Limited water area safety route auxiliary decision-making method based on stress intensity model

Abstract

The invention relates to a stress intensity model-based auxiliary decision-making method for a water-area-limited safe route, which comprises the following steps of: firstly, introducing heading angle disturbance psi b in a navigation process of an intelligent ship, and simulating track deviation in a real environment to enable the track deviation to deviate from a set path and course; calculating the track offset of the ship when the ship recovers to the specified path and course under the action of the path tracking controller, controlling the ship to return to the preset route by using the autonomous navigation module, and recording the maximum transverse offset distance eta max; then, according to the relationship between the track offset and the channel space, estimating the probability of collision between the ship and the quay wall by adopting a stress intensity model; evaluating the safety of the route based on the collision probability, and assisting in deciding an optimal safe route. According to the method, firstly, the adjusting capacity of the intelligent ship for dealing with the track deviation disturbance is comprehensively evaluated, and the relation between the adjusting capacity and the safe bearing capacity of the channel space is analyzed, so that quantitative evaluation of the navigation safety is achieved. And safe route decision support is provided for the intelligent ship in the limited water area.

Description

Technical Field

[0001] The present invention relates to the technical field of safe route auxiliary decision-making for intelligent ships, and specifically, to a ship navigation safety assessment method based on a stress interference model, which can calculate the collision probability between a ship and a quay wall, and overcomes the problems of not giving priority to navigation safety in traditional route decision-making methods, requiring large historical trajectory data of ships, and being unsuitable for navigation safety assessment of intelligent ships in small-scale spaces. It can assist in the decision-making of a safe route in an environment with complex and intertwined waterways. Background Art

[0002] With the rapid development of intelligent ship technology, improving the navigation safety of intelligent ships has become a hot research topic. Restricted waters pose a particularly significant challenge to the navigation safety of ships. In order to improve the navigation safety of intelligent ships in restricted waters, safe routes can be selected at key locations such as channel intersections by assisting decision-making on safe routes, thereby effectively reducing the risk of collision between ships and quay walls.

[0003] Common route decision-making methods include path planning algorithms and data-driven methods.

[0004] The path planning algorithm aims to plan a feasible route between a specified starting point and a designated end point that avoids obstacles. Classic path planning algorithms, such as the A* algorithm, the fast marching method, and the random exploration tree, usually aim to optimize the route length and generate the shortest route from the starting point to the end point. In order to improve the safety of the route, researchers have optimized the distance between the route and the quay wall based on the shortest route to ensure that there is a certain safety space when the ship is maneuvering. However, on the one hand, this strategy puts route safety behind route length and fails to focus on the study of safe route decisions; on the other hand, it only optimizes a single factor and it is difficult to fully consider the comprehensive impact of multiple factors on the safety of ship navigation.

[0005] The data-driven method aims to utilize a large amount of historical ship trajectory data, mainly relying on the ship's Automatic Identification System (AIS) data, and quantify the impact of multiple factors on the safety of ship navigation, so as to estimate the probability of a ship's safe passage on a specific route. Classic data-driven methods, such as Bayesian networks, quantify each factor by grade and establish conditional probability relationships between them to calculate the probability of a ship's safe passage. However, the data-driven method relies on a large amount of historical ship trajectory data and is more suitable for manned ships in busy waters; in restricted waters, especially in the stage where smart ships have not yet been widely used, it is difficult to obtain the corresponding historical trajectory data. More importantly, the quantitative method of dividing factors into "excellent", "good", "qualified" and other levels is difficult to accurately characterize the actual performance of the ship's autonomous navigation module and key factors such as ship maneuverability.

[0006] The path planning algorithms commonly used for route decision-making mainly include: graph search algorithms such as the A* algorithm, the Rapidly-exploring Random Tree (RRT) algorithm, and the Dynamic Window Approach (DWA). The A* algorithm is a heuristic search method that aims to explore the optimal path length from the starting point to the target point. Researchers increase the safe distance between path points and obstacles by dilating obstacles. The RRT algorithm randomly samples feasible path points and continuously iterates to generate paths to optimize the length of the path. The DWA algorithm explores feasible paths to the destination under kinematic constraints such as the ship's speed and acceleration. The safety of the path is improved by increasing the weight of the distance between the ship and the obstacle in the objective function.

