A dynamic calibration system for proliferation release optimal performance area

Abstract

This invention relates to the field of fishery resource enhancement and release management technology, specifically disclosing a dynamic calibration system for optimal enhancement and release areas. The system acquires release batch information, environmental raster data, fishing vessel trajectories, and recapture points. It constructs a heterogeneous graph topology using spatial units, fishing vessels, and release batches as nodes, establishing occupancy and ownership relationships based on operational stoppage and release coordinates. The system transmits fishing effort to spatial nodes and accumulates pressure values, using recapture points as anchors for reverse transmission and forward propagation, generating an implicit probability field by combining spatial distance attenuation. The probability values ​​are fused as soft and hard labels, and the weights are adjusted based on the multi-scale variation coefficient of the environmental vector, calculating a suitability index. The highest value is selected as the recommended release area. This system solves the problem of calibration inaccuracies caused by sparse recapture labels and fixed environmental weights, and has the beneficial effects of expanding feedback label density and achieving adaptive weight correction.

Description

Technical Field

[0001] This invention relates to the field of fishery resource enhancement and release management technology, specifically to a dynamic calibration system for the optimal effectiveness of enhancement and release areas. Background Technology

[0002] Stock enhancement is an important management method for restoring fishery resources and replenishing the populations of economically important fish species in natural waters. Currently, before carrying out stock enhancement operations, fishery management departments in various regions pre-select several year-round stock enhancement areas in the target waters based on historical experience or environmental monitoring data from fixed sampling sections, and then release the fry into these fixed locations.

[0003] In existing methods for calibrating stock enhancement and release areas, the recapture tags of tagged fish, which are relied upon for effectiveness evaluation, are extremely sparse due to the poor reporting mechanism at the grassroots level. This results in a severe lack of monitoring signals for the effectiveness of stock enhancement and release. At the same time, the influence weights of various environmental factors are fixed in the long term once set, and cannot be adaptively corrected based on the ecological response data after actual stock enhancement and release. Both of these factors together cause the deviation between the output of the calibration system and the actual effectiveness of stock enhancement and release to accumulate continuously over time and cannot be automatically converged. This has become the core technical problem restricting the reliability of dynamic optimization decisions for stock enhancement and release areas. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic calibration system for the optimal efficiency area of ​​stock enhancement and release, so as to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions: A dynamic calibration system for determining the optimal effectiveness of restocking and release areas includes: The data acquisition module acquires the coordinates and quantity of release batches in the target waters, environmental raster data, the trajectory sequence of fishing vessels and the type of fishing nets used, as well as the reported locations of the marked fish recapture points. The heterogeneous graph construction module uses each spatial unit as a spatial node, each fishing vessel as a ship node, and each release batch as an event node. It determines the operational occupancy relationship between ship nodes and spatial nodes based on the trajectory dwell time and frequency, and determines the spatial affiliation relationship between event nodes and spatial nodes based on the release coordinates, thus constructing the heterogeneous graph topology. The feature transmission and pressure accumulation module transmits the fishing effort features of each vessel node to the associated spatial node along the operation occupancy relationship edge, accumulates them to form the comprehensive fishing pressure value of each spatial node, and transmits the release batch attribute to the corresponding spatial node according to the spatial ownership relationship edge. The implicit probability estimation module uses the recapture point as an anchor and infers the probability of a marked fish being captured but not reported at each spatial node by inferring the relationship between the edges in the heterogeneous graph topology, thus generating an implicit recapture probability field. The weight update and region labeling module uses the probability values ​​with confidence levels higher than the threshold in the probability field as soft labels, merges them with the hard labels of the actual recapture, updates the weights of each environmental factor on the release effectiveness using the merged label set, calculates the suitability index of each spatial unit based on the updated weights, and labels the spatial unit with the highest suitability index as the recommended release area.

[0006] As a further aspect of the present invention: the construction of the heterogeneous graph topology specifically includes: Each spatial unit obtained after discretizing the target water area is used as a spatial node. Each spatial node is assigned a spatial code, and a spatial node index table is established. The trajectory sequence of each fishing vessel is divided into operation and rest segments and navigation segments according to the speed difference between adjacent trajectory points. The spatial code corresponding to the spatial coordinates of each operation and rest segment is used as the matching key to search the spatial node index table, associate the fishing vessel node with the matched spatial node, and establish the operation occupancy relationship. Using the spatial code corresponding to the spatial unit where the release coordinates of each release batch fall as the matching key, the spatial node index table is retrieved, and the release event node is associated with the matched spatial node to establish a spatial affiliation relationship. Based on the operation occupancy relationship and spatial ownership relationship, spatial nodes, ship nodes and event nodes are combined into a heterogeneous graph topology.

