Nickel ore ship transportation route optimization method considering risk and emission

Abstract

The invention discloses a ship transportation nickel ore route optimization method considering risk and emission, and the method comprises the steps: S1, obtaining risk factors affecting nickel ore liquefaction in marine transportation of nickel ore, and taking the risk factors as network nodes to construct an initial Bayesian network affecting a nickel ore liquefaction risk state; s2, obtaining historical event data of nickel ore marine transportation, and training the initial Bayesian network based on the historical event data to obtain a Bayesian risk inference network; s3, performing sensitivity analysis on each network node according to the conditional probability to obtain a key network node, and obtaining a risk numerical map of a ship transportation nickel ore route based on the key network node; s4, on the basis of the risk value map, assumed conditions for ship nickel ore transportation route optimization are set; s5, constructing a ship route transportation multi-objective optimization model according to assumed conditions; and S6, taking the lowest ship route transportation risk, the lowest energy consumption and the lowest emission as optimization objectives, based on an improved NSDBO algorithm, according to the ship route transportation multi-objective optimization model, solving and obtaining an optimal solution corresponding to the optimization objectives, and according to the optimal solution, realizing ship transportation nickel ore route optimization considering the risk and the emission. The nickel ore liquefaction risk assessment system solves the problem that a process for systematic risk assessment of nickel ore liquefaction is lacked in previous researches, and meanwhile, in the aspect of measures for coping with nickel ore liquefaction risks, previous researchers tend to reduce the probability of nickel ore liquefaction and the caused adverse effects by improving a ship structure and adjusting ship safety management, and the risk assessment efficiency is improved. And the method for improving the nickel ore transportation safety and environmental performance by optimizing the route design is lacked.

Description

Technical Field

[0001] The present invention relates to the technical field of ship route optimization, and in particular to a method for optimizing a ship route for transporting nickel ore taking risks and emissions into consideration. Background Art

[0002] Nickel ore is of great significance in key areas such as modern battery technology and stainless steel production, and its transportation is crucial to the stability of the global supply chain. However, the maritime transportation of nickel ore faces many challenges, especially the liquefaction of nickel ore during transportation, which reduces the stability of the ship and may even cause the ship to capsize or sink, posing a serious threat to the safety of the crew and the environment. Therefore, it is particularly important to optimize the safety and environmental impact of nickel ore transportation.

[0003] Past studies have mainly focused on the analysis of the causes of nickel ore transportation accidents and the analysis of the impact of nickel ore liquefaction on ship stability in combination with ship motion theory. These studies have emphasized the importance of factors such as cargo type, weather conditions and ship structure in liquefaction risk, as well as the importance of water limit (TML) for nickel ore transportation. However, previous studies lacked a systematic risk assessment process for nickel ore liquefaction. At the same time, in terms of measures to deal with the risk of nickel ore liquefaction, previous researchers tended to reduce the probability of nickel ore liquefaction and the adverse effects caused by improving the hull structure and adjusting ship safety management, and lacked the method of optimizing route design to improve the safety and environmental friendliness of nickel ore transportation. Summary of the invention

[0004] The present invention provides a method for optimizing a route for transporting nickel ore by ship taking risks and emissions into consideration, so as to overcome the above technical problems.

[0005] In order to achieve the above object, the technical solution of the present invention is:

[0006] A method for optimizing a route for transporting nickel ore by ship considering risks and emissions specifically comprises the following steps:

[0007] S1: Obtain risk factors affecting nickel ore liquefaction during nickel ore marine transportation, and use the risk factors as network nodes to construct an initial Bayesian network affecting the risk status of nickel ore liquefaction;

[0008] The risk factors include environmental factors, ship factors and human factors;

[0009] S2: Obtain historical event data of nickel ore marine transportation, and train the initial Bayesian network based on the historical event data to obtain a Bayesian risk inference network;

[0010] The Bayesian risk inference network is used to infer the risk law that affects nickel ore liquefaction during nickel ore marine transportation and obtain the conditional probability of each network node;

[0011] S3: Conduct sensitivity analysis on each network node according to conditional probability to obtain key network nodes, and obtain a risk numerical map of the nickel ore transport route based on the key network nodes;

[0012] S4: Set the assumptions for optimizing the shipping route for nickel ore based on the risk numerical map;

[0013] S5: constructing a multi-objective optimization model for ship route transportation according to the assumptions;

[0014] The multi-objective optimization model for ship route transportation includes a ship fuel consumption model, a ship sulfur emission model and a ship carbon emission model;

[0015] S6: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, solve and obtain the optimal solution of the corresponding optimization objectives according to the multi-objective optimization model of ship route transportation, and optimize the ship transportation nickel ore route considering risks and emissions according to the optimal solution.

[0016] Furthermore, the S1 specifically includes the following steps:

[0017] S11: Obtain risk factors affecting nickel ore liquefaction during marine transportation of nickel ore;

[0018] The risk factors include environmental factors, ship factors and human factors;

[0019] The environmental factors include at least wind speed / direction, temperature and sea waves;

[0020] The ship factors at least include ship age, ship tonnage, ship speed and hull strength;

[0021] The human factors include at least the age of the crew and the corresponding crew safety inspection and crew safety awareness assessment results;

[0022] S12: Based on expert experience, determine the evidence nodes, intermediate nodes and query nodes that affect nickel ore liquefaction according to the risk factors;

[0023] And the characteristics of the evidence nodes are the risk factors affecting nickel ore liquefaction;

[0024] The characteristics of the intermediate node are the nickel ore vibration and the nickel ore water content caused by the risk factors;

[0025] The query node is the liquefaction risk level during nickel ore transportation;

[0026] And obtain the association relationship between evidence nodes, intermediate nodes and query nodes to confirm the edge set between nodes, and then establish the initial Bayesian network that affects the risk status of nickel ore liquefaction.

[0027] Furthermore, the S2 specifically includes the following steps:

[0028] S21: Based on the GeNIe Academic 4.1 software, a corresponding visual Bayesian network was constructed according to the initial Bayesian network;

[0029] S22: Obtain historical event data of nickel ore marine transportation, and import the historical event data into GeNIe Academic 4.1 software based on a data-driven approach;

[0030] The EM algorithm is used to train the visual Bayesian network model to obtain the Bayesian risk reasoning network;

[0031] The Bayesian risk inference network is used to infer the risk patterns that affect nickel ore liquefaction during marine transportation of nickel ore and obtain the conditional probability of each network node.

[0032] Furthermore, in S3, sensitivity analysis is performed on each network node according to conditional probability to obtain key network nodes, specifically:

[0033] Based on the joint tree algorithm, converting the Bayesian risk inference network into a joint tree;

[0034] At the same time, the conditional probability of each network node obtained through the Bayesian risk inference network is converted into a joint tree, and the joint tree that satisfies global consistency is obtained through message passing;

[0035] According to the joint tree that satisfies global consistency, the probability distribution of each network node can be obtained to obtain the key network nodes in the Bayesian risk reasoning network;

[0036] The message transmission mode includes a first transmission process of transmitting information from top to bottom based on the joint tree and a second transmission process of transmitting information from bottom to top based on the joint tree;

[0037] The model formula of the first transfer process is:

[0038]

[0039] The model formula of the second transfer process is:

[0040]

[0041] And the conditions satisfied when the join tree reaches a globally consistent state are

[0042]

[0043] C={C i}; S = {S ij}

[0044] Where: r M represents the product of the evidence function assigned to the network node set and the conditional probability distribution function; kr Represents network node C k Passed to network node C r message; C and S represent the network node set and edge set in the joint tree respectively.

