A ballast water inlet and outlet control system, method, apparatus and medium

Abstract

This invention relates to a ballast water inlet and outlet control system, method, apparatus, and medium for controlling the automatic inlet and outlet of ballast water tanks. The ballast water tanks consist of multiple sub-tanks, each equipped with an inlet valve, an outlet valve, and a vent valve. Multiple depth gauges are also installed on the side walls. The system further includes a drive valve box installed in the control cabin and several levels installed at the bow, midship, and aft points of the hull to detect the ship's roll and pitch angles. An inlet and outlet neural network controller, also installed in the control cabin, acquires the measurement results from the depth gauges and levels, calculates and judges the results from the multiple depth gauges, and issues control commands to the drive valve box based on the judgment results. This causes the drive valve box to control the high-pressure water pump, the inlet and outlet valves, and the vent valve to achieve the inlet and outlet of the ballast tanks. Compared with existing technologies, this invention has advantages such as high automation and high efficiency.

Description

Technical Field

[0001] This invention relates to the field of dynamic distribution and control technology of ballast water in floating crane ballast systems and ballast tanks, and in particular to a ballast water inlet and outlet control system, method, device and medium. Background Technology

[0002] Floating cranes are widely used in marine rescue, salvage, and large-scale offshore infrastructure projects, playing a vital role in marine engineering. Considering the complexity of marine engineering operations, floating crane design tends towards large-scale, multi-ship collaborative, intelligent, and automated deep-sea operations. Lifting and lowering operations can cause the floating crane to tilt and sag. This problem becomes more prominent and critical as the size of floating cranes increases. The overturning moment generated by lifting cargo can cause tilt angles of 7-8 degrees or even greater, seriously affecting operational safety. Ballast water systems, by transferring ballast water between ballast tanks, regulate draft and hull balance, counteracting tilt and sag moments and ensuring the safe operation of the vessel. Therefore, the distribution of ballast water between the ballast system and ballast tanks is crucial for ensuring the safe operation of floating cranes. Offshore operations are frequently affected by the marine environment, resulting in short periods suitable for offshore operations. Furthermore, floating cranes need to perform lifting operations quickly under safe conditions. Therefore, researching efficient algorithms for solving floating crane distribution schemes has become an urgent task. This not only ensures the safety of floating crane operations at sea but also improves the level of automation in offshore operations.

[0003] In recent years, some researchers have studied the mathematical modeling of ship ballast processes, the solution algorithms for ballast water allocation, and optimization design methods. Regarding optimization modeling and solution algorithms, Samyn proposed a six-degree-of-freedom dynamic ballast water control system for a semi-submersible platform, integrating the effects of ballast tank mass, inertia, and moment; Chen proposed a submarine hovering control method based on LI adaptive theory, establishing dynamic models of the ballast tank and the submarine; and Zhou proposed using a multi-objective optimization algorithm based on decomposition technology to optimize ballast water configuration.

[0004] From an engineering optimization perspective, ballast water allocation optimization is a complex engineering optimization problem with constraints, presenting multiple challenges including engineering mathematical complexity and modeling / solution complexity. Factors influencing the dynamic allocation of ballast water include cargo handling status, marine environmental load, ballast system status, and ship hull status. Ballast system status includes ballast method, ballast tank arrangement, and ballast tank water level. Previous studies have shown that relying solely on optimization models and general solution algorithms is insufficient to meet engineering requirements in terms of efficiency and quality. Summary of the Invention

[0005] The purpose of this invention is to provide a ballast water inlet and outlet control system, method, device and medium that collects ship attitude information and uses a multi-variable multi-objective task reinforcement learning control method based on the attitude information to automatically adjust and control the ballast water inlet and outlet volume, thereby controlling the inlet and outlet process of the ballast water tanks and achieving the purpose of controlling the ship's stability. This solves the technical problems of low automation and slow efficiency in the current ballast water tank inlet and outlet process.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A ballast water inlet and outlet control system, wherein the ballast water tank is composed of multiple sub-tanks, the system comprising:

[0008] Each sub-tank is equipped with two electronic valves to control the water intake and drainage of the ballast water tank.

[0009] The vent valve is installed on the top of the sub-water tank. The vent valve opens when the sub-water tank is being filled or drained, and closes when the sub-water tank is not in operation.

[0010] Depth gauges, installed on the sidewalls of each sub-tank, are used to measure the water level in the ballast tanks;

[0011] The drive valve box is installed in the control compartment and is connected to the water inlet valve, drain valve and vent valve via signal lines;

[0012] The level is installed at the bow, midship and aft of the hull to detect the hull's roll and pitch angles.