[0007] As a further aspect of the present invention: the establishment of spatial affiliation specifically includes: Extract the release coordinates and ship trajectory segments within the start and end time of each release batch, and record the spatial codes corresponding to each spatial unit traversed by the ship trajectory segments one by one to form the trajectory spatial code sequence of each release batch. Using each spatial code in the trajectory spatial coding sequence as a matching key, the spatial node index table is retrieved, and all spatial nodes that successfully match any spatial code in the trajectory spatial coding sequence are associated with the release event nodes to form a spatial association group for the release batch. All spatial nodes in the spatial association group of each release event node are determined as the spatial range of the release impact of the release batch.

[0008] As a further aspect of the present invention: the step of transferring the release batch attributes to the corresponding spatial node specifically includes: The trajectory dwell time and turning frequency of each vessel node are obtained, and the product of dwell time and turning frequency is used as the fishing effort feature of the vessel node. The fishing effort characteristics are assigned to each associated spatial node along the edge of the operation occupancy relationship. When the same spatial node receives values ​​from multiple ship nodes, the values ​​are accumulated to form the comprehensive fishing pressure value of the spatial node. The release batch number of each release event node is assigned to each associated spatial node along the spatial affiliation edge, so that each spatial node records the release batch number to which it belongs.

[0009] As a further aspect of the present invention: the generation of the implicit capture probability field specifically includes: The spatial nodes where the actual recapture points are located are selected as anchor nodes, and the comprehensive fishing pressure value of each anchor node is obtained. Based on the comprehensive fishing pressure value of each anchoring node, trace back to each vessel node along the edge of the operation occupation relationship, and transmit the comprehensive fishing pressure value of each anchoring node back according to the pressure contribution ratio of each vessel node to the anchoring node to obtain the recapture correlation strength of each vessel node. Starting with the recapture association strength of each ship node, the algorithm propagates forward along the operation occupancy relationship edge to all spatial nodes associated with the ship node. The capture-out correlation strength received by each space node from different ship nodes is accumulated according to the spatial distance attenuation value to obtain the capture-out-of-report probability value of each space node. The implicit capture-out probability field is composed of the probability values ​​of all space nodes.

[0010] As a further aspect of the present invention: obtaining the recapture correlation strength of each ship node specifically includes: Obtain all vessel nodes associated with each anchoring node and the operational dwell time of each vessel node at the anchoring node. The operational dwell time of each vessel node is multiplied by the capture capability coefficient of the type of net equipped by the vessel node as the operational contribution of the vessel node to the anchoring node. The comprehensive fishing pressure value of the anchored node is allocated according to the proportion of the operational contribution of each vessel node to the total operational contribution of all related vessel nodes, and the allocated value is used as the recapture correlation strength value of each vessel node relative to the anchored node. The final capture association strength of a vessel node is obtained by summing up the multiple capture association strength values ​​obtained for the same vessel node relative to multiple anchored nodes.

[0011] As a further aspect of the present invention: the forward propagation specifically includes: Obtain all spatial nodes associated with each ship node and the corresponding operational dwell time of each spatial node, and use the ratio of the operational dwell time of the ship node at each spatial node to the total operational dwell time of the ship node as the propagation allocation coefficient; The final recapture association strength of the ship node is allocated according to the propagation allocation coefficient of each space node, and the allocated value is assigned to the corresponding space node as the recapture strength contribution value received by the space node from the ship node. The capture intensity contribution values ​​received by the same space node from multiple ship nodes are summed to obtain the capture non-reporting probability value of the space node.

[0012] As a further aspect of the present invention: the calibration process for the recommended release area is as follows: The probability values ​​of each spatial node in the implicit capture probability field are used as corresponding soft labels, which together with the true capture hard labels constitute a label set with confidence attributes for each spatial node. The environmental raster data of each spatial node is used to construct the environmental vector of the spatial node. Based on the difference in environmental vectors of the same spatial node in different spatial scale neighborhoods, the coefficient of variation of each environmental factor in the spatial dimension is determined. The initial weight of each environmental factor is corrected by the coefficient of variation. The environmental factor with the larger coefficient of variation is assigned a higher weight correction value. The environmental vector is weighted and synthesized using the corrected weights of each environmental factor, and the suitability index of each spatial node is calculated. The spatial unit with the highest suitability index is selected as the recommended release area.

[0013] The beneficial effects of this invention are: (1) By deeply mining the Beidou monitoring trajectory data of the deployed fishing vessels, the inference link from fishing vessel operation behavior to implicit capture events is constructed by using the operation dwell time, turning frequency and net capture capability coefficient. An implicit recapture probability field covering the target water area is generated. Without increasing the reporting burden of grassroots fishermen and without changing the existing administrative management process, the spatial coverage density of recapture supervision tags is expanded, and a large amount of field data that was originally lost due to poor manual registration is transformed into usable training signals, providing a sufficient data input basis for the evaluation of release effect.

