Method for detecting, regulating and controlling ammonia nitrogen content of aquaculture pond in real time

Abstract

The invention relates to the technical field of aquaculture environment monitoring and regulation, in particular to an aquaculture pond ammonia nitrogen content real-time detection and regulation method, which comprises the steps of main / standby ammonia nitrogen sensor collaborative acquisition, temperature compensation correction and data reliability verification, key feature extraction based on historical water quality data and day and night temperature difference, a long-short-term memory network model is adopted to predict the ammonia nitrogen concentration trend, and a collaborative regulation and control instruction for regulating the glacier water supply amount, controlling the operation mode of an aerator and regulating the feeding strategy is automatically generated in combination with the standard exceeding grade. According to the invention, by establishing a prediction-decision-execution-feedback closed loop mechanism, high-precision prediction and dynamic response regulation and control of the ammonia nitrogen concentration in the cold water culture environment are realized, and the stability of culture water quality and the intelligent level of the system are improved.

Description

Technical Field

[0001] This invention relates to the field of aquaculture environment monitoring and control technology, and in particular to a method for real-time detection and control of ammonia nitrogen content in aquaculture ponds. Background Technology

[0002] In cold-water fish farming, especially in the factory or pond aquaculture of high-value species such as salmon, ammonia nitrogen concentration is one of the core water quality parameters reflecting aquatic environmental quality and fish health. Excessive ammonia nitrogen not only inhibits fish metabolism and damages gill tissue but can even lead to mass mortality. Therefore, real-time detection and dynamic control of ammonia nitrogen concentration are crucial for ensuring the stability and economic benefits of aquaculture. In cold-water environments, problems such as decreased ammonia nitrogen conversion efficiency, sluggish sensor response, and weakened microbial activity are more pronounced. There is an urgent need to construct a high-precision detection and intelligent control system adapted to the characteristics of low-temperature water bodies to achieve closed-loop control throughout the entire process from "detection to prediction to response."

[0003] However, the following technical problems are common in the existing technical solutions: (1) The single sensor working mode is easily affected by electrode aging and low temperature interference, lacks backup mechanism and real-time verification capability, resulting in insufficient accuracy of ammonia nitrogen concentration data; (2) Most detection systems do not consider the impact of day and night temperature difference on data fluctuation, and lack prediction models for cold water environment, making it impossible to effectively predict the trend of ammonia nitrogen concentration; (3) Water quality control usually relies on manual experience judgment, with delayed response and coarse control parameters, making it difficult to form a refined control strategy that matches the ammonia nitrogen risk level. Summary of the Invention

[0004] This invention provides a method for real-time detection and control of ammonia nitrogen content in aquaculture ponds, which is an integrated solution that can adapt to cold water environments and has high-precision detection, intelligent prediction and multi-channel control capabilities.

[0005] A method for real-time detection and control of ammonia nitrogen content in aquaculture ponds includes the following steps: S1: The main ammonia nitrogen sensor and the backup ammonia nitrogen sensor deployed in the cold water aquaculture pond synchronously collect the raw signal of the ammonia nitrogen concentration at the current moment and the backup ammonia nitrogen concentration data. The backup ammonia nitrogen sensor is in a low-temperature dormant state during non-working periods. The temperature, pH value and dissolved oxygen data of the pond water are collected synchronously, and the sensor readings are temperature compensated and corrected according to the characteristics of low-temperature water to generate real-time ammonia nitrogen concentration data. S2: Periodically wake up the standby ammonia nitrogen sensor in low-temperature dormancy state, obtain its standby ammonia nitrogen concentration data and convert it into calibration ammonia nitrogen concentration data; compare the real-time ammonia nitrogen concentration data with the calibration ammonia nitrogen concentration data under low-temperature conditions. If the difference is within the allowable error range, output reliable real-time ammonia nitrogen concentration data; if it exceeds the error range, start the sensor self-cleaning program for low-temperature environments and recalibrate. S3: Based on the environmental characteristics of large diurnal temperature differences, combined with historical reliable real-time ammonia nitrogen concentration data and temperature and pH value change patterns, a prediction model adapted to low-temperature water environment is used to predict the ammonia nitrogen concentration in a specific future period. S4: Compare the predicted ammonia nitrogen concentration with a specific safety threshold for salmon farming. When the predicted value is close to or exceeds the safety threshold, generate and execute a coordinated control instruction suitable for cold-water aquaculture, including one or more combinations of adjusting the glacial water supply, controlling the aerator operation mode, and adjusting the feeding strategy.

