Fish pond water environment anomaly detection and automatic alarm method and system

Abstract

The invention discloses a fishpond water environment anomaly detection and automatic alarm method and system, relates to the technical field of water environment detection, and overcomes the limitation of false alarm and missing alarm of a single sensor by simultaneously collecting and analyzing water physical and chemical parameters, fish motion videos and environment audios. The accuracy and the reliability of abnormity judgment are greatly improved; the dynamic abnormal degree of water quality parameters is quantified by calculating a'water symptom index ', and a'biological stress index' is calculated and fused with fish behavior abnormal characteristics, so that the transformation from'alarm after exceeding 'to'trend abnormal early warning' is realized, and precious advance is gained for intervention measures; according to the method, a grading alarm strategy is automatically executed according to the abnormal grade, the closed-loop automatic management from monitoring, judgment to primary treatment is realized from sending of an early warning notification and automatic starting of the aerator to triggering of a field sound-light alarm, and the refinement level and emergency response efficiency of fishpond management are improved.

Description

Technical Field

[0001] This application relates to the field of water environment monitoring technology, specifically to a method and system for detecting and automatically alarming abnormal water environment in fishponds. Background Technology

[0002] Traditional aquaculture, especially high-density fishpond farming, relies primarily on two methods for water quality management and early warning of anomalies: one is manual observation and regular water sampling based on the experience of aquaculture workers. This method suffers from strong subjectivity, delayed response, and the inability to conduct continuous monitoring around the clock, especially with large blind spots at night or during inclement weather. Therefore, current technology lacks a method that can integrate the dynamic trends of physicochemical parameters with multidimensional characteristics of biological behavior to achieve earlier, more accurate, and more comprehensive integrated perception and intelligent early warning of anomalies in fishpond water environments.

[0003] Existing technology, such as the invention application patent with publication number CN113841655A, discloses a fish farming environment purification system and method based on water environment treatment, belonging to the field of fish farming. It includes a purification pond and oxygen supply equipment distributed within the purification pond. The key feature is that the purification pond has several evenly distributed partitions inside, one side of each partition has a flow guide plate that cooperates with it, the side of the partition away from the flow guide plate has a functional plate, a flow outlet is provided between the functional plate and the partition, and several evenly distributed support rods are provided on one side of the bottom of the partition. The purification system operates normally by setting up the farming pond and the oxygen supply equipment. During operation, the purification system can effectively clean the water surface, supply oxygen to the farming pond, and continuously clean fish feces and other residues, allowing the purification system to integrate with the farming system and environment for a long time, forming a complete and diversified system.

[0004] Regarding the above-mentioned solutions, the inventors of this application have found that the above-mentioned technologies have at least the following technical problems: 1. Currently, there is a lack of simultaneous collection and correlation of three types of heterogeneous data: water body physicochemical parameter sequences, fish activity video streams, and environmental audio streams. There is no comprehensive monitoring data foundation for "water body chemical environment - fish visual behavior - fish acoustic behavior". It is impossible to overcome the limitations of traditional single water quality sensor monitoring, and it is impossible to simultaneously capture environmental physicochemical changes and biological behavioral responses, and it is impossible to provide multi-dimensional data support for a comprehensive assessment of the health status of fish ponds.

[0005] 2. Currently, there is a lack of two independent comprehensive quantitative indices—the water symptom index and the biological stress index. The abnormalities are not quantified from two independent but related channels: "environmental physical and chemical stress" and "biological behavioral response." Complex multi-parameter information cannot be condensed into two intuitive quantitative indicators, which is not convenient for the system to make logical judgments and risk classifications, nor is it convenient for managers to understand.

[0006] 3. Currently, there is a lack of decision-making based on a logical combination of water body anomaly markers and biological anomaly markers. The physicochemical and biological evidence is not integrated through logical rules, so the anomaly judgment is not simply a matter of crossing a threshold. There is no comprehensive reasoning based on multiple evidence chains, which cannot improve the accuracy and intelligence level of the judgment. There is also a lack of clear three-level classification, which cannot provide a direct and clear basis for taking differentiated response measures, and cannot achieve the leap from "whether to alarm" to "according to what level of urgency to alarm". Summary of the Invention

[0007] To address the aforementioned technical shortcomings, the purpose of this application is to provide a method and system for detecting and automatically alarming abnormalities in fishpond water environment.

[0008] To solve the above-mentioned technical problems, this application adopts the following technical solution: In the first aspect, this application provides a method for detecting and automatically alarming abnormal water environment in fish ponds. The method includes the following steps: S1, generating standardized water quality data, fish movement data sequence and audio energy time series data based on pre-acquired fish pond water physicochemical parameter sequence, fish activity video stream and environmental audio stream.

[0009] S2. Based on the standardized water quality data, fish movement data sequence, and audio energy time series data, key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics are analyzed and obtained.

[0010] S3. Calculation of water body symptom index and biological stress index.

[0011] S4. Dynamic early warning threshold comparison and anomaly determination.

[0012] S5, Execution of hierarchical alarm strategy.

