Intelligent fishery environment sensing and adaptive intelligent adjustment method based on big data

Abstract

The application discloses a big data-based intelligent fishery environment sensing and self-adaptive intelligent adjusting method and relates to the technical field of intelligent fishery, and comprises the following steps: step one: collecting aquaculture environment data, equipment operation data and control threshold values to form an environment data set; step two: performing space-time registration to form an environment state record, a water body exchange record and a safety domain; step three: generating phase state units, environment focal elements and basic probability distributions, constructing a time phase identification framework and a phase state spectrum mapping; step four: constructing a water body conduction map and migrating environment focal element probability mass; step five: dividing into arrival quality, in-transit quality and retention quality to generate in-transit focal elements; step six: distributing conflict mass by using an improved Dubois-Prade algorithm to generate a fusion result; step seven: forming a control homoeomorphous framework and a joint basic probability distribution; and step eight: determining a target action and controlling equipment operation. The application improves sensing and adjusting accuracy.

Description

Technical Field

[0001] This invention relates to the field of smart fisheries technology, and in particular to a smart fisheries environmental perception and adaptive intelligent regulation method based on big data. Background Technology

[0002] With the increasing scale of aquaculture and the growing demand for smart fisheries, technologies for continuous sensing, risk identification, and adaptive equipment adjustment of aquatic environments have received widespread attention. Existing smart fisheries environmental control systems primarily rely on single-point sensor data and fixed thresholds to trigger aeration, water exchange, and circulation equipment. However, these systems suffer from the following problems in practical applications: The data collection times for dissolved oxygen, temperature, pH, and water exchange within the aquaculture area are inconsistent, and water flow between areas exhibits conduction delays. Existing treatment methods struggle to accurately match environmental conditions with equipment operating status, easily leading to biases in judging the location and arrival time of anomalies. The aquatic environment exhibits continuous evolution, splitting migration, and convergence transformation characteristics. Traditional evidence fusion methods typically allocate conflicts between environmental focal elements to a focal element union, making it difficult to distinguish between time-delayed conflicts, transformation conflicts, and unreachable conflicts, and easily losing phase inheritance relationships and water conduction relationships. Existing equipment control methods primarily determine adjustment actions directly based on the results of individual index exceedances, lacking joint analysis between environmental focal elements, action reachability results, and the safety domain. This makes it difficult to assess the effectiveness of adjustment actions in bringing the aquaculture environment into the safety domain, resulting in insufficient equipment control precision and poor continuity of environmental regulation.

[0003] Therefore, how to provide a smart fisheries environment perception and adaptive intelligent regulation method based on big data is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose a smart fishery environment perception and adaptive intelligent regulation method based on big data. This invention integrates aquaculture environment data through water body conduction diagrams, phase spectrum mapping, and an improved Dubois-Prade algorithm to distinguish between time-delay conflicts, transformation conflicts, and unreachable conflicts, and determines target actions based on the control equivalence quotient framework. It has the advantages of accurate environmental identification, reasonable conflict handling, timely equipment adjustment, and interpretable control results.

[0005] The intelligent fisheries environment perception and adaptive intelligent regulation method based on big data according to an embodiment of the present invention includes the following steps: Step 1: Collect aquaculture environment data and equipment operation data, obtain control thresholds, and form an environmental dataset by region and time series; Step 2: Perform spatiotemporal registration on the environmental dataset to generate environmental status records and water exchange records, and establish security domains according to control thresholds; Step 3: Generate phase state units and environmental focal elements based on environmental state records, generate basic probability assignments, and construct a temporal phase identification framework and phase spectrum mapping; Step 4: Construct a water transfer map based on water exchange records, and map the probabilistic mass of the migration environment focal elements according to the phase spectrum. Step 5: Divide the probabilistic mass into arriving mass, en route mass, and stationary mass according to the water body conduction diagram, and generate en route focal elements containing arrival time windows according to the en route mass; Step 6: Construct an improved Dubois-Prade algorithm, write the time delay conflict quality into the in-transit focal element, transform the conflict quality into the minimum propagation envelope, write the unreachable conflict quality into the focal element union, and generate the fusion result; Step 7: Form an action set from the equipment operation data, construct a control equivalence quotient framework based on the reachability of environmental focal elements into the safety domain under the action set, and generate a joint basic probability allocation by combining the fusion results; Step 8: Determine the target action based on the joint basic probability allocation, control the equipment operation, and incorporate the operation results into the environmental dataset.

[0006] Optionally, step one specifically includes: Configure area identifiers for aquaculture areas, collect aquaculture environmental data carrying area identifiers, collection time, and environmental indicator identifiers, organize the aquaculture environmental data according to area identifiers and collection time, and generate environmental collection records; Collect equipment operation data carrying equipment identification, area identification and operation time, organize the equipment operation data according to equipment identification and operation time, generate equipment operation record, read threshold data corresponding to area identification and environmental indicator identification, and generate control threshold; Environmental data collection records and equipment operation records are associated with regional identifiers, a time-series correspondence is established between collection time and operation time, control thresholds are associated with environmental data collection records, and environmental data collection records, equipment operation records and control thresholds are aggregated to form an environmental dataset.

[0007] Optionally, step two specifically involves: The environmental dataset is collected according to the regional identifier. The time difference between the collection time and the operation time and the registration time is calculated respectively. The aquaculture environment data and equipment operation data are written into the registration time with the smallest time difference to form a registration data unit. The environmental indicator identifier and indicator value are read from the registration data unit. The indicator value is arranged according to the environmental indicator identifier to form an environmental state vector. The environmental state vector is bound to the area identifier and the registration time to generate an environmental state record. Water flow direction and water exchange volume are extracted from the aquaculture environment data in the registration data unit. Outflow and inflow areas are determined according to the water flow direction. The outflow area, inflow area, water exchange volume and registration time are associated to generate water exchange records. The control thresholds are converted into allowable ranges based on the environmental indicator identifiers, and the range in which all indicator values ​​in the environmental state vector are within the corresponding allowable ranges is defined as the safe domain.

