Laying hen epidemic disease biological prevention and control system and method

Abstract

The application relates to the technical field of laying hen epidemic disease biological prevention and control knowledge graph reasoning, and discloses a laying hen epidemic disease biological prevention and control system and method. The system converts multi-medium prevention and control flow data of a laying hen farm into a flow medium state fragment set, generates fragment entities from the flow medium state fragment set, generates area entities from contact areas, constructs an explicit flow subgraph, a pollution residue subgraph and a rule knowledge layer for pollution residue closed reasoning, forms a laying hen farm multi-medium prevention and control knowledge graph, extracts a preceding fragment and a subsequent fragment in time sequence in the same contact area from the knowledge graph, calls the rule knowledge layer to reason about a pollution residue closed relationship, obtains a pollution residue closed state set, takes an abnormal occurrence area and an abnormal occurrence time corresponding to an epidemic disease anomaly as a reverse reasoning starting point, generates a candidate hidden transmission chain and calculates a transmission chain contribution result, and outputs a partitioned biological prevention and control strategy.

Description

Technical Field

[0001] This application relates to the field of knowledge graph reasoning technology for the biological control of diseases in laying hens, and in particular to a biological control system and method for diseases in laying hens. Background Technology

[0002] Large-scale layer hen farms typically employ multi-building, zoned management, and continuous production processes including egg collection, manure removal, feed delivery, dead animal transport, inspection and maintenance, and disinfection. In actual operation, personnel, vehicles, egg trays, manure, and shared equipment repeatedly circulate between chicken houses, passageways, egg collection areas, manure removal areas, disinfection areas, and material storage areas. Different items in different circulations form cross-contact relationships at different times and in different contact areas. These circulation relationships are also affected by temperature and humidity, ventilation conditions, ground moisture levels, manure exposure time, disinfection intervals, and the implementation of personnel protective measures. When there are weak links in disease prevention in a particular chicken house, passageway, or circulation area, pathogens may spread to other areas via shoe soles, egg tray surfaces, equipment contact surfaces, manure residue, or vehicle tires, forming a hidden transmission chain that is difficult to detect directly through a single inspection. Therefore, biological control of diseases in layer hen farms requires a comprehensive analysis of the spatiotemporal contact relationships and residual impact relationships of various circulation items in the continuous production process.

[0003] In existing technologies, knowledge graphs, graph neural networks, or graph reasoning methods have been used for tracing the source of livestock and poultry diseases, analyzing abnormal transmission paths in farms, and identifying the correlation of epidemic prevention events. These methods typically abstract farmed objects, farmed areas, test results, and epidemic prevention events into entity nodes, and infer potential transmission paths through ordinary association edges, node adjacency relationships, or path lengths. However, in the multi-media flow scenario of large-scale egg farms, personnel, egg trays, manure, vehicles, manure removal equipment, feed delivery batches, and tools have different contamination carrying capacities, contact surfaces, disinfection reduction conditions, and residue retention conditions. If these objects are expressed according to a uniform entity relationship edge, it is easy to confuse simultaneous contact, sequential residue carrying, and cross-regional flow into the same type of transmission relationship, leading to problems in the graph reasoning process. High-risk link misconnections, misjudgments of disinfection blocking, and missed detections of residual effects occur in the process. Existing graph reasoning methods usually use node adjacency, path reachability, or feature similarity as the basis for inferring the transmission chain. They do not organize the rules of bearing strength, surface compatibility, time decay, disinfection reduction, environmental maintenance, and closure establishment into a rule knowledge layer that can participate in reasoning. They also do not independently express the non-simultaneous contact relationship between the residual effect of the preceding segment and the subsequent segment as a candidate relationship of pollution residue. As a result, existing technologies have the problem of being unable to convert multi-media epidemic prevention records of egg-laying chicken farms into a multi-layer graph structure for knowledge reasoning, and it is difficult to reason about the closed state of pollution residues based on rule knowledge and reverse attribute the implicit transmission chain. This affects the accurate generation of zonal biosecurity strategies for egg-laying chicken farms. Summary of the Invention

[0004] This application proposes a biological control system and method for diseases in laying hens to address the problems mentioned in the background art.

[0005] To achieve the above objectives, this application adopts the following technical solution: a biological control system for laying hen diseases, comprising: The status segment generation module receives multi-media control and circulation data from the egg farm, generates circulation media status segments according to the continuous dwell process of the circulation object in the contact area, and segments the data when the circulation object changes, the contact area changes, or the time interruption exceeds the continuous judgment interval determined based on the working rhythm of the contact area, forming a collection of circulation media status segments. The knowledge graph construction module for prevention and control generates fragment entities from the set of state fragments of the flowing medium and regional entities from the contact area. It forms an explicit flow subgraph according to the fragment flow order and regional connectivity. It forms a contamination residue subgraph based on the time interval between the preceding and following fragments that are established in the same contact area, the surface compatibility result, and the disinfection interval. It also configures a rule knowledge layer for closed reasoning of contamination residue to form a multi-media prevention and control knowledge graph for egg farms. The residual closure knowledge reasoning module extracts the preceding and following fragments that are established in the same contact area from the multi-media prevention and control knowledge graph of the egg farm, calls the rule knowledge layer to reason about the pollution residue closure relationship, and obtains the pollution residue closure state set; The abnormal reverse attribution prevention and control module takes the abnormal occurrence area and time corresponding to the epidemic abnormality as the starting point for reverse reasoning. It generates candidate hidden transmission chains based on the explicit flow rotor diagram and the set of closed states of pollution residue. It calculates the contribution results of the transmission chain according to the link time continuity, the consistency of the abnormal direction and the closed state of pollution residue, and generates a regional biocontrol strategy.

[0006] A biological control method for diseases in laying hens includes the following steps: S1 receives multi-media control and circulation data from the egg farm, generates circulation medium status segments according to the continuous dwelling process of the circulation object in the contact area, and segments are cut when the circulation object changes, the contact area changes, or the time interruption exceeds the continuous judgment interval determined based on the working rhythm of the contact area, forming a set of circulation medium status segments. S2 generates fragment entities from the set of state fragments of the flowing medium and generates regional entities from the contact area, constructing an explicit flow rotor diagram, a pollution residue sub-diagram, and a rule knowledge layer for closed-loop reasoning of pollution residue, forming a multi-media prevention and control knowledge graph for egg farms; S3: Extract the preceding and following segments that are in the same contact area from the knowledge graph of multi-media prevention and control in egg farms, and call the rule knowledge layer to reason about the closed relationship of pollution residue to obtain the set of closed states of pollution residue. S4 uses the abnormal occurrence area and time of the abnormal occurrence corresponding to the epidemic abnormality as the starting point for reverse reasoning. Based on the explicit flow rotor diagram and the set of closed states of pollution residue, candidate hidden transmission chains are generated. The contribution results of the transmission chain are calculated according to the link time continuity, the consistency of the abnormal direction and the closed state of pollution residue, and a regional biocontrol strategy is generated.

[0007] The beneficial effects of this invention are as follows: 1. This invention converts multi-media prevention and control data from egg farms into a set of state fragments of the circulating media, and constructs a multi-media prevention and control knowledge graph of egg farms using fragment entities, regional entities, explicit circulation relationships, and candidate relationships of contamination residues. This transforms the circulation processes of personnel, vehicles, egg trays, manure, and shared equipment from scattered ledgers into a graph structure capable of performing knowledge reasoning, solving the problem that existing epidemic prevention records are difficult to support the attribution of hidden transmission chains.

[0008] 2. This invention configures a rule knowledge layer in the multi-media prevention and control knowledge graph of egg farms, and calls the load-bearing strength rule, surface compatibility rule, time decay rule, disinfection reduction rule, environmental maintenance rule and closure establishment rule to infer the closure status of contamination residues. This avoids misjudgment caused by judging risk based solely on the area passed or disinfection records, and enables the residual connection relationship between the preceding and following segments in the same contact area to be independently identified.

[0009] 3. This invention uses the abnormal occurrence area and time of the abnormal occurrence corresponding to the epidemic abnormality as the starting point for reverse reasoning, converts the closed state of the contamination residue into the residue transmission relationship and generates candidate hidden transmission chains. Based on the contribution results of the transmission chain, it determines the high-contribution hidden transmission chain, high-risk circulation medium, high-risk carrying area and the link that needs to be reviewed and disinfected. It can provide an interpretable basis for prevention and control for zoned review and disinfection, temporary blocking of paths and targeted tracing. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort: Figure 1 This is a system framework diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0011] 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.

[0012] Example 1 like Figure 1 As shown, this invention discloses a biological control system for laying hen diseases, comprising: a state fragment generation module, a control knowledge graph construction module, a residual closed knowledge reasoning module, and an abnormal reverse attribution control module.

[0013] In this embodiment, the state fragment generation module receives multi-media epidemic prevention and control flow data from the egg farm. This data includes flow operation records, epidemic prevention and control records, spatial carrying capacity records, environmental auxiliary records, and anomaly feedback records. Flow operation records characterize the flow of personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools within the egg farm. Epidemic prevention and control records characterize the disinfection execution process and verification results. Spatial carrying capacity records characterize the regional connectivity between chicken houses, passageways, egg collection areas, manure cleaning areas, disinfection areas, and material storage areas. Environmental auxiliary records... Auxiliary records are used to characterize the temperature, humidity, ventilation status, and ground condition within the contact area; abnormal feedback records are used to characterize the abnormal occurrence area and time corresponding to the disease abnormality. The above data can be obtained from access control systems, vehicle registration systems, egg tray turnover ledgers, manure removal equipment operation records, feed delivery records, tool borrowing records, disinfection execution records, environmental sensors, inspection terminals, and disease abnormality reporting records. Data collection, timestamp writing, equipment number reading, and area number configuration are existing information management methods. This implementation method uses them as data support for forming a set of state fragments of the circulating medium.

[0014] The status fragment generation module first performs fragment generation preprocessing on the multi-media prevention and control data flow data of the layer chicken farm. This preprocessing includes unified time base conversion, unified identification of flow objects, contact area coding mapping, and record reliability marking. Unified time base conversion refers to converting the recording times from different data sources to the same farm time base, which is the farm server time. When there is a discrepancy between the time of the acquisition device and the farm server time, the subsequent recording times of the acquisition device are corrected based on the time offset formed when the acquisition device last synchronized with the farm server. When no valid synchronization time exists, the recording time is not calculated or corrected, and the record reliability of the corresponding record is reduced. The unified identification of transfer objects refers to mapping the numbers of the same personnel, the same vehicle, the same egg tray batch, the same manure transfer batch, the same shared equipment, the same feed delivery equipment, the same feed delivery batch, or the same tool in different records to a unified transfer object identifier. The contact area coding mapping refers to mapping each transfer record to a contact area in the chicken house, passage, egg collection area, manure cleaning area, disinfection area, or material temporary storage area based on the spatial carrying record. When a record covers multiple contact areas, it is split into multiple sub-records according to the area boundaries and the corresponding entry and exit times. When a record cannot be mapped to a contact area, the record is not included in the transfer medium state fragment set and is written to the abnormal mapping record.

