Fishing boat cabin CO2 concentration monitoring method and device

Abstract

The invention provides a fishing boat cabin CO2 concentration monitoring method and device. The method provided by the invention comprises the following steps: acquiring CO2 concentrations of a plurality of positions in a fishing boat cabin, CO2 concentration influence factors and a CO2 propagation curve in a simulation cabin; when the CO2 concentration at any position reaches a first alarm threshold value, starting a first alarm, monitoring the change of the CO2 concentration in real time, and calculating the weight of each CO2 concentration influence factor based on an eagle hunting algorithm; when the CO2 concentration reaches a second alarm threshold value, on the basis of the CO2 concentration, the CO2 concentration influence factor and the weight of the CO2 concentration influence factor, a random forest algorithm is utilized to predict CO2 concentration change trends of multiple positions, and multiple random forest regression curves are obtained; a safe evacuation time is determined based on comparison of the plurality of random forest regression curves to the CO2 propagation curve, and a second alarm is initiated. The method and the device provided by the invention are used for monitoring the CO2 concentration of the cabin of the fishing boat in real time.

Description

Technical Field

[0001] The present application relates to the technical field of ship safety, and particularly to a method and device for monitoring the CO2 concentration in a fishing vessel engine room. Background Art

[0002] With the continuous development and utilization of global fishery resources, the operating environment and safety of modern fishing vessels have gradually attracted attention. Among them, the engine room, as the core part of a fishing vessel, carries key equipment such as the propulsion system and power supply, and its stable operation is crucial for the navigation safety of the ship and the fishing production efficiency. However, due to the long-term continuous operation of mechanical equipment, a large amount of exhaust gas, especially high-concentration carbon dioxide (CO2), will be emitted in the engine room. When the CO2 concentration exceeds the safety threshold, serious health risks may be triggered, such as asphyxiation, incapacitation, and even threatening the lives of crew members. Therefore, timely and accurate monitoring of the CO 2 concentration in the fishing vessel engine room is a necessary means to ensure the safety of crew members and improve the safety of the operating environment.

[0003] Traditional monitoring of the CO2 concentration in a fishing vessel engine room usually relies on regular manual inspections and manual sampling analysis. Specifically, operators regularly collect air samples in the fishing vessel engine room and measure the concentration through laboratory analysis or portable equipment. Although this method can provide concentration data under specific circumstances, it has obvious deficiencies, such as data lag, slow response, and high labor costs, and it is difficult to meet the requirements of modern fishing vessels for efficient, accurate, and continuous monitoring.

[0004] Therefore, there is an urgent need for a method to achieve real-time monitoring of the CO2 concentration in the fishing vessel engine room and trigger an alarm in a timely manner through a preset alarm threshold to help crew members evacuate safely. Summary of the Invention

[0005] In view of this, the present application provides a method and device for monitoring the CO2 concentration in a fishing vessel engine room to achieve real-time monitoring of the CO2 concentration in the fishing vessel engine room and trigger an alarm in a timely manner through a preset alarm threshold to help crew members evacuate safely.

[0006] Specifically, the present application is implemented through the following technical solutions:

[0007] The first aspect of the present application provides a method for monitoring the CO2 concentration in a fishing vessel engine room, the method comprising:

[0008] Obtaining the CO2 concentration and CO2 concentration influencing factors at multiple positions in the fishing vessel engine room, and simulating the CO2 propagation curve in the engine room; wherein, based on the ranking of the influence degree on the change of the CO2 concentration, a preset number of CO2 concentration influencing factors are determined;

[0009] When the CO2 concentration at any position reaches the first alarm threshold, start the first alarm, monitor the change of CO2 concentration in real time, and calculate the weights of various CO2 concentration influencing factors based on the goshawk hunting algorithm; the weights of different CO2 concentration influencing factors are different, and the weight represents the influence degree of the CO2 concentration influencing factor on the CO2 concentration. The first alarm is to give a safety warning on the fishing boat and set the ventilation system to the first ventilation mode;

[0010] When the CO2 concentration reaches the second alarm threshold, based on the CO2 concentration, the CO2 concentration influencing factors, and the weights of the CO2 concentration influencing factors, use the random forest algorithm to predict the CO2 concentration change trends at multiple positions, and obtain multiple random forest regression curves corresponding to the multiple positions; the first alarm threshold is less than the second alarm threshold;

[0011] Based on the comparison between multiple random forest regression curves and the simulated CO2 propagation curve, determine the safe evacuation time and start the second alarm. The second alarm is to broadcast the safe evacuation time on the fishing boat and set the ventilation system to the second ventilation mode.

[0012] The second aspect of this application provides a CO2 concentration monitoring device for a fishing boat engine room. The device includes an acquisition module, a calculation module, a prediction module, and a determination module;

[0013] Among them, the acquisition module is used to acquire the CO2 concentration and CO2 concentration influencing factors at multiple positions in the fishing boat engine room, and simulate the CO2 propagation curve in the engine room; among them, based on the sorting of the influence degree on the change of the CO2 concentration, a preset number of CO2 concentration influencing factors are determined;

[0014] The calculation module is used to start the first alarm when the CO2 concentration at any position reaches the first alarm threshold, monitor the change of CO2 concentration in real time, and calculate the weights of various CO2 concentration influencing factors based on the goshawk hunting algorithm; the weights of different CO2 concentration influencing factors are different, and the weight represents the influence degree of the CO2 concentration influencing factor on the CO2 concentration. The first alarm is to give a safety warning on the fishing boat and set the ventilation system to the first ventilation mode;

[0015] The prediction module is used to, when the CO2 concentration reaches the second alarm threshold, based on the CO2 concentration, the CO2 concentration influencing factors, and the weights of the CO2 concentration influencing factors, use the random forest algorithm to predict the CO2 concentration change trends at multiple positions, and obtain multiple random forest regression curves corresponding to the multiple positions; the first alarm threshold is less than the second alarm threshold;

[0016] The determining module is configured to compare a plurality of the random forest regression curves with the CO2 propagation curve obtained by simulation to determine the safe evacuation time and activate a second alarm, where the second alarm is to broadcast the safe evacuation time on the fishing boat and set the ventilation system to a second ventilation mode.

[0017] The fishing boat engine room CO2 concentration monitoring method and device provided by this application are designed to significantly extend the escape time in case of CO2 leakage and accurately give early warnings of CO2 leakage. The method provided by the present invention combines the goshawk hunting algorithm and the random forest method to predict the CO2 leakage propagation curve. That is, first, the goshawk hunting algorithm is used to estimate the weights of the CO2 propagation influence factors, which can be adapted to various engine room environments, evaluate the influence degrees of various factors such as space and wind direction on propagation, and further rely on the random forest to estimate the CO2 propagation according to the above weights, so as to obtain the most accurate escape time adapted to the current engine room. In addition, in order to accurately warn of the alarm situation in the fishing boat engine room, this application adopts different alarm modes and different ventilation levels according to different concentration thresholds, enabling the crew to understand the CO2 concentration level only through the alarm degree. At the same time, different ventilation levels can increase the ventilation volume at dangerous position points, improving the safety of the cabin. In the first aspect, when the CO2 concentration at any position in the fishing boat engine room is monitored to reach the first alarm threshold, the goshawk hunting algorithm is used to efficiently calculate the weights of multiple CO2 concentration influence factors, and then judge the specific influence degrees of these factors on the CO2 concentration. The goshawk hunting algorithm is used in this application to optimize the weights of each CO2 concentration influence factor, thus realizing adaptive weight distribution. The goshawk hunting algorithm is essentially a heuristic optimization method based on the hunting strategies in nature, which can quickly and effectively search for the optimal solution. Specifically in this application, the goshawk hunting algorithm can automatically adjust and optimize the weights of the CO2 concentration influence factors. Based on real-time data and environmental condition changes, the system can flexibly adjust the influence degrees of these factors on the concentration change, avoiding the limitations of traditional fixed weight settings, being able to adaptively adjust according to environmental changes, and optimizing the accuracy of the alarm. For example, if the space size of the fishing boat engine room or the volume of the CO2 device changes, these changes may cause the law of CO2 concentration fluctuations to change. The goshawk hunting algorithm can help the system more accurately understand the contributions of these factors to the CO2 concentration change by optimizing the weight of each influence factor, so as to more precisely set the alarm threshold and early warning level. This adaptive ability enables the system to continuously maintain high-efficiency monitoring and response capabilities in a changing environment, avoiding the problem in traditional methods that the static weight setting cannot cope with environmental changes. In the second aspect, when the CO2 concentration at any position in the fishing boat engine room is monitored to reach the second alarm threshold, the random forest algorithm is used to predict the future CO2 concentration change trends at multiple positions. The random forest algorithm can handle complex multi-dimensional inputs, has strong fault tolerance ability, and can still maintain good prediction performance when there is noise in the data. Compared with traditional regression analysis, the random forest algorithm has more advantages in dealing with multi-factor, high-complexity non-linear relationships. By training the random forest model, the system can not only predict the current CO2 concentration level, but also predict the future CO2 concentration change trends within a certain period of time.This is particularly important for the fishing vessel engine room because the change in CO2 concentration usually does not occur instantaneously but is a gradual accumulation process. Through multi-dimensional and multi-position data analysis, the random forest algorithm can generate multiple regression curves, which can provide accurate predictions of future CO2 concentrations. The crew can make timely decisions based on this prediction information, such as starting the ventilation system in advance or adjusting the operation process to avoid a further increase in CO2 concentration. Thirdly, the present application calculates the safe evacuation time by comparing the random forest regression curve with the CO2 propagation curve obtained by simulation. Through scientific simulation and prediction, a more accurate evacuation time is provided for the crew. This calculation is not just a simple response after the CO2 concentration reaches a certain fixed value, but by dynamically simulating the diffusion process of CO2 in the entire engine room and combining the CO2 concentration changes at different positions, it accurately estimates the safest evacuation opportunity under the current concentration and conditions. Once the safe evacuation time is calculated, the system will trigger a second alarm and announce the safe evacuation time, providing the crew with an early time window to avoid the emergency panic and missed evacuation opportunities caused by the rapid increase in CO2 concentration. By knowing the safe evacuation time in advance, the crew can take corresponding measures in an orderly manner, which not only improves the survival rate of personnel but also reduces the misoperations or delays that may occur under high-pressure situations. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of the method for monitoring the CO2 concentration in the fishing vessel engine room provided by the first embodiment of the present application;