[0006] Optionally, S1 includes: S11: The main ammonia nitrogen sensor continuously collects the raw signal of ammonia nitrogen concentration in the aquaculture pond water, while controlling the backup ammonia nitrogen sensor to collect backup ammonia nitrogen concentration data during the synchronous collection period when it is awakened, and to keep it in a low-temperature dormant state during the asynchronous collection period. S12: Temperature data, pH data and dissolved oxygen data are collected synchronously through temperature sensor, pH sensor and dissolved oxygen sensor. The pH data and dissolved oxygen data are used for subsequent prediction of ammonia nitrogen change trend and intelligent collaborative regulation. S13: Using the built-in temperature compensation algorithm, the temperature data is used to perform nonlinear compensation correction on the original ammonia nitrogen concentration signal collected by the main ammonia nitrogen sensor for the characteristics of low-temperature water bodies, and the corrected real-time ammonia nitrogen concentration data is generated.

[0007] Optionally, in S11, controlling the backup ammonia nitrogen sensor to collect backup ammonia nitrogen concentration data during the synchronized acquisition period after being awakened specifically involves sending a synchronized trigger signal to the backup ammonia nitrogen sensor through a preset data processing unit, so that the sampling timing of the backup ammonia nitrogen sensor is completely consistent with the sampling timing of the main ammonia nitrogen sensor.

[0008] Optionally, in S11, the sensor is placed in a low-temperature sleep state by: the data processing unit cutting off the power supply to the backup ammonia nitrogen sensor except for the wake-up receiving circuit, and covering its sensing probe with a mechanical protective cover to isolate it from the water.

[0009] Optionally, S2 includes: S21: The data processing unit sends a wake-up command according to a preset cycle to wake up the standby ammonia nitrogen sensor in low-temperature dormancy state, and controls it to perform synchronous signal acquisition with the main ammonia nitrogen sensor to obtain the standby ammonia nitrogen concentration data collected by the standby ammonia nitrogen sensor. S22: The data processing unit performs signal conversion and temperature compensation correction on the backup ammonia nitrogen concentration data to generate verification ammonia nitrogen concentration data for comparison. S23: The data processing unit compares the real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor with the calibration ammonia nitrogen concentration data in real time, and calculates the concentration difference between the two. S24: If the concentration difference is within the preset allowable error range, the main ammonia nitrogen sensor data is determined to be reliable, and the real-time ammonia nitrogen concentration data is output as reliable real-time ammonia nitrogen concentration data; if the concentration difference exceeds the preset allowable error range, the sensor self-cleaning procedure and recalibration process for low-temperature environments are triggered.

[0010] Optionally, the preset period in S21 is dynamically adjusted according to the eutrophication level of the water body. Specifically, the data processing unit calculates the rate of change of ammonia nitrogen concentration based on historical and reliable real-time data, and automatically shortens the preset period when the rate of change exceeds a threshold.

[0011] Optionally, the real-time deviation comparison in S23 adopts the sliding window mean comparison method, specifically: the average value of multiple real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor in the most recent time window is taken, and the concentration difference is calculated with the single calibration ammonia nitrogen concentration data generated by the backup ammonia nitrogen sensor.

[0012] Optionally, S3 includes: S31: Construct a time series dataset containing continuous and reliable real-time ammonia nitrogen concentration data, temperature data, pH value data, and dissolved oxygen data within a historical period, and preprocess the time series dataset to eliminate outliers and noise interference; S32: Based on the preprocessed time series dataset, extract key input features to characterize the environmental characteristics of cold-water aquaculture ponds. The key input features include the changing trend of reliable real-time ammonia nitrogen concentration data, the diurnal temperature difference characteristics in temperature data, the changing pattern of pH data, and the correlation fluctuation pattern of dissolved oxygen data. S33: Input the key input features into a pre-trained prediction model adapted to low-temperature water environment. The prediction model is based on a long short-term memory network architecture and trained using high-latitude cold water aquaculture data. The model calculates and generates a predicted sequence of ammonia nitrogen concentration for a specific future period. S34: The ammonia nitrogen concentration prediction sequence is smoothed and optimized, and its peak value is extracted as the final ammonia nitrogen concentration prediction value. The ammonia nitrogen concentration prediction value is then output for subsequent intelligent collaborative regulation and control decisions.

[0013] Optionally, S4 includes: S41: Compare and analyze the predicted ammonia nitrogen concentration with the pre-stored specific safety threshold for salmon farming, and determine the exceedance level of the predicted ammonia nitrogen concentration based on the exceedance magnitude and the rate of increase. S42: Based on the level of exceeding the standard, a corresponding set of coordinated control instructions is generated from the preset control strategy library. The set of coordinated control instructions includes specific parameters for adjusting the glacier water supply, parameters for controlling the aerator operation mode, and parameters for adjusting the feeding strategy. S43: Send the set of coordinated control instructions to the corresponding execution devices, specifically including sending an opening instruction to the glacier water inlet valve controller to adjust the glacier water supply, sending an start / stop and power instruction to the aerator controller to control the aerator operation mode, and sending an instruction to the feeder controller to adjust the feeding amount and interval of the feeding strategy. S44: After each executing device receives and executes the corresponding instruction, it continuously monitors the changes in reliable real-time ammonia nitrogen concentration data in the next round, verifies the control effect, and dynamically adjusts the subsequent set of coordinated control instructions.