[0015] Record credibility is used to characterize the degree of trustworthiness of a record when forming a state fragment of a circulation medium. Its value ranges from 0 to 1. Record credibility is jointly determined by the completeness of the record source, the completeness of the timestamp, the completeness of the circulation object identifier, and the completeness of the contact area mapping. As a preferred implementation, records that are automatically collected and have all four pieces of information are recorded with a credibility of 0.90 to 1.00; records that are automatically collected but lack one non-time information are recorded with a credibility of 0.70 to 0.89; records that are manually supplemented and can correspond to the circulation object, contact area, and contact time period are recorded with a credibility of 0.50 to 0.69; records whose timestamps cannot be corrected or whose contact areas require manual confirmation are recorded with a credibility of 0.30 to 0.49; records whose contact areas cannot be determined and cannot be supplemented by spatial carrying records are not used to generate state fragments of a circulation medium; records that are missing circulation object identifiers but have complete contact areas and contact time periods are not processed as object-level fragments, but as region-level low-credibility fragments, and their credibility is limited to 0.30 to 0.49. The above values ​​are based on the fact that object-level circulating medium status segments need to have three basic types of information: object, region, and time period. Region-level low-reliability segments need to have at least two basic types of information: contact area and contact time period. The errors in automatic collection and recording mainly come from device synchronization and region mapping. The uncertainties in manual supplementary recording mainly come from input lag and inconsistent description granularity. Therefore, the reliability of the records is classified according to the completeness of information and the type of segment.

[0016] After completing the preprocessing for fragment generation, the status fragment generation module determines the continuous judgment interval. The continuous judgment interval is used to determine whether adjacent records belong to the continuous dwelling process of the same transfer object in the same contact area. The continuous judgment interval is determined based on the historical continuous operation interval, shift operation rules and equipment operating cycle of the same type of transfer object in the corresponding contact area. The historical continuous operation interval refers to the time interval between two consecutive valid records of the same transfer object in the same contact area.

[0017] When the number of historical continuous operation interval samples for the same type of circulating objects within the corresponding contact area is no less than 30 sets, the median of this type of sample is used as the base interval, and the upper quartile value is used to limit the upper limit of the continuous judgment interval, so that occasional long intervals are not mistakenly included in the continuous dwell process; when the number of historical continuous operation interval samples is less than 30 sets, the continuous judgment interval is determined according to the operation procedures, shift operation rules or equipment operating cycle of the corresponding contact area, and the record reliability of the corresponding records is reduced. The preferred range of the continuous judgment interval is 30 seconds to 30 minutes, where egg tray turnover and The continuous judgment interval for short-term tool borrowing is 30 seconds to 5 minutes, the continuous judgment interval for personnel inspection is 1 minute to 10 minutes, the continuous judgment interval for continuous operation of manure removal equipment is 3 minutes to 20 minutes, and the continuous judgment interval for vehicle parking and feed delivery is 5 minutes to 30 minutes. The above ranges are consistent with the common access control sampling granularity, egg tray turnover rate, inspection dwell time, manure removal equipment operation cycle, and vehicle parking management process in egg farms. This can reduce the risk of excessive segmentation caused by second-level recording fluctuations and also avoid the incorrect merging of long periods without records into the same segment.

[0018] The state segment generation module then generates state segments of the transfer medium according to the continuous dwelling process of the transfer object in the contact area. For continuous records of the same transfer object, the records are sorted according to the record time after a unified time base. When adjacent records correspond to the same contact area, have continuous contact time periods, and the time interruption does not exceed the continuous judgment interval, the adjacent records are merged into one state segment of the transfer medium. When there is a change in the transfer object or contact area between the current record and the adjacent subsequent record, or when the time interruption between the two exceeds the continuous judgment interval, the current state segment of the transfer medium ends and a new state segment of the transfer medium is generated. For continuous operation records across regions, the state segment generation module generates state segments of the transfer medium corresponding to the contact area according to the contact area coding mapping result, and retains the regional transfer relationship through the previous contact area and the next contact area. This process can simultaneously retain the movement order of the transfer object and the dwelling boundary in the specific contact area, providing a traceable segment basis for the subsequent explicit transfer sub-map and pollution residue sub-map.

[0019] Each fluid medium state segment is configured with a fluid object type, fluid object identifier, contact area, entry time, exit time, previous contact area, next contact area, contact surface type, corresponding epidemic prevention and control record, environmental auxiliary record, and record reliability. The fluid object type is used to distinguish personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools; the fluid object identifier is used to distinguish specific objects or batches within the same type; the contact area is used to characterize the spatial location where the fluid medium state segment occurs; the entry time and exit time are used to determine the time boundary of the segment; the previous contact area and the next contact area are used to retain the regional transfer relationship before and after the segment; the contact surface type is used to characterize shoe soles, tires, egg tray surfaces, tool surfaces, manure removal equipment contact surfaces, ground, or turnover table surfaces; the corresponding epidemic prevention and control record is used to associate the disinfection execution process and verification results before and after the segment; the environmental auxiliary record is used to record the temperature, humidity, ventilation status, and ground status during the corresponding period of the segment; and the record reliability is used to mark the completeness and acceptability of the data sourced from the segment.

[0020] Personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools in the transfer operation record are included in the segment body as transfer objects; chicken houses, passages, egg collection areas, manure cleaning areas, disinfection areas, and material storage areas in the spatial carrying capacity record are included in the segment location attribute as contact areas; temperature, humidity, ventilation status, and ground status in the environmental auxiliary record are included in the segment environmental attribute as conditional variables. The environmental auxiliary record is not included in the segment segmentation as a transfer object, nor is the epidemic prevention and control record. The environmental auxiliary record is used for the environmental maintenance result in the subsequent pollution residue closed inference, and the epidemic prevention and control record is used for the disinfection reduction credibility in the subsequent pollution residue closed inference.

[0021] For anomalous data, the state fragment generation module performs boundary processing. Records with missing object identifiers but complete contact areas and contact periods are generated as low-confidence fragments at the regional level, with the record confidence level limited to 0.30 to 0.49. Records with missing contact areas that cannot be supplemented by spatial carrying records are not generated as state fragments for the circulating medium and are written into the anomalous mapping record. Records with entry times later than exit times are corrected first based on adjacent records from the same source; if correction is not possible, the generation of fragments for that time period is discarded. Records with overlapping times within the same circulating object and the same contact area are merged into the same continuous stay process. When epidemic prevention and control records or environmental auxiliary records are missing, it does not affect the generation of state fragments for the circulating medium, but the missing corresponding related information is marked in the fragment to reduce the confidence level of the corresponding disinfection reduction or environmental maintenance results for subsequent closed-loop inference of pollution residues.

[0022] In this embodiment, the prevention and control knowledge graph construction module uses the set of state fragments of the circulating medium generated by the state fragment generation module as the graph generation object to construct a multi-media prevention and control knowledge graph for egg-laying chicken farms. The multi-media prevention and control knowledge graph for egg-laying chicken farms includes an explicit circulating medium graph and a contamination residue subgraph. The explicit circulating medium graph is used to characterize the continuous movement relationship of circulating objects between different contact areas, as well as the regional contact relationship formed by different circulating objects in the same contact area. The contamination residue subgraph is used to characterize the indirect residual influence relationship of subsequent circulating medium state fragments after the previous circulating medium state fragment forms a residual carrier in the contact area. The graph entity storage, relation storage, and relation index retrieval can be implemented using a graph database, relation database, or time-series database. The data storage method belongs to the underlying management means of realizing the multi-media prevention and control knowledge graph for egg-laying chicken farms. The focus of this embodiment is on the organization of the entity layer, relation layer, and rule knowledge layer, as well as the knowledge reasoning process of the contamination residue closed state.

[0023] In this embodiment, the multi-media prevention and control knowledge graph of the egg-laying hen farm includes an entity layer, a relationship layer, and a rule knowledge layer. The entity layer includes fragment entities formed by the state fragments of the flowing medium and regional entities formed by the contact areas. The relationship layer includes explicit flow relationships, object flow relationships, regional contact relationships, and contamination residue candidate relationships. The rule knowledge layer includes load-bearing strength rules, surface compatibility rules, time decay rules, disinfection reduction rules, environmental maintenance rules, and closure establishment rules. The load-bearing strength rule is used to determine the contact load-bearing strength based on the type of flowing object, contact surface type, residence time, and record reliability of the preceding flowing medium state fragment. The surface compatibility rule is used to determine the surface compatibility result based on the contact surface compatibility mapping table. The time decay rule is used to determine the surface compatibility result based on the contact surface compatibility mapping table. The time decay result is determined based on the departure time of the preceding circulating medium state fragment, the entry time of the subsequent circulating medium state fragment, and the corresponding epidemic prevention operation procedures in the contact area; the disinfection reduction rule is used to determine the reliability of disinfection reduction based on the coverage, execution sequence, execution completeness, record reliability, and review results of the epidemic prevention and control records; the environmental maintenance rule is used to determine the environmental maintenance result based on the temperature, humidity, ventilation status, and ground status of the environmental auxiliary records within the corresponding time range; the closure establishment rule is used to determine the closure establishment conditions based on the preceding and subsequent circulating medium state fragments being located in the same contact area, the time sequence being established, the area identifier and timestamp being valid, and combined with the valid status of the object identifier or the area-level low-reliability fragment marker.

[0024] When applying conventional knowledge graph or graph neural network methods to trace the source of disease in egg-laying hen farms, they typically construct graphs by treating personnel, vehicles, equipment, areas, and test results as nodes of the same type or relationships of the same type, and inferring transmission paths based on node adjacency, path length, or node feature similarity. However, this approach has limitations in the multi-media control scenarios of egg-laying hen farms: First, personnel, egg trays, manure, vehicles, manure removal equipment, and tools have different contamination carrying capacities, and simply whether they pass through the same area cannot be considered as equivalent transmission edges. Second, after a preceding segment leaves the same contact area, a subsequent segment, even if not appearing simultaneously with the preceding segment, may still carry residual effects through the ground, turnover table, egg tray surface, or manure removal channel; conventional explicit contact edges cannot adequately represent such non-simultaneous contact relationships. Third, disinfection records do not necessarily equate to the elimination of contamination residues; judgment must be made in conjunction with disinfection coverage, execution sequence, execution completeness, and record reliability. Fourth, temperature, humidity, ventilation, ground condition, and manure exposure status are only conditions for residue retention and should not be directly used as flow object nodes in the construction of transmission edges using conventional graph models.

