A ship automatic collision avoidance path planning method based on population size adaptability of differential evolution algorithm

Abstract

The application provides a ship automatic collision avoidance path planning method based on a population size adaptivity differential evolution algorithm, and the method comprises the following steps: acquiring sea state information of a surrounding environment and other ships, establishing a ship field model, and calculating a safe meeting distance of the ship and movement parameters of ship reorientation; aiming at the premature convergence problem of a standard differential evolution, a population size adaptivity differential evolution algorithm is introduced; a ship collision risk model is constructed based on the population size adaptivity differential evolution algorithm. A multi-objective function is converted into a linear function for solving by a weight method, and the multi-objective function problem is solved by an improved differential evolution algorithm. The technical scheme of the application solves the optimization decision problem in the automatic collision avoidance path planning in various complex environments in the prior art.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more particularly to a method for automatic collision avoidance path planning for ships based on a differential evolution algorithm with population size adaptability. Background Technology

[0002] Maritime transport, as a fundamental mode of transportation, plays a vital role in global economic and trade exchanges. The safety, economy, and efficiency of ships, as carriers of maritime transport, have always been issues of common concern to both academia and the business community. Meanwhile, with the increasing number of ships, maritime encounters are becoming increasingly complex, requiring crew members to make timely and accurate judgments about collision hazards and avoidance decisions in a shorter period of time. Therefore, achieving autonomous collision avoidance for ships has significant practical implications.

[0003] According to relevant data in my country, over 80% of ship collisions are caused by human factors. With the increasing number of ships, maritime encounters are becoming increasingly complex, requiring crew members to make timely and accurate judgments about collision hazards and avoidance decisions in a shorter time, thus placing higher demands on their professional skills. With the gradual promotion of intelligent algorithms, the level of ship automation is also gradually improving. Addressing collisions caused by improper human operation through ship autonomous identification and automatic collision avoidance is an inevitable trend. Automatic collision avoidance path planning is a key and challenging issue in ship automation research. Through automatic collision avoidance path planning, the level of collision risk can be monitored in real time, thereby effectively assessing the degree of danger between ships.

[0004] Compared to two-vehicle collision avoidance, multi-vehicle collision avoidance scenarios are far more complex and unpredictable. Even fishing boats and other vessels sometimes disregard rules and arbitrarily cross waterways, posing a significant challenge to multi-vehicle collision avoidance. However, collision avoidance rules do not specify how to avoid encounters between multiple vessels, adding to the difficulties in actual collision avoidance. When multiple vessels are navigating, the situation becomes more complex, requiring calculations of the collision risk between the vessel and multiple target vessels, as well as a clear understanding of the encounter situation. Simultaneously, artificial intelligence algorithms are needed to optimize a multi-objective function constructed during the actual navigation process to determine the optimal navigation route for the vessel itself or for both the vessel and the target vessels. Summary of the Invention

[0005] To address the aforementioned technical problems of inaccurate collision risk assessment during ship navigation and premature convergence of differential evolution algorithms (DEA) in optimizing multi-objective variable functions, this invention provides an automatic collision avoidance path planning method for ships based on population-scale adaptive DEA. This invention constructs an automatic collision avoidance path planning system framework, comprehensively analyzes the ship navigation process to determine the occurrence of ship collision accidents, analyzes the causes of collisions, and determines the ship domain model and the calculation method for safe encounter distances. Simultaneously, for port entry / exit and open water navigation environments, a reasonable collision risk assessment method is established to solve the problem of inaccurate collision risk assessment during ship navigation. Based on this, according to actual ship navigation conditions, a multi-objective function model based on navigation economy, collision risk, and turning angle is established. An improved DEA strategy is proposed to address the premature convergence problem in optimizing multi-objective variable functions. Experimental simulations are conducted in complex environments such as multi-ship encounters and port entry / exit, yielding good avoidance timing, turning timing, and rerouting timing.

[0006] The technical means employed in this invention are as follows:

[0007] An automatic collision avoidance path planning method for ships based on a differential evolution algorithm with population size adaptability includes:

[0008] Acquire information on the surrounding environment and sea conditions of other vessels, establish a vessel domain model, and calculate the safe encounter distance and motion parameters for vessel course changes;

[0009] Introduce a differential evolution algorithm with population size adaptability;

[0010] A differential evolution algorithm based on population size adaptability is used to construct a ship collision risk model.

