Intelligent fish feed dispensing system, control method for feeding machines, and computer-readable storage device

Abstract

An intelligent fish feed dispensing system includes a sensing module, a monitoring module, a processing module, a control module, and a feeding machine. The sensing module is installed in the fish pond and is used to detect at least one environmental parameter of the fish pond. The monitoring module captures real-time images of fish activities in the fish pond, generating real-time image data. The processing module is coupled to both the sensing module and the monitoring module to analyze the obtained environmental parameter data and real-time image data, generating a predicted feed demand and control commands. The control module receives the predicted feed demand and control commands from the processing module and controls the feeding machine to execute feeding operations based on the predicted feed demand. After a preset time has elapsed from the starting time, the control module determines, based on the control commands, whether to continue or stop dispensing fish feed.

Description

Technical Field

[0001] This disclosure relates to a smart aquaculture technology, and in particular to a fish feed dispensing system based on artificial intelligence and its control method. Prior Technology

[0002] Traditional fish feeding methods are primarily manual. While this method offers some flexibility, it relies heavily on the operator's experience and is prone to overfeeding or underfeeding. Overfeeding results in uneaten food accumulating at the bottom of the pond, leading to waste and increased cleaning and maintenance needs. Conversely, underfeeding can negatively impact fish growth and health, reducing farming efficiency. Furthermore, the frequency and accuracy of manual feeding are highly dependent on human labor, making it susceptible to errors due to fatigue or negligence, and significantly increasing labor costs. This method presents challenges to efficiency and economic viability, especially in large-scale aquaculture.

[0003] To address the shortcomings of manual feeding, some automated feeding devices have been applied in aquaculture. These devices typically operate based on timed and quantitative settings. However, this method can only operate according to preset times and amounts of feed, lacking the ability to adjust feeding strategies based on the fish's condition, resulting in persistent problems of overfeeding or underfeeding. Furthermore, the fixed feeding patterns of automated devices struggle to adapt to changes in the fish's feed requirements at different growth stages, leading to a lack of flexibility and precision in aquaculture management.

[0004] Therefore, how to accurately calculate the amount of feed to be given and dynamically adjust the amount of feed to be given according to the actual situation has become an important issue in feeding technology. Summary of the Invention

[0005] In view of this, this disclosure provides a smart fish feed dispensing system and a control method for the feeder, which can accurately estimate the amount of fish feed to be dispensed and dynamically adjust the amount of fish feed dispensed according to the actual situation.

[0006] The intelligent fish feeding system disclosed herein includes a sensing module, a monitoring module, a processing module, a control module, and a feeder. The sensing module is installed in the fishpond to detect at least one environmental parameter. The monitoring module captures images of the fish in the pond and generates image data. The processing module is coupled to both the sensing and monitoring modules to generate predicted demand and control commands based on the obtained environmental parameter data. The control module receives the predicted demand and control commands from the processing module and controls the feeder to perform the feeding operation based on the predicted demand.

[0007] In some embodiments, the control module determines whether to continue or stop feeding fish based on control commands after a preset time has elapsed from the start time.

[0008] In some embodiments, the aforementioned at least one environmental parameter includes water temperature, fish weight and length, fish population density, and at least one or a combination of two or more of the pH and dissolved oxygen levels of the water in the fishpond.

[0009] This disclosure also provides a method for controlling a fish feeder, including detecting at least one environmental parameter in a fishpond, capturing images of a fish school, and generating a predicted demand and control command based on the environmental parameter data at a start time. At the start time, the feeder is controlled to dispense fish feed according to the predicted demand. Then, after a first preset time has elapsed from the start time, a control command is generated based on the image data to determine whether to continue or stop feeding the fish.

[0010] In some embodiments, after a first preset time has elapsed from the start time, the control method disclosed herein determines whether the amount of fish feed on the surface of the fishpond exceeds a first threshold. If it does, a control command is issued to stop feeding. Conversely, if it is below the threshold, a second preset time is then elapsed, and a decision is made on whether to continue feeding or stop feeding based on whether the amount of fish feed is below the second threshold.