[0014] The beneficial effects of this invention are: This invention, by constructing a highly reliable detection mechanism that enables the main ammonia nitrogen sensor and the backup ammonia nitrogen sensor to work together, and combining low-temperature dormancy control, synchronous sampling triggering, mechanical protective cover isolation and temperature compensation correction algorithm, achieves high-precision, low-drift, and long-term stable monitoring of ammonia nitrogen concentration in cold-water aquaculture environment. This significantly improves the accuracy and stability of real-time ammonia nitrogen data and provides a reliable data foundation for subsequent judgment and control.

[0015] This invention, based on data reliability assurance, introduces a low-temperature environment prediction model based on long short-term memory networks. Combining multi-variable time-series characteristics such as diurnal temperature difference, pH fluctuation, and dissolved oxygen linkage, it dynamically generates future ammonia nitrogen concentration prediction values. The trend smoothing and credibility determination are performed by exponential weighted moving average method, realizing a functional leap from passive monitoring to proactive early warning, and significantly improving the ability to predict water quality changes.

[0016] This invention further establishes a multi-dimensional collaborative regulation system for the needs of cold-water fish farming. Based on the predicted level of exceeding the standard, it drives different intensities of glacial water replenishment, aerator power adjustment and feeding strategy adjustment. Combined with the slope reverse judgment mechanism, it dynamically closes the loop to verify the regulation effect, and achieves the synergistic goals of precise control, energy-saving operation and rapid water quality recovery. This improves the intelligent management level and farming safety assurance capability of the cold-water fish farming system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this 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 for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the S2 process in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] It should be noted that the use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0021] Generally, terms can be understood at least partly from their use in context. For example, depending at least partly on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood not necessarily to convey an exclusive set of factors, but rather, alternatively, depending at least partly on the context, to allow for the presence of other factors that are not necessarily explicitly described.

[0022] like Figures 1-2 As shown, a method for real-time detection and control of ammonia nitrogen content in aquaculture ponds includes the following steps: S1: The main ammonia nitrogen sensor and the backup ammonia nitrogen sensor deployed in the cold-water aquaculture pond synchronously collect the raw signal of the current ammonia nitrogen concentration and the backup ammonia nitrogen concentration data. The backup ammonia nitrogen sensor is in a low-temperature dormant state during non-operating periods. Simultaneously, the temperature, pH value, and dissolved oxygen data of the pond water are collected, and the readings of the main ammonia nitrogen sensor are temperature compensated and corrected for the characteristics of the low-temperature water to generate real-time ammonia nitrogen concentration data, specifically: S11: The main ammonia nitrogen sensor is deployed at a representative water layer in the middle of the cold-water aquaculture pond. It continuously collects raw ammonia nitrogen concentration signals at a preset sampling period and transmits the collected raw ammonia nitrogen concentration signals to the data processing unit in real time. During asynchronous sampling periods, the backup ammonia nitrogen sensor is in a low-temperature dormant state. In this state, the preset data processing unit cuts off all power supply channels to the backup ammonia nitrogen sensor except for the wake-up receiving circuit. Simultaneously, it controls the mechanical protective cover to descend to the probe position, completely isolating the sensor probe from the pond water, forming a physically isolated static dormant environment to reduce biological adhesion to the sensor probe surface and electrode aging under low-temperature conditions. Upon reaching the preset synchronous sampling period, the data processing unit sends a synchronization trigger signal to the backup ammonia nitrogen sensor, simultaneously activating its power supply system. The mechanical protective cover automatically opens, exposing its sampling device to the water. The backup ammonia nitrogen sensor responds to the synchronization trigger signal, completing the acquisition of backup ammonia nitrogen concentration data within the set sampling time window (the backup ammonia nitrogen concentration data is used for reliability verification in subsequent step S2), and maintaining complete consistency with the sampling sequence of the main ammonia nitrogen sensor. After synchronous data acquisition is completed, the data processing unit enters sleep management mode again, power off the ammonia nitrogen sensor for backup, and resumes the mechanical protective cover closing action.

[0023] S12: The temperature sensor, pH sensor, and dissolved oxygen sensor are linked and controlled in the data acquisition module via a parallel interface structure. While the main ammonia nitrogen sensor acquires the raw ammonia nitrogen concentration signal, it simultaneously acquires temperature, pH, and dissolved oxygen data. Temperature data is obtained through a water temperature acquisition electrode with a sampling accuracy better than ±0.1℃, used for subsequent temperature compensation and correction calculations. pH data is obtained using the glass electrode method, with sampling intervals consistent with the raw ammonia nitrogen concentration signal. Its internal circuitry uses components with extremely low low-temperature drift coefficients, and a temperature-based compensation algorithm is integrated into the firmware to enhance its stability in cold water environments. Dissolved oxygen data is acquired through a fluorescence-based dissolved oxygen sensor, equipped with an optical probe and temperature control unit to ensure that the acquired data accurately reflects the actual dissolved oxygen level. After preliminary filtering and time-series alignment, the pH and dissolved oxygen data are used as input feature parameters for subsequent ammonia nitrogen change trend prediction and collaborative control models, and stored in the real-time data buffer for use by S3 and S4.