[0045] Furthermore, in S4, based on the risk value map, the assumptions for optimizing the shipping route for nickel ore transportation are set, specifically:

[0046] During the marine weather data update period, it is assumed that wind speed / direction and sea waves remain unchanged;

[0047] During the sea voyage, the ship is considered as a point mass;

[0048] It is assumed that the ship can sail according to the expected route plan.

[0049] Furthermore, a multi-objective optimization model for ship route transportation is constructed in S5, including a ship fuel consumption model, a ship sulfur emission model, and a ship carbon emission model;

[0050] The ship fuel consumption model is:

[0051] P E =f total v1

[0052]

[0053] MF=P B ×SFOC·10 -6

[0054] SFOC=α1·(B ME ) 3 +α2·(B ME ) 2 -α3·B ME +α4

[0055] B ME =P B / P SMCR ×100%

[0056] Where: P E represents the effective power of the ship during navigation; v1 represents the speed of the ship in still water; f totalIt represents the total resistance of the ship at any time during the navigation process; P B Indicates the actual power of the ship's main engine; η S Indicates the working efficiency of the main engine shaft; η D represents the propulsion efficiency of the ship; SFOC represents the conversion rate of the ship's main engine fuel consumption; MF represents the ship's main engine fuel consumption; α1, α2, α3, α4 are all set polynomial fitting coefficients; B ME represents the intermediate parameter variable; P SMCR Indicates the rated power of the ship's main engine;

[0057] The ship sulfur emission model is:

[0058]

[0059] Where: S S represents the total sulfur emission; α SC It represents the sulfur emission coefficient of light diesel; Indicates the consumption of light diesel; β SC It represents the sulfur emission coefficient of heavy diesel; Indicates heavy diesel fuel consumption;

[0060] The ship carbon emission model is:

[0061]

[0062] Where: S S represents the total carbon emissions; α SC Indicates the carbon emission coefficient of light diesel; Indicates the consumption of light diesel; β SC represents the carbon emission coefficient of heavy diesel; Indicates heavy diesel consumption.

[0063] Furthermore, the S6 specifically includes the following steps:

[0064] S61: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, the multi-objective optimization model of ship route transportation is solved and the optimal solution of the corresponding optimization objectives is obtained;

[0065] S611: define and initialize the dung beetle population and algorithm parameters of the dung beetle optimization algorithm;

[0066] The algorithm parameters include the size of the dung beetle population, the maximum number of iterations, the total number of dung beetles, and the ratio of the role dung beetles;

[0067] The role dung beetles include rolling dung beetles and breeding dung beetles;

[0068] S612: The risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore are used as the individual position of a dung beetle in the dung beetle population; and the multi-objective optimization model of ship route transportation is used as the evaluation function for optimizing the individual fitness value of the dung beetle in the process of searching for trajectories;

[0069] And based on the evaluation function, the individual fitness value of each dung beetle individual in the current dung beetle population is obtained;

[0070] S613: performing non-dominated sorting on each dung beetle individual in the current dung beetle population according to the individual fitness value, and obtaining an initial sorted dung beetle population including different non-dominated levels;

[0071] S614: Calculate and obtain the crowding distance between dung beetle individuals in the same non-dominated level in the initially sorted dung beetle population;

[0072] And based on the crowding distance, the dung beetle individuals in the initially sorted dung beetle population are sorted in descending order to obtain the currently sorted dung beetle population;

[0073] S615: Based on the Pareto dominance theory, dominance sorting is performed according to the individual fitness values ​​in the current sorted dung beetle population to obtain the Pareto frontier set;

[0074] And select the best individual in the Pareto front set as the global optimal solution;

[0075] S616: based on the position update strategy, obtaining the updated position of each dung beetle individual in the currently sorted dung beetle population to obtain the next generation of dung beetle population;

[0076] And the next generation of dung beetle population and the currently sequenced dung beetle population are taken as the latest dung beetle population;

[0077] S617: Obtaining a new individual fitness value of each dung beetle individual in the latest dung beetle population based on the evaluation function;

[0078] The latest dung beetle population is used as the currently sorted dung beetle population, and steps S615 to S616 are repeatedly performed until the maximum number of iterations is reached to obtain the final global optimal solution;

[0079] S62: Optimizing the shipping route for nickel ore transportation taking into account risks and emissions according to the final global optimal solution, wherein the final global optimal solution is to optimize the risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore under the constraints of satisfying the optimization objectives of the shipping route with the lowest shipping route risk, the lowest energy consumption and the lowest emissions.

[0080] Furthermore, the non-dominated degree calculation formula for performing non-dominated sorting on each dung beetle individual in the current dung beetle population in S613 is:

[0081]

[0082] Where: n i represents the non-dominance degree of the i-th dung beetle individual; H(x) represents the step function, and x = f j -f i ;f j represents the individual fitness value of the jth dung beetle; f i represents the individual fitness value of the i-th dung beetle individual; N represents the total number of dung beetle individuals.

[0083] Furthermore, in S61, the crowding distance between dung beetle individuals in the same non-dominated level in the initially sorted dung beetle population is calculated and obtained;

[0084] The calculation formula of the crowding distance is:

[0085]

[0086] Where: CDi represents the crowding distance between dung beetle individuals in the same non-dominated hierarchy; M represents the number of dung beetle individuals in any non-dominated hierarchy; and Respectively represent the maximum and minimum values ​​of individual fitness values; and Represents the adjacent solutions of the fitness value of the i-th individual.

[0087] Beneficial effects: The present invention provides a method for optimizing the route of nickel ore transported by ships taking into account risks and emissions. Aiming at the risk assessment and optimization problems in the process of nickel ore transportation, a risk assessment framework based on a Bayesian network is first constructed. The historical event data of nickel ore marine transportation is obtained, and the initial Bayesian network is trained to obtain a Bayesian risk inference network for inferring the risk law affecting nickel ore liquefaction in the marine transportation of nickel ore to obtain the conditional probability of each network node, and the main risk factors in the transportation process are identified and quantified; and the application of the Bayesian network makes the risk assessment not only systematic and scientific, but also able to dynamically reflect the relationship between various risk factors, thereby providing a solid theoretical basis for risk management. Based on the improved NSDBO algorithm, the nickel ore transportation risk, ship fuel consumption, carbon emissions and sulfur emissions are included in the optimization consideration range for multi-objective optimization, and the ship transportation route of nickel ore taking into account risks and emissions is optimized according to the optimal solution of the improved NSDBO algorithm, thereby achieving a dual improvement in transportation safety and environmental protection performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0089] Figure 1 A flow chart of a method for optimizing a route for transporting nickel ore by ship considering risks and emissions according to the present invention;

[0090] Figure 2 This is a core flow chart of the method for optimizing the route of nickel ore transported by ships taking into account risks and emissions in this embodiment;

[0091] Figure 3 This is a schematic diagram of the nickel ore liquefaction process in this embodiment;

[0092] Figure 4 A schematic diagram of the direct logical relationship between the nodes of the Bayesian network of this embodiment;