[0013] Preferably, the step of generating standardized water quality data, fish movement data sequence, and audio energy time-series data based on pre-acquired fishpond water physicochemical parameter sequences, fish activity video streams, and environmental audio streams includes: acquiring fishpond water physicochemical parameter sequences collected by pre-deployed sensors, performing missing value imputation and standardization processing on the water physicochemical parameter sequences to generate standardized water quality data; acquiring fish activity video streams captured by pre-deployed cameras, extracting keyframes from the fish activity video streams and performing background subtraction and target detection to generate fish movement data sequences; and acquiring environmental audio streams collected by pre-deployed microphones, performing noise reduction and specific frequency band energy extraction on the environmental audio streams to generate audio energy time-series data.

[0014] Preferably, the step of analyzing and obtaining key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics based on the standardized water quality data, fish movement data sequences, and audio energy time series data includes: performing key parameter extraction operations on dissolved oxygen concentration time series data, ammonia nitrogen concentration data, and pH value data based on the standardized water quality data to obtain a set of water body characteristic parameters.

[0015] Based on fish movement data sequences, video behavioral features are extracted to obtain video behavioral characteristics.

[0016] Based on audio energy time-series data, the average intensity of the energy data of feeding activity frequency bands within a unit time is calculated to obtain audio behavior characteristics.

[0017] Preferably, the step of extracting video behavioral features based on fish motion data sequences to obtain video behavioral features includes: extracting video behavioral features based on fish motion data sequences, calculating average swimming speed, motion disorder, spatial clustering degree, and water surface activity ratio; average swimming speed is represented as the average displacement speed of all fish within a time window; motion disorder is represented as the standard deviation of the trajectory positions of all fish; spatial clustering degree is represented as the proportion of fish forming a cluster; and water surface activity ratio is represented as the proportion of trajectory points located within a preset "water surface area" at the top of the screen.

[0018] Preferably, the calculation of the water body symptom index and the biological stress index includes: calculating the water body symptom index based on the water body characteristic parameters and a preset water body symptom index algorithm; and generating the biological stress index by fusing the video behavior features and audio behavior features using a preset biological stress index algorithm.

[0019] Preferably, the calculation of the water body symptom index based on the water body characteristic parameters and a preset water body symptom index algorithm includes: calculating the water body symptom index based on the dissolved oxygen change slope, ammonia nitrogen concentration, and pH deviation in the water body characteristic parameter set, using the preset calculation formula of the water body symptom index algorithm. The water symptom index was obtained. , Expressed as the slope of the change in dissolved oxygen, This represents the current measured value of ammonia nitrogen concentration. This represents the preset upper limit of the safe threshold for ammonia nitrogen concentration. This is expressed as pH deviation. , and These are respectively represented as the weighting factor corresponding to the slope of dissolved oxygen change, the weighting factor corresponding to ammonia nitrogen concentration, and the weighting factor corresponding to pH deviation.

[0020] Preferably, the step of generating a biological stress index by fusing the video and audio behavioral features using a preset biological stress index algorithm includes: calculating the biological stress index using the preset biological stress index algorithm formula. Biological stress index was derived ,in Represented as video behavioral features, Represented as audio behavior features. This is represented by the maximum value of the audio behavior feature. and These are represented as the weighting factors corresponding to video behavior features and the weighting factors corresponding to audio behavior features.

[0021] Preferably, the dynamic early warning threshold comparison and anomaly determination includes: performing a numerical comparison operation based on the water symptom index and a preset water symptom index early warning threshold to obtain a water anomaly indicator.

[0022] Based on the biological stress index and the preset biological stress index warning threshold, a numerical comparison operation is performed to obtain biological abnormality markers.

[0023] Based on water body anomaly markers and biological anomaly markers, a decision is made through anomaly level classification logic to generate the anomaly level.

[0024] Preferably, the execution of the graded alarm strategy includes: based on the anomaly determination flag and the anomaly level, executing a preset graded alarm strategy decision to obtain the alarm strategy type to be executed; when the anomaly level is equal to the preset first-level warning level, triggering the execution of the first-level warning strategy; when the anomaly level is equal to the preset second-level emergency level, triggering the execution of the second-level emergency alarm strategy.

[0025] In a second aspect, this application provides a system for detecting and automatically alarming abnormalities in fishpond water environment, comprising: preferably, a data acquisition module, which generates standardized water quality data, fish movement data sequence, and audio energy time series data based on pre-acquired fishpond water physicochemical parameter sequences, fish activity video streams, and environmental audio streams.

[0026] The feature analysis module, based on the standardized water quality data, fish movement data sequences, and audio energy time series data, analyzes and derives key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics.

[0027] Module for calculating water symptom index and biological stress index.

[0028] Dynamic early warning threshold comparison and anomaly detection module.

[0029] Tiered alarm strategy execution module.

[0030] The beneficial effects of this application are as follows: 1. The method and system for detecting and automatically alarming abnormal water environment in fishponds provided by this application overcomes the limitations of false alarms and missed alarms caused by a single sensor by simultaneously collecting and analyzing water body physicochemical parameters, fish movement videos, and environmental audio, and greatly improves the accuracy and reliability of abnormality judgment; by calculating the "water body symptom index" to quantify the dynamic abnormality of water quality parameters, and by calculating the "biological stress index" to integrate abnormal fish behavior characteristics, the transformation from "alarm after exceeding the standard" to "early warning of abnormal trends" is realized, providing valuable advance time for intervention measures; according to the level of abnormality, a graded alarm strategy is automatically executed, from sending early warning notifications and automatically starting aerators to triggering on-site audible and visual alarms, realizing closed-loop automated management from monitoring, judgment to preliminary treatment, improving the precision level of fishpond management and emergency response efficiency.