[0025] Therefore, this implementation breaks down the multi-media control process in egg-laying hen farms into an explicit flow diagram and a contamination residue sub-diagram. The explicit flow diagram is used to express the explicit contact relationships between the same flow object moving across regions and different flow objects forming within the same region. The contamination residue sub-diagram is used to express the indirect residual influence relationships on subsequent flow media state fragments after the preceding flow media state fragment leaves the contact area, based on surface compatibility results, time intervals, disinfection intervals, and environmental conditions. This two-layer diagram structure replaces the adjacent-potential-propagation in ordinary graph reasoning with two types of constrained propagation relationships: explicit flow-propagation and residual closed-loop propagation. This adapts to the epidemic prevention scenario where personnel, egg trays, manure, vehicles, and shared equipment flow together in egg-laying hen farms.

[0026] When constructing the explicit flow graph, the prevention and control knowledge graph construction module uses the state fragments of the flowing medium as fragment entities and the contact areas as region entities. The fragment entities carry the type of the flowing object, the identification of the flowing object, the entry time, the exit time, the previous contact area, the next contact area, the contact surface type, the corresponding epidemic prevention and control record, the environmental auxiliary record, and the record credibility. The region entities carry the contact area code, the region type, the region connectivity relationship, and the region epidemic prevention level. The contact areas include chicken houses, passages, egg collection areas, manure cleaning areas, disinfection areas, and material storage areas. The region connectivity relationship comes from the spatial carrying record and is used to determine whether there is a reachable path in the production process between two contact areas, such as the connectivity relationship between chicken houses and adjacent passages, passages and egg collection areas, manure cleaning areas and manure transfer paths, and disinfection areas and vehicle parking areas. Through this node setting, the explicit flow graph simultaneously retains the dwell boundary of the flowing object and the spatial connectivity basis between contact areas.

[0027] Explicit flow relationships include object flow relationships and regional contact relationships. Object flow relationships describe the continuous movement relationship formed by the same flow object across contact areas. The prevention and control knowledge graph construction module sorts the state segments of each flow medium under the same flow object identifier according to their entry and exit times. When the next contact area of ​​the previous flow medium state segment is consistent with the contact area of ​​the next flow medium state segment, or when the corresponding contact areas of the two have a direct connection relationship in the spatial carrying record, and the time sequence is valid, an object flow relationship is established between the two segment entities. The time sequence is valid when the exit time of the previous flow medium state segment is earlier than or equal to the entry time of the next flow medium state segment. The entry time, exit time, and time interval are all the result of the unified time base conversion in the state segment generation module, and are uniformly converted to minutes before comparison. The object flow relationship records the time interval between the segments, the direction of regional transfer, the type of flow object, and the record credibility.

[0028] When the time ranges of two segments overlap, the prevention and control knowledge graph construction module first determines whether the overlap is due to acquisition delay or upload delay. If it can be corrected based on the time synchronization offset of the same acquisition device, the time order is re-determined using the corrected entry and exit times. If it cannot be corrected, no object flow relationship is established, and the record credibility of the corresponding segment is reduced. If there is no reachable path between the two contact areas in the spatial carrying record, and there are no records of allocation, handling, transfer, or manual confirmation, no object flow relationship is established. This process can prevent segments with abnormal time records or those that are spatially inaccessible from being incorrectly connected as actual flow paths.

[0029] Regional contact relationships are used to describe the explicit contact relationships formed by different circulating objects within the same contact area. The prevention and control knowledge graph construction module selects the state segments of the circulating medium corresponding to different circulating objects within the same contact area, and continuously generates regional contact relationships based on overlapping contact time periods, compatible contact surfaces, and continuous operation sequence. Overlapping contact time periods refer to the existence of a common time range between two segments; compatible contact surfaces refer to the ability of the contact surfaces of two segments to form a contact relationship through the same bearing surface, such as shoe sole and ground, tire and ground, egg tray surface and turnover table surface, tool surface and equipment contact surface; continuous operation sequence refers to the fact that after the previous segment ends, the next segment enters the same contact area within a continuous judgment interval. Regional contact relationship records the contact area, contact surface type, overlap duration, adjacent duration, operation sequence direction, and the reliability of associated records.

[0030] For regional contact relationships, if two segments only satisfy the overlap of contact time periods but lack contact surface compatibility relationships, then the edge is marked as a low-confidence regional contact relationship. Low-confidence regional contact relationships are retained in the explicit flow chart for manual verification or subsequent supplementary verification, and are not included in the pollution residue candidate relationship generation process of the pollution residue subchart. If two segments are located in different contact areas and do not share a contact surface, regional contact relationships are not established. If the flow object identifier is missing, a low-confidence regional contact relationship is allowed to be established only when the segment has been marked as a regional-level low-confidence segment by the state segment generation module and the contact area and contact time period are complete. Object flow relationships are not established. Through this restriction, the explicit flow chart can express the explicit contact facts in real production, while reducing erroneous edge connections caused by missing entries in the ledger.

[0031] The construction of the contamination residue sub-graph is based on the explicit flow rotor graph and the set of flow medium state segments. The explicit flow rotor graph reflects the actual path and explicit contact relationship of the flow object. The contamination residue sub-graph reflects the possible relationship where subsequent segments inherit the residual impact in the same area after the preceding segment leaves residue in the contact area. The prevention and control knowledge graph construction module selects the preceding and subsequent flow medium state segments with a valid time sequence from the same contact area. Based on the time interval between the two, the surface compatibility result, and the disinfection interval, candidate contamination residue relationships are generated. The preceding flow medium state segment is the segment whose departure time is earlier than the subsequent flow medium state segment's entry time; the subsequent flow medium state segment is the segment that enters the same contact area after the preceding segment leaves. If the preceding and subsequent segments are not located in the same contact area, or the time sequence is not valid, no candidate contamination residue relationships are generated.

[0032] The time interval is used to characterize the duration from when the preceding flowing medium state segment leaves the contact area to when the subsequent flowing medium state segment enters the contact area. The time interval is uniformly converted to minutes before participating in the screening of pollution residue candidate relationships. The time interval does not directly represent the pollution risk, but is only used to form pollution residue candidate relationships in the pollution residue subgraph and to provide the residue closure knowledge reasoning module with time decay results. In order to reduce edges that have no actual connection for a long time, the time interval of pollution residue candidate relationships does not exceed the residue observation upper limit of the corresponding contact area. The residue observation upper limit is determined by the contact area type and epidemic prevention operation procedures. It is used for screening pollution residue candidate relationships and is not used to directly determine pollution residue closure relationships.

[0033] The upper limit for residual observation is 0.5 hours to 48 hours, with 0.5 hours to 6 hours for dry and ventilated passages and disinfection areas, 2 hours to 24 hours for egg collection areas and material storage areas, and 6 hours to 48 hours for manure cleaning areas, damp ground areas, and manure-related areas. This range matches the cleaning frequency, ventilation conditions, and manure exposure of different contact areas in the laying hen farm, and can filter out fragment combinations that obviously lack temporal correlation, while retaining candidate residual relationships that need to be further judged by the residual closure knowledge reasoning module.

[0034] The surface compatibility result is determined by the contact surface compatibility mapping table, which is established based on the egg farm's disease prevention operation procedures and historical contamination review results. It is used to describe whether the contact surface of the preceding segment can form a residual contact relationship with the contact surface of the subsequent segment. The contact surface compatibility mapping table includes at least the correspondence between shoe soles and the ground, tires and the ground, manure removal equipment contact surfaces and manure channels, egg tray surfaces and turnover table surfaces, tool surfaces and equipment contact surfaces, and feed packaging and material storage areas.

[0035] The surface compatibility result ranges from 0 to 1. For surface combinations confirmed by historical pollution review results to have residual connection relationships, the surface compatibility result is 0.70 to 1.00. For surface combinations confirmed by epidemic prevention operation procedures to have contact opportunities but lacking historical pollution review support, the surface compatibility result is 0.30 to 0.69. For surface combinations lacking direct or indirect connection relationships, the surface compatibility result is 0. When the surface compatibility result is greater than 0, a pollution residue candidate relationship can be generated. When the surface compatibility result is 0, no pollution residue candidate relationship is generated. This surface compatibility result is used for pollution residue candidate relationship screening in the epidemic prevention knowledge graph construction module and further participates in the pollution residue closure relationship determination in the residue closure knowledge reasoning module.

[0036] The disinfection interval is used to characterize the temporal position of the epidemic prevention and control record between the preceding and subsequent circulating media state segments. The epidemic prevention and control knowledge graph construction module retrieves epidemic prevention and control records covering the same contact area after the preceding circulating media state segment leaves the contact area and before the subsequent circulating media state segment enters the contact area. If multiple epidemic prevention and control records exist, the last disinfection execution process completed within this time period that covers the same contact area is used as the epidemic prevention and control record corresponding to the contamination residue candidate relationship. The disinfection interval is the time difference between the completion time of the disinfection execution process and the entry time of the subsequent circulating media state segment, and is uniformly converted to minutes. The contamination residue candidate relationship records the disinfection execution area, disinfection execution completion time, disinfection method, verification result, and disinfection interval. If there is no epidemic prevention and control record covering the same contact area and with a valid temporal position, the contamination residue candidate relationship is marked as having no valid disinfection record. The disinfection interval does not directly delete the contamination residue candidate relationship because whether disinfection reduction is valid also depends on the coverage, execution integrity, and verification result. The subsequent residue closure knowledge reasoning module continues to judge based on the credibility of disinfection reduction. This process can avoid directly equating the execution of disinfection with the disappearance of residual effects.

[0037] When the preceding medium state segment, contact area, subsequent medium state segment, and time sequence corresponding to a candidate relationship for contamination residue are all valid, and the surface compatibility result is greater than 0, the prevention and control knowledge graph construction module writes the candidate relationship for contamination residue into the contamination residue subgraph. Each candidate relationship for contamination residue records the preceding medium state segment identifier, the subsequent medium state segment identifier, the contact area, the time interval, the surface compatibility result, the disinfection interval, the environmental auxiliary record index, and the credibility of the candidate relationship for contamination residue. The environmental auxiliary record index is used to point to the temperature, humidity, ventilation status, and ground status within the time range and contact area corresponding to the candidate relationship for contamination residue. The credibility of the candidate relationship for contamination residue is determined by the preceding medium state segment identifier, the subsequent medium state segment identifier, the contact area, the time interval, the surface compatibility result, the disinfection interval, the environmental auxiliary record index, and the contamination residue candidate relationship credibility. When the reliability of segment records, the reliability of subsequent segment records, the integrity of contact area mapping, and the source of surface compatibility mapping are determined, and both preceding and following segments are automatically collected, the contact area is complete, and the surface compatibility result is supported by historical pollution verification results, the reliability of the pollution residue candidate relationship is taken as 0.80 to 1.00; when there are manually supplemented records in the preceding and following segments, or when the surface compatibility result is only supported by epidemic prevention operation procedures, the reliability of the pollution residue candidate relationship is taken as 0.50 to 0.79; when there are regional-level low-reliability segments in the preceding and following segments, the reliability of the pollution residue candidate relationship is taken as 0.30 to 0.49; when the contact area cannot be determined, the time sequence is not valid, or the surface compatibility result is 0, no pollution residue candidate relationship is generated.