[0008] Optionally, step three specifically includes: The environmental status records are arranged according to the region identifier and registration time. The change direction of the same environmental indicator at adjacent registration times is extracted. Environmental status records with the same change direction and continuous registration time are merged to generate a phase unit with the record region identifier, start and end time and indicator value range. A temporal phase identification framework is constructed by collecting phase units according to the registration time. Phase units with consistent regional identifiers and overlapping index value ranges are combined to generate environmental focal elements. The number of environmental indicators falling within the range of environmental foci index values ​​in the environmental status record is counted. The ratio of the number of environmental indicators to the total number of environmental indicators in the environmental status record is calculated. The ratio is normalized to generate the basic probability allocation corresponding to the environmental foci. By comparing phase units in adjacent phase identification frameworks, inheritance relationships are established for phase units with consistent regional identifiers and overlapping index value ranges. The correspondence between phase units is recorded according to single-source inheritance, split inheritance, and convergence inheritance to generate a phase spectrum mapping.

[0009] Optionally, step four specifically involves: Read the water exchange records, map the area identifiers to the transmission nodes, establish transmission edges according to the outflow area to the inflow area, write the water exchange volume and registration time into the transmission edges, and construct the water transmission graph; According to the regional identifier and registration time, the environmental focal elements are mapped to the transmission nodes. The phase state units in the adjacent temporal identification framework are compared, and the inheritance relationship of the environmental focal elements is determined according to the phase state spectrum mapping. For single-source inheritance relationships, the probabilistic quality of the focal element in the previous time phase is transferred to the focal element in the next time phase. For split inheritance relationships, the probabilistic quality of the focal element in the previous time phase is allocated according to the proportion of the overlap length of the index value range to the total overlap length. For converging inheritance relationships, the probability mass of the environmental focal elements in the previous time phase is accumulated to the environmental focal elements in the next time phase. The direction of probability mass migration is determined by the propagation edge, thus completing the probability mass migration of the environmental focal elements.

[0010] Optionally, step five specifically includes: Extract the conduction nodes and conduction edges corresponding to environmental focal elements from the water conduction map, connect the conduction nodes along the conduction edges to form a conduction path, and read the water exchange volume and registration time in the conduction path. Calculate the proportion of water exchange volume at the conduction edge to the total outflow water exchange volume at the conduction node, multiply the environmental focal element probabilistic mass by the proportion to generate the conduction mass, and subtract the conduction mass from the environmental focal element probabilistic mass to generate the retention mass. The start registration time and end registration time are determined according to the order of registration times in the transmission path. The period from the start registration time to the end registration time is defined as the arrival time window. When the registration time is not earlier than the end registration time, the transmission quality is defined as the arrival quality. When the registration time is earlier than the end registration time, the transmission quality is defined as the in-transit quality. By binding environmental focal elements, conduction paths, arrival windows, and in-transit quality, in-transit focal elements are generated, and arrival quality, in-transit quality, and lingering quality are aggregated to form a probabilistic quality classification result.

[0011] Optionally, step six specifically includes: An improved Dubois-Prade algorithm is constructed using a temporal identification framework, basic probability assignment, probability quality partitioning results, and water body conduction map. Environmental focal elements with empty intersections are combined into conflict focal element pairs, and conflict quality is generated based on the probability quality of environmental focal elements in the conflict focal element pairs. Map the environmental focal elements in the conflict focal pair to the transmission node, search the transmission path along the transmission edge, and when there are in-transit focal elements in the transmission path and the registration time is within the arrival time window, determine the conflict quality as the time delay conflict quality and assign it to the in-transit focal elements. When there are no in-transit focal elements in the transmission path and the conflicting focal elements point to the same environmental focal element through phase spectrum mapping, the conflict quality is determined as the transformed conflict quality. The transmission path with the smallest interval between the start registration time and the end registration time is selected, the phase units carried by the transmission path are formed into the minimum transmission envelope, and the transformed conflict quality is assigned to the minimum transmission envelope. When there is no transmission path connecting conflict focal pairs in the water conduction map and no focal point pointing to the same environment in the phase spectrum mapping, the conflict quality is determined as unreachable conflict quality and assigned to the focal union formed by the conflict focal pairs. The probability quality corresponding to the non-empty environmental focal elements in the intersection is assigned to the focal elements in the intersection. The probability quality carried by the focal elements in the intersection, the focal elements in transit, the minimum transmission envelope, and the focal element union is collected to generate the fusion result.

[0012] Optionally, the generation of the minimum conduction envelope includes: determining the conduction node corresponding to the conflicting focal element pair as the starting conduction node, determining the conduction node corresponding to the same pointing environment focal element as the ending conduction node, searching for a conduction path connecting the starting and ending conduction nodes along the conduction edge pointing, calculating the time interval between the starting and ending registration times in each conduction path, selecting the conduction path with the smallest time interval, extracting the phase units carried by the conduction path according to the connection order of the conduction nodes, and arranging the phase units according to the registration time to form the minimum conduction envelope.

[0013] Optionally, step seven specifically includes: Extract equipment operating parameters and changes in operating status from equipment operating data, generate adjustment actions according to the consistent relationship between equipment operating parameters and changes in operating status, and configure action identifiers for the adjustment actions to form an action set; Establish a temporal association between the action and the environmental state records before and after the action execution. When the environmental state record before the action execution falls within the range of the corresponding index value of the environmental focal element, generate the action reachable result based on whether the environmental state vector after the action execution enters the safe domain. Arrange the action reachable results corresponding to the environmental focal elements according to the action identifier, compare the action reachable results among the environmental focal elements, classify the environmental focal elements with the same action reachable results into the same control similarity class, and construct the control similarity quotient framework based on the control similarity class. The probability quality carried by the environmental focal element in the fusion result is mapped to the corresponding control equivalent class. Based on the action reachability result, the adjustment action that enables the control equivalent class to enter the safe domain is determined. The probability quality corresponding to the control equivalent class and the adjustment action is collected to generate a joint basic probability assignment.