[0031] 2. This application simultaneously collects and correlates three types of heterogeneous data: water body physicochemical parameter sequences, fish activity video streams, and environmental audio streams, constructing a comprehensive monitoring data foundation of "water body chemical environment - fish visual behavior - fish acoustic behavior"; it overcomes the limitations of traditional single water quality sensor monitoring, and can simultaneously capture environmental physicochemical changes and biological behavioral responses, providing multidimensional data support for a comprehensive assessment of fishpond health status.

[0032] 3. This application constructs two independent comprehensive quantitative indices—the water body symptom index and the biological stress index—to quantify anomalies from two independent but related channels: “environmental physical and chemical stress” and “biological behavioral response”. Complex multi-parameter information is condensed into two intuitive quantitative indicators, which facilitates logical judgment and risk classification by the system and is also easy for managers to understand.

[0033] 4. This application makes decisions based on a logical combination of water body anomaly markers and biological anomaly markers. By integrating physicochemical and biological evidence through logical rules, the anomaly determination is no longer a simple threshold crossing, but a comprehensive reasoning based on multiple evidence chains, which greatly improves the accuracy and intelligence of the determination. The clear three-level classification provides a direct and clear basis for taking differentiated response measures, realizing the leap from "whether to alarm" to "according to what level of urgency to alarm". Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart illustrating the steps involved in implementing the method described in this application.

[0036] Figure 2 This is a schematic diagram of the system structure connection of this application. Detailed Implementation

[0037] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0038] Please see Figure 1 As shown, this application provides a method for detecting and automatically alarming abnormal water environment in fishponds in the first aspect, including: S1, generating standardized water quality data, fish movement data sequence and audio energy time series data based on pre-acquired fishpond water body physicochemical parameter sequence, fish activity video stream and environmental audio stream.

[0039] In a specific example, the process of generating standardized water quality data, fish movement data sequences, and audio energy time-series data based on pre-acquired fishpond water physicochemical parameter sequences, fish activity video streams, and environmental audio streams includes: acquiring fishpond water physicochemical parameter sequences collected by pre-deployed sensors, performing missing value imputation and standardization on the water physicochemical parameter sequences to generate standardized water quality data; acquiring fish activity video streams captured by pre-deployed cameras, extracting keyframes from the fish activity video streams and performing background subtraction and target detection to generate fish movement data sequences; and acquiring environmental audio streams collected by pre-deployed microphones, performing noise reduction and specific frequency band energy extraction on the environmental audio streams to generate audio energy time-series data.

[0040] It should be noted that the process of imputing missing values ​​and standardizing the water body physicochemical parameter sequence is as follows: First, based on the values ​​of adjacent valid sampling points in the water body physicochemical parameter sequence, a linear interpolation algorithm is used to estimate the missing data points in the sequence to generate a complete parameter sequence. The calculation formula for the linear interpolation algorithm is as follows. Obtain the physicochemical parameter values ​​of the filled missing points. ,in and These represent the physicochemical parameter values ​​of the previous and next valid sampling points, respectively. The sampling time points are represented as missing points. and These represent the sampling time points of the previous and next valid sampling points, respectively. Further, the essence of the linear interpolation algorithm is to linearly estimate the parameter values ​​at missing moments based on adjacent valid data points in time sequence, maintaining the continuity of the data time series. After missing value imputation, the complete parameter sequence is standardized. The standardization process is based on the Z-Score standardization algorithm, and the calculation process is as follows: First, calculate the arithmetic mean of all data points in the entire parameter sequence; second, calculate the standard deviation of all data points relative to this mean; finally, subtract the mean from the value of each data point and divide by the standard deviation to obtain the standardized value. Further, the Z-Score standardization algorithm transforms water quality parameters of different dimensions and orders of magnitude (such as dissolved oxygen concentration and pH value) into a standard normal distribution with a mean of 0 and a standard deviation of 1, eliminating the influence of unit and scale differences between different parameters on subsequent fusion calculations and generating unified standardized water quality data.

[0041] It should be noted that the extraction of keyframes from the fish activity video stream refers to selecting images representing key moments in the movement of the fish school from consecutive image frames based on the inter-frame content change rate of the video stream, thereby reducing data redundancy. Specifically, this involves calculating the sum of differences between corresponding pixels in consecutive frames, and determining the current frame as a keyframe when the sum of differences exceeds a preset threshold. Background subtraction of the extracted keyframes involves modeling the video background using a Gaussian mixture model, subtracting the modeled static background from each keyframe, thereby separating the dynamic foreground region containing the fish to highlight the moving target. Target detection of the keyframes after background subtraction involves scanning the dynamic foreground region based on a preset target detection algorithm, identifying and locating fish targets in the image, and outputting the bounding box coordinates of each fish in the image, thus obtaining the fish position set for each keyframe. Furthermore, the process of generating the fish motion data sequence is as follows: Based on the bounding box coordinates of the fish targets detected in all keyframes, the bounding boxes belonging to the same fish in different frames are associated through a target tracking algorithm (such as Kalman filtering combined with the Hungarian algorithm) to form a data sequence of each fish changing over time. This data sequence represents the motion data of the fish.

