Automatic sea kelp harvester and winch-cutter combined control method thereof
By combining the Mixer-MLP model with PID control, the rotational speeds of the winch and cutters are adjusted in real time, solving the winch and cutter control problem of the kelp harvester under complex sea conditions, and achieving efficient and precise kelp harvesting.
Patent Information
- Application Number
- CN202511057821.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-30
- Publication Date
- 2025-11-14
AI Technical Summary
When existing kelp harvesters operate at sea, the control of the winch and blades is difficult to cope with the effects of waves and currents, resulting in fluctuations in the tension of the kelp ropes and problems such as missed cuts or blade jamming.
By employing the Mixer-MLP model combined with classic PID control, parameters such as seawater flow velocity, flow direction, and rope traction force are collected in real time through sensors to predict the winch and cutter speeds and PID parameters in the next few seconds, thereby achieving adaptive control.
It improves the computational efficiency and control precision of kelp harvesting, reduces missed cuts and jamming, enhances operational quality and consistency, and improves stability in complex sea conditions.
Smart Images
Figure CN120949538A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of aquatic product processing, and in particular to an automatic kelp harvester for marine use and its winch-cutter combined control method. Background Technology
[0002] Currently, the main methods for harvesting kelp are: 1. Manual harvesting. Harvesters sit in small boats and manually pull mature kelp ropes from floating rafts, cutting the kelp from the ropes or loading the entire rope onto the boat to return to shore for harvesting. 2. Harvesting by directly trawl the base of the kelp plant. 3. Machine-assisted harvesting. For example, a boom crane, using a hydraulically controlled robotic arm to drive a toothed claw with a gripping device, lifts the ropes with kelp attached section by section, grabs the kelp, and transfers it to the deck. Alternatively, a spiral harvester, using a rotating metal spiral structure, wraps and cuts kelp leaves or the entire plant, and is widely used in shallow-sea aquaculture.
[0003] Of the methods mentioned above: manual harvesting is inefficient and labor-intensive, but causes less damage to the harvested kelp and less disruption to the marine ecosystem. Trawling is only suitable for large kelp in the deep sea; it is efficient but may impact the seabed ecosystem and is unsuitable for my country's kelp raft aquaculture model. Machine-assisted harvesting is one of the fastest-growing and most promising kelp harvesting methods. It combines efficiency with a degree of selectivity and is suitable for my country's mainstream raft aquaculture model. Specifically, the rotary boom crane reduces manual labor intensity to some extent and better preserves the integrity of the kelp plants. The spiral harvester achieves high-efficiency harvesting, but causes greater damage to the kelp plants.
[0004] The aforementioned mechanical solutions all face a common challenge when applied to kelp rope raft aquaculture: the harvester, acting as a floating platform, experiences six-degree-of-freedom oscillation due to hydrological conditions such as waves and tides. In this situation, using a constant speed to control the winch and cutter head causes tension fluctuations in the kelp rope, preventing it from maintaining tautness. This leads to missed cuts or cutter jams during the cutting process, hindering the effective removal of kelp from the rope. Furthermore, the kelp rope winch cannot be driven and controlled using constant torque or constant speed methods. Summary of the Invention
[0005] In view of the above-mentioned deficiencies of the prior art, the purpose of this invention is to provide an automatic kelp harvester for marine use and a winch-cutter joint control method thereof, so as to solve the deficiencies existing in the prior art.
[0006] To achieve the above objectives, the present invention provides a winch-cutter combined control method for an automatic kelp harvester at sea, used to control the winch of the kelp harvester to wind up and collect kelp ropes, and to control the rotating cutter to rotate and cut the kelp off the kelp ropes; the method includes:
[0007] Input parameters are collected synchronously by sensors at a predetermined sampling frequency to form an input parameter sequence; the input parameters include seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, rotary cutter speed, PID control parameters of the winch drive motor, and PID control parameters of the rotary cutter drive motor.
[0008] At predetermined intervals, an input sequence is extracted from the input parameter sequence, starting from the current time and proceeding backward for a first time length. This input sequence is then input into a trained Mixer-MLP model to obtain a control parameter sequence for a second time length following the current time. The control parameter sequence includes several sets of control parameters, each set including the winch target speed, the rotary cutter target speed, the PID control parameters of the winch drive motor, and the PID control parameters of the rotary cutter drive motor. The PID controllers of the winch drive motor and the rotary cutter drive motor are tuned according to the control parameter sequence to control the speed of the winch and the rotary cutter.
[0009] A further improvement of the present invention is that the predetermined sampling frequency is 20Hz, the first time length is 60s, and the second time length is 5s.