[0014] (2) By establishing the correlation between the release batch and the spatial range of its impact, the calibration results of the recommended release area are linked with the spatial distribution of the actual fishing pressure of the batch. When the fishing vessel operation mode shifts due to changes in environmental conditions in a certain area, the comprehensive fishing pressure value and the implicit recapture probability field respond accordingly. The calibration results can automatically shift with the shift of the fishermen's actual operation focus, avoiding the mismatch between the release area and the current water conditions due to the fixed setting of the release area at historical experience points. This improves the adaptability of the release area recommendation results in actual technical service applications. Attached Figure Description

[0015] The invention will now be further described with reference to the accompanying drawings.

[0016] Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart of the process for constructing a heterogeneous graph topology in this invention; Figure 3This is a flowchart illustrating the process of transferring the release batch attributes to the corresponding spatial nodes in this invention. Detailed Implementation

[0017] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 As shown, this invention is a dynamic calibration system for determining the optimal effectiveness area of ​​stock enhancement and release, comprising: The data acquisition module acquires the coordinates and quantity of release batches in the target waters, environmental raster data, the trajectory sequence of fishing vessels and the type of fishing nets used, as well as the reported locations of the marked fish recapture points. The heterogeneous graph construction module uses each spatial unit as a spatial node, each fishing vessel as a ship node, and each release batch as an event node. It determines the operational occupancy relationship between ship nodes and spatial nodes based on the trajectory dwell time and frequency, and determines the spatial affiliation relationship between event nodes and spatial nodes based on the release coordinates, thus constructing the heterogeneous graph topology. The feature transmission and pressure accumulation module transmits the fishing effort features of each vessel node to the associated spatial node along the operation occupancy relationship edge, accumulates them to form the comprehensive fishing pressure value of each spatial node, and transmits the release batch attribute to the corresponding spatial node according to the spatial ownership relationship edge. The implicit probability estimation module uses the recapture point as an anchor and infers the probability of a marked fish being captured but not reported at each spatial node by inferring the relationship between the edges in the heterogeneous graph topology, thus generating an implicit recapture probability field. The weight update and region labeling module uses the probability values ​​with confidence levels higher than the threshold in the probability field as soft labels, merges them with the hard labels of the actual recapture, updates the weights of each environmental factor on the release effectiveness using the merged label set, calculates the suitability index of each spatial unit based on the updated weights, and labels the spatial unit with the highest suitability index as the recommended release area.

[0019] The data acquisition module acquires the coordinates and quantity of release batches in the target waters, environmental raster data, the trajectory sequence of fishing vessels and the type of fishing nets used, as well as the reported locations of tagged fish recapture points. Specifically, this includes: The coordinates and quantities of the released batches are obtained from the release registration forms recorded by the stock enhancement and release management units when carrying out the release operations.

[0020] Before the release operation, the management unit used a mobile terminal with satellite positioning function to collect the latitude and longitude coordinates of the release point at the release site, with a coordinate accuracy of meters.

[0021] During the release process, on-site staff counted the actual number of seedlings released by species one by one, and entered the species name, release coordinates, number of released seedlings, and release date and time for each batch into the release registration form.

[0022] The registration form is archived by the release management unit. This invention retrieves the above fields from the archived database through a data interface, serving as the basic source of the spatial location and quantity of each release batch.

[0023] The environmental raster data is derived from publicly available environmental monitoring databases and remote sensing inversion products of the target water area.

[0024] Water environment parameters such as water temperature, dissolved oxygen, and pH are measured hourly by fixed water quality monitoring buoy stations deployed in the target water area, and the results are uploaded to the provincial water environment monitoring data platform via wireless network.

[0025] Chlorophyll concentration and turbidity parameters were obtained by satellite remote sensing inversion. By retrieving remote sensing images of the target water area from satellites passing over it, the raster data with a spatial resolution of hundreds of meters was calculated using a band ratio inversion algorithm.

[0026] The water depth data is derived from electronic nautical charts or underwater topographic mapping results of the target water area. This invention processes the data from the above sources into a grid according to spatial location, and uniformly resamples it to a regular grid with the same spatial resolution. Each environmental parameter within each grid cell is assigned a value based on measured or inverted values.

[0027] The trajectory sequence and net type of the fishing vessels are derived from the dynamic monitoring platform for fishing vessels that has been deployed and operated by the fisheries management department in the target waters.

[0028] All fishing vessels within the jurisdiction that meet the legal length standard have been equipped with Beidou satellite navigation terminals. These terminals report the vessel's longitude, latitude, speed, course, and positioning time to the monitoring platform at a frequency of minutes.

[0029] This invention extracts all fishing vessel trajectory records during the release operation window from the monitoring platform through a data interface, and sorts them by fishing vessel code and positioning time to form a trajectory sequence for each fishing vessel.