[0042] It should be noted that the noise reduction of the ambient audio stream refers to processing the ambient audio stream using the Wiener filtering algorithm to suppress steady-state background noise in the audio signal and enhance acoustic event signals related to fish activity. Wiener filtering adjusts the gain of each frequency component in the frequency domain based on the power spectrum estimation of the signal and noise.

[0043] It should be noted that the energy extraction of a specific frequency band from the denoised audio stream refers to performing a short-time Fourier transform on the audio stream to convert the time-domain signal into a time-spectrum graph, and then extracting the energy values ​​within a specific frequency range related to the splashing sound of fish feeding. The specific frequency band is typically set to the high-frequency portion of the human audible range, such as 2kHz to 8kHz. The energy extraction calculation formula is used to... Time of Result audio energy value , For audio signals at time frequency The short-time Fourier transform coefficients at the given location, and These are the lower and upper frequency limits for the feeding splash sound band, respectively. Furthermore, the essence of this energy extraction process is to calculate the sum of squares of the amplitudes of all frequency components within the selected frequency band, characterizing the total intensity of sound wave vibrations in that band per unit time. By traversing all time periods, a time-varying audio energy temporal data is generated, reflecting the intensity changes of acoustic activity related to feeding behavior in the fishpond environment.

[0044] This application simultaneously collects and correlates three types of heterogeneous data: water body physicochemical parameter sequences, fish activity video streams, and environmental audio streams, constructing a comprehensive monitoring data foundation encompassing "water body chemical environment - fish visual behavior - fish acoustic behavior." It overcomes the limitations of traditional single water quality sensor monitoring, enabling the simultaneous capture of environmental physicochemical changes and biological behavioral responses, providing multidimensional data support for a comprehensive assessment of fishpond health.

[0045] S2. Based on the standardized water quality data, fish movement data sequence, and audio energy time series data, key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics are analyzed and obtained.

[0046] In a specific example, the analysis of key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics based on the standardized water quality data, fish movement data sequences, and audio energy time series data includes: performing key parameter extraction operations on dissolved oxygen concentration time series data, ammonia nitrogen concentration data, and pH value data based on the standardized water quality data to obtain a set of water body characteristic parameters.

[0047] Based on fish movement data sequences, video behavioral features are extracted to obtain video behavioral characteristics.

[0048] Based on audio energy time-series data, the average intensity of the energy data of feeding activity frequency bands within a unit time is calculated to obtain audio behavior characteristics.

[0049] It should be noted that the key parameter extraction operation on the dissolved oxygen concentration time-series data, ammonia nitrogen concentration data, and pH value data specifically includes: based on the dissolved oxygen concentration time-series data in standardized water quality data, calculating the rate of change of dissolved oxygen concentration over time using the least squares linear fitting method to obtain the slope of dissolved oxygen change. Further, the calculation formula for the slope of dissolved oxygen change is used... The slope of dissolved oxygen change was obtained. , This is represented by the number corresponding to the sampling time point. , This represents the total number of sampling time points. Represented as the first Each sampling time point It is expressed as the average value at all sampling time points. Represented as the first Standardized dissolved oxygen concentration values ​​corresponding to each sampling time point It is expressed as the average of all dissolved oxygen concentration values. Furthermore, the slope of dissolved oxygen change is a quantitative indicator that measures the trend of dissolved oxygen concentration change in water body per unit time. Its magnitude and sign reflect the intensity and direction of the reoxygenation or oxygen consumption process in the water body, respectively, and it is a key parameter for assessing the self-purification capacity of water bodies and organic pollution load.

[0050] It should be noted that the key parameter extraction operation on the dissolved oxygen concentration time-series data, ammonia nitrogen concentration data, and pH value data also includes: directly extracting the ammonia nitrogen concentration value at the corresponding sampling time from the standardized water quality data as the ammonia nitrogen concentration in the water body characteristic parameter set. Furthermore, the ammonia nitrogen concentration reading at a specified time point in the standardized water quality data sequence is directly used as the output ammonia nitrogen concentration parameter. Ammonia nitrogen concentration is a direct measurement of the ammonia nitrogen content in water bodies. Its value directly reflects the degree of pollution from organic matter decomposition in the water body and is a core physicochemical indicator for judging eutrophication and the risk of toxicity to fish.

[0051] It should be noted that the key parameter extraction operation on the dissolved oxygen concentration time-series data, ammonia nitrogen concentration data, and pH value data also includes: calculating the absolute difference between the pH value at each sampling point and the median of the preset suitable pH range for fish based on the pH value data in the standardized water quality data, thus obtaining the pH deviation. Furthermore, the pH deviation reflects the degree to which the water's acidity or alkalinity deviates from the optimal survival range for fish. A larger value indicates a more severe acidity or alkalinity anomaly, directly affecting fish metabolism, respiration, and ammonia nitrogen toxicity, and is an important parameter for assessing the stability of the aquatic chemical environment.