[0014] Optionally, step eight specifically includes: Based on the probability quality of the control class and the corresponding adjustment action in the joint basic probability allocation of action identifiers, the aggregation result is determined as the action support value corresponding to the adjustment action. The adjustment actions in the action set are arranged in descending order of action support value. The adjustment action with the largest action support value is determined as the target action. If the action support values ​​are the same, the adjustment action with the first action icon is determined as the target action. Read the equipment operating parameters and changes in operating status corresponding to the target action, control the equipment operation based on the equipment operating parameters, and collect the equipment operation data generated by the equipment operation; The system associates equipment operation data with environmental status records based on region identifiers and operation times to generate operation results, which are then incorporated into the environmental dataset according to region identifiers and time sequence.

[0015] The beneficial effects of this invention are: This invention establishes environmental status records and water exchange records by regionally correlating and spatiotemporally registering aquaculture environment data and equipment operation data. A safety domain is then established based on control thresholds, thereby unifying the temporal and spatial references of environmental perception data and equipment operation data. Based on phase units, environmental focal elements, a temporal identification framework, and phase spectrum mapping, this invention can continuously characterize the direction of change, inheritance relationships, and regional migration processes of aquaculture environmental states. Furthermore, by utilizing water conduction maps to divide arrival quality, transit quality, and retention quality, it generates transit focal elements containing arrival time windows. This effectively reduces the interference caused by water exchange delays in environmental state judgment and improves the accuracy of identifying the location, direction, and arrival time of environmental anomalies.

[0016] This invention constructs an improved Dubois-Prade algorithm, classifying environmental focal element conflicts into time-delayed conflicts, transformation conflicts, and unreachable conflicts. It then assigns conflict quality to in-transit focal elements, minimum conduction envelopes, and focal element unions, respectively, avoiding information loss caused by the uniform handling of conflict quality in traditional evidence fusion processes. This ensures that the fusion results preserve water conduction relationships and phase evolution relationships. Based on the fusion results, this invention constructs a control equivalence quotient framework using action reachability results, generating joint basic probability assignments between control equivalence classes and corresponding regulatory actions. Target actions are determined based on action support values, achieving a closed-loop connection between aquaculture environment perception, state fusion, action selection, and equipment control, thus improving the accuracy, timeliness, and interpretability of equipment regulation. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 The flowchart shows the intelligent fishery environment perception and adaptive intelligent regulation method based on big data proposed in this invention. Figure 2 This is a flowchart illustrating the conflict quality allocation process of the big data-based intelligent fisheries environmental perception and adaptive intelligent regulation method proposed in this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0019] refer to Figures 1-2 The intelligent fisheries environment perception and adaptive intelligent regulation method based on big data includes the following steps: Step 1: Collect aquaculture environment data and equipment operation data, obtain control thresholds, and form an environmental dataset by region and time series; Step 2: Perform spatiotemporal registration on the environmental dataset to generate environmental status records and water exchange records, and establish security domains according to control thresholds; Step 3: Generate phase state units and environmental focal elements based on environmental state records, generate basic probability assignments, and construct a temporal phase identification framework and phase spectrum mapping; Step 4: Construct a water transfer map based on water exchange records, and map the probabilistic mass of the migration environment focal elements according to the phase spectrum. Step 5: Divide the probabilistic mass into arriving mass, en route mass, and stationary mass according to the water body conduction diagram, and generate en route focal elements containing arrival time windows according to the en route mass; Step 6: Construct an improved Dubois-Prade algorithm, write the time delay conflict quality into the in-transit focal element, transform the conflict quality into the minimum propagation envelope, write the unreachable conflict quality into the focal element union, and generate the fusion result; Step 7: Form an action set from the equipment operation data, construct a control equivalence quotient framework based on the reachability of environmental focal elements into the safety domain under the action set, and generate a joint basic probability allocation by combining the fusion results; Step 8: Determine the target action based on the joint basic probability allocation, control the equipment operation, and incorporate the operation results into the environmental dataset.

[0020] In this embodiment, step one specifically includes: Each aquaculture area is assigned a unique area identifier. When collecting aquaculture environmental data, the area identifier, collection time, environmental indicator identifier, and indicator value are written into the same data record. The data records are grouped according to the area identifier, and the data records in the same area are arranged in ascending order according to the collection time. The correspondence between indicator value and collection time is established according to the environmental indicator identifier to generate an environmental collection record. When collecting equipment operation data, the equipment identifier, area identifier, operation time and operation status are written into the same data record. The data records are grouped according to the equipment identifier, and the data records corresponding to the same equipment are arranged in ascending order of operation time to generate equipment operation records. The threshold data corresponding to both the area identifier and the environmental indicator identifier is read, and the threshold data is written into the threshold record according to the environmental indicator identifier to generate control thresholds. According to the area identifier, the environmental collection records and equipment operation records are divided into the same breeding area. The time difference between the collection time in the environmental collection records and the operation time in the equipment operation records is calculated. The environmental collection records with the smallest time difference are established with the equipment operation records to form a time sequence correspondence. Then, according to the environmental indicator identifier, the control threshold is written into the corresponding environmental collection record. The environmental collection records, equipment operation records and control thresholds with time sequence correspondence are collected to form an environmental dataset.

[0021] In this embodiment, step two specifically involves: The environmental dataset is grouped according to the regional identifier. The aquaculture environment data and equipment operation data with the same regional identifier are grouped into the same regional data group. The collection time and operation time in the regional data group are read, and the registration time is generated according to the preset time interval. The absolute time difference between each collection time and each registration time is calculated. The aquaculture environment data is written into the registration time with the smallest absolute time difference. The absolute time difference between each operation time and each registration time is calculated. The equipment operation data is written into the registration time with the smallest absolute time difference. Then, the aquaculture environment data and equipment operation data with the same regional identifier and the same registration time are aggregated into a registration data unit. The environmental indicator identifiers and values ​​are read from the registration data unit. The element positions in the environmental state vector are determined according to the fixed arrangement order of the environmental indicator identifiers. Each indicator value is written into the element position corresponding to the same environmental indicator identifier, so that each region forms an environmental state vector with the same dimension and the same element meaning at each registration time. Then, the environmental state vector is bound to the region identifier and the registration time to generate an environmental state record that records the changes in the regional environmental state with the registration time. The water flow direction and water exchange volume are read from the aquaculture environment data contained in the registration data unit. The region identifier is matched according to the starting position of the water flow direction and the outflow region is determined. The region identifier is matched according to the ending position of the water flow direction and the inflow region is determined. The direction from the outflow region to the inflow region is determined as the water exchange direction. The water exchange volume and registration time are written into the exchange relationship between the outflow region and the inflow region to generate a water exchange record that records the water exchange direction and exchange scale between regions. According to the environmental indicator identifier, the lower limit value and the upper limit value of the indicator are read from the control threshold. The closed interval between the lower limit value and the upper limit value of the indicator is determined as the indicator allowable interval. The indicator values ​​in the environmental state vector are compared with the indicator allowable intervals corresponding to the same environmental indicator identifier. The environmental state vectors in which all indicator values ​​are located within the corresponding indicator allowable intervals are classified into the safe state set, and the safe state set is determined as the safe domain.