[0030] Each fishing vessel has its main fishing gear type entered in its vessel registration information. This invention associates and calls the vessel code of each fishing vessel with the fishing gear type field in the registration information.

[0031] The reported locations for recaptured tagged fish are derived from the recapture registration records collected and compiled by various grassroots fisheries management stations.

[0032] During the release operation, some seedlings are tagged before being released. The tagged methods include external hanging of tags or internal implantation of coded tags.

[0033] After catching a tagged fish, fishermen must take their catch to the local fisheries management station for registration. The management personnel will use a positioning terminal to collect the latitude and longitude coordinates of the catch point and record the tagged fish number, the catch date, and the vessel code of the fishing boat that caught the fish.

[0034] Each grassroots station compiles the recapture registration records into the database of the superior management department on a monthly basis. This invention retrieves the above records from the database through a data interface as the source of the real recapture hard tags.

[0035] Each data acquisition process uses the release time of the release batch as the time benchmark, and only calls the fishing vessel trajectory data and recapture registration data within a specified time after the release time to ensure the logical consistency of data from various sources in the time dimension.

[0036] Please see Figure 2 As shown, in the heterogeneous graph construction module, each spatial unit is a spatial node, each fishing vessel is a vessel node, and each release batch is an event node. The operational occupancy relationship between vessel nodes and spatial nodes is determined based on the trajectory dwell time and frequency. The spatial affiliation relationship between event nodes and spatial nodes is determined based on the release coordinates. This process constructs the heterogeneous graph topology, specifically including: The target water area is divided into several regular rectangular spatial units at 0.1-degree intervals of latitude and longitude. Each spatial unit has a unique spatial code. These spatial codes are arranged according to the row and column positions of the spatial unit to form the spatial code of that spatial unit. At the same time, a one-to-one correspondence record between the spatial code and the latitude and longitude range of the boundary of the spatial unit is established as a spatial node index table.

[0037] After obtaining the trajectory sequence of each fishing vessel, the speed values ​​between two adjacent trajectory points are calculated sequentially. The absolute value of the speed difference between two adjacent trajectory points is used as the criterion. When the absolute value of the speed difference is greater than 0.2 meters per second, the next trajectory point is taken as the starting boundary point of the operation stop segment. Multiple consecutive trajectory points that meet the criterion are grouped into the same operation stop segment to obtain the set of operation stop segments for the fishing vessel. Each operation stop segment contains the spatial coordinates of all trajectory points within the stop segment.

[0038] For any operational stop segment, extract the spatial coordinates of all trajectory points within that segment, determine which spatial unit's boundary latitude and longitude range each coordinate falls into, extract the spatial code corresponding to the spatial unit, and obtain the spatial code set for that operational stop segment. Using each spatial code in this set as a matching key, search for the corresponding spatial unit in the spatial node index table, establish an operational occupancy relationship between the fishing vessel node and the found spatial unit, and record the fishing vessel node's stay duration and turning frequency in that spatial unit.

[0039] Extract the release coordinates of each release batch, determine the latitude and longitude range of the boundary of the spatial unit into which the release coordinates fall, extract the spatial code corresponding to the spatial unit, use the spatial code as the matching key to search the spatial node index table, establish a spatial affiliation relationship between the found spatial unit and the release event node, and record the spatial unit as the release landing point spatial unit of the release batch.

[0040] After establishing the spatial attribution relationship of each release batch, the start and end time interval of the release batch is extracted. From the ship trajectory sequence within the time interval, trajectory segments whose spatial coordinates fall within the range of nine spatial units, including the row and two adjacent rows, the column and two adjacent columns of the release landing point spatial unit of the release batch, are selected. The spatial code corresponding to each spatial unit traversed by the trajectory segment is recorded sequentially to form the trajectory spatial code sequence of the release batch.

[0041] Using each spatial code in the trajectory spatial coding sequence as a matching key, the spatial node index table is retrieved, and all successfully matched spatial units are associated with the release event node. All successfully matched spatial units are gathered to form a spatial association group for the release batch. The latitude and longitude range of the boundary of all spatial units covered by the spatial association group is determined as the release influence space range of the release batch.

[0042] Based on the established operation occupancy relationships and spatial ownership relationships, all spatial nodes, all ship nodes, and all event nodes are classified according to node type. Ship nodes are connected to spatial nodes by operation occupancy relationship edges, and event nodes are connected to spatial nodes by spatial ownership relationship edges. All nodes and edge relationships are aggregated to form a heterogeneous graph topology.

