Intelligent perception-based precise feeding method and system for fermented feed of procambarus clarkii

Abstract

The application discloses a precise feeding method and system for red claw crayfish fermented feed based on intelligent sensing, and particularly relates to the field of automatic control technology, and is used for solving the problem that the existing automatic feeding method based on threshold triggering cannot effectively handle the dynamic changes of multiple water quality parameter coupling, nonlinearity and time delay characteristics caused by the allelopathy effect of fermented feed, resulting in inaccurate feeding decision; by collecting multiple water quality parameter data in real time to determine whether an alternating fluctuation relationship occurs, and matching it with a predefined dominant cause category to determine the main cause, the evolution trend of the current state is analyzed in the phase space composed of water quality parameters, the future water quality comprehensive change is predicted based on the trend, and the feeding control instruction is generated and executed according to the deviation degree of the prediction result and the water quality interval of the shrimp group feeding demand; the prospective analysis and precise feedforward control of the complex dynamic water quality system are realized, and the precision of the feeding decision and the real state of the breeding water body and the demand of the shrimp group are improved.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, and more specifically, to a method and system for precise feeding of fermented feed for red claw crayfish based on intelligent sensing. Background Technology

[0002] In the industrialized farming of redclaw crayfish, automating and precisely controlling feed feeding is a key aspect of improving farming efficiency. Existing technologies have disclosed various automatic feeding systems based on environmental sensing. These systems typically use sensors deployed in the aquaculture water to collect key water quality parameters such as dissolved oxygen and temperature in real time, and transmit this data to a controller. The controller has preset feeding trigger thresholds or simple control logic corresponding to each parameter. When the sensor data reaches specific conditions, it drives the feeder to perform quantitative feeding actions, aiming to replace human experience and achieve automatic triggering and stopping of feeding through technical means.

[0003] However, when the aforementioned automation methods based on single-point threshold triggering or simple linear feedback control are applied to the feeding scenario of fermented feed, their control effectiveness faces challenges. The complex allelopathic effects triggered by the introduction of fermented feed into the water body lead to interrelated and nonlinear dynamic changes in multiple water quality parameters such as dissolved oxygen, pH, and ammonia nitrogen. Moreover, these changes have a significant time lag relative to the feeding action. Existing automation control methods, whose control models are usually based on the assumption of independent parameter changes or instantaneous responses, cannot accurately describe and handle this dynamic process of a multi-parameter coupled, nonlinear system with time lag characteristics. The direct consequence is that the feeding decisions made by the control system based on instantaneous or single-point sensor data are often seriously out of sync with the actual dynamic state of the water environment and the actual feeding needs of the shrimp population. This not only makes it difficult to achieve the design goal of precise feeding but may also cause water quality fluctuations due to misjudgment, thus affecting the stability of aquaculture. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for precise feeding of fermented feed for red claw crayfish based on intelligent sensing to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing includes the following steps:

[0007] S1. Data on multiple water quality parameters related to the allelopathic effect of fermented feed are collected in real time by multiple sensors deployed in the aquaculture water body.

[0008] S2. Based on the data of multiple water quality parameters collected at present, determine whether there is an alternating fluctuation relationship in the current water quality status;

[0009] S3. If an alternating fluctuation relationship is determined, it is matched with multiple predefined dominant cause categories to determine the dominant cause category with the highest degree of correlation.

[0010] S4. Based on the dominant cause category with the highest correlation, analyze the evolution trend of the current water quality status in the phase space composed of multiple water quality parameters.

[0011] S5. Based on evolutionary trends, predict the comprehensive change trend of multiple water quality parameters within a set future time period;

[0012] S6. Based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs, generate and execute feeding control instructions for fermented feed.

[0013] Furthermore, multiple sensors deployed in the aquaculture water body are used to collect data in real time on several water quality parameters related to the allelopathic effects of fermented feed, including:

[0014] Multiple water quality parameters are periodically sampled simultaneously using multiple sensors.

[0015] Within each sampling period, acquire the sensor readings corresponding to each water quality parameter;

[0016] Real-time noise filtering is performed on the sensor readings of each acquired water quality parameter;

[0017] The sensor readings for each water quality parameter, after noise filtering, are combined with the time information of the corresponding sampling period to form data for multiple water quality parameters.

[0018] Furthermore, based on the data from multiple water quality parameters currently collected, it is determined whether the current water quality status exhibits an alternating fluctuation relationship, including:

[0019] Select at least two key water quality parameters from multiple water quality parameters;

[0020] Analyze the direction of change of at least two key water quality parameters over a set number of consecutive sampling periods;

[0021] If, within a set number of consecutive sampling periods, the number of times the data changes in the opposite direction of at least two key water quality parameters reach a preset threshold for the number of times the changes are reversed, it is preliminarily determined that there are alternating fluctuation characteristics.

[0022] Verify whether the alternating fluctuation characteristics persist within a set number of subsequent sampling periods;

[0023] If this continues, it indicates that the current water quality is exhibiting alternating fluctuations.

[0024] Furthermore, if an alternating fluctuation relationship is determined, it is matched with multiple predefined dominant cause categories to identify the dominant cause category with the highest degree of correlation, including:

[0025] Extract the characteristic information of the alternating fluctuation relationship. The characteristic information should include at least the combination of key water quality parameters that cause the alternating fluctuation and the alternation period.

[0026] The similarity between the feature information and the feature pattern corresponding to each predefined dominant cause category is calculated;

[0027] The dominant cause category with the highest similarity calculation result is determined as the dominant cause category with the highest degree of association.

[0028] Furthermore, the establishment of feature patterns corresponding to multiple predefined dominant cause categories includes:

[0029] Collect examples of multiple alternating fluctuations that occurred in historical aquaculture cycles and their corresponding aquaculture environment records;

[0030] Based on common factors in aquaculture environment records, multiple instances of alternating fluctuations were clustered into several dominant cause categories;

[0031] Statistical features of all alternating fluctuation relationship instances under each dominant cause category in terms of key water quality parameter combinations and alternation cycles are extracted to form a feature pattern.

[0032] Furthermore, based on the dominant cause category with the highest correlation, the evolution trend of the current water quality state is analyzed in a phase space composed of multiple water quality parameters, including:

[0033] Based on the dominant cause category with the highest correlation, obtain the reference trajectory set formed in phase space by the historical water quality parameter data associated with the dominant cause category;

[0034] Based on the data of multiple water quality parameters collected at present, determine the current state point in phase space corresponding to the current water quality state;

[0035] Calculate the directional deviation between the current state point and each historical trajectory in the reference trajectory set within the neighborhood of the current position;

[0036] Based on the statistical results of directional deviation, the most likely direction of movement of the current water quality state in phase space is determined as the evolution trend.