[0011] This disclosure further provides a computer-readable storage device that stores at least one piece of code to control the operation of a feeding machine. Once the code is read, an artificial intelligence engine can be activated to generate predicted demand and control commands based on environmental parameters and images of the fishpond. Additionally, at the start time, the feeding machine is controlled to dispense fish feed according to the predicted demand. Then, after a preset time has elapsed from the start time, the artificial intelligence engine determines whether to continue or stop feeding based on the amount of fish feed, further ensuring the accuracy and efficiency of the feeding.

[0012] Since the intelligent fish feed dispensing system and control method disclosed herein can calculate the amount of feed to be dispensed based on environmental parameters and dynamically control the dispensing of fish feed according to image data, it can not only avoid feed waste and improve the aquaculture environment, but also achieve the purpose of intelligent aquaculture and reduce labor operating costs. Simple Explanation of the Diagram

[0013] To make the above and other objects, features, advantages and embodiments of this disclosure more apparent and understandable, the accompanying drawings are described below: Figure 1 is a functional block diagram of a smart fish feed dispensing system according to a preferred embodiment of the present disclosure. Figure 2 is a functional block diagram of a processing module according to a preferred embodiment of the present disclosure. Figure 3 illustrates a flowchart of the steps of a control method for a feeder according to a preferred embodiment of the present disclosure. Figure 4 is a functional block diagram of a control module according to a preferred embodiment of the present disclosure. Figure 5 illustrates a detailed flowchart of step S310 of Figure 3 according to a preferred embodiment of the present disclosure. Figure 6 is a functional block diagram of a smart fish feed dispensing system according to a second embodiment of the present disclosure. Implementation

[0014] The following drawings disclose multiple embodiments of this disclosure. For clarity, many practical details will be described in the following description. However, it should be understood that these practical details should not be used to limit this disclosure. That is, in some embodiments of this disclosure, these practical details are not essential. Furthermore, for the sake of simplicity, some conventional structures and elements will be shown in a simple schematic manner in the drawings, and in all drawings, the same reference numerals will be used to denote the same or similar elements. And, where feasible, features of different embodiments can be applied interchangeably.

[0015] Unless otherwise defined, all terms used herein (including technical and scientific terms) have their ordinary meanings, meanings that are understandable to those skilled in the art. Furthermore, the definitions of the aforementioned terms in commonly used dictionaries should be interpreted in the context of this specification as having the same meaning as those applicable to the field of this disclosure. Unless specifically defined, these terms should not be interpreted as having idealized or overly formal meanings. The term "comprising" as used in the specification and claims is an open-ended term and should therefore be interpreted as "comprising but not limited to".

[0016] In addition, the term "coupled" or "connected" as used in this article may refer to two or more components making direct physical or electrical contact with each other, or making indirect physical or electrical contact with each other, or may refer to two or more components operating or acting on each other.

[0017] In this article, the term "module" is used to refer to an object consisting of one or more transistors and / or one or more active and passive components connected in a certain way to process signals.

[0018] Figure 1 is a functional block diagram of a smart fish feed dispensing system according to a preferred embodiment of the present disclosure. Referring to Figure 1, the smart fish feed dispensing system 10 provided in this disclosure is suitable for controlling a feeder 20 to feed fish to a fishpond. In addition, the smart fish feed dispensing system 10 also includes a processing module 12, a sensing module 14, a monitoring module 16, a control module 18, and a feeder 20. The processing module 12 is coupled to the sensing module 14, the monitoring module 16, and the control module 18, respectively, while the control module 18 is also coupled to the feeder 20.