[0093] Figure 5 Schematic diagram of the Bayesian network structure and conditional probability of each node in the GeNIe software of this embodiment;

[0094] Figure 6 This is a flowchart of the improved NSDBO algorithm in this embodiment;

[0095] Figure 7 It is a Pareto curve comparison diagram of the improved NSDBO algorithm, the multi-objective particle swarm optimization algorithm (MOPSO) and the decomposition-based multi-objective evolutionary algorithm (MOEAD) in this embodiment;

[0096] Figure 8 This is a risk analysis comparison diagram of the improved NSDBO algorithm, the multi-objective particle swarm optimization algorithm (MOPSO) and the decomposition-based multi-objective evolutionary algorithm (MOEAD) in this embodiment;

[0097] Fig. 9 This is a schematic diagram of the impact of different carbon emission prices on ship routes in this embodiment. DETAILED DESCRIPTION

[0098] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0099] This embodiment provides a method for optimizing the route of nickel ore transport by ship considering risks and emissions, such as Figure 1 to Figure 2 As shown, the specific steps include:

[0100] S1: Obtain risk factors affecting nickel ore liquefaction during nickel ore marine transportation, and use the risk factors as network nodes to construct an initial Bayesian network affecting the risk status of nickel ore liquefaction;

[0101] The risk factors include environmental factors, ship factors and human factors;

[0102] In this embodiment, nickel ore liquefaction refers to the physical properties of the relatively dry liquefiable cargo on the surface suddenly changing from a solid state to a viscous fluid state under the influence of the vibration and movement of the ship at sea. The liquefiable cargo refers to the material loaded on the ship in a loose form, which contains fine particles and water deposited between them. Under the action of dynamic loads at sea, the loose materials on the ship tend to be dense, resulting in a decrease in the gaps between the particles and an increase in the pressure of the water in the gaps. With compaction, the friction between the particles decreases, the shear strength decreases, and the cargo finally behaves as a viscous fluid, such as Figure 3 The figure shows the process of nickel ore gradually transitioning from a normal state to a liquefied state during ship transportation;

[0103] The specific steps include:

[0104] S11: Obtain risk factors affecting nickel ore liquefaction during marine transportation of nickel ore;

[0105] The risk factors include environmental factors, ship factors and human factors;

[0106] The environmental factors include at least wind speed / direction, temperature and sea waves;

[0107] The ship factors at least include ship age, ship tonnage, ship speed and hull strength;

[0108] The human factors include at least the age of the crew and the corresponding crew safety inspection and crew safety awareness assessment results;

[0109] S12: Based on expert experience, determine the evidence nodes, intermediate nodes and query nodes that affect nickel ore liquefaction according to the risk factors;

[0110] And the characteristics of the evidence nodes are the risk factors affecting nickel ore liquefaction;

[0111] The characteristics of the intermediate node are the nickel ore vibration and the nickel ore water content caused by the risk factors;

[0112] The query node is the liquefaction risk level during nickel ore transportation;

[0113] And obtain the node association relationship between evidence nodes, intermediate nodes and query nodes to confirm the edge set between nodes, and then establish the initial Bayesian network affecting the risk status of nickel ore liquefaction;

[0114] In this embodiment, by comprehensively considering ship factors, human factors and environmental factors, through analysis of existing literature, combined with the actual working experience of experts and bulk carriers, the evidence nodes of nickel ore liquefaction are set as shown in Table 1 below;

[0115] Table 1. Evidence points for nickel ore liquefaction

[0116]

[0117] To meet data requirements, this embodiment reduces the number of states of most nodes to three and avoids the computational complexity caused by the exponential increase in the number of states. For example, in actual production work, this embodiment divides the ship speed into high speed, medium speed and low speed levels according to the speed of bulk carriers, and the other node states are set as shown in Table 2 below;

[0118] Table 2. Node and node status table

[0119]

[0120] In addition to the above nodes, this embodiment also sets the status of the "nickel ore liquefaction" node to "liquefied" and "not liquefied", sets the status of the wind speed node to 13 levels such as no wind, light wind, gentle wind, strong wind, etc. according to the Pu-type wind scale, and sets the status of the wave node to 9 levels such as no wave, slight wave, small wave, and light wave;

[0121] In this embodiment, the node association relationship between expert experience and evidence nodes, intermediate nodes, and query nodes is as follows: Taking ship speed as a child node, and ship age and crew safety awareness as its parent node, it is based on the following logic:

[0122] (1) The age of a ship usually directly or indirectly affects the performance of its main engine and the condition of its mechanical equipment, which in turn affects its speed;

[0123] (2) The crew's safety awareness usually affects their control of the ship's speed;

[0124] Therefore, the status of the two parent nodes, crew safety awareness and ship age, will affect the status of its child node, ship speed;

[0125] The logical relationship between the three nodes of crew experience, safety inspection and safety awareness is as follows:

[0126] Crew experience affects the number of safety checks and crew safety awareness. Generally, experienced crew members are more likely to conduct more safety checks and have a higher safety awareness.

[0127] The remaining node relationships are similar to the above examples, and are all inferred based on actual conditions and expert experience; then the initial Bayesian network affecting the risk status of nickel ore liquefaction is established, such as Figure 4 The figure shows the relationship between nodes in the proposed Bayesian network model;

[0128] S2: Obtain historical event data of nickel ore marine transportation, and train the initial Bayesian network based on the historical event data to obtain a Bayesian risk inference network;

[0129] The Bayesian risk inference network is used to infer the risk law that affects nickel ore liquefaction during nickel ore marine transportation and obtain the conditional probability of each network node;

[0130] The specific steps include:

[0131] S21: Based on the GeNIe Academic 4.1 software, a corresponding visual Bayesian network was constructed according to the initial Bayesian network;

[0132] S22: Obtain historical event data of nickel ore marine transportation, and import the historical event data into GeNIe Academic 4.1 software based on a data-driven approach;

[0133] The EM algorithm is used to train the visualized Bayesian network model to obtain a Bayesian risk inference network; and the Bayesian risk inference network is used to infer the risk law that affects the liquefaction of nickel ore during the marine transportation of nickel ore, and to obtain the conditional probability of each network node; wherein obtaining the conditional probability of each network node through the Bayesian risk inference network is a prior art, and this embodiment is only used to train the initial Bayesian network through the historical event data of the marine transportation of nickel ore to obtain the final Bayesian risk inference network model; wherein Figure 5 As shown, the research data of this embodiment is based on 200 cases of bulk transportation of laterite nickel ore, which include 80 liquefaction events and 120 cases close to liquefaction or normal transportation conditions;

[0134] S3: Conduct sensitivity analysis on each network node according to conditional probability to obtain key network nodes, and obtain a risk numerical map of the nickel ore transport route based on the key network nodes;

[0135] In a specific embodiment, sensitivity analysis is performed on each network node according to conditional probability to obtain key network nodes, specifically:

[0136] On the basis of calculating the conditional probability of each node, this embodiment uses the sensitivity analysis function in the GeNIe software to perform sensitivity analysis on each node, and the software implementation logic is as follows:

[0137] The sensitivity analysis of the Bayesian network is to analyze the probability change of the query node, that is, the degree of reduction of the information entropy of the query node, by changing the probability estimation of the network evidence node, thereby judging the evidence node with the greatest impact on the query node; the evidence node of this embodiment refers to the observable root node (such as sea breeze, temperature), and the query node refers to the leaf node, that is, the key node risk, the intermediate node (such as ship vibration, cargo moisture content, etc.), and the query node that can be used as a root node or a leaf node is selected according to experience;

[0138] The information entropy is a statistic that describes the degree of discreteness of random variables. The greater the information, the higher the degree of uncertainty of the variable. Its calculation formula is as follows:

[0139]

[0140] It also includes the mutual information used to calculate the strength of the relationship between the query node and the evidence node. It represents the amount of information shared between the two variables and is a measure of the degree of mutual dependence of the variables. It is used to calculate the strength of the relationship between the query node and the evidence node; calculate the degree of reduction of the information entropy of the query node under different states of the evidence node; the mutual information of two discrete random variables X and Y can be defined as:

[0141]

[0142] Where P(x,y) represents the joint distribution probability of variable X and variable Y; p(x) and p(y) represent the marginal probability distribution functions of variable X and variable Y respectively;

[0143] The influencing factors with the greatest degree of dependence on the query node are obtained through mutual information, and the training results of the Bayesian network model parameters are used to perform reasoning analysis on the query node;

[0144]

[0145] Wherein, Q and E represent query nodes and evidence nodes respectively. Bayesian network sensitivity analysis reasoning is an NP-hard problem. Therefore, this embodiment uses the joint tree algorithm to reason about the Bayesian network model sensitivity analysis. Specifically,

[0146] Based on the joint tree algorithm (divided into Shenoy-Shafer algorithm and Hugin algorithm), the Bayesian risk inference network is converted into a joint tree;

[0147] At the same time, the conditional probability of each network node obtained through the Bayesian risk inference network is converted into a joint tree, and the joint tree that satisfies global consistency is obtained through message passing;

[0148] According to the joint tree that satisfies global consistency, the probability distribution of each network node can be obtained to obtain the key network nodes in the Bayesian risk reasoning network;

[0149] The message transmission mode includes a first transmission process of transmitting information from top to bottom based on the joint tree and a second transmission process of transmitting information from bottom to top based on the joint tree;

[0150] The model formula of the first transfer process is:

[0151]

[0152] Where: M rj Represents node C r Passed to Node C j Information; S rj Represents the slave node C in the union tree r To Node C j The edge set of

[0153] The model formula of the second transfer process is:

[0154]

[0155] Where: M ri Represents node C r Passed to Node C i Information; S ir Represents the slave node C in the union tree r To Node C i The edge set of

[0156] And the conditions satisfied when the join tree reaches a globally consistent state are

[0157]

[0158] C={C i}; S = {S ij}

[0159] Where: P(S) represents the probability distribution of S; P(C i ) means C i The probability distribution of S ij Represents node C in the joint tree i To Node C j The edge set of

[0160] Where: rM represents the product of the evidence function assigned to the network node set and the conditional probability distribution function; kr Represents network node C k Passed to network node C r Message; C and S represent the network node set and edge set in the joint tree respectively;

[0161] Among them, when the joint tree reaches a globally consistent state, the Shenoy-Shafer algorithm of the joint tree algorithm realizes the credibility update of the connection tree through the credibility update formula, and the Hugin algorithm realizes the credibility update of the network node C during the message transmission process. r After receiving the i-th neighbor node C i After all the neighboring nodes transmit the probability message, according to the neighboring node C i The potential function Φ i The update formula of passes the message to its i-th neighbor node and updates the neighbor node C i The potential function Φ i ;

[0162] The confidence update formula is:

[0163]

[0164] P(S ri ,e)=M ri ×M ri

[0165]

[0166] Where: P(C r ,e) represents node C r The probability distribution under given evidence e; m represents the neighboring nodes; Represents node C r The local potential function between its neighboring node m; P(S ri ,e) represents node C r To Node C i The probability distribution under given evidence e; P(e) represents the probability and given evidence e;

[0167] The neighboring node C i The potential function Φ i The update formula is

[0168]

[0169] Φ i =Φ i ×M ri

[0170] Where: Φ r Represents node Cr The potential function of

[0171] Based on the above software implementation logic, after implementing sensitivity analysis through software, the sensitivity values ​​of each node are shown in Table 3 below;

[0172] Table 3. Sensitivity analysis numerical table

[0173]

[0174]

[0175] In this embodiment, the sensitivity analysis shows that cargo vibration is the largest parent node that causes nickel ore liquefaction, while at the same time, wind and wave factors (sea waves and swells) have the greatest impact on the vibration of nickel ore;

[0176] In a specific embodiment, a risk numerical map of a route for transporting nickel ore by ship is obtained based on key network nodes, specifically:

[0177] Based on the sensitivity analysis, this embodiment confirms that wind speed and waves are the key factors affecting nickel ore transportation. Therefore, this embodiment will provide a nickel ore transportation risk map under given wind speed and wave conditions through a visualization method.

[0178] This example is based on wind speed and wave data from the European weather website (https: / / cds.climate.copernicus.eu / ), with a data accuracy of 0.25 degrees. The software versions used to draw the risk map are: Python 3.9.7, ArcGIS 10.8, GeNIe Academic 4.1, and the steps are as follows:

[0179] Step 1: Read initial data, and the initial data includes wind speed and wave data;

[0180] S101: Import necessary libraries: netCDF4, pandas, numpy;

[0181] S102: Use the netCDF4 library to read the wind speed data file;

[0182] S103: extracting longitude, latitude and wind speed data from the wind speed data file;

[0183] S104: Read the wave data file using the netCDF4 library;

[0184] S105: extracting longitude, latitude and wave data from the wave data file;

[0185] S106: Creating a longitude-latitude grid for wind speed data / wave data;

[0186] Step 2: Save the initial data in Excel format;

[0187] S201: Convert wind speed data into a "DataFrame" with longitude and latitude as indexes;

[0188] S202: Save the wind speed “DataFrame” as “wind speed data 2.xlsx”;

[0189] S203: converting the ocean wave data into a "DataFrame" with longitude and latitude as indexes;

[0190] S204: Save the ocean wave “DataFrame” as “ocean wave data 2.xlsx”.