[0037] Furthermore, obtaining the set of reference trajectories formed in phase space by historical water quality parameter data associated with the dominant cause category includes:

[0038] Filter out all historical water quality parameter data segments from the historical database that are marked as belonging to the dominant cause category with the highest correlation.

[0039] Each historical water quality parameter data segment is mapped to a phase space composed of multiple water quality parameters to form a historical trajectory;

[0040] The water quality parameter data within a set length prior to the current state point corresponding to the current water quality state are also mapped to the phase space and used as the recent trajectory;

[0041] Multiple historical trajectories are combined with recent trajectories to form a reference trajectory set.

[0042] Furthermore, based on evolutionary trends, the comprehensive changing trends of multiple water quality parameters within a future set time period are predicted, including:

[0043] Based on the most likely direction of motion indicated by the evolutionary trend, the motion is extrapolated forward in phase space along the corresponding direction of motion;

[0044] The inference step size is determined by combining the historical water quality parameter change rate characteristics corresponding to the dominant cause category with the highest correlation.

[0045] During the forward extrapolation process, the direction of extrapolation is dynamically adjusted based on the stability of the system state reflected by the historical trajectory in phase space.

[0046] The phase space state point corresponding to the end of the set future time period will be extrapolated and mapped in reverse to the predicted values ​​of multiple water quality parameters, forming a comprehensive trend of change.

[0047] Furthermore, based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs, feeding control instructions for fermented feed are generated and executed, including:

[0048] Calculate the degree of deviation between the predicted values ​​of multiple water quality parameters in the overall trend and the preset boundary values ​​of the water quality range for shrimp feeding requirements;

[0049] Based on the calculated deviation, the preset control strategy mapping table is consulted to determine the corresponding feeding control parameters;

[0050] Based on the determined feeding control parameters, a feeding control instruction for fermented feed, including feeding amount and feeding timing, is generated.

[0051] The generated fermented feed feeding control command is sent to the feeder to drive the feeder to perform the feeding action.

[0052] On the other hand, the present invention provides a precision feeding system for fermented feed of red claw crayfish based on intelligent sensing, comprising the following modules:

[0053] The sensor acquisition module is used to collect data on multiple water quality parameters related to the allelopathic effect of fermented feed in real time through multiple sensors deployed in the aquaculture water.

[0054] The fluctuation discrimination module is used to determine whether the current water quality status shows an alternating fluctuation relationship based on the data of multiple water quality parameters collected at the moment;

[0055] The trigger matching module is used to perform matching analysis with multiple predefined dominant trigger categories if an alternating fluctuation relationship is determined, in order to determine the dominant trigger category with the highest degree of correlation.

[0056] The trend analysis module is used to analyze the evolution trend of the current water quality status in a phase space composed of multiple water quality parameters, based on the dominant cause category with the highest correlation.

[0057] The water quality prediction module is used to predict the comprehensive change trend of multiple water quality parameters within a set future time period based on evolution trends.

[0058] The feeding control module is used to generate and execute feeding control instructions for fermented feed based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. By identifying and analyzing the alternating fluctuations among water quality parameters, the inherent coupled oscillation patterns caused by the allelopathic effects of fermented feed can be captured from complex real-time data. This identification breaks away from the limitations of traditional methods that only focus on whether a single parameter exceeds a threshold, laying a precise perceptual foundation for subsequent processing of nonlinear and multi-parameter interactive system dynamics. By matching this fluctuation pattern with a variety of predefined dominant cause categories, the most likely root cause of the current abnormal dynamics of the system can be traced, making control decisions not simply responses to appearances, but differentiated responses to different disturbance causes, significantly improving the pertinence and root cause of control strategies.

[0061] 2. Analyzing the evolution trend of water quality in phase space involves geometrically representing and extrapolating the dynamics of a multidimensional, coupled system as a whole. This allows for the essential description and prediction of the continuous evolution of the system state, effectively overcoming the decision-making lag problem caused by time delay effects in traditional methods. Based on this evolution trend, the prediction of future comprehensive water quality changes enables a forward-looking judgment of the system's development trend. Feeding instructions are generated based on the deviation between the prediction results and the ideal demand range, transforming feeding control from a passive response based on past or instantaneous states into a predictive, proactive, and precise feedforward intervention. This forms a closed-loop control method capable of adaptively handling complex dynamic systems with multivariable coupling, nonlinearity, and time delay characteristics. At the automatic control level, this achieves precise temporal and spatial synchronization and matching between feeding actions and the actual dynamics of the water body and the feeding needs of shrimp populations, improving the accuracy, stability, and reliability of control. Attached Figure Description

[0062] Figure 1 This is a flowchart of the precise feeding method for fermented feed of redclaw crayfish based on intelligent sensing according to the present invention.

[0063] Figure 2 This is a schematic diagram of the structure of the intelligent sensing-based precision feeding system for fermented feed of red claw crayfish according to the present invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0065] Example 1: Figure 1 This invention presents a method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing, which includes the following steps:

[0066] S1. Data on multiple water quality parameters related to the allelopathic effect of fermented feed are collected in real time by multiple sensors deployed in the aquaculture water body.

[0067] S2. Based on the data of multiple water quality parameters collected at present, determine whether there is an alternating fluctuation relationship in the current water quality status;

[0068] S3. If an alternating fluctuation relationship is determined, it is matched with multiple predefined dominant cause categories to determine the dominant cause category with the highest degree of correlation.

[0069] S4. Based on the dominant cause category with the highest correlation, analyze the evolution trend of the current water quality status in the phase space composed of multiple water quality parameters.

[0070] S5. Based on evolutionary trends, predict the comprehensive change trend of multiple water quality parameters within a set future time period;

[0071] S6. Based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs, generate and execute feeding control instructions for fermented feed.

[0072] S1. By deploying multiple sensors in the aquaculture water body, data on multiple water quality parameters related to the allelopathic effect of fermented feed are collected in real time, including:

[0073] To achieve accurate perception of the water body used for redclaw crayfish farming, it is necessary to collect data on multiple water quality parameters related to the allelopathic effects of fermented feed in real time. Water quality parameters sensitive to microbial metabolic activity and organic matter decomposition processes were selected as the multiple parameters to be collected. These parameters include dissolved oxygen concentration, pH value, ammonia nitrogen concentration, nitrite concentration, and redox potential. The reason for selecting these parameters is that after fermented feed is added to the water body, the organic matter and microorganisms it contains will significantly affect the oxygen consumption and reoxygenation balance, acid-base balance, and nitrogen cycle of the water body. Dissolved oxygen concentration directly reflects the oxygen consumption degree of aerobic microbial metabolism; pH value changes are related to the production of organic acids and the conversion of ammonia nitrogen; ammonia nitrogen and nitrite concentrations are key intermediate product indicators of nitrogen-containing organic matter decomposition; and redox potential comprehensively characterizes the overall biochemical reaction environment of the water body.