[0043] Please see Figure 3 As shown, in the feature transmission and pressure accumulation module, the fishing effort features of each vessel node are transmitted to the associated spatial nodes along the operation occupancy relationship edge, accumulating to form the comprehensive fishing pressure value of each spatial node, and the release batch attribute is transmitted to the corresponding spatial node according to the spatial ownership relationship edge, specifically including: For each vessel node, extract each operational stop segment from the vessel node's trajectory sequence, obtain the start and end times of each operational stop segment, and use the difference between the end time and the start time as the stop duration of the operational stop segment, with the unit of stop duration being hours; within the operational stop segment, calculate the difference in heading angle between two adjacent trajectory points sequentially, and use the number of times the heading angle difference is greater than 30 degrees as the turning frequency of the operational stop segment, with the unit of turning frequency being times; multiply the stop duration value of the operational stop segment by the turning frequency value, and the product is used as the fishing effort feature of the vessel node on the spatial node associated with the operational stop segment, with the unit of fishing effort feature being times per hour.

[0044] For any spatial node, obtain a list of all vessel nodes associated with that spatial node through the operation occupancy relationship edge, and the fishing effort characteristic value of each associated vessel node on that spatial node. Sum the fishing effort characteristic values ​​of all vessel nodes in the list, and use the sum as the comprehensive fishing pressure value of that spatial node.

[0045] After calculating the comprehensive fishing pressure value for all spatial nodes, for any spatial node, obtain the release batch number of the release event node associated with the spatial node through the spatial affiliation edge, and record the release batch number as the affiliation batch number of the spatial node; if a spatial node is associated with multiple release event nodes at the same time, then record the spatial node as belonging to multiple release batches at the same time, and record the release batch numbers of the multiple release batches one by one in the affiliation batch number list of the spatial node.

[0046] In the implicit probability estimation module, using the recapture point as an anchor, the probability of a marked fish being captured but not reported is inferred through reverse estimation of the edge relationships in the heterogeneous graph topology, generating an implicit recapture probability field, specifically including: The spatial node where the actual capture point is located is selected as the anchor node. For each anchor node, all ship nodes associated with the anchor node through the operation occupancy relationship edge are extracted from the heterogeneous graph topology, along with the operation dwell time records of these ship nodes at the anchor node, as the basic data for subsequent reverse tracing.

[0047] For each anchoring node, the comprehensive fishing pressure value of that anchoring node is transmitted back along the operation occupancy relationship edge to each of its associated vessel nodes.

[0048] The specific method for returning the data is as follows: Obtain a list of all vessel nodes associated with the anchoring node, as well as the operating time of each vessel node at the anchoring node. At the same time, obtain the catch capacity coefficient of the net type registered to each vessel node. The catch capacity coefficient is the fishing efficiency constant corresponding to the net type. The value is 0.8 for trawls, 0.6 for purse seines, 0.4 for gillnets, and 0.3 for angling. The product of the operational dwell time of each vessel node and its respective capture capability coefficient is taken as the operational contribution of each vessel node to the anchoring node. The comprehensive fishing pressure value of the anchoring node is allocated according to the proportion of the operational contribution of each vessel node to the total operational contribution of all associated vessel nodes, and the allocated value is used as the recapture association strength value of each vessel node relative to the anchoring node.

[0049] The capture correlation strength value of each vessel node relative to the anchoring node is determined according to the following mathematical formula: ; In this mathematical formula, the parameters have the following meanings: Indicates the first The ship node is relative to the first The captured association strength value obtained from each anchor node; Indicates the first The overall fishing pressure value of each anchored node; Indicates the first The ship node at the Duration of operation at each anchor node; Indicates the first The capture capability coefficient of the type of net equipment equipped at each ship node; Indicates the first The total number of all ship nodes associated with each anchoring node; Indicates the first The ship node at the Duration of operation at each anchor node; Indicates the first The capture capability coefficient of the type of net equipment equipped at each ship node; The denominator represents the sum of the operational contributions of all associated vessel nodes.

[0050] For any given vessel node, obtain the capture association strength values ​​of that vessel node relative to all anchored nodes, sum these values, and use the total as the final capture association strength of that vessel node.

[0051] The final recapture correlation strength reflects the overall probability that the vessel node will make implicit contact with the marker fish throughout its entire operational trajectory.

[0052] For any given ship node, extract a list of all spatial nodes associated with that ship node through operation occupancy relationship edges from the heterogeneous graph topology, as well as the operation dwell time of that ship node on each spatial node.

[0053] The ratio of the operating dwell time of the ship node at each spatial node to the total operating dwell time of the ship node at all spatial nodes is used as the propagation distribution coefficient of each spatial node.

[0054] The final recapture association strength of each vessel node is distributed to each associated spatial node in a positive direction along the operation occupancy relationship edge, according to the propagation allocation coefficient of each spatial node.