[0007] Data-driven methods mainly include clustering algorithms and Bayesian networks. Clustering algorithms analyze a large amount of navigation trajectory data in specific waters to determine all routes between the starting point and the end point. This type of method mainly extracts the connection between waterways, but does not consider the safety factors of the waterways. Bayesian networks can construct conditional probability relationships between various risk factors to estimate the probability of a ship's safe passage on a specific route.

[0008] When generating routes based on path planning algorithms, researchers have increased the distance between the route and the quay to ensure that the ship has enough maneuvering space, or smoothed the route to improve path tracking performance, thereby improving the safety of the route. However, such strategies fail to take navigation safety as the primary goal, and it is difficult to comprehensively consider the impact of more factors on navigation safety.

[0009] Researchers often face the problem of data scarcity when using data-driven methods to generate routes or estimate the probability of safe passage of ships on specific routes. Especially in the context of the fact that smart ships have not yet been widely used, the corresponding navigation data is very limited. More importantly, when evaluating the navigation safety of ships based on Bayesian networks, the influencing factors are usually roughly graded and quantified, such as dividing the ship's autonomous navigation system into "excellent", "good", "qualified" and other levels. Such a quantitative method makes it difficult to accurately evaluate the impact of various factors on the navigation safety of ships. When smart ships are sailing in restricted waters, the tolerance range of ship maneuvers is extremely small, and Bayesian networks are difficult to adapt to the needs of refined evaluation in small-scale spaces.

[0010] The above methods are difficult to help smart ships comprehensively consider navigation safety in restricted waters, thereby supporting safe route decisions. Summary of the invention

[0011] In order to support the safe navigation of intelligent ships in restricted waters, this paper proposes a navigation safety assessment method based on stress intensity model. By comprehensively evaluating the adjustment ability of intelligent ships to cope with track deviation disturbances and its relationship with the safety tolerance of waterway space, the impact of autonomous navigation modules on navigation risks is quantitatively analyzed, thereby achieving navigation safety assessment.

[0012] To achieve the above object, the technical solution of the present invention is: a method for assisting decision-making of safe routes in restricted waters based on a stress intensity model, comprising the following steps:

[0013] (a) Introducing heading angle disturbance ψ during the navigation process of the intelligent ship b , simulate the track deviation in real environment;

[0014] (b) Use the autonomous navigation module to control the ship to return to the scheduled route and record the maximum lateral deviation η max ;

[0015] (c) Based on the relationship between the maximum lateral deviation and the channel width, the probability of collision between the ship and the quayside is calculated using the stress intensity model;

[0016] (d) Evaluate the safety of the route based on the collision probability and assist in deciding the optimal safe route.

[0017] Furthermore, the heading angle disturbance ψ b Follow a certain probability distribution, which is established based on a small amount of historical navigation data. Without loss of generality, the present invention uses the normal distribution N(μ,σ 2 ).

[0018] Furthermore, the autonomous navigation module includes a LOS guidance method and a PD controller, wherein: the LOS guidance method dynamically calculates the desired heading angle ψr , the discrete scheduled route is the target point sequence P k ; PD controller generates rudder angle input δ c , through the control law Achieve heading error e ψ The convergence of is the heading error, K p and K d are the proportional and derivative gain parameters of the PD controller.

[0019] Furthermore, the LOS guidance method dynamically calculates the desired heading angle ψ r , guide the ship back to the scheduled route; first, the scheduled route can be discretized into multiple target points Calculate the scheduled route corresponding to the ship from P k-1 To P k Tangent angle a of the segment k

[0020]

[0021] On this basis, the expected heading angle ψ of the ship is calculated r

[0022]

[0023] Where [x, y] represents the current position of the ship and Δ represents the forward distance.

[0024] Furthermore, the stress intensity model calculates the probability density distribution of the lateral offset by a kernel density estimation method, and performs interference analysis with the channel width to determine the collision probability.