[0019] Figure 2 It is a schematic structural diagram of the device for monitoring the CO2 concentration in the fishing vessel engine room provided by the second embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application.

[0021] The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. The singular forms "a", "the" and "said" used in the present application are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0022] It should be understood that although terms such as first, second, and third may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0023] Specific embodiments are given below to introduce the technical solutions of this application in detail.

[0024] Figure 1 It is a flowchart of the CO2 concentration monitoring method for the fishing boat engine room provided in the first embodiment of this application. Please refer to Figure 1 , the method provided in this embodiment may include:

[0025] S101. Obtain the CO2 concentrations and CO2 concentration influencing factors at multiple positions in the fishing boat engine room, and simulate the CO2 propagation curve in the engine room.

[0026] Specifically, the CO2 concentration influencing factors will affect the change of the CO2 concentration. The CO2 concentration influencing factors include the internal space size of the fishing boat engine room, the volume of the CO2 device, and the leakage time of the CO2 device. The multiple positions are set according to actual needs and are not limited in this embodiment. It should be noted that the CO2 concentrations at different positions are different, and the CO2 concentrations are obtained by installing gas sensors at different positions in the fishing boat engine room. This application selects a preset number of CO2 concentration influencing factors based on the influence degree of different CO2 concentration influencing factors on the change of the CO2 concentration.

[0027] In this embodiment, the gas sensor used is a high-performance non-dispersive infrared gas sensor. The high-performance non-dispersive infrared gas sensor has an extremely short response time and can usually complete a complete measurement cycle within milliseconds, ensuring that the change in CO 2 concentration can be captured in a dynamically changing engine room environment in a timely manner. The high-performance non-dispersive infrared gas sensor is built-in with an advanced temperature compensation algorithm, and can maintain stable measurement performance even in the case of large temperature fluctuations. Its measurement range is wide, covering 0 to 5000 ppm, and the accuracy reaches ±3% FS, fully meeting the requirements for CO 2 concentration monitoring in the fishing boat engine room. The high-performance non-dispersive infrared gas sensor has good selectivity and is only sensitive to CO 2The gas generation reaction is not affected by other common gases such as water vapor and oxygen, ensuring the reliability of the measurement results. In addition, the sensor housing is made of special materials and has excellent waterproof and dustproof performance, making it suitable for long-term use in harsh environments such as humidity and salt spray. The high-performance non-dispersive infrared gas sensor adopts a low-power design. Considering the possible power resource limitations faced by fishing vessels, the high-performance non-dispersive infrared gas sensor has been optimized with extremely low power consumption, which not only extends the battery life but also reduces the operating cost, making it suitable for applications with long-term unattended operation scenarios.

[0028] During specific implementation, gas sensors are installed at different positions in the fishing vessel engine room. The detection ranges of all gas sensors cover the entire area of the fishing vessel engine room. Each gas sensor periodically collects CO2 concentration data to obtain the CO2 concentrations at multiple positions in the fishing vessel engine room. According to the design drawings of the fishing vessel, the size of the fishing vessel engine room is obtained, and the total volume of the fishing vessel engine room, that is, the internal space size of the fishing vessel engine room, is calculated. According to the specification manual of the CO2 device or actual measurement, the volume of the CO2 device is determined. Further, by installing leakage monitoring equipment (such as pressure sensors or gas leakage sensors), the pressure change in the CO2 device is monitored. When a leakage is detected, the start time is recorded, the duration of the leakage is monitored, and the end time of the leakage is obtained to get the time when the CO2 device leaks.

[0029] Optionally, obtaining the CO2 concentrations at multiple positions in the fishing vessel engine room includes: obtaining the CO2 concentrations at multiple first positions in the fishing vessel engine room, where the ratio of the union of the monitoring areas of the sensors at the multiple first positions to the area of the fishing vessel engine room is greater than a preset threshold; establishing a spatial structure model of the fishing vessel engine room and establishing a propagation curve of CO2 in the spatial structure model; predicting the CO2 increase at each of the first positions according to the propagation curve and the CO2 concentrations at the multiple first positions; screening multiple positions with an increase greater than the preset threshold according to the CO2 increase to obtain second positions; and obtaining the CO2 concentrations at the second positions in the fishing vessel engine room.

[0030] In specific implementation, multiple sensors are installed in the fishing vessel engine room to monitor the CO2 concentration. The multiple sensors are arranged in multiple different areas of the engine room, and the monitoring areas of these sensors are combined together, and the ratio of the total monitoring area to the area of the entire engine room is greater than a preset threshold. That is, it is ensured that the arrangement of the sensors covers a sufficient area so as to accurately reflect the CO2 concentration situation of the entire engine room. Further, through the structure diagram and layout of the fishing vessel, a spatial structure model of the engine room is established. The spatial structure model is a three-dimensional model that describes different areas in the engine room and the relationships between them (for example: different cabin areas, passageways, partition walls, etc.). According to the spatial structure model of the engine room, combined with the laws of fluid mechanics and gas propagation, factors such as the diffusion rate, flow direction, and obstacles of CO2 at different positions are calculated, and a propagation curve of the CO2 concentration is established. According to the established CO2 propagation curve, by inputting the CO2 concentration data at multiple first positions, how the CO2 concentration at these positions will change over time is predicted, and the increase in the CO2 concentration is obtained. According to the prediction results, the positions where the CO2 increase exceeds the preset threshold are screened out. The CO2 concentration at these positions changes greatly, and these positions are marked as second positions. By obtaining the sensors corresponding to the second positions, the CO2 concentration at the second positions is read. Subsequently, concentration monitoring and alarm detection are performed based on the CO2 concentration at the second positions.

[0031] The method provided in this embodiment, by arranging multiple sensors in the fishing vessel engine room and combining the spatial structure model and the CO2 propagation curve, provides a comprehensive and efficient CO2 concentration monitoring and early warning method. First, through the multi-position sensor layout, it is ensured that the CO2 monitoring of the engine room covers a sufficient area, which can accurately reflect the CO2 concentration situation of the entire engine room, and the ratio of the union of the sensor monitoring areas to the engine room area is greater than the preset threshold, further ensuring the integrity of the coverage. Combining with the spatial structure model of the fishing vessel, which accurately describes the different areas in the engine room and their interrelationships, this enables the propagation of CO2 in the engine room to not only be monitored point-to-point, but also to accurately simulate the change of CO2 concentration by considering complex factors such as air flow and cabin area isolation. In addition, by introducing the principles of fluid mechanics and gas propagation laws, the diffusion rate, flow direction, and obstacles of CO2 at different positions can be calculated more accurately, and thus a reliable CO2 concentration propagation curve can be obtained. Through the increase prediction based on this propagation curve, it is possible to accurately identify which positions have a large change in CO2 concentration, timely screen out potential dangerous areas and mark them as "second positions", thereby avoiding ignoring key areas. Further, by obtaining the real-time CO2 concentration data of these areas, real-time monitoring can be carried out to ensure timely discovery and response to abnormal increases in CO2 concentration, trigger the alarm system, and remind the operator to take appropriate measures. This method can not only accurately locate dangerous areas, but also reduce resource waste, avoid the increase of redundant hardware by reasonably arranging sensors, and at the same time improve the working efficiency of the system.