[0038] During the construction of the pollution residue subgraph, environmental auxiliary records are not used as independent nodes or as the basis for edge generation. Instead, they are used as condition indexes for pollution residue candidate relationships. Temperature, humidity, ventilation status, and ground status are residue maintenance conditions used by the residue closure knowledge reasoning module to form environmental maintenance results. If environmental auxiliary records are directly used as transfer objects or pollution source nodes in the prevention and control knowledge graph construction module, it will cause confusion in physical meaning. Therefore, this implementation method only associates environmental auxiliary records with pollution residue candidate relationships, so that the residue closure knowledge reasoning module can call the environmental conditions of the corresponding time period and the corresponding contact area when determining the pollution residue closure relationship.

[0039] Through the knowledge graph construction module for epidemic prevention and control, a multi-media epidemic prevention and control knowledge graph for egg-laying chicken farms is obtained, consisting of an explicit flow subgraph and a contamination residue subgraph. The explicit flow subgraph retains the actual flow sequence and regional contact relationships of personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools within the egg-laying chicken farm. The contamination residue subgraph independently expresses the possible residue succession relationships between successive segments within the same area. Thus, the system can simultaneously represent the actual path traversed by the flowing objects and the indirect relationship between the residue of the preceding segment and the subsequent segment, solving the problem that ledger-style epidemic prevention records are difficult to support the reasoning of implicit transmission chains. The subsequent residue closure knowledge reasoning module can directly extract the preceding and subsequent segments with the established time sequence within the same area from the multi-media epidemic prevention and control knowledge graph of the egg-laying chicken farm, and combine the surface compatibility results, time decay results, disinfection reduction credibility, and environmental maintenance results to generate a set of contamination residue closure states.

[0040] In this embodiment, the residual closure knowledge reasoning module reads the preceding and subsequent flow media state fragments corresponding to the candidate relationships of contamination residues from the multi-media prevention and control knowledge graph of the egg farm, and calls the rule knowledge layer to generate contact bearing strength, surface compatibility result, time decay result, disinfection reduction confidence, environmental maintenance result, and closure establishment condition. Then, based on the contact bearing strength, surface compatibility result, time decay result, disinfection reduction confidence, environmental maintenance result, and closure establishment condition, it infers the contamination residue closure strength, and generates the contamination residue closure state when the contamination residue closure strength reaches the residual closure threshold.

[0041] The residual closure knowledge reasoning module, based on the multi-media prevention and control knowledge graph of the egg-laying chicken farm obtained from the prevention and control knowledge graph construction module, performs residual contamination closure reasoning on the state fragments of the circulating media arranged in chronological order within the same contact area, generating a set of residual contamination closure states. The multi-media prevention and control knowledge graph of the egg-laying chicken farm includes an explicit circulating media graph and a residual contamination subgraph. The explicit circulating media graph provides the actual movement sequence of the circulating objects and the basis for regional connectivity, while the residual contamination subgraph provides candidate relationships of residual contamination between preceding and subsequent circulating media state fragments. Graph entity storage, relation index retrieval, and fragment sorting can be implemented using graph databases, relational databases, or time-series databases. This type of data management method belongs to the underlying management means of realizing the multi-media prevention and control knowledge graph of the egg-laying chicken farm. The focus of this implementation is to jointly reason about the residual contamination closure relationship through contact bearing strength, surface compatibility results, time decay results, disinfection reduction credibility, environmental maintenance results, and closure establishment conditions, so that the candidate relationships of residual contamination are further transformed into residual contamination closure states that can be used for subsequent anomaly backtracking.

[0042] The residual closure knowledge reasoning module first reads the candidate relationships of pollution residues from the pollution residue subgraph, and extracts the preceding flow medium state fragment, the following elements: contact area, time interval, surface compatibility result, disinfection interval, environmental auxiliary record index, and the credibility of the pollution residue candidate relationship. Before a pollution residue candidate relationship enters the pollution residue closure reasoning, it needs to meet three preconditions: the preceding flow medium state fragment and the following flow medium state fragment are located in the same contact area; the departure time of the preceding flow medium state fragment is earlier than the entry time of the following flow medium state fragment; and the surface compatibility result is greater than 0. If any precondition is not met, the residual closure knowledge reasoning module does not generate a pollution residue closure state and marks the corresponding pollution residue candidate relationship as an invalid pollution residue candidate relationship.

[0043] The residual closed-loop knowledge reasoning module performs a unified unit conversion on the time data involved in the calculation. The departure time of the preceding flow medium state segment, the entry time of the subsequent flow medium state segment, and the completion time of disinfection are all based on the results of the unified time reference conversion in the state segment generation module. The time interval is uniformly converted to hours before participating in the time decay calculation. The residual decay coefficient uses the reciprocal of the hour as the unit. Thus, the time interval and the residual decay coefficient are multiplied to form a dimensionless exponential term, avoiding inconsistencies in dimensions.

[0044] After extracting candidate pollution residue relationships, the residue closure knowledge reasoning module quantifies the pollution residue closure relationship between the preceding and subsequent flow medium state segments. This judgment adopts a multiplication of constraints item by item because the pollution residue closure relationship needs to simultaneously satisfy the following conditions: the source segment has pollution carrying capacity, the contact surface of the preceding and subsequent segments can bear the pollution, the time interval is still within the range where the residue can be affected, the epidemic prevention and control measures have not been sufficient to reduce the pollution, the environmental conditions support the retention of the residue, and the basic closure conditions are met. If any key condition is not met, the corresponding segment combination is not recognized as a valid pollution residue closure relationship.

[0045] When determining the contamination residue closure relationship between preceding and subsequent flowing medium state segments, the residue closure knowledge reasoning module first checks whether the basic closure conditions are met. These basic closure conditions include: the preceding and subsequent flowing medium state segments are located in the same contact area; the departure time of the preceding flowing medium state segment is earlier than the entry time of the subsequent flowing medium state segment; the area identifier and timestamp are valid; and both have valid object identifiers; or one of them has been marked as a low-confidence segment at the region level by the state segment generation module and has a complete corresponding contact area and contact time period. If the basic closure conditions are not met, the residue closure knowledge reasoning module does not generate a contamination residue closure state.

[0046] When the basic closure conditions are met, the residual closure knowledge reasoning module further determines the contact bearing strength of the preceding flow medium state segment, the surface compatibility results between the preceding and subsequent segments, the time decay results between the preceding and subsequent segments, the disinfection reduction confidence, the environmental maintenance results, and the effective disinfection blocking conditions. Contact bearing strength characterizes the basic ability of the preceding flow medium state segment to form residual contamination; surface compatibility results characterize whether there is a possibility of residual adhesion between the contact surfaces of the preceding and subsequent segments; time decay results characterize the degree to which the residual impact decreases with increasing time interval after the preceding flow medium state segment leaves the contact area; disinfection reduction confidence characterizes the degree of confidence in the reduction of residual impact by the disinfection treatment from the departure of the preceding segment to the entry of the subsequent segment; and environmental maintenance results characterize the likelihood of residual contamination continuing under corresponding temperature, humidity, ventilation, ground conditions, and fecal exposure conditions after disinfection.

[0047] Effective disinfection blocking conditions are jointly determined by disinfection coverage matching results, disinfection timing matching results, disinfection execution completeness, disinfection record reliability, and verification results. If, after the departure of a current sequence of circulating media state segments and before the arrival of a subsequent sequence of circulating media state segments, there exists a record of epidemic prevention and control covering the same contact area, occurring between the two sequences, with complete disinfection execution, reliable disinfection records, and satisfactory verification results, then the effective disinfection blocking condition is deemed established. When the effective disinfection blocking condition is established, the residual closure knowledge reasoning module will determine that the corresponding contamination residual candidate relationship has been effectively blocked, and no contamination residual closure state will be generated.

[0048] When the effective disinfection blocking conditions are not met, the residual closure knowledge reasoning module continues to determine whether the remaining residue after disinfection can still be maintained under environmental conditions. Specifically, it first forms a contact bearing strength based on the type of the object being transported, the type of the contact surface, the residence time, and the reliability of the segment record of the preceding flow medium state segment; then it forms a surface compatibility result based on the contact surface compatibility mapping table; then it forms a time decay result based on the time interval between the departure of the preceding segment and the entry of the subsequent segment and the residual decay coefficient of the corresponding contact area; then it forms a disinfection reduction reliability based on the coverage, execution sequence, execution completeness, record reliability, and review results of the epidemic prevention and control records; and finally, it forms an environmental maintenance result based on the temperature, humidity, ventilation status, ground condition, and fecal exposure status within the corresponding time range.

[0049] The residual closure knowledge reasoning module uses contact bearing strength, surface compatibility results, time decay results, post-disinfection residual level, and environmental maintenance results as the basis for forming the residual closure strength. Specifically, the post-disinfection residual level is determined inversely by the disinfection reduction confidence; the higher the disinfection reduction confidence, the lower the post-disinfection residual level. The environmental maintenance result is only used to adjust the retention level of post-disinfection residuals and is not used as an independent source of contamination. A residual closure state is generated when contact bearing strength, surface compatibility results, time decay results, post-disinfection residual level, and environmental maintenance results collectively indicate that the residual impact of the preceding segment can still be inherited by the subsequent segment, and the corresponding results reach the residual closure threshold. No residual closure state is generated when any key condition is not met, or when the effective disinfection blocking condition is met.

[0050] The above process ensures that the pollution residue closure relationship is formed in the following order: basic closure condition judgment, effective disinfection blocking judgment, post-disinfection residual judgment, environmental maintenance judgment, and residual closure threshold judgment. This order is used to reflect that disinfection belongs to the active reduction condition and the environmental state belongs to the residual pollution maintenance condition, so as to avoid the pollution residue closure state being mistakenly generated simply because the environmental maintenance result is high when disinfection is sufficient and effective and the verification is qualified.

[0051] For circulating medium state segments that have been marked as regional-level low-confidence segments by the state segment generation module, they are allowed to participate in the contamination residue closure relationship judgment if their contact area and contact period are complete, the time sequence is valid, and the surface compatibility result is greater than 0. The missing object identifier state of such segments does not directly set the closure establishment indicator to 0; instead, its contribution to the contamination residue closure strength is reduced through the segment record confidence and the contamination residue candidate relationship confidence. When regional-level low-confidence segments exist in the preceding and subsequent circulating medium state segments, the corresponding segment record confidence is set to a value between 0.30 and 0.49, and is simultaneously marked as a low-confidence contamination residue candidate relationship in the contamination residue candidate relationship confidence.