[0074] For each water quality parameter, a dedicated sensor is configured. A fluorescence dissolved oxygen sensor measures dissolved oxygen concentration, a glass electrode pH sensor measures pH value, an ion-selective electrode sensor measures ammonia nitrogen and nitrite concentrations, and a platinum electrode sensor measures oxidation-reduction potential. These sensors are deployed in different spatial locations within the aquaculture water body according to their measurement principles and the hydrodynamic characteristics of the water. The dissolved oxygen and pH sensors are placed in the middle layer of the water body, away from the direct water inlet and aeration plate. The ammonia nitrogen and nitrite sensors are placed near the bottom of the pond, and the oxidation-reduction potential sensor is placed in the lower middle layer of the water body. The sensing probes of all sensors are submerged at a preset measurement depth below the water surface. Each sensor is connected to a data acquisition unit located at the edge of the pond via a waterproof cable. This data acquisition unit provides an independent signal channel for each sensor.

[0075] The data acquisition unit drives all sensors to perform synchronous periodic sampling according to a unified and fixed time reference. The time interval of this periodic sampling is called the sampling period, which is set according to the typical time scale of the occurrence and development of allelopathic effects in fermented feed. Considering that the response of water quality parameters to the feeding of fermented feed may change significantly from several minutes to several hours, the sampling period is set to, for example, 5 minutes. At the end of each sampling period, the data acquisition unit sequentially reads and acquires the raw electrical signal output by the sensor corresponding to each water quality parameter through its analog-to-digital conversion channel. This raw electrical signal is the sensor reading corresponding to each water quality parameter, such as acquiring the current value output by the dissolved oxygen sensor, the voltage value output by the pH sensor, and the potential value output by the ion-selective electrode sensor.

[0076] Due to interference in the aquaculture water environment, the sensor readings of each water quality parameter may contain noise. Real-time noise filtering is performed on the sensor readings of each water quality parameter using a moving average filter combined with outlier removal. For each water quality parameter data stream, a first-in-first-out queue containing sensor readings for the parameter from the most recent sampling periods is maintained. Whenever a new sensor reading is acquired, it is added to the end of the queue, and the median and mean absolute difference (MAD) of all values ​​in the queue are calculated. If the difference between the new sensor reading and the median exceeds a preset multiple of the MAD, for example, more than 3 times the MAD, it is considered an outlier, removed, and replaced with the median of the remaining values ​​in the queue. If the new value is not an outlier, it is retained. Subsequently, the arithmetic mean of all valid values ​​in the queue is calculated, and this average is used as the valid reading of the water quality parameter in the current sampling period after real-time noise filtering. The queue length is, for example, 10 sampling periods, and the preset multiple is, for example, 3 times. This method effectively smooths random impulse interference and preserves the true trend of water quality parameter changes.

[0077] The effective reading of each water quality parameter in the current sampling period, after noise filtering, is bound to the precise time information corresponding to that sampling period. The data acquisition unit has a high-precision real-time clock that records the current year, month, day, hour, minute, and second information as a timestamp when each sampling period is triggered. This timestamp, along with the readings of all water quality parameters collected and processed at the same time, is packaged together to form a structured data record. This data record contains a timestamp field and numerical fields for multiple water quality parameters, including dissolved oxygen concentration, pH value, ammonia nitrogen concentration, nitrite concentration, and redox potential. This data record is stored in the non-volatile memory of the data acquisition unit and transmitted to the backend server via a communication network. This completes the process of real-time acquisition of multiple water quality parameters related to the allelopathic effects of fermented feed, obtaining a data set of multiple water quality parameters that are time-synchronized and noise-filtered.

[0078] S2. Based on the data of multiple water quality parameters collected so far, determine whether the current water quality status shows an alternating fluctuation relationship. This includes:

[0079] At least two key water quality parameters are selected from multiple water quality parameters. The selection criteria are the sensitivity of water quality parameters to the allelopathic effects of fermented feed and the historical correlation patterns between parameters. The absolute value of the correlation coefficient of each pair of parameters in the data sequence within a recent time window is calculated. The recent time window is, for example, the last 24 hours. The two water quality parameters with the largest absolute values ​​of correlation coefficients are selected as the initial candidate key water quality parameters. At the same time, the physiological and ecological significance of the changes in water quality parameters is considered, and parameter pairs that have direct biochemical links in the decomposition of fermented feed and often show antagonistic or synergistic effects are given priority. The final key water quality parameters include at least dissolved oxygen concentration and ammonia nitrogen concentration, because dissolved oxygen concentration and ammonia nitrogen concentration are directly related to the oxygen consumption process and the nitrogen-containing product release process, respectively, in the decomposition of organic matter.

[0080] After selecting at least two key water quality parameters, analyze the direction of data change for these parameters over a set number of consecutive sampling periods. The direction of data change refers to whether the water quality parameter value increases, decreases, or remains essentially unchanged from one sampling period to the next. The criterion for determining whether it remains essentially unchanged is that the absolute value of the difference between the values ​​of two adjacent sampling periods is less than a multiple of the sensor's measurement accuracy for that water quality parameter; for example, if it is less than three times the measurement accuracy, it is considered unchanged. The number of consecutive sampling periods needs to be sufficient to cover a potential fluctuation cycle, and its setting is based on the shortest duration of alternating fluctuations typically observed in historical observations. For example, if the sampling period is 5 minutes and historical data shows that the fluctuation cycle is usually not less than 30 minutes, then the number of consecutive sampling periods can be set to 6, corresponding to a 30-minute data window. During the analysis, traverse the set number of consecutive sampling periods in chronological order, recording the direction of change for each key water quality parameter relative to the previous period in each period, forming a sequence of the direction of change for each key water quality parameter.