[0052] It should be noted that the operation of calculating the average intensity of the feeding activity frequency band energy data per unit time refers to calculating the arithmetic mean of the audio energy values ​​at all times within a unit time window, based on audio energy time-series data and within a set feeding activity-related frequency band. Further, the calculation formula for the average intensity of audio energy is... Calculate the unit time Average intensity of energy in the internal feeding activity frequency band This is denoted as audio behavior feature. , This is expressed as the total number of time points per unit time. Furthermore, the average intensity of the feeding activity frequency band per unit time characterizes the average activity level of the acoustic signals generated by the feeding behavior of fish. During normal feeding, this value remains at a certain level; when aquatic environmental abnormalities lead to decreased appetite or stress in fish, feeding activity decreases or ceases, and this average intensity value drops significantly. Therefore, it can serve as an important audio characteristic reflecting the physiological and behavioral state of fish.

[0053] In a specific example, the extraction of video behavior features based on fish movement data sequences to obtain video behavior features includes: the extraction of video behavior features based on fish movement data sequences, calculating average swimming speed, motion disorder, spatial clustering degree, and water surface activity ratio; the average swimming speed is represented as the average displacement speed of all fish within a time window; the motion disorder degree is represented as the standard deviation of the trajectory positions of all fish; the spatial clustering degree is represented as the proportion of fish forming a cluster; and the water surface activity ratio is represented as the proportion of trajectory points located within a preset "water surface area" at the top of the screen.

[0054] It should be noted that the process of calculating the average swimming speed is as follows: First, based on the movement trajectory sequence, calculate the displacement distance of each fish within a given time window; then, divide this displacement distance by the duration of the time window to obtain the instantaneous speed or average speed of a single fish; finally, calculate the arithmetic mean of this speed value for all fish. The average swimming speed reflects the overall activity level of the fish school during the observation period and is a fundamental indicator for assessing the intensity of fish school movement.

[0055] It should be noted that when calculating the degree of dysregulation, the method for combining the standard deviations of the two directions can be to calculate the square root of the sum of their squares (i.e., the standard deviation of the Euclidean distance of the location point relative to its average position), or to perform a weighted summation of the standard deviations of the two directions. The degree of dysregulation is a quantitative indicator of the spatial dispersion of a fish school. The larger the value, the more dispersed the fish members are in space, the more inconsistent their overall movement directions are, and the more disordered their behavior tends to be.

[0056] It should be noted that the spatial clustering degree uses the DBSCAN clustering algorithm (a density-based spatial clustering algorithm with noise). This algorithm clusters samples based on their density, without requiring a pre-specified number of clusters. It groups densely connected sample points into clusters and identifies discrete points (noise points) far from any cluster. This algorithm can distinguish whether a fish school exhibits a clustered or discrete state. Furthermore, spatial clustering degree represents the proportion of individuals forming effective clusters within the total population. A higher value indicates more pronounced social behaviors within the fish school (such as cooperative feeding and collective avoidance); a very low value may indicate a loss of organization and abnormal individual behavior within the fish school.

[0057] It should be noted that the preset water surface area usually refers to the rectangular or strip-shaped area near the top of the video frame, representing the surface layer of the water. Statistical analysis of the water surface activity ratio calculates the frequency of "surfacing" behavior. This is one of the key behavioral indicators for determining whether dissolved oxygen in the water is sufficient; an abnormally high ratio often directly indicates an emergency situation such as hypoxia or stress.

[0058] S3. Calculation of water body symptom index and biological stress index.

[0059] In a specific example, the calculation of the water body symptom index and the biological stress index includes: calculating the water body symptom index based on the water body characteristic parameters and a preset water body symptom index algorithm; and generating the biological stress index by fusing the video behavior features and audio behavior features using a preset biological stress index algorithm.

[0060] In a specific example, the calculation of the water body symptom index based on the water body characteristic parameters and a preset water body symptom index algorithm includes: calculating the water body symptom index based on the dissolved oxygen change slope, ammonia nitrogen concentration, and pH deviation in the water body characteristic parameter set, using the preset calculation formula of the water body symptom index algorithm. The water symptom index was obtained. , Expressed as the slope of the change in dissolved oxygen, This represents the current measured value of ammonia nitrogen concentration. This represents the preset upper limit of the safe threshold for ammonia nitrogen concentration. This is expressed as pH deviation. , and These are respectively represented as the weighting factor corresponding to the slope of dissolved oxygen change, the weighting factor corresponding to ammonia nitrogen concentration, and the weighting factor corresponding to pH deviation.

[0061] It should be noted that, , , , We obtained the weighting factors corresponding to the slope of dissolved oxygen change, the ammonia nitrogen concentration, and the pH deviation through factor analysis. First, we condensed the information of dissolved oxygen change slope, ammonia nitrogen concentration, and pH deviation, and then obtained the variance explained after rotation. We obtained the weights by dividing the cumulative variance explained.