[0052] Contact bearing strength The result is based on the type of the object being transferred. Contact surface bearing results Normalized results of dwell time Credibility of fragmented records form: ; in, This indicates that the calculation result will be limited to the range of 0 to 1; The basic carrying capacity result of the type of transferred object is determined by the historical residual review results or the epidemic prevention risk level of the site, which includes personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches and tools. The results of surface bearing capacity are determined by the material properties, cleaning difficulty, and surface wiping test results of shoe soles, tires, egg tray surfaces, manure removal equipment contact surfaces, ground surfaces, turnover table surfaces, and tool surfaces. This represents the normalized result of the dwell time, which is calculated by the ratio of the dwell time of the segment to the standard contact time of the contact area. When the dwell time of the segment reaches or exceeds the standard contact time of the contact area, it is taken as 1. The reliability of the segment record is determined by the integrity of the record source, the integrity of the timestamp, the integrity of the transfer object identifier, and the integrity of the contact area mapping in the state segment generation module. The design logic of this formula is to unify the pollution carrying capacity of the transfer object itself, the adhesion properties of the contact surface, the influence of the residence time, and the reliability of the record to the same dimensionless scale, so that the contact carrying capacity can be used as the basic input of the pollution residue closure strength.

[0053] The value range for the basic carrying capacity result of the transfer object type is 0 to 1, with personnel transfer objects corresponding to 0.70 to 1.00; when personnel work continuously across chicken houses or across zone passages, or when there are missing records of shoe disinfection, changing clothes, or personnel protection execution, the value for personnel transfer objects is 0.85 to 1.00; when personnel work only within the same zone and the protection records, shoe disinfection records, and changing clothes records are complete, the value for personnel transfer objects is 0.70 to 0.84. The value for manure-related transfer objects and manure removal equipment is 0.60 to 0.90; when manure is exposed, manure removal equipment operates across zones, or cleaning verification is not completed after manure removal, the value is 0.75 to 0.90; when manure is treated through a fixed manure removal process, closed pipeline, or fixed transfer path and cleaning verification is complete, the value is 0.60 to 0.74. For vehicle circulation objects, the value ranges from 0.50 to 0.85 in tire contact scenarios; for egg tray turnover objects, it ranges from 0.50 to 0.85; and for tool and feed delivery batches, it ranges from 0.30 to 0.75. The value range for the bearing capacity of the contact surface is 0 to 1, with rough, easily adhered, and difficult-to-clean surfaces corresponding to 0.60 to 1.00, and smooth, easy-to-clean surfaces with a fixed cleaning process corresponding to 0.20 to 0.59. The normalized dwell time result is calculated by converting the fragment dwell time to the standard contact time of the contact area, and the standard contact time of the contact area is not less than 1 minute. When the fragment dwell time reaches or exceeds the standard contact time of the contact area, the dwell time normalization result is set to 1. The reliability of the fragment record ranges from 0 to 1, and the standard contact time of the contact area is determined by the site operation procedures and historical continuous operation interval samples. When historical samples or operation procedures cannot provide a valid standard contact time, the standard contact time of the same type of contact area is used, and the dwell time normalization result is marked as a low reliability result.

[0054] The surface compatibility result is determined by the contact surface compatibility mapping table in the prevention and control knowledge graph construction module. For surface combinations that have been confirmed to have residual contact relationship by historical contamination review results, the surface compatibility result is 0.70 to 1.00. For surface combinations that have been confirmed to have contact opportunities by epidemic prevention operation procedures but lack historical contamination review support, the surface compatibility result is 0.30 to 0.69. For surface combinations that lack direct or indirect contact relationship, the surface compatibility result is 0. When the surface compatibility result is 0, no contamination residue closure state is generated. This setting is used to avoid directly identifying two fluid medium state segments as having a contamination residue closure relationship just because they are located in the same contact area.

[0055] The residual decay coefficient is determined by the contact area type, ventilation status, cleaning frequency, and historical residual verification records. It is used to characterize the overall trend of the residual impact that can be carried over by subsequent segments in the contact area decreasing over time. It is not equivalent to the pathogen half-life. The residual decay coefficient for dry ventilation areas and regularly cleaned areas is taken as the reciprocal of 0.60 to 3.00 per hour; the residual decay coefficient for egg collection areas, material storage areas, and ordinary passages is taken as the reciprocal of 0.15 to 0.80 per hour; the residual decay coefficient for damp ground areas, manure removal areas, and manure-related areas is taken as the reciprocal of 0.02 to 0.30 per hour. The specific values ​​are corrected by the historical wiping verification results of the site, the frequency of epidemic prevention cleaning, and ventilation operation records. When historical residual verification records are missing, the initial value is determined based on the area type and epidemic prevention operation procedures in the spatial carrying capacity record, and is updated after the verification results are generated.

[0056] Disinfection reduction credibility Matching results based on disinfection coverage Disinfection time sequence matching results Completeness of disinfection implementation And the credibility of disinfection records form: ; in, Indicates whether the disinfection area is covered by the fragment. and fragments The contact area; Indicate whether the disinfection completion time is within the segment. After leaving, until the fragment Before entering; The completeness of the disinfection process is indicated by the disinfection duration, disinfection method, status of the equipment used, and confirmation records of the personnel involved. This indicates the source of the disinfection record, the integrity of the timestamp, and the credibility of the disinfection record corresponding to the review result. All four items are normalized results, with values ​​ranging from 0 to 1. When there is no epidemic prevention and control record covering the contact area and with a valid time sequence, the credibility of disinfection reduction is set to 0. When the disinfection record lacks dosage, duration, coverage, or execution confirmation information, the corresponding normalized result is reduced and the disinfection step that needs to be reviewed is output. The design logic of this formula is that only when the disinfection action simultaneously meets the requirements of area coverage, time sequence validity, execution integrity, and record credibility can it generate a high credibility for residue reduction, avoiding the direct equivalent of the existence of a disinfection record to the disappearance of residue risk.

[0057] The environmental maintenance results are formed by normalizing and fusing temperature maintenance results, humidity maintenance results, ventilation suppression results, ground condition results, and fecal exposure results. Temperature maintenance results are obtained by comparing the average temperature within the time range corresponding to the candidate pollution residue relationship with the historical normal temperature range of the contact area. When the temperature is within the historical normal temperature range, a value of 0.30 to 0.60 is used; when the temperature continuously deviates from the historical normal temperature range and the historical residue verification results show that the corresponding residue retention is enhanced, a value of 0.61 to 1.00 is used; and when the temperature deviation corresponds to a decrease in residue retention, a value of 0.10 to 0.29 is used. Humidity maintenance results are obtained by comparing the average humidity within the time range corresponding to the candidate pollution residue relationship with the historical normal humidity range of the contact area. When the humidity is below the historical normal humidity range and the historical residue verification results show that the residue retention is reduced, a value of 0.10 to 0.29 is used; when the humidity is within the historical normal humidity range, a value of 0.30 to 0.60 is used; and when the humidity continuously exceeds the upper limit of the historical normal humidity range, a value of 0.61 to 1.00 is used. Ventilation suppression results are obtained by comparing the ventilation operation status... The results are determined by the air exchange status of the area; 0.10 to 0.40 is used when ventilation is normal, and 0.41 to 1.00 is used when ventilation is insufficient or stopped. The ground condition result is determined by whether the ground is dry, wet, has standing water, has fecal residue, or is properly cleaned; 0.10 to 0.30 is used when the ground is dry and properly cleaned, and 0.60 to 1.00 is used when the ground is wet, has standing water, or has fecal residue. Fecal exposure results are determined by fecal transfer records and manure removal equipment operation records; 0.10 to 0.30 is used when there are no fecal exposure records, and 0.30 is used when fecal exposure exists. Furthermore, when cleaning verification is not completed, the value is taken as 0.60 to 1.00. The environmental maintenance result is formed by the weighted average of the above results, and each weight is pre-normalized to a weight sum of 1. When there are sufficient historical residual verification samples, each weight is determined according to the contribution of the corresponding environmental result to the positive result of residual verification. When there are insufficient historical residual verification samples, each weight is configured equally and updated after the verification results are accumulated. Temperature, humidity, ventilation status and ground status are only used as conditional variables of pollution residue and are not used as flow medium objects to participate in the construction of contact edges.

[0058] When a certain type of environmental auxiliary record is missing within the time range corresponding to a candidate relationship for pollution residues, the candidate relationship is not directly deleted. If adjacent valid environmental records exist within the anomaly backtracking window for the same contact area, the valid environmental record closest to the time range of the candidate relationship for pollution residues is used to form the corresponding environmental result, and this result is marked as a low-confidence environmental result. If no adjacent valid environmental records exist, the corresponding environmental result is determined based on the epidemic prevention level of the contact area: 0.30 for low-risk contact areas, 0.45 for ordinary contact areas, and 0.60 for high-risk contact areas, damp ground areas, or areas related to feces and sewage. This result is marked as a low-confidence environmental result, reducing the confidence level of the environmental maintenance result. The environmental maintenance result is formed by the weighted average result after renormalization of the weights corresponding to the valid environmental results. When there are fewer than two types of valid environmental results, the closed state of pollution residues is not directly confirmed as a valid state, and the corresponding candidate relationship for pollution residues is written into the manual review list.

[0059] When the residual contamination closure intensity reaches the residual closure threshold, the residual closure knowledge reasoning module generates a residual contamination closure state. This state includes the residual source fragment, the residual carrying area, the effective residual time, the affected subsequent fragments, the residual transmission direction, and the corresponding disinfection reduction confidence level. The residual source fragment is the preceding circulating medium state fragment that forms the residual closure intensity; the residual carrying area is the contact area shared by the preceding and subsequent circulating medium state fragments; the effective residual time starts from the departure time of the residual source fragment and ends when the residual contamination closure intensity falls below the residual closure threshold; when a residual observation upper limit is set for the corresponding contact area, the effective residual time does not exceed this limit; the affected subsequent fragment is the subsequent circulating medium state fragment that receives the impact of the residual contamination; the residual transmission direction is from the preceding circulating medium state fragment to the subsequent circulating medium state fragment; and the corresponding disinfection reduction confidence level is used to indicate whether this closure state involves effective epidemic prevention and control records.

[0060] The residual closure threshold is used to determine whether a residual contamination closure state is established. The residual closure threshold is determined by historical negative operating samples and re-verified abnormal samples. When the number of historical samples is no less than 50, the 95th percentile of the residual contamination closure intensity in the historical negative operating samples is used as the initial threshold. Then, the initial threshold is corrected using the hidden transmission chains that can be manually confirmed in the re-verified abnormal samples, so that accidental transfer chains during negative operations are unlikely to form a residual contamination closure state, while retaining the effective residual relationships in the re-verified abnormal samples. When the number of historical samples is less than 50, the initial residual closure threshold is determined according to the site's epidemic prevention level, regional level, and operating procedures, with a value range of 0.45 to 0.75. Areas with high epidemic prevention levels, high regional levels, or stricter operating procedures use lower thresholds to improve early prevention and control sensitivity; ordinary operating areas use higher thresholds to reduce the number of invalid closure states. The residual closure threshold is updated based on the re-verification results during subsequent operations.