[0081] If, within a set number of consecutive sampling periods, the number of times that the data changes in the opposite direction of at least two key water quality parameters reach a preset threshold for the number of times they change in opposite directions, then an alternating fluctuation characteristic is preliminarily determined. An opposite change refers to the opposite direction of change of two key water quality parameters between the same pair of adjacent sampling periods; for example, one key water quality parameter increases while the other decreases. The preset threshold for the number of times they change in opposite directions is set to avoid misjudging occasional, sporadic opposite changes as regular alternating fluctuations. This threshold is set based on probability statistics combined with experience. For example, if the set number of consecutive sampling periods is 6, there may be a maximum of 5 opportunities to compare adjacent periods. If at least 3 opposite changes are required for a preliminary determination, it means that opposite changes have occurred in more than half of the comparison opportunities. The specific determination process involves counting the number of times dissolved oxygen concentration and ammonia nitrogen concentration show opposite changes in opposite directions in the 5 adjacent period comparisons corresponding to the selected 6 consecutive sampling periods. If the count is greater than or equal to the preset threshold of 3 times, an alternating fluctuation characteristic is preliminarily determined.

[0082] After initially determining the existence of alternating fluctuation characteristics, it is necessary to verify whether these characteristics persist within a set number of subsequent sampling periods. The number of subsequent sampling periods can be the same as the number used in the initial determination, for example, 6 sampling periods. The verification method is to move the analysis window forward and perform the same analysis steps again for the next 6 consecutive sampling periods. Analyze the direction of change in dissolved oxygen concentration and ammonia nitrogen concentration within the next 6 consecutive sampling periods and count the number of times reverse changes occur. If the number of reverse changes counted in the subsequent window reaches or exceeds the preset reverse number threshold of 3 times again, then the alternating fluctuation characteristics are considered to persist.

[0083] If the alternating fluctuation characteristics are verified to continue to appear within a set number of subsequent sampling periods, it is finally determined that the current water quality status exhibits an alternating fluctuation relationship. The entire judgment process is based on clear quantitative rules, including rules for selecting key parameters, rules for analyzing the direction of change, rules for counting the number of reversals, and rules for continuous verification. The parameters such as the number of consecutive sampling periods and the preset threshold for the number of reversals are calibrated and set according to the historical data of the specific aquaculture pond and the characteristics of the sensors. For example, the empirical range of these parameters is determined by analyzing the periods in historical data where oscillations are clearly observed.

[0084] S3. If an alternating fluctuation relationship is determined, it is matched with multiple predefined dominant cause categories to determine the dominant cause category with the highest correlation. This includes:

[0085] The feature information of the alternating fluctuation relationship is extracted. This feature information includes at least the combination of key water quality parameters exhibiting the alternating fluctuation and the alternation period. The key water quality parameter combination refers to the specific identities of at least two key water quality parameters identified as exhibiting an alternating fluctuation relationship during the judgment process. For example, a key water quality parameter combination is the combination of dissolved oxygen concentration and ammonia nitrogen concentration. The alternation period refers to the average time required for the two key water quality parameters in the alternating fluctuation relationship to complete one complete reverse cycle. The method for extracting the alternation period is to identify consecutive peaks and sub-peaks in the data sequence of each key water quality parameter within the continuous time window in which the alternating fluctuation relationship is determined. A trough and a peak are defined as points where the value of a point is greater than the values ​​of the two adjacent sampling points before and after it, and a trough is defined as a point where the value of a point is less than the values ​​of the two adjacent sampling points before and after it. The time interval between adjacent peaks or adjacent troughs of the same key water quality parameter is calculated, and the arithmetic mean of these time intervals is taken as the fluctuation period of the key water quality parameter. The arithmetic mean of the fluctuation periods of two key water quality parameters is taken as the alternation period of the alternating fluctuation relationship. For example, if the fluctuation period of dissolved oxygen concentration is calculated to be 35 minutes and the fluctuation period of ammonia nitrogen concentration is 37 minutes, then the alternation period is 36 minutes.

[0086] After extracting the feature information of the alternating fluctuation relationship, the similarity of the feature information with the feature patterns corresponding to each predefined dominant cause category is calculated. The feature pattern of each predefined dominant cause category is a set of statistically representative feature value ranges. For a specific dominant cause category, its feature pattern contains two parts: the first part is the matching rule for key water quality parameter combinations, and the second part is the typical value range for the alternation cycle. The similarity calculation is performed in two steps. The first step calculates the matching degree of the key water quality parameter combinations. If the key water quality parameter combinations of the current alternating fluctuation relationship are completely consistent with the parameter combinations defined in the feature pattern, the matching degree is recorded as 1; otherwise, the matching degree is recorded as 0. The second step calculates the similarity of the alternation cycle. Proximity is calculated by subtracting the absolute difference between the current alternation cycle and the median typical value of the alternation cycle defined in the feature pattern from the numerical value of 1, and then dividing by a normalization coefficient. The normalization coefficient is used to scale the difference; for example, setting the normalization coefficient to 60 minutes ensures that the proximity calculation result falls between 0 and 1. The final similarity calculation result is a weighted sum of matching degree and proximity degree. The weight coefficients of matching degree and proximity degree are set according to historical matching experience. The weight coefficient of matching degree is set higher than that of proximity degree to emphasize the consistency of parameter combination. For example, the weight coefficient of matching degree is set to 0.6 and the weight coefficient of proximity degree is set to 0.4. A similarity calculation result is calculated in this way for each dominant cause category.

[0087] After completing the similarity calculation for all predefined dominant cause categories, the dominant cause category with the highest similarity calculation result is determined as the dominant cause category with the highest degree of association with the alternating fluctuation relationship. The determination method is to compare the similarity calculation results corresponding to all dominant cause categories and select the dominant cause category corresponding to the largest similarity calculation result. If two or more categories have the same similarity calculation result and both are the highest value, then the alternating cycle proximity calculation results of these categories are further compared, and the category with the higher alternating cycle proximity calculation result is selected as the dominant cause category with the highest degree of association.

[0088] The process of establishing the characteristic patterns corresponding to multiple predefined dominant cause categories is completed offline; multiple alternating fluctuation relationship instances and their corresponding aquaculture environment records are collected in the historical aquaculture cycle; the historical aquaculture cycle refers to the red claw crayfish aquaculture batches that have been completed in the past and have complete data records; alternating fluctuation relationship instances are identified by reviewing historical water quality data and applying the same logic to judge whether alternating fluctuation relationships occur in the current water quality state; the aquaculture environment record is the environmental parameter log corresponding to the time point of each alternating fluctuation relationship instance, and the aquaculture environment record includes at least the batch number of fermented feed, feeding amount, water body base temperature before feeding, aquaculture density, and continuous weather conditions before the occurrence of the instance.