[0011] Furthermore, the acquisition of surrounding environment and sea state information of other vessels, the establishment of a vessel domain model, and the calculation of safe encounter distance and vessel motion parameters specifically include:

[0012] Construct a quaternion-based ship domain model (QSD), whose analytical expression and boundary equation expressions are as follows:

[0013] QSD={(x,y)∣f(x,y,Q)≤1,Q={R f o re ,R aft ,R starb ,R p o rt},k≥1}

[0014]

[0015] To determine the analytical expression for the safe encounter distance, calculations were performed based on a quaternion ship domain model, yielding the ship domain boundary values ​​for different target ship hull angles:

[0016]

[0017] Calculate the relative motion parameters DCPA and TCPA between the ship and the target ship using the following formulas:

[0018]

[0019]

[0020] Assuming the relative positions of the two ships remain unchanged at the moment of turning, calculate the ship's kinematic parameters DCPA' and TCPA' during the course change using the following formulas:

[0021]

[0022]

[0023] Furthermore, the differential evolution algorithm that incorporates population size adaptability specifically includes:

[0024] Population initialization using a differential evolution algorithm for population size adaptation:

[0025] The differential evolution algorithm uses real-number encoding to generate a population of dimension (NP, Q) within the feasible region of the search space, where NP represents the population size and Q represents the number of decision variables. Then, the i-th individual x... i For x i ={x i，1 x i，2 , ..., x i，j , ..., x i，Q};x i，j x represents the value of the j-th dimension of the i-th individual. i，j It is generated randomly within the decision space, and the initialization expression is:

[0026] x i =x min , j +rand i , j (0, 1)·(x) max , j -x min , j ),

[0027] Among them, rand i，j (0, i) represents generating uniformly distributed random numbers between 0 and 1;

[0028] At the generation of each generation of the population, the lifetime value of each individual is calculated based on the fitness function value:

[0029] When the population is initialized, the age of an individual is 0, meaning that the individual's generation number and age are the same. When the age value of an individual in the population is greater than its lifespan, it is deleted from the population. The specific calculation formula is as follows:

[0030]

[0031] If the fitness function value of a population remains unchanged for a consecutive number of times, the current generation of the population will fall into a local extreme value. It will then escape the local extreme value through the extinction mechanism, thereby increasing the diversity of the population.

[0032] When some individuals are removed through the lifespan mechanism, the integrity of the population is maintained by replenishing them with new individuals:

[0033] By replicating elite individuals with high fitness function values ​​from the population, the following formula is used for calculation:

[0034] num=ζ·(NP max -NP g ),

[0035] The population size changed as follows after the improvement:

[0036] NP g+1 =NP g -NP dead +num

[0037] When the extinction mechanism comes into play, in order to ensure population diversity and randomness, a new population is established using the following formula:

[0038]

[0039] Adding the best individuals from the extinct population to the new population increases the population size by "1", thus changing the population size to:

[0040] NP g+1 =NP g +1.

[0041] Furthermore, the differential evolution algorithm based on population size adaptability, used to construct the ship collision hazard model, specifically includes:

[0042] Based on fuzzy comprehensive evaluation, the risk level of ship collision is calculated;

[0043] Construct a fitness function for ship collision avoidance path planning;

[0044] The fitness function is solved based on the improved differential evolution algorithm;

[0045] Based on the degree of ship collision risk, collision avoidance decisions are designed.

[0046] Furthermore, the calculation of ship collision risk based on fuzzy comprehensive evaluation specifically includes:

[0047] Establish a factor set:

[0048] The six indicators used as inputs when calculating the collision risk of the vessel with other vessels or obstacles are: nearest encounter distance, nearest encounter time, distance between vessels, relative bearing, speed ratio, and channel dimensions, denoted as DCPA, TCPA, D, B, K, and R, respectively.

[0049] The factor set U is constructed using 6 indicator parameters as follows:

[0050] U={DCPA,TCPA,D,B,K,R}

[0051] Establish an evaluation set:

[0052] The target vessel evaluation set contains two items, one indicating that the target vessel is dangerous and the other indicating that the target vessel is safe, denoted as U1 and U2 respectively;

[0053] U1 represents a situation where there is a collision hazard between the vessel and other vessels or obstacles, requiring evasive action. U2 represents a situation where there is no collision hazard between the vessel and other vessels or obstacles, meaning no evasive action is required. Therefore, the evaluation set V is established as follows:

[0054] V = {U1, U2}

[0055] Determine the indicator weights:

[0056] The weight set W is constructed based on the factor set as follows:

[0057] W={wD CPA w TCPA w D w B w K w R}

[0058] In the formula, w DCPA w TCPA w D w B w K w R These represent the weight values ​​of DCPA, TCPA, D, B, K, and R, respectively.

[0059] Among the indicators affecting the risk of ship collisions, the weights of each indicator satisfy the following:

[0060] w DCPA ≥w TCPA≥w D ≥w B ≥w K ≥w R >0,

[0061] And it satisfies the normalization condition:

[0062] w DCPA +w TCPA +w D +w B +w K +w R =1

[0063] The weights of each indicator are determined as follows: w DCPA =0.3, w TCPA =0.3, wD ＝ 0.1, w B =0.1, w K =0.1, w R =0.1, the weight matrix is ​​as follows:

[0064] W = [w DCPA wT CPA w D w B w K w R ]

[0065] Establish the membership function of the evaluation index;

[0066] A comprehensive evaluation function for calculating the risk of ship collisions:

[0067] The target evaluation matrix includes the membership functions of the above six indicators, as shown in the following formula.

[0068]

[0069] Among them, u TCPA U DCPA u B u D u K u R It is the membership function of the factors affecting the collision risk between ships, and its value ranges from 0 to 1.