[0061] In this embodiment, the anomaly reverse attribution control module uses the abnormal occurrence area and time corresponding to the disease anomaly as the starting point for reverse reasoning. It reversely invokes explicit flow relationships along the explicit flow map in the multi-media control knowledge graph of the egg farm, and converts the closed state of contamination residue into a residue transmission relationship. When the explicit flow relationship and the residue transmission relationship satisfy the time progression relationship, regional connectivity relationship, and anomaly direction consistency, they are linked together to form a candidate latent transmission chain. The transmission chain contribution result is used to rank the candidate latent transmission chains, and candidate latent transmission chains that reach the high contribution judgment threshold are identified as high-contribution latent transmission chains.

[0062] The anomaly back-attribution prevention and control module, based on the explicit flow graph obtained from the prevention and control knowledge graph construction module and the set of pollution residue closed states obtained from the residue closed knowledge reasoning module, performs anomaly back-tracing contribution attribution on the abnormal occurrence area and time corresponding to the epidemic anomaly, generates candidate implicit transmission chains, calculates the transmission chain contribution results, and generates a zonal biocontrol strategy accordingly. Graph retrieval, link sorting, logistic function mapping, and supervised learning model training can be implemented using existing graph computing or machine learning processing methods, which are used as supporting technologies. The key point of this implementation is that it incorporates explicit flow relationships and pollution residue closed states into the anomaly back-tracing window, so that the transmission chain contribution results simultaneously reflect the actual flow path and pollution residue closed relationships, avoiding the generation of prevention and control strategies based solely on single-point anomaly records, single personnel and vehicle records, or single epidemic prevention and control records.

[0063] When anomaly feedback records show that an abnormal disease occurred in the target chicken house or target production area, the anomaly reverse attribution and control module establishes an anomaly backtracking window based on the area and time of the anomaly. The end time of the anomaly backtracking window is the time of the anomaly, and the start time is determined based on the anomaly type label, the farm's epidemic prevention level, the retention period of circulation records, and the duration of historical review transmission chains. The anomaly backtracking window for ordinary anomaly feedback is 24 hours to 7 days, while the anomaly backtracking window for high epidemic prevention level areas or areas with continuous anomaly feedback is 3 to 14 days. If the farm's epidemic prevention operation procedures specify the traceability period for the corresponding disease type, this traceability period is used as the priority basis. The length of the anomaly backtracking window is not less than 1 hour to avoid the denominator being 0 when normalizing the total time span later. The role of the anomaly backtracking window is to limit the search scope of candidate hidden transmission chains and prevent historical circulation records lacking temporal correlation from entering the contribution attribution process.

[0064] Within the anomaly backtracking window, the anomaly reverse attribution and control module retrieves explicit flow relationships from the anomaly occurrence area forward in the explicit flow diagram and converts the contamination residue closure state into a residue transmission relationship for reverse reasoning. Explicit flow relationships are used to characterize the actual movement relationship of the flowing object between contact areas or the explicit contact relationship within the same contact area. Residue transmission relationships are formed by the contamination residue closure state and are used to characterize the indirect residual impact relationship of the residue source fragment on the affected subsequent fragment. During the conversion, the residue source fragment is used as the starting point of the residue transmission relationship, and the affected subsequent fragment is used as the ending point of the residue transmission relationship. The residue bearing area, residue effective time, residue transmission direction, contamination residue closure strength, and disinfection reduction confidence are used as edge attributes of the residue transmission relationship.

[0065] Candidate latent propagation chains are formed by connecting explicit flow relationships and residual transmission relationships. When connecting them, they simultaneously satisfy temporal progression, regional connectivity, and anomaly direction consistency. Temporal progression means that the occurrence time of the previous link edge in the chain is earlier than the occurrence time of the next link edge, and the link endpoint can reach the anomaly occurrence area before the anomaly occurrence time, or there is a pollution residue closure state with the anomaly occurrence area. Regional connectivity is determined by spatial carrying records and is used to exclude paths that cannot be reached in the spatial topology. Anomaly direction consistency is used to determine whether the link direction can point from the candidate source area to the anomaly occurrence area through the flow path or residual transmission relationship. If the occurrence time of the link edge is later than the anomaly occurrence time, or the regional flow direction cannot reach the anomaly occurrence area, the link edge will not participate in the candidate latent propagation chain connection.

[0066] To prevent the infinite expansion of candidate latent propagation chains, the anomaly reverse attribution control module sets a boundary for candidate chain generation. The starting point of a candidate latent propagation chain is located within the anomaly backtracking window, and the ending point is the anomaly occurrence area or a flow medium state segment with a valid pollution residue closed state related to the anomaly occurrence area. The number of link edges traversed by the candidate latent propagation chain is between 1 and 20. The total time span of the candidate latent propagation chain does not exceed the length of the anomaly backtracking window. When repeated contact areas appear in the candidate latent propagation chain, if the repeated path does not form a new pollution residue closed state, the repeated paths are merged and the link edge with more complete propagation chain contribution basis is retained. If no link combination that meets the above conditions is found, the result of no valid candidate latent propagation chain is output, and the anomaly occurrence area is added to the manual review list.

[0067] After obtaining candidate latent transmission chains, the anomaly reverse attribution prevention and control module calculates the transmission chain contribution result for each candidate latent transmission chain. The transmission chain contribution result is used to characterize the degree of contribution of the candidate latent transmission chain to the epidemic anomaly, and the value ranges from 0 to 1. Before calculation, it is first determined whether the candidate latent transmission chain contains link edges. If the candidate latent transmission chain does not contain link edges, the candidate latent transmission chain is not valid, the transmission chain contribution result is not calculated, and it enters the manual review list. If the candidate latent transmission chain contains at least one link edge, it enters the link feature extraction process.

[0068] For each established candidate latent transmission chain, the anomaly reverse attribution prevention and control module extracts the following results for each link edge in the candidate latent transmission chain: link residual support, explicit flow continuity, anomaly direction consistency, and disinfection reduction insufficiency. These four types of link features are all converted to a dimensionless scale of 0 to 1 before being used in subsequent processing. The link residual support result characterizes whether the link edge has a basis for residual transmission or residual risk support; explicit flow continuity characterizes the degree of continuity of the link edge in terms of object identification, regional connectivity, and temporal sequence; anomaly direction consistency characterizes whether the temporal and spatial flow directions of the link edge can point to the area where the anomaly occurred; and disinfection reduction insufficiency result characterizes the degree to which the corresponding epidemic prevention and control measures for the link edge are insufficient in reducing residuals.

[0069] When the link edge represents a residual transmission relationship, the link residual support result uses the residual closure strength corresponding to the residual closure state in the residual closure knowledge reasoning module. When the link edge represents an explicit flow relationship, the link residual support result is formed based on the contact bearing strength of the flow object corresponding to the explicit flow relationship, the epidemic prevention level of the contact area, and the credibility of the explicit flow relationship. The contact bearing strength uses the normalized results for the flow object and contact surface in the residual closure knowledge reasoning module. The epidemic prevention level of the contact area is determined by the spatial bearing record and the site epidemic prevention operation procedure. The production area, manure cleaning area, and egg collection area correspond to 0.70 to 1.00, the passage and material temporary storage area correspond to 0.40 to 0.69, and the disinfection area that has completed effective disinfection verification corresponds to 0.10 to 0.39. The credibility of the explicit flow relationship is determined by the recording credibility of the flow medium state fragments at both ends of the edge, the integrity of regional connectivity, and the validity of the time sequence. All the above results are normalized to 0 to 1 to form the link residual support result of the explicit flow relationship.

[0070] Explicit continuity is determined by object identifier continuity, regional connectivity, and temporal continuity. Object identifier continuity is used to determine whether consecutive segments belong to the same person, vehicle, egg tray batch, shared equipment, feed delivery equipment, feed delivery batch, or tool; a value of 0.80 to 1.00 is used when they belong to the same object identifier, 0.50 to 0.79 is used when they belong to the same batch but lack individual identifiers, and 0 is used when object continuity cannot be confirmed. Regional connectivity is determined by spatial carrying records; a value of 0.80 to 1 is used when there is a direct connection between two contact areas. The time interval is 0.00. When a transition through an intermediate contact area is required, the interval is 0.40 to 0.79. When the space carrying capacity record indicates unreachability, the interval is 0. Temporal continuity is determined by the time interval between the departure time of the previous segment and the arrival time of the next segment. All time intervals use the unified time base in the state segment generation module and are converted to minutes. The time interval is 0.80 to 1.00 when it does not exceed the continuous judgment interval of the corresponding contact area; 0.30 to 0.79 when it exceeds the continuous judgment interval but does not exceed the allowable interval of the anomaly backtracking window; and 0 when the time sequence is reversed or exceeds the allowable interval of the anomaly backtracking window. The allowable interval of the anomaly backtracking window is determined based on the continuous judgment interval of the corresponding contact area, the length of the anomaly backtracking window, and the site operation procedures. It is used to determine whether adjacent link edges still have traceable temporal continuity during anomaly backtracking. As a preferred implementation, the allowable interval for the anomaly backtracking window is 2 to 6 times the continuous judgment interval of the corresponding contact area, and does not exceed one-tenth of the anomaly backtracking window length. When the site operation procedures specify a maximum allowable interruption time for personnel inspection, egg tray turnover, manure removal equipment operation, or vehicle parking, this maximum allowable interruption time shall be used as the priority. The allowable interval for the anomaly backtracking window shall not be less than the continuous judgment interval, and the anomaly backtracking window length shall not be less than 1 hour.

[0071] The consistency of anomaly direction is determined by the relationship between the occurrence time of the link edge, the direction of regional flow, and the location of the anomaly occurrence area. When the occurrence time of the link edge is earlier than the time of the anomaly occurrence, and its spatial flow direction can reach the anomaly occurrence area through the spatial carrying record, the consistency of anomaly direction is 0.70 to 1.00. When the link edge needs to indirectly point to the anomaly occurrence area through the closed state of the contamination residue, the consistency of anomaly direction is 0.30 to 0.69. When the occurrence time of the link edge is later than the time of the anomaly occurrence, or the spatial flow direction cannot reach the anomaly occurrence area, the consistency of anomaly direction is 0. This result is used to exclude time inversion and spatially unreachable paths to avoid reverse attribution of candidate hidden propagation chains.