[0089] Based on common factors in aquaculture environmental records, instances of alternating fluctuations were clustered into multiple dominant cause categories. Common factors refer to combinations of environmental conditions that repeatedly occur among different alternating fluctuation instances and may cause water quality fluctuations. The clustering analysis method first constructs a feature vector for each alternating fluctuation instance. The feature vector not only includes information on the alternation cycle and key water quality parameter combinations extracted from water quality data, but also includes key factors quantified from aquaculture environmental records, such as the ratio of feed amount to water body base temperature and the number of consecutive cloudy days before feeding. Then, the K-means clustering algorithm is used to group the feature vectors of all historical alternating fluctuation instances. The number of clusters K is determined by the silhouette coefficient method, which calculates the average silhouette coefficient of all instances under different K values ​​and selects the K value that maximizes the average silhouette coefficient. Finally, each cluster is a dominant cause category, and each dominant cause category has a set of central feature vectors that are distinct from other categories.

[0090] Statistical features of key water quality parameter combinations and alternation cycles for all alternating fluctuation relationship instances under each dominant cause category are extracted to form a feature pattern. For key water quality parameter combinations, the parameter combinations with the highest frequency among all alternating fluctuation relationship instances under the dominant cause category are statistically analyzed, and parameter combinations with a frequency exceeding, for example, 50% are defined as parameter combinations for the feature pattern of the dominant cause category. For alternation cycles, the mean and standard deviation of alternation cycles for all alternating fluctuation relationship instances under the dominant cause category are calculated, and the mean is used as the median of the typical value of the alternation cycle. The range of the mean plus or minus twice the standard deviation is used as the reasonable range of the typical value of the alternation cycle. The feature pattern of each dominant cause category consists of two parts: high-frequency parameter combinations and the range of typical values ​​of the alternation cycle.

[0091] S4. Based on the dominant inducing factor category with the highest correlation, analyze the evolution trend of the current water quality state in the phase space composed of multiple water quality parameters, including:

[0092] Based on the dominant cause category with the highest correlation, a set of reference trajectories formed in phase space by historical water quality parameter data associated with that dominant cause category is obtained. The specific steps for obtaining the reference trajectory set include: filtering all historical water quality parameter data segments marked as belonging to the dominant cause category with the highest correlation from the historical database; a historical water quality parameter data segment refers to a data sequence that has been recorded and classified in history, lasts for a period of time, and includes multiple water quality parameters collected in real time; filtering is completed by querying whether the database label field matches the identifier of the dominant cause category with the highest correlation; mapping each filtered historical water quality parameter data segment to a phase space composed of multiple water quality parameters to form a historical trajectory; the mapping method is to treat the values ​​of multiple water quality parameters recorded at each sampling time for each historical water quality parameter data segment as a multiple... The phase space is a five-dimensional coordinate space. Multiple sampling points are connected in chronological order to form a trajectory line. The multiple water quality parameters specifically refer to five parameters: dissolved oxygen concentration, pH value, ammonia nitrogen concentration, nitrite concentration, and redox potential. The phase space is also a five-dimensional space. Water quality parameter data within a predetermined length preceding the current water quality state point are also mapped to the same phase space as the recent trajectory. The predetermined length of water quality parameter data refers to continuous water quality data traced back a fixed time period from the current moment. This fixed time period must be sufficient to reflect the recent dynamics of the system; for example, the fixed time period is set to 2 hours. The mapping method is the same as for historical water quality parameter data segments, i.e., generating a trajectory line connecting all coordinate points within the traced time period. Multiple historical trajectories obtained through filtering and mapping, together with the currently generated recent trajectory, constitute a reference trajectory set.

[0093] The current state point in phase space is determined based on the data of multiple water quality parameters collected at the moment. The data of these multiple water quality parameters refers to the specific values ​​of the five water quality parameters collected and processed in the latest sampling period. The current state point is a five-dimensional coordinate point composed of the values ​​of these five water quality parameters. For example, if the latest sampling period measured dissolved oxygen concentration of 6.2 mg / L, pH value of 7.5, ammonia nitrogen concentration of 0.3 mg / L, nitrite concentration of 0.05 mg / L, and redox potential of 250 mV, then the coordinates of the current state point are (6.2, 7.5, 0.3, 0.05, 250). This coordinate system is used to differentiate the current state from other parameters in calculations. The trajectory points are standardized in dimensions, and the values ​​of each parameter are normalized. The normalization method is to divide each parameter value by its typical range of variation within the healthy aquaculture range. The typical range of variation is determined based on the difference between the maximum and minimum values ​​of the parameter within the normal fluctuation range in long-term historical data. For example, the typical range of variation for dissolved oxygen concentration is 4 mg / L, for pH value it is 1.0, for ammonia nitrogen concentration it is 0.5 mg / L, for nitrite concentration it is 0.1 mg / L, and for redox potential it is 200 mV. After normalization, the example coordinates become (1.55, 7.5, 0.6, 0.5, 1.25).

[0094] Calculate the directional deviation between the current state point and each historical trajectory in the reference trajectory set within the neighborhood of the current position. The neighborhood of the current position is defined as a five-dimensional hyperspherical spatial region centered on the current state point and with a preset distance threshold as its radius. The preset distance threshold is set based on the statistical distribution of the pairwise distances between all historical data points in the phase space; for example, it calculates the set of Euclidean distances between all historical data points and takes the 10th percentile value of this distance set as the preset distance threshold. For each historical trajectory in the reference trajectory set, find all trajectory points that fall within the neighborhood of the current position. Calculate the average direction of motion of this historical trajectory at these neighborhood points. The direction of motion is expressed as... The vector representation of each neighboring point pointing to its next time point; calculate the vector from the current state point to each neighboring point; the direction deviation is defined as the average value of the cosine of the angle between the vector from the current state point to a neighboring point and the historical trajectory motion direction vector at that neighboring point; the cosine of the angle is calculated by dividing the dot product of the two vectors by the product of their magnitudes; in specific calculation, the above cosine values ​​are calculated for each historical trajectory at all its neighboring points, and then the arithmetic mean of these cosine values ​​is taken as the direction deviation of the historical trajectory; if a historical trajectory has no trajectory points in the neighborhood of the current position, the direction deviation of the historical trajectory is not included in the subsequent statistics.

[0095] Based on the statistical results of directional deviation, the most probable direction of movement of the current water quality state in phase space is determined as the evolutionary trend. The statistical results of directional deviation refer to the set of directional deviation values ​​corresponding to all historical trajectories with trajectory points in the neighborhood of the current location. Historical trajectories with directional deviation values ​​less than or equal to zero are excluded from the statistical results of directional deviation. From the remaining historical trajectories, several historical trajectories with the largest directional deviation values ​​are selected, for example, the top 3 historical trajectories with the largest directional deviation values ​​are selected. The weighted average vector of the average movement direction vector of these selected historical trajectories at their corresponding neighborhood points is calculated. The weight of each historical trajectory in the weighted average vector calculation is its directional deviation value. The weighted average vector is normalized to obtain a unit vector. The direction of this unit vector represents the most probable direction of movement of the current water quality state in phase space, which is the evolutionary trend. The evolutionary trend is represented as a unit direction vector in five-dimensional phase space, and each component corresponds to the tendency of change in a water quality parameter dimension.