[0032] S102: When the CO2 concentration at any location reaches a first alarm threshold, a first alarm is activated, CO2 concentration changes are monitored in real time, and weights of various CO2 concentration influencing factors are calculated based on a goshawk hunting algorithm.

[0033] Specifically, the first alarm threshold is set according to actual needs, and this is not limited in this embodiment. For example, in one embodiment, the first alarm threshold is 700ppm. The first alarm is to perform a safety alarm on the fishing boat and set the ventilation system to the first ventilation mode. In combination with the above description, the CO2 concentration influencing factor affects the change of CO2 concentration, and different CO2 concentration influencing factors have different degrees of influence on CO2 concentration, that is, different CO2 concentration influencing factors have different weights.

[0034] It should be noted that the first alarm threshold is formulated according to relevant standards. 2 The concentration shall not be higher than 1% (i.e. below 10,000 ppm). At the same time, the oxygen concentration in the engine room of a fishing vessel shall be greater than or equal to 19.5% (by volume) and less than or equal to 23.5% (by volume). These standards are intended to ensure the safety and health of crew members in the engine room of a fishing vessel and prevent accidents caused by lack of oxygen or CO. 2 Dangerous conditions may occur due to excessive concentrations. 2 When the concentration approaches or exceeds this value, it may cause discomfort symptoms such as dizziness, shortness of breath, fatigue, etc., which poses a great threat to health and safety. Therefore, it is necessary to ensure that the concentration is controlled within the alarm threshold through reasonable ventilation and other means. However, the specific alarm threshold may vary slightly depending on factors such as cabin type and personnel situation. 2 Concentration alarm thresholds vary. For cabins, the following common standards are usually used: For general living or working areas, indoor CO 2 The concentration should be controlled below 1000ppm (parts per million). When the concentration is between 1000-1500ppm for a long time, people may feel stuffy and have difficulty breathing. When the concentration exceeds 1500ppm, it will more obviously affect comfort and health. For special operations and other areas with higher requirements, such as some precision operations, long-term closed areas with concentrated personnel, it is best to control it within 700ppm.

[0035] During specific implementation, the CO2 concentrations at multiple positions in the fishing boat engine room are obtained through real-time analysis, and the CO2 concentration at each position is compared with the first alarm threshold. When it is determined that the CO2 concentration at any position is greater than or equal to the first alarm threshold, the first alarm is activated, a safety alarm is issued on the fishing boat, the ventilation system is turned on, the ventilation system is set to the first ventilation mode, and the CO2 concentration detected by the gas sensor is continuously obtained in the future time period to monitor the change in the CO2 concentration, and the weights of various CO2 concentration influencing factors are calculated through the goshawk hunting algorithm.

[0036] Optionally, calculating the weights of various CO2 concentration influencing factors based on the goshawk hunting algorithm includes:

[0037] (1) Determining the CO2 concentration influencing factors as population individuals and determining the weights of the CO2 concentration influencing factors as the fitness solutions of the population individuals; wherein, calculating the predicted CO2 concentration corresponding to different combinations of population individuals, and determining the mean square error between the actual CO2 concentration and the predicted CO2 concentration as the weight of each CO2 concentration influencing factor.

[0038] Specifically, by introducing the goshawk hunting algorithm to determine the weights of various CO2 concentration influencing factors, each CO2 concentration influencing factor is determined as a population individual, and the weight of each CO2 concentration influencing factor is determined as the fitness solution of the population individual. Among them, the fitness solutions of each population individual are different, and the fitness solution is obtained by calculating the mean square error between the predicted CO2 concentration corresponding to different combinations of population individuals and the actual CO2 concentration actually measured by the gas sensor. During specific implementation, the fitness solution can be calculated according to the following formula: ; ; Among them, the is the mean square error between the predicted CO2 concentration and the actual CO2 concentration; the is the fitness solution; the is the number of different combinations of population individuals; the is the actual CO2 concentration; the is the predicted CO2 concentration.

[0039] Furthermore, the population individual can be expressed as: ; Among them, the is the population individual; the is the number of population individuals; the is the number of CO2 concentration influencing factors.

[0040] The fitness solution of the population individual can be expressed as: ; wherein, the is the fitness solution of the population individuals; the is the number of the population individuals.

[0041] (2) Each individual of the first population randomly identifies other individuals of the second population, and updates the individual of the first population and the fitness solution of the individual of the first population based on the fitness solution of the individual of the second population and the fitness solution of the individual of the first population, and adjusts the individual of the first population and the corresponding fitness solution based on the fitness solution of the updated individual of the first population and the fitness solution of the individual of the first population before update.

[0042] Specifically, each individual of the first population in the population individuals identifies other population individuals in the population individuals, randomly selects an individual of the second population, determines the first fitness solution corresponding to the individual of the second population from the calculated fitness solutions corresponding to each population individual, determines the corresponding population individual update method based on the magnitude relationship between the first fitness solution and the fitness solution corresponding to the individual of the first population, and updates the individual of the first population and the fitness solution corresponding to the individual of the first population based on the population individual update method. After updating the individual of the first population, determine the fitness solution of the updated individual of the first population from the calculated fitness solutions corresponding to each population individual, and adjust the individual of the first population and the corresponding fitness solution by comparing the magnitude of the fitness solution of the updated individual of the first population and the fitness solution of the individual of the first population before update. When the fitness solution of the updated individual of the first population is less than the fitness solution of the individual of the first population before update, determine the updated individual of the first population and the corresponding fitness solution as the individual of the first population and the corresponding fitness solution. When the fitness solution of the updated individual of the first population is not less than the fitness solution of the individual of the first population before update, do not update the individual of the first population and the corresponding fitness solution (the individual of the first population remains in the initial state, that is, no adjustment or update is performed).

[0043] It should be noted that each individual of the first population randomly identifies one other individual of the second population each time, and repeats the same steps until the update times reach the maximum update times and stops identifying other individuals of the second population.

[0044] Among them, when the first fitness solution is less than the fitness solution corresponding to the individuals in the first population, calculate the first product of the first random parameter and the individuals in the first population. The first random parameter is 1 or 2. Determine the difference between the second population individual and the first product as the intermediate value, and determine the sum value of the product of the second random parameter and the intermediate value and the individuals in the first population as the updated individuals in the first population; the value of the second random parameter ranges from 0 to 1. When the first fitness solution is less than the fitness solution corresponding to the individuals in the first population, calculate the difference between the individuals in the first population and the individuals in the second population, and determine the sum value of the product of the second random parameter and the difference and the individuals in the first population as the updated individuals in the first population.

[0045] In specific implementation, based on the following formula, each individual in the first population randomly identifies other individuals in the second population: ; where, the is an individual in the first population; the is an individual in the second population; the is the number of individuals in the population.

[0046] Update the individuals in the first population and the corresponding fitness solutions based on the following formula: ; where, the is the updated individual in the first population; the is the individual in the first population before update; the is an individual in the second population; the is the second random parameter, with a value ranging from 0 to 1; the is the first random parameter, with a value of 1 or 2; the is the fitness solution of the individual in the second population; the is the fitness solution of the individual in the first population.

[0047] Adjust the individuals in the first population and the corresponding fitness solutions based on the following formula: ; where, the is the individual in the first population before update; the is the updated individual in the first population; the is the fitness solution of the updated individual in the first population; the is the fitness solution of the individual in the first population before update.

[0048] (3) Determine the update radius based on the update times of the first population individuals. Update the first population individuals and their fitness solutions based on the update radius and the adjusted first population individuals. Adjust the first population individuals and their corresponding fitness solutions based on the fitness solutions of the updated first population individuals and the fitness solutions of the first population individuals before the update. When the update times reach the maximum update times, determine the weights of each CO2 concentration influencing factor based on the fitness solutions of each first population individual.