[0019] Figure 2 illustrates a functional block diagram of a processing module according to a preferred embodiment of the present disclosure. Referring to Figure 2, the processing module 12 includes a processing unit 202 and an artificial intelligence engine 204. The processing unit 202 may be a central processing unit, a graphics processing unit, or other processor. The artificial intelligence engine 204 may be a deep learning network model. Table 1 below lists the deep learning network models tested. Model Name Score WeightedEnsemble_L2 -0.18583 ExtraTreesMSE -0.188854 LightGBMLarge -0.189062 LightGBMXT -0.190615 LightGBM -0.192586 CatBoost -0.194555 KNeighborsUnif -0.194934 RandomForestMSE -0.198267 KNeighborsDist -0.200184 NeuralNetTorch -0.206107 NeuralNetFastAI -0.212266 XGBoost -0.235968 Table 1

[0020] In Table 1, the scores represent the model's evaluation score on the validation set. Here, a smaller score is better. Additionally, in Table 1, negative scores represent evaluation metrics for error types, such as mean squared error. Table 1 shows that WeightedEnsemble_L2 performs better and is therefore selected as the artificial intelligence engine 204 in this embodiment. However, those skilled in the art may choose other models as the artificial intelligence engine 204 without affecting the main spirit of this disclosure. In some embodiments, the artificial intelligence engine 204 can be stored in a computer-readable storage device, such as a hard drive.

[0021] Referring to Figure 1, the sensing module 14 also includes at least one sensing unit for sensing at least one environmental parameter of the fishpond and sending it to the processing module 12 for analysis. In practice, various environmental parameters of the fishpond, such as water temperature, fish weight and length, fish density, and the pH and dissolved oxygen levels of the water, all affect the feeding speed of the fish. Table 2 below shows the characteristic analysis of various environmental parameters of the fishpond. feature importance Standard deviation p-value High confidence interval low confidence interval Temperature 0.118054 0.016223 0.000042 0.151457 0.08465 Fish weight 0.116053 0.021093 0.000125 0.159484 0.072623 pH value 0.027907 0.003013 0.000016 0.034112 0.021703 Dissolved oxygen (ox) 0.026464 0.006477 0.000398 0.039801 0.013127 The fish's body length (Height) 0.025393 0.005184 0.000197 0.036067 0.014718 Cluster density (group) 0.018235 0.002346 0.000032 0.023066 0.013404 Table 2

[0022] The values ​​for each column in Table 2 are described below: The "Importance" column displays the importance score for each environmental parameter. A higher score indicates a greater impact of that parameter on the model's (i.e., the aforementioned AI engine 204) prediction results. Table 2 shows that the environmental parameter with the greatest influence on the model's prediction results is the temperature of the fishpond. The "Standard Deviation" column displays the standard deviation of the importance estimate, reflecting the stability of the importance value. A smaller standard deviation indicates a more reliable importance value. Table 2 also shows that the temperature parameter has the smallest standard deviation. The “P-value” field indicates the significance test result of the variable’s importance value. The smaller the value, the higher the statistical significance of the variable. The "High Confidence Interval" and "Low Confidence Interval" fields represent the 99% confidence interval range for the importance of the variable, with an upper bound (p99_high) and a lower bound (p99_low). The narrower the range, the more stable the estimate of the variable's importance.

[0023] This disclosure utilizes environmental parameters such as water temperature, fish weight and length, fish population density, and the pH and dissolved oxygen levels of the water in the fishpond to train the artificial intelligence engine 204. As can be seen from the analysis in Table 2, the temperature parameter of the fishpond has a significant impact on the feeding speed of the fish. Therefore, in some embodiments, the artificial intelligence engine 204 can adjust the weighting of parameters based on actual conditions, prioritizing certain environmental parameters; however, this disclosure is not limited to this.

[0024] Please refer to Figure 1. The monitoring module 16 includes at least one camera unit, such as a camera, which can capture real-time images of the fishpond and transmit them to the processing module 12, for example, in the form of a streaming or network packet.

[0025] Figure 3 illustrates a flowchart of the control method for a feeding machine according to a preferred embodiment of this disclosure. Referring to Figures 1 and 3 together, in this embodiment, the sensing module 14, as described in step S302, senses at least one environmental parameter in the fishpond, such as the water temperature, fish weight and length, fish population density, and at least one or a combination of the pH and dissolved oxygen levels of the water in the fishpond. Furthermore, the monitoring module 16 can also capture images of the fishpond, as described in step S304. The sensed environmental parameter is then sent to the processing module 12 for analysis. At this time, the artificial intelligence model in the processing module 12, such as the artificial intelligence engine 204 in Figure 2, is activated, as described in step S306.