[0096] S5. Based on evolutionary trends, predict the comprehensive changing trends of multiple water quality parameters within a future set time period, including:

[0097] The evolutionary trend refers to the most likely direction of motion in phase space, expressed as a five-dimensional unit direction vector. Based on the most likely direction of motion indicated by the evolutionary trend, the phase space is extrapolated forward along the corresponding direction of motion. The specific process of extrapolation is to start from the current state point and move step by step in the five-dimensional phase space along the direction specified by the evolutionary trend unit direction vector, generating a new extrapolated state point with each move. The first move starts from the current state point and moves a distance of one extrapolation step to reach the first extrapolated state point. The second move starts from the first extrapolated state point and moves a distance of one extrapolation step along the direction of the evolutionary trend to reach the second extrapolated state point. The extrapolation is iterated, and the total extrapolation time is equal to the future set time period. The future set time period is determined according to the warning time required for the feeding decision, for example, the future set time period is set to 3 hours. The extrapolation process continues until the cumulative extrapolation time reaches the future set time period, and the extrapolated state point obtained at this time is the phase space state point corresponding to the end of the future set time period.

[0098] The extrapolation step size is determined by combining the historical water quality parameter change rate characteristics corresponding to the most correlated dominant cause category. The historical water quality parameter change rate characteristics refer to the statistical characteristics of the system state point's movement speed in phase space within the historical water quality parameter data segment belonging to the most correlated dominant cause category. Specifically, the method for obtaining these historical water quality parameter change rate characteristics is as follows: from all historical trajectories of the most correlated dominant cause category, calculate the Euclidean distance between adjacent sampling points on the trajectory, divide this Euclidean distance by the sampling period length to obtain the instantaneous rate within each time interval; calculate the instantaneous rate values ​​for all instantaneous rate values. The arithmetic mean is taken as the typical rate of change for the dominant cause category with the highest correlation. The extrapolation step size is equal to the typical rate of change multiplied by a base time unit. The base time unit is the basic time interval for extrapolation iterations. The base time unit must be set smaller than the sampling period to maintain the continuity of the prediction. For example, the base time unit is set to 1 minute. The extrapolation step size is equal to the typical rate of change multiplied by 1 minute, and its dimension is distance units in phase space. For example, if the typical rate of change is calculated to be 0.5 phase space distance units per hour, then the extrapolation step size per minute is 0.5 / 60≈0.0083 distance units.

[0099] During the forward extrapolation process, the extrapolation direction is dynamically adjusted based on the system state stability reflected by the historical trajectories in phase space. The evaluation of system state stability is based on the proximity of the current extrapolation state point to the historical trajectory and the degree of local divergence of the historical trajectory. Specifically, in each extrapolation iteration, when a new extrapolation state point is reached, all historical trajectory segments falling into the reference trajectory set are searched with the new extrapolation state point as the center and a preset neighborhood radius as the range. The preset neighborhood radius is the same as the preset distance threshold used in step S4. For each historical trajectory segment falling into the neighborhood, the motion direction of the historical trajectory segment in the local region is calculated. The average direction of all these local motion direction vectors and the dispersion of the direction distribution are calculated. The dispersion is quantified by calculating the average angle between each direction vector and the average direction vector. The preset stability threshold is set according to the statistical value of the direction dispersion of the stable motion phase in the historical data, for example, taking the average dispersion of all stable segments in the historical data plus one standard value. The dispersion is used as a preset stability threshold. If the dispersion is lower than the preset stability threshold, the system is considered stable in this region, and the inference direction remains the most likely direction of motion indicated by the evolution trend. If the dispersion is higher than or equal to the preset stability threshold, the historical trajectory is considered to have significant divergence in this region, and the system's stability is low. In this case, the inference direction is dynamically adjusted. The adjustment method is to weight and fuse the current evolution trend direction with the calculated average direction of the local historical trajectory to generate a new adjusted direction. The weighting is set according to the magnitude of the dispersion. A baseline dispersion value is set. When the actual dispersion is equal to the baseline dispersion, the weight of the current evolution trend direction and the weight of the average direction of the local historical trajectory are both 0.5. When the actual dispersion is higher than the baseline dispersion, the weight of the average direction of the local historical trajectory is increased proportionally. The increase ratio is equal to the actual dispersion divided by the baseline dispersion and then multiplied by 0.5. The new adjusted direction vector is renormalized into a unit vector and used for the next inference movement.

[0100] The phase space state point corresponding to the end of the future set time period is back-mapped to predicted values ​​of multiple water quality parameters, forming a comprehensive trend of change. Back-mapping is the reverse process of normalization. The phase space state point corresponding to the end of the future set time period is a five-dimensional coordinate point, where each coordinate value is a normalized value. Back-mapping requires multiplying the normalized value of each dimension by its corresponding typical variation span value to restore water quality parameter values ​​with actual physical units and dimensions. The typical variation span value is exactly the same as the value used for normalization in step S4; for example, the typical variation span of dissolved oxygen concentration is 4 mg / L. Assuming a coordinate value of 1.6 for the future state point, the predicted dissolved oxygen concentration value is 1.6 × 4. =6.4 mg / L; This calculation is performed on all five dimensions to obtain the predicted values ​​of dissolved oxygen concentration, pH value, ammonia nitrogen concentration, nitrite concentration, and redox potential at the end of the future set time period; The expression of the comprehensive trend includes not only the predicted value at the end time, but also a series of intermediate predicted values ​​obtained by back-mapping the intermediate simulation state points recorded during the simulation process at several key time points within the future set time period; Key time points are selected at equal intervals, for example, one point is selected every 30 minutes; All key time points and their corresponding predicted values ​​of multiple water quality parameters are organized in chronological order to form the comprehensive trend of multiple water quality parameters within the future set time period.