[0024] S13: The raw ammonia nitrogen concentration signal acquired by the main ammonia nitrogen sensor is nonlinearly compensated and corrected for the characteristics of low-temperature water using a temperature compensation algorithm. The temperature compensation algorithm employs a piecewise linear compensation algorithm. Based on the specific temperature range of the temperature data (e.g., 0-2℃, 2-4℃, 4-6℃, etc.), it calls the compensation coefficients from a pre-set compensation coefficient lookup table corresponding to that range and applies them to the raw ammonia nitrogen concentration signal correction process. The compensation coefficients are obtained from the laboratory low-temperature calibration curve. After piecewise coefficient correction, the raw ammonia nitrogen concentration signal generates corrected real-time ammonia nitrogen concentration data. This corrected real-time ammonia nitrogen concentration data serves as the basic input data for the ammonia nitrogen concentration prediction model and subsequent deviation comparison, and is simultaneously uploaded to a host computer or edge processing platform for visual monitoring and remote data analysis.

[0025] The temperature compensation algorithm calculation formula is as follows: ; in, This represents the corrected real-time hydrogen and nitrogen concentration data. This represents the raw hydrogen and nitrogen concentration signal collected by the main oxygen and nitrogen sensor. This indicates the current water temperature collected by the temperature sensor. This indicates the reference temperature value, set to 15°C. This represents the temperature compensation coefficient, which is a piecewise function of temperature and is obtained by looking up a table. The compensation coefficient varies in different temperature ranges. For example: when At °C, ; when At °C, ; when At °C, .

[0026] compensation coefficient The value is derived from laboratory calibration data under cold water conditions, obtained by statistical regression fitting of the ammonia nitrogen sensor measurement error at multiple temperature points.

[0027] Automatically select the corresponding temperature range based on the current temperature data. Perform the correction operation. The output after correction is complete. This refers to the effective real-time ammonia nitrogen concentration data under the current operating conditions, which serves as the basic input data for the subsequent ammonia nitrogen concentration prediction model and deviation comparison module. At the same time, it is uploaded to the host computer or edge processing platform for real-time display and remote monitoring of the aquaculture system.

[0028] S2: Periodically wake up the standby ammonia nitrogen sensor in low-temperature dormancy state, acquire its standby ammonia nitrogen concentration data and convert it into calibration ammonia nitrogen concentration data; compare the real-time ammonia nitrogen concentration data with the calibration ammonia nitrogen concentration data under low-temperature conditions. If the difference is within the allowable error range, output reliable real-time ammonia nitrogen concentration data; if it exceeds the error range, start the sensor self-cleaning program for low-temperature environments and recalibrate, specifically as follows: S21: The preset data processing unit sends a wake-up command according to a preset cycle to wake up the standby ammonia nitrogen sensor in low-temperature dormancy state, and controls it to perform synchronous signal acquisition with the main ammonia nitrogen sensor to obtain the standby ammonia nitrogen concentration data collected by the standby ammonia nitrogen sensor.

[0029] The preset cycle is dynamically adjusted by the data processing unit based on the eutrophication level of the water body. Specifically, the data processing unit analyzes the rate of change of historically reliable real-time ammonia nitrogen concentration data. When the rate of change of ammonia nitrogen concentration per unit time exceeds the system's preset dynamic threshold (e.g., 0.2 mg / L / h), the wake-up cycle is automatically shortened from the original 30 minutes to 15 minutes. If the rate of change is within a stable range, the original cycle is maintained, ensuring that while maintaining monitoring accuracy, the energy consumption and usage frequency of the backup ammonia nitrogen sensor are reduced. After the backup ammonia nitrogen sensor is woken up, it performs a data acquisition with the same timestamp as the main ammonia nitrogen sensor through a synchronous triggering mechanism, ensuring that the output data of both sensors are consistent under the same water body conditions.

[0030] S22: The data processing unit performs signal conversion and temperature compensation correction on the backup ammonia nitrogen concentration data to generate verification ammonia nitrogen concentration data for comparison. The signal conversion process includes analog-to-digital conversion of analog signals, numerical normalization of sampled data, and adjustment for consistency with the output format of the main ammonia nitrogen sensor, ensuring that the backup ammonia nitrogen concentration data has a unified data structure and dimensional standard. The temperature compensation correction process uses the same piecewise linear compensation algorithm as the temperature compensation correction algorithm used in S13 for the original ammonia nitrogen concentration signal acquired by the main ammonia nitrogen sensor.