[0025] Furthermore, the method further includes constructing a route network model. The process of establishing the route network model can be intuitively understood as simulating an object moving at a constant speed in a waterway, optimizing the distance between the object and the quay, and always keeping the object as far away from the quay as possible, thereby obtaining the central axis of the waterway as a reference route:

[0026] (1) Simplify the channel boundary into point elements H p and line element H l ;

[0027] (2) Based on the transformation relationship between the distances of objects and geometric elements, the centerline of the waterway is optimized as the reference route.

[0028] Furthermore, in the route network model, objects and line elements H l The distance change relationship is:

[0029]

[0030] Among them, [x, y] is the coordinate of the object, [x′, y′] is the coordinate of the object projected on the line element, v is the constant moving speed of the object, φ is the moving direction angle of the object, is the distance between the object and the line element; the distance between the object and the point element H p The transformation relationship between the distances is:

[0031]

[0032] in, Is the distance between the object and the point feature.

[0033] Furthermore, the ship maneuverability model adopts the Mariner model.

[0034] The Mariner model is formally expressed as:

[0035]

[0036] Among them, s=[x,y,ψ,u,v,r,δ] T , represents the state vector of the ship, including the position coordinates x and y of the 3-DOF ship model in the fixed coordinate system, the heading angle ψ, the longitudinal velocity u, the lateral velocity v and the bow angular velocity r, and the rudder angle δ in the body coordinate system. c represents the rudder angle input command of the ship, U0 represents the forward speed input of the ship, W=[w s , w θ ] T Indicates wind speed w s and wind direction θ .

[0037] Furthermore, the wind speed w s The Weibull distribution is used for modeling, and its probability density function is:

[0038]

[0039] Among them, the parameters k and λ are fitted to the historical wind speed data by the maximum likelihood estimation method; the maximum likelihood estimation method is used to fit the historical wind speed data to solve the parameters k and λ of the Weibull distribution.

[0040] Furthermore, the method is applicable to restricted waters, including narrow irregular waterways, areas with crisscrossing waterways, and low-speed scenarios, and for specific smart ships, it evaluates their collision risk under uncertain track disturbances under a given path tracking controller.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. Conduct navigation safety assessment for smart ships. Analyze the ship's ability to adjust to track deviation disturbances under a given path tracking control, quantify the multiple impacts of channel space limitations, uncertainty of wind load interference, ship maneuverability constraints, and autonomous navigation module performance on navigation safety, and provide support for safe route decision-making;

[0043] 2. For the special navigation environment of restricted waters. Under the limited spatial scale, the margin of error for ship maneuvering is small, and the risk factors need to be accurately characterized;

[0044] 3. Low reliance on historical ship trajectory data. Based on ship motion simulation modeling, the navigation risk of a specific ship under a specific voyage is estimated.

[0045] The present invention uses a stress intensity model to comprehensively evaluate the impact of various risk factors on the navigation safety of intelligent ships. In order to simulate the uncertainty of the track of intelligent ships in a real environment, the ship is first deviated from the set route by introducing a bow disturbance. Subsequently, the present invention uses the LOS guidance method and the PD controller as a path tracking module to control the ship to return to the specified route under the influence of constant wind force. During the ship's return process, the lateral deviation between the actual trajectory and the specified route is recorded. The size of the lateral deviation is comprehensively affected by the ship's maneuverability, wind force and control module performance. When the lateral deviation exceeds the channel width, it is considered that the ship has collided with the shore. On this basis, using the stress intensity model, the bow angle of the ship disturbance is randomly sampled, the lateral deviation distribution at a specific position during the ship's travel is obtained, and the interference between the lateral deviation distribution and the channel width at this position is calculated, thereby obtaining the probability of the ship colliding with the shore. Therefore, the present invention can comprehensively evaluate the impact of channel space restrictions, uncertain wind load interference, ship maneuverability constraints and autonomous navigation module performance on the navigation safety of intelligent ships, and support safe route decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 Help smart ships return to the prescribed route through autonomous navigation modules;

[0047] Figure 2 It is a schematic diagram of the relationship between the movement of objects and geometric elements;

[0048] Among them: (a) is a line feature, (b) is a point feature;

[0049] Figure 3 This is a flow chart of the auxiliary decision-making of the safe route of the present invention. DETAILED DESCRIPTION

[0050] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0051] The embodiment of the present invention proposes a navigation safety assessment method based on a stress intensity model, the steps of which include:

[0052] In the first step, in order to comprehensively evaluate the impact of multiple risk factors on the navigation safety of smart ships, a navigation safety assessment method for smart ships under track disturbance is adopted. This method can effectively reflect the adjustment ability of smart ships to cope with track deviation, especially the impact of the ship's own maneuverability and the autonomous navigation module on navigation safety. It provides a basis for calculating the probability of a ship colliding with a quay wall at a specific location.