[0062] It should be noted that factor analysis is a well-known technique. It is a multivariate statistical analysis method that starts by studying the internal dependencies of variables and reduces some variables with complex relationships to a few comprehensive factors. Information condensation is expressed as the calculation of the median. The variance explained rate is the amount of information extracted by the factors. Variance explained rate = eigenvalues ​​ / total number of analysis terms. The rotated variance explained rate is expressed as the variance explained by the factors after maximum variance rotation.

[0063] It should be noted that the Water Symptom Index is a comprehensive quantitative indicator of water quality anomalies. The higher the value, the more severe the deviation of the water body from normal conditions in the three core physicochemical dimensions of dissolved oxygen dynamics, ammonia nitrogen pollution load, and pH stability.

[0064] In a specific example, the step of generating a biological stress index by fusing the video and audio behavioral features using a preset biological stress index algorithm includes: calculating the biological stress index using the preset biological stress index algorithm formula. Biological stress index was derived ,in Represented as video behavioral features, Represented as audio behavior features. This is represented by the maximum value of the audio behavior feature. and These are represented as the weighting factors corresponding to video behavior features and the weighting factors corresponding to audio behavior features.

[0065] It should be noted that the weight factors corresponding to video behavior features and audio behavior features are obtained through factor analysis. First, the information of video behavior features and audio behavior features is condensed, and then the variance explained after rotation is obtained. The weights are obtained by dividing the cumulative variance explained.

[0066] It should be noted that the biological stress index is a comprehensive indicator that integrates visual and auditory behavioral monitoring information. A higher value indicates a stronger stress response in the fish population, specifically manifested in the degree of disorder in individual movement and a decrease in feeding activity. By weighted fusion of behavioral characteristics from two different modalities, the limitations of single-sensor information are overcome, enabling more sensitive and comprehensive capture of abnormal behavioral signals in fish caused by environmental discomfort.

[0067] This application constructs two independent comprehensive quantitative indices—the water body symptom index and the biological stress index—to quantify anomalies from two independent but related channels: “environmental physical and chemical stress” and “biological behavioral response.” It condenses complex multi-parameter information into two intuitive quantitative indicators, which facilitates logical judgment and risk classification by the system and makes it easier for managers to understand.

[0068] S4. Dynamic early warning threshold comparison and anomaly determination.

[0069] In a specific instance, the dynamic early warning threshold comparison and anomaly determination includes: performing a numerical comparison operation based on the water body symptom index and a preset water body symptom index early warning threshold to obtain a water body anomaly indicator.

[0070] Based on the biological stress index and the preset biological stress index warning threshold, a numerical comparison operation is performed to obtain biological abnormality markers.

[0071] Based on water body anomaly markers and biological anomaly markers, a decision is made through anomaly level classification logic to generate the anomaly level.

[0072] It should be noted that the numerical comparison operation based on the water symptom index and the preset water symptom index warning threshold is implemented by directly comparing the real-time calculated value of the water symptom index with a preset water symptom index warning threshold, which may be dynamically adjusted. If the water symptom index is greater than or equal to the preset water symptom index warning threshold, it is determined that the water physicochemical parameters are abnormal, and an abnormal water symptom flag (e.g., logical value 1) is output; otherwise, it is determined that the water physicochemical parameters are within the normal range, and a normal water symptom flag (e.g., logical value 0) is output. Furthermore, the water symptom flag is a binary logical flag (whether the current water symptom index exceeds the safety warning line). When the flag is abnormal, it indicates that one or more core water quality parameters, such as dissolved oxygen changes, ammonia nitrogen concentration, or pH deviation, have deviated from the normal range, and the aquatic environment faces potential chemical stress risks.

[0073] It should be noted that the numerical comparison operation based on the biological stress index and the preset biological stress index warning threshold is a judgment of the degree of abnormality in the behavioral monitoring dimension. The real-time calculated value of the biological stress index is compared with the preset biological stress index warning threshold. If the biological stress index is greater than or equal to the preset biological stress index warning threshold, it is determined that the fish behavior is under significant stress, and a biological anomaly flag (e.g., logical value 1) is output to indicate abnormality; otherwise, it is determined that the fish behavior is within the normal fluctuation range, and a biological anomaly flag (e.g., logical value 0) is output to indicate normality. Furthermore, the biological anomaly flag is a binary logical flag (whether the current biological stress index exceeds the threshold of normal behavioral fluctuation). When the flag is abnormal, it indicates that the fish school is exhibiting statistically significant abnormal patterns in its movement state (such as swimming speed and turbulence) or feeding acoustic activity, which is a biological behavioral response signal of fish to environmental discomfort.

[0074] It should be noted that, based on the aforementioned water anomaly markers and biological anomaly markers, decision-making is made through anomaly level classification logic. This is implemented by executing preset multi-level judgment rules based on the combined state of the two anomaly markers to comprehensively assess the severity of the anomaly. The anomaly level classification logic formula is used... Generate anomaly level ,in Indicated as a water body anomaly marker, This is represented as a biological anomaly marker.