[0055] The value obtained from the forward allocation is the recapture intensity contribution value received by each spatial node from the ship node. The principle of using the propagation allocation coefficient as the basis for allocation is that the longer a fishing vessel stays and operates in a certain spatial unit, the greater the probability that it will catch the marked fish in that spatial unit. Therefore, the recapture association intensity value allocated from the ship node to that spatial node should be greater.

[0056] While each space node receives the capture intensity contribution value from each ship node, the impact of the adjacency relationship between space units on the capture probability is considered.

[0057] For any spatial node, extract the latitude and longitude coordinates of the center point of the spatial node, as well as the latitude and longitude coordinates of the center points of all adjacent spatial nodes within the same release influence space. Calculate the geographic straight-line distance between the spatial node and each adjacent spatial node. Substitute the geographic straight-line distance value into the preset attenuation process to obtain the spatial distance attenuation value of each adjacent spatial node to the spatial node.

[0058] The preset attenuation process is as follows: Let the geographical straight-line distance be... (Unit: kilometers), base distance is ( The value is taken as the length of the diagonal of the spatial unit (in kilometers), then the spatial distance attenuation value is... Calculated using the following piecewise function: when hour, ; when hour, ,in is a natural constant with a value of 2.71828; when hour, .

[0059] when The exponential decay form is used to reflect the natural law that the spatial dispersal probability of the marker fish decreases continuously and smoothly with increasing distance; when The decay value is set to 0 directly to limit the effective influence radius of the marker fish, avoiding the interference of the accuracy of subsequent accumulation calculations caused by the exponential function being mathematically always greater than 0 at the far end and generating too small a non-zero probability value.

[0060] For any spatial node, the capture intensity contribution values ​​received by the spatial node from all associated vessel nodes are summed up, and then the summed result is multiplied by the spatial distance attenuation values ​​obtained by the spatial node from all adjacent spatial nodes. The product is used as the capture non-reporting probability value of the spatial node.

[0061] Among them, the spatial distance attenuation values ​​obtained by the same spatial node from multiple adjacent spatial nodes are first summed and then multiplied by the sum of the recapture intensity contribution values ​​of that spatial node.

[0062] The unreported capture probability value represents the inferred probability that a marker fish was actually caught at this spatial node in the absence of records of fishermen actively reporting the event.

[0063] The capture-without-reporting probability values ​​of all spatial nodes are organized according to the spatial arrangement order of each spatial node in the target water area to form an implicit recapture probability field covering the entire target water area. The value of each spatial node in this probability field is the capture-without-reporting probability value corresponding to that spatial node.

[0064] In the weight update and region labeling module, probability values ​​with confidence levels above a threshold in the probability field are used as soft labels and fused with the true capture hard labels. The weights of each environmental factor on the release effectiveness are updated using the fused label set. Based on the updated weights, the suitability index of each spatial unit is calculated, and the spatial unit with the highest suitability index is labeled as the recommended release area. Specifically, this includes: The probability value of each spatial node in the implicit capture probability field is used as the soft label of that spatial node. The value of the soft label is a real number between 0 and 1. The true capture hard label is defined as a binary label with a value of only 0 or 1, where the value 1 indicates that there is a true capture record for the spatial node, and the value 0 indicates that there is no true capture record for the spatial node. For each spatial node, its soft label and hard label are stored together to form the label set of the spatial node. The label set is accompanied by a confidence attribute, where the confidence of the hard label is set to 1, and the confidence of the soft label is the probability value of the soft label itself.

[0065] For each spatial node, the corresponding environmental raster data is retrieved. This environmental raster data includes water temperature, dissolved oxygen, salinity, pH, and chlorophyll values. The values ​​of these five environmental parameters are arranged in a fixed order to form the environmental vector of the spatial node.

[0066] Centered on each spatial node, three square neighborhoods with side lengths of three, five, and seven spatial units are selected respectively, serving as three spatial scales. For each spatial scale neighborhood, the environmental vectors of all spatial nodes within that neighborhood are extracted. The maximum and minimum values ​​of each environmental parameter across all spatial nodes within that neighborhood are calculated, and the difference between the maximum and minimum values ​​is taken as the range of that environmental parameter at that spatial scale. For each environmental parameter, the average of the three ranges obtained at the three spatial scales is taken to obtain the average range of that environmental parameter. The average range is then divided by the average value of that environmental parameter across all spatial nodes, and the resulting ratio is taken as the spatial variation coefficient of that environmental parameter.

[0067] The initial weights of all five environmental parameters are set to be equal values, which is 1. For each environmental parameter, its spatial coefficient of variation is divided by the sum of the spatial coefficients of variation of all five environmental parameters to obtain the coefficient of variation ratio of that environmental parameter. The coefficient of variation ratio of each environmental parameter is multiplied by a coefficient of 5 and used as the weight correction value of each environmental parameter, so that the sum of the weight correction values ​​of all five environmental parameters is equal to 5. The initial weight of each environmental parameter is 1 and the corresponding weight correction value is added to obtain the updated weight of each environmental parameter.

[0068] For each spatial node, multiply the value of each environmental parameter in its environment vector by the updated weight corresponding to that environmental parameter to obtain the weighted value of that environmental parameter; sum the weighted values ​​of all five environmental parameters of the spatial node, and use the sum as the suitability index of the spatial node.

[0069] Traverse all spatial nodes and compare the suitability index of each spatial node with the highest recorded suitability index. If the suitability index of the current spatial node is higher than the highest recorded value, replace the highest recorded value with the suitability index of the current spatial node and record the spatial code of the spatial node. After traversing all spatial nodes, use the spatial unit corresponding to the recorded spatial code as the recommended release area.

[0070] The working principle of this invention is as follows: The coordinates and quantity of release batches in the target water area, environmental raster data, trajectory sequences of fishing vessels, net types, and reported recapture points of marked fish are acquired. The target water area is discretized into spatial units and assigned spatial codes to establish a spatial node index table. Based on the matching relationship between the trajectory dwell segments of fishing vessels and spatial units, operational occupancy relationships are established between fishing vessel nodes and spatial nodes. Based on the spatial units into which the release coordinates fall, spatial attribution relationships are established between release event nodes and spatial nodes. A heterogeneous graph topology is constructed using node and edge relationships. The fishing effort characteristics of each fishing vessel node are transmitted to the spatial nodes along the operational occupancy relationship edges and accumulated into a comprehensive fishing pressure value. The release batch number is assigned to the spatial nodes along the spatial attribution relationship edges. The spatial node where the actual recapture point is located is selected as the anchor node. The comprehensive fishing pressure value is transmitted back to each vessel node along the operation occupancy relationship edge, and the final recapture association strength of each vessel node is obtained by summing them up. Then, it is propagated forward to each spatial node along the operation occupancy relationship edge, and multiplied with the spatial distance attenuation value to obtain the capture non-reporting probability value of each spatial node. The probability values ​​of all spatial nodes constitute an implicit recapture probability field. The probability values ​​in the implicit recapture probability field are fused with the actual recapture hard label as soft labels. The environmental raster data of each spatial node is used to construct an environmental vector, and the spatial variation coefficient of each environmental factor is calculated based on the difference of environmental vectors in the neighborhood of different spatial scales. The initial weight of each environmental factor is corrected with the variation coefficient, and the environmental vector of each spatial node is weighted and synthesized to obtain the suitability index of each spatial node. The spatial unit with the highest suitability index is selected as the recommended release area.

[0071] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A dynamic calibration system for determining the optimal effectiveness area of ​​stock enhancement and release, characterized in that, include: The data acquisition module acquires the coordinates and quantity of release batches in the target waters, environmental raster data, the trajectory sequence of fishing vessels and the type of fishing nets used, as well as the reported locations of the marked fish recapture points. The heterogeneous graph construction module uses each spatial unit as a spatial node, each fishing vessel as a ship node, and each release batch as an event node. It determines the operational occupancy relationship between ship nodes and spatial nodes based on the trajectory dwell time and frequency, and determines the spatial affiliation relationship between event nodes and spatial nodes based on the release coordinates, thus constructing the heterogeneous graph topology. The feature transmission and pressure accumulation module transmits the fishing effort features of each vessel node to the associated spatial node along the operation occupancy relationship edge, accumulates them to form the comprehensive fishing pressure value of each spatial node, and transmits the release batch attribute to the corresponding spatial node according to the spatial ownership relationship edge. The implicit probability estimation module uses the recapture point as an anchor and infers the probability of a marked fish being captured but not reported at each spatial node by inferring the relationship between the edges in the heterogeneous graph topology, thus generating an implicit recapture probability field. The weight update and region labeling module uses the probability values ​​with confidence levels higher than the threshold in the probability field as soft labels, merges them with the hard labels of the actual recapture, updates the weights of each environmental factor on the release effectiveness using the merged label set, calculates the suitability index of each spatial unit based on the updated weights, and labels the spatial unit with the highest suitability index as the recommended release area.

2. The system of claim 1, wherein, The construction of the heterogeneous graph topology specifically includes: Each spatial unit obtained after discretizing the target water area is used as a spatial node. Each spatial node is assigned a spatial code, and a spatial node index table is established. The trajectory sequence of each fishing vessel is divided into operation and rest segments and navigation segments according to the speed difference between adjacent trajectory points. The spatial code corresponding to the spatial coordinates of each operation and rest segment is used as the matching key to search the spatial node index table, associate the fishing vessel node with the matched spatial node, and establish the operation occupancy relationship. Using the spatial code corresponding to the spatial unit where the release coordinates of each release batch fall as the matching key, the spatial node index table is retrieved, and the release event node is associated with the matched spatial node to establish a spatial affiliation relationship. Based on the operation occupancy relationship and spatial ownership relationship, spatial nodes, ship nodes and event nodes are combined into a heterogeneous graph topology.

