Intelligent alarm method and system for breeding house

Abstract

The invention relates to the technical field of intelligent alarm, in particular to an intelligent alarm method and system for a breeding house, and the method comprises the following steps: collecting temperature, humidity and ventilation data in the breeding house through a sensor, recognizing abnormal events, building a label and a weight for each event, and obtaining an event identifier set; and connecting events in the event identifier set according to a causal relationship to form a directed connection graph, and layering each node according to an influence degree to generate an event hierarchical structure. According to the method, the directed connection graph is formed through labeling and weighting processing of the abnormal events in combination with causal relationship analysis, so that a complex association structure between the events can be analyzed on the whole, and the priority and the influence range of the abnormal events are highlighted. In the process of data updating and node scoring, the initial trigger path is dynamically adjusted, and the response capability to real-time changes is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of intelligent alarm technology, and in particular to an intelligent alarm method and system for a breeding house. Background Art

[0002] The intelligent alarm method for breeding houses is a technical solution for real-time monitoring and early warning of key factors such as temperature, humidity, ventilation, etc. in modern breeding environments. It aims to detect abnormal conditions in breeding houses in an intelligent way and issue alarm signals in a timely manner.

[0003] However, in the prior art, the processing of untriggered event nodes is relatively simple, and the implicit correlation between them and the current triggering nodes is not included in the analysis. This omission may increase the safety risks of the breeding environment. Therefore, improvements are needed. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent alarm method and system for a breeding house.

[0005] In order to achieve the above object, the present invention adopts the following technical scheme, an intelligent alarm method for a breeding house, comprising the following steps: The temperature, humidity and ventilation data in the breeding house are collected through sensors, abnormal events are identified, and labels and weights are established for each event to obtain an event identification set; the events in the event identification set are connected according to causal relationships to form a directed connection graph, and each node is layered according to the degree of influence to generate an event hierarchy; Extract nodes from the event hierarchy, compare and score the data of each node, find out the emergency path, and obtain the initial trigger path; then update and re-score the event nodes in the initial trigger path in real time to generate an updated trigger path; Calculate the total credibility according to the updated trigger path, compare it with the preset threshold, and if it exceeds the threshold, activate the alarm, and perform time detection at the same time. If the alarm is repeatedly activated within the set time, delay the alarm again to obtain the final alarm signal; After receiving the final alarm signal, the event nodes that are not directly triggered but affected are analyzed, the correlation with the triggered events is calculated, and it is determined whether to dynamically activate the nodes to obtain the auxiliary event activation analysis.

[0006] Preferably, the steps of obtaining the event identification set are: By installing sensors in the breeding house, we can collect data on ambient temperature, humidity, and ventilation in real time to obtain an environmental monitoring data set; Based on the environmental monitoring data set, the anomaly index is calculated to identify abnormal points in the data. The calculation formula is: ; in, is the abnormal index of the ith data point, is the fth measurement value of the ith data point, is the average value of the fth measurement, and F is the number of types of measurement values; Based on the anomaly index, labels and weights are assigned to the identified abnormal events to form an event identification set.

[0007] Preferably, the steps of obtaining the event hierarchy are: By analyzing the data in the event identification set, identifying the causal relationship between events, constructing a directed connection graph, forming an event association network, and obtaining a directed association network; Based on the directed association network, an event impact degree analysis is performed, and each event node is hierarchical according to its influence to form an event hierarchy.

[0008] Preferably, the steps of acquiring the initial trigger path are: Traversing the event hierarchy, extracting node data one by one, verifying and recording the attributes of each node, and forming a node attribute data set; Based on the node attribute data set, each node is scored according to the data difference with other nodes. The scoring calculation formula is: ; in, Representative Node The comprehensive score of Is a node With Node The correlation coefficient of Is a node The priority of is the base of natural logarithms, is the total number of nodes; Based on the comprehensive score, the urgency of the node is determined according to the comprehensive score, and the node path that needs to be processed first is identified and marked as the initial trigger path.