[0075] It should be noted that the anomaly level is a discrete grading index (an overall environmental risk assessment result that integrates abnormalities in aquatic chemical parameters and abnormal fish biological behavior). An anomaly level of 0 represents a normal environmental state; an anomaly level of 1 represents an anomaly in only one dimension (aquatic body or organisms), indicating the need to pay attention to potential risks; an anomaly level of 2 represents anomalies in both dimensions simultaneously, indicating that environmental stress has been transmitted from physicochemical parameters to the organisms and has triggered a significant stress response, representing the highest level of risk and requiring immediate intervention. This level provides a clear decision-making basis for subsequent implementation of differentiated alarm and control strategies.

[0076] This application makes decisions based on a logical combination of water anomaly markers and biological anomaly markers. By integrating physicochemical and biological evidence through logical rules, the anomaly determination is no longer a simple threshold crossing, but a comprehensive reasoning based on multiple evidence chains, which greatly improves the accuracy and intelligence of the determination. The clear three-level classification provides a direct and clear basis for taking differentiated response measures, realizing the leap from "whether to alarm" to "according to what level of urgency to alarm".

[0077] S5, Execution of hierarchical alarm strategy.

[0078] In a specific example, the execution of the hierarchical alarm strategy includes: based on the anomaly determination flag and the anomaly level, executing a preset hierarchical alarm strategy decision to obtain the alarm strategy type to be executed; when the anomaly level is equal to the preset first-level warning level, triggering the execution of the first-level warning strategy; when the anomaly level is equal to the preset second-level emergency level, triggering the execution of the second-level emergency alarm strategy.

[0079] It should be noted that, based on the aforementioned anomaly determination flag and anomaly level, the preset hierarchical alarm strategy decision is executed. This is achieved by reading the specific numerical value of the anomaly level and, according to a preset decision mapping relationship, mapping different anomaly level values ​​to corresponding alarm strategy types to be executed. Further, the decision mapping relationship is a preset conditional judgment rule; for example, if the anomaly level = 1, it is mapped to a Level 1 early warning strategy type; if the anomaly level = 2, it is mapped to a Level 2 emergency alarm strategy type. Based on the comprehensively assessed environmental risk level, the alarm and intervention intensity is intelligently selected to match it.

[0080] It should be noted that the Level 1 early warning strategy sends a Level 1 early warning notification to the pre-set administrator's mobile terminal. This is achieved by pushing text or voice information containing a brief description of the anomaly, the anomaly level, a timestamp, and suggested inspection items to the administrator's terminal via a wireless communication network. Furthermore, the purpose of this action is to promptly and remotely notify management personnel to pay attention and conduct preliminary verification when a primary or single-dimensional environmental anomaly is detected, achieving immediacy of the early warning. Further, the automatic control command to activate pre-set remedial equipment such as aerators is implemented by sending activation commands to environmental control equipment such as aerators connected to the fishpond via an IoT control module based on the trigger signal of the Level 1 early warning strategy. In essence, this action automatically takes preliminary physical remedial measures when the water body is determined to have a potential risk of hypoxia or mild pollution, attempting to improve the dissolved oxygen status of the water body and prevent further deterioration of the environmental anomaly, demonstrating the method's automatic response and preliminary intervention capabilities.

[0081] It should be noted that triggering the Level 2 emergency alarm is achieved by generating a high-priority emergency alarm signal after the system determines that the anomaly level has reached the highest level. This signal will activate all preset emergency alarm channels within the system. The system will also activate the on-site audible and visual alarms by simultaneously sending the Level 2 emergency alarm signal to the audible and visual alarm devices deployed at the fishpond site via wired or wireless means, driving them to emit a high-pitched alarm and flashing lights. Furthermore, the purpose of this action is to generate a strong visual and auditory warning at the accident site when it is determined that a serious anomaly has occurred in the fishpond environment and caused significant stress to the fish, thereby maximizing the attention of on-site personnel and buying time for emergency manual intervention. This serves as the last automated alarm barrier in response to high-risk situations.

[0082] Please see Figure 2 As shown, in its second aspect, this application provides a system for detecting and automatically alarming abnormalities in fishpond water environment.

[0083] The system 100 of the fishpond water environment anomaly detection and automatic alarm method described in this invention can be installed in an electronic device. Depending on the functions implemented, the system 100 may include a data acquisition module 101, a feature analysis module 102, a water symptom index and biological stress index calculation module 103, a dynamic early warning threshold comparison and anomaly judgment module 104, and a graded alarm strategy execution module 105. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0084] In this embodiment, the functions of each module / unit are as follows: The data acquisition module generates standardized water quality data, fish movement data sequence, and audio energy time series data based on the pre-acquired fishpond water physicochemical parameter sequence, fish activity video stream, and environmental audio stream.

[0085] The feature analysis module, based on the standardized water quality data, fish movement data sequences, and audio energy time series data, analyzes and derives key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics.

[0086] Module for calculating water symptom index and biological stress index.

[0087] Dynamic early warning threshold comparison and anomaly detection module.

[0088] Tiered alarm strategy execution module.

[0089] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0090] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0091] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0092] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0093] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting and automatically alarming abnormal water environment in fishponds, characterized in that, include: S1. Based on the pre-acquired sequence of physicochemical parameters of fishpond water, video stream of fish activity and environmental audio stream, generate standardized water quality data, fish movement data sequence and audio energy time series data; S2. Based on the standardized water quality data, fish movement data sequence, and audio energy time series data, key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics are analyzed and obtained. S3, Calculation of Water Body Symptom Index and Biological Stress Index; S4. Dynamic early warning threshold comparison and anomaly determination; S5, Execution of hierarchical alarm strategy.

2. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 1, characterized in that, The process generates standardized water quality data, fish movement data sequences, and audio energy time-series data based on pre-acquired fishpond water physicochemical parameter sequences, fish activity video streams, and environmental audio streams, including: The system acquires a sequence of physicochemical parameters of the fishpond water collected by pre-deployed sensors, performs missing value imputation and standardization on the sequence to generate standardized water quality data; it acquires a video stream of fish activity captured by a pre-deployed camera, extracts keyframes from the video stream and performs background subtraction and target detection to generate a fish motion data sequence; and it acquires an ambient audio stream collected by a pre-deployed microphone, performs noise reduction and energy extraction of specific frequency bands on the ambient audio stream to generate audio energy time-series data.

3. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 1, characterized in that, Based on the standardized water quality data, fish movement data sequences, and audio energy time-series data, key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics are analyzed and derived, including: Based on standardized water quality data, key parameters were extracted from time-series dissolved oxygen concentration data, ammonia nitrogen concentration data, and pH value data to obtain a set of water body characteristic parameters. Based on fish movement data sequences, video behavioral features are extracted to obtain video behavioral characteristics; Based on audio energy time-series data, the average intensity of the energy data of feeding activity frequency bands per unit time is calculated to obtain audio behavior characteristics.

4. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 3, characterized in that, The extraction of video behavioral features based on fish movement data sequences to obtain video behavioral features includes: Video behavior feature extraction is based on fish movement data sequences, calculating average swimming speed, movement disorder, spatial clustering degree, and surface activity ratio. Average swimming speed is represented by the average displacement speed of all fish within the time window; movement disorder is represented by the standard deviation of the trajectory positions of all fish; spatial clustering degree is represented by the proportion of fish forming a cluster; and surface activity ratio is represented by the proportion of trajectory points located within the preset "water surface area" at the top of the screen.

5. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 1, characterized in that, The calculation of the water symptom index and biological stress index includes: Based on the water body characteristic parameters and the preset water body symptom index algorithm, the water body symptom index is calculated; based on the video behavior characteristics and audio behavior characteristics, the biological stress index is generated by fusion calculation through the preset biological stress index algorithm.

6. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 5, characterized in that, The water body symptom index is calculated based on the water body characteristic parameters and a preset water body symptom index algorithm, including: Based on the slope of dissolved oxygen change, ammonia nitrogen concentration, and pH deviation from the set of water body characteristic parameters, the water body symptom index is calculated using a pre-defined algorithm. The water symptom index was obtained. , Expressed as the slope of the change in dissolved oxygen, This represents the current measured value of ammonia nitrogen concentration. This represents the preset upper limit of the safe threshold for ammonia nitrogen concentration. This is expressed as pH deviation. , and These represent the weighting factors corresponding to the slope of dissolved oxygen change, the weighting factor corresponding to ammonia nitrogen concentration, and the weighting factor corresponding to pH deviation, respectively.

7. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 5, characterized in that, The process of generating a biological stress index by fusing the video and audio behavioral features using a preset biological stress index algorithm includes: The calculation formula of the biological stress index algorithm is established. Biological stress index was derived ,in Represented as video behavioral features, Represented as audio behavior features. This is represented by the maximum value of the audio behavior feature. and These are represented as the weighting factors corresponding to video behavior features and the weighting factors corresponding to audio behavior features.

8. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 1, characterized in that, The dynamic early warning threshold comparison and anomaly determination include: Based on the water body symptom index and the preset water body symptom index warning threshold, a numerical comparison operation is performed to obtain water body abnormality indicators. Based on the biological stress index and the preset biological stress index warning threshold, a numerical comparison operation is performed to obtain biological abnormality markers. Based on water body anomaly markers and biological anomaly markers, a decision is made through anomaly level classification logic to generate the anomaly level.

9. The method for detecting and automatically alarming abnormal water environment in fishponds according to claim 1, characterized in that, The execution of the tiered alarm strategy includes: Based on the anomaly determination flag and anomaly level, a preset hierarchical alarm strategy decision is executed to obtain the alarm strategy type to be executed; when the anomaly level is equal to the preset first-level warning level, the first-level warning strategy is triggered; when the anomaly level is equal to the preset second-level emergency level, the second-level emergency alarm strategy is triggered.

10. A method and system for detecting and automatically alarming abnormal water environment in fishponds according to any one of claims 1-9, characterized in that, include: The data acquisition module generates standardized water quality data, fish movement data sequences, and audio energy time-series data based on the pre-acquired fishpond water physicochemical parameter sequences, fish activity video streams, and environmental audio streams. The feature analysis module, based on the standardized water quality data, fish movement data sequence, and audio energy time series data, analyzes and derives key water body characteristic parameters, video behavior characteristics, and audio behavior characteristics. Module for calculating water body symptom index and biological stress index; Dynamic early warning threshold comparison and anomaly detection module; Tiered alarm strategy execution module.

Images