[0049] Specifically, the maximum update times are set according to actual needs. In this embodiment, no limitation is imposed on this.

[0050] In specific implementation, determine the update radius based on the quotient of the update times of the first population individuals and the maximum update times. Determine the update amount as the product of the update radius, the second random parameter, and the adjusted first population individuals. Determine the updated first population individuals as the sum of the adjusted first population individuals and the update amount. Determine the fitness solutions of the updated first population individuals from the calculated fitness solutions corresponding to each population individual. Adjust the first population individuals and their corresponding fitness solutions by comparing the sizes of the fitness solutions of the updated first population individuals and the fitness solutions of the first population individuals before the update. Among them, when the fitness solutions of the updated first population individuals are less than the fitness solutions of the first population individuals before the update, determine the updated first population individuals and their corresponding fitness solutions as the first population individuals and their corresponding fitness solutions. When the fitness solutions of the updated first population individuals are not less than the fitness solutions of the first population individuals before the update, do not update the first population individuals and their corresponding fitness solutions (the first population individuals remain in the initial state, that is, no adjustment or update is performed).

[0051] In specific implementation, determine the update radius based on the following formula: ; Among them, the is the update radius; the is the update times; the is the maximum update times.

[0052] Update the first population individuals and their corresponding fitness solutions based on the following formula: ; Among them, the is the updated first population individuals; the is the first population individuals before the update; the is the update radius; the is the second random parameter, and its value ranges from 0 to 1.

[0053] Adjust the individuals of the first population and the corresponding fitness solutions based on the following formula: ; wherein, the is the individual of the first population before update; the is the individual of the first population after update; the is the fitness solution of the individual of the first population after update; the is the fitness solution of the individual of the first population before update.

[0054] Furthermore, when the update times of the individuals of the first population reach the maximum update times, at this time, determine the fitness solution corresponding to each individual of the first population as the weight of the CO2 concentration influencing factor corresponding to each individual of the first population, and determine the weight of each CO2 concentration influencing factor.

[0055] S103. When the CO2 concentration reaches the second alarm threshold, based on the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor, use the random forest algorithm to predict the CO2 concentration change trends at multiple positions, and obtain multiple random forest regression curves corresponding to the multiple positions.

[0056] Specifically, the second alarm threshold is set according to actual needs, and in this embodiment, it is not limited herein. It should be noted that the second alarm threshold is greater than the first alarm threshold. For example, the second alarm threshold is 7000 ppm. Each random forest regression curve represents the change trend of the CO2 concentration at the corresponding position in the fishing boat engine room in the future for a period of time, and the random forest regression curves corresponding to different positions represent the change trends of the CO2 concentrations at different positions in the fishing boat engine room in the future for a period of time.

[0057] Furthermore, the random forest algorithm makes predictions by constructing multiple decision trees. Each decision tree is trained on a random subset of the data and the final prediction result is synthesized by voting or averaging. By combining the prediction results of multiple decision trees, the overfitting problem that may exist in a single decision tree is reduced, and the generalization ability of the random forest algorithm is improved. The random forest takes random sampling during the training process, randomly draws multiple sample subsets from the original training dataset with replacement, and each subset is used to train a decision tree. Since it is sampling with replacement, some samples will appear repeatedly in different trees, while some samples will not appear in the training set of a certain tree. Moreover, at each split node of each decision tree, a part of the features are randomly selected instead of using all features to determine the best split. This can avoid over-reliance on certain features and improve the stability and generalization ability of the random forest model.

[0058] In specific implementation, based on the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor, the random forest algorithm is used to predict the CO2 concentration change trends at multiple positions, and multiple random forest regression curves corresponding to the multiple positions are obtained, including:

[0059] (1) For each position in the fishing vessel engine room, using the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor at each position during the first time period as inputs, and the CO2 concentration at each position during the second time period as the output, a training set is constructed; the end time of the first time period is the same as the start time of the second time period.

[0060] Specifically, the first time period and the second time period are two adjacent time periods, that is, the end time of the first time period is the same as the start time of the second time period.

[0061] In specific implementation, from the past historical time periods, any two adjacent time periods are selected as the first time period and the second time period, ensuring that the end time of the first time period is the same as the start time of the second time period. Further, the CO2 concentration, the CO2 concentration influencing factor, the weight of the CO2 concentration influencing factor at each position in the fishing vessel engine room during the first time period, and the CO2 concentration at each position in the fishing vessel engine room during the second time period are determined from the data obtained in the above embodiments. Using the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor at each position in the fishing vessel engine room during the first time period as inputs, and the CO2 concentration at each position in the fishing vessel engine room during the second time period as the output, a data pair of the training set is formed.

[0062] (2) Initialize the random forest model, train each decision tree in the random forest model based on the training set, each decision tree outputs a corresponding CO2 concentration at each position in the future time period based on the input of the training set, comprehensively determine the CO2 concentration at each position in the future time period according to the output results of each decision tree, and connect the CO2 concentrations at each position in multiple future time periods to obtain the CO2 concentration change trend at each position.

[0063] In specific implementation, the parameters of the random forest model are set, including the number of decision trees (n_estimators), the maximum depth of the decision trees (max_depth), the maximum number of features for each decision tree (max_features), etc. Multiple decision trees are initialized, and each decision tree learns based on the input data and output data in the training set. Further, each decision tree is trained based on the training set data. Each decision tree uses a randomly selected part of the samples from the training set (i.e., the input data, including CO2 concentration, CO2 concentration influencing factors, and weights of CO2 concentration influencing factors) during training, and uses a randomly selected subset of features at each splitting node to build a prediction model. The output data of each decision tree during training is the true label (CO2 concentration) in the training set. Each decision tree learns how to predict the target output (CO2 concentration in future time periods) based on the input data. Each decision tree learns the relationship between the input data and the target output by continuously splitting nodes, and finally obtains a model that can predict future concentrations based on historical data. For each location, the prediction results of all decision trees are integrated, and the final prediction value is obtained by calculating the average of all prediction results or using weighted average, that is, the CO2 concentration at each location in multiple future time periods. The CO2 concentrations at each location in multiple future time periods are concatenated to obtain the CO2 concentration change trend at each location.

[0064] (3) Input the CO2 concentration, CO2 concentration influencing factors, and weights of CO2 concentration influencing factors at multiple locations to be predicted into the trained random forest model. The random forest model predicts the change trends of the CO2 concentrations at multiple locations, and obtains multiple random forest regression curves corresponding to multiple locations.

[0065] In specific implementation, the CO2 concentration, CO2 concentration influencing factors, and weights of CO2 concentration influencing factors at multiple locations to be predicted are used as input data and input into the trained random forest model. Through the trained random forest model, the predicted values of the CO2 concentrations at each location in multiple future time periods are output. The predicted values of the CO2 concentrations at each location in multiple future time periods are concatenated to obtain the change trends of the CO2 concentrations at multiple locations, that is, multiple random forest regression curves.

[0066] Optionally, after obtaining the multiple random forest regression curves corresponding to the multiple locations, the method further includes: performing a first correction on each random forest regression curve; the first correction includes: establishing a structural model of the fishing vessel engine room, determining the ventilation points and their corresponding ventilation volumes corresponding to the first ventilation mode according to the structural model; determining the CO2 monitoring location points affected by ventilation according to the ventilation points and the structural model, obtaining multiple corrected location points, and correcting the random forest regression curves of each of the corrected location points according to the ventilation volume.

[0067] Specifically, for each correction position point, calculate the first distance from the correction position point to the ventilation-affecting point. Each correction position point can correspond to one or more ventilation points. Using the reciprocal of the first distance as the correction coefficient, correct the ventilation volume to obtain the actual ventilation volume at the correction position point. If a correction position point corresponds to multiple ventilation points, calculate the actual ventilation volume of each ventilation point for this correction position point and then sum them up to obtain the actual ventilation volume at this correction position point. Correct the random forest regression curve according to the actual ventilation volume.