[0053] The second step is to build a simulation environment: digitally model the waterway and establish a corresponding route network to explore all feasible routes; construct a normal distribution model of the ship's disturbance heading angle; establish a ship maneuverability model using the classic Mariner model; build a ship autonomous navigation module, using the LOS guidance method and PD controller as the path tracking module; and construct a Weibull distribution model of wind speed.

[0054] The third step is to select multiple locations at equal intervals on the scheduled route and randomly sample the ship's disturbance heading angle and wind speed at each location. According to the content of Section 4.1, the lateral deviation distribution of the ship at each location is obtained, and the interference between the lateral deviation distribution and the channel width at that location is calculated, so as to determine the probability of collision at each location when the ship travels along a specific route.

[0055] Among them: The first step is the navigation safety assessment method under the track disturbance of intelligent ships. The specific implementation steps are as follows: Figure 1 .

[0056] Step 1: Considering the uncertainty of the ship's track in the real environment, a heading angle disturbance ψ is introduced during its travel. b The probability distribution of heading angle disturbance can be established based on a small amount of historical track data;

[0057] Step 2: Calculate the maximum lateral deviation η when the ship returns to the specified route under the control of the autonomous navigation module max . Maximum lateral deviation η max It mainly reflects the impact of the ship's own maneuverability and the performance of the autonomous navigation module on navigation safety.

[0058] Step 3: Determine whether the ships will collide based on the relationship between the maximum lateral deviation of the ship and the width of the channel.

[0059] The above process can be used as the basis for calculating the probability of collision when ships are traveling along a specific route.

[0060] The second step is to establish a simulation environment. The specific implementation steps are as follows:

[0061] Step 1: Establish a route network model to identify all potential routes and key intersection nodes in complex waters. The process of establishing a route network model can be intuitively understood as simulating an object moving at a constant speed in the channel and optimizing the distance between the object and the shore, always keeping the object as far away from the shore as possible, so as to obtain the central axis of the channel as a reference route. The specific implementation steps are:

[0062] Step 1.1: Define the waterway boundary by point element H p , line element H l Simplify expressions;

[0063] Step 1.2: Establish the transformation relationship between the distance between the object and the geometric elements during the movement, see Figure 2 . Object and line element H l The transformation relationship between the distances is:

[0064]

[0065] Among them, [x, y] is the coordinate of the object, [x′, y′] is the coordinate of the object projected on the line element, v is the constant moving speed of the object, φ is the moving direction angle of the object, is the distance between the object and the line feature. p The transformation relationship between the distances is:

[0066]

[0067] in, Is the distance between the object and the point feature.

[0068] Step 1.3: Establish a route network model, calculate the moving direction angle of the object based on formulas (1) and (2), and optimize the distance between the object and the quayside to identify all potential routes and key intersection nodes. When the number of geometric elements closest to the object is 1, calculate the moving direction angle φ of the particle so that the distance change relationship between the object and the geometric element H1 satisfies

[0069]

[0070] When the number of geometric elements closest to the object is 2, the moving direction angle φ of the particle is calculated so that the distance change relationship between the object and the geometric elements H1 and H2 satisfies

[0071]

[0072] When the number of geometric elements closest to the object exceeds 2, the object is located at the key node of the channel intersection. At the channel intersection, the object can choose a channel so that the number of geometric elements closest to the object is 2, and repeat the calculation process of formula (4). After traversing all intersections, the exploration of all potential routes is completed.