[0031] S23: The data processing unit performs a real-time deviation comparison between the real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor and the calibration ammonia nitrogen concentration data, and calculates the concentration difference between the two. The real-time deviation comparison uses the sliding window mean comparison method, specifically: The data processing unit extracts multiple real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor within the most recent set time window (e.g., 5 minutes) and calculates their average value. This average value is then compared with the single calibration ammonia nitrogen concentration data generated by the backup ammonia nitrogen sensor during the current synchronization acquisition cycle. The concentration difference is calculated using the following formula: ; in, The concentration difference (mg / L) The mean value (mg / L) of the sliding window of the main ammonia nitrogen sensor. This provides calibration ammonia nitrogen concentration data (mg / L) for the backup ammonia nitrogen sensor. The sliding window mean comparison method effectively eliminates the influence of instantaneous fluctuations in the primary ammonia nitrogen sensor on the results, improving the stability and accuracy of the difference calculation.

[0032] S24: If the concentration difference If the concentration difference is within the system's set allowable error range (e.g., ±0.3 mg / L), the data processing unit determines that the main ammonia nitrogen sensor data is reliable and outputs the real-time ammonia nitrogen concentration data as reliable real-time ammonia nitrogen concentration data for subsequent steps S3 and S4. If the concentration difference exceeds the preset allowable error range, the data processing unit triggers the sensor self-cleaning procedure and recalibration process for low-temperature environments.

[0033] The sensor self-cleaning procedure for low-temperature environments specifically includes: the data processing unit controls the miniature heating element surrounding the main ammonia nitrogen sensor probe to start a low-power heating program, locally raising the water around the probe to a preset temperature (e.g., 18°C) for 2 minutes to soften dirt and low-temperature deposits on the electrode surface; after the heating process is completed, a miniature ultrasonic generator is activated to generate pulsed ultrasonic oscillations in the 1 MHz frequency range for 30 seconds to perform high-frequency oscillation cleaning of the electrode surface, removing microbial films and deposits.

[0034] The recalibration process is as follows: After completing the sensor self-cleaning procedure, the data processing unit controls the main ammonia nitrogen sensor and the backup ammonia nitrogen sensor to simultaneously acquire measurements of multiple sets of standard solution samples (including standard solutions with known ammonia nitrogen concentrations of 0.5 mg / L, 1.0 mg / L, and 2.0 mg / L). The measured response signals are then fitted and analyzed against the standard solution concentrations. Based on the least squares fitting principle, the calibration curve parameters of the main ammonia nitrogen sensor are recalibrated. The updated calibration curve is loaded into the data processing module of the main ammonia nitrogen sensor, replacing the original curve and achieving adaptive and accurate calibration.

[0035] S3: Based on the environmental characteristics of large diurnal temperature differences, combined with historical and reliable real-time ammonia nitrogen concentration data and temperature and pH value variation patterns, a prediction model adapted to low-temperature water environments is used to predict ammonia nitrogen concentration values ​​for a specific future period. Specifically: S31: After completing the reliability verification of real-time ammonia nitrogen concentration data in S2, the data processing unit arranges the reliable real-time ammonia nitrogen concentration data, temperature data, pH value data, and dissolved oxygen data within a continuous time period in an ordered manner according to timestamps to construct a time series dataset. The time series dataset covers the continuous monitoring period within the historical time period, with the time step consistent with the data acquisition frequency, ensuring no missing data. For outliers and noise interference in the time series dataset, a sliding window median filtering method is used for preprocessing. The window size of the sliding window median filtering method is dynamically adjusted according to the actual data sampling frequency. For example, when the sampling frequency is 1 time / 5 minutes, a window size of 5 is used; when the sampling frequency is 1 time / 10 minutes, a window size of 3 is used. Through this median filtering process, pulse noise caused by sudden interference or short-term signal jitter is effectively eliminated, maintaining the stability and representativeness of the data.

[0036] S32: The data processing unit extracts structured features from the preprocessed time-series dataset to generate a set of key input features characterizing the environmental characteristics of cold-water aquaculture ponds. These key input features include four categories of indicators: (1) The first difference value of reliable real-time ammonia nitrogen concentration data is used to calculate the rate of change between consecutive concentration values ​​in the past 6 hours and capture its upward or downward trend. (2) The diurnal temperature range characteristics in the temperature data are expressed by extracting the difference between the maximum and minimum temperatures in the past 24 hours; (3) The variation pattern of pH value data is reflected by the fluctuation amplitude index of pH value series, that is, the difference between the maximum and minimum values, which reflects the degree of acidity and alkalinity change. (4) The correlation fluctuation pattern of dissolved oxygen data is calculated by the Pearson correlation coefficient with ammonia nitrogen concentration data, reflecting the coupling strength between dissolved oxygen level and ammonia nitrogen change.