[0068] The method provided in this embodiment further improves the accuracy of CO2 concentration prediction by performing the first correction on the random forest regression curves corresponding to multiple positions, especially after considering the structure and ventilation factors of the fishing vessel engine room. By establishing an engine room structure model and combining the ventilation mode, the ventilation points and their corresponding ventilation volumes can be identified, thereby determining the ventilation position points that affect the CO2 concentration. Correcting according to the relationship between the distance between each correction position point and the ventilation point and the ventilation volume can make the CO2 concentration prediction at each position more conform to the air flow and ventilation conditions in the actual environment. Especially the correction of the ventilation volume can accurately reflect the influence of air flow on the change of CO2 concentration, and then correct the output of the random forest regression curve, making the prediction result of the model more accurate. If a correction position point corresponds to multiple ventilation points, the system will comprehensively correct according to the reciprocal of the distance and the ventilation volumes of each ventilation point, further improving the adaptability and prediction accuracy for complex ventilation environments. This method makes the model dynamically consider the actual changes in the air flow in the engine room by introducing ventilation factors, avoiding the errors that may be brought by traditional models ignoring environmental factors, so that the prediction of CO2 concentration is more in line with the actual situation, and provides a more reliable basis for the prediction of the safe evacuation time. Generally speaking, this setting improves the prediction accuracy and the intelligent level of the system, can better adapt to complex engine room structures and ventilation conditions, ensure that the evacuation time can be accurately calculated under any environmental changes, and improve the safety and decision-making efficiency.

[0069] S104. Compare the multiple random forest regression curves with the CO2 propagation curve obtained by simulation to determine the safe evacuation time and activate the second alarm.

[0070] Specifically, the safe evacuation time refers to the time when the crew on the fishing vessel can safely evacuate the fishing vessel. The CO2 concentration varies at different positions on the fishing vessel, and the corresponding safe evacuation times for the crew at different positions on the fishing vessel also vary. The second alarm is to broadcast the safe evacuation time on the fishing vessel, notify the personnel on the fishing vessel to evacuate the fishing vessel as soon as possible, and turn on the ventilation system, setting the ventilation system to the second ventilation mode. As an optional method, the ventilation mode at different positions is determined according to the random forest regression curve. The ventilation modes at positions with different CO2 concentrations are different, and the ventilation intensity corresponding to the ventilation mode is proportional to the CO2 concentration.

[0071] Furthermore, the CO2 propagation curve is the changing trend of the CO2 concentration with time and space distribution in a specific environment (such as the engine room of a fishing vessel). It describes the process of CO2 diffusion starting from the source (such as the leakage point), takes into account factors such as spatial structure, air flow, and ventilation effect, and shows the change of the CO2 concentration in the engine room of the fishing vessel over time. This curve is usually generated based on fluid mechanics, aerodynamic models, or simulation results, and can predict the distribution of the CO2 concentration at different positions and its changing trend under specific conditions.

[0072] When specifically implemented, comparing the CO2 propagation curve obtained by simulation with multiple random forest regression curves to determine the safe evacuation time includes: traversing each position, determining the associated position where the difference in CO2 concentration from the position is less than a preset difference based on the CO2 propagation curve; calculating the distance between the associated position and the position, and determining a correction coefficient based on the distance; calculating the actual concentration reaching the position based on the correction coefficient and the random forest regression curve corresponding to the position.

[0073] Specifically, at each position in the fishing vessel's engine room, the CO2 concentration at each position is determined using the simulated CO2 propagation curve. For the CO2 concentration at each position, by comparing it with the CO2 concentrations at other positions in the fishing vessel's engine room, the associated positions where the difference in CO2 concentration from the currently traversed position is less than the preset difference are determined, that is, other positions where the change in CO2 concentration is relatively close to the currently traversed position. For each associated position, the spatial distance between the associated position and the currently traversed position is calculated through the spatial structure model in the engine room. Further, based on the calculated distance and the relationship between the distance and the correction coefficient, the amplitude of the concentration change is adjusted. The correction coefficient is determined according to the distance and the CO2 concentration propagation law. The closer the area, the larger the correction coefficient; while for the farther area, the correction coefficient is relatively smaller. Based on the correction coefficient, the random forest regression curve is adjusted, and the product of the CO2 concentration at each position and the correction coefficient is determined as the adjusted CO2 concentration at each position, that is, the actual CO2 concentration at each position. The actual CO2 concentration at each position and the adjusted CO2 concentration at each position are summed, and the sum value is determined as the corrected CO2 concentration at each position. Based on the comparison between the corrected CO2 concentration at each position and the set safety threshold, the safe evacuation time is determined.

[0074] The method provided in this embodiment determines the safe evacuation time by combining multiple random forest regression curves and the simulated CO2 propagation curve. First, the present application can accurately consider the local environmental differences in the engine room. By traversing each position and calculating the difference in CO2 concentration from other positions, the system can identify areas similar to the target position, thus effectively determining the associated positions. This provides a more accurate prediction of the CO2 concentration change at each position, rather than relying solely on a single regression model or simulation curve. Second, after introducing the correction coefficient, the output of the random forest regression curve can be adjusted to make it more in line with the actual environmental conditions. The correction coefficient is calculated based on factors such as the distance between positions and the air flow path, ensuring the accuracy of the concentration prediction, enabling the system to dynamically adapt to the changes in the engine room environment, considering different spatial layouts and fluid mechanics factors. Since this method finely adjusts the prediction results, it can accurately determine which areas are about to exceed the CO2 concentration standard, thus more accurately calculating the safe evacuation time, avoiding premature or late evacuation, ensuring the safety of the crew and reducing resource waste. In addition, through the corrected concentration calculation, this method can not only optimize the evacuation time, but also propose more reasonable evacuation strategies for different areas, effectively improving the evacuation efficiency and avoiding unnecessary personnel or resource waste. This optimization strategy also enables the system to adjust the evacuation plan according to the specific conditions of each position, ensuring that the evacuation decision is more accurate, thereby enhancing the intelligence level and response ability of the entire system.

[0075] Optionally, determining the safe evacuation time based on the corrected CO2 concentration at each location includes: for each corrected CO2 concentration at a location, determining the time corresponding to the target danger threshold from the random forest regression curve corresponding to that location as the evacuation time for each location; the target danger threshold being the danger threshold of the CO2 concentration at that location; comprehensively comparing the evacuation times of each location and determining the one with the smallest value as the safe evacuation time.

[0076] Specifically, the danger threshold refers to the concentration value at which, under a certain specific condition, when the CO2 concentration reaches or exceeds this value, it will pose a serious threat to human health or life safety. In the fishing vessel engine room, when the CO2 concentration exceeds a certain danger threshold, it may cause crew members to be poisoned, suffocated or suffer other hazards, and even may cause death. It should be noted that the danger thresholds of CO2 concentrations at different locations are different.

[0077] In specific implementation, on the CO2 propagation curve, determine at which moments the CO2 concentration exceeds different human health hazard standards (such as 5000 ppm, 10000 ppm, etc.). Combine the CO2 propagation curve to determine the time required for the CO2 concentration to reach a certain danger value, and determine the corresponding concentration value as the danger threshold according to the human health hazard standard.

[0078] In specific implementation, determining the safe evacuation time based on the danger thresholds of CO2 concentrations at different locations and multiple random forest regression curves includes: for the CO2 concentration at each location, determining the time corresponding to the target danger threshold from the random forest regression curve corresponding to that location as the evacuation time for each location; the target danger threshold being the danger threshold of the CO2 concentration at that location; comprehensively comparing the evacuation times of each location and determining the one with the smallest value as the safe evacuation time.

[0079] Specifically, the target danger threshold is the danger threshold of the CO2 concentration at each location. For each location in the fishing vessel engine room, each location corresponds to a random forest regression curve and a target danger threshold. Therefore, the evacuation times for each location are different.

[0080] In specific implementation, for each corrected CO2 concentration at a location, according to the forest regression curve corresponding to that location, determine the time corresponding to the target danger threshold of that location as the evacuation time. Traverse the corresponding forest regression curve to determine the time point when the CO2 concentration reaches or exceeds the target danger threshold, and use this time point as the evacuation time for that location. Repeat this process to obtain the evacuation times for each location. By comparing the evacuation times of all locations, determine the smallest evacuation time among them, that is, the safe evacuation time, and determine the safe evacuation time as the safe evacuation time for the entire fishing vessel.