[0022] In this embodiment, step three specifically includes: Environmental status records are grouped according to regional identifiers. Environmental status records within the same region are arranged in ascending order of registration time. The difference in index value between adjacent registration times for the same environmental indicator is calculated item by item. The direction of change is determined based on the sign of the index value difference. Environmental status records with the same direction of change and continuous registration times are grouped into the same change segment. The first and last registration times corresponding to the change segment are read as the start and end times. The minimum and maximum values ​​of environmental index values ​​within the change segment are extracted to form the index value range. The regional identifier, start and end times, and index value range are associated to generate phase units. The phase units that cover the corresponding registration time are collected according to the registration time. The collected phase units are used as the basic propositions in the temporal identification framework. The index value range of the phase units is compared according to the region identifier. The phase units with the same region identifier and the index value range have a common value part are combined. The set of phase units corresponding to the combination relationship is determined as the environmental focal element. The environmental index values ​​in the environmental status record are compared with the index value range corresponding to the environmental focal element. The number of environmental indicators falling within the index value range is counted. The total number of environmental indicators contained in the environmental status record is read. The ratio of the number of environmental indicators to the total number of environmental indicators is calculated as the focal element support value of the environmental focal element. The focal element support value of each environmental focal element in the same temporal identification frame is divided by the sum of the focal element support values ​​to make the sum of the corresponding values ​​of each environmental focal element equal to one, thus generating a basic probability allocation. Phase units in adjacent time phase identification frameworks are compared according to the registration time sequence. Inheritance relationships are established for phase units with consistent regional identifiers and common value ranges in their index values. When a phase unit in the current time phase corresponds to a phase unit in the next time phase, it is recorded as single-source inheritance. When a phase unit in the current time phase corresponds to two or more phase units in the next time phase, it is recorded as split inheritance. When two or more phase units in the current time phase correspond to a phase unit in the next time phase, it is recorded as merged inheritance. The inheritance relationships are aggregated to generate a phase spectrum mapping.

[0023] In this embodiment, step four specifically includes: Read the water exchange records, extract the outflow area, inflow area, water exchange volume and registration time, map each type of area identifier to a transmission node, establish transmission edges according to the direction of the transmission node corresponding to the outflow area to the transmission node corresponding to the inflow area, and write the water exchange volume and registration time into the transmission edges, and collect the transmission nodes and transmission edges to construct a water transmission graph; Read the region identifier and registration time corresponding to the environmental focal element, write the environmental focal element into the transmission node corresponding to the region identifier, and classify it into the corresponding temporal identification frame according to the registration time. Compare the phase state units contained in the environmental focal elements in adjacent temporal identification frames, and determine the inheritance relationship of the environmental focal element according to the single-source inheritance, split inheritance and convergence inheritance of the phase state spectrum mapping record. Identify the previous and subsequent time-phase environmental focal elements corresponding to the single-source inheritance relationship, transfer the probability quality of the previous time-phase environmental focal elements to the subsequent time-phase environmental focal elements, identify the subsequent time-phase environmental focal elements corresponding to the split inheritance relationship, calculate the overlap length between the index value range of the previous time-phase phase state unit and the index value range of each subsequent time-phase phase state unit, divide each overlap length by the sum of the overlap lengths to obtain the allocation ratio, and allocate the probability quality of the previous time-phase environmental focal elements according to the allocation ratio. Identify the previous temporal environmental focal element corresponding to the confluence inheritance relationship, add the probabilistic masses of the previous temporal environmental focal elements to obtain the confluence probabilistic mass, and transfer the confluence probabilistic mass to the corresponding subsequent temporal environmental focal element. During the transfer process, the probabilistic mass of the environmental focal element is transferred from the outflow area to the inflow area according to the direction of the conduction edge in the water body conduction diagram, thus completing the transfer of the probabilistic mass of the environmental focal element.

[0024] In this embodiment, step five specifically includes: Locate the conduction node where the environmental focal element is located from the water conduction map. Using the conduction node where the environmental focal element is located as the starting point, read the connected conduction nodes step by step along the conduction edge. Form a conduction path from the starting conduction node to the ending conduction node according to the connection order of the conduction nodes. Then read the water exchange volume and registration time recorded by each conduction edge in the conduction path. Arrange the water exchange volume and registration time according to the connection order of the conduction edges to form the path record of the environmental focal element migrating along the water conduction map. For each transmission edge in the transmission path, the water exchange volume from the transmission node corresponding to the starting point of the transmission edge to the external transmission node is summarized to obtain the total outflow water exchange volume. The water exchange volume of the current transmission edge is divided by the total outflow water exchange volume to obtain the transmission ratio. The environmental focal element probability mass is multiplied by the transmission ratio to obtain the transmission mass migrating along the current transmission edge. The environmental focal element probability mass is then subtracted from the transmission mass to obtain the retention mass retained at the current transmission node. The registration times are arranged according to the connection order of the transmission edges in the transmission path. The registration time recorded by the first transmission edge is determined as the start registration time, and the registration time recorded by the last transmission edge is determined as the end registration time. The time range between the start registration time and the end registration time is determined as the arrival time window. The current registration time is then compared with the end registration time. If the current registration time is not earlier than the end registration time, the transmission quality is determined as the arrival quality. If the current registration time is earlier than the end registration time, the transmission quality is determined as the in-transit quality. The environmental focal element, conduction path, arrival window, and in-transit quality are written into the same focal element record to form an in-transit focal element that characterizes the environmental focal element's inter-regional migration state. The arrival quality is written into the conduction node corresponding to the end point of the conduction path, and the retention quality is written into the conduction node corresponding to the start point of the conduction path. The probabilistic quality is then verified according to the relationship that the probabilistic quality of the environmental focal element is equal to the sum of the arrival quality, in-transit quality, and retention quality. The verified arrival quality, in-transit quality, and retention quality are then collected to form the probabilistic quality classification result.