3. The system of claim 2, wherein the system is configured to: The establishment of spatial affiliation specifically includes: Extract the release coordinates and ship trajectory segments within the start and end time of each release batch, and record the spatial codes corresponding to each spatial unit traversed by the ship trajectory segments one by one to form the trajectory spatial code sequence of each release batch. Using each spatial code in the trajectory spatial coding sequence as a matching key, the spatial node index table is retrieved, and all spatial nodes that successfully match any spatial code in the trajectory spatial coding sequence are associated with the release event nodes to form a spatial association group for the release batch. All spatial nodes in the spatial association group of each release event node are determined as the spatial range of the release impact of the release batch.

4. The system of claim 1, wherein the system is configured to: The process of transferring the release batch attributes to the corresponding spatial nodes specifically includes: The trajectory dwell time and turning frequency of each vessel node are obtained, and the product of dwell time and turning frequency is used as the fishing effort feature of the vessel node. The fishing effort characteristics are assigned to each associated spatial node along the edge of the operation occupancy relationship. When the same spatial node receives values ​​from multiple ship nodes, the values ​​are accumulated to form the comprehensive fishing pressure value of the spatial node. The release batch number of each release event node is assigned to each associated spatial node along the spatial affiliation edge, so that each spatial node records the release batch number to which it belongs.

5. The dynamic calibration system for determining the optimal effectiveness area of ​​stock enhancement and release according to claim 1, characterized in that, The generation of the implicit capture probability field specifically includes: The spatial nodes where the actual recapture points are located are selected as anchor nodes, and the comprehensive fishing pressure value of each anchor node is obtained. Based on the comprehensive fishing pressure value of each anchoring node, trace back to each vessel node along the edge of the operation occupation relationship, and transmit the comprehensive fishing pressure value of each anchoring node back according to the pressure contribution ratio of each vessel node to the anchoring node to obtain the recapture correlation strength of each vessel node. Starting with the recapture association strength of each ship node, the algorithm propagates forward along the operation occupancy relationship edge to all spatial nodes associated with the ship node. The capture-out correlation strength received by each space node from different ship nodes is accumulated according to the spatial distance attenuation value to obtain the capture-out-of-report probability value of each space node. The implicit capture-out probability field is composed of the probability values ​​of all space nodes.

6. The system of claim 5, wherein the system is configured to: The obtained recapture correlation strength of each ship node specifically includes: Obtain all vessel nodes associated with each anchoring node and the operational dwell time of each vessel node at the anchoring node. The operational dwell time of each vessel node is multiplied by the capture capability coefficient of the type of net equipped by the vessel node as the operational contribution of the vessel node to the anchoring node. The comprehensive fishing pressure value of the anchored node is allocated according to the proportion of the operational contribution of each vessel node to the total operational contribution of all related vessel nodes, and the allocated value is used as the recapture correlation strength value of each vessel node relative to the anchored node. The final capture association strength of a vessel node is obtained by summing up the multiple capture association strength values ​​obtained for the same vessel node relative to multiple anchored nodes.

7. The system of claim 5, wherein the system further comprises a plurality of sensors for detecting the presence of the one or more species of interest in the one or more zones of interest. The forward propagation specifically includes: Obtain all spatial nodes associated with each ship node and the corresponding operational dwell time of each spatial node, and use the ratio of the operational dwell time of the ship node at each spatial node to the total operational dwell time of the ship node as the propagation allocation coefficient; The final recapture association strength of the ship node is allocated according to the propagation allocation coefficient of each space node, and the allocated value is assigned to the corresponding space node as the recapture strength contribution value received by the space node from the ship node. The capture intensity contribution values ​​received by the same space node from multiple ship nodes are summed to obtain the capture non-reporting probability value of the space node.

8. The system of claim 1, wherein: The calibration process for the recommended release area is as follows: The probability values ​​of each spatial node in the implicit capture probability field are used as corresponding soft labels, which together with the true capture hard labels constitute a label set with confidence attributes for each spatial node. The environmental raster data of each spatial node is used to construct the environmental vector of the spatial node. Based on the difference in environmental vectors of the same spatial node in different spatial scale neighborhoods, the coefficient of variation of each environmental factor in the spatial dimension is determined. The initial weight of each environmental factor is corrected by the coefficient of variation. The environmental factor with the larger coefficient of variation is assigned a higher weight correction value. The environmental vector is weighted and synthesized using the corrected weights of each environmental factor, and the suitability index of each spatial node is calculated. The spatial unit with the highest suitability index is selected as the recommended release area.

Images