[0101] S6. Based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs, generate and execute feeding control instructions for fermented feed, including:

[0102] This study calculates the deviation between the predicted values ​​of multiple water quality parameters in the overall trend and the preset boundary values ​​of the water quality range for shrimp feeding requirements. The predicted values ​​of these parameters include the predicted dissolved oxygen concentration, pH value, ammonia nitrogen concentration, nitrite concentration, and redox potential at key time points within a future set time period. The preset water quality range for shrimp feeding requirements is a numerical range preset for each water quality parameter, with the upper and lower limits constituting the boundary values. These boundary values ​​are determined through aquaculture experiments. Different water quality parameter levels are controlled during the experiments, and the feeding rate of redclaw crayfish is observed. The parameter range where the feeding rate reaches 95% or higher of the optimal feeding rate is recorded as the water quality range for shrimp feeding requirements. When calculating the deviation, for each key time point in the overall trend, the difference between the predicted value of each water quality parameter at that time point and its corresponding boundary value of the water quality range for shrimp feeding requirements is calculated. For a single water quality parameter at a given time... The deviation degree calculation is as follows: if the predicted value falls within the preset water quality range for shrimp feeding needs, the deviation degree is recorded as zero; if the predicted value is higher than the upper limit of the range, the deviation degree is equal to the difference between the predicted value and the upper limit, divided by the upper limit, resulting in a positive value indicating a higher proportion; if the predicted value is lower than the lower limit of the range, the deviation degree is equal to the difference between the lower limit and the predicted value, divided by the lower limit, resulting in a negative value indicating a lower proportion. For each key time point, the deviation degree values ​​calculated from the five water quality parameters are weighted and summed according to the preset weights of each parameter's impact on feeding to obtain the comprehensive deviation degree for that time point. The preset weights of each parameter are determined based on historical feeding experiment data. In the historical feeding experiment data, the correlation coefficient between the actual decrease in feeding rate and the deviation degree of each parameter when different water quality parameters deviate from their respective shrimp feeding needs ranges is statistically analyzed. The correlation coefficients of each parameter are normalized and used as the preset weights of that parameter. The maximum value of the comprehensive deviation degree of all key time points within a future set period is taken as the final deviation degree value for decision-making.

[0103] Based on the calculated deviation, a preset control strategy mapping table is consulted to determine the corresponding feeding control parameters. This preset control strategy mapping table is a lookup table reflecting the correspondence between deviation levels and feeding control parameters. The table is constructed based on historical aquaculture data analysis and expert experience. Specifically, it involves collecting historical aquaculture records, which include the degree of water quality deviation, the feeding control parameters used, and subsequent aquaculture effect evaluations. By analyzing the aquaculture effects of different feeding control parameters under different deviation levels, the feeding control parameters that bring the best aquaculture results are associated with the corresponding deviation ranges, forming a mapping table. The deviation level range is divided based on... The final deviation value is divided into levels based on its statistical distribution. For example, the final deviation values ​​in historical data are sorted by size, and the minimum value to the 20th percentile is taken as level one, the 20th percentile to the 40th percentile as level two, and so on to divide multiple level intervals. The feeding control parameters include at least the base feeding amount and the feeding timing correction coefficient. The base feeding amount represents the weight of feed fed in a single feeding under standard conditions, and the feeding timing correction coefficient is used to adjust the feeding time. When querying, the calculated final deviation value is compared with the level interval in the preset control strategy mapping table to determine its level, thereby obtaining the base feeding amount and the feeding timing correction coefficient corresponding to that level.

[0104] Based on the defined feeding control parameters, a fermented feed feeding control instruction is generated, including the feeding amount and timing. The feeding amount is derived by fine-tuning the base feeding amount combined with the current stocking density and the average weight of the shrimp population. Specifically, the stocking density is first calculated based on the pond area and the current number of shrimp. Then, the average weight of the shrimp population is retrieved. This average weight is calculated by periodically catching at least 30 shrimp from the pond, weighing them, and then averaging the results. The final feeding amount equals the base feeding amount multiplied by the stocking density, and then multiplied by a correction factor for the average weight of the shrimp population. This correction factor is an empirical value, determined through the use of... The breeding experiment determined that by setting different feeding levels and observing the growth rate and feed conversion rate of shrimp populations, the feeding amount per unit density and per unit body weight when both growth rate and feed conversion rate reached their optimal values ​​was determined as the correction coefficient. The feeding timing was obtained by multiplying the planned feeding time by the feeding timing correction coefficient. The planned feeding time was a fixed daily feeding time preset according to the breeding management procedures. The generated control command was a structured data packet containing the command type identifier, the target feeder number, the calculated feeding amount, the calculated feeding timing, and the fermented feed type code.

[0105] The generated fermented feed feeding control command is sent to the feeder, driving the feeder to perform the feeding action. The sending is completed through wired or wireless communication protocols, and the communication protocol selection is consistent with the protocol supported by the feeder controller. The command data packet is encapsulated according to the frame format specified by the feeder controller, including a start character, address code, function code, data field, check code, and end character. After receiving the command, the feeder controller parses the feeding amount and feeding timing information in the data field and stores it in the execution queue. When the system clock reaches the feeding timing time specified by the command, the controller drives the stepper motor to rotate, causing the feeding screw to rotate a preset number of revolutions. The conversion relationship between the preset number of rotations of the feeding screw and the weight of feed to be fed is determined through calibration experiments. In the calibration experiments, the actual weight of feed fed under different motor rotation numbers is recorded, and the correspondence between the number of revolutions and the weight is established. Thus, the corresponding feeding amount of fermented feed is pushed from the hopper to the feeding pipe. The feed is evenly scattered into the aquaculture water through the pipe, completing a precise feeding action.

[0106] Example 2: Figure 2 A schematic diagram of the intelligent sensing-based precision feeding system for fermented feed of redclaw crayfish is provided. The intelligent sensing-based precision feeding system for fermented feed of redclaw crayfish includes the following modules:

[0107] The sensor acquisition module is used to collect data on multiple water quality parameters related to the allelopathic effect of fermented feed in real time through multiple sensors deployed in the aquaculture water.

[0108] The fluctuation discrimination module is used to determine whether the current water quality status shows an alternating fluctuation relationship based on the data of multiple water quality parameters collected at the moment;

[0109] The trigger matching module is used to perform matching analysis with multiple predefined dominant trigger categories if an alternating fluctuation relationship is determined, in order to determine the dominant trigger category with the highest degree of correlation.

[0110] The trend analysis module is used to analyze the evolution trend of the current water quality status in a phase space composed of multiple water quality parameters, based on the dominant cause category with the highest correlation.

[0111] The water quality prediction module is used to predict the comprehensive change trend of multiple water quality parameters within a set future time period based on evolution trends.

[0112] The feeding control module is used to generate and execute feeding control instructions for fermented feed based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs.