[0013] An intake and drainage neural network controller is installed in the control compartment and is connected to the drive valve box, depth gauge, and level via signal lines. The controller acquires the measurement results from the depth gauge and level, calculates and judges the measurement results, and issues control commands to the drive valve box based on the judgment results. The drive valve box controls the high-pressure water pump, intake and drainage valve, and vent valve to realize the intake and drainage of the ballast tank.

[0014] The inlet valve is located on the upper side of the sub-water tank and is used to control the opening and closing of the inlet of the sub-water tank; the drain valve is located on the lower side of the sub-water tank and is used to control the opening and closing of the drain outlet of the sub-water tank.

[0015] The depth gauges are symmetrically mounted on the side walls of the sub-water tank.

[0016] A method for controlling the inflow and outflow of ballast water includes the following steps:

[0017] S1. Obtain the roll and pitch angles of the hull as measured by the level.

[0018] S2. Obtain the depth detection information measured by the depth gauge of each sub-water tank;

[0019] S3. Repeat steps S1 and S2, record the measured values ​​at multiple times as the state variables of the ballast water loading process, and input the state variables into the inflow and outflow neural network controller based on deep reinforcement learning. Treat the ballast water loading process as a partially observable Markov process, and use deep reinforcement learning to perform self-learning to realize the control strategy output.

[0020] S4. The inlet and outlet neural network controller outputs the control strategy to the drive valve box, so that the drive valve drives the inlet valve, outlet valve, vent valve and high-pressure water pump connected to it respectively, so as to realize the automatic control of the inlet and outlet of the water tank.

[0021] S5. Return to steps S1 and S2, measure the hull roll angle and depth detection information, calculate the reward function value in real time, and update the parameters of the intake and drainage neural network controller according to the reward value;

[0022] S6. Based on the results measured in S5, if balance has not been achieved or the operation has not stopped, return to S3 to continue the ballast water balance loading operation; if balance has been achieved and the operation has ended, calculate the final reward function value and the cumulative reward, and update the parameters of the influent and drainage neural network controller according to the cumulative reward and the strategy gradient.

[0023] The side roll angle and fore and aft roll angle of the hull are the average values ​​of the corresponding level measurements.

[0024] The depth detection information of the sub-water tank is represented in vector form.

[0025] In step S5, the reward function value is calculated as follows:

[0026]

[0027] Where, α M For the roll safety angle, β M Let |α(t)| be the heel safety angle, |β(t)| be the absolute value of the heel angle at time t, and |β(t)| be the absolute value of the heel angle at time t.

[0028] The inlet and outlet neural network controller uses the minimum total ballast time as the optimization objective. The total ballast time of the ballast system includes the time for transferring ballast water between ballast water tanks, the response time of the tank valves, and the discharge time of the ballast water.

[0029] A ballast water inlet and outlet control device includes a memory, a processor, and a program stored in the memory, wherein the processor executes the program to implement the method described above.

[0030] A storage medium having a program stored thereon, which, when executed, implements the method described above.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] (1) The ballast water tank inlet and outlet control system of the present invention has a simple structure, does not require a symmetrical structure, is flexible in installation, and is highly practical.

[0033] (2) The ballast water tank inlet and outlet control method of the present invention utilizes the reward feedback of reinforcement learning to optimize control parameters, automatically determine control measurements, and is highly efficient. It is also robust to the interference of sea waves, providing a feasible solution for automatic ballast water loading. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the arrangement of the sub-water tank and the level gauge in one embodiment.

[0035] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0036] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0037] This embodiment first provides a ballast water inlet and outlet control system, wherein the ballast water tank is composed of multiple sub-tanks, and the system includes:

[0038] (1) Water inlet valve and water outlet valve: Each sub-water tank is equipped with two electronic valves to control the water inlet and water outlet of the ballast water tank respectively. The water inlet valve is located on the upper side of the sub-water tank and is used to control the opening and closing of the water inlet of the sub-water tank; the water outlet valve is located on the lower side of the sub-water tank and is used to control the opening and closing of the water outlet of the sub-water tank.

[0039] (2) Several vent valves are installed on the top of the sub-water tank. The vent valves are opened when the sub-water tank is filled or drained, and closed when the sub-water tank is not in operation.

[0040] (3) Several depth gauges are symmetrically installed on the side wall of each sub-tank to measure the water level of the ballast tank;

[0041] (4) Drive valve box, installed in the control compartment, connected to the inlet valve, drain valve and vent valve via signal lines;

[0042] (5) Several levels are installed at the front, middle and rear of the ship to detect the roll angle and fore and aft angle of the ship;

[0043] (6) Inlet and outlet neural network controller, which is installed in the control cabin and is connected to the drive valve box, depth gauge and level through signal lines to obtain the measurement results of the depth gauge and level, and to calculate and judge the measurement results. Based on the judgment results, the controller sends control commands to the drive valve box to control the high-pressure water pump, the inlet and outlet valve and the vent valve to realize the inlet and outlet of the ballast tank.