[0010] A further improvement of the present invention is that: the traction force of the kelp rope is detected by a tension sensor on the drive shaft of the winch; the rotational speed of the winch is detected by a Hall effect sensor on the drive shaft of the winch; the rotational speed of the rotating cutter is detected by a Hall effect sensor on the drive shaft of the rotating cutter; and the sensors for measuring seawater flow velocity and direction include a biaxial / triaxial ultrasonic velocimeter and an electromagnetic velocimeter.
[0011] A further improvement of this invention is that, during the process of acquiring the dataset for training the Mixer-MLP model, the seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, and rotary cutter speed are measured by sensors; the kelp rope is kept taut by manually adjusting the PID control parameters of the winch drive motor and the rotary cutter drive motor; each sensor measurement value and each PID control parameter is timestamped, and the time series obtained after aligning the timestamps is used as the dataset.
[0012] A further improvement of the present invention is that the PID controllers of the rotary tool drive motor and the winch drive motor adjust the drive circuit and drive voltage of the corresponding motors according to the detected rotational speed and the target rotational speed, thereby controlling the corresponding motors to rotate at a predetermined rotational speed.
[0013] The present invention also provides an automatic kelp harvester for marine use, including a floating platform, on which a winch drive motor and a rotary cutter drive motor are mounted; the winch is driven by the winch drive motor; and the rotary cutter is driven by the rotary cutter drive motor.
[0014] The winch drive motor and the rotary cutter drive motor are each controlled by a corresponding PID controller; the PID controller adjusts the drive current and drive voltage according to the target speed and the actual speed.
[0015] The controller is communicatively connected to the sensor assembly and each PID controller; the controller is configured to implement the above method.
[0016] The beneficial effects of this invention include:
[0017] I. High computational efficiency: Compared to Transformer-based temporal modeling methods, the Mixer-MLP architecture is more lightweight, does not rely on complex attention mechanisms, and has lower computational complexity, making it suitable for deployment in embedded devices or edge computing units. Its fully connected structure supports batch parallel computing, resulting in fast inference speed and low resource consumption, making it suitable for offshore operation systems with high real-time requirements.
[0018] 2. High control precision: By integrating multi-source data such as seawater flow velocity, flow direction, and traction force, Mixer-MLP can accurately capture the dynamic change trend of the system and predict the target winch and cutter speed and PID parameters within the next few seconds. This makes the winch and cutter run closer to the ideal state, significantly reducing problems such as slippage, rope jamming, and incomplete cutting during the harvesting process, and improving the overall operation quality and consistency.
[0019] Third, high stability: This method achieves adaptive control through parameter tuning assisted by deep neural networks, effectively mitigating the impact of environmental uncertainties. Meanwhile, PID control itself possesses the advantages of disturbance rejection robustness and ease of implementation, enabling the entire control system to maintain good stability and reliability even under varying sea conditions. Attached Figure Description
[0020] Figure 1 This is a schematic diagram of an automated seaweed harvester at sea.
[0021] Figure 2 This is a schematic diagram of the Mixer-MLP model;
[0022] Figure 3 The flowchart shows the winch-tool joint control method.
[0023] Figure 4 Examples of measured data;
[0024] Figure 5 The graph shows the predicted results of the winch speed, the rotary cutter speed, and its PID parameters. Detailed Implementation
[0025] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.
[0026] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.
[0027] Some exemplary embodiments of the invention have been described for illustrative purposes. It should be understood that the invention may be implemented in other ways not specifically shown in the accompanying drawings.
[0028] like Figure 1 As shown, an embodiment of the present invention provides an automatic kelp harvester for marine use, including a floating platform 10. A winch drive motor and a rotary cutter drive motor are installed on the floating platform 10. The winch 20 is driven by the winch drive motor. The rotary cutter 30 is driven by the rotary cutter drive motor. The winch 20 is used to wind and pull the kelp rope 40, gradually pulling the kelp rope 40 onto the floating platform 10. The rotary cutter 30 is disc-shaped and can cut the kelp on the kelp rope 40 when rotating.
[0029] The winch drive motor and the rotary cutter drive motor are each controlled by a corresponding PID controller; the PID controller adjusts the drive current and drive voltage according to the target speed and the actual speed.
[0030] The controller is communicatively connected to the sensor assembly and each PID controller; the controller is configured to implement the winch-tool joint control method of the present invention.