[0070] Taking into account the impact of the six evaluation indicators, the evaluation results of the target ship are as follows:

[0071]

[0072] Therefore, the expression for collision hazard level is:

[0073] CRI = w DCPA uDCPA +w TCPA u TCPA +w DuD +w BuB +w K u K +w R u R .

[0074] Furthermore, the fitness function for constructing the ship collision avoidance path planning specifically includes:

[0075] Construct an objective function based on ship collision risk:

[0076] f1 = max CRI

[0077] Construct an objective function based on the total flight distance:

[0078]

[0079] Construct an objective function based on steering amplitude:

[0080] f3=|θ i |

[0081] Construct an objective function based on collision avoidance rules:

[0082]

[0083] Construct an objective function based on the timing of avoidance:

[0084]

[0085] Furthermore, the step of solving the fitness function based on the improved differential evolution algorithm specifically includes:

[0086] An improved differential evolution algorithm is used to optimize a multi-objective function composed of decision variables x, where x is a vector set of turning magnitudes.

[0087] S332. Based on the five constructed objective functions, the fitness function is established as follows during navigation within the waterway:

[0088] Fitness=(ω1f1+ω2f2+ω3f3)·f4·f5

[0089] The weight coefficients ω1, ω2, and ω3, which serve as the objective function, satisfy the following:

[0090] ω1+ω2+ω3=1

[0091] Then, within the waterway, the fitness function optimized based on the improved differential evolution algorithm is expressed as:

[0092] min{Fitness}

[0093] s.t. -30≤x≤30

[0094] f4 = 0 or 1

[0095] f5 = 0 or 1

[0096] In the case of open water, the fitness function optimized based on the improved differential evolution algorithm is expressed as:

[0097] nin{Fitness}

[0098] st-60°≤x≤60°

[0099] f4 = 0 or 1

[0100] f5 = 0 or 1.

[0101] Furthermore, the design of collision avoidance decisions based on the degree of ship collision risk specifically includes:

[0102] Assuming the navigation area is free from the influence of wind, waves, and currents; when designing collision avoidance decisions, only the collision avoidance measure of turning the ship is considered, and the measure of changing speed is not used for collision avoidance; at the same time, according to the requirements of good seamanship, the turning amplitude function is designed.

[0103] Selecting the collision avoidance timing involves determining the vessel's collision avoidance opportunity based on a set threshold.

[0104] The multi-objective function is transformed into a single-objective function with the steering amplitude as the independent variable by using the weighting method.

[0105] Compared with the prior art, the present invention has the following advantages:

[0106] 1. The present invention provides a differential evolution algorithm for automatic collision avoidance path planning of ships based on population size adaptability. Addressing the premature convergence problem inherent in standard differential evolution, this invention proposes a differential evolution algorithm with population size adaptability. This algorithm introduces extinction and lifespan mechanisms. When the age of an individual in the population exceeds its lifespan, it is either removed from the population or a new population is established to increase population size diversity, thus achieving adaptive adjustment of population size.

[0107] 2. This invention provides an automatic collision avoidance path planning method for ships based on a differential evolution algorithm with adaptive population size. It establishes a ship collision avoidance algorithm based on an improved differential evolution algorithm and constructs a ship collision hazard model, which can be used for ships to simultaneously avoid waterways, static obstacles, and dynamic vessels. A multi-objective function is constructed based on collision hazard, total distance, turning radius, waterway boundary, collision avoidance rules, and avoidance timing. The multi-objective function is transformed into a linear function using a weighted method for solution, and through the improved differential evolution algorithm, optimal ship avoidance timing, turning radius, and rerouting timing are obtained.

[0108] 3. Based on the Matlab platform, comparative experiments were conducted between the improved differential evolution algorithm and the standard differential evolution algorithm and genetic algorithm in typical scenarios such as multiple ships simultaneously avoiding multiple ships and static obstacles. The improved differential evolution algorithm of this invention shows significant improvements in iteration speed and heading stability. Simultaneously, simulation experiments were performed using the improved algorithm under various encounter situations, such as ships meeting each other in the channel and ships crossing paths with fishing boats in the channel, verifying the effectiveness of the collision avoidance decision. This invention can effectively solve the optimization decision problem in automatic collision avoidance path planning under various complex environments.

[0109] Based on the above reasons, this invention can be widely applied in fields such as artificial intelligence. Attached Figure Description

[0110] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0111] Figure 1 This is a flowchart of the method of the present invention.

[0112] Figure 2 This is a flowchart of the ship collision avoidance process of the present invention.

[0113] Figure 3 This establishes the ship collision hazard index for the present invention.

[0114] Figure 4 This is the optimal path in a multi-ship encounter situation according to the present invention.

[0115] Figure 5 This invention describes the iterative process under multi-ship encounter situations.

[0116] Figure 6 This invention relates to the course of ships in multi-ship encounter situations.