[0044] like Figure 1 As shown, this embodiment has 10 sub-ballast water tanks, which are symmetrically distributed to form a ballast water tank. The ballast water tank is rectangular with a volume of 10×10×2m. 3 Each ballast tank is equipped with four depth gauges. Additionally, two levels are installed at the bow, hull, and stern to detect the ship's heel. The ship is also equipped with six 200m depth gauges. 3 / h water pump. It should be noted that the ship ballast tanks targeted by this invention do not need to have a symmetrical structure, and there are no restrictions on the number, shape, or size of the ship's ballast tanks, as long as they can meet the requirements of full load balance.

[0045] Ballast water allocation optimization decision-making is a continuous and dynamic process. The allocation decision at each moment depends not only on the current state of the ballast water in each ballast tank but also on subsequent optimal allocation decisions. Therefore, based on the above system, this invention employs a deep reinforcement learning strategy, treating the ballast water allocation process as a partially observable Markov process, and utilizes deep reinforcement learning for control strategy self-learning, thereby providing a ballast water inflow and outflow control method, such as... Figure 2 As shown, it includes the following steps:

[0046] At time S1 and t, the roll and fore-and-aft angles of the ship are measured by n levels installed on the ship. The roll and fore-and-aft angles are the average values ​​of the corresponding level measurements.

[0047] by Figure 1 Taking the system shown as an example, at time t, the ship's roll and pitch angles are recorded by six levels installed on the hull. The measurement result of the i-th level is represented as s. i (t), where the heel and roll angles are the average of all measurements, i.e.:

[0048]

[0049] in, The summation symbol represents...

[0050] S2. Obtain the depth detection information measured by the depth gauge of each sub-water tank. The depth detection information of all sub-water tanks is represented by a vector.

[0051] by Figure 1 Taking the system shown as an example, let the depth detected by the j-th depth gauge of the i-th sub-ballast tank be h. ij (t), the water depth of the i-th ballast tank at time t is:

[0052]

[0053] The depth of all sub-ballast tanks is represented by the vector h(t) = [h1(t),…h2(t)]. 10 [(t)] represents.

[0054] S3. Repeat steps S1 and S2, and denote the measured values ​​S(t) = [s(t), s(t+1), s(t+2), h(t), h(t+1), h(t+2)] at multiple times as the state variables of the ballast water loading process at time t. Input the state variables into the influent and effluent neural network controller based on deep reinforcement learning to realize the control strategy output A(t) = F(S(t)), where A(t) represents the control output at time t and F(·) represents the neural network strategy function.

[0055] S4. The inlet and outlet neural network controller outputs the control strategy A(t) to the drive valve box, so that the drive valve drives the inlet valve, outlet valve, vent valve and high-pressure water pump connected to it respectively, so as to realize the automatic control of the inlet and outlet of the pressurized water tank.

[0056] S5. Return to steps S1 and S2, measure the ship's roll angle and depth detection information, and calculate the reward function value R1 in real time:

[0057]

[0058] Where, α M For the roll safety angle, 5° is used in this embodiment; β M The trim angle is set to 2° in this embodiment; |α(t)| is the absolute value of the trim angle of the hull at time t, and |β(t)| is the absolute value of the trim angle of the hull at time t.

[0059] At the same time, the parameters of the inflow and outflow neural network controller are updated based on the reward value.

[0060] S6. Based on the results measured in S5, if balance has not been achieved or the operation has not yet stopped, return to S3 to continue the ballast water balance loading operation; if balance has been achieved and the operation has ended, calculate the final reward function value R2 = (1 + T). -1 The cumulative reward is calculated, and the parameters of the water inlet and outlet neural network controller are updated based on the cumulative reward and the policy gradient.

[0061] The intake and drainage neural network controller uses the minimum total ballast time as its optimization objective. The total ballast time of the ballast system includes the time for transferring ballast water between ballast tanks, the response time of the tank valves, and the discharge time of the ballast water, i.e.:

[0062] T = t1 + t2 + t3

[0063] t1 is the time for transferring ballast water between ballast water tanks.

[0064]

[0065] Where, ρ=1.02g / cm 3 S is the density of seawater. i Let Δh be the bottom area of ​​the i-th sub-ballast tank. i (t) represents the water level change of the i-th sub-ballast tank at time t, N=6 represents the number of water pumps, and q p =200m 3 / h represents the pump flow rate, and η = 0.8 represents the pump efficiency.