[0031] Specifically, the construction process of the Mixer-MLP model used in the winch-tool joint control method in this embodiment includes the following steps:
[0032] Data Acquisition: The ocean current velocity and direction (two-dimensional horizontal component) are measured using a seawater velocity sensor (either a dual-axis / triaxial ultrasonic velocimeter or an electromagnetic velocimeter) on an automated kelp harvester; sensors (tension sensors and rotary encoders or Hall effect sensors) on the kelp rope winch measure the traction force of the kelp rope and the winch rotation speed; sensors (rotary encoders or Hall effect sensors) on the rotating cutter measure the rotation speed of the rotating cutter. All sensor data are synchronized with the same timestamp, and the sampling frequency is set to 20Hz per second. The data is recorded, with the seawater velocity v = [v1, v2, ..., v k The seawater flows in the direction d = [d1, d2, ..., d]. k The traction force of the kelp rope is F = [F1, F2, ..., F]. k winch speed Tool speed
[0033] PID controller parameter tuning: Two independent PID controllers are used to control the winch speed and cutter speed respectively. For the winch speed, the motor voltage / current is adjusted in real time based on the target winch speed and the current feedback error to achieve smooth adjustment of the winch speed. For the cutter speed, the motor voltage / current is adjusted in real time based on the target cutter speed and the current feedback error to achieve stable cutter speed, improving cutting efficiency and safety. The PID controller parameters are adjusted and tuned by the operator according to the on-site environment. Finally, the corresponding PID controller parameters are recorded. The timestamps are then aligned with the data detected by the sensors. The goal of tuning the PID controller is to keep the kelp ropes taut.
[0034] In addition, in some embodiments, during the actual deployment of the trained Mixer-MLP model, the actual measurement values of each sensor and the PID controller parameters of each PID controller can be collected. Sensor measurement values and corresponding PID controller parameters with the kelp rope traction force in the preferred range (300N to 500N) can be selected and added to the supplementary dataset for iterative training of the Mixer-MLP model to obtain a better Mixer-MLP model.
[0035] Constructing and training a hybrid deep neural network (Mixer-MLP) model: 1. Establish a dataset based on collected sensor data and PID controller parameters (dividing it into training and testing sets). 2. Train the model using the collected dataset. Input historical multivariate sequence data within a continuous time window (e.g., the first 60 seconds) into a time-series modeling network composed of a Mixer structure and MLP layers. Input features include: seawater flow velocity and direction, rope traction force, winch speed, cutter speed, and corresponding PID parameter values. Output features are: the target speed of the winch and the target speed of the cutter within a future continuous time window (e.g., 5 seconds), and the corresponding PID controller parameter values. 3. After training, the model can accurately predict and output the required speeds of the winch and cutter within a future continuous time window, along with the corresponding PID controller parameters. This enables the control of the winch and cutter of an automated kelp harvester at sea.
[0036] The aforementioned control method proposes a control approach combining Mixer-MLP time-series modeling and classical PID control for the winch and rotating cutter of an automated kelp harvester at sea. This approach enables precise control of the winch and cutter, improving control accuracy, stability, and adaptability to complex sea conditions. The method utilizes a Mixer-MLP deep learning model to model and predict seawater flow velocity and direction, kelp rope traction force, winch rotation speed, and rotating cutter rotation speed across the entire time domain, thereby enhancing control accuracy. Combined with classical PID control, this method leverages the direct output of the control target values for the winch rotation speed and rotating cutter, along with relevant PID parameters, from the Mixer-MLP model. This achieves adaptive PID tuning, improving adaptability and stability to complex sea conditions.
[0037] The following uses specific data and experiments to verify the prediction results of the present invention for the control targets and corresponding PID parameters of winch speed and rotating cutter. 1. A training dataset is formed by using sensors on an automated kelp harvester to collect data on seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, and rotating cutter speed. 2. A corresponding Mixer-MLP deep learning model is established. The input is historical time-series data of seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, and rotating cutter speed. The output is the target winch speed and its corresponding PID parameters, and the rotating cutter speed and its corresponding PID parameters. The Mixer-MLP deep learning model is trained using the training dataset. 3. The trained Mixer-MLP deep learning model is used to control the winch speed and rotating cutter speed.
[0038] Please see Figure 4 , Figure 4This is a partial sample of measured data, including historical time series of seawater flow velocity, seawater flow direction, traction force of kelp ropes, winch speed, and rotary cutter speed (only measurement data from 5:00 to 18:00 daily from May 21, 2025 to May 27, 2025 are selected).
[0039] In this embodiment, using Figure 4 The data shown is used to train the Mixer-MLP model (the first 80% of the recorded data is used as the training set, and the last 20% as the test set) to obtain the winch speed, the rotary tool speed, and their corresponding PID parameters for the next moment, thereby achieving automatic control of the winch and rotary tool. The effect is as follows: Figure 5 As shown in the figure, the relative error of the winch speed is 5.16%, with corresponding PID parameter errors of 5.16%, 5.15%, and 5.15%; the error of the rotating cutter is 5.16%, with corresponding PID parameter errors of 5.17%, 5.17%, and 5.13%. It can be seen that a good result was achieved.