[0117] Figure 7This invention defines the shortest passing distance for ships in multi-ship encounter situations. Detailed Implementation

[0118] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0119] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0120] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0121] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0122] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention. The directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0123] For ease of description, spatial relative terms such as "above," "over," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation besides the orientation of the device as described in the figures. For example, if the device in the figures is inverted, a device described as "above" or "above" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0124] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0125] like Figure 1 As shown, this invention provides a method for automatic collision avoidance path planning for ships based on a differential evolution algorithm with population size adaptability, comprising:

[0126] S1. Obtain information on the surrounding environment and sea conditions of other vessels, establish a vessel domain model, and calculate the safe encounter distance and motion parameters for vessel reversal.

[0127] S2, Introducing a differential evolution algorithm with adaptive population size;

[0128] S3. A differential evolution algorithm based on population size adaptability is used to construct a ship collision risk model.

[0129] In a preferred embodiment of the present invention, step S1, which involves acquiring information about the surrounding environment and sea conditions of other vessels, establishing a vessel domain model, and calculating the safe encounter distance and motion parameters of the vessel, specifically includes:

[0130] S11. Construct the quaternion-based ship domain model QSD, whose analytical expression and boundary equation expressions are as follows:

[0131]

[0132]

[0133] S12. To determine the analytical expression for the safe meeting distance, calculations are performed based on the four-tuple ship domain model to obtain the ship domain boundary values ​​for different target ship hull angles:

[0134]

[0135] S13. Calculate the relative motion parameters DCPA and TCPA between the vessel and the target vessel. These relative motion parameters are the data source for calculating the collision risk. The calculation formula is as follows:

[0136]

[0137]

[0138] S14. Assuming the relative positional relationship between the two ships does not change at the moment of turning, calculate the ship's kinematic parameters DCPA' and TCPA' during the course change. The calculation formulas are as follows:

[0139]

[0140]

[0141] In a specific implementation, as a preferred embodiment of the present invention, step S2 introduces a population-size adaptive differential evolution algorithm to address the premature convergence problem that standard differential evolution is prone to. Specifically, this includes:

[0142] S21. Population initialization using a differential evolution algorithm for population size adaptation:

[0143] Differential evolution is a parallel search algorithm based on population heuristics. Before starting the algorithm, population initialization is performed for differential evolution with adaptive population size.

[0144] The differential evolution algorithm uses real-number encoding to generate a population of dimension (NP, Q) within the feasible region of the search space, where NP represents the population size and Q represents the number of decision variables. Then, the i-th individual x... i For x i ={x i，1 x i，2 , ..., x i，j , ..., x i，Q};x i，j x represents the value of the j-th dimension of the i-th individual. i，j It is generated randomly within the decision space, and the initialization expression is:

[0145] x i =x min，j +rand i，j (0, 1)·(x) max，j -x min，j ),

[0146] Among them, rand i，j (0, 1) represents generating uniformly distributed random numbers between 0 and 1. Therefore, for different solution vectors, the initial values ​​of each dimension element are independent of each other, ensuring the differences between different individuals and the diversity of the population.

[0147] S22. At the generation of each generation of individuals, calculate the lifetime value of each individual based on the fitness function value:

[0148] This genetic algorithm employs the concepts of individual age and lifespan in population-sized genetic algorithms. When the population is initialized, the individual age is 0, meaning the individual's generation number and age are the same. When an individual's age exceeds its lifespan, it is removed from the population. The specific calculation formula is as follows:

[0149]

[0150] S23. When the fitness function value of the population remains unchanged for a consecutive number of times, the current generation of the population falls into a local extreme value. It escapes the local extreme value through the extinction mechanism, thereby increasing the diversity of the population. The parameter p is the extinction coefficient, which ensures the occurrence of population extinction and requires the generation of new population individuals.

[0151] S24. When some individuals are removed through the lifespan mechanism, the integrity of the population is maintained by replenishing them with new individuals:

[0152] By replicating elite individuals with high fitness function values ​​from the population, the following formula is used for calculation:

[0153] num=ζ·(NP max -NP g ),

[0154] The population size changed as follows after the improvement:

[0155] NP g+1 =NP g -NP dead +num

[0156] S25. When the extinction mechanism is in effect, in order to ensure population diversity and randomness, a new population is established using the following formula:

[0157]

[0158] In the extinction mechanism, individuals with better performance but weaker evolutionary ability are removed from the population through extinction. The fitness function value of the newly established population needs to be compared one by one with the fitness function values ​​of individuals in the original population. To prevent the best individuals from failing to enter the population during the comparison process, the best individuals from the extinct population are added to the new population, increasing the population size by "1". That is, the population size becomes:

[0159] NP g+1 =NP g +1.

[0160] In a specific implementation, as a preferred embodiment of the present invention, step S3, which involves constructing a ship collision risk model based on a differential evolution algorithm with population size adaptability, specifically includes:

[0161] S31. Calculate ship collision risk based on fuzzy comprehensive evaluation. Ship collision risk is an indicator used to measure the magnitude of the risk of a ship colliding with other dynamic ships or static obstacles. In research on ship path planning and automatic collision avoidance, collision risk is the basis and important basis for relevant decision-making, and it is also a core and difficult problem in this field. Therefore, modeling and quantifying this indicator is significant for the development of ship intelligence. The quantification process requires an overall assessment of the encounter state between two ships, involving many influencing factors such as people, ships, and the environment. Therefore, to improve the accuracy of the assessment results, fuzzy comprehensive evaluation theory is needed for quantification. The core issues in the process lie in the standardization of the factor set, weight allocation, comprehensive evaluation, and finally, the numerical value is calculated. The value ranges from [0, 1]. When the value is 0, it means that there is no risk of collision between the vessel and the target vessel or obstacle. When the value is between 0 and 1, the probability of a collision increases with the value. When the value is 1, it means that the collision cannot be avoided by the yielding vessel alone. If the vessel with the risk of collision is a moving vessel, both vessels need to take reasonable actions at the same time to eliminate the risk of collision.