[0009] Preferably, the step of acquiring the update trigger path is: Collecting the environment and operation data of each event node in the initial trigger path in real time to obtain a node data set updated in real time; Based on the real-time updated node data set, re-evaluate the urgency and influence of each event node, and re-score to obtain a re-scored node set; From the re-scored node set, the path with the highest score is analyzed and selected as the update trigger path to generate the update trigger path.

[0010] Preferably, the steps of obtaining the final alarm signal are: Based on the data of each event node in the update trigger path, the total credibility is calculated using the following formula: ; in, Representative Node activity, Is a node The alert level, is the median of all node alert levels, is the total number of nodes, is the total reliability; Based on the calculated total reliability, it is compared with the preset threshold. If the total reliability exceeds the threshold, an alarm is triggered and the alarm frequency is monitored. If it is triggered multiple times within ten seconds, the alarm is triggered again according to the set delay to form a final alarm signal.

[0011] Preferably, the acquisition steps of the auxiliary event activation analysis are: After receiving the final alarm signal, all event node data that are not directly triggered but affected are collected to obtain a potential impact node data set; Based on the potential impact node data set, the correlation between each node and the triggered event is calculated using the following formula: ; in, and Represents nodes and triggered nodes In Properties The value of is the total number of attributes, is the relevance of each node; Based on the correlation, the activity potential of the nodes is evaluated, and it is determined which nodes need to be dynamically activated to deal with potential risks, generating auxiliary event activation analysis.

[0012] The present invention provides an intelligent alarm system, comprising: The data collection and event identification module collects temperature, humidity and ventilation data of the breeding house through multiple sensors, monitors the collected data in real time, and identifies abnormal events when the data deviates from the normal range. It assigns a unique label and weight to each abnormal event and generates an event identification set. The event association and path analysis module connects events according to causal relationships based on the event identification set, constructs a directed connection graph, hierarchically processes the nodes in the directed connection graph according to the impact of the event, extracts key nodes, compares node data, scores and determines the emergency path, and obtains the initial trigger path; The alarm trigger and response module performs real-time data updates and re-scoring of event nodes in the initial trigger path, generates an updated trigger path, calculates the total credibility and compares it with the set threshold. If the threshold is exceeded, the alarm is activated, the final alarm signal is obtained, and based on the final alarm signal, the event nodes that are not directly triggered but affected are analyzed and the correlation is calculated.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are: The present invention forms a directed connection graph by labeling and weighting abnormal events, combined with causal analysis, which helps to analyze the complex correlation structure between events as a whole and highlight the priority and impact range of abnormal events. In the process of data update and node scoring, the initial trigger path is dynamically adjusted to enhance the ability to respond to real-time changes. The calculation of total confidence combined with dynamic time detection makes the triggering of alarm signals more accurate. The analysis of untriggered but potentially affected nodes further improves the coverage capability and ensures the early assessment and processing of hidden risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] See also Figure 1 The present invention provides a technical solution, an intelligent alarm method for a breeding house, comprising the following steps: The temperature, humidity and ventilation data in the breeding house are collected through sensors, abnormal events are identified, and labels and weights are established for each event to obtain a set of event identifiers. Events in the set of event identifiers are connected according to causal relationships to form a directed connection graph, and each node is layered according to the degree of influence to generate an event hierarchy. Extract nodes from the event hierarchy, compare and score the data of each node, find out the urgency path, and obtain the initial trigger path; then update and re-score the event nodes in the initial trigger path in real time to generate an updated trigger path; The total credibility is calculated based on the updated trigger path and compared with the preset threshold. If the threshold is exceeded, the alarm is activated and time detection is performed at the same time. If the alarm is repeatedly activated within the set time, the alarm is delayed again to obtain the final alarm signal; After receiving the final alarm signal, the event nodes that are not directly triggered but affected are analyzed, the correlation with the triggered events is calculated, and it is determined whether to dynamically activate the nodes to obtain the auxiliary event activation analysis.

[0017] The steps to obtain the event identifier set are: By installing sensors in the breeding house, we can collect data on ambient temperature, humidity, and ventilation in real time to obtain an environmental monitoring data set; Based on the environmