Fresh water aquaculture water quality monitoring method and system based on Internet of Things

Abstract

The invention discloses a fresh water aquaculture water quality monitoring method and system based on the Internet of Things, and relates to the technical field of fresh water aquaculture, and the method comprises the following steps: firstly, determining an initial dissolved oxygen concentration threshold value based on previous aquaculture data and experience; a high-precision dissolved oxygen sensor network is deployed in a culture environment, it is ensured that dissolved oxygen sensors can cover a key area of a whole water body, and real-time dissolved oxygen data is transmitted to a central control system through a wireless communication technology. Through the high-precision dissolved oxygen and metabolic data sensor, deep learning is combined to intelligently predict and monitor the metabolic state of aquatic products in real time, dynamically adjust the threshold value of the dissolved oxygen and automatically control the oxygenation equipment, so that timely oxygen supply in the metabolic peak period is ensured, the growth problem caused by oxygen deficit is prevented, the breeding efficiency is effectively improved, and the energy utilization is optimized; management difficulty is reduced, aquatic product health is guaranteed, and a win-win situation of economic and ecological benefits is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of freshwater aquaculture, and specifically relates to a method and system for monitoring the water quality of freshwater aquaculture based on the Internet of Things. Background Art

[0002] The monitoring of the water quality of freshwater aquaculture based on the Internet of Things refers to a systematic solution that uses Internet of Things (IoT) technology to monitor key water quality parameters in the freshwater aquaculture environment in real time, perform data analysis, and carry out intelligent regulation. This system usually includes a distributed water quality sensor network, a wireless data transmission module, a cloud data processing and storage platform, and an intelligent decision-making and control unit. By deploying sensors such as dissolved oxygen, pH value, temperature, ammonia nitrogen, nitrate, and conductivity, water quality data is collected in real time, and the data is uploaded to the cloud through wireless communication technologies (such as LoRa, NB-IoT, 4G / 5G). The cloud uses big data analysis and machine learning algorithms to predict the trend of water quality fluctuations and identify anomalies, so as to achieve precise regulation, such as automatic oxygenation, feeding, or water body replacement, to ensure the stability of the aquaculture environment and the healthy growth of aquatic products. In addition, this system can provide visual water quality monitoring reports and remote warning notifications, helping aquaculture farmers improve management efficiency, reduce labor costs, and effectively prevent economic losses caused by sudden water quality changes.

[0003] In the water quality monitoring system of freshwater aquaculture based on the Internet of Things, the role of automatic oxygenation is to ensure that the concentration of dissolved oxygen (DO) in the water body is maintained within an appropriate range to meet the normal physiological needs of aquatic organisms, promote healthy growth, and at the same time prevent aquaculture risks caused by oxygen deficiency. When the system detects that the dissolved oxygen concentration is lower than the set threshold, the control unit will automatically start the oxygenation equipment, such as an oxygenation pump, a micro-nano aeration system, or a waterwheel aerator, to quickly increase the oxygen content in the water body and prevent stress reactions, growth retardation, and even death risks of fish and shrimp caused by oxygen deficiency. After the oxygenation system reaches the preset target dissolved oxygen concentration, it will automatically shut down to achieve precise control and avoid unnecessary energy consumption. In addition, reasonable oxygenation control can also prevent negative effects that may be caused by excessive oxygenation, such as increased ammonia nitrogen accumulation, fluctuations in the pH value of the water body, and imbalance of the aquaculture water environment, thereby ensuring the long-term stability of the water quality, improving aquaculture efficiency, and environmental sustainability.

[0004] The existing technologies have the following deficiencies:

[0005] When monitoring the water quality of freshwater aquaculture based on the Internet of Things, the existing technology usually sets a constant dissolved oxygen concentration threshold. However, during the peak metabolic period, the oxygen demand of aquaculture organisms rises sharply, and the fixed dissolved oxygen concentration threshold may not be able to meet the actual needs in time, resulting in the water body quickly entering an anoxic state. At a minimum, it can cause a decline in the vitality of aquatic products, a reduction in food intake, and affect the growth rate; at worst, it can cause large-scale asphyxiation deaths of aquatic products due to severe hypoxia, resulting in huge economic losses, and even may lead to the collapse of the entire farm.

[0006] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, so it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The object of the present invention is to provide a method and system for monitoring the water quality of freshwater aquaculture based on the Internet of Things. By deploying high-precision dissolved oxygen and metabolic data sensors and combining the intelligent prediction of deep learning models, it realizes real-time monitoring of the metabolic state of aquatic products, and dynamically adjusts the dissolved oxygen threshold and the automatic control of oxygenation equipment. The dissolved oxygen threshold is raised in advance during the peak metabolic period to ensure sufficient oxygen content in the water body, prevent the decline in the vitality of aquatic products, reduce food intake and death, ensure the healthy growth of organisms, and improve the breeding efficiency and output. At the same time, the method optimizes energy utilization, reduces the ineffective operation of equipment and operating costs, reduces human intervention, improves management stability and sustainability, and ultimately achieves a win-win situation of economic and ecological benefits to solve the problems in the above background art.

[0008] To achieve the above object, the present invention provides the following technical solutions: A method for monitoring the water quality of freshwater aquaculture based on the Internet of Things, comprising the following steps:

[0009] First, based on past breeding data and experience, determine an initial dissolved oxygen concentration threshold;

[0010] Deploy a high-precision dissolved oxygen sensor network in the breeding environment to ensure that the dissolved oxygen sensors can cover the key areas of the entire water body, and transmit the real-time dissolved oxygen data to the central control system through wireless communication technology;

[0011] When the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water body is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues an instruction to start the oxygenation equipment to increase the oxygen content in the water body, quickly raise the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold;

[0012] Deploy a variety of sensors in the breeding environment to collect the metabolic data of aquatic products in real time, and transmit the metabolic data to the central processing system through Internet of Things devices;

[0013] Preprocess the real-time obtained metabolic data to ensure the consistency and integrity of the data;

[0014] In the preprocessed metabolic data, use feature extraction technology to extract key indicators that can reflect the metabolic peak of aquatic products, and form a core dataset for model analysis;

[0015] Input the extracted key metabolic data into a pre-trained deep learning model to intelligently predict the metabolic situation of aquatic products, predict the metabolic state of aquatic products in a future period of time, and classify the results into "metabolic peak" or "normal metabolism";

[0016] For aquatic products in the "normal metabolism" state, continue to conduct intelligent control with the set initial dissolved oxygen threshold to maintain efficient energy utilization; for aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use, improve the oxygen supply level, and start the aeration equipment in advance to ensure sufficient oxygen concentration in the water body during the high oxygen demand period.

[0017] Preferably, the initial dissolved oxygen concentration threshold refers to the lowest dissolved oxygen concentration standard preset according to historical aquaculture data, the physiological needs of aquaculture varieties, and environmental conditions. When the dissolved oxygen concentration in the water body is lower than the initial dissolved oxygen concentration threshold, the aeration equipment will be automatically triggered to maintain the dissolved oxygen demand of the aquaculture environment and ensure the normal growth and survival of aquatic organisms.