[0191] Step 3: Process wind speed data

[0192] S301: Use the pandas library to read the "wind speed data 2.xlsx" file and set "header = None";

[0193] S302: Remove the first row and the first column of data;

[0194] S303: Take the absolute value of the data and keep one decimal place;

[0195] S304: Save the processed data as “wind speed data 3.xlsx”;

[0196] Step 4: Processing Wave Data

[0197] S401: Use the pandas library to read the "wave data 2.xlsx" file and set "header = None";

[0198] S402: Remove the first row and the first column of data;

[0199] S403: Take the absolute value of the data and keep one decimal place;

[0200] S404: Save the processed data as “wave data 3.xlsx”;

[0201] Step 5: Change the wind speed level

[0202] S501: define a wind speed level replacement rule, and map the wind speed range to the corresponding level of the Pu-type wind scale (such as "breeze", "typhoon", etc.);

[0203] S502: define an "apply_replacement" function to replace the wind speed value according to a replacement rule;

[0204] S503: using the “map” method of the pandas library to apply a replacement function to the wind speed data;

[0205] S504: Save the replaced wind speed data as “wind speed data 3.xlsx”;

[0206] Step 6: Replace the wave level

[0207] S601: define the wave level replacement rule, mapping the wave height range to the corresponding level in the wave table (such as "calm", "smooth", etc.);

[0208] S602: define an "apply_replacement" function to replace the wave value according to the replacement rule;

[0209] S603: Use the “map” method of the pandas library to apply a replacement function to the ocean wave data;

[0210] S604: Save the replaced wave data as “wave data 3.xlsx”;

[0211] Step 7: Bayesian network inference

[0212] S701: Import the data from Step 5 and Step 6 into the pysmile library and set the license information;

[0213] S702: Define the “infer_vibration_grievous_probability” function:

[0214] S703: Create a Bayesian network object, that is, read the structure and parameters from the specified network file, and set the state of the parent node (wind speed and waves);

[0215] S704: Update the credibility in the network to obtain the conditional probability of the child nodes;

[0216] Step 8: Read the formatted data

[0217] S801: Use the pandas library to read the formatted wind speed data "wind speed data 3.xlsx";

[0218] S802: Use the pandas library to read the formatted ocean wave data "Ocean wave data 3.xlsx";

[0219] Step 9: Conditional Probability Inference

[0220] S901: Create an empty "DataFrame" to store the results, with the rows and columns consistent with the wind speed and wave data;

[0221] S902: traverse each combination of wind speed and wave state, obtain the wind speed and wave state at the current location, and call the "infer_vibration_grievous_probability" function for probability inference;

[0222] Store the inference results in the result "DataFrame";

[0223] Step 10: Save the inference results

[0224] S101: Save the result "DataFrame" as "risk data", where the result is the risk value under the wind speed and wave conditions in the set area;

[0225] Step 11: Numerical visualization processing using ArcGIS 10.8 software

[0226] S1101: Import the risk data file "risk data" into ArcMap 10.8;

[0227] S1102: Setting a set scale and rendering it to obtain a risk value map;

[0228] S4: Based on the risk value map, the assumptions for optimizing the shipping route for nickel ore are set, specifically:

[0229] During the marine weather data update period, it is assumed that wind speed / direction and sea waves remain unchanged;

[0230] During the sea voyage, the ship is considered as a point mass;

[0231] It is assumed that the ship can sail according to the expected route plan, that is, after determining the speed, heading and heading angle, it can accurately follow the specified navigation task;

[0232] S5: constructing a multi-objective optimization model for ship route transportation according to the assumptions;

[0233] The multi-objective optimization model for ship route transportation includes a ship fuel consumption model, a ship sulfur emission model and a ship carbon emission model;

[0234] In a specific embodiment, during the navigation of a ship, the generation of hull resistance is caused by the interaction between the hull and the water when the hull moves in still water. When the hull is moving, the water flow moves along the surface of the hull, and also exerts an obstacle on the hull; this obstacle depends on factors such as the shape and size of the hull and the viscosity of water. The hull resistance during the navigation of the ship is composed of friction resistance and residual resistance; During the navigation of a ship, the generation of hull resistance is caused by the interaction between the hull and the water when the hull moves in still water. When the hull is moving, the water flow moves along the surface of the hull, and also exerts an obstacle on the hull. This obstacle depends on factors such as the shape and size of the hull and the viscosity of water. The hull resistance during the navigation of the ship is composed of friction resistance and residual resistance;

[0235] The friction resistance is calculated using a plane formula, which is:

[0236]

[0237] Among them, the friction resistance coefficient C f The Schoenherr formula is used for calculation:

[0238]

[0239] Roughness compensation coefficient ΔC f According to ITTC-1957, the calculation formula is:

[0240]

[0241] In this embodiment, the residual resistance coefficient is obtained by using the Lap-Keller diagram. The Lap-Keller diagram is a graphical tool commonly used in ship fluid mechanics research to represent the resistance characteristics of a ship. The calculation formula of the residual resistance is:

[0242]

[0243] The residual resistance coefficient is obtained by polynomial fitting of the Lap-Keller diagram and interpolation, and the residual resistance correction coefficient is introduced

[0244] α=A1sin(B1L WL +C1)+A2sin(B2L WL +C2)

[0245] The generation of additional wave resistance can be traced back to the interaction between the hull and the waves. When the hull passes through the waves, the waves will cause the hull to move up and down in the vertical direction, resulting in wave lift and the generation of wave resistance. In order to more accurately describe the additional wave resistance, this embodiment refers to the additional wave resistance formula proposed by LIU et al.

[0246] f wave =f awr +f awm

[0247] The calculation formula for the additional resistance component brought to the ship by the diffraction effect of waves is:

[0248]

[0249] The calculation formula for the additional resistance brought to the ship by the radiation effect of waves is:

[0250]

[0251] For a ship in sailing, the additional resistance can also be calculated by the additional resistance in regular waves and the JONSWAP spectrum, specifically:

[0252]

[0253] Where: T P is the spectrum peak period, ω P is the spectrum peak frequency, H S is the significant wave height, S is the wave spectrum density, f wave is the additional wave resistance in irregular sea areas, f wave is the additional wave resistance in regular sea area, γ is the peak gain factor;

[0254] Ship wind resistance refers to the resistance caused by wind when a ship is running in water. In this embodiment, ITTC-1957 is used as a reference, and the calculation formula is:

[0255]

[0256] Then the expression of the hull resistance model is

[0257] f total =f water +f r +f wave +f wind

[0258] Where: f total It represents the total resistance of the ship at any time during the voyage; f water It indicates the hull resistance of the ship at any time during the voyage; fwave It indicates the ocean current resistance that the ship encounters at any time during the voyage; f wind The wind resistance that a ship encounters at any time during its voyage; C f Indicates the friction coefficient; ΔC f represents the roughness compensation coefficient; ρ sea represents the density of seawater; v1 represents the speed of the ship in still water; S represents the wet surface area of ​​the ship; Re represents the Reynolds number; L WL represents the design waterline length; ν represents the viscosity coefficient; Ks represents the hull surface roughness; f r represents the residual resistance; ξ represents the correction percentage of the length and width of the ship; α represents the residual resistance correction coefficient; C r represents the residual resistance coefficient; A1 , B1 , C1 as well as A2 , B2 , C2 represents the setting parameters obtained by fitting the ship model experimental data; f awr Indicates that the diffraction effect of waves produces additional resistance; f awm represents the additional resistance caused by the radiation effect of waves; B represents the length of the ship beam; a T represents the ship's draft correction factor; E represents the ship's waterline entrance angle; L pp represents the length between the vertical lines of the hull; λ represents the wavelength; Fr represents the Froude coefficient; C B represents the square coefficient; ζ a represents the incident wave amplitude; represents the effective encounter frequency of ships; a1, a2, a3 and b1, d1, d2 represent the setting parameters determined by the ship speed and wave spectrum; ρ air Indicates air density; A T represents the orthographic projection area of ​​the waterline surface; θ wa Indicates the angle between the wind direction and the bow during the ship's voyage;

[0259] The ship fuel consumption model is:

[0260] P E =f total v1

[0261]

[0262] MF=P B ×SFOC·10 -6

[0263] Since the total fuel consumption of a ship can be regarded as the fuel consumption generated by overcoming the hull resistance, wind additional resistance and wave additional resistance, the total fuel consumption of the ship is calculated as follows:

[0264] MF=MF water +MF wind +MF wave

[0265]

[0266] Since the main engine model of each ship is different, the main engine fuel consumption rate is related to the main engine model. In this embodiment, the following formula is used to fit the main engine fuel consumption conversion rate:

[0267] SFOC=α1·(B ME ) 3 +α2·(B ME ) 2 -α3·B ME +α4

[0268] B ME =P B / P SMCR ×100%

[0269] Where: P E represents the effective power of the ship during navigation; v1 represents the speed of the ship in still water; f total It represents the total resistance of the ship at any time during the navigation process; P B Indicates the actual power of the ship's main engine; η S Indicates the working efficiency of the main engine shaft; η D represents the propulsion efficiency of the ship; SFOC represents the conversion rate of the ship's main engine fuel consumption; MF represents the ship's main engine fuel consumption; α1, α2, α3, α4 are all set polynomial fitting coefficients; B ME represents the intermediate parameter variable; P SMCR Indicates the rated power of the ship's main engine;

[0270] The fuel prices referenced in this embodiment are all based on the prices on Clarkson Shipping Network on September 3, 2024, as shown in Table 4, where the total price of fuel consumed by the ship is calculated as follows:

[0271]

[0272] Where: P fuel Represents the total price of fuel consumed, P LFO Represents the LFO price, P HFO represents the HFO price;

[0273] Table 4. Fuel price list

[0274]

[0275] During the voyage of a ship, since ships usually use light diesel (LFO) and heavy diesel (HFO) to burn alternately, the ship sulfur emission model is:

[0276]

[0277] Where: S S represents the total sulfur emission; α SC It represents the sulfur emission coefficient of light diesel; Indicates the consumption of light diesel; β SC It represents the sulfur emission coefficient of heavy diesel; Indicates heavy diesel fuel consumption;

[0278] During the voyage of the ship, the carbon conversion coefficient (C F ) are also different, which directly affects the carbon emissions of different fuels. The carbon conversion coefficients calculated for reference in this embodiment are shown in Table 5 below;

[0279] Table 5. Carbon conversion coefficient table

[0280]

[0281] The ship carbon emission model is:

[0282]

[0283] Where: S S represents the total carbon emissions; α SC Indicates the carbon emission coefficient of light diesel; Indicates the consumption of light diesel; β SC represents the carbon emission coefficient of heavy diesel; Indicates heavy diesel consumption.

[0284] This embodiment calculates the carbon emissions of a ship's voyage by measuring the carbon emission price corresponding to the ship's carbon dioxide emissions. The carbon dioxide emission price referenced in this embodiment is derived from the data of Clarkson Shipping Network on August 30, 2024, following the European Union Emissions Quota (EUA, European Union Allowances). The carbon dioxide EUA price refers to the amount required to buy and sell one unit of carbon dioxide emission quota;

[0285] P carbon =S c ×EUA

[0286] Among them, P carbon represents the cost of emitting carbon dioxide, S C represents the total carbon emissions;

[0287] S6: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, solve and obtain the optimal solution of the corresponding optimization objectives according to the multi-objective optimization model of ship route transportation, and optimize the nickel ore shipping route considering risks and emissions according to the optimal solution;

[0288] like Figure 6 As shown, the specific steps include:

[0289] S61: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, the multi-objective optimization model of ship route transportation is solved and the optimal solution of the corresponding optimization objectives is obtained;

[0290] S611: define and initialize the dung beetle population and algorithm parameters of the dung beetle optimization algorithm;

[0291] The algorithm parameters include the size of the dung beetle population, the maximum number of iterations, the total number of dung beetles, and the ratio of the role dung beetles;

[0292] The role dung beetles include rolling dung beetles and breeding dung beetles;

[0293] S612: The risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore are used as the individual position of a dung beetle in the dung beetle population; and the multi-objective optimization model of ship route transportation is used as the evaluation function for optimizing the individual fitness value of the dung beetle in the process of searching for trajectories;

[0294] And based on the evaluation function, the individual fitness value of each dung beetle individual in the current dung beetle population is obtained;

[0295] S613: performing non-dominated sorting on each dung beetle individual in the current dung beetle population according to the individual fitness value, and obtaining an initial sorted dung beetle population including different non-dominated levels;

[0296] The improvement of NSDBO algorithm lies in the introduction of non-dominated sorting strategy. Specifically, the algorithm performs non-dominated sorting on the solutions, divides the solution set into multiple levels, and performs non-dominated sorting on each dung beetle individual in the current dung beetle population. The non-dominated degree calculation formula is:

[0297]

[0298] Where: n i represents the non-dominance degree of the i-th dung beetle individual; H(x) represents the step function, and x = f j -f i; When x≤0, H(x)=1, otherwise H(x)=0; and the Pareto solution set formed by the non-dominated solution is coordinated between the objective functions, and the optimization is achieved by weighing different objective functions. The objective functions here include transportation risk, energy consumption, and carbon and sulfur emissions; f j represents the individual fitness value of the jth dung beetle; f i represents the individual fitness value of the i-th dung beetle individual; N represents the total number of dung beetle individuals;

[0299] S614: Calculate and obtain the crowding distance between dung beetle individuals in the same non-dominated level in the initially sorted dung beetle population;

[0300] And based on the crowding distance, the dung beetle individuals in the initially sorted dung beetle population are sorted in descending order to obtain the currently sorted dung beetle population;

[0301] In this embodiment, in order to maintain the diversity of the population, the NSDBO algorithm introduces the crowding distance CD after the non-dominated sorting. i The concept of is used to measure the density of each solution in its non-dominated layer; the calculation formula of the crowding distance is

[0302]

[0303] Where: CDi represents the crowding distance between dung beetle individuals in the same non-dominated hierarchy; M represents the number of dung beetle individuals in any non-dominated hierarchy; and Respectively represent the maximum and minimum values ​​of individual fitness values; and Represents the adjacent solutions of the fitness value of the i-th individual;

[0304] Each layer of solutions specifically refers to the solutions in the same non-dominated level. The target values ​​of these solutions are rearranged after non-dominated sorting to ensure diversity in the target space. The "density" of the solution refers to the number of solutions in a certain area in the target space. The greater the density, the more concentrated the distribution of solutions in the area, which helps to increase the mining efficiency of the algorithm;

[0305] S615: Based on the Pareto dominance theory, dominance sorting is performed according to the individual fitness values ​​in the current sorted dung beetle population to obtain the Pareto frontier set;

[0306] And select the best individual in the Pareto front set as the global optimal solution;

[0307] S616: based on the position update strategy, obtaining the updated position of each dung beetle individual in the currently sorted dung beetle population to obtain the next generation of dung beetle population;

[0308] And the next generation of dung beetle population and the currently sequenced dung beetle population are taken as the latest dung beetle population;

[0309] Among them, the location update strategy includes the location update of the rolling dung beetle and the location update of the breeding dung beetle, and the location update of the rolling dung beetle and the location update of the breeding dung beetle are existing location update methods, which will not be described in detail here;

[0310] S617: Obtaining a new individual fitness value of each dung beetle individual in the latest dung beetle population based on the evaluation function;

[0311] The latest dung beetle population is used as the currently sorted dung beetle population, and steps S615 to S616 are repeatedly performed until the maximum number of iterations is reached to obtain the final global optimal solution;

[0312] S62: Optimizing the shipping route for nickel ore transportation taking into account risks and emissions according to the final global optimal solution, wherein the final global optimal solution is to optimize the risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore under the constraints of satisfying the optimization objectives of the shipping route with the lowest shipping route risk, the lowest energy consumption and the lowest emissions.