[0025] This invention establishes environmental focal element migration paths based on the transmission nodes, transmission edges, water exchange volume, and registration time in the water body transmission diagram. By calculating the proportion of water exchange volume at the transmission edge to the total outflow water exchange volume, the probabilistic quality of the environmental focal element is divided into transmission quality and retention quality. Combining the initial registration time, the final registration time, and the current registration time, the transmission quality is further divided into arrival quality and in-transit quality, and in-transit focal elements containing the transmission path and arrival time window are generated. This allows the probabilistic quality of the environmental focal element to represent the states of having reached the target area, migrating between areas, and still remaining in the starting area, avoiding misjudging time delays in the water exchange process as environmental state conflicts. At the same time, the probability quality conservation check ensures that the sum of arrival quality, in-transit quality, and retention quality is consistent with the probabilistic quality of the environmental focal element, thereby improving the continuity, accuracy, and interpretability of the water body environmental state migration process, and providing a reliable data foundation for the allocation and fusion results of time delay conflict quality.

[0026] In this embodiment, step six specifically includes: The range of environmental focal elements is limited by the temporal identification framework. The probability mass corresponding to each environmental focal element in the basic probability allocation is read. The probability mass division result and the water body conduction map are introduced into the conflict handling process of the Dubois-Prade algorithm to form an improved Dubois-Prade algorithm. The intersection of environmental focal elements in the same temporal identification framework is calculated pairwise. Two environmental focal elements with empty intersection are formed into conflict focal element pairs. The probability mass corresponding to the two environmental focal elements is multiplied to generate the conflict mass corresponding to the conflict focal element pair. Read the region identifier corresponding to the environmental focal element in the conflict focal element pair, match the region identifier to the conduction node in the water conduction map, and search for the conduction path connecting the conduction node along the conduction edge. Then match the conduction path with the conduction path contained in the focal element in transit. When the match is successful, read the arrival time window corresponding to the focal element in transit, and compare the registration time with the start registration time and end registration time of the arrival time window respectively. When the registration time is within the arrival time window, the conflict quality is determined as the time delay conflict quality, and the time delay conflict quality is assigned to the successfully matched focal element in transit. When there are no successfully matched in-transit focal elements in the transmission path, the inheritance relationship of the corresponding phase units of the environmental focal elements in the conflict focal element pair is traced according to the phase spectrum mapping. Environmental focal elements that point to the same combination of phase units are identified as the same pointing environmental focal elements. The time interval between the start registration time and the end registration time of the corresponding transmission path of the conflict focal element pair is calculated. The transmission path with the smallest time interval is selected. The phase units are arranged according to the transmission nodes in the transmission path. The arranged phase units form the minimum transmission envelope. The conflict quality is determined as the transformed conflict quality and is included in the minimum transmission envelope. When no conduction path connecting conflict focal pairs is found in the water conduction map and no focal point pointing to the same environment is found in the phase spectrum mapping, the conflict focal pairs are determined to have no water conduction relationship or phase inheritance relationship, the conflict quality is determined to be unreachable conflict quality, the environmental focal points in the conflict focal pairs are merged to form a focal union, and the unreachable conflict quality is assigned to the focal union. For environmental focal elements with non-empty intersection within the same temporal identification framework, the product of the probability quality corresponding to the environmental focal elements is calculated, and the product is included in the intersection focal elements formed by the intersection of environmental focal elements. Then, the probability quality carried by the intersection focal elements, the in-transit focal elements, the minimum transmission envelope, and the focal element union is summarized respectively, and the fusion result is formed according to the corresponding focal elements.

[0027] This invention addresses the problem that the Dubois-Prade algorithm fails to characterize the conduction delay, environmental phase transformation, and regional inaccessibility in aquaculture water. It integrates a temporal identification framework, phase spectrum mapping, probability mass partitioning results, and water conduction maps into the conflict resolution process. By identifying time-delay conflicts arising from environmental focal elements that have not yet completed their regional migration through in-transit focal elements and arrival windows, the invention allocates the time-delay conflict mass to in-transit focal elements. Through phase spectrum mapping, it searches for environmental focal elements that conflict focal elements commonly point to, and uses the conduction path with the shortest time interval to carry phase units, forming a minimum conduction envelope. The transformation conflict mass is then allocated to the minimum conduction envelope, thus resolving the issues related to water conduction and phase inheritance. When none of the conditions are met, the unreachable conflict quality is assigned to the focal element union, while preserving the allocation relationship between the probability quality corresponding to the focal element in the intersection and the focal element in the intersection. The improved Dubois-Prade algorithm can distinguish between the in-transit state, phase evolution state, and regional isolation state of the environment during the water exchange process, avoid misjudging time delay as real conflict, reduce the information loss caused by excessive inclusion of conflict quality into the focal element union, and maintain the probability quality transmission process, the evolution direction of environmental focal elements, and the spatial transmission relationship in the fusion result. This improves the accuracy, continuity, and interpretability of aquaculture environment state identification and provides a reliable basis for generating joint basic probability allocation for the control equivalence quotient framework.

[0028] In this embodiment, the generation of the minimum conduction envelope includes: determining the conduction node corresponding to the conflicting focal element pair as the starting conduction node, determining the conduction node corresponding to the same pointing environment focal element as the ending conduction node, searching for a conduction path connecting the starting and ending conduction nodes along the conduction edge, calculating the time interval between the starting and ending registration times in each conduction path, selecting the conduction path with the smallest time interval, extracting the phase units carried by the conduction path according to the connection order of the conduction nodes, and arranging the phase units according to the registration time to form the minimum conduction envelope.