[0026] Once the artificial intelligence model is activated, the processing module 12 can perform step S308, which involves generating a predicted demand and a control command for the control module 18 based on at least one transmitted environmental parameter and image data at the start time. As shown in Figure 4, the control module 18 has a control unit 402, such as a single-chip microcomputer, embedded system, control circuit, etc., and a control relay 404. In some embodiments, the Raspberry Pi computer is used as the hardware to implement the control unit 402 because of its small size, light weight, full functionality, and programmability; however, this disclosure is not limited thereto. In other embodiments, the control unit 402 can also be implemented using software.

[0027] When the control unit 402 receives the predicted demand and control command from the processing module 12, it can control the feeder 20 to deliver a fixed quantity, such as 100 fish food pellets, to the fishpond according to the predicted demand generated by the processing module 12, as described in step S308. Next, the processing module 12 times the feeder for a first preset time, such as 2 minutes, and then proceeds to step S310, which generates a new control command based on the image data to determine whether the feeder 20 should continue feeding the fish.

[0028] Figure 5 illustrates a detailed flowchart of step S310 according to a preferred embodiment of this disclosure. Referring to Figures 1, 4, and 5 together, after a first preset time has elapsed from the start time, the processing module 12 can activate the image recognition function and execute step S502, which determines whether the amount of fish feed on the surface of the fishpond exceeds a first threshold based on the image data generated by the monitoring module 16. If step S502 finds that after a first preset time has elapsed from the start time, there is too much fish feed in the fishpond, for example, more than 35 pellets (i.e., the first threshold), it means that too much fish feed has been fed. Therefore, step S504 is performed, whereby the control unit 402 controls the control relay 404 to stop the feeder 20 from feeding the fish.

[0029] Conversely, in step S502, if the processing module 12 determines from the image that the amount of fish food on the water surface does not exceed a first threshold (e.g., is less than the first threshold), then step S506 can be performed to continue timing for a second preset time, such as 1 minute. After the first and second preset times have elapsed from the start time, for example, after 3 minutes, the processing module 12 can determine from the image data whether the amount of fish food on the surface of the fishpond is lower than a second threshold, as described in step S508. If step S508 finds that the amount of fish food on the water surface is not lower than the second threshold (e.g., is greater than 5 pieces), it means that the fish are almost full, and therefore the rate at which they eat the food slows down. At this time, the processing module 12 can issue a new control command to the control module 18. The control unit 402 can then, according to the new control command, control the relay 404 to stop the feeding machine 20 from feeding the fish, as described in step S504.

[0030] Conversely, if in step S504 it is found that the amount of fish feed on the water surface has fallen below the second critical value, it means the fish are still very hungry, so the fish feed is being consumed quickly and needs to be fed again. At this time, the processing module 12 can proceed to step S510, instructing the control module 18 to control the feeder 20 to continue feeding the fish, and to repeat step S506. Of course, in other embodiments, after completing step S510, the amount of fish feed on the water surface can be checked every third preset time, such as 30 seconds, until the goal of feeding the fish is achieved.

[0031] As can be seen from the flowcharts in Figures 3 and 5 above, the intelligent fish feed feeding system 10 disclosed herein can accurately control the amount of fish feed fed by the feeder 20 to achieve the feeding target. It will not cause waste or damage to water quality due to overfeeding, nor will it fail to achieve the goal of keeping the fish population healthy due to underfeeding.

[0032] In some embodiments, the processes disclosed in Figures 3 and 5 above can be written as program code into a computer-readable storage device, such as a memory. When the processing unit 202 in Figure 4 reads this program code, it can execute the above-described steps.

