t |ψ0>| 2 ; Wherein, P(x, t) represents the quantum walk probability of node x in the t-th step of quantum evolution; <x| represents the projection of the quantum state; |ψ0> represents the starting node of the quantum walk; U C represents the coin operator; U T represents the transformation operator.

5. A method for controlling the pressure of a low-pressure carbon dioxide tank for a ship according to claim 1, characterized in that, The combination of calculating the probability change of each node in the path by the martingale theory and evaluating the difference of the nodes according to the anomaly score includes: S2321. Construct the path and set the initial state of all nodes in the path to the uncovered state, and set the number of overlapping times of each node to the non-overlapped state; S2322. Select a node in the path as the central node, calculate the ratio of the exclusion mass to the centrality, and divide the other nodes within a specified radius from the central node into the same group as the preliminary central node path; S2323. Mark the current central node as the covered node, and among the remaining uncovered nodes, select the next central node according to the node with the smallest ratio as the new central node; S2324. Repeat the operations of selecting the central node and dividing the path until all nodes in the path are completed for division and each node is marked as covered; S2325. Combine the martingale theory to calculate the probability change of each node in the path, evaluate the difference between the state evolution of the node and other nodes, obtain the anomaly score, and evaluate the difference of the nodes based on the anomaly score.

6. A method for controlling the pressure of a low-pressure carbon dioxide tank for a ship according to claim 5, characterized in that, The formula for calculating the probability change of each node in the path by the martingale theory is: Where, ΔP(x, t) represents the change degree of the quantum walk probability of node x in the t-th step of quantum evolution; i represents the i-th node in the path; k represents the k-th iteration; represents the probability change factor of the i-th node at the k-th iteration.

7. A method for controlling the pressure of a marine low-pressure carbon dioxide tank according to claim 2, characterized in that, The generation of the optimal pressure regulation scheme through the adjustment optimization algorithm and the output of the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank include: S241. Set the parameters of the adjustment optimization algorithm and randomly generate an initial scheme set; S242. Form multiple niche groups based on the adjustment factor weights and solution distributions, with each group focusing on a preset pressure adjustment region; S243. Perform local pressure adjustment search and global pressure search respectively within each niche group; S244. Calculate the fitness values of each solution, evaluate its contribution to the pressure stability of the low-pressure carbon dioxide tank, and obtain the optimal pressure adjustment solution; S245. Output the optimal pressure adjustment solution, and analyze to obtain the key factors affecting the pressure stability in the low-pressure carbon dioxide tank.

8. A method for controlling the pressure of a low-pressure carbon dioxide tank for a ship according to claim 1, characterized in that, Based on the key factors, establish a pressure regulation model and adjust the pressure parameters of the low-pressure carbon dioxide tank in real time. The optimization control strategy includes: S31. Preset the parameters of the model algorithm, obtain the key factor set, and divide the key factor set into a test set and a training set to construct a pressure regulation model; S32. Set the upper and lower limits of the pressure of the low-pressure carbon dioxide tank, and randomly initialize several adjustment solutions, with each solution corresponding to a pressure control setting; S33. Calculate the fitness value of each adjustment solution and evaluate its impact on pressure stability; S34. Apply the explosion operator and mutation operator, and apply the mapping rule to the adjustment solutions that exceed the boundaries; S35. Use roulette wheel selection to select the best-performing individual from the current adjustment solutions to enter the next-generation optimization process; S35 If the number of iterations reaches the maximum value, output the optimal adjustment solution; otherwise, return to step S33 to continue optimization; S36. Based on the training set, use the optimal adjustment solution to train the pressure regulation model, use the test set to verify the accuracy of the model, and adjust the parameters of the pressure regulation model according to the verification results; S37. Map the optimal adjustment solution to the weight matrix of the pressure regulation model and adjust the pressure of the low-pressure carbon dioxide tank in real time.

9. A method for controlling the pressure of a marine low-pressure carbon dioxide tank according to claim 8, characterized in that, The process of using roulette wheel selection to select the best-performing individual from the current adjustment solutions to enter the next-generation optimization process includes: S351. Calculate the fitness value of each individual and convert it into a probability; S352. Construct a roulette wheel for roulette wheel selection according to the fitness ratio. Each sector on the roulette wheel corresponds to an individual, and the size of the sector is proportional to the selection probability of the individual; S353. Randomly select a starting point, simulate the rotation of the roulette wheel, and select the individual corresponding to the sector where the stopping point is located as a member of the next generation; S354. Repeat the roulette wheel selection process, select the best-performing individual, and expand it into a new population according to the optimization strategy to enter the next-generation optimization.

10. A marine low-pressure carbon dioxide tank pressure control system for implementing the marine low-pressure carbon dioxide tank pressure control method described in any one of claims 1-9, characterized in that, The system includes: A data acquisition module for acquiring the real-time pressure data inside the low-pressure carbon dioxide tank and extracting the pressure characteristic data; A diagnosis and analysis module for analyzing the pressure characteristic data using a pressure diagnosis algorithm to identify the key factors affecting the pressure stability inside the low-pressure carbon dioxide tank; An adjustment and control module for establishing a pressure regulation model based on the key factors and adjusting the pressure parameters of the low-pressure carbon dioxide tank in real time to optimize the control strategy.

Images