[0029] In this embodiment, step seven specifically includes: The system reads equipment identification, equipment operating parameters, operating status, and operating time from the equipment operation data. It collects the equipment operation data according to the equipment identification and arranges them in ascending order by operating time. It compares the equipment operating parameters and operating status of adjacent operating times and groups the equipment operation data with the same equipment operating parameters and the same changes in operating status into the same operating segment. It combines the changes in equipment operating parameters and operating status corresponding to the operating segment into adjustment actions, assigns a unique action identifier to the adjustment action, and collects the adjustment actions to form an action set. According to the regional identifier, the adjustment actions in the action set are associated with the environmental state records. Taking the running time corresponding to the adjustment action as the boundary, the environmental state record with the smallest distance before the running time is selected from the same region as the environmental state record before the action is executed. The environmental state record with the smallest distance after the running time is selected as the environmental state record after the action is executed. The environmental state vector before the action is executed is compared with the index value range corresponding to the environmental focal element. When the environmental state vector before the action is executed falls into the index value range, the environmental state vector after the action is executed is compared with the index allowable interval corresponding to the safety domain. When the environmental state vector after the action is executed enters the safety domain, an reachable marker is generated. When it does not enter the safety domain, an unreachable marker is generated. The reachable marker or unreachable marker is determined as the action reachable result. Arrange the action reachable results corresponding to each environmental focal element according to the fixed arrangement order of action identifiers to form a sequence of action reachable results corresponding to environmental focal elements. Compare the action reachable results at the same action identifier position among environmental focal elements one by one. Group environmental focal elements with consistent action reachable results at each position into the same control similarity class. Configure class identifiers for control similarity classes and aggregate control similarity classes according to class identifiers to construct the control similarity quotient framework. Based on the attribution relationship between environmental focal elements and control equivalent classes, the probability quality carried by environmental focal elements in the fusion result is transferred to the control equivalent class. The action reachability result corresponding to the environmental focal element in the control equivalent class is read. The adjustment action with the reachability mark is established as a joint relationship with the control equivalent class. The probability quality carried by the environmental focal element corresponding to the joint relationship is accumulated, and the accumulated result is normalized to generate the joint basic probability assignment corresponding to the control equivalent class and the adjustment action.

[0030] In this embodiment, step eight specifically includes: The joint basic probability allocation is grouped according to the action identifier. The probability mass corresponding to the joint relationship between the control class and the regulation action is read item by item. The probability mass with the same action identifier is grouped into the same action group. The cumulative value of the probability mass within the action group is calculated, and the cumulative value is determined as the action support value corresponding to the regulation action. The action support value represents the total probability mass of the control class in the regulation action coverage fusion result. The adjustment actions in the action set are arranged from largest to smallest according to their action support values. The action support values ​​of adjacent adjustment actions are compared in turn. If the action support values ​​are different, the adjustment action with the larger value is arranged first. If the action support values ​​are the same, the order is determined according to the arrangement of the action labels. The adjustment action at the top of the list is determined as the target action. Based on the action identifier of the target action, the corresponding equipment operation parameters and operation status changes are matched in the equipment operation data. The equipment that performs the target action is determined according to the equipment identifier. The equipment operation parameters are converted into equipment operation set values, and the equipment operation status is adjusted according to the operation status changes. During the period when the equipment is running according to the equipment operation set values ​​and operation status, the equipment operation parameters, operation status and operation time are continuously collected to form the equipment operation data generated by the execution of the target action. Read the area identifier and running time carried in the equipment operation data generated by the execution of the target action, filter the environmental status records with the same area identifier in the environmental status records, calculate the absolute time difference between the running time and the registration time, establish a correlation between the environmental status record with the smallest absolute time difference and the equipment operation data to generate the operation result, then collect the operation results according to the area identifier, arrange the operation results in ascending order of the running time, and merge the arranged operation results into the environmental dataset.

[0031] Example 1: To verify the feasibility of this invention in practice, it was applied to a recirculating aquaculture system with six aquaculture zones. Each zone is connected by inlet, outlet, and circulation pipes, forming a water exchange relationship. Each zone is equipped with dissolved oxygen, temperature, pH, ammonia nitrogen, and flow sensors, as well as aeration, circulation, and water exchange equipment. During the aquaculture process, water exchange between zones causes changes in environmental indicators to gradually propagate along the pipes. The timing of sensor data collection differs from the timing of equipment operation. Fixed threshold control methods can easily identify environmental changes that have not yet reached adjacent zones as data conflicts, leading to repeated startups of the aeration equipment, premature operation of the circulation equipment, and continuous operation of the water exchange equipment.

[0032] During operation, aquaculture environmental data was collected at five-minute intervals. Each data point included a region identifier, collection time, environmental indicator identifier, and indicator value. Simultaneously, equipment identifiers, region identifiers, operation time, equipment operating parameters, and operating status were also collected. After thirty consecutive days of operation, 51,840 environmental status records and 17,280 equipment operation records were generated. Control thresholds were set at dissolved oxygen 5.5 mg / L to 8.5 mg / L, temperature 24℃ to 30℃, pH 7.2 to 8.4, and an upper limit of ammonia nitrogen concentration of 0.20 mg / L. A safety domain was established based on these control thresholds. After spatiotemporal registration, phase units were generated according to the direction of change of environmental indicators at adjacent registration times. Phase units with consistent region identifiers and overlapping indicator value ranges were combined into environmental focal elements, and basic probability assignments were generated based on the environmental status records.