[0033] Figure 6 illustrates a functional block diagram of a smart fish feed dispensing system according to a second embodiment of this disclosure. Referring to Figure 6, the smart fish feed dispensing system 10' provided in this embodiment, compared to the smart fish feed dispensing system 10, is further equipped with a database module 22, which is coupled to the processing module 12. The database module 22 is, for example, a database constructed using MySQL, used to store image data and environmental parameter data of the fish feeding in the fishpond each time. When the processing module 12 performs the process illustrated in Figure 5, it can also write all the process data into this database module 22. In this way, the artificial intelligence engine 204 in the processing module 12 can use the image data, at least one environmental parameter, and historical process data in the database module 22 for training to adjust the predicted demand for the next time.

[0034] In summary, the intelligent fish feeding system disclosed in this paper integrates an artificial intelligence engine, a sensing module, a monitoring module, and a database module to form a precise and efficient fish feed management solution. Based on environmental parameters (such as water temperature, dissolved oxygen, and pH) and image data, the system uses an artificial intelligence model for analysis and prediction to dynamically adjust the feeding amount. Especially under different environmental conditions, the system can intelligently determine the feed requirements of the fish population, effectively avoiding water pollution and feed waste caused by overfeeding, while ensuring the healthy growth of the fish and improving the economic benefits of the fish farm.

[0035] Furthermore, the control method disclosed herein can detect feeding results at different time intervals using image data and preset parameters (such as critical values ​​for fish feed), and dynamically adjust the feed delivery strategy based on the feeding status of the fish population. This not only improves the accuracy of feeding but also flexibly responds to the needs of the fish population, avoiding underfeeding or overfeeding, ensuring the automation of aquaculture farms, and further reducing labor costs, thus achieving commercial success.

[0036] The various components, method steps, or technical features in the foregoing embodiments can be combined with each other, and are not limited to the order of textual description or graphical presentation in this disclosure.

[0037] Although the present disclosure has been described above with reference to embodiments, it is not intended to limit the present disclosure. Anyone skilled in the art may make various modifications and alterations without departing from the spirit and scope of the present disclosure. Therefore, the scope of protection of the present disclosure shall be determined by the claims defined in the appended patent application.

[0038] 10, 10': Intelligent fish feed dispensing system 12: Processing Module 14: Sensing Module 16: Monitoring Module 18: Control Module 20: Feeder 22: Database Module 202: Processing Unit 204: Artificial Intelligence Engine 402: Control Unit 404: Control relay S302, S304, S306, S308, S310, S502, S504, S506, S508, S510: Steps

Claims

1. A smart fish feeding system, comprising: A sensing module (14) is configured in a fishpond to detect at least one environmental parameter data of the fishpond; a monitoring module (16) is configured in the fishpond to capture images of the fish in the fishpond and generate image data; a processing module (12) is coupled to the sensing module (14) and the monitoring module (16) to obtain the at least one environmental parameter data and execute an artificial intelligence engine (204) to analyze the situation of the fish in the fishpond based on the at least one environmental parameter data, and generate a predicted demand and a control command; a control module (18) is coupled to the processing module (12) to receive the predicted demand and the control command; and a feeder (20) is configured in the fishpond and coupled to the control module (18), wherein at the beginning of the fishpond, the control module (18) controls the feeder (20) to feed the fish according to the predicted demand. After a preset time has elapsed from the start time, the control module (18) controls the feeder (20) to continue feeding the fish feed or stop feeding the fish feed according to the control command.

2. The intelligent fish feed feeding system as described in claim 1, wherein the control module (18) controls the feeder (20) to continue feeding the fish feed or stop feeding the fish feed according to the control command after a preset time.

3. The intelligent fish feed dispensing system as described in claim 1 further includes: A database module (22) is coupled to the processing module (12) and stores image data of the fish feeding in the fishpond each time and the at least one environmental parameter data; wherein the processing module (12) accesses the image data and the at least one environmental parameter data in the database module (22) to train the artificial intelligence engine (204).

4. The intelligent fish feed system as described in claim 3, wherein the database module (22) is used to record all processes processed by the processing module (12) for use in training the artificial intelligence engine (204).

5. The intelligent fish feeding system as described in claim 1, wherein the at least one environmental parameter data includes water temperature, fish weight and length, fish population density, and at least one or a combination of two or more of the pH and dissolved oxygen levels of the water in the fishpond.