[0040] Figure 2 The diagram shows the Mixer-MLP model used in this embodiment; its main parameters are: inputs are seawater flow velocity v, seawater flow direction d, kelp rope traction force F, and winch rotation speed w. s Rotary cutting tool speed w * The outputs are the winch speed control target, the rotary cutter control target, the PID values for controlling the winch speed, and the PID values for controlling the rotary cutter speed; the number of MixerLayer layers is 1; the Embedding dimension is set to 128; the input time series length is 60; the token mixing size is 60; and the channel mixing size is 60. After training with historical data, the corresponding Mixer-MLP model is obtained.
[0041] Figure 3 The diagram shows the specific control steps for the winch and rotary cutter using a trained Mixer-MLP deep model.
[0042] Step S101: The sensor equipment on the automatic kelp harvester at sea collects the seawater flow rate, flow direction, kelp rope traction force, winch speed and rotating cutter speed, and forms the corresponding time series.
[0043] Step S102: Input the time-series data collected in step S101 into the trained Mixer-MLP model. In actual deployment, the input data format of the Mixer-MLP model is consistent with the input data format during training.
[0044] Step S103: The Mixer-MLP deep model is used to predict the winch speed, the rotary tool speed control target, and the corresponding PID parameters for the next moment.
[0045] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.
Claims
1. A winch-cutter combined control method for an automatic kelp harvester at sea, used to control the winch of the kelp harvester to wind and collect kelp ropes, and to control the rotating cutter to rotate and cut the kelp off the kelp ropes; characterized in that, The method includes: Input parameters are collected synchronously by sensors at a predetermined sampling frequency to form an input parameter sequence; the input parameters include seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, rotary cutter speed, PID control parameters of the winch drive motor, and PID control parameters of the rotary cutter drive motor. At predetermined intervals, an input sequence is extracted from the input parameter sequence, trunculated backward from the current time for a first time length. This input sequence is then input into a trained Mixer-MLP model to obtain a control parameter sequence for a second time length following the current time. The control parameter sequence includes several sets of control parameters, each set including the winch target speed, the rotary cutter target speed, the PID control parameters of the winch drive motor, and the PID control parameters of the rotary cutter drive motor. The PID controllers of the winch drive motor and the rotary cutter drive motor are tuned according to the control parameter sequence to control the speed of the winch and the rotary cutter.
2. The winch-cutter combined control method for an automatic kelp harvester at sea according to claim 1, characterized in that: The predetermined sampling frequency is 20Hz, the first time length is 60s, and the second time length is 5s.
3. The winch-cutter combined control method for an automatic kelp harvester at sea according to claim 1, characterized in that: The traction force of the kelp rope is detected by a tension sensor on the drive shaft of the winch; the speed of the winch is detected by a Hall effect sensor on the drive shaft of the winch; the speed of the rotating cutter is detected by a Hall effect sensor on the drive shaft of the rotating cutter; the sensors for measuring seawater flow velocity and direction include biaxial / triaxial ultrasonic velocimeters and electromagnetic velocimeters.
4. The winch-cutter combined control method for an automatic kelp harvester at sea according to claim 1, characterized in that: In the process of acquiring the dataset for training the Mixer-MLP model, sensors were used to measure seawater flow velocity, seawater flow direction, kelp rope traction force, winch speed, and rotary cutter speed. The PID control parameters of the winch drive motor and the rotary cutter drive motor were manually adjusted to keep the kelp rope taut. Each sensor measurement and each PID control parameter was timestamped, and the time series obtained by aligning the timestamps was used as the dataset.
5. The winch-cutter combined control method for an automatic kelp harvester at sea according to claim 1, characterized in that: The PID controllers of the rotary tool drive motor and the winch drive motor adjust the drive circuit and drive voltage of the corresponding motor according to the detected rotational speed and the target rotational speed, thereby controlling the corresponding motor to rotate at the predetermined speed.
6. An automated kelp harvester for marine use, comprising a floating platform, characterized in that: The floating platform is equipped with a winch drive motor and a rotary cutter drive motor; the winch is driven by the winch drive motor; the rotary cutter is driven by the rotary cutter drive motor. The winch drive motor and the rotary cutter drive motor are each controlled by a corresponding PID controller; Each of the PID controllers adjusts the drive current and drive voltage according to the target speed and the actual speed; The controller is communicatively connected to the sensor assembly and each PID controller; the controller is configured to implement the method of claims 1 to 5.