[0033] A water conduction map is constructed based on the outflow area, inflow area, water exchange volume, and registration time recorded by the flow sensor. When the dissolved oxygen in the first aquaculture area decreases from 6.3 mg / L to 4.8 mg / L, the water exchange record indicates that the low-dissolved oxygen water flows along the conduction path to the second aquaculture area, with a conduction time interval of fifteen minutes. During the period before the low-dissolved oxygen water reaches the second aquaculture area, the dissolved oxygen in the second aquaculture area remains between 6.0 mg / L and 6.2 mg / L. Traditional fusion methods would classify the environmental focal elements of the two areas as conflicting. This invention determines the corresponding probabilistic quality as in-transit quality based on the conduction path, registration time, and arrival time window, and generates in-transit focal elements, so that conflicting quality is identified as time-delayed conflicting quality. When environmental focal elements point to the same environmental focal element through phase spectrum mapping, the conflicting quality is determined as transformed conflicting quality and assigned to the minimum conduction envelope. When there is no conduction path in the water conduction map and no environmental focal element is pointed to by the same phase spectrum mapping, the conflicting quality is determined as unreachable conflicting quality and assigned to the focal element union.

[0034] Equipment operation data generates oxygenation, circulation, and water exchange adjustment actions. Environmental state records before and after action execution are correlated according to the operating time. After action execution, an reachable marker is generated when the environmental state vector enters the safe domain; otherwise, an unreachable marker is generated. Environmental focal elements with consistent action reachability are grouped into the same control-equivalent class, and a control-equivalent quotient framework is constructed. The probability quality carried by the environmental focal elements in the fusion result is mapped to the control-equivalent class. A joint basic probability allocation is generated according to the action reachability result, and the target action is determined based on the action support value. The same environmental dataset was processed using a fixed threshold control method, as described in Table 1. Table 1 Comparison of the effects of environmental regulation in aquaculture

[0035] As shown in Table 1, this invention improves the accuracy of environmental state identification to 95.8% and reduces the false judgment rate of time-delay conflicts to 4.3%, indicating that the in-transit focal elements and arrival time windows can effectively distinguish environmental states that have not yet completed migration during water body conduction, reducing erroneous conflict judgments caused by water body exchange time delays. By processing the transformation of conflict quality through minimum conduction envelope processing and processing unreachable conflict quality through focal element union, the fusion result can retain the phase inheritance relationship and regional conduction relationship, making the target action efficiency reach 93.6%, shortening the average time for the environmental state to recover to the safe domain by 11.7 minutes, reducing the number of repeated starts of the aeration equipment by 28, reducing the cumulative running time of the equipment by 112.7 hours, and reducing the cumulative time for dissolved oxygen to exceed the safe domain by 31.3 hours. This proves that this invention can improve the accuracy of aquaculture environment perception and the effectiveness of equipment adjustment, and reduce energy consumption caused by ineffective equipment operation.

[0036] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A smart fisheries environment perception and adaptive intelligent regulation method based on big data, characterized in that, Includes the following steps: Step 1: Collect aquaculture environment data and equipment operation data, obtain control thresholds, and form an environmental dataset by region and time series; Step 2: Perform spatiotemporal registration on the environmental dataset to generate environmental status records and water exchange records, and establish security domains according to control thresholds; Step 3: Generate phase state units and environmental focal elements based on environmental state records, generate basic probability assignments, and construct a temporal phase identification framework and phase spectrum mapping; Step 4: Construct a water transfer map based on water exchange records, and map the probabilistic mass of the migration environment focal elements according to the phase spectrum. Step 5: Divide the probabilistic mass into arriving mass, en route mass, and stationary mass according to the water body conduction diagram, and generate en route focal elements containing arrival time windows according to the en route mass; Step 6: Construct an improved Dubois-Prade algorithm, write the time delay conflict quality into the in-transit focal element, transform the conflict quality into the minimum propagation envelope, write the unreachable conflict quality into the focal element union, and generate the fusion result; Step 7: Form an action set from the equipment operation data, construct a control equivalence quotient framework based on the reachability of environmental focal elements into the safety domain under the action set, and generate a joint basic probability allocation by combining the fusion results; Step 8: Determine the target action based on the joint basic probability allocation, control the equipment operation, and incorporate the operation results into the environmental dataset.

2. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step one specifically involves: Configure area identifiers for aquaculture areas, collect aquaculture environmental data carrying area identifiers, collection time, and environmental indicator identifiers, organize the aquaculture environmental data according to area identifiers and collection time, and generate environmental collection records; Collect equipment operation data carrying equipment identification, area identification and operation time, organize the equipment operation data according to equipment identification and operation time, generate equipment operation record, read threshold data corresponding to area identification and environmental indicator identification, and generate control threshold; Environmental data collection records and equipment operation records are associated with regional identifiers, a time-series correspondence is established between collection time and operation time, control thresholds are associated with environmental data collection records, and environmental data collection records, equipment operation records and control thresholds are aggregated to form an environmental dataset.

3. The intelligent fisheries environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step two specifically involves: The environmental dataset is collected according to the regional identifier. The time difference between the collection time and the operation time and the registration time is calculated respectively. The aquaculture environment data and equipment operation data are written into the registration time with the smallest time difference to form a registration data unit. The environmental indicator identifier and indicator value are read from the registration data unit. The indicator value is arranged according to the environmental indicator identifier to form an environmental state vector. The environmental state vector is bound to the area identifier and the registration time to generate an environmental state record. Water flow direction and water exchange volume are extracted from the aquaculture environment data in the registration data unit. Outflow and inflow areas are determined according to the water flow direction. The outflow area, inflow area, water exchange volume and registration time are associated to generate water exchange records. The control thresholds are converted into allowable ranges based on the environmental indicator identifiers, and the range in which all indicator values ​​in the environmental state vector are within the corresponding allowable ranges is defined as the safe domain.

4. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step three specifically involves: The environmental status records are arranged according to the region identifier and registration time. The change direction of the same environmental indicator at adjacent registration times is extracted. Environmental status records with the same change direction and continuous registration time are merged to generate a phase unit with the record region identifier, start and end time and indicator value range. A temporal phase identification framework is constructed by collecting phase units according to the registration time. Phase units with consistent regional identifiers and overlapping index value ranges are combined to generate environmental focal elements. The number of environmental indicators falling within the range of environmental foci index values ​​in the environmental status record is counted. The ratio of the number of environmental indicators to the total number of environmental indicators in the environmental status record is calculated. The ratio is normalized to generate the basic probability allocation corresponding to the environmental foci. By comparing phase units in adjacent phase identification frameworks, inheritance relationships are established for phase units with consistent regional identifiers and overlapping index value ranges. The correspondence between phase units is recorded according to single-source inheritance, split inheritance, and convergence inheritance to generate a phase spectrum mapping.

5. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step four specifically involves: Read the water exchange records, map the area identifiers to the transmission nodes, establish transmission edges according to the outflow area to the inflow area, write the water exchange volume and registration time into the transmission edges, and construct the water transmission graph; According to the regional identifier and registration time, the environmental focal elements are mapped to the transmission nodes. The phase state units in the adjacent temporal identification framework are compared, and the inheritance relationship of the environmental focal elements is determined according to the phase state spectrum mapping. For single-source inheritance relationships, the probabilistic quality of the focal element in the previous time phase is transferred to the focal element in the next time phase. For split inheritance relationships, the probabilistic quality of the focal element in the previous time phase is allocated according to the proportion of the overlap length of the index value range to the total overlap length. For converging inheritance relationships, the probability mass of the environmental focal elements in the previous time phase is accumulated to the environmental focal elements in the next time phase. The direction of probability mass migration is determined by the propagation edge, thus completing the probability mass migration of the environmental focal elements.

6. The intelligent fisheries environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step five specifically involves: Extract the conduction nodes and conduction edges corresponding to environmental focal elements from the water conduction map, connect the conduction nodes along the conduction edges to form a conduction path, and read the water exchange volume and registration time in the conduction path. Calculate the proportion of water exchange volume at the conduction edge to the total outflow water exchange volume at the conduction node, multiply the environmental focal element probabilistic mass by the proportion to generate the conduction mass, and subtract the conduction mass from the environmental focal element probabilistic mass to generate the retention mass. The start registration time and end registration time are determined according to the order of registration times in the transmission path. The period from the start registration time to the end registration time is defined as the arrival time window. When the registration time is not earlier than the end registration time, the transmission quality is defined as the arrival quality. When the registration time is earlier than the end registration time, the transmission quality is defined as the in-transit quality. By binding environmental focal elements, conduction paths, arrival windows, and in-transit quality, in-transit focal elements are generated, and arrival quality, in-transit quality, and lingering quality are aggregated to form a probabilistic quality classification result.

7. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step six specifically involves: An improved Dubois-Prade algorithm is constructed using a temporal identification framework, basic probability assignment, probability quality partitioning results, and water body conduction map. Environmental focal elements with empty intersections are combined into conflict focal element pairs, and conflict quality is generated based on the probability quality of environmental focal elements in the conflict focal element pairs. Map the environmental focal elements in the conflict focal pair to the transmission node, search the transmission path along the transmission edge, and when there are in-transit focal elements in the transmission path and the registration time is within the arrival time window, determine the conflict quality as the time delay conflict quality and assign it to the in-transit focal elements. When there are no in-transit focal elements in the transmission path and the conflicting focal elements point to the same environmental focal element through phase spectrum mapping, the conflict quality is determined as the transformed conflict quality. The transmission path with the smallest interval between the start registration time and the end registration time is selected, the phase units carried by the transmission path are formed into the minimum transmission envelope, and the transformed conflict quality is assigned to the minimum transmission envelope. When there is no transmission path connecting conflict focal pairs in the water conduction map and no focal point pointing to the same environment in the phase spectrum mapping, the conflict quality is determined as unreachable conflict quality and assigned to the focal union formed by the conflict focal pairs. The probability quality corresponding to the non-empty environmental focal elements in the intersection is assigned to the focal elements in the intersection. The probability quality carried by the focal elements in the intersection, the focal elements in transit, the minimum transmission envelope, and the focal element union is collected to generate the fusion result.

8. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, The generation of the minimum conduction envelope includes: determining the conduction node corresponding to the conflicting focal element pair as the starting conduction node, determining the conduction node corresponding to the same pointing environment focal element as the ending conduction node, searching for a conduction path connecting the starting and ending conduction nodes along the conduction edge, calculating the time interval between the starting and ending registration times in each conduction path, selecting the conduction path with the smallest time interval, extracting the phase state units carried by the conduction path according to the connection order of the conduction nodes, arranging the phase state units according to the registration time to form the minimum conduction envelope.

9. The intelligent fishery environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step seven specifically involves: Extract equipment operating parameters and changes in operating status from equipment operating data, generate adjustment actions according to the consistent relationship between equipment operating parameters and changes in operating status, and configure action identifiers for the adjustment actions to form an action set; Establish a temporal association between the action and the environmental state records before and after the action execution. When the environmental state record before the action execution falls within the range of the corresponding index value of the environmental focal element, generate the action reachable result based on whether the environmental state vector after the action execution enters the safe domain. Arrange the action reachable results corresponding to the environmental focal elements according to the action identifier, compare the action reachable results among the environmental focal elements, classify the environmental focal elements with the same action reachable results into the same control similarity class, and construct the control similarity quotient framework based on the control similarity class. The probability quality carried by the environmental focal element in the fusion result is mapped to the corresponding control equivalent class. Based on the action reachability result, the adjustment action that enables the control equivalent class to enter the safe domain is determined. The probability quality corresponding to the control equivalent class and the adjustment action is collected to generate a joint basic probability assignment.

10. The intelligent fisheries environment perception and adaptive intelligent regulation method based on big data according to claim 1, characterized in that, Step eight specifically involves: Based on the probability quality of the control class and the corresponding adjustment action in the joint basic probability allocation of action identifiers, the aggregation result is determined as the action support value corresponding to the adjustment action. The adjustment actions in the action set are arranged in descending order of action support value. The adjustment action with the largest action support value is determined as the target action. If the action support values ​​are the same, the adjustment action with the first action icon is determined as the target action. Read the equipment operating parameters and changes in operating status corresponding to the target action, control the equipment operation based on the equipment operating parameters, and collect the equipment operation data generated by the equipment operation; The system associates equipment operation data with environmental status records based on region identifiers and operation times to generate operation results, which are then incorporated into the environmental dataset according to region identifiers and time sequence.

Images