6. A method for controlling a feeder, wherein the feeder is used to feed fish food to a fishpond, the control method comprising at least the following steps: detecting at least one environmental parameter data of the fishpond; capturing an image of a group of fish in the fishpond to generate an image data; activating an artificial intelligence engine (204) and, at a start time, generating a predicted demand and a control command based on the at least one environmental parameter data to control the feeder (20) to feed fish food according to the predicted demand; and, after a first preset time has elapsed from the start time, generating the control command based on the image data to determine whether the feeder should continue feeding fish food.

7. The control method for the feeder as described in claim 6, wherein after the first preset time has elapsed from the start time, the artificial intelligence engine (204) performs the following steps: determining, based on the image data, whether the amount of fish feed on the surface of the fishpond exceeds a first threshold; and when the amount of fish feed on the surface of the fishpond exceeds the first threshold after the first preset time has elapsed from the start time, issuing the control command to stop the feeder from feeding fish.

8. The control method for the feeding machine as described in claim 7, wherein when the amount of fish feed on the surface of the fishpond is less than the first threshold value after the first preset time has elapsed since the start time, the artificial intelligence engine (204) performs the following steps: after a second preset time has elapsed, it is determined, based on the image data, whether the amount of fish feed on the surface of the fishpond is lower than a second threshold value; when the amount of fish feed on the surface of the fishpond is lower than the second threshold value after the first preset time and the second preset time have elapsed since the start time, a control command is issued to make the feeding machine continue feeding fish; and when the amount of fish feed on the surface of the fishpond is greater than the second threshold value after the first preset time and the second preset time have elapsed since the start time, a control command is issued to make the feeding machine stop feeding fish.

9. The control method for the feeder as described in claim 8, wherein the first critical value is 35 and the second critical value is 5.

10. The method for controlling the feeder as claimed in claim 7, wherein the at least one environmental parameter data includes water temperature, fish weight and length, fish population density, and at least one or a combination of two or more of the pH and dissolved oxygen levels of the water in the fishpond.

11. A computer-readable storage device adapted to store at least one piece of code for controlling a feeder (20), wherein when the at least one piece of code is read, an artificial intelligence engine (204) is activated and executes at least the following procedures: obtaining at least one environmental parameter data of a fishpond from a sensing module (14); obtaining image data generated from images of fish in the fishpond captured by a monitoring module (16); generating a predicted demand and a control command based on the at least one environmental parameter data and the image data at a starting time, to control the feeder to feed fish to the fishpond according to the predicted demand; and generating the control command based on the image data after a first preset time elapsed from the starting time, to determine whether the feeder should continue feeding fish.

12. The computer-readable storage device as described in claim 11, wherein the artificial intelligence engine (204) performs the following steps after the first preset time has elapsed from the start time: determining, based on the image data, whether the amount of fish feed on the surface of the fishpond exceeds a first threshold; and when the amount of fish feed on the surface of the fishpond exceeds the first threshold after the first preset time has elapsed from the start time, issuing the control command to stop the feeding machine from feeding the fish.

13. The computer-readable storage device as described in claim 12, wherein when, after the first preset time has elapsed from the start time, it is determined that the amount of fish feed on the surface of the fishpond is less than the first threshold, the artificial intelligence engine performs the following steps: after a second preset time has elapsed, it determines, based on the image data, whether the amount of fish feed on the surface of the fishpond is lower than a second threshold; when, after the first and second preset times have elapsed from the start time, the amount of fish feed on the surface of the fishpond is lower than the second threshold, it issues the control command to cause the feeder to continue feeding fish; and when, after the first and second preset times have elapsed from the start time, the amount of fish feed on the surface of the fishpond is greater than the second threshold, it issues the control command to cause the feeder to stop feeding fish.

14. The computer-readable storage device as described in claim 13, wherein the first threshold value is 35 and the second threshold value is 5.

15. The computer-readable storage device as described in claim 13, wherein the first preset time is 1 minute and the second preset time is 3 minutes.

Images