[0066] t2 is the opening and closing response time of each valve.

[0067] t3 is the total time for loading and unloading ballast water.

[0068]

[0069] Where, Δh i This represents the total difference in water level change of the i-th sub-ballast tank before and after the start.

[0070] Therefore, the cumulative reward is:

[0071]

[0072] In the formula, γ is a discount factor of 0.99.

[0073] Finally, the inflow and outflow neural network controller π is updated based on the reward value. θ The parameter θ:

[0074]

[0075] In the formula, λ is the update step size, which is 0.0001.

[0076] This embodiment also provides a ballast water inlet and outlet control device, including a memory, a processor, and a program stored in the memory, wherein the processor executes the program to implement the method described above.

[0077] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for controlling the inflow and outflow of ballast water, the method being implemented based on a corresponding system, wherein the ballast water tank of the system is composed of multiple sub-tanks, characterized in that, The system includes: Each sub-tank is equipped with two electronic valves to control the water intake and drainage of the ballast water tank. The vent valve is installed on the top of the sub-water tank. The vent valve opens when the sub-water tank is being filled or drained, and closes when the sub-water tank is not in operation. Depth gauges, installed on the sidewalls of each sub-tank, are used to measure the water level in the ballast tanks; The drive valve box is installed in the control compartment and is connected to the water inlet valve, drain valve and vent valve via signal lines; The level is installed at the bow, midship and aft of the ship to detect the ship's roll and pitch angles. An inlet / outlet neural network controller is installed in the control compartment and is connected to the drive valve box, depth gauge, and level via signal lines. It acquires the measurement results from the depth gauge and level, calculates and judges the measurement results, and issues control commands to the drive valve box based on the judgment results. The drive valve box controls the high-pressure water pump, inlet / outlet valve, and vent valve to realize the inlet / outlet of the ballast water tank. The method includes the following steps: S1. Obtain the roll and trim angles of the ship as measured by the level. S2. Obtain the depth detection information measured by the depth gauge of each sub-water tank; S3. Repeat steps S1 and S2, record the measured values ​​at multiple times as the state variables of the ballast water loading process, and input the state variables into the inflow and outflow neural network controller based on deep reinforcement learning. Treat the ballast water loading process as a partially observable Markov process, and use deep reinforcement learning to perform self-learning to realize the control strategy output. S4. The inlet and outlet neural network controller outputs the control strategy to the drive valve box, so that the drive valve drives the inlet valve, outlet valve, vent valve and high pressure water pump connected to it respectively, so as to realize the automatic control of the inlet and outlet of the ballast water tank. S5. Return to steps S1 and S2, measure the hull roll angle and depth detection information, calculate the reward function value in real time, and update the parameters of the intake and drainage neural network controller according to the reward value; S6. Based on the results measured in S5, if balance has not been achieved or the operation has not stopped, return to S3 to continue the ballast water balance loading operation; if balance has been achieved and the operation has ended, calculate the final reward function value and the cumulative reward, and update the parameters of the influent and drainage neural network controller according to the cumulative reward and the strategy gradient.

2. The ballast water inlet and outlet control method according to claim 1, characterized in that, The inlet valve is located on the upper side of the sub-water tank and is used to control the opening and closing of the inlet of the sub-water tank; the drain valve is located on the lower side of the sub-water tank and is used to control the opening and closing of the drain outlet of the sub-water tank.

3. The ballast water inlet and outlet control method according to claim 1, characterized in that, The depth gauges are symmetrically mounted on the side walls of the sub-water tank.

4. The ballast water inlet and outlet control method according to claim 1, characterized in that, The side tilt angle and pitch angle of the hull are the average values ​​of the corresponding level measurements.

5. The ballast water inlet and outlet control method according to claim 1, characterized in that, The depth detection information of the sub-water tank is represented in vector form.

6. The ballast water inlet and outlet control method according to claim 1, characterized in that, In step S5, the reward function value is calculated as follows: in, For the roll safety angle, For the safety angle of pitch, for The absolute value of the ship's heel angle at any given moment. for The absolute value of the trim angle of the ship at any given moment.

7. The ballast water inlet and outlet control method according to claim 1, characterized in that, The inlet and outlet neural network controller uses the minimum total ballast time as the optimization objective. The total ballast time of the ballast system includes the time for transferring ballast water between ballast water tanks, the response time of the tank valves, and the discharge time of the ballast water.

8. A ballast water inlet and outlet control device, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-7.

9. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the method as described in any one of claims 1-7.

Images