[0072] The disinfection reduction deficiency result is used to characterize the impact of insufficient epidemic prevention and control measures at the link edge on the contribution of the transmission chain. For residual transmission relationships, the disinfection reduction deficiency result is obtained by subtracting the disinfection reduction confidence level in the corresponding contaminated residual closed state from 1. Since the disinfection reduction confidence level ranges from 0 to 1, the disinfection reduction deficiency result is also a dimensionless result of 0 to 1. For explicit transmission relationships, the disinfection reduction deficiency result is determined by the epidemic prevention and control records that cover the same contact area and are valid in time before and after the occurrence of the link edge. When there are no epidemic prevention and control records that cover the corresponding contact area and are valid in time, the disinfection reduction deficiency result is 0.70 to 1.00. When there are epidemic prevention and control records that cover the area, are valid in time, are fully executed, and have passed the review, the disinfection reduction deficiency result is 0 to 0.30. When there are epidemic prevention and control records but lack dosage, duration, coverage, or execution confirmation information, the disinfection reduction deficiency result is 0.31 to 0.69, and the disinfection link that needs to be reviewed is output.

[0073] After feature extraction for each link edge is completed, the anomaly reverse attribution prevention and control module weights the link residual support result, explicit flow continuity, anomaly direction consistency, and insufficient disinfection reduction result with the corresponding model weights to obtain the comprehensive link contribution of each link edge. All four positive model weights are non-negative and are pre-normalized to a weight sum of 1. This setting is used to avoid the over-amplification of a single feature, so that the contribution judgment of the candidate hidden transmission chain is simultaneously constrained by residual support, actual flow, anomaly direction, and insufficient epidemic prevention and control measures.

[0074] After obtaining the total contribution of each link edge in the candidate latent propagation chain, the anomaly reverse attribution prevention module averages the total contribution of all links to obtain the basic contribution of the candidate latent propagation chain. The reason for using averaging is that different candidate latent propagation chains contain different numbers of links. If the contributions of each link are directly accumulated, candidate latent propagation chains with a large number of links are easily mechanically amplified. Averaging can unify candidate latent propagation chains of different lengths to a comparable scale. When averaging, the number of links in the candidate latent propagation chain is used as the denominator. The number of links has been limited to not less than 1 in the pre-judgment, so there is no problem of the denominator being 0.

[0075] Subsequently, the anomaly reverse attribution prevention module determines the link length penalty result of the candidate latent propagation chain. The link length penalty result is formed by normalizing the number of regions traversed by the candidate latent propagation chain, the number of link edges, and the total time span, with a value range of 0 to 1. The more regions the candidate latent propagation chain traverses, the more link edges it has, and the closer the total time span is to the upper limit of the anomaly backtracking window, the higher the link length penalty result. The number of regions is normalized to the number of reachable contact areas within the anomaly backtracking window, and this normalization upper limit is not less than 1; the number of link edges is normalized to 20; the total time span is normalized to the length of the anomaly backtracking window, and the anomaly backtracking window length is not less than 1 hour. The link length penalty result is used to reduce the probability that candidate latent propagation chains with long spans, excessive cross-regional jumps, and sparse evidence are judged as high-contribution chains.

[0076] The anomaly reverse attribution prevention module inputs the basic contribution of candidate latent propagation chains, the link length penalty result, and the model bias term into the contribution mapping process to obtain the propagation chain contribution result. Specifically, the impact of the link length penalty result is first subtracted from the basic contribution, and then the model bias term is combined to form the attribution score of the candidate latent propagation chain. Subsequently, the attribution score is converted to the 0 to 1 range using the existing logistic function mapping method to obtain the propagation chain contribution result. The logistic function mapping method is a conventional numerical mapping method in existing machine learning, used to convert the real-number attribution score into a comparable result of 0 to 1. The focus of this implementation method is on the input source and combination logic of the attribution score, that is, the link residual support result, explicit flow continuity, anomaly direction consistency, disinfection and reduction insufficient result, and link length penalty result are all included in the candidate latent propagation chain contribution attribution.

[0077] The model bias terms and weights were obtained through training on historical epidemic anomaly records, manually verified transmission chain records, epidemic prevention and control verification records, and negative operation records. When the number of historical samples was no less than 100, a supervised learning model with a logistic function mapping output layer was used to train the model bias terms and weights, with confirmed transmission chains as positive samples and accidental transfer chains during negative operation as negative samples. When the number of historical samples was less than 100, the initial weights were determined based on the epidemic prevention level of the site, regional importance, and expert-calibrated samples. The weights corresponding to the link residual support results, explicit transfer continuity, anomaly direction consistency, and insufficient disinfection reduction results were all non-negative weights, and the sum of the normalized weights of the four positive weights was 1. The weights corresponding to the link length penalty results were taken from 0.10 to 0.50. The initial weights were updated in subsequent operations based on the results of manual verification, epidemic prevention and control verification, and anomaly feedback.

[0078] The anomaly reverse attribution prevention and control module sorts candidate latent transmission chains based on their contribution to the transmission chain. The higher the contribution of the transmission chain, the greater its contribution to the epidemic anomaly. The threshold for determining a high-contribution latent transmission chain is determined by historical negative operating samples and reviewed abnormal samples. As a preferred implementation method, when the number of historical samples is not less than 50, the 95th percentile of the transmission chain contribution result in the historical negative operating samples is used as the initial threshold, and it is corrected in conjunction with the reviewed abnormal samples. When the number of historical samples is less than 50, the initial threshold is determined according to the epidemic prevention level of the site. The threshold is 0.60 for sites with high epidemic prevention level or sites with abnormal feedback records in the past 30 days, 0.70 for sites with ordinary epidemic prevention level, and 0.80 for sites with low epidemic prevention level and no abnormal feedback records in the past 30 days. Candidate latent transmission chains that reach the high contribution threshold are determined as high-contribution latent transmission chains. Candidate latent transmission chains that are below the high contribution threshold and lack sufficient evidence of key timestamps, regional connectivity, or surface compatibility are determined as low-confidence candidate chains and enter the manual review list.

[0079] After completing the transmission chain contribution ranking, the anomaly reverse attribution prevention and control module identifies high-risk transfer media, high-risk carrying areas, and disinfection links that need to be reviewed. High-risk transfer media are determined by the transfer objects that appear in high-contribution hidden transmission chains with a frequency that meets the frequency judgment criteria and whose transmission chain contribution results reach the high contribution judgment threshold. These include personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools. The frequency judgment criteria are determined by the frequency distribution of each transfer object in the high-contribution hidden transmission chain within the anomaly backtracking window. As a preferred implementation, when the number of historical samples is no less than 50, the 75th percentile of the frequency of high-risk circulating media in the historically reviewed abnormal samples is taken as the frequency determination criterion; when the number of historical samples is less than 50, the circulating object appears in no less than two high-contribution latent transmission chains, and the contribution results of the corresponding transmission chains all reach the high contribution determination threshold, thus satisfying the frequency determination criterion. The high-risk carrying area is determined by the contact area in the high-contribution latent transmission chain that continuously forms a closed state of contamination residue. The disinfection link that needs to be reviewed is determined by the link edge where the disinfection reduction result reaches the review trigger range, the disinfection record lacks dosage, duration, coverage, or execution confirmation information; as a preferred implementation, the review is triggered when the disinfection reduction result reaches 0.31 to 1.00, where 0.31 to 0.69 corresponds to incomplete information review, and 0.70 to 1.00 corresponds to review without effective epidemic prevention and control records.

[0080] The zoned biosecurity strategy includes restricting the cross-building movement of media corresponding to high-contribution latent transmission chains, performing re-disinfection on high-risk areas, setting up temporary blocking paths for contact areas that have continuously formed closed states of contamination residue, and performing targeted tracing of objects in candidate latent transmission chains with consistent abnormal directions. Based on the re-disinfection results, the reliability of records for corresponding media status segments, the reliability of disinfection reduction for corresponding residue transmission relationships, and the closed state of contamination residue are updated. Restricting cross-building movement includes adjusting personnel inspection routes, restricting the cross-zone turnover of egg trays, suspending the cross-building use of shared equipment, or restricting vehicles from entering the production area. Re-disinfection includes supplementing the disinfection duration, re-disinfecting coverage area, checking the status of disinfection equipment, and performing surface wiping re-disinfection. Temporary blocking paths include adjusting the order of channel use, setting up temporary isolation routes, or suspending the turnover of corresponding materials. Targeted tracing includes tracking the contact areas and surfaces that the same object passed through within the abnormal tracing window. All of the above control actions correspond to the transmission chain contribution results and the closed state of contamination residue, and are not based solely on a single abnormal report or a single ledger record.

[0081] If there are no abnormal disease records, the abnormal reverse attribution prevention and control module will not perform abnormal backtracking contribution attribution, and will retain the pollution residue closure state set generated by the residual closure knowledge reasoning module as the basis for early warning; if the candidate hidden transmission chain lacks key timestamps, spatial topology cannot be connected, or surface compatibility evidence is insufficient, the candidate hidden transmission chain will be marked as a low-confidence candidate chain and entered into the manual review list; if the manual review results confirm that there are errors in the corresponding circulation object, contact area, or epidemic prevention and control records, the review results will be fed back to the corresponding circulation medium state fragment, explicit circulation relationship, residual transmission relationship, and pollution residue closure state for the calculation of the next round of transmission chain contribution results.

[0082] Through the abnormal reverse attribution prevention and control module, the system links the set of closed states of contamination residues with the abnormal occurrence area and time of the abnormality corresponding to the epidemic abnormality. It identifies high-contribution hidden transmission chains from candidate hidden transmission chains and further outputs high-risk circulation media, high-risk carrying areas, disinfection links that need to be reviewed, and regional biocontrol strategies.

[0083] Example 2 like Figure 2 As shown, the present invention also discloses a method for biological control of diseases in laying hens, comprising the following steps: S1 receives multi-media control and circulation data from the egg farm, generates circulation medium status segments according to the continuous dwelling process of the circulation object in the contact area, and segments are cut when the circulation object changes, the contact area changes, or the time interruption exceeds the continuous judgment interval determined based on the working rhythm of the contact area, forming a set of circulation medium status segments. S2, taking the set of state fragments of the flowing medium as the graph generation object, forms an explicit flow rotor graph according to the flow order of the fragments and the regional connectivity relationship, and forms a contamination residue subgraph based on the time interval between consecutive fragments in the same region, surface compatibility results and disinfection interval. The explicit flow rotor graph and the contamination residue subgraph constitute a multi-media prevention and control knowledge graph for egg farms. S3: Extract the preceding and following segments that are established in time sequence from the knowledge graph of multi-media prevention and control in the egg farm. Combine the surface compatibility results, time decay results, disinfection reduction confidence and environmental maintenance results to determine the pollution residue closure relationship and obtain the pollution residue closure state set. S4. Establish an anomaly backtracking window in the abnormal occurrence area and time corresponding to the epidemic anomaly. Generate candidate latent transmission chains along the explicit flow diagram and the set of closed states of contamination residue. Calculate the transmission chain contribution results according to the link time continuity, anomaly direction consistency and contamination residue closed state, and generate a regional biocontrol strategy.