[0073] Step 2: Construct the distribution model of ship disturbance heading angle

[0074] When a smart ship is sailing in a real environment, there is a certain uncertainty deviation in the track. We use the bow disturbance angle to describe this deviation. In a real environment, the bow disturbance angle is affected by factors such as wind, current and the ship itself, and obeys a certain probability distribution, which can usually be established based on a small amount of navigation data. The present invention directly uses the normal distribution to describe this disturbance. The ship bow disturbance angle ψ b ψ b The probability distribution of follows a normal distribution with a mean of 10° and a variance of 2.

[0075] ψ b ~N(10°,2 2 ). (5)

[0076] Step 3: Build a ship maneuverability model

[0077] The present invention adopts the Mariner model to describe the dynamic response relationship between the ship's own state (such as position, heading, speed, acceleration, etc.) and the control input under the combined action of external environmental loads (such as wind torque) and internal control input (such as rudder angle) at a given speed. The Mariner model can be formally expressed as:

[0078]

[0079] in, Represents the state vector of the ship, including the position coordinates x and y of the 3-DOF unmanned boat model in the fixed coordinate system, the heading angle ψ, and the longitudinal velocity u, lateral velocity v and bow angular velocity r, and rudder angle δ in the body coordinate system. c represents the rudder angle input command of the ship, U0 represents the forward speed input of the ship, W=[w s , w θ ] T Indicates wind speed w s and wind direction θ For more details, please refer to the Fossen toolbox at https: / / github.com / cybergalactic / MSS.

[0080] Step 4: Establish ship autonomous navigation module

[0081] The present invention adopts LOS guidance method and PD controller as the autonomous navigation module of the ship to realize tracking control of the set route. Under the action of external wind load, by adjusting the rudder angle input δ, the ship returns to the desired route and maintains a stable navigation state.

[0082] Step 4.1: LOS guidance method dynamically calculates the desired heading angle ψ r , guide the ship back to the scheduled route. First, the scheduled route can be discretized into multiple target points Calculate the scheduled route corresponding to the ship from P k-1 To P k Tangent angle a of the segment k

[0083]

[0084] On this basis, the expected heading angle ψ of the ship is calculated r

[0085]

[0086] Where [x, y] represents the current position of the ship, and Δ represents the foresight distance, which is usually set manually based on factors such as the ship's length and speed. The larger the foresight distance Δ, the smoother the track.

[0087] Step 4.1: PD controller is used to generate the rudder angle input command δ c , so that the ship heading angle ψ gradually converges to the desired heading angle ψ r The rudder angle control law of the PD controller can be expressed as:

[0088]

[0089] In, e ψ =ψ r -ψ is the heading error, K p and K d are the proportional and derivative gain parameters of the PD controller.

[0090] Step 5: Build a Weibull model for wind speed

[0091] Step 5.1: Wind speed data collection

[0092] The present invention is based on the wind speed data collected by the German Meteorological Bureau in the Hamburg Port area.

[0093] Website: https: / / www.dwd.de / EN / Home / home_node.html.

[0094] Step 5.2: Probability density function modeling

[0095] Usually the Weibull distribution is used to calculate the wind speed w s To model the model, its probability density function is defined as follows:

[0096]

[0097] Among them, k is the shape parameter and λ is the scale parameter, both of which determine the characteristics of the distribution.

[0098] Step 5.3: Parameter estimation

[0099] The maximum likelihood estimation method is used to fit the historical wind speed data and solve the parameters k and λ of the Weibull distribution.

[0100] The third step is to evaluate the safety of the route. The specific implementation steps are as follows:

[0101] Based on the first and second steps, the probability of a ship colliding with a quay wall when traveling along a specific route can be evaluated to support safe route decision-making. For detailed process, please refer to the process Figure 3 .

[0102] Step 1: Select a predetermined route from the route network. The route is also the central axis of the channel and can be used as a reference path for the ship. Discrete the route into several discrete points to calculate the probability of collision between the ship and the shore at different locations; Step 2: At each discrete point, randomly sample the ship's bow disturbance angle and wind speed;

[0103] Step 3: Under the action of the autonomous navigation module, the ship is made to return to the predetermined route and the corresponding maximum lateral deviation is calculated;

[0104] Step 4: Use the kernel density estimation method to estimate the probability density distribution of the maximum horizontal deviation at each discrete point;

[0105] Step 5: Calculate the interference between the probability density distribution of the maximum lateral deviation at each discrete point and the channel width, that is, calculate the probability that the maximum lateral deviation exceeds the channel width, which is the collision risk. After obtaining the probability of collision with the quay at each discrete point, the overall risk of collision with the quay when the ship travels along the scheduled route can be evaluated;

[0106] Step 6: Calculate the collision probability of each feasible route to the destination to support safe route decision.

Claims

1. A method for assisting decision making of safe routes in restricted waters based on a stress intensity model, characterized in that: The following steps are involved: (a) Introducing heading angle disturbance ψ during the navigation process of the intelligent ship b , simulate the track deviation in real environment; (b) Use the autonomous navigation module to control the ship to return to the scheduled route and record the maximum lateral deviation η max ; (c) Based on the relationship between the maximum lateral deviation and the channel width, the probability of collision between the ship and the quayside is calculated using the stress intensity model; (d) Evaluate the safety of the route based on the collision probability and assist in deciding the optimal safe route.

2. The method according to claim 1, characterized in that The heading angle disturbance ψ b It follows a probability distribution, which is built on a small amount of historical navigation data.

3. The method according to claim 1, characterized in that: The autonomous navigation module includes a LOS guidance method and a PD controller, wherein: the LOS guidance method dynamically calculates the desired heading angle ψ r , the discrete scheduled route is the target point sequence P k ; PD controller generates rudder angle input δ c , through the control law Achieve heading error e ψ The convergence of ψ =ψ r -ψ is the heading error, K p and K d are the proportional and derivative gain parameters of the PD controller.

4. The method according to claim 3, characterized in that The LOS guidance method dynamically calculates the desired heading angle ψ r , guide the ship back to the scheduled route; first, the scheduled route can be discretized into multiple target points Calculate the scheduled route corresponding to the ship from P k-1 To P k Tangent angle a of the segment k On this basis, the expected heading angle ψ of the ship is calculated r Where [x, y] represents the current position of the ship and Δ represents the forward distance.

5. The method according to claim 1, characterized in that The stress intensity model calculates the probability density distribution of the lateral deviation through the kernel density estimation method, and performs interference analysis with the channel width to determine the collision probability.

6. The method according to claim 1, characterized in that The method further includes constructing a route network model. The process of establishing the route network model can be intuitively understood as simulating an object moving at a constant speed in a waterway, optimizing the distance between the object and the quay, and always keeping the object away from the quay, thereby obtaining the central axis of the waterway as a reference route: (1) Simplify the channel boundary into point elements H p and line element H l ; (2) Based on the transformation relationship between the distances of objects and geometric elements, the centerline of the waterway is optimized as the reference route.

7. The method according to claim 6, characterized in that In the route network model, objects and line elements H l The distance change relationship is: Among them, [x, y] is the coordinate of the object, [x′, y′] is the coordinate of the object projected on the line element, v is the constant moving speed of the object, φ is the moving direction angle of the object, is the distance between the object and the line feature; Objects and point elements H p The transformation relationship between the distances is: in, Is the distance between the object and the point feature.

8. The method according to claim 1, characterized in that The ship maneuverability model adopts the Mariner model, which is formally expressed as: in, Represents the state vector of the ship, including the position coordinates x and y of the 3-DOF ship model in the fixed coordinate system, the heading angle ψ, the longitudinal velocity u, the lateral velocity v and the bow angular velocity r in the body coordinate system, and the rudder angle δ, δ c represents the rudder angle input command of the ship, U0 represents the forward speed input of the ship, W=[w s , w θ ] T Indicates wind speed w s and wind direction θ .

9. The method according to claim 1, characterized in that: The wind speed w s The Weibull distribution is used for modeling, and its probability density function is: Among them, the parameters k and λ are fitted to the historical wind speed data by the maximum likelihood estimation method; the maximum likelihood estimation method is used to fit the historical wind speed data to solve the parameters k and λ of the Weibull distribution.

10. The method according to any one of claims 1 to 9, characterized in that: The method is applicable to restricted waters, including narrow irregular waterways, areas with crisscrossing waterways, and low-speed scenarios. It also evaluates the collision risk of specific smart ships under uncertain track disturbances under a given path tracking controller.

Images