[0313] In the process of transporting nickel ore at sea, ships often face many risks, which come from human factors, natural conditions and the state of the ship itself. The existence of these risk factors may not only cause serious economic losses, but also endanger the safety of crew members and the health of the marine environment. Therefore, it is extremely valuable for shipping industry practitioners to reasonably identify and evaluate the comprehensive risks in the transportation of nickel ore and identify the key risk factors in the transportation of nickel ore. In addition, the route of the ship is usually planned before sailing. However, if the ship only sails according to the scheduled route, and ignores the impact of real-time weather and sea conditions on the operating status of the ship, it is very easy to cause shipping accidents or unnecessary fuel consumption. This embodiment can effectively formulate a more optimized nickel ore transportation route by combining a visual risk map and a route optimization strategy (improved NSDBO algorithm), thereby improving the safety and economic efficiency of navigation.

[0314] The verification experiment of this embodiment is as follows:

[0315] This example selects part of the route for nickel ore transportation from Da Nang, Vietnam (15N, 109E) to Zhoushan Port, China (22N, 121E) as a demonstration example; the total voyage is expected to take 60 hours to complete. The common parameters of the experiment in this example are set after a large number of experimental analyses, as shown in Table 6;

[0316] Table 6. Experimental common parameters

[0317]

[0318] In order to verify the superiority of the algorithm, this example compares the proposed NSDBO algorithm with the multi-objective particle swarm optimization algorithm (MOPSO) and the multi-objective evolutionary algorithm based on decomposition (MOEAD), which are widely used by scholars. By analyzing their performance on the Pareto curve, Figure 7 In order to ensure the reliability of the results and reduce the randomness of the results, this embodiment uses MATLAB 2023a as the simulation environment, the model is repeated ten times, and the simulation results are output;

[0319] This example compares the performance of three algorithms: MOEDA, MOPSO, and NSDBO. By analyzing the Pareto projection results of any two objective combinations, the solution set of the NSDBO algorithm shows higher smoothness and uniformity as a whole. At the same time, the number of solution sets generated by the NSDBO algorithm is also significantly higher than that of the MOEDA and MOPSO algorithms. These results show that the NSDBO algorithm can provide a better solution set in multi-objective optimization problems. Although in very few projections, the effect of NSDBO is slightly inferior to that of MOEDA and MOPSO, which may be due to the paradoxical relationship between the objectives or the inspiration of the algorithm itself, but overall, NSDBO still performs better than the other two algorithms in the solution set.

[0320] Since sulfur emissions and fuel consumption are both derived from energy consumption, there is a significant linear correlation between these two objectives. Specifically, an increase in energy consumption usually leads to an increase in sulfur emissions and fuel consumption, and this relationship is statistically expressed as linear dependence. Therefore, in multi-objective optimization problems, this linear correlation may cause the distribution of optimization results in the objective space to tend to a certain linear structure.

[0321] In order to verify the effect of the optimization algorithm in risk optimization, this embodiment selected actual route data of 100 segments on a certain route, and conducted a detailed analysis of the actual route risk and the risk after optimization by various algorithms. Three different optimization algorithms were used in the experiment: multi-objective evolutionary algorithm (MOEAD), multi-objective particle swarm optimization algorithm (MOPSO) and non-dominated sorting dung beetle optimization algorithm (NSDBO). The analysis results show that the risks after optimization by MOEAD, MOPSO and NSDBO algorithms are all lower than the actual route risks. Figure 8 As shown in the figure, all three algorithms effectively reduce the route risk during the optimization process. However, among the three algorithms, the optimization result of the NSDBO algorithm is the most significant, and its risk value is the lowest. This result verifies the effectiveness and superiority of the NSDBO algorithm in dealing with route risk optimization problems. The NSDBO algorithm can provide a better risk control solution in multi-objective optimization, which proves the potential and advantages of the algorithm in practical applications.

[0322] In the context of increasingly stringent environmental regulations, the fluctuation of carbon emission prices has become a key factor affecting the operational decisions of the shipping industry. Fig. 9 The figure shows the impact of different carbon emission prices on ship routes. The study simulates the route selection behavior of ships under different economic incentives by setting the carbon emission price at two different levels of 75$ / t and 90$ / t. Under low carbon emission prices, ships give priority to the shortest route to reduce energy consumption and operating costs; while under high carbon emission prices, ships tend to detour to avoid high-emission areas, although this may result in longer voyages. Although detours may cause time delays, the overall economic benefits of reducing carbon emission costs may be more attractive in the long run. Therefore, the increase in carbon emission prices prompts ships to take more environmental avoidance measures to optimize costs and reduce carbon emissions.

[0323] In summary, this embodiment provides a systematic solution by combining the risk assessment of Bayesian network with the multi-objective optimization of improved NSDBO algorithm, which is of great significance to the risk management and environmental protection in nickel ore transportation. It can further expand the data sample, simplify the algorithm application process, and optimize the weather data processing method to improve the practical application value and feasibility of the research results.

[0324] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the route of nickel ore transport by ship considering risks and emissions, characterized in that: The specific steps include: S1: Obtain risk factors affecting nickel ore liquefaction during nickel ore marine transportation, and use the risk factors as network nodes to construct an initial Bayesian network affecting the risk status of nickel ore liquefaction; The risk factors include environmental factors, ship factors and human factors; S2: Obtain historical event data of nickel ore marine transportation, and train the initial Bayesian network based on the historical event data to obtain a Bayesian risk inference network; The Bayesian risk inference network is used to infer the risk law that affects nickel ore liquefaction during nickel ore marine transportation and obtain the conditional probability of each network node; S3: Conduct sensitivity analysis on each network node according to conditional probability to obtain key network nodes, and obtain a risk numerical map of the nickel ore transport route based on the key network nodes; S4: Set the assumptions for optimizing the shipping route for nickel ore based on the risk numerical map; S5: constructing a multi-objective optimization model for ship route transportation according to the assumptions; The multi-objective optimization model for ship route transportation includes a ship fuel consumption model, a ship sulfur emission model and a ship carbon emission model; S6: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, solve and obtain the optimal solution of the corresponding optimization objectives according to the multi-objective optimization model of ship route transportation, and optimize the ship transportation nickel ore route considering risks and emissions according to the optimal solution.

2. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 1, characterized in that: The S1 specifically includes the following steps: S11: Obtain risk factors affecting nickel ore liquefaction during marine transportation of nickel ore; The risk factors include environmental factors, ship factors and human factors; The environmental factors include at least wind speed / direction, temperature and sea waves; The ship factors at least include ship age, ship tonnage, ship speed and hull strength; The human factors include at least the age of the crew and the corresponding crew safety inspection and crew safety awareness assessment results; S12: Based on expert experience, determine the evidence nodes, intermediate nodes and query nodes that affect nickel ore liquefaction according to the risk factors; And the characteristics of the evidence nodes are the risk factors affecting nickel ore liquefaction; The characteristics of the intermediate node are the nickel ore vibration and the nickel ore water content caused by the risk factors; The query node is the liquefaction risk level during nickel ore transportation; And obtain the association relationship between evidence nodes, intermediate nodes and query nodes to confirm the edge set between nodes, and then establish the initial Bayesian network that affects the risk status of nickel ore liquefaction.

3. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 2, characterized in that: The S2 specifically includes the following steps: S21: Based on the GeNIe Academic 4.1 software, a corresponding visual Bayesian network was constructed according to the initial Bayesian network; S22: Obtain historical event data of nickel ore marine transportation, and import the historical event data into GeNIe Academic 4.1 software based on a data-driven approach; The EM algorithm is used to train the visual Bayesian network model to obtain the Bayesian risk reasoning network; The Bayesian risk inference network is used to infer the risk patterns that affect nickel ore liquefaction during marine transportation of nickel ore and obtain the conditional probability of each network node.

4. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 3, characterized in that: In S3, sensitivity analysis is performed on each network node according to conditional probability to obtain key network nodes. Specifically, Based on the joint tree algorithm, converting the Bayesian risk inference network into a joint tree; At the same time, the conditional probability of each network node obtained through the Bayesian risk inference network is converted into a joint tree, and the joint tree that satisfies global consistency is obtained through message passing; According to the joint tree that satisfies global consistency, the probability distribution of each network node can be obtained to obtain the key network nodes in the Bayesian risk reasoning network; The message transmission mode includes a first transmission process of transmitting information from top to bottom based on the joint tree and a second transmission process of transmitting information from bottom to top based on the joint tree; The model formula of the first transfer process is: The model formula of the second transfer process is: And the conditions satisfied when the join tree reaches a globally consistent state are C={C i };S={S ij } Where: r M represents the product of the evidence function assigned to the network node set and the conditional probability distribution function; kr Represents network node C k Passed to network node C r message; C and S represent the network node set and edge set in the joint tree respectively.

5. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 4, characterized in that: In S4, the assumptions for optimizing the shipping route for nickel ore transportation are set based on the risk value map, specifically: During the marine weather data update period, it is assumed that wind speed / direction and sea waves remain unchanged; During the sea voyage, the ship is considered as a point mass; It is assumed that the ship can sail according to the expected route plan.

6. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 5, characterized in that: In S5, a multi-objective optimization model for ship route transportation is constructed, including a ship fuel consumption model, a ship sulfur emission model, and a ship carbon emission model; The ship fuel consumption model is P E =f total v1 MF=P B ×SFOC·10 -6 SFOC=α1·(B ME ) 3 +α2·(B ME ) 2 -α3·B ME +a4 B ME =P B / P SMCR ×100% Where: P E represents the effective power of the ship during navigation; v1 represents the speed of the ship in still water; f total It represents the total resistance of the ship at any time during the navigation process; P B Indicates the actual power of the ship's main engine; η S Indicates the working efficiency of the main engine shaft; η D represents the propulsion efficiency of the ship; SFOC represents the conversion rate of the ship's main engine fuel consumption; MF represents the ship's main engine fuel consumption; α1, α2, α3, α4 are all set polynomial fitting coefficients; B ME represents the intermediate parameter variable; P SMCR Indicates the rated power of the ship's main engine; The ship sulfur emission model is: Where: S S represents the total sulfur emission; α SC It represents the sulfur emission coefficient of light diesel; Indicates the consumption of light diesel; β SC It represents the sulfur emission coefficient of heavy diesel; Indicates heavy diesel fuel consumption; The ship carbon emission model is: Where: S S represents the total carbon emissions; α SC Indicates the carbon emission coefficient of light diesel; Indicates the consumption of light diesel; β SC represents the carbon emission coefficient of heavy diesel; Indicates heavy diesel consumption.

7. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 5, characterized in that: The S6 specifically comprises the following steps: S61: Taking the lowest risk, lowest energy consumption and lowest emission of ship route transportation as the optimization objectives, based on the improved NSDBO algorithm, the multi-objective optimization model of ship route transportation is solved and the optimal solution of the corresponding optimization objectives is obtained; S611: define and initialize the dung beetle population and algorithm parameters of the dung beetle optimization algorithm; The algorithm parameters include the size of the dung beetle population, the maximum number of iterations, the total number of dung beetles, and the ratio of the role dung beetles; The role dung beetles include rolling dung beetles and breeding dung beetles; S612: The risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore are used as the individual position of a dung beetle in the dung beetle population; and the multi-objective optimization model of ship route transportation is used as the evaluation function for optimizing the individual fitness value of the dung beetle in the process of searching for trajectories; And based on the evaluation function, the individual fitness value of each dung beetle individual in the current dung beetle population is obtained; S613: performing non-dominated sorting on each dung beetle individual in the current dung beetle population according to the individual fitness value, and obtaining an initial sorted dung beetle population including different non-dominated levels; S614: Calculate and obtain the crowding distance between dung beetle individuals in the same non-dominated level in the initially sorted dung beetle population; And based on the crowding distance, the dung beetle individuals in the initially sorted dung beetle population are sorted in descending order to obtain the currently sorted dung beetle population; S615: Based on the Pareto dominance theory, according to the individual fitness values ​​in the currently sorted dung beetle population, dominance sorting is performed to obtain the Pareto frontier set; And select the best individual in the Pareto front set as the global optimal solution; S616: based on the position update strategy, obtaining the updated position of each dung beetle individual in the currently sorted dung beetle population to obtain the next generation of dung beetle population; And the next generation of dung beetle population and the currently sequenced dung beetle population are taken as the latest dung beetle population; S617: Obtaining a new individual fitness value of each dung beetle individual in the latest dung beetle population based on the evaluation function; The latest dung beetle population is used as the currently sorted dung beetle population, and steps S615 to S616 are repeatedly performed until the maximum number of iterations is reached to obtain the final global optimal solution; S62: Optimizing the shipping route for nickel ore transportation taking into account risks and emissions according to the final global optimal solution, wherein the final global optimal solution is to optimize the risk factors affecting nickel ore liquefaction during the marine transportation of nickel ore under the constraints of satisfying the optimization objectives of the shipping route with the lowest shipping route risk, the lowest energy consumption and the lowest emissions.

8. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 7, characterized in that: The non-dominated degree calculation formula for performing non-dominated sorting on each dung beetle individual in the current dung beetle population in S613 is: Where: n i represents the non-dominance degree of the i-th dung beetle individual; H(x) represents the step function, and x = f j -f i ;f j represents the individual fitness value of the jth dung beetle individual; f i represents the individual fitness value of the i-th dung beetle individual; N represents the total number of dung beetle individuals.

9. A method for optimizing the shipping route of nickel ore considering risks and emissions according to claim 7, characterized in that: In the S61, the crowding distance between dung beetle individuals in the same non-dominated level in the initially sorted dung beetle population is calculated and obtained; The calculation formula of the crowding distance is: Where: CDi represents the crowding distance between dung beetle individuals in the same non-dominated hierarchy; M represents the number of dung beetle individuals in any non-dominated hierarchy; and Respectively represent the maximum and minimum values ​​of individual fitness values; and Represents the adjacent solutions of the fitness value of the i-th individual.

Images