[0081] Optionally, connect the position points with the longest safe evacuation time at different stages to form an evacuation route, and guide the crew on the fishing vessel to evacuate based on the evacuation route. By connecting the position points with the longest safe evacuation time at different stages to form an evacuation route, the main purpose of this application is to improve the efficiency of safe evacuation, ensure that the crew evacuates along the most suitable route, avoid areas with high concentrations of CO2, and safeguard the lives of the crew. Specifically, when implementing, first, for the CO2 concentration at each position, use the corresponding random forest regression curve to determine the time corresponding to the target danger threshold as the evacuation time for each position. Then, by comprehensively comparing the evacuation times of each position, determine the minimum evacuation time as the overall safe evacuation time. This minimum value represents the earliest time point for evacuation, ensuring that the most critical areas are evacuated in the shortest time. However, to further improve the evacuation efficiency, connect the position points with the longest safe evacuation time at different stages to form a reasonable evacuation route. This route is generated based on the sorting of the evacuation times of each position, usually starting from the position with the longest safe evacuation time, ensuring that the crew can choose the safest and most effective route during each stage of evacuation. The generation of the evacuation route can take into account factors such as the passage layout, air flow direction, and personnel distribution inside the ship to avoid areas with high CO2 concentrations. By optimizing the evacuation route, the crew can reach the safe area faster and more safely, reducing potential risks.

[0082] Optionally, the first alarm is a light alarm. When the CO2 concentration reaches the first alarm threshold, the light alarm device on the fishing vessel emits a light alarm, and the flashing frequency of the light alarm changes dynamically based on the alarm level; the second alarm is a sound and light alarm. When the CO2 concentration reaches the second alarm threshold, the light alarm device and the sound alarm device on the fishing vessel simultaneously emit a light alarm and a sound alarm. The volume of the sound alarm changes dynamically based on the alarm level. The flashing frequency of the light alarm corresponding to the second alarm is greater than that of the first alarm, and the wind speed of the first ventilation mode is less than that of the second ventilation mode.

[0083] Specifically, the first alarm is a light alarm using a light alarm device. When the CO2 concentration at any position in the fishing vessel engine room reaches the first alarm threshold, the light alarm device emits a light alarm. It should be noted that the light alarm device uses a high-brightness LED lamp group, which can provide a strong visual warning in a dim or light-deficient environment. The LED lamp group supports multiple colors and flashing modes, and the LED lamp group sets different colors and flashing frequencies according to different levels of alarms. The flashing frequency of the light alarm changes dynamically based on the alarm level. The higher the alarm level, the faster the flashing frequency of the light alarm.

[0084] The second alarm is to use both the sound alarm device and the light alarm device to give a sound alarm and a light alarm. When the CO2 concentration at any position in the fishing vessel engine room reaches the second alarm threshold, the light alarm device gives a light alarm, and the sound alarm device gives a sound alarm. It should be noted that the sound alarm device uses a high-power speaker and can emit an alarm sound exceeding 100 decibels to ensure that it can be clearly heard by the crew in the noisy engine room environment. At the same time, the alarm sound has multiple frequencies and rhythm patterns, and different sound characteristics can be set according to different levels of alarms to enhance the recognition degree and warning effect. The volume of the sound alarm changes dynamically based on the alarm level. The higher the alarm level, the higher the volume of the sound alarm. The flashing frequency of the light alarm under the second alarm is greater than that of the light alarm under the first alarm, and the wind speed of the corresponding ventilation system under the second alarm is greater than that of the corresponding ventilation system under the first alarm.

[0085] The method provided in this embodiment enhances the alarm effect through multiple sensory means (vision and hearing), and flexibly adjusts the warning intensity according to different alarm levels of the CO2 concentration, which helps to ensure that the crew in the fishing vessel engine room can timely and effectively identify potential dangers in different environments, thereby improving safety. On the one hand, using the light alarm device to give light alarms with different flashing frequencies can ensure that even in a dim or insufficient light environment, the crew can visually detect the danger of rising CO2 concentration in a timely manner. The high brightness and multiple color options of the LED light group can make the alarm more prominent visually and easily distinguish different alarm levels. The dynamic adjustment of the flashing frequency is to enhance the sense of urgency of the alarm by increasing the flashing frequency. The higher the alarm level, the faster the flashing frequency, which can remind the crew of the severity and urgency of the danger. On the other hand, when the CO2 concentration reaches the second alarm threshold, the sound alarm and the light alarm are activated simultaneously, further strengthening the warning effect. The sound of more than 100 decibels emitted by the sound alarm device can effectively penetrate in the noisy environment in the engine room, enabling the crew to hear the alarm no matter where they are. Especially in the case where the crew needs to concentrate on operation or judgment, the high volume and different frequency and rhythm changes of the sound alarm can also help the crew quickly identify the alarm level. The light flashing frequency corresponding to the second alarm is higher than that of the first alarm. This setting makes the visual warning at a higher alarm level more conspicuous and avoids the crew ignoring serious alarms due to visual adaptation. At the same time, the wind speed of the ventilation mode corresponding to the second alarm is greater than that of the first alarm. At a high alarm level, more drastic ventilation measures need to be taken immediately to quickly reduce the CO2 concentration to ensure the safety of the crew. Generally speaking, the combination of light and sound alarms and the differential settings of multi-level alarms and ventilation modes make the alarm system more comprehensive and accurate, and can provide timely, effective and hierarchical warnings and responses at different danger levels, significantly improving the crew's reaction speed and safety guarantee in case of emergency.

[0086] Optionally, when the CO2 concentration reaches the third alarm threshold, increase the flashing frequency of the light alarm emitted by the light alarm device and the volume of the sound alarm emitted by the sound alarm device, and the crew on the fishing boat helps the passengers on the fishing boat to evacuate. The third alarm threshold is greater than the second alarm threshold.

[0087] Specifically, the third alarm threshold is set according to actual needs and is not limited in this embodiment. It should be noted that the third alarm threshold is greater than the second alarm threshold. For example, the third alarm threshold is 1000 ppm.

[0088] In specific implementation, when the CO2 concentration at any position in the fishing boat engine room reaches the third alarm threshold, the light alarm device and the sound alarm device on the fishing boat simultaneously emit a light alarm with a higher flashing frequency and a sound alarm with a higher volume. At the same time, the crew on the fishing boat helps the passengers on the fishing boat to evacuate, and all personnel on the fishing boat evacuate the fishing boat within the safe evacuation time.

