Circulating aquaculture dissolved oxygen and water temperature regulation and control method fusing physical information

Abstract

The invention relates to the technical field of agricultural informatization, in particular to a recirculating aquaculture dissolved oxygen and water temperature regulation and control method fusing physical information. The method comprises the following steps: firstly, selecting liquid oxygen aeration and hot water temperature regulation as manipulated variables, carrying out aeration and heating pre-experiments, analyzing the response characteristics of dissolved oxygen and water temperature in the aeration and temperature regulation process, and establishing a kinetic model of dissolved oxygen and water temperature of the recirculating aquaculture system; and then, integrating the dynamic model into a neural network training process, and constructing a physical information neural network for dynamic response prediction of dissolved oxygen and water temperature. And finally, embedding the physical information neural network into a prediction controller, and providing a new model prediction control method to realize simultaneous accurate control of dissolved oxygen and water temperature in recirculating aquaculture.

Description

Technical Field

[0001] This invention relates to the field of agricultural information technology, specifically to a method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems that integrates physical information. Background Technology

[0002] During the hot summer months, aquaculture faces numerous severe challenges. High summer temperatures, with water temperatures reaching as high as 38 degrees Celsius, create an environment highly conducive to the growth of various pathogens. The proliferation of these pathogens poses a serious threat to the health of fish, increasing their susceptibility to disease and consequently affecting both yield and quality.

[0003] Meanwhile, water temperature is closely related to the dissolved oxygen content in the water. Generally speaking, the higher the water temperature, the lower the dissolved oxygen content. In high-temperature environments, the decrease in dissolved oxygen in the water can lead to the risk of oxygen deficiency for fish. When fish are oxygen deficient, their normal physiological activities, such as respiration, feeding, and growth, will be affected, and in severe cases, it can even lead to fish death, causing huge economic losses to fish farmers.

[0004] Currently, most recirculating aquaculture systems (RAS) use aeration for oxygenation, but this method is inefficient and cannot meet the needs of high-density aquaculture. With the rapid development of intensive aquaculture, RAS have an urgent need for efficient and precise decoupled control solutions for dissolved oxygen and water temperature in high-density aquaculture, making liquid oxygen aeration increasingly important.

[0005] When liquid oxygen dissolves in water, it not only causes a rapid increase in dissolved oxygen levels but also a sudden drop in water temperature. However, if the liquid oxygen flow rate is not properly managed, it can easily lead to drastic fluctuations in dissolved oxygen and water temperature within a short period, such as excessively high or low levels. This not only results in significant energy waste but is also harmful to fish growth.

[0006] Current methods for regulating liquid oxygen involve analyzing water level and dissolved oxygen data, and then using a PLC controller to adjust the flow rate of the aeration device to replenish dissolved oxygen in the aquaculture water. However, this approach does not integrate water temperature and dissolved oxygen levels for comprehensive regulation. In recirculating aquaculture systems (RAS) water quality control, achieving simultaneous and efficient regulation of both dissolved oxygen and water temperature is of significant practical importance. Summary of the Invention

[0007] To address the aforementioned technical problems, the present invention aims to provide a method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information. The specific technical solution adopted is as follows: Establish a dynamic model of dissolved oxygen and water temperature during the oxygenation and temperature regulation process of a recirculating aquaculture system; By incorporating the dissolved oxygen and water temperature dynamics model into a neural network model, a physical information neural network model is constructed to predict the dynamic response of dissolved oxygen and water temperature. A physical information neural network model is embedded in the predictive controller to adjust the flow rates of liquid oxygen and hot water in real time.

[0008] Furthermore, the establishment of the dissolved oxygen and water temperature kinetic model for the oxygenation and temperature regulation process of the recirculating aquaculture system includes: In the dynamic modeling of dissolved oxygen, the coupling effects of circulating water flow, liquid oxygen aeration, air-water surface reoxygenation, and temperature regulation are analyzed to determine the overall rate equation of dissolved oxygen change in the water body. In dynamic water temperature modeling, the coupling effects of circulating water flow rate, hot water inflow rate, air-water convective heat transfer, and oxygenation are analyzed to determine the overall rate equation for water temperature change. By combining the total rate equation for dissolved oxygen change and the total rate equation for water temperature change, a dynamic model of dissolved oxygen and water temperature is constructed.

[0009] Furthermore, in the dynamic modeling of dissolved oxygen, the coupling effects of circulating water flow, liquid oxygen aeration, air-water surface reoxygenation, and temperature regulation are analyzed to determine the overall rate equation for changes in dissolved oxygen in the water body, including: ; in, This refers to the circulating water flow rate; To oxygenate liquid oxygen; Reoxygenation of air and water surface; For temperature regulation coupling effect; denoted as , where is the total rate of change in dissolved oxygen in the water body; C is the dissolved oxygen concentration.

[0010] Furthermore, in the dynamic water temperature modeling, the coupling effects of circulating water flow rate, hot water inflow rate, air-water convective heat transfer, and oxygenation are analyzed to determine the overall rate equation for water temperature change, including: ; in, This refers to the circulating water flow rate; For hot water inflow; Air-water convective heat transfer; The coupling effect is for oxygenation; The total rate of change of water temperature is denoted as ; T is the water temperature.

[0011] Furthermore, the construction of the physical information neural network model includes: The data error term and the physical information error term are used as loss terms in the training process of the physical information neural network model; The formula for calculating the data error term is: ; in, This represents the data error value. This represents the dissolved oxygen or water temperature predicted by the network. Represents the actual observed dissolved oxygen or water temperature; N is the number of times the sample is traversed. For the dissolved oxygen or water temperature predicted by the network in the i-th iteration; The dissolved oxygen or water temperature is the i-th actual observation. The formula for calculating the physical information error term is as follows: ; in, This represents the error value of physical information. This represents the dissolved oxygen or water temperature predicted by the network. The output of the differential equation is either dissolved oxygen or water temperature. The dissolved oxygen or water temperature is the output of the i-th differential equation; By combining the data error term and the physical information error term, a loss function for the physical information neural network model is constructed.

[0012] Furthermore, the loss function for constructing the physical information neural network model by combining the data error term and the physical information error term includes: The data error term and the physical information error term are weighted separately, and the weighted summation error term is used as the loss function of the physical information neural network.

[0013] Furthermore, the method of embedding a physical information neural network model into a predictive controller to adjust the flow rates of liquid oxygen and hot water in real time includes: A prediction model based on a physical information neural network model is constructed, wherein the prediction model is the internal dynamic model of the prediction controller; Design a multi-objective optimization function with the core objectives of minimizing tracking error and minimizing energy consumption; Within each control cycle, the predictive controller, based on the predicted trajectory provided by the current system state and physical information neural network model, solves the optimal control sequence in the future finite time domain through a rolling optimization algorithm. The optimal control sequence includes the adjustment amounts of liquid oxygen flow rate and hot water flow rate.

[0014] The embodiments of the present invention have at least the following beneficial effects: To address the coupling problem between dissolved oxygen and water temperature in high-density recirculating aquaculture systems (RAS), making simultaneous and precise control difficult, this invention first selects liquid oxygen aeration and hot water temperature regulation as manipulated variables. Preliminary experiments on aeration and heating are conducted to analyze the response characteristics of dissolved oxygen and water temperature during these processes, establishing a dynamic model of dissolved oxygen and water temperature in the RAS system. Then, this dynamic model is integrated into a neural network training process to construct a physical information neural network for predicting the dynamic response of dissolved oxygen and water temperature. Finally, the physical information neural network is embedded into a predictive controller, proposing a novel model predictive control method to achieve simultaneous and precise control of dissolved oxygen and water temperature in RAS. This invention has practical significance for improving the quality and efficiency of aquaculture and promoting the green and low-carbon development of fisheries. This invention achieves simultaneous and precise control of dissolved oxygen and water temperature in high-density RAS, reducing aquaculture costs and improving aquaculture efficiency. This invention can simultaneously achieve precise control of dissolved oxygen and water temperature; it uses a physical information neural network model as the content prediction model of the predictive controller to improve the prediction accuracy and robustness of the controller; at the same time, it utilizes the dual effects of liquid oxygen dissolving in water, which has the effects of oxygenation and cooling, to reduce energy consumption and improve economic efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems that integrates physical information, provided in one embodiment of the present invention. Figure 2 This is a flowchart of another method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information, provided in one embodiment of the present invention. Figure 3 This is a schematic diagram of a recirculating aquaculture system provided in one embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems that integrates physical information, based on the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] This invention provides a specific implementation method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information. This method is applicable to recirculating aquaculture scenarios. The recirculating aquaculture system mainly includes aquaculture tanks, a collection pond, and a regulating pond.

[0020] The water collection tank mainly includes a waste filtration device, an ammonia nitrogen treatment device, and an ultraviolet sterilization device.

[0021] The water conditioning pond is used to optimize and regulate water quality, and the water flows into the aquaculture tank after meeting the required indicators.

[0022] The water conditioning tank includes a liquid oxygen aeration system and a temperature control system, with an embedded control method that can automatically adjust the liquid oxygen flow rate and hot water flow rate to ensure that the dissolved oxygen and water temperature in the water conditioning tank always meet the effluent standards.

[0023] The liquid oxygen enrichment system includes a liquid oxygen tank, solenoid valves, a liquid oxygen flow meter, and piping. The temperature control system includes a constant temperature hot water tank, solenoid valves, a hot water flow meter, and piping.

[0024] The following description, in conjunction with the accompanying drawings, details a specific scheme for a recirculating aquaculture system that integrates physical information to regulate dissolved oxygen and water temperature, as provided by this invention.

[0025] Please see Figure 1 The diagram illustrates a flowchart of a method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information, according to an embodiment of the present invention. The method includes the following steps: Step S100: Establish a dynamic model of dissolved oxygen and water temperature in the oxygenation and temperature regulation process of the recirculating aquaculture system.

[0026] A preliminary experiment on oxygenation and heating was conducted using target variables and manipulated variables. Dissolved oxygen and water temperature were selected as target variables, and the corresponding manipulated variables were liquid oxygen flow rate and hot water flow rate, respectively.

[0027] Dissolved oxygen is regulated by liquid oxygen aeration, and its flow rate is controlled by the opening of a solenoid valve; water temperature is regulated by injecting constant-temperature hot water, and its flow rate is also controlled by the opening of a solenoid valve.

[0028] Based on the principles of mass and energy conservation, the dissolved oxygen concentration and temperature in the water conditioning tank are always in a dynamic balance. Considering the actual operating conditions of recirculating aquaculture, and ignoring secondary factors, we selected the main influencing factors for dynamic modeling of dissolved oxygen and water temperature. In the dynamic modeling of dissolved oxygen, the circulating water flow rate (…) was taken into account. ), liquid oxygen oxygenation ( ), air-water reoxygenation ( ) and the coupling effect of the temperature regulation process ( The influence of these four main factors on dissolved oxygen changes led to the establishment of the following expression characterizing the overall rate of dissolved oxygen change in water bodies. Specifically, in the dynamic modeling of dissolved oxygen, the coupling effects of circulating water flow, liquid oxygen aeration, air-water surface reoxygenation, and temperature regulation were analyzed to determine the equation for the overall rate of dissolved oxygen change in water bodies: ; in, This refers to the circulating water flow rate; To oxygenate liquid oxygen; Reoxygenation of air and water surface; For temperature regulation coupling effect; denoted as , where is the total rate of change in dissolved oxygen in the water body; C is the dissolved oxygen concentration.

[0029] In dynamic water temperature modeling, the circulating water flow rate is considered ( ), hot water inflow ( ), air-water convective heat transfer ( The coupling effect of ) and oxygenation ( The influence of four main factors on water temperature changes was analyzed to establish the overall rate of water temperature change. Specifically, in water temperature dynamic modeling, the coupling effects of circulating water flow, hot water inflow, air-water convective heat transfer, and oxygenation were analyzed to determine the overall rate equation for water temperature change. ; in, This refers to the circulating water flow rate; For hot water inflow; Air-water convective heat transfer; The coupling effect is for oxygenation; The total rate of change of water temperature is denoted as ; T is the water temperature.

[0030] After obtaining the total rate equations for the changes in dissolved oxygen and water temperature over time, the two equations can be combined to obtain the total rate equations for the dynamic changes in dissolved oxygen and water temperature. In other words, by combining the total rate equations for the changes in dissolved oxygen and water temperature, a dynamic model of dissolved oxygen and water temperature can be constructed.

[0031] ; In this model, general parameters can be determined based on the hardware and environmental parameters of the circulating water system, while the two coupling factors... and The main analytical target is dissolved oxygen concentration, which is expressed in mg / L, and water temperature is expressed in °C.

[0032] Step S200: The dissolved oxygen and water temperature dynamics model is introduced into the neural network model to construct a physical information neural network model to predict the dynamic response of dissolved oxygen and water temperature.

[0033] In a real recirculating aquaculture system, multiple sets of aeration and heating pre-experiments were conducted to record dissolved oxygen and water temperature response data, obtaining a considerable number of input-output sample datasets for subsequent modeling training. In a single sample dataset, the inputs are liquid oxygen flow rate, hot water flow rate, initial dissolved oxygen value, and initial water temperature value; the outputs are the future dissolved oxygen and water temperature values.

[0034] A multi-input multi-output (MIMO) physical information neural network is used to fit and train the collected data. The MIMO model used in this invention considers two error factors as the network's loss function: a data error term and a physical information error term.

[0035] The formula for calculating the data error term is: ; in, This represents the data error value. This represents the dissolved oxygen or water temperature predicted by the network. Represents the actual observed dissolved oxygen or water temperature; N is the number of times the sample is traversed. For the dissolved oxygen or water temperature predicted by the network in the i-th iteration; Let be the dissolved oxygen or water temperature observed in the i-th actual observation.

[0036] The formula for calculating the physical information error term is as follows: ; in, This represents the error value of physical information. This represents the dissolved oxygen or water temperature predicted by the network. The output of the differential equation is either dissolved oxygen or water temperature. is the dissolved oxygen or water temperature output by the i-th differential equation.

[0037] Finally, combining the data error term and the physical information error term, a loss function for the physical information neural network model is constructed: the data error term and the physical information error term are weighted respectively, and the weighted summation error term is used as the loss function of the physical information neural network.

[0038] The final loss function of the physical information neural network model is a combination of the two, as shown in the following formula: ; In the formula, and It is a hyperparameter that measures the importance of two error terms. Its value is in the range of 0-1. The optimal combination of the two parameters can be obtained through multiple sets of control experiments.

[0039] The physical information neural network model can be a BP neural network, a long short-term memory network, or other neural networks with modeling capabilities.

[0040] In step S300, the physical information neural network model is embedded into the predictive controller to adjust the flow rates of liquid oxygen and hot water in real time.

[0041] A prediction model based on a physical information neural network model is constructed and used as the internal dynamic model of the prediction controller to predict the dynamic response trajectory of dissolved oxygen concentration and water temperature in future time periods.

[0042] Conducting on-site investigations to understand production needs and determine optimization indicators. This method selects high control accuracy and low energy consumption as objective functions, that is, designing a multi-objective optimization function with small tracking error and low energy consumption as the core objectives.

[0043] Within each control cycle, the predictive controller, based on the predicted trajectory provided by the physical information neural network model of the current system state, solves the optimal control sequence in the future finite time domain through a rolling optimization algorithm. This optimal control sequence includes the adjustment amounts of liquid oxygen flow rate and hot water flow rate.

[0044] The objective function is defined in the following form: ; Among them, Y all Y represents the predicted dissolved oxygen or water temperature value obtained through a physical information neural network; ref U represents the setpoint for dissolved oxygen or water temperature; Q represents the control variable; Q and R are the error weight matrix and control weight matrix, respectively; J is the objective function value.

[0045] It should be noted that the solution oxygen or water temperature setpoints can be set by the implementer according to the needs of the aquatic species.

[0046] After completing the above steps, the obtained physical information neural network model and the defined objective optimization function are combined to design a predictive controller. During the control process, for the safety and stability of system operation, excessively drastic changes or excessively large control variables are not desired. Therefore, this invention imposes restrictions on the control variables, with the following constraints: ; in, , , and All of these are constants, and their values ​​can be set by the implementer according to the actual situation of the control system; This indicates the liquid oxygen flow rate, used to control the oxygen content in water. This indicates the flow rate of hot water, used to control the temperature in the water.

[0047] The response trajectories to dissolved oxygen and water temperature are predicted based on the current input, and this prediction is used as a benchmark to calculate the error index. The controller, based on the predicted trajectory provided by the current system state and physical information neural network, uses a rolling optimization algorithm to solve for the optimal control sequence within the future finite time domain, i.e., let... The optimal liquid oxygen flow rate and hot water flow rate are determined. Based on the obtained optimal liquid oxygen flow rate and hot water flow rate, the dissolved oxygen and water temperature in the next cycle are controlled to complete the closed-loop precise control, thereby enhancing the system's anti-interference ability and adaptability.

[0048] During the control phase, the control quantity of the first time step of the optimized control sequence in each control cycle is actually applied to the controlled process to achieve closed-loop precise control of dissolved oxygen and water temperature. In the next sampling cycle, the dissolved oxygen content and water temperature data in the current water body are remeasured, and prediction and optimization are performed again based on the latest measurement data.

[0049] Please see Figure 2 , Figure 2 This is another flowchart of a method for regulating dissolved oxygen and water temperature in recirculating aquaculture that integrates physical information, provided as an embodiment of the present invention.

[0050] Please see Figure 3 , Figure 3 This is a schematic diagram of a recirculating aquaculture system provided in one embodiment of the present invention.

[0051] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0052] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information, characterized in that, The method includes the following steps: Establish a dynamic model of dissolved oxygen and water temperature during the oxygenation and temperature regulation process of a recirculating aquaculture system; By incorporating the dissolved oxygen and water temperature dynamics model into a neural network model, a physical information neural network model is constructed to predict the dynamic response of dissolved oxygen and water temperature. A physical information neural network model is embedded in the predictive controller to adjust the flow rates of liquid oxygen and hot water in real time.

2. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 1, characterized in that, The establishment of the dissolved oxygen and water temperature dynamic model for the oxygenation and temperature regulation process of the recirculating aquaculture system includes: In the dynamic modeling of dissolved oxygen, the coupling effects of circulating water flow, liquid oxygen aeration, air-water surface reoxygenation, and temperature regulation are analyzed to determine the overall rate equation of dissolved oxygen change in the water body. In dynamic water temperature modeling, the coupling effects of circulating water flow rate, hot water inflow rate, air-water convective heat transfer, and oxygenation are analyzed to determine the overall rate equation for water temperature change. By combining the total rate equation for dissolved oxygen change and the total rate equation for water temperature change, a dynamic model of dissolved oxygen and water temperature is constructed.

3. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 2, characterized in that, In the dynamic modeling of dissolved oxygen, the coupling effects of circulating water flow, liquid oxygen aeration, air-water surface reoxygenation, and temperature regulation are analyzed to determine the overall rate equation for changes in dissolved oxygen in the water body, including: ； in, This refers to the circulating water flow rate; To oxygenate liquid oxygen; Reoxygenation of air and water surface; For temperature regulation coupling effect; denoted as , where is the total rate of change in dissolved oxygen in the water body; C is the dissolved oxygen concentration.

4. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 2, characterized in that, In the dynamic water temperature modeling, the coupling effects of circulating water flow rate, hot water inflow rate, air-water convective heat transfer, and oxygenation are analyzed to determine the overall rate equation for water temperature change, including: ； in, This refers to the circulating water flow rate; For hot water inflow; Air-water convective heat transfer; The coupling effect is for oxygenation; The total rate of change of water temperature is denoted as ; T is the water temperature.

5. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 1, characterized in that, The construction of the physical information neural network model includes: The data error term and the physical information error term are used as loss terms in the training process of the physical information neural network model; The formula for calculating the data error term is: ； in, This represents the data error value. This represents the dissolved oxygen or water temperature predicted by the network. Represents the actual observed dissolved oxygen or water temperature; N is the number of times the sample is traversed. For the dissolved oxygen or water temperature predicted by the network in the i-th iteration; The dissolved oxygen or water temperature is the i-th actual observation; The formula for calculating the physical information error term is as follows: ； in, This represents the error value of physical information. This represents the dissolved oxygen or water temperature predicted by the network. The output of the differential equation is either dissolved oxygen or water temperature. The dissolved oxygen or water temperature is the output of the i-th differential equation; By combining the data error term and the physical information error term, a loss function for the physical information neural network model is constructed.

6. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 5, characterized in that, The loss function for constructing the physical information neural network model by combining the data error term and the physical information error term includes: The data error term and the physical information error term are weighted separately, and the weighted summation error term is used as the loss function of the physical information neural network.

7. The method for regulating dissolved oxygen and water temperature in recirculating aquaculture systems by integrating physical information as described in claim 1, characterized in that, The method of embedding a physical information neural network model into a predictive controller to adjust the flow rates of liquid oxygen and hot water in real time includes: A prediction model based on a physical information neural network model is constructed, wherein the prediction model is the internal dynamic model of the prediction controller; Design a multi-objective optimization function with the core objectives of minimizing tracking error and minimizing energy consumption; Within each control cycle, the predictive controller, based on the predicted trajectory provided by the current system state and physical information neural network model, solves the optimal control sequence in the future finite time domain through a rolling optimization algorithm. The optimal control sequence includes the adjustment amounts of liquid oxygen flow rate and hot water flow rate.

Images