[0018] Preferably, in the preprocessed metabolic data, use feature extraction technology to extract key indicators that can reflect the metabolic peak of aquatic products. The specific steps are as follows:

[0019] Screen out the gill opening and closing frequency of aquatic products and the generation rate of aquatic product excrement from the preprocessed metabolic data. Under the monitoring window, perform feature engineering on the screened gill opening and closing frequency of aquatic products and the generation rate of aquatic product excrement to generate a gill movement frequency reference value and a metabolic waste excretion reference value respectively. Establish a core dataset with the gill movement frequency reference value and the metabolic waste excretion reference value to quantify the physiological load and energy consumption level of aquatic products during the metabolic peak period, so as to accurately identify the behavioral characteristics and metabolic intensity in the high metabolic state.

[0020] Preferably, input the extracted gill movement frequency reference value and metabolic waste excretion reference value into a pre-trained deep learning model, generate a dynamic metabolic vitality evaluation factor based on the deep learning model, and intelligently predict the metabolic situation of aquatic products through the dynamic metabolic vitality evaluation factor.

[0021] Preferably, when the dynamic evaluation factor of metabolic vitality generated by the pre-trained deep learning model for intelligent prediction of the metabolic situation of aquatic products under the monitoring window is compared with the reference threshold of the pre-set dynamic evaluation factor of metabolic vitality, the results are divided into "metabolic peak" or "normal metabolism", and the specific steps are as follows:

[0022] If the dynamic evaluation factor of metabolic vitality is greater than the reference threshold of the dynamic evaluation factor of metabolic vitality, the metabolic level of the aquatic product is divided into "metabolic peak"; if the dynamic evaluation factor of metabolic vitality is less than or equal to the reference threshold of the dynamic evaluation factor of metabolic vitality, the metabolic level of the aquatic product is divided into "normal metabolism".

[0023] Preferably, under the monitoring window, the specific steps for generating the reference value of gill movement frequency by feature engineering of the gill opening and closing frequency of the selected aquatic products are as follows:

[0024] By performing frequency domain decomposition on the original signal of the gill opening and closing frequency and calculating the weighted sum of its dynamic components through integration, the core frequency characteristics of gill movement are extracted, and the extraction expression is:

[0025] ,

[0026] where F osc is the dynamically extracted component of the gill opening and closing frequency, which is used to quantify the frequency change trend of gill movement within the monitoring window, S osc is the time series of the original signal of gill movement, which reflects the time series of gill opening and closing actions, ω is the frequency component of the movement, which describes the frequency domain characteristics of the gill opening and closing frequency within the monitoring window, λ is the exponential decay coefficient, which is used to reduce the influence of high-frequency noise while retaining the key information of low-frequency characteristic signals, and τ is the length of the monitoring window, which defines the time range of signal processing;

[0027] Using the obtained dynamically gill opening and closing frequency component, combined with the temporal curvature and non-linear characteristics of the gill movement signal, a reference value of gill movement frequency is generated. By performing feature enhancement on the second derivative of the signal and introducing a phase correction term, the finally generated result can accurately reflect the physiological load level of the aquatic product during the metabolic peak period. The generation expression of the reference value of gill movement frequency is as follows:

[0028] ,

[0029] where G index is the finally generated reference value of gill movement frequency, which quantifies the physiological load level of the aquatic product during the metabolic peak period, κ dyn is the dynamic curvature correction coefficient, which describes the non-linear change characteristics of the gill opening and closing frequency, is the second derivative of the gill movement signal, revealing the acceleration and non-linear dynamic changes of gill movement. α is the non-linear enhancement factor, used to amplify the signal characteristics in the high dynamic frequency region. φ is the phase of the gill movement signal, reflecting the periodic characteristics of gill movement and correcting the stability of feature calculation. cos(φ) is used to adjust the symmetry of the signal characteristics and reduce external interference. β is the weight factor, used to balance the influence of time series characteristics and frequency domain characteristics and optimize the accuracy of calculation.

[0030] Preferably, under the monitoring window, the specific steps for generating the metabolic waste excretion reference value through feature engineering on the generation rate of the screened aquatic excrement are as follows:

[0031] Within the monitoring window, first construct a time series trend model for the collected data of the generation rate of aquatic excrement, adopt a non-uniform smoothing filtering method, combined with a weight allocation strategy, to emphasize the influence of recent data on the metabolic state. The constructed expression is:

[0032] ,

[0033] Where: is the smoothed sequence of the excrement generation rate, R w (i) represents the aquatic excretion rate collected within the monitoring window, δ is the weight decay factor, 0 < δ < 1, reflecting the importance of recent data, n is the total number of data points within the monitoring window, α' is the adjustment coefficient to ensure that the system adapts to different variation ranges of excretion rates, i represents the data sampling serial number within the monitoring window, that is, the index position of the current sampling point, max(R w ) represents the maximum value among all the aquatic excretion rate data within the monitoring window, min(R w ) represents the minimum value among all the aquatic excretion rate data within the monitoring window;

[0034] After obtaining the smoothed sequence of the excrement generation rate, construct the metabolic waste excretion reference value, adopt a non-linear incremental analysis model, through the change amplitude of the cumulative excretion rate, combined with the metabolic fluctuation amplitude index for feature enhancement. The constructed expression is:

[0035] ,

[0036] Where: E wi is the metabolic waste excretion reference value, is the value of the smoothed excrement generation rate at the jth sampling point, γ is the non-linear exponential coefficient, amplifying the influence of drastic metabolic changes, β' is the basic weight factor, used to adjust the sensitivity, μ is the weight coefficient of the fluctuation amplitude, Φ m is the metabolic fluctuation amplitude index, measuring the degree of drastic fluctuation of the excretion rate, defined as:

[0037] ,

[0038] Among them, is the value of the smoothed excrement production rate at the k-th sampling point, is the value of the smoothed excrement production rate at the (k - 1)-th sampling point, and η is the smoothing adjustment coefficient, which controls the influence degree of drastic changes.

[0039] Preferably, for aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use. The specific steps are as follows:

[0040] When the aquatic products are in the "metabolic peak" state, based on the metabolic vitality dynamic evaluation factor, dynamically adjust the dissolved oxygen threshold to meet their oxygen demand during the high metabolism period. The dynamic adjustment expression of the dissolved oxygen threshold is:

[0041] ,

[0042] Among them, DO adj is the dynamically adjusted dissolved oxygen threshold, which is used for actual oxygenation decision-making to ensure meeting the high metabolism demand. DO init is the initial dissolved oxygen threshold, Meta activity is the metabolic vitality dynamic evaluation factor, which represents the current metabolic intensity of the aquatic products. Meta ref is the reference threshold of the metabolic vitality dynamic evaluation factor, which represents the reference value under the normal metabolic state. α x is the adjustment coefficient, which controls the sensitivity of the dissolved oxygen threshold adjustment and determines the weight of the adjustment range. β x is the exponential weight, which adjusts the non-linearity degree of the influence of the metabolic vitality dynamic evaluation factor, and the value range is 1 < β x < 3.

[0043] An Internet of Things-based freshwater aquaculture water quality monitoring system includes an initial dissolved oxygen threshold setting module, a dissolved oxygen monitoring module, an oxygenation control module, a metabolic data collection module, a metabolic data preprocessing module, a key feature extraction module, an intelligent metabolism prediction module, and an intelligent dissolved oxygen regulation module;

[0044] The initial dissolved oxygen threshold setting module first determines an initial dissolved oxygen concentration threshold based on past aquaculture data and experience;

[0045] The dissolved oxygen monitoring module deploys a high-precision dissolved oxygen sensor network in the aquaculture environment to ensure that the dissolved oxygen sensors can cover key areas of the entire water body, and transmits real-time dissolved oxygen data to the central control system through wireless communication technology;

[0046] Oxygenation control module: When the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water body is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues an instruction to start the oxygenation equipment to increase the oxygen content in the water body, quickly raise the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold;

[0047] Metabolic data acquisition module: Deploy a variety of sensors in the aquaculture environment to collect the metabolic data of aquatic products in real time, and transmit the metabolic data to the central processing system through Internet of Things devices;

[0048] Metabolic data preprocessing module: Preprocess the real-time obtained metabolic data to ensure the consistency and integrity of the data;

[0049] Key feature extraction module: In the preprocessed metabolic data, use feature extraction technology to extract key indicators that can reflect the metabolic peak of aquatic products, and form a core data set for model analysis;

[0050] Intelligent metabolic prediction module: Input the extracted key metabolic data into a pre-trained deep learning model to intelligently predict the metabolic situation of aquatic products, predict the metabolic state of aquatic products in the future for a period of time, and divide the results into "metabolic peak" or "normal metabolism";

[0051] Intelligent dissolved oxygen regulation module: For aquatic products in the "normal metabolism" state, continue to carry out intelligent control with the set initial dissolved oxygen threshold to maintain efficient energy utilization; for aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use, improve the oxygen supply level, and start the oxygenation equipment in advance to ensure sufficient oxygen concentration in the water body during the high oxygen demand period.

[0052] In the above technical solution, the technical effects and advantages provided by the present invention:

[0053] The present invention deploys high-precision dissolved oxygen sensors and a variety of metabolic data sensors, combines the intelligent prediction of the deep learning model, monitors and analyzes the metabolic state of aquatic products in real time, realizes the automatic control of dynamically adjusting the dissolved oxygen threshold and the oxygenation equipment. When it is predicted that the metabolic peak is about to come, automatically increase the dissolved oxygen threshold and start the oxygenation equipment in advance to ensure sufficient oxygen content in the water body, avoid the decline in the vitality of aquatic products, reduced feeding, and even large-scale death caused by insufficient dissolved oxygen. It not only guarantees the healthy growth of aquatic organisms, improves the aquaculture efficiency and output, but also optimizes the energy utilization, reduces the ineffective operation of the oxygenation equipment and related operation costs. At the same time, the intelligent water quality management reduces the human intervention and management difficulty, improves the stability and sustainability of the aquaculture process, and finally achieves a win-win situation of economic and ecological benefits. Description of the Drawings

[0054] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required in the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0055] Figure 1 This is the method flow chart of a method for monitoring the water quality of freshwater aquaculture based on the Internet of Things according to the present invention.

[0056] Figure 2 This is the module schematic diagram of a system for monitoring the water quality of freshwater aquaculture based on the Internet of Things according to the present invention. Detailed implementation manners

[0057] Now, the exemplary embodiments will be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these exemplary embodiments are provided so that the present disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art.

[0058] The present invention provides a method for monitoring the water quality of freshwater aquaculture based on the Internet of Things as shown in Figure 1 the following, which includes the following steps:

[0059] First, based on past aquaculture data and experience, determine an initial dissolved oxygen concentration threshold;

[0060] The initial dissolved oxygen concentration threshold should consider factors such as the oxygen demand of the aquaculture species, water body capacity, aquaculture density, environmental temperature, and water quality conditions. Usually, by analyzing the dissolved oxygen requirements at different growth stages and environmental conditions in historical records, a basic threshold that can meet the healthy growth of aquatic products in most cases is set. The setting of the initial threshold provides a reference point for the entire water quality monitoring system, ensuring a stable reference standard at the initial stage of the system. This step ensures that the system can maintain basic water quality stability before dynamic adjustment, preventing system failure caused by missing parameters in the initial stage. At the same time, setting the threshold based on historical data can improve the scientificity and rationality of the initial setting and reduce errors caused by manual setting.

[0061] The initial dissolved oxygen concentration threshold refers to the minimum dissolved oxygen concentration standard preset according to historical aquaculture data, the physiological requirements of aquaculture varieties, and environmental conditions (such as water temperature, aquaculture density, etc.).

[0062] Deploy a high-precision dissolved oxygen sensor network in the aquaculture environment to ensure that the dissolved oxygen sensors can cover the key areas of the entire water body, and transmit the real-time dissolved oxygen data to the central control system through wireless communication technologies (such as LoRa, NB-IoT, 4G / 5G);

[0063] Obtaining real-time dissolved oxygen data is the basis of the entire monitoring system, ensuring that the system can immediately understand the oxygen status in the water body. When the dissolved oxygen concentration changes, the dissolved oxygen sensor can capture these changes in a timely manner, providing accurate data support for subsequent automatic regulation. The accuracy of this link is directly related to the timeliness and effectiveness of the system response, and is the key to preventing hypoxia problems.

[0064] When the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water body is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues an instruction to start the aeration equipment (such as an aeration pump, a micro-nano aeration system or a waterwheel aerator) to increase the oxygen content in the water body, quickly raise the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold;

[0065] The automated aeration response mechanism ensures that in the case of insufficient dissolved oxygen, the system can take prompt action to prevent the water body from entering an anoxic state. By starting the aeration equipment in a timely manner, it ensures the physiological needs of aquatic organisms and avoids the risk of growth stagnation or death caused by hypoxia. At the same time, the automatic shutdown mechanism effectively saves energy and avoids the negative impacts of over-aeration, such as pH fluctuations and ammonia nitrogen accumulation, maintaining the long-term stability of water quality.

[0066] Deploy multiple sensors (such as feeding sensors, activity monitoring devices, in-vivo sensors, etc.) in the aquaculture environment, collect the metabolic data of aquatic products in real time, and transmit the metabolic data to the central processing system through Internet of Things devices;

[0067] Metabolic data reflects the physiological state and oxygen demand of aquatic organisms, and is an important basis for predicting metabolic peaks. By obtaining this data in real time, the system can comprehensively understand the growth dynamics and behavioral changes of aquatic products, providing rich input information for subsequent data analysis and deep learning models. This step lays a data foundation for realizing intelligent prediction and dynamic regulation, and improves the accuracy and response ability of the system.

[0068] Preprocess the real-time obtained metabolic data to ensure the consistency and integrity of the data;

[0069] The preprocessing step is a crucial link in ensuring data quality and analysis accuracy. High-quality preprocessing can eliminate interfering factors in the data, extract representative and usable feature data, and provide reliable data input for subsequent key data extraction and deep learning model training. This process effectively improves the overall prediction ability and decision-making accuracy of the system, avoiding misjudgments and incorrect responses caused by data quality problems.

[0070] In the preprocessed metabolic data, feature extraction techniques are used to extract key indicators that can reflect the metabolic peak of aquatic products, forming a core dataset for model analysis;

[0071] The extracted key metabolic data is input into a pre-trained deep learning model (such as Long Short-Term Memory Network LSTM, Convolutional Neural Network CNN, or hybrid model) to intelligently predict the metabolic situation of aquatic products, predict the metabolic state of aquatic products in a future period of time, and classify the results into "metabolic peak" or "normal metabolism";

[0072] The purpose of key data extraction is to simplify the data dimension and retain the features that have the most influence on the prediction of the metabolic peak. This not only improves the data processing efficiency but also enhances the training effect and prediction accuracy of the deep learning model. By accurately identifying the key indicators of the metabolic peak, the system can more effectively capture the dynamic changes in the oxygen demand of aquatic products, give early warnings and take corresponding measures to prevent sudden hypoxia problems.

[0073] In the preprocessed metabolic data, feature extraction techniques are used to extract key indicators that can reflect the metabolic peak of aquatic products. The specific steps are as follows:

[0074] Screen out the gill opening and closing frequency of aquatic products and the production rate of aquatic product excrement from the preprocessed metabolic data. Under the monitoring window, perform feature engineering on the screened gill opening and closing frequency of aquatic products and the production rate of aquatic product excrement to generate the gill movement frequency reference value and the metabolic waste excretion reference value respectively. Establish a core dataset with the gill movement frequency reference value and the metabolic waste excretion reference value to quantify the physiological load and energy consumption level of aquatic products during the metabolic peak period, so as to accurately identify the behavioral characteristics and metabolic intensity in the high metabolic state.

[0075] When the gill opening and closing frequency of aquatic products is relatively high, it usually indicates that the metabolism of aquatic products is at a peak. The gill is the main organ for gas exchange in aquatic organisms, and its opening and closing frequency is closely related to the physiological metabolic needs. During the peak of metabolism, due to the significant enhancement of physiological activities such as feeding, digestion, and movement of aquatic products, the demand for oxygen and nutrients in the body also increases accordingly. As a result, the gills need to open and close more frequently to accelerate the gas exchange rate, so as to meet the high-intensity metabolic needs in the body. At the same time, a high gill opening and closing frequency is often accompanied by accelerated energy consumption and the discharge of metabolic wastes such as ammonia nitrogen and carbon dioxide. Therefore, the gill opening and closing frequency of aquatic products can be used as an important physiological indicator to identify the entry of aquatic products into a high metabolic state.

[0076] Under the monitoring window, the specific steps for generating the reference value of gill movement frequency through feature engineering for the gill opening and closing frequency of the selected aquatic products are as follows:

[0077] By performing frequency domain decomposition on the original signal of the gill opening and closing frequency and calculating the weighted sum of its dynamic components through integration, the core frequency characteristics of gill movement are extracted. The extraction expression is:

[0078] ,

[0079] where F osc is the dynamically extracted component of the gill opening and closing frequency, used to quantify the frequency change trend of gill movement within a specific monitoring window. S osc is the time series of the original signal of gill movement, reflecting the time series of gill opening and closing actions. ω is the frequency component of the movement, describing the frequency domain characteristics of the gill opening and closing frequency within the monitoring window. λ is the exponential decay coefficient, used to reduce the influence of high-frequency noise while retaining the key information of low-frequency characteristic signals. τ is the length of the monitoring window, defining the time range of signal processing;

[0080] The exponential decay term in the formula effectively suppresses high-frequency noise and emphasizes the movement characteristics in the middle and low frequency bands, making the result more accurately reflect the physiological change trend of gill movement. This feature extraction process is the basis for generating the reference value of gill movement frequency, ensuring the reliability and accuracy of subsequent data analysis.

[0081] The gill opening and closing data of aquatic products can be obtained through a variety of advanced monitoring technologies, mainly including high-frequency image / video analysis, micro biosensors, ultrasonic monitoring, and laser interferometry. Among them, the high-frequency image / video analysis technology uses an underwater camera or an infrared camera to capture the movement of the gills of aquatic products in real time, and automatically identifies the opening and closing frequency and amplitude of the gills through computer vision algorithms (such as optical flow method, deep learning object detection model, etc.). The micro biosensor can be directly attached to the individual of aquatic products, and the gill movement data can be recorded in real time by detecting the weak vibration or electromyogram signal of the skin or gills. The ultrasonic monitoring technology emits low-frequency sound waves to aquatic products and realizes non-contact monitoring based on the echo analysis of the periodic movement changes of the gills. Laser interferometry uses a high-precision laser interferometer to detect the tiny displacement of the gills of aquatic products and provides high-precision movement frequency data. These methods have their own advantages and disadvantages, and the combined use can provide more comprehensive and accurate gill movement data to meet the monitoring needs in different aquaculture environments.

[0082] Using the obtained dynamic frequency components of gill opening and closing, combined with the temporal curvature and non-linear characteristics of the gill movement signal, a gill movement frequency reference value is generated. By enhancing the characteristics of the second derivative of the signal and introducing a phase correction term, the final generated result can accurately reflect the physiological load level of aquatic products during the metabolic peak period. The generation expression of the gill movement frequency reference value is as follows:

[0083] ,

[0084] where G index is the finally generated gill movement frequency reference value, quantifying the physiological load level of aquatic products during the metabolic peak period, κ dyn is the dynamic curvature correction coefficient, describing the non-linear change characteristics of the gill opening and closing frequency, is the second derivative of the gill movement signal, revealing the acceleration and non-linear dynamic changes of the gill movement. α is the non-linear enhancement factor, used to amplify the signal characteristics in the high dynamic frequency region, usually obtained by fitting experimental data. φ is the phase of the gill movement signal, reflecting the periodic characteristics of the gill movement, correcting the stability of the characteristic calculation, and cos(φ) is used to adjust the symmetry of the signal characteristics to reduce external interference. β is the weight factor, used to balance the influence of temporal characteristics and frequency domain characteristics and optimize the calculation accuracy.

[0085] From the reference value of gill movement frequency, it can be seen that under the monitoring window, the larger the performance value of the reference value of gill movement frequency generated by feature engineering for the gill opening and closing frequency of the selected aquatic products, the higher the metabolic level of the aquatic products generally indicates. On the contrary, if the reference value of gill movement frequency is relatively low, it means that the metabolic activities of the aquatic products are relatively gentle. The gill is a key organ for aquatic organisms to carry out gas exchange and metabolic regulation, and its opening and closing frequency is directly affected by the physiological oxygen demand. In a high metabolic state, such as when activities such as feeding, digestion, and movement are enhanced, aquatic products need more oxygen to support energy metabolism. Therefore, the opening and closing frequency of the gills will increase accordingly, resulting in an increase in the performance value of the reference value of gill movement frequency. In a low metabolic state, such as during rest, in a low-temperature environment, or in a stress state, the energy consumption of aquatic products decreases, the demand for oxygen drops, and the opening and closing frequency of the gills decreases, resulting in a corresponding decrease in the performance value of the reference value of gill movement frequency. Therefore, by monitoring the changes in the reference value of gill movement frequency, the metabolic dynamics of aquatic products can be accurately evaluated, providing a scientific basis for aquaculture management.

[0086] When the production rate of aquatic product excrement is relatively high, it usually indicates that the metabolic activities of the aquatic products are at a peak period. Specifically, aquatic products will experience significant energy consumption during the high metabolic period, which is mainly reflected in increased feeding, frequent activities, and accelerated growth. These physiological activities require more oxygen and nutrients, and at the same time will accelerate the generation and excretion of waste. For example, an increase in food intake will lead to the active operation of the digestive system, thereby generating more ammonia nitrogen and other metabolic wastes. In addition, a high activity level will prompt aquatic products to excrete frequently to maintain the stability of the internal environment. The accumulation of these wastes in the water body is not only a direct manifestation of the metabolic peak, but also affects the water quality, increasing the ammonia nitrogen concentration and the content of other harmful substances in the water body. Therefore, by monitoring and analyzing the production rate of aquatic product excrement, it is possible to effectively judge whether its metabolic state is at a peak period, thereby providing an important reference basis for aquaculture management and ensuring the healthy growth of aquatic products and the good maintenance of water quality.

[0087] Under the monitoring window, the specific steps for generating the reference value of metabolic waste excretion by feature engineering for the production rate of the selected aquatic product excrement are as follows:

[0088] Within the monitoring window, first construct a time series trend model for the collected data of the production rate of aquatic product excrement, adopt a non-uniform smoothing filtering method, and combine a weight allocation strategy to emphasize the impact of recent data on the metabolic state. The constructed expression is:

[0089] ,

[0090] Where: is the smoothed sequence of the production rate of excrement, R w(i) represents the aquatic excretion rate collected within the monitoring window, δ is the weight attenuation factor, 0<δ<1, reflecting the importance of recent data, n is the total number of data points in the monitoring window, α' is the adjustment coefficient to ensure that the system adapts to different excretion rate changes, i represents the data sampling sequence in the monitoring window, that is, the index position of the current sampling point, max(R w ) represents the maximum value of all aquatic discharge rate data within the monitoring window, min(R w ) represents the minimum value of all aquatic discharge rate data within the monitoring window;

[0091] The purpose of this step is to eliminate random fluctuations in the data and enhance the ability to capture metabolic trends, making the trends after data processing more representative and providing accurate input for subsequent feature construction. After introducing exponential decay and amplitude adjustment, the system can more accurately identify metabolic peak periods and avoid misjudgments caused by data fluctuations.

[0092] After obtaining the smoothed excretion rate sequence, the metabolic waste excretion reference value is constructed. The nonlinear incremental analysis model is used to enhance the characteristics by accumulating the change amplitude of the excretion rate and combining the metabolic fluctuation amplitude index. The constructed expression is:

[0093] ,

[0094] Where: E wi is the reference value for metabolic waste excretion, is the value of the smoothed excretion production rate at the jth sampling point, γ is the nonlinear exponential coefficient, which amplifies the impact of drastic changes in metabolism, β' is the basic weight factor used to adjust the sensitivity, μ is the weight coefficient of the fluctuation amplitude, and Φ m The metabolic fluctuation index, which measures the degree of fluctuation in excretion rate, is defined as:

[0095] ,

[0096] in, is the value of the smoothed excrement production rate at the kth sampling point, is the value of the smoothed excrement production rate at the k-1th sampling point, η is the smoothing adjustment coefficient, which controls the impact of drastic changes;

[0097] The purpose of this step is to comprehensively analyze the intensity of changes and fluctuation characteristics of the excretion rate and extract the significant characteristics of the peak of aquatic metabolism. By combining the cumulative nonlinear changes and the amplitude of fluctuations, the metabolic waste excretion reference value can quantify the energy consumption level of aquatic products during the peak of metabolism, provide more sensitive and accurate metabolic characteristic indicators, and provide a scientific basis for subsequent dissolved oxygen regulation strategies.

[0098] From the reference values of metabolic waste excretion, it can be seen that under the monitoring window, the larger the performance value of the metabolic waste excretion reference value generated by feature engineering on the production rate of the screened aquatic excreta, the higher the metabolic level of the aquatic product usually indicates. On the contrary, it indicates a lower metabolic level of the aquatic product. The construction of the metabolic waste excretion reference value is based on the fluctuation situation and cumulative change range of the production rate of aquatic excreta, and can effectively reflect the energy consumption intensity of the aquatic product within the monitoring window. A higher metabolic waste excretion reference value means that the feeding, digestion and growth activities of the aquatic product are enhanced during this period, and the production rate of excreta is accelerated, reflecting vigorous metabolic activities, such as after feeding, under suitable temperature conditions or during the growth peak period. When the metabolic waste excretion reference value is low, it indicates that the feeding and activity levels of the aquatic product are reduced, and the metabolic process tends to be stable or enters a dormant state, such as at night or in a low-temperature environment. Therefore, the size of the metabolic waste excretion reference value directly quantifies the intensity of aquatic metabolism.

[0099] The extracted gill movement frequency reference value and metabolic waste excretion reference value are input into a pre-trained deep learning model. Based on the deep learning model, a dynamic evaluation factor for metabolic vitality is generated, and the metabolic situation of the aquatic product is intelligently predicted through the dynamic evaluation factor for metabolic vitality.

[0100] The pre-trained deep learning model refers to being trained and optimized based on a large amount of historical data and specific features through deep learning algorithms (such as long short-term memory network LSTM, convolutional neural network CNN or hybrid model) before actual application to have the ability to efficiently predict the metabolic state of aquatic products. Specifically, this model uses a large amount of historical data containing gill movement frequency reference values and metabolic waste excretion reference values during the training stage. These data cover the physiological characteristics and behavior patterns of aquatic products in different metabolic states. Through repeated iterative training, the model can learn and identify the subtle differences and complex correlations between the metabolic peak period and the normal metabolic period, so that when facing new real-time data, it can accurately predict the current metabolic state of the aquatic product. During the training process, the model will perform parameter optimization, cross-validation and performance evaluation to ensure its high accuracy and robustness under different environmental conditions and breeding stages. In addition, the pre-trained model usually undergoes specific regularization techniques and methods to prevent overfitting, so that it still shows good generalization ability when dealing with unseen data. This means that even if the breeding environment changes or the aquatic product variety is slightly different, the pre-trained deep learning model can still rely on its inherent learning ability to adapt to new situations and provide reliable prediction results.

[0101] In practical applications, the pre-trained deep learning model generates a comprehensive dynamic assessment factor for metabolic vitality by inputting the extracted reference values of gill movement frequency and metabolic waste excretion into the model, thereby making an intelligent prediction of the metabolic situation of aquatic products. This process involves multiple key steps: First, the system collects and preprocesses the raw data of gill movement frequency and metabolic waste excretion in real time, and generates the corresponding reference values of gill movement frequency and metabolic waste excretion through feature engineering. The reference values of gill movement frequency and metabolic waste excretion are used as input features of the model and are transmitted to the pre-trained deep learning model. The model analyzes the current input data according to the patterns and rules learned during the training stage to evaluate whether the aquatic product is in the metabolic peak period or the normal metabolic period. The generated dynamic assessment factor for metabolic vitality not only reflects the current metabolic state of the aquatic product but also can predict the metabolic trend in the future for a period of time, providing forward-looking decision-making support for aquaculture management. In addition, as the system continuously accumulates new data, the pre-trained deep learning model can also continuously optimize and improve its prediction ability through retraining or online learning to ensure high prediction performance under different aquaculture stages and environmental changes. This intelligent, data-driven management method significantly improves the refinement and automation levels of freshwater aquaculture, reduces aquaculture risks, and enhances economic benefits and the health level of aquatic products.

[0102] The deep learning model is not specifically limited here, and any deep learning model that can achieve the comprehensive analysis of the reference value G of gill movement frequency index and the reference value E of metabolic waste excretion wi to generate the dynamic assessment factor Meta for metabolic vitality activity is acceptable. To implement the technical solution of the present invention, the present invention provides a specific implementation method; the calculation formula for generating the dynamic assessment factor Meta for metabolic vitality activity is as follows:

[0103] ,

[0104] where ψ 1 , ψ 2 are respectively the preset proportionality coefficients of the reference value G of gill movement frequency index and the reference value E of metabolic waste excretion wi , and both ψ 1 , ψ 2 are greater than 0. The preset proportionality coefficients (ψ 1 and ψ 2 ) refer to when calculating the dynamic assessment factor for metabolic vitality, the reference value G of gill movement frequency index and the reference value E of metabolic waste excretion wiThe assigned weighted weights. The role of these preset proportional coefficients is to balance the relative contributions of the two indicators in the final dynamic assessment factor of metabolic vitality, so as to ensure that the calculation results can accurately reflect the actual situation of the aquaculture metabolic peak. The selection of the preset proportional coefficients is usually based on historical data analysis and professional experience to ensure the adaptability and accuracy of the model under different environmental and metabolic conditions. For example, if the gill movement frequency has a more significant impact on the metabolic peak, ψ 1 can be set to be larger, so as to assign it a higher weight in the calculation. On the contrary, if the change in the metabolic waste excretion rate can better reflect the metabolic activity, the weight of ψ 2 can be increased. Since both of these preset proportional coefficients are greater than zero, it ensures the rationality and stability of the formula calculation, making the evaluation results of the dynamic assessment factor of metabolic vitality more accurately adapt to the actual needs of different aquaculture environments.

[0105] It can be seen from the dynamic assessment factor of metabolic vitality that under the monitoring window, the larger the performance value of the gill movement frequency reference value generated by feature engineering on the gill opening and closing frequency of the selected aquaculture products, and the larger the performance value of the metabolic waste excretion reference value generated by feature engineering on the production rate of the excreta of the selected aquaculture products. That is, when the metabolic situation of aquaculture products is intelligently predicted through a pre-trained deep learning model under the monitoring window, the larger the performance value of the dynamic assessment factor of metabolic vitality generated, it indicates that the metabolic level of the aquaculture products is higher, and vice versa, it indicates that the metabolic level of the aquaculture products is lower.

[0106] Compare and analyze the dynamic assessment factor of metabolic vitality generated when the metabolic situation of aquaculture products is intelligently predicted through a pre-trained deep learning model under the monitoring window with the preset reference threshold of the dynamic assessment factor of metabolic vitality, and divide the results into "metabolic peak" or "normal metabolism". The specific steps are as follows:

[0107] If the dynamic assessment factor of metabolic vitality is greater than the reference threshold of the dynamic assessment factor of metabolic vitality, the metabolic level of the aquaculture products will be classified as "metabolic peak"; if the dynamic assessment factor of metabolic vitality is less than or equal to the reference threshold of the dynamic assessment factor of metabolic vitality, the metabolic level of the aquaculture products will be classified as "normal metabolism".

[0108] For aquaculture products in the "normal metabolism" state, continue to carry out intelligent control with the set initial dissolved oxygen threshold to maintain efficient energy utilization; for aquaculture products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use, improve the oxygen supply level, and start the aeration equipment in advance to ensure that the oxygen concentration in the water body is sufficient during the high oxygen demand period and avoid sudden hypoxia problems;

[0109] For aquatic products in the "normal metabolism" state, continuing with intelligent control using the set initial dissolved oxygen threshold means that in the aquaculture environment, when the deep learning model analyzes that the current metabolic state of the aquatic products is at a normal level, oxygenation management will be carried out according to the pre-set initial dissolved oxygen concentration threshold. This intelligent control strategy aims to keep the dissolved oxygen concentration in the water body stable within a reasonable range required for growth, while avoiding unnecessary oxygenation operations, reducing energy consumption, and improving the operating efficiency of the system. In the normal metabolic state, the oxygen demand of aquatic products is relatively stable. Continuing to use the set initial dissolved oxygen concentration threshold can ensure that their basic physiological needs are met without causing hypoxic stress or dissolved oxygen surplus problems. In addition, this intelligent control process combines real-time monitoring data to ensure accurate dissolved oxygen management while maintaining the stability of water quality and the state of cultured organisms, maximizing energy utilization efficiency, reducing aquaculture costs, and minimizing the impact on the environment while ensuring aquaculture benefits.

[0110] For aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use. The specific steps are as follows:

[0111] When the aquatic products are in the "metabolic peak" state, based on the dynamic evaluation factor of metabolic vitality, dynamically adjust the dissolved oxygen threshold to meet their oxygen demand during the high metabolic period. The dynamic adjustment expression of the dissolved oxygen threshold is:

[0112] ,

[0113] where DO adj is the dynamically adjusted dissolved oxygen threshold for actual oxygenation decision-making to ensure meeting the high metabolic demand. DO init is the initial dissolved oxygen threshold. Meta activity is the dynamic evaluation factor of metabolic vitality, representing the current metabolic intensity of the aquatic products. Meta ref is the reference threshold of the dynamic evaluation factor of metabolic vitality, representing the reference value in the normal metabolic state. α x is the adjustment coefficient, controlling the sensitivity of the dissolved oxygen threshold adjustment and determining the weight of the adjustment amplitude. β x is the exponential weight, adjusting the non-linearity of the influence of the dynamic evaluation factor of metabolic vitality. Usually, the value range is 1 < β x < 3;

[0114] The core of this step is to use the dynamic evaluation factor of metabolic vitality generated by the deep learning model to real-time evaluate the current metabolic demand of the aquatic products. When the dynamic evaluation factor of metabolic vitality is greater than the reference threshold of the dynamic evaluation factor of metabolic vitality, the system will dynamically increase the dissolved oxygen threshold to cope with the higher oxygen consumption demand. The adjustment amplitude is determined by the adjustment coefficient α x and the exponential weight β xControl to ensure sufficient dissolved oxygen supply while avoiding unnecessary energy consumption.

[0115] By dynamically increasing the dissolved oxygen threshold during actual use, it ensures that the sudden increase in oxygen demand during the peak of aquatic metabolism is met in a timely manner, thus maintaining the dissolved oxygen concentration in the water body within the optimal range and avoiding serious problems such as physiological stress, reduced feeding, growth retardation, and even large-scale death caused by insufficient dissolved oxygen. Through intelligent dynamic regulation, the system can identify the metabolic peak in advance and start the aeration equipment in a timely manner, avoiding the lag of aeration response and ensuring the continuity and stability of dissolved oxygen supply. At the same time, this measure helps to optimize energy utilization and prevent water quality fluctuations caused by excessive aeration, such as abnormal pH value and ammonia nitrogen accumulation, thus ensuring the health and stability of the aquaculture environment, improving the aquaculture efficiency and economic benefits.

[0116] Through the above-mentioned Internet of Things-based freshwater aquaculture water quality monitoring method, it can significantly improve the intelligent management level of the aquaculture environment, ensure timely and accurate regulation of the dissolved oxygen concentration during the peak of aquatic metabolism, and thus effectively prevent hypoxia problems. This solution deploys high-precision dissolved oxygen sensors and a variety of metabolic data sensors, combines the intelligent prediction of deep learning models, and monitors and analyzes the metabolic status of aquatic products in real time to achieve automatic control of dynamically adjusting the dissolved oxygen threshold and aeration equipment. Specifically, when the system predicts that the metabolic peak is about to come, it automatically increases the dissolved oxygen threshold and starts the aeration equipment in advance to ensure sufficient oxygen content in the water body and avoid the decline in the vitality of aquatic products, reduced feeding, and even large-scale death caused by insufficient dissolved oxygen. This not only guarantees the healthy growth of aquatic organisms, improves the aquaculture efficiency and output, but also optimizes energy utilization, reduces the ineffective operation of aeration equipment and related operating costs. At the same time, intelligent water quality management reduces human intervention and management difficulties, improves the stability and sustainability of the aquaculture process, and finally achieves a win-win situation of economic and ecological benefits. This comprehensive solution provides efficient and reliable technical support for modern freshwater aquaculture, significantly reduces the aquaculture risk, and promotes the intelligent and modern development of the aquaculture industry.

[0117] The present invention provides a Figure 2 Internet of Things-based freshwater aquaculture water quality monitoring system as shown, including an initial dissolved oxygen threshold setting module, a dissolved oxygen monitoring module, an aeration control module, a metabolic data collection module, a metabolic data preprocessing module, a key feature extraction module, an intelligent metabolic prediction module, and an intelligent dissolved oxygen regulation module;

[0118] The initial dissolved oxygen threshold setting module first determines an initial dissolved oxygen concentration threshold based on past aquaculture data and experience;

[0119] Dissolved oxygen monitoring module, deploying a high-precision dissolved oxygen sensor network in the aquaculture environment to ensure that the dissolved oxygen sensors can cover the key areas of the entire water body, and transmitting real-time dissolved oxygen data to the central control system through wireless communication technology;

[0120] Oxygenation control module, when the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water body is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues an instruction to start the oxygenation equipment to increase the oxygen content in the water body, quickly raise the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold;

[0121] Metabolic data acquisition module, deploying various sensors in the aquaculture environment to collect the metabolic data of aquatic products in real time, and transmitting the metabolic data to the central processing system through Internet of Things devices;

[0122] Metabolic data preprocessing module, preprocessing the real-time obtained metabolic data to ensure the consistency and integrity of the data;

[0123] Key feature extraction module, using feature extraction technology to extract key indicators that can reflect the metabolic peak of aquatic products from the preprocessed metabolic data, and forming a core data set for model analysis;

[0124] Intelligent metabolic prediction module, inputting the extracted key metabolic data into a pre-trained deep learning model to intelligently predict the metabolic situation of aquatic products, predict the metabolic state of aquatic products in a future period of time, and classify the results into "metabolic peak" or "normal metabolism";

[0125] Intelligent dissolved oxygen regulation module, for aquatic products in the "normal metabolism" state, continue to perform intelligent control with the set initial dissolved oxygen threshold to maintain efficient energy utilization; for aquatic products in the "metabolic peak" state, dynamically increase the actual dissolved oxygen threshold during use, improve the oxygen supply level, and start the oxygenation equipment in advance to ensure sufficient oxygen concentration in the water body during the high oxygen demand period;

[0126] A method for monitoring the water quality of freshwater aquaculture based on the Internet of Things provided by the embodiments of the present invention is implemented through the above-mentioned system for monitoring the water quality of freshwater aquaculture based on the Internet of Things. The specific methods and processes of the system for monitoring the water quality of freshwater aquaculture based on the Internet of Things are detailed in the embodiments of the above-mentioned method for monitoring the water quality of freshwater aquaculture based on the Internet of Things, and will not be elaborated here.

[0127] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0128] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

[0129] Only some exemplary embodiments of the present invention have been described above by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

Claims

1. A method for monitoring water quality of freshwater aquaculture based on the Internet of Things, characterized in that: The following steps are involved: First, based on past aquaculture data and experience, determine an initial dissolved oxygen concentration threshold; Deploy a high-precision dissolved oxygen sensor network in the aquaculture environment to ensure that the dissolved oxygen sensors can cover key areas of the entire water body and transmit real-time dissolved oxygen data to the central control system through wireless communication technology; When the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues a command to start the oxygenation equipment to increase the oxygen content in the water, rapidly increase the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold; Deploy a variety of sensors in the aquaculture environment to collect metabolic data of aquatic products in real time, and transmit the metabolic data to the central processing system through IoT devices; Preprocess the metabolic data acquired in real time to ensure the consistency and integrity of the data; From the pre-processed metabolic data, feature extraction technology is used to extract key indicators that can reflect the peak of aquatic metabolism, forming a core data set for model analysis; The extracted key metabolic data are input into the pre-trained deep learning model to make intelligent predictions on the metabolic status of aquatic products, predict the metabolic status of aquatic products in the future, and classify the results into "metabolic peak" or "normal metabolism"; For aquatic products in the "normal metabolism" state, continue to perform intelligent control with the set initial dissolved oxygen threshold to maintain efficient energy utilization; for aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use, improve the oxygen supply level, start the oxygen enrichment equipment in advance, and ensure that the oxygen concentration in the water is sufficient during the high oxygen demand period.

2. A freshwater aquaculture water quality monitoring method based on the Internet of Things according to claim 1, characterized in that: The initial dissolved oxygen concentration threshold refers to the minimum dissolved oxygen concentration standard pre-set based on historical aquaculture data, the physiological needs of aquaculture species and environmental conditions. When the dissolved oxygen concentration in the water body is lower than the initial dissolved oxygen concentration threshold, the oxygen enrichment equipment will be automatically triggered to maintain the dissolved oxygen demand of the aquaculture environment and ensure the normal growth and survival of aquatic organisms.

3. A freshwater aquaculture water quality monitoring method based on the Internet of Things according to claim 1, characterized in that: In the pre-processed metabolic data, feature extraction technology is used to extract key indicators that can reflect the peak of aquatic metabolism. The specific steps are as follows: The gill opening and closing frequency of aquatic products and the production rate of aquatic excrement are screened out from the preprocessed metabolic data. Under the monitoring window, feature engineering is performed on the screened gill opening and closing frequency and the production rate of aquatic excrement to generate reference values ​​for gill movement frequency and metabolic waste excretion, respectively. The gill movement frequency reference values ​​and metabolic waste excretion reference values ​​are used to establish a core data set to quantify the physiological load and energy consumption level of aquatic products at the peak of metabolism, so as to accurately identify the behavioral characteristics and metabolic intensity under high metabolic state.

4. A method for monitoring water quality of freshwater aquaculture based on the Internet of Things according to claim 3, characterized in that: The extracted gill movement frequency reference value and metabolic waste excretion reference value are input into a pre-trained deep learning model, and a metabolic activity dynamic evaluation factor is generated based on the deep learning model. The metabolic status of aquatic products is intelligently predicted through the metabolic activity dynamic evaluation factor.

5. A method for monitoring water quality of freshwater aquaculture based on the Internet of Things according to claim 4, characterized in that: The metabolic activity dynamic evaluation factor generated by the pre-trained deep learning model in the monitoring window when intelligently predicting the metabolic status of aquatic products is compared with the pre-set reference threshold of the metabolic activity dynamic evaluation factor, and the results are divided into "metabolic peak" or "normal metabolism". The specific steps are as follows: If the metabolic activity dynamic assessment factor is greater than the metabolic activity dynamic assessment factor reference threshold, the metabolic level of the aquatic product is classified as "metabolic peak"; if the metabolic activity dynamic assessment factor is less than or equal to the metabolic activity dynamic assessment factor reference threshold, the metabolic level of the aquatic product is classified as "normal metabolism".