[0162] In this embodiment, the specific implementation process of step S31 is as follows:

[0163] S311. Establish the factor set:

[0164] The six indicators used as inputs when calculating the collision risk of the vessel with other vessels or obstacles are: nearest encounter distance, nearest encounter time, distance between vessels, relative bearing, speed ratio, and channel dimensions, denoted as DCPA, TCPA, D, B, K, and R, respectively.

[0165] The above indicators can be directly read or indirectly calculated using shipboard electronic equipment (such as AIS, ARPA, ECDIS, etc.). Therefore, the factor set U is established using the six indicator parameters as follows:

[0166] U={DCPA,TCPA,D,B,K,R}

[0167] S312. Establish the evaluation set:

[0168] The target vessel evaluation set contains two items, one indicating that the target vessel is dangerous and the other indicating that the target vessel is safe, denoted as U1 and U2 respectively;

[0169] U1 represents a situation where there is a collision hazard between the vessel and other vessels or obstacles, requiring evasive action. U2 represents a situation where there is no collision hazard between the vessel and other vessels or obstacles, meaning no evasive action is required. Therefore, the evaluation set V is established as follows:

[0170] V = {U1, U2}

[0171] S313. Determine the weights of the indicators:

[0172] The weight set W is constructed based on the factor set as follows:

[0173] W = {w DCPA w TCPA w D w B w K w R}

[0174] In the formula, w DCPA w TCPA w D w B w K w R These represent the weight values ​​of DCPA, TCPA, D, B, K, and R, respectively.

[0175] Among the indicators affecting the risk of ship collisions, the weights of each indicator satisfy the following:

[0176] w DCPA ≥w TCPA≥w D ≥w B ≥w K ≥w R >0,

[0177] And it satisfies the normalization condition:

[0178] w DCPA +w TCPA +w D +w B +w K +w R =1

[0179] The weights of each indicator are determined as follows: w DCPA =0.3, w TCPA =0.3, w D =0.1, w B =0.1, w K =0.1, w R =0.1, the weight matrix is ​​as follows:

[0180] W = [w DCPA w TCPA w D w B w K w R ]

[0181] S314. Establish the membership function of the evaluation index; for the established fuzzy set, the membership function not only reflects the fuzzy characteristics of the index concept, but also performs reasonable quantification of each index through mathematical calculation. Accurately defining the membership function is the bridge for applying fuzzy theory to quantitatively characterize fuzzy concepts, and it is also the key to applying fuzzy theory to practical engineering situations.

[0182] S315. Comprehensive evaluation function for calculating the risk of ship collision:

[0183] The target evaluation matrix includes the membership functions of the above six indicators, as shown in the following formula.

[0184]

[0185] Among them, u TCPA u DCPA u B u D u K u R It is the membership function of the factors affecting the collision risk between ships, and its value ranges from 0 to 1.

[0186] Taking into account the impact of the six evaluation indicators, the evaluation results of the target ship are as follows:

[0187]

[0188] Therefore, the expression for collision hazard level is:

[0189] CRI = w DCPA u DCPA +w TCPA u TCPA +w D u D +w B u B +w K u K +w R u R .

[0190] S32. Construct the fitness function for ship collision avoidance path planning;

[0191] In this embodiment, the specific implementation process of step S32 is as follows:

[0192] S321. Construct an objective function based on ship collision risk:

[0193] The entire voyage is divided into several sub-segments, within which the ship maintains its course and speed. To minimize the collision risk within a segment, the objective function needs to be set to the maximum value of that segment. Minimizing this maximum value ensures that the values ​​for the entire segment are relatively small. The objective function based on collision risk is:

[0194] f1 = mac CRI

[0195] S322. Construct an objective function based on total flight distance:

[0196] Among all optimized sailing routes, the total voyage distance is an economic evaluation metric. Generally speaking, under the premise of ensuring absolute safety, the shorter the total voyage distance, the better. Assuming that the ship originally planned to sail from one place to another in a straight line, the requirement of minimizing the total voyage distance can also be understood as minimizing the deviation from the original route. The smaller the deviation from the original route, the higher the economic efficiency of the route.

[0197] Assume the coordinates of a turning point in the collision avoidance path are (xi, yi), and the coordinates of the next adjacent turning point in the path are (xi+1, yi+1), with n turning points. Since the ship maintains its course and speed within each sub-segment, the distance of each sub-segment can be calculated. The objective function based on the total distance is:

[0198]

[0199] S323. Construct an objective function based on steering amplitude:

[0200] Similar to the economic function that minimizes the total voyage distance, the best economic efficiency is achieved when the ship performs no maneuvers during navigation and obstacle avoidance, i.e., maintaining its course and speed. Avoiding frequent course changes also aligns with maritime practice. Therefore, an objective function can be designed based on the ship's course-changing amplitude. If the ship can complete the maneuver with the minimum turning amplitude during obstacle avoidance, then its deviation from the original course will also be minimal. The better the course, i.e., the more economical the navigation path, the better. Assuming the ship's turning amplitude at each turning point is [value], then the objective function based on the turning amplitude is:

[0201] f3=|θ i |

[0202] S324. Construct an objective function based on collision avoidance rules:

[0203] Based on the division of various encounter situations and avoidance responsibilities, when the target vessel is positioned relative to it in areas A, B, or F, the vessel has the responsibility to actively avoid it. In other areas (C, D), the vessel is a straight-ahead vessel and should maintain its course and speed. The objective function based on the collision avoidance rules is:

[0204]

[0205] S325. Construct an objective function based on avoidance timing:

[0206] The timing of evasive action by a vessel is related to the Collision Risk Index (CRI). Generally, a CRI threshold is set, and when the distance between two vessels exceeds this threshold, the vessel giving way should immediately take evasive action. This paper uses a CRI of 0.5 as the threshold to determine the timing of evasive action. Therefore, the objective function based on the timing of evasive action is:

[0207]

[0208] S33. Solve the fitness function based on the improved differential evolution algorithm;

[0209] In this embodiment, the specific implementation process of step S33 is as follows:

[0210] S331. An improved differential evolution algorithm is used to optimize a multi-objective function composed of decision variables x, where x is a vector set of turning amplitudes. When a vessel is navigating in open water, collision avoidance rules require that turning measures be "early," "large," "wide," and "clear," generally involving a turning amplitude of not less than a certain value. In general navigation practice, crew members will not take turning amplitudes exceeding a certain value to avoid a collision. Therefore, when leaving a channel, the constraint x is set to a range of -60° ≤ x ≤ 60°. However, when navigating within a channel, due to the channel width limitation, vessels generally do not take large or frequent turns. In navigation practice, the turning amplitude x within a channel is set to -30° ≤ x ≤ 30°.

[0211] S332. Among the five objective functions mentioned above, f1, f2, and f3 serve as indicators measuring the safety and economy during collision avoidance, playing a dominant role in the calculation of the fitness value. f4 and f5, on the other hand, act as indicators for determining the yielding vessel, the timing of the avoidance maneuver, and whether the vessel has touched the channel boundary, respectively, thus providing constraints. Therefore, based on the constructed five objective functions, the fitness function "Fitness" is established as follows during navigation within the channel:

[0212] Fitness=(ω1f1+ω2f2+ω3f3)·f4·f5

[0213] The weight coefficients ω1, ω2, and ω3, which serve as the objective function, satisfy the following:

[0214] ω1+ω2+ω3=1

[0215] Then, within the waterway, the fitness function optimized based on the improved differential evolution algorithm is expressed as:

[0216] min{Fitness}

[0217] st-30°≤x≤30°

[0218] f4 = 0 or 1

[0219] f5 = 0 or 1

[0220] In the case of open water, the fitness function optimized based on the improved differential evolution algorithm is expressed as:

[0221] min{Fitness}

[0222] st·-60°≤x≤60°

[0223] f4 = 0 or 1

[0224] f5 = 0 or 1.

[0225] S34. Design collision avoidance decisions based on the degree of ship collision risk.

[0226] In this embodiment, the specific implementation process of step S34 is as follows:

[0227] S341. In-depth research on ship collision avoidance decision-making, in this embodiment, it is assumed that the ship's navigation area is unaffected by wind, waves, and currents. Since the research environment in this embodiment is divided into areas such as inside the channel, outside the channel, and open water, and since changes in speed can easily lead to more complex situations, this embodiment only considers collision avoidance measures such as ship turning when studying and designing ship collision avoidance decisions, and does not use speed change measures for collision avoidance actions. At the same time, according to the requirements of good seamanship, a turning amplitude function is designed to avoid frequent turning and using small-angle turns for collision avoidance operations, and the ship maintains a sufficient safe distance from each target ship and the channel during the collision avoidance process.

[0228] S342. Selecting the collision avoidance timing: The collision avoidance timing of the vessel is determined by setting a threshold value. In this embodiment, when the value is set to 0.5, the yielding vessel begins to take collision avoidance action, ensuring early collision avoidance. If, during the actual collision avoidance process, the yielding vessel merely makes a turning maneuver to reduce the risk of collision, this is clearly not in line with practical operation. To solve this problem, an economic function is set to ensure that the vessel can effectively restore its original course, combining this with navigation economics, thus effectively solving the problem of the timing of resuming course during collision avoidance.

[0229] S343. In the collision avoidance path planning decision-making process, issues such as the safety, economy, and rationality of the turning radius of the route must be satisfied. In this embodiment, the multi-objective function is transformed into a single-objective function with the turning radius as the independent variable through the weighting method. Simultaneously, to solve the optimal solution problem of this objective function, the advantages of multi-point parallel search of the differential evolution algorithm are utilized to solve the multi-objective function problem. An improved strategy is proposed to address the premature convergence problem of the algorithm and applied to the optimization problem in ship collision avoidance path planning. Combined with the ship collision avoidance theory system, it effectively solves the optimization problems of avoidance timing, turning timing, and rerouting timing in multi-ship collision avoidance path planning.

[0230] To verify the effectiveness of the method of this invention, comparative experiments were conducted on the Matlab platform, comparing the improved differential evolution algorithm with the standard differential evolution algorithm and the genetic algorithm in typical scenarios such as multiple ships and simultaneous avoidance of multiple ships and static obstacles, and the prediction results were obtained. The specific implementation process of the experiment is as follows:

[0231] The process of ship collision avoidance path planning:

[0232] Initialize the ship positions: the initial position of this ship is (x0, y0), and the initial positions of other ships are (x1, y1), (x2, y2), and (x3, y3), respectively. The position of static obstacles is (x0, y0). target y target ), calculate the basic parameter values ​​such as DCPA, TCPA, D, B, K between the vessel and other vessels or static obstacles;

[0233] A multi-objective function is established using the indicators selected above, and then the multi-objective function is transformed into a single-objective function with the turning amplitude as the independent variable. The piecewise function is set according to different navigation environments, and the values ​​of the independent variables in the piecewise function are selected according to different environments.

[0234] Based on the selection of the multi-objective function and the application of collision avoidance rules, the ship's responsibility for active avoidance is determined according to different encounter situations. Simultaneously, a collision hazard function is calculated based on initial parameters, using collision hazard as the optimization objective. As the ship's turning angle changes, the collision hazard gradually changes, thus obtaining a reasonable turning angle under a reasonable collision hazard level.

[0235] Random numbers are generated to initialize the independent variables, the fitness function values ​​of the independent variables are calculated, and new individuals are generated by crossover, mutation, and selection comparison based on the fitness function values.

[0236] After the above operations, most of the individuals in the algorithm population are the best individuals selected and compared. At this point, the introduction of lifespan and extinction mechanisms increases the diversity of the population, thereby increasing the global search capability.

[0237] Simulation experiments were conducted on collision avoidance path planning for vessels in open waters and on collision avoidance path planning algorithms for vessels near waterways; the experimental results are as follows: Figure 4-7 As shown, from Figure 4 As can be seen, the optimal paths of the three algorithms are shown. The red path represents the optimization result of the improved standard difference evolutionary algorithm, the green path represents the optimization result obtained by the genetic algorithm, the original planned route of the ship is represented by the red dashed line, and the route optimized by the DE algorithm is represented by the blue solid line. From Figure 7 As shown in the distance graph, the improved differential evolution algorithm ensures safe and effective avoidance of both target ships by maximizing the closest distance between any two vessels. However, the ship navigation paths clearly demonstrate that the red path (result of the improved differential evolution algorithm) is significantly better than the blue path (result of the standard differential evolution algorithm) and the green path (result of the genetic algorithm) in terms of both deviation from the original route and course change. Furthermore, the iterative convergence curves of the three algorithms show that the improved differential evolution algorithm converges slightly faster than the standard differential evolution algorithm, exhibits better stability, and provides a larger closest passing distance to the two ships, indicating enhanced safety.

[0238] Simulation results show that: First, comparative experiments were conducted using the improved differential evolution algorithm with the standard differential evolution algorithm and the genetic algorithm in various scenarios. The experimental results demonstrate that the improved differential evolution algorithm significantly improves both iteration speed and heading stability, verifying the algorithm's superiority and effectiveness. Second, simulation experiments were performed using the improved algorithm under various encounter situations, such as ships meeting each other in the channel and ships crossing paths with fishing vessels in the channel, verifying the effectiveness of the collision avoidance decision-making.

[0239] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions 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 automatic collision avoidance path planning for ships based on differential evolution algorithm with population size adaptability, characterized in that, include: Acquire information on the surrounding environment and sea conditions of other vessels, establish a vessel domain model, and calculate the safe encounter distance and motion parameters for vessel course changes; Differential evolution algorithms that incorporate population size adaptability include: Population initialization using a differential evolution algorithm for population size adaptation: The differential evolution algorithm uses real-number encoding to generate a population of dimension (NP, Q) within the feasible region of the search space, where NP represents the population size and Q represents the number of decision variables. Then, the i-th... i individual x i for ; Representing the i The individual j Dimension value, It is generated randomly within the decision space, and the initialization expression is: ， in, This represents generating uniformly distributed random numbers between 0 and 1; At the generation of each generation of the population, the lifetime value of each individual is calculated based on the fitness function value: When the population is initialized, the age of an individual is 0, meaning that the individual's generation number and age are the same. When the age value of an individual in the population is greater than its lifespan, it is deleted from the population. The specific calculation formula is as follows: If the fitness function value of a population remains unchanged for a consecutive number of times, the current generation of the population will fall into a local extreme value. It will then escape the local extreme value through the extinction mechanism, thereby increasing the diversity of the population. When some individuals are removed through the lifespan mechanism, the integrity of the population is maintained by replenishing them with new individuals: By replicating elite individuals with high fitness function values ​​from the population, the following formula is used for calculation: ， The population size changed as follows after the improvement: When the extinction mechanism comes into play, in order to ensure population diversity and randomness, a new population is established using the following formula: Adding the best individuals from an extinct population to a new population increases the population size by "1", thus changing the population size to: A differential evolution algorithm based on population size adaptability is used to construct a ship collision risk model, including: Based on fuzzy comprehensive evaluation, the risk level of ship collision is calculated; Construct a fitness function for ship collision avoidance path planning; The fitness function is solved based on the improved differential evolution algorithm; Based on the degree of ship collision risk, collision avoidance decisions are designed.

2. The ship automatic collision avoidance path planning method based on differential evolution algorithm with population size adaptability according to claim 1, characterized in that, The process of acquiring information on the surrounding environment and sea conditions of other vessels, establishing a vessel domain model, and calculating the safe encounter distance and motion parameters of the vessel specifically includes: Construct a quaternion-based ship domain model (QSD), whose analytical expression and boundary equation expressions are as follows: To determine the analytical expression for the safe encounter distance, calculations were performed based on a quaternion ship domain model, yielding the ship domain boundary values ​​for different target ship hull angles: Calculate the relative motion parameters DCPA and TCPA between the ship and the target ship using the following formulas: Assuming the relative positions of the two ships remain unchanged at the moment of turning, calculate the ship's kinematic parameters DCPA' and TCPA' during the course change using the following formulas: 。 3. The ship automatic collision avoidance path planning method based on differential evolution algorithm with population size adaptability according to claim 1, characterized in that, The calculation of ship collision risk based on fuzzy comprehensive evaluation specifically includes: Establish a factor set: The six indicators used as inputs when calculating the collision risk of this vessel with other vessels or obstacles are: nearest encounter distance, time of nearest encounter, distance between vessels, relative bearing, speed ratio, and channel dimensions, denoted as […]. , , , , , ; The factor set U is constructed using 6 indicator parameters as follows: Establish an evaluation set: The target vessel evaluation set contains two items: one indicating the target vessel's danger and the other indicating its safety, denoted as [insert items here]. and ; This indicates a situation where evasive action must be taken if the vessel poses a collision hazard to other vessels or obstacles. This indicates that there is no risk of collision between the vessel and other vessels or obstacles, meaning no evasive action is required; therefore, an evaluation set is established. as follows: Determine the indicator weights: The weight set W is constructed based on the factor set as follows: In the formula, , , , , , Represent , , , , , The weight value; Among the indicators affecting the risk of ship collisions, the weights of each indicator satisfy the following: , And it satisfies the normalization condition: The weights of each indicator are determined as follows: , , , , , The weight matrix is ​​as follows: Establish the membership function of the evaluation index; A comprehensive evaluation function for calculating the risk of ship collisions: The target evaluation matrix includes the membership functions of the above six indicators, as shown in the following formula. in, , , , It is the membership function of the factors influencing the risk of collisions between ships; Taking into account the impact of the six evaluation indicators, the evaluation results of the target ship are as follows: Therefore, the expression for collision hazard level is: 。 4. The ship automatic collision avoidance path planning method based on differential evolution algorithm with population size adaptability according to claim 3, characterized in that, The fitness function for constructing the ship collision avoidance path planning specifically includes: Construct an objective function based on ship collision risk: Construct an objective function based on the total flight distance: ； Construct an objective function based on steering amplitude: Construct an objective function based on collision avoidance rules: Construct an objective function based on the timing of avoidance: 。 5. The ship automatic collision avoidance path planning method based on differential evolution algorithm with population size adaptability according to claim 4, characterized in that, The method of solving the fitness function based on the improved differential evolution algorithm specifically includes: An improved differential evolution algorithm is used to target decision variables. x The multi-objective function is optimized, and the decision variables are... x It is a vector set of steering amplitude; S332. Based on the five constructed objective functions, establish a fitness function during navigation within the waterway. as follows: Among them, the weight coefficients of the objective function are... , and satisfy: Then, within the waterway, the fitness function optimized based on the improved differential evolution algorithm is expressed as: In the case of open water, the fitness function optimized based on the improved differential evolution algorithm is expressed as: 。 6. The ship automatic collision avoidance path planning method based on differential evolution algorithm with population size adaptability according to claim 5, characterized in that, The design of collision avoidance decisions based on the degree of ship collision risk specifically includes: Assuming the navigation area is free from the influence of wind, waves, and currents; when designing collision avoidance decisions, only the collision avoidance measure of turning the ship is considered, and the measure of changing speed is not used for collision avoidance; at the same time, according to the requirements of good seamanship, the turning amplitude function is designed. Selecting the collision avoidance timing involves determining the vessel's collision avoidance opportunity based on a set threshold. The multi-objective function is transformed into a single-objective function with the steering amplitude as the independent variable by using the weighting method.

Images