[0037] The aforementioned key input features are all normalized and stored in a feature vector set, which is then used as input for subsequent prediction models.

[0038] S33: Key input features are fed into a prediction model adapted to low-temperature water environments. This prediction model is based on a Long Short-Term Memory (LSTM) network architecture and employs a two-layer LSTM network structure, with each layer containing 128 neurons. The input layer dimension matches the number of key input features, currently 4 dimensions; the output layer is the predicted ammonia nitrogen concentration sequence for the next 12 hours, with a time step of 1 hour, outputting 12 consecutive predicted values. Before deployment, the prediction model is pre-trained using a large amount of historical data from high-latitude cold-water aquaculture farms. During training, a mean squared error loss function and an adaptive gradient optimization algorithm are used to ensure that the model has good generalization ability and temporal prediction accuracy in low-temperature environments. The data processing unit inputs the key input features extracted at the current moment into the LSTM prediction model to obtain the corresponding predicted ammonia nitrogen concentration sequence for the next 12 hours.

[0039] S34: For the ammonia nitrogen concentration prediction sequence output by the LSTM network, an exponentially weighted moving average method is used to smooth and optimize the prediction sequence. This method assigns dynamically changing weighting coefficients to the predicted values ​​at different time steps, making the prediction sequence closer to the actual trend and suppressing random disturbances. The specific weight calculation method is as follows: ; in, This is a smoothed forecast value for the current moment. This is the current raw prediction value output by the LSTM. The weighting factor is dynamically adjusted based on the prediction time step; the closer the predicted value is to the current time point, the higher the weight. The default setting decreases from 0.7 to 0.3. The maximum value of the entire smoothed prediction sequence is extracted as the final predicted ammonia nitrogen concentration.

[0040] In addition, to ensure the reliability of the predicted values, a reliability assessment is performed on the prediction results. When the predicted values ​​of ammonia nitrogen concentration in the prediction sequence exceed 80% of the safety threshold for salmon farming for three consecutive time steps (for example, if the threshold is set to 2.5 mg / L, then the judgment condition is that the predicted value is greater than 2.0 mg / L), the predicted ammonia nitrogen concentration value is marked as a high-reliability warning signal, which serves as the trigger for the generation of the synergistic regulation strategy.

[0041] S4: Compare the predicted ammonia nitrogen concentration with a specific safety threshold for salmon farming. When the predicted value approaches or exceeds the safety threshold, generate and execute coordinated control instructions suitable for cold-water aquaculture, including one or more combinations of: adjusting glacial water supply, controlling aerator operation mode, and adjusting feeding strategies. Specifically: S41: The data processing unit receives the predicted ammonia nitrogen concentration output in step S3 and compares it with the specific safety threshold for salmon farming pre-stored in the system. The specific safety threshold for salmon farming is determined based on the physiological tolerance of high-latitude cold-water fish to ammonia nitrogen concentration, for example, it can be set to 2.5 mg / L. By calculating the percentage of the predicted ammonia nitrogen concentration exceeding this safety threshold, the corresponding exceedance level is determined. The specific classification rules are as follows: When the predicted ammonia nitrogen concentration is within 100%–120% of the safe threshold (i.e., 2.5–3.0 mg / L), it is marked as a Level 1 warning; when it is within 120%–150% (i.e., 3.0–3.75 mg / L), it is marked as a Level 2 warning; and when it exceeds 150% (i.e., above 3.75 mg / L), it is marked as a Level 3 warning.

[0042] The system jointly determines the final exceedance level based on the current exceedance magnitude and the rising rate of the predicted ammonia nitrogen concentration in the time series, and outputs the result to the collaborative control module.

[0043] S42: The coordinated control module generates a corresponding set of coordinated control instructions from a pre-set control strategy library based on the level of exceedance. The pre-set control strategy library is a parameterized instruction mapping table built based on historical aquaculture control experience, covering three levels of exceedance: Level 1, Level 2, and Level 3. It matches differentiated parameters for adjusting glacial water replenishment, controlling aerator operation mode, and adjusting feeding strategy parameters, respectively.

[0044] The set of coordinated control instructions corresponding to the Level 1 warning is as follows: reduce feeding strategy parameters (reduce feeding amount by 10% and increase feeding interval by 10%), fine-tune aerator operation mode parameters (adjust to intermittent operation, stop for 5 minutes every 10 minutes of operation), and do not activate glacial water regulation.

[0045] The set of coordinated control instructions corresponding to the Level II early warning is as follows: significantly adjust the glacier water replenishment parameter (increase it by 10%-30%, with the specific value determined by the prediction deviation), strengthen the aerator operation mode parameter (switch to continuous low power operation, with the power controlled within 60% of the maximum power), and do not adjust the feeding strategy parameter.