[0113] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0114] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0115] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0116] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0118] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0119] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing, characterized in that, Includes the following steps: S1. Data on multiple water quality parameters related to the allelopathic effect of fermented feed are collected in real time by multiple sensors deployed in the aquaculture water body. S2. Based on the data of multiple water quality parameters collected so far, determine whether the current water quality status shows an alternating fluctuation relationship, including: Select at least two key water quality parameters from multiple water quality parameters; Analyze the direction of data change for at least two key water quality parameters over a set number of consecutive sampling periods. The direction of data change refers to whether the water quality parameter value increases, decreases, or remains essentially unchanged from one sampling period to the next. The criterion for determining whether the value remains essentially unchanged is that the absolute value of the difference between the values ​​of two adjacent sampling periods is less than a multiple of the sensor's measurement accuracy for that water quality parameter. If, within a set number of consecutive sampling periods, the number of times the data changes in the opposite direction of at least two key water quality parameters reach a preset threshold for the number of times the changes are reversed, it is preliminarily determined that there are alternating fluctuation characteristics. Verify whether the alternating fluctuation characteristics persist within a set number of subsequent sampling periods; If this continues, it indicates that the current water quality is exhibiting alternating fluctuations. S3. If an alternating fluctuation relationship is determined, it is matched with multiple predefined dominant cause categories to determine the dominant cause category with the highest degree of correlation. S4. Based on the dominant inducing factor category with the highest correlation, analyze the evolution trend of the current water quality state in the phase space composed of multiple water quality parameters, including: Based on the dominant cause category with the highest correlation, obtain the reference trajectory set formed in phase space by the historical water quality parameter data associated with the dominant cause category; Based on the data of multiple water quality parameters collected at present, determine the current state point in phase space corresponding to the current water quality state; Calculate the directional deviation between the current state point and each historical trajectory in the reference trajectory set within the neighborhood of the current position; Based on the statistical results of directional deviation, the most likely direction of movement of the current water quality state in phase space is determined as the evolution trend; S5. Based on evolutionary trends, predict the comprehensive change trend of multiple water quality parameters within a set future time period; S6. Based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs, generate and execute feeding control instructions for fermented feed.

2. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 1, characterized in that, Multiple sensors deployed in the aquaculture water body are used to collect real-time data on several water quality parameters related to the allelopathic effects of fermented feed, including: Multiple water quality parameters are periodically sampled simultaneously using multiple sensors. Within each sampling period, acquire the sensor readings corresponding to each water quality parameter; Real-time noise filtering is performed on the sensor readings of each acquired water quality parameter; The sensor readings for each water quality parameter, after noise filtering, are combined with the time information of the corresponding sampling period to form data for multiple water quality parameters.

3. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 1, characterized in that, If an alternating fluctuation relationship is identified, it is matched with multiple predefined dominant cause categories to determine the dominant cause category with the highest degree of association, including: Extract the characteristic information of the alternating fluctuation relationship. The characteristic information should include at least the combination of key water quality parameters that cause the alternating fluctuation and the alternation period. The similarity between the feature information and the feature pattern corresponding to each predefined dominant cause category is calculated; The dominant cause category with the highest similarity calculation result is determined as the dominant cause category with the highest degree of association.

4. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 3, characterized in that, The establishment of feature patterns corresponding to multiple predefined dominant cause categories includes: Collect examples of multiple alternating fluctuations that occurred in historical aquaculture cycles and their corresponding aquaculture environment records; Based on common factors in aquaculture environment records, multiple instances of alternating fluctuations were clustered into several dominant cause categories; Statistical features of all alternating fluctuation relationship instances under each dominant cause category in terms of key water quality parameter combinations and alternation cycles are extracted to form a feature pattern.

5. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 1, characterized in that, The set of reference trajectories formed in phase space by historical water quality parameter data associated with the dominant cause category includes: Filter out all historical water quality parameter data segments from the historical database that are marked as belonging to the dominant cause category with the highest correlation. Each historical water quality parameter data segment is mapped to a phase space composed of multiple water quality parameters to form a historical trajectory; The water quality parameter data within a set length prior to the current state point corresponding to the current water quality state are also mapped to the phase space and used as the recent trajectory; Multiple historical trajectories are combined with recent trajectories to form a reference trajectory set.

6. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 1, characterized in that, Based on evolutionary trends, the comprehensive changing trends of multiple water quality parameters within a given future time period are predicted, including: Based on the most likely direction of motion indicated by the evolutionary trend, the motion is extrapolated forward in phase space along the corresponding direction of motion; The inference step size is determined by combining the historical water quality parameter change rate characteristics corresponding to the dominant cause category with the highest correlation. During the forward extrapolation process, the direction of extrapolation is dynamically adjusted based on the stability of the system state reflected by the historical trajectory in phase space. The phase space state point corresponding to the end of the set future time period will be extrapolated and mapped in reverse to the predicted values ​​of multiple water quality parameters, forming a comprehensive trend of change.

7. The method for precise feeding of fermented feed for redclaw crayfish based on intelligent sensing according to claim 1, characterized in that, Based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding requirements, feeding control instructions for fermented feed are generated and executed, including: Calculate the degree of deviation between the predicted values ​​of multiple water quality parameters in the overall trend and the preset boundary values ​​of the water quality range for shrimp feeding requirements; Based on the calculated deviation, the preset control strategy mapping table is consulted to determine the corresponding feeding control parameters; Based on the determined feeding control parameters, a feeding control instruction for fermented feed, including feeding amount and feeding timing, is generated. The generated fermented feed feeding control command is sent to the feeder to drive the feeder to perform the feeding action.

8. A precision feeding system for fermented feed of redclaw crayfish based on intelligent sensing, used to implement the precision feeding method for fermented feed of redclaw crayfish based on intelligent sensing as described in any one of claims 1-7, characterized in that, Includes the following modules: The sensor acquisition module is used to collect data on multiple water quality parameters related to the allelopathic effect of fermented feed in real time through multiple sensors deployed in the aquaculture water. The fluctuation discrimination module is used to determine whether the current water quality status shows an alternating fluctuation relationship based on the data of multiple water quality parameters collected at the moment; The trigger matching module is used to perform matching analysis with multiple predefined dominant trigger categories if an alternating fluctuation relationship is determined, in order to determine the dominant trigger category with the highest degree of correlation. The trend analysis module is used to analyze the evolution trend of the current water quality status in a phase space composed of multiple water quality parameters, based on the dominant cause category with the highest correlation. The water quality prediction module is used to predict the comprehensive change trend of multiple water quality parameters within a future set period based on evolution trends. The feeding control module is used to generate and execute feeding control instructions for fermented feed based on the deviation between the predicted overall trend and the preset water quality range for shrimp feeding needs.

Images