[0089] The CO2 concentration monitoring method for fishing vessel engine rooms provided in this embodiment aims to significantly extend the escape time during CO2 leakage and accurately give early warnings of CO2 leakage. The method provided by the present invention combines the goshawk hunting algorithm and the random forest method to predict the CO2 leakage propagation curve. That is, first, the goshawk hunting algorithm is used to estimate the weights of the CO2 propagation influence factors, which can be adapted to various engine room environments, evaluate the influence degrees of various factors such as space and wind direction on propagation, and further rely on the random forest to estimate the CO2 propagation according to the above weights, so as to obtain the most accurate escape time adapted to the current engine room. In addition, in order to accurately warn of the alarm situation in the fishing vessel engine room, this application adopts different alarm modes and different ventilation levels according to different concentration thresholds, enabling the crew to understand the CO2 concentration level only through the alarm degree. At the same time, different ventilation levels can increase the ventilation level at the dangerous location points, improving the safety of the cabin. In the first aspect, when the CO2 concentration at any position in the fishing vessel engine room is monitored to reach the first alarm threshold, the goshawk hunting algorithm is used to efficiently calculate the weights of multiple CO2 concentration influence factors, and then judge the specific influence degrees of these factors on the CO2 concentration. The goshawk hunting algorithm is used in this application to optimize the weights of each CO2 concentration influence factor, thereby realizing adaptive weight allocation. The goshawk hunting algorithm is essentially a heuristic optimization method based on the hunting strategy in nature, which can quickly and effectively search for the optimal solution. Specifically in this application, the goshawk hunting algorithm can automatically adjust and optimize the weights of the CO2 concentration influence factors. Based on real-time data and environmental condition changes, the system can flexibly adjust the influence degrees of these factors on the concentration change, which can avoid the limitations of traditional fixed weight settings, can be adaptively adjusted according to environmental changes, and optimize the accuracy of the alarm. For example, if the space size of the fishing vessel engine room or the volume of the CO2 device changes, these changes may cause the law of CO2 concentration fluctuation to change. The goshawk hunting algorithm can help the system more accurately understand the contributions of these factors to the CO2 concentration change by optimizing the weight of each influence factor, so as to more accurately set the alarm threshold and early warning level. This adaptive ability enables the system to continuously maintain high-efficiency monitoring and response capabilities in a changing environment, avoiding the problem that traditional methods cannot cope with environmental changes due to static weight settings. In the second aspect, when the CO2 concentration at any position in the fishing vessel engine room is monitored to reach the second alarm threshold, the random forest algorithm is used to predict the CO2 concentration change trends at multiple future positions. The random forest algorithm can handle complex multi-dimensional inputs, has strong fault tolerance, and can still maintain good prediction performance when there is noise in the data. Compared with traditional regression analysis, the random forest algorithm has more advantages in dealing with multi-factor and high-complexity non-linear relationships. By training the random forest model, the system can not only predict the current CO2 concentration level, but also predict the CO2 concentration change trends in a future period of time.This is particularly important for the fishing vessel engine room because the change in CO2 concentration usually does not occur instantaneously but is a gradual accumulation process. Through multi-dimensional and multi-location data analysis, the random forest algorithm can generate multiple regression curves, which can provide accurate predictions of future CO2 concentrations. The crew can make timely decisions based on this predictive information, such as starting the ventilation system in advance or adjusting the operation process to avoid a further increase in CO2 concentration. Thirdly, in this application, by comparing the random forest regression curve with the CO2 propagation curve obtained through simulation, the safe evacuation time is calculated. Through scientific simulation and prediction, a more accurate evacuation time is provided for the crew. This calculation is not just a simple reaction after the CO2 concentration reaches a certain fixed value, but by dynamically simulating the diffusion process of CO2 throughout the engine room and combining the CO2 concentration changes at different locations, the safest evacuation timing under the current concentration and conditions is accurately estimated. Once the safe evacuation time is calculated, the system will trigger the second alarm and announce the safe evacuation time, providing the crew with an early time window and avoiding the emergency panic and missed evacuation opportunities caused by the rapid increase in CO2 concentration. By knowing the safe evacuation time in advance, the crew can take response measures in an orderly manner, not only increasing the survival rate of personnel but also reducing the misoperations or delays that may occur under high-pressure situations. Fourthly, in this application, by deploying CO2 sensors at multiple locations in the fishing vessel engine room, comprehensive monitoring of the CO2 concentration in each key area of the fishing vessel engine room can be achieved. The advantage of this design is that it avoids the limitations of single-location monitoring. The fishing vessel engine room usually has a complex spatial layout and local air flow phenomena, and the CO2 concentration may vary significantly at different locations. Especially in the case of poor ventilation or leakage, the CO2 concentration in some areas may rise rapidly while that in other areas changes less. Multi-point monitoring can accurately obtain the concentration data of different areas, helping to provide a comprehensive understanding of the overall engine room environment and avoiding missing potential danger points. In addition, by obtaining this multi-location data and combining other CO2 concentration influencing factors, the system can more accurately identify and judge the impact of each factor on the CO2 concentration change, thereby performing more precise early warning and processing. This comprehensive and dynamic monitoring method greatly improves the sensitivity and response speed of the system. Fifthly, this application adopts a hierarchical alarm mechanism. When the CO2 concentration reaches the first alarm threshold, the system will activate the first alarm and the primary ventilation mode. When the CO2 concentration further rises to reach the second alarm threshold, the system will not only increase the alarm level but also start a more efficient ventilation system. This multi-level alarm design enables the system to make different responses according to different danger levels of the CO2 concentration, making the alarms at the corresponding levels play corresponding warning roles and avoiding the problem of starting the ventilation system too early or too late.The ventilation system in the first alarm stage is set to a lower wind speed, which means that the system can provide basic ventilation protection at lower concentrations to prevent the further rise of CO2 concentration. When the second alarm is triggered, the system will start a ventilation mode with a higher wind speed to quickly reduce the CO2 concentration in the engine room. The efficient ventilation system can be linked with the alarm system to quickly reduce the concentration, reduce the exposure time of the crew in the dangerous environment, and provide a safer working environment for the crew.

[0090] Corresponding to the embodiment of the method for monitoring the CO2 concentration in the engine room of a fishing boat described above, the present application also provides an embodiment of a device for monitoring the CO2 concentration in the engine room of a fishing boat.

[0091] Figure 2 It is a schematic structural diagram of the device for monitoring the CO2 concentration in the engine room of a fishing boat provided in the second embodiment of the present application. Please refer to Figure 2 The device provided in this embodiment includes an acquisition module 210, a calculation module 220, a prediction module 230, and a determination module 240;

[0092] Among them, the acquisition module 210 is used to acquire the CO2 concentration and the CO2 concentration influencing factors at multiple positions in the engine room of the fishing boat, and simulate the CO2 propagation curve in the engine room; among them, based on the ranking of the influence degree on the change of the CO2 concentration, a preset number of CO2 concentration influencing factors are determined;

[0093] The calculation module 220 is used to start the first alarm when the CO2 concentration at any position reaches the first alarm threshold, monitor the change of the CO2 concentration in real time, and calculate the weights of each CO2 concentration influencing factor based on the goshawk hunting algorithm; the weights of different CO2 concentration influencing factors are different, and the weight represents the influence degree of the CO2 concentration influencing factor on the CO2 concentration. The first alarm is to give a safety warning on the fishing boat and set the ventilation system to the first ventilation mode;

[0094] The prediction module 230 is used to, when the CO2 concentration reaches the second alarm threshold, predict the change trend of the CO2 concentration at multiple positions based on the CO2 concentration, the CO2 concentration influencing factors, and the weights of the CO2 concentration influencing factors by using the random forest algorithm, and obtain multiple random forest regression curves corresponding to the multiple positions; the first alarm threshold is less than the second alarm threshold;

[0095] The determination module 240 is used to compare multiple random forest regression curves with the simulated CO2 propagation curve to determine the safe evacuation time and start the second alarm. The second alarm is to broadcast the safe evacuation time on the fishing boat and set the ventilation system to the second ventilation mode.

[0096] The device of this embodiment can be used to execute Figure 1The steps of the method embodiments shown are similar in specific implementation principles and implementation processes, and will not be elaborated here.

[0097] Optionally, the device further includes a power management module, which is used to provide power for the operation of the device. The power management module includes a lithium battery and a solar charging panel, and the lithium battery is connected to the solar charging panel;

[0098] The solar charging panel is made of monocrystalline silicon or polycrystalline silicon, and is installed on the outer wall of the fishing boat engine room, and is used to convert the received solar energy into electrical energy to charge the lithium battery;

[0099] The lithium battery is used to provide power for each device of the device by discharging;

[0100] The power management module is further used to monitor the power of the lithium battery in real time, start the solar charging panel to charge the lithium battery when the power is lower than the first preset power, and close the charging circuit to stop charging when the power reaches the second preset power.

[0101] Specifically, the CO2 concentration monitoring device for the fishing boat engine room further includes a power management module, which is used to provide power for the operation of the CO2 concentration monitoring device for the fishing boat engine room. The power management module includes a lithium battery and a solar charging panel. The power management module uses a high-performance lithium battery as the main power source, and the lithium battery has characteristics such as high energy density, long life, and low self-discharge rate. The lithium battery can provide a stable voltage output to ensure that each device of the entire device maintains high performance during long-term operation.

[0102] Furthermore, in order to solve the power supply problem of the fishing boat during long-term offshore operations, the power management module is equipped with an efficient solar charging panel. The solar charging panel is installed outside the fishing boat engine room and is made of monocrystalline silicon or polycrystalline silicon materials, with a high photoelectric conversion efficiency, and can effectively generate electricity under various lighting conditions. Through the solar charging panel, the entire device can use solar energy to charge the lithium battery during the day, extend the battery life, and reduce the dependence on the ship's power grid. The power management module is built-in with advanced power management chips and algorithms, which can monitor the lithium battery power, charging status, and load conditions in real time, and realize intelligent management of the power supply. When the lithium battery power is lower than the first preset power, the power management module automatically starts the solar charging panel to charge the lithium battery; when the lithium battery is fully charged, the power management module automatically cuts off the charging circuit to prevent overcharging.

[0103] Optionally, the power management module supports multiple power inputs, including solar charging, lithium battery power supply, and external DC power supply. When the main power fails, the power management module can automatically switch to the backup power supply to ensure the continuous operation of the device. The multiple power input design also supports redundant backup, further improving the fault tolerance of the device.

[0104] For the implementation processes of the functions and roles of each unit in the above device, please refer to the implementation processes of the corresponding steps in the above method for details, which will not be elaborated here.

[0105] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can be referred to the partial description of the method embodiment. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0106] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included within the scope of protection of this application.