6. A method for monitoring water quality of freshwater aquaculture based on the Internet of Things according to claim 3, characterized in that: In the monitoring window, the specific steps for performing feature engineering on the gill opening and closing frequency of the selected aquatic products to generate the gill movement frequency reference value are as follows: The core frequency characteristics of gill movement are extracted by decomposing the original signal of gill opening and closing frequency in the frequency domain and calculating the weighted sum of its dynamic components by integration. The extracted expression is: ， Among them, F osc is the initially extracted dynamic component of the gill opening and closing frequency, which is used to quantify the frequency change trend of gill movement within the monitoring window. osc is the original signal time series of gill movement, reflecting the time series of gill opening and closing movements, ω is the frequency component of the movement, describing the frequency domain characteristics of the gill opening and closing frequency within the monitoring window, λ is the exponential attenuation coefficient, used to reduce the influence of high-frequency noise while retaining the key information of low-frequency characteristic signals, and τ is the length of the monitoring window, defining the time range of signal processing; The obtained gill opening and closing dynamic frequency components are combined with the temporal curvature and nonlinear characteristics of the gill movement signal to generate the gill movement frequency reference value. By enhancing the characteristics of the second-order derivative of the signal and introducing the phase correction term, the final result can accurately reflect the physiological load level of aquatic products during the peak metabolic period. The generation expression of the gill movement frequency reference value is as follows: ， Among them, G index The gill movement frequency reference value is finally generated to quantify the physiological load level of aquatic products during the peak metabolic period, κ dyn is the dynamic curvature correction coefficient, which describes the nonlinear change characteristics of the gill opening and closing frequency. is the second-order derivative of the gill movement signal, which reveals the acceleration and nonlinear dynamic changes of the gill movement. α is the nonlinear enhancement factor, which is used to amplify the signal characteristics in the high dynamic frequency area. φ is the phase of the gill movement signal, which reflects the periodic characteristics of the gill movement and corrects the stability of the feature calculation. cos(φ) is used to adjust the symmetry of the signal characteristics and reduce external interference. β is the weight factor, which is used to balance the influence of the timing characteristics and the frequency domain characteristics and optimize the accuracy of the calculation.

7. A method for monitoring water quality of freshwater aquaculture based on the Internet of Things according to claim 3, characterized in that: Under the monitoring window, the specific steps for characteristic engineering of the production rate of screened aquatic excrement to generate reference values ​​for metabolic waste excretion are as follows: In the monitoring window, we first construct a time series trend model for the collected aquatic excrement generation rate data. We use a non-uniform smoothing filtering method and a weight distribution strategy to emphasize the impact of recent data on metabolic status. The constructed expression is: ， in: is the smoothed excrement production rate series, R w (i) represents the aquatic excretion rate collected within the monitoring window, δ is the weight attenuation factor, 0<δ<1, reflecting the importance of recent data, n is the total number of data points in the monitoring window, α' is the adjustment coefficient to ensure that the system adapts to different excretion rate changes, i represents the data sampling sequence in the monitoring window, that is, the index position of the current sampling point, max(R w ) represents the maximum value of all aquatic discharge rate data within the monitoring window, min(R w ) represents the minimum value of all aquatic discharge rate data within the monitoring window; After obtaining the smoothed excretion rate sequence, the metabolic waste excretion reference value is constructed. The nonlinear incremental analysis model is used to enhance the characteristics by accumulating the change amplitude of the excretion rate and combining the metabolic fluctuation amplitude index. The constructed expression is: ， Where: E wi is the reference value for metabolic waste excretion, is the value of the smoothed excretion production rate at the jth sampling point, γ is the nonlinear exponential coefficient, which amplifies the impact of drastic changes in metabolism, β' is the basic weight factor used to adjust the sensitivity, μ is the weight coefficient of the fluctuation amplitude, and φ m The metabolic fluctuation index, which measures the degree of fluctuation in excretion rate, is defined as: ， in, is the value of the smoothed excrement production rate at the kth sampling point, is the value of the smoothed excrement production rate at the k-1th sampling point, and η is the smoothing adjustment coefficient, which controls the impact of drastic changes.

8. The method for monitoring water quality of freshwater aquaculture based on the Internet of Things according to claim 5, characterized in that: For aquatic products in the "metabolic peak" state, dynamically increase the dissolved oxygen threshold during actual use. The specific steps are as follows: When aquatic products are in the "metabolic peak" state, the dissolved oxygen threshold is dynamically adjusted based on the metabolic activity dynamic evaluation factor to meet their oxygen demand during the high metabolic period. The dynamic adjustment expression of the dissolved oxygen threshold is: ， Among them, DO adj The dynamically adjusted dissolved oxygen threshold is used for actual oxygenation decisions to ensure that high metabolic demands are met. init is the initial dissolved oxygen threshold, Meta activity Meta is a dynamic evaluation factor of metabolic activity, indicating the current metabolic intensity of aquatic products. ref is the reference threshold of the metabolic activity dynamic evaluation factor, indicating the reference value under normal metabolic state, α x is the adjustment coefficient, which controls the sensitivity of the dissolved oxygen threshold adjustment and determines the weight of the adjustment range. x is the index weight, which adjusts the nonlinear degree of the influence of the dynamic evaluation factor of metabolic activity, and its value range is 1<β x <3.

9. A freshwater aquaculture water quality monitoring system based on the Internet of Things, used to implement the freshwater aquaculture water quality monitoring method based on the Internet of Things as described in any one of claims 1 to 8, characterized in that: It includes an initial dissolved oxygen threshold setting module, a dissolved oxygen monitoring module, an oxygenation control module, a metabolic data acquisition module, a metabolic data preprocessing module, a key feature extraction module, an intelligent metabolic prediction module, and an intelligent dissolved oxygen regulation module; The initial dissolved oxygen threshold setting module first determines an initial dissolved oxygen concentration threshold based on past aquaculture data and experience; Dissolved oxygen monitoring module, deploying a high-precision dissolved oxygen sensor network in the aquaculture environment to ensure that the dissolved oxygen sensors can cover key areas of the entire water body and transmit real-time dissolved oxygen data to the central control system through wireless communication technology; Oxygenation control module: when the dissolved oxygen sensor detects that the dissolved oxygen concentration in the water is lower than the preset initial dissolved oxygen concentration threshold, the central control unit immediately issues a command to start the oxygenation equipment to increase the oxygen content in the water, rapidly increase the dissolved oxygen level, and automatically shut down after reaching or exceeding the set initial dissolved oxygen concentration threshold; Metabolic data acquisition module, deploys multiple sensors in the aquaculture environment to collect metabolic data of aquatic products in real time, and transmits the metabolic data to the central processing system through IoT devices; Metabolic data preprocessing module preprocesses the metabolic data acquired in real time to ensure the consistency and integrity of the data; The key feature extraction module uses feature extraction technology to extract key indicators that can reflect the peak of aquatic metabolism from the preprocessed metabolic data to form a core data set for model analysis; The intelligent metabolism prediction module inputs the extracted key metabolic data into a pre-trained deep learning model to make intelligent predictions on the metabolic status of aquatic products, predict the metabolic status of aquatic products in the future, and classify the results into "metabolic peak" or "normal metabolism"; The intelligent dissolved oxygen control module continues to perform intelligent control based on the set initial dissolved oxygen threshold for aquatic products in the "normal metabolism" state to maintain efficient energy utilization; for aquatic products in the "metabolic peak" state, it dynamically increases the dissolved oxygen threshold during actual use, improves the oxygen supply level, and starts the oxygen enrichment equipment in advance to ensure sufficient oxygen concentration in the water during the high oxygen demand period.

Images