[0046] The set of coordinated control instructions corresponding to the Level 3 early warning is a comprehensive control combination, which activates all three parameter channels: the glacier water replenishment parameter is executed at the maximum threshold (increased by more than 30%), the aerator operation mode parameter is switched to continuous high power mode (power reaches maximum output), and the feeding strategy parameter is to suspend feeding.

[0047] S43: The coordinated control module distributes the set of coordinated control instructions to the corresponding execution devices. The specific execution method is as follows: A command to adjust the glacial water supply is sent to the glacial water inlet valve controller. The opening command is calculated using the following formula: ; in, The predicted value of ammonia nitrogen concentration (mg / L) The safety threshold (mg / L) The adjustment coefficient is set based on the pond volume and water exchange efficiency (e.g., pond volume is 100 m³). 3 If the water exchange efficiency is 50%, then the adjustment coefficient is 40.

[0048] Send start / stop and power commands to the aerator controller to control the aerator's operating mode, specifically including: intermittent aeration mode (start / stop cycle control) under Level 1 warning; continuous low-power aeration mode (power at 60% of rated value) under Level 2 warning; and continuous high-power aeration mode (power between 90% and 100% of rated value) under Level 3 warning.

[0049] Send instructions to the feeder controller to adjust the feeding amount and interval of the feeding strategy. The specific parameters tend to be conservative as the level of exceedance increases, so as to avoid additional nitrogen input under high ammonia nitrogen load conditions.

[0050] S44: After each executing device receives and executes the corresponding coordinated control command, the data processing unit initiates a new round of real-time monitoring to evaluate the effect of the ammonia nitrogen concentration change trend after the execution action. The specific method for verifying the control effect is as follows: the system continuously acquires reliable real-time ammonia nitrogen concentration data for 3 hours after the control, and calculates the linear change slope of the ammonia nitrogen concentration during this period. When the slope changes from a positive value to a negative value, and the absolute value exceeds the threshold set by the system (e.g., ≥0.05 mg / L / h), the current round of control is deemed effective, and the current control parameters and effect are recorded to update the strategy library. When the control effect is deemed insufficient or without improvement, the system automatically starts the dynamic adjustment program of the next round of coordinated control command set, increases the control level or extends the command duration, and ensures that the ammonia nitrogen concentration returns to within the safe threshold.

[0051] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0052] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for real-time detection and regulation of ammonia nitrogen content in a culture pond, characterized in that, The method comprises the following steps: S1: The main ammonia nitrogen sensor and the standby ammonia nitrogen sensor deployed in the cold water aquaculture pond synchronously collect the ammonia nitrogen concentration original signal and the standby ammonia nitrogen concentration data at the current time, the standby ammonia nitrogen sensor is in a low-temperature dormant state during a non-working period; the temperature, pH value and dissolved oxygen data of the pond water body are synchronously collected, and the temperature compensation correction is performed on the main ammonia nitrogen sensor reading for the low-temperature water body characteristics to generate real-time ammonia nitrogen concentration data; S2: The standby ammonia nitrogen sensor in the low-temperature dormant state is regularly awakened, standby ammonia nitrogen concentration data is obtained and converted into verification ammonia nitrogen concentration data; the real-time ammonia nitrogen concentration data and the verification ammonia nitrogen concentration data are compared for deviation in a low-temperature environment, if the difference is within an allowable error range, reliable real-time ammonia nitrogen concentration data is output; If the error range is exceeded, a sensor self-cleaning program for the low-temperature environment is started and recalibration is performed; S3: Based on the environmental characteristics of large diurnal temperature difference, in combination with the historical reliable real-time ammonia nitrogen concentration data and the temperature and pH value change law, a prediction model suitable for the low-temperature water body environment is used to predict the ammonia nitrogen concentration prediction value in a specific period in the future; S4: The ammonia nitrogen concentration prediction value is compared with a specific safety threshold value for salmon farming, when the prediction value approaches or exceeds the safety threshold value, a coordinated control instruction suitable for cold water aquaculture is generated and executed, including one or more combinations of adjusting the glacial water supply amount, controlling the operation mode of the oxygenator, adjusting the feeding strategy.

2. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 1, characterized in that, The S1 comprises: S11: The main ammonia nitrogen sensor continuously collects the ammonia nitrogen concentration original signal of the aquaculture pond water body, while the standby ammonia nitrogen sensor is controlled to collect standby ammonia nitrogen concentration data during the awakened synchronous collection period, and is in a low-temperature dormant state during a non-synchronous collection period; S12: The temperature data, pH value data and dissolved oxygen data are synchronously collected by the temperature sensor, the pH sensor and the dissolved oxygen sensor, and the pH value data and the dissolved oxygen data are used for subsequent ammonia nitrogen change trend prediction and intelligent coordinated control; S13: The temperature data is used to perform nonlinear compensation correction on the ammonia nitrogen concentration original signal collected by the main ammonia nitrogen sensor for the low-temperature water body characteristics by using a temperature compensation algorithm, to generate corrected real-time ammonia nitrogen concentration data.

3. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 2, characterized in that, In the S11, the standby ammonia nitrogen sensor is controlled to collect standby ammonia nitrogen concentration data during the awakened synchronous collection period, specifically, a preset data processing unit sends a synchronous trigger signal to the standby ammonia nitrogen sensor to make the sampling time sequence of the standby ammonia nitrogen sensor completely consistent with that of the main ammonia nitrogen sensor.

4. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 3, characterized in that, In the S11, the standby ammonia nitrogen sensor is made to be in a low-temperature dormant state, specifically, the data processing unit cuts off the power supply of the standby ammonia nitrogen sensor except for the awakening receiving circuit, and covers the sensing probe of the standby ammonia nitrogen sensor with a mechanical protective cover to isolate it from the water body.

5. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 4, characterized in that, The S2 comprises: S21: The data processing unit sends a wakeup instruction according to a preset period to awaken the standby ammonia nitrogen sensor in the low-temperature dormant state, and controls it to perform synchronous signal collection with the main ammonia nitrogen sensor to obtain the standby ammonia nitrogen concentration data collected by the standby ammonia nitrogen sensor; S22: The data processing unit performs signal conversion and temperature compensation correction on the standby ammonia nitrogen concentration data to generate verification ammonia nitrogen concentration data for comparison; S23: The data processing unit performs real-time deviation comparison between the real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor and the verification ammonia nitrogen concentration data, and calculates the concentration difference therebetween; S24: If the concentration difference is within the preset allowable error range, it is determined that the main ammonia nitrogen sensor data is reliable, and the real-time ammonia nitrogen concentration data is output as reliable real-time ammonia nitrogen concentration data; if the concentration difference exceeds the preset allowable error range, a sensor self-cleaning program and a recalibration process for low-temperature environment are triggered.

6. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 5, characterized in that, The preset period in S21 is dynamically adjusted according to the degree of water eutrophication, specifically: the data processing unit calculates the change rate of the historical reliable real-time ammonia nitrogen concentration data, and automatically shortens the preset period when the change rate exceeds a threshold.

7. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 6, characterized in that, The real-time deviation comparison in S23 uses a sliding window mean comparison method, specifically: the average value of multiple real-time ammonia nitrogen concentration data generated by the main ammonia nitrogen sensor in the latest time window is taken, and the concentration difference is calculated with the single verification ammonia nitrogen concentration data generated by the standby ammonia nitrogen sensor.

8. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 7, characterized in that, S3 includes: S31: Construct a time series data set containing continuous reliable real-time ammonia nitrogen concentration data, temperature data, pH value data and dissolved oxygen data in a historical period, and pre-process the time series data set to eliminate abnormal values and noise interference; S32: Based on the pre-processed time series data set, extract key input features for representing the environmental characteristics of cold water aquaculture ponds, including the change trend of reliable real-time ammonia nitrogen concentration data, the day-night temperature difference characteristics in temperature data, the change law of pH value data and the associated fluctuation mode of dissolved oxygen data; S33: Input the key input features into a pre-trained prediction model adapted to low-temperature water body environment, which is based on a long short-term memory network architecture and trained with high-latitude cold water aquaculture data, and generate an ammonia nitrogen concentration prediction sequence in a future specific period through model calculation; S34: Smooth and optimize the ammonia nitrogen concentration prediction sequence, extract its peak value as the final ammonia nitrogen concentration prediction value, and output the ammonia nitrogen concentration prediction value for subsequent intelligent collaborative regulation decision.

9. The method for real-time detection and control of ammonia nitrogen content in aquaculture ponds according to claim 8, characterized in that, S4 includes: S41: Compare and analyze the ammonia nitrogen concentration prediction value with the pre-stored specific safety threshold of salmon farming, and determine the over-standard level of the ammonia nitrogen concentration prediction value according to the over-standard amplitude and rising rate; S42: According to the over-standard level, match and generate a corresponding set of collaborative regulation instructions from the pre-set regulation strategy library, which includes specific adjustment of glacial water supply amount parameters, control of oxygenation machine operation mode parameters and adjustment of feeding strategy parameters; S43: The set of coordinated control instructions is issued to the corresponding execution equipment, specifically including sending the opening degree instruction of adjusting the supply amount of glacial water to the glacial water inlet valve controller, sending the start-stop and power instruction of controlling the operation mode of the oxygenator to the oxygenator controller, and sending the feeding amount and interval instruction of adjusting the feeding strategy to the feeder controller; S44: After each execution equipment receives and executes the corresponding instruction, it continues to monitor the reliable real-time ammonia nitrogen concentration data change of the new round, verifies the control effect and dynamically adjusts the subsequent set of coordinated control instructions.

Images