Claims

1. A method for monitoring CO2 concentration in a fishing vessel engine room, characterized in that: The method comprises: The CO2 concentration and CO2 concentration influencing factors at multiple locations in the engine room of the fishing boat are obtained, and the CO2 propagation curve in the engine room is simulated; wherein, based on the ranking of the degree of influence on the CO2 concentration change, a preset number of CO2 concentration influencing factors are determined; When the CO2 concentration at any location reaches the first alarm threshold, the first alarm is activated, the CO2 concentration change is monitored in real time, and the weights of various CO2 concentration influencing factors are calculated based on the goshawk hunting algorithm; different CO2 concentration influencing factors have different weights, and the weights represent the degree of influence of the CO2 concentration influencing factors on the CO2 concentration. The first alarm is a safety warning on the fishing boat and the ventilation system is set to the first ventilation mode; When the CO2 concentration reaches the second alarm threshold, based on the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor, a random forest algorithm is used to predict the CO2 concentration change trend of multiple locations, and multiple random forest regression curves corresponding to the multiple locations are obtained; the first alarm threshold is less than the second alarm threshold; Based on the comparison between the multiple random forest regression curves and the CO2 propagation curve obtained by simulation, the safe evacuation time is determined and the second alarm is activated. The second alarm is to broadcast the safe evacuation time on the fishing boat and set the ventilation system to the second ventilation mode.

2. The method according to claim 1, characterized in that: The weights of various CO2 concentration influencing factors are calculated based on the goshawk hunting algorithm, including: The CO2 concentration influencing factor is determined as a population individual, and the weight of the CO2 concentration influencing factor is determined as the fitness solution of the population individual; wherein the corresponding predicted CO2 concentration under different population individual combinations is calculated, and the mean square error between the actual CO2 concentration and the predicted CO2 concentration is determined as the weight of each CO2 concentration influencing factor; Each first population individual randomly identifies other second population individuals, updates the first population individuals and the fitness solutions of the first population individuals based on the fitness solutions of the second population individuals and the fitness solutions of the first population individuals, and adjusts the first population individuals and the corresponding fitness solutions based on the fitness solutions of the first population individuals after the update and the fitness solutions of the first population individuals before the update; An update radius is determined based on the number of updates of the first population individuals; the first population individuals and the fitness solutions of the first population individuals are updated based on the update radius and the adjusted first population individuals; the first population individuals and the corresponding fitness solutions are adjusted based on the fitness solutions of the updated first population individuals and the fitness solutions of the first population individuals before the update; when the number of updates reaches the maximum number of updates, the weight of each of the CO2 concentration influencing factors is determined based on the fitness solutions of each of the first population individuals.

3. The method according to claim 1, characterized in that The method of predicting the CO2 concentration change trend at multiple locations using a random forest algorithm based on the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor, and obtaining multiple random forest regression curves corresponding to the multiple locations, includes: For each position in the engine room of the fishing boat, the CO2 concentration, the CO2 concentration influencing factor and the weight of the CO2 concentration influencing factor of each position in the first time period are used as input, and the CO2 concentration of each position in the second time period is used as output to construct a training set; the end time of the first time period is the same as the start time of the second time period; Initialize the random forest model, train each decision tree in the random forest model based on the training set, each decision tree outputs a corresponding CO2 concentration of each location in the future time period based on the input of the training set, determine the CO2 concentration of each location in the future time period by combining the output results of each decision tree, connect the CO2 concentrations of each location in multiple future time periods, and obtain the CO2 concentration change trend of each location; The CO2 concentrations at multiple locations to be predicted, CO2 concentration influencing factors, and weights of CO2 concentration influencing factors are input into a trained random forest model. The random forest model predicts the changing trends of CO2 concentrations at multiple locations, and obtains multiple random forest regression curves corresponding to the multiple locations.

4. The method according to claim 1, characterized in that: The step of comparing the multiple random forest regression curves with the simulated CO2 propagation curve to determine the safe evacuation time includes: Traversing each position, and determining, based on the CO2 propagation curve, an associated position having a CO2 concentration difference with the position that is less than a preset difference; calculating a distance between the associated position and the position, and determining a correction factor based on the distance; Calculating the actual concentration reaching the location based on the correction coefficient and the random forest regression curve corresponding to the location; Calculating the sum of the actual concentration and the CO2 concentration at the location, and determining the sum as the corrected CO2 concentration at the location; A safe evacuation time is determined based on the corrected CO2 concentration at the location.

5. The method according to claim 4, characterized in that The determining of the safe evacuation time based on the corrected CO2 concentration at the location includes: For the corrected CO2 concentration at each location, determine the time corresponding to the target danger threshold from the random forest regression curve corresponding to the location as the evacuation time for each location; the target danger threshold is the danger threshold of the CO2 concentration at the location; The evacuation time of each location is comprehensively compared, and the smallest value is determined as the safe evacuation time.

6. The method according to claim 1, characterized in that The first alarm is a light alarm. When the CO2 concentration reaches the first alarm threshold, the light alarm device on the fishing boat sends out a light alarm. The flashing frequency of the light alarm changes dynamically based on the alarm level. The second alarm is a sound and light alarm. When the CO2 concentration reaches the second alarm threshold, the light alarm device and the sound alarm device on the fishing boat will simultaneously send out a light alarm and a sound alarm. The volume of the sound alarm changes dynamically based on the alarm level. The flashing frequency of the light alarm corresponding to the second alarm is greater than that of the first alarm, and the wind speed of the first ventilation mode is less than that of the second ventilation mode.

7. The method according to claim 6, characterized in that When the CO2 concentration reaches the third alarm threshold, the flashing frequency of the light alarm emitted by the light alarm device and the volume of the sound alarm emitted by the sound alarm device are increased, and the staff on the fishing boat help the passengers of the fishing boat to evacuate. The third alarm threshold is greater than the second alarm threshold.

8. The method according to claim 1, characterized in that: The method of obtaining CO2 concentrations at multiple locations in the engine room of the fishing vessel includes: Acquire the CO2 concentrations at multiple first positions in the engine room of the fishing boat, where the ratio of the union of the monitoring areas of the sensors at the multiple first positions to the area of ​​the engine room of the fishing boat is greater than a preset threshold; Establishing a spatial structure model of the fishing boat engine room and establishing a propagation curve of CO2 in the spatial structure model; Predicting a CO2 increase at each first location based on the propagation curve and the CO2 concentration at the plurality of first locations; According to the CO2 increase rate, multiple positions having an increase rate greater than a preset threshold are selected to obtain a second position; The CO2 concentration at the second location in the engine room of the fishing vessel is obtained.

9. A CO2 concentration monitoring device for a fishing boat engine room, characterized in that: The device includes an acquisition module, a calculation module, a prediction module and a determination module; The acquisition module is used to acquire the CO2 concentration and CO2 concentration influencing factors at multiple locations in the engine room of the fishing boat, and simulate the CO2 propagation curve in the engine room; based on the ranking of the degree of influence on the CO2 concentration change, a preset number of CO2 concentration influencing factors are determined; The calculation module is used to activate the first alarm when the CO2 concentration at any position reaches the first alarm threshold, monitor the CO2 concentration change in real time, and calculate the weight of each CO2 concentration influencing factor based on the goshawk hunting algorithm; different CO2 concentration influencing factors have different weights, and the weights represent the degree of influence of the CO2 concentration influencing factors on the CO2 concentration. The first alarm is to issue a safety warning on the fishing boat and set the ventilation system to the first ventilation mode; The prediction module is used to predict the CO2 concentration change trend of multiple locations using a random forest algorithm based on the CO2 concentration, the CO2 concentration influencing factor, and the weight of the CO2 concentration influencing factor when the CO2 concentration reaches the second alarm threshold, and obtain multiple random forest regression curves corresponding to the multiple locations; the first alarm threshold is less than the second alarm threshold; The determination module is used to determine the safe evacuation time based on comparison between the multiple random forest regression curves and the simulated CO2 propagation curve, and activate the second alarm, wherein the second alarm is to broadcast the safe evacuation time on the fishing boat and set the ventilation system to the second ventilation mode.

10. The device according to claim 9, characterized in that The device further comprises a power management module, the power management module is used to provide power for the operation of the device, the power management module comprises a lithium battery and a solar charging panel, the lithium battery is connected to the solar charging panel; The solar charging panel is made of monocrystalline silicon or polycrystalline silicon and is installed on the outer wall of the fishing boat engine room to convert the received solar energy into electrical energy to charge the lithium battery; The lithium battery is used to provide power to various devices of the device through discharge; The power management module is also used to monitor the power level of the lithium battery in real time, start the solar charging panel to charge the lithium battery when the power level is lower than a first preset power level, and close the charging circuit and stop charging when the power level reaches a second preset power level.

Images