[0084] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A biological control system for diseases in laying hens, characterized in that, include: The status segment generation module receives multi-media control and circulation data from the egg farm, generates circulation media status segments according to the continuous dwell process of the circulation object in the contact area, and segments the data when the circulation object changes, the contact area changes, or the time interruption exceeds the continuous judgment interval determined based on the working rhythm of the contact area, forming a collection of circulation media status segments. The knowledge graph construction module for prevention and control generates fragment entities from the set of state fragments of the flowing medium and regional entities from the contact area. It forms an explicit flow subgraph according to the fragment flow order and regional connectivity. It forms a contamination residue subgraph based on the time interval between the preceding and following fragments that are established in the same contact area, the surface compatibility result, and the disinfection interval. It also configures a rule knowledge layer for closed reasoning of contamination residue to form a multi-media prevention and control knowledge graph for egg farms. The residual closure knowledge reasoning module extracts the preceding and following fragments that are established in the same contact area from the multi-media prevention and control knowledge graph of the egg farm, calls the rule knowledge layer to reason about the pollution residue closure relationship, and obtains the pollution residue closure state set; The abnormal reverse attribution prevention and control module takes the abnormal occurrence area and time corresponding to the epidemic abnormality as the starting point for reverse reasoning. It generates candidate hidden transmission chains based on the explicit flow rotor diagram and the set of closed states of pollution residue. It calculates the contribution results of the transmission chain according to the link time continuity, the consistency of the abnormal direction and the closed state of pollution residue, and generates a regional biocontrol strategy.

2. The biological control system for laying hen diseases according to claim 1, characterized in that, The multi-media prevention and control data for layer hen farms includes operation records, disease prevention and control records, spatial carrying capacity records, environmental auxiliary records, and anomaly feedback records. Operation records include the movement of personnel, vehicles, egg trays, manure, shared equipment, feed delivery equipment, feed delivery batches, and tools within the layer hen farm. Disease prevention and control records include the disinfection process and verification results. Spatial carrying capacity records include the connectivity between chicken houses, passageways, egg collection areas, manure cleaning areas, disinfection areas, and material storage areas. Environmental auxiliary records include temperature, humidity, ventilation status, and ground condition within the contact areas. Anomaly feedback records include the area and time of occurrence of any disease-related anomalies.

3. The biological control system for laying hen diseases according to claim 2, characterized in that, The specific operations of the status fragment generation module for preprocessing multi-media prevention and control data transfer in layer chicken farms include: performing unified time base conversion, unified identification of transfer objects, contact area coding mapping, and record credibility marking on the multi-media prevention and control data transfer in layer chicken farms; determining the continuous judgment interval based on the historical continuous operation interval, shift operation rules, and equipment operating cycle of similar transfer objects in the corresponding contact area; and determining the continuous judgment interval based on the operation procedures of the corresponding contact area when the historical continuous operation interval is insufficient, and reducing the record credibility of the corresponding record.

4. The biological control system for laying hen diseases according to claim 3, characterized in that, The specific operations of the state fragment generation module in generating a set of state fragments for the circulating medium include: merging records with the same contact area, continuous contact time, and no time interruption exceeding the continuous judgment interval into a single state fragment for the same circulating object; when a change in the circulating object or contact area occurs between the current record and an adjacent subsequent record, or when the time interruption between them exceeds the continuous judgment interval, ending the current state fragment for the circulating medium and generating a new state fragment for the circulating medium; configuring the circulating object type, circulating object identifier, contact area, entry time, exit time, previous contact area, next contact area, contact surface type, corresponding epidemic prevention and control record, environmental auxiliary record, and record credibility for the state fragment for the circulating medium; when the circulating object identifier is missing but the contact area and contact time are complete, generating a regional-level low-credibility fragment and configuring a regional-level low-credibility fragment marker for the regional-level low-credibility fragment, forming a set of state fragments for the circulating medium that serves as the source of the fragment entities.

5. The biological control system for laying hen diseases according to claim 1, characterized in that, The specific operations for constructing an explicit flow graph in the knowledge graph construction module for epidemic prevention and control include: using the state segments of the flowing medium as segment entities and the contact areas as region entities, establishing explicit flow relationships according to the flow object identifier, entry time, exit time, previous contact area, and next contact area corresponding to the segment entity; explicit flow relationships include object flow relationships formed by the same flow object across contact areas, and regional contact relationships formed by different flow objects in the same contact area due to overlapping contact time periods, identical contact surfaces, or continuous operation sequences; the explicit flow graph representing the flow sequence of segments and the connectivity of regions is formed by the segment entities, region entities, and explicit flow relationships.

6. The biological control system for laying hen diseases according to claim 5, characterized in that, The specific operations for constructing the pollution residue subgraph in the knowledge graph construction module for epidemic prevention and control include: selecting the preceding and subsequent flow media state segments that are in chronological order from the same contact area, and generating candidate relationships for pollution residues based on the time interval between them, surface compatibility results, and disinfection intervals; the surface compatibility results are determined by the contact surface compatibility mapping table, which is established based on the epidemic prevention operation procedures of the egg-laying chicken farm and the results of historical pollution verification; when the preceding flow media state segment, contact area, subsequent flow media state segment, and chronological order corresponding to the candidate relationship for pollution residues are all valid, and the surface compatibility result is greater than 0, the candidate relationship for pollution residues is written into the pollution residue subgraph.

7. A biological control system for laying hen diseases according to claim 6, characterized in that, The rule knowledge layer includes load-bearing strength rules, surface compatibility rules, time decay rules, disinfection reduction rules, environmental maintenance rules, and closure establishment rules. The load-bearing strength rules are used to determine the contact load-bearing strength based on the type of the object being transported, the type of the contact surface, the residence time, and the reliability of the record of the preceding flow medium state segment. Surface compatibility rules are used to determine surface compatibility results based on the contact surface compatibility mapping table; The time decay rule is used to determine the time decay result based on the departure time of the preceding flowing medium state segment, the entry time of the subsequent flowing medium state segment, and the epidemic prevention operation procedures of the corresponding contact area. The disinfection reduction rules are used to determine the credibility of disinfection reduction based on the coverage, execution sequence, execution completeness, record credibility, and review results of epidemic prevention and control records. The environmental maintenance rule is used to determine the environmental maintenance result based on the temperature, humidity, ventilation status and ground status of the environmental auxiliary records within the corresponding time range; the closure establishment rule is used to determine the closure establishment condition based on the fact that the preceding and subsequent flowing medium state segments are located in the same contact area, the time sequence is established, the area identifier and timestamp are valid, and combined with the valid object identifier status or the area-level low-confidence segment mark.

8. A biological control system for laying hen diseases according to claim 7, characterized in that, The specific operations of the residual closure knowledge reasoning module to generate a set of residual contamination closure states include: reasoning the residual contamination closure strength based on contact bearing strength, surface compatibility results, time decay results, disinfection reduction confidence, environmental maintenance results, and closure establishment conditions; when the residual contamination closure strength reaches the residual closure threshold, a residual contamination closure state is generated, which includes the residual source fragment, residual bearing area, residual effective time, affected subsequent fragments, residual transmission direction, and corresponding disinfection reduction confidence; the residual closure threshold is determined by historical negative operating samples and re-examined abnormal samples, and is determined based on the site's epidemic prevention level, area level, and operating procedures when historical samples are insufficient.

9. A biological control system for laying hen diseases according to claim 8, characterized in that, The specific operations of the anomaly reverse attribution control module in generating candidate latent transmission chains, calculating the contribution results of transmission chains, and generating regional biocontrol strategies include: establishing an anomaly backtracking window based on the anomaly occurrence area and time corresponding to the anomaly; inferring the explicit flow relationship along the explicit flow diagram within the anomaly backtracking window, and converting the closed state of contamination residue into a residue transmission relationship for reverse inference; connecting explicit flow relationships that satisfy time progression, regional connectivity, and anomaly direction consistency with residue transmission relationships to form candidate latent transmission chains; for each candidate latent transmission chain, determining the transmission chain contribution result based on the link time continuity, contamination residue closed state, anomaly direction consistency, the disinfection reduction insufficiency result determined by the disinfection reduction confidence level, link length, number of cross-regions, and total time span; and applying the transmission chain contribution result to the candidate latent transmission chain. The transmission chains are sorted, and candidate latent transmission chains whose contribution results reach the high contribution judgment threshold are identified as high contribution latent transmission chains. Based on the high contribution latent transmission chains, high-risk transfer media, high-risk carrying areas, and disinfection links that need to be reviewed are identified. The transfer media corresponding to the high contribution latent transmission chains are restricted from cross-building, high-risk carrying areas are reviewed and disinfected, temporary blocking paths are set for contact areas that continuously form a closed state of contamination residue, and targeted tracing is performed on transfer objects with consistent abnormal directions. The record credibility of the corresponding transfer media state segment, the disinfection reduction credibility of the corresponding residue transmission relationship, and the closed state of contamination residue are updated according to the review results. Among them, the high contribution judgment threshold is determined by historical abnormal disease records, manually reviewed transmission chain records, epidemic prevention and control review records, and negative operation records. When historical samples are insufficient, it is determined according to the epidemic prevention level of the site.

10. A method for biological control of diseases in laying hens, employing a biological control system for diseases in laying hens as described in any one of claims 1-9, characterized in that, Includes the following steps: S1 receives multi-media control and circulation data from the egg farm, generates circulation medium status segments according to the continuous dwelling process of the circulation object in the contact area, and segments are cut when the circulation object changes, the contact area changes, or the time interruption exceeds the continuous judgment interval determined based on the working rhythm of the contact area, forming a set of circulation medium status segments. S2 generates fragment entities from the set of state fragments of the flowing medium and generates regional entities from the contact area, constructing an explicit flow rotor diagram, a pollution residue sub-diagram, and a rule knowledge layer for closed-loop reasoning of pollution residue, forming a multi-media prevention and control knowledge graph for egg farms; S3: Extract the preceding and following segments that are in the same contact area from the knowledge graph of multi-media prevention and control in egg farms, and call the rule knowledge layer to reason about the closed relationship of pollution residue to obtain the set of closed states of pollution residue. S4 uses the abnormal occurrence area and time of the abnormal occurrence corresponding to the epidemic abnormality as the starting point for reverse reasoning. Based on the explicit flow rotor diagram and the set of closed states of pollution residue, candidate hidden transmission chains are generated. The contribution results of the transmission chain are calculated according to the link time continuity, the consistency of the abnormal direction and the closed state of pollution residue, and a regional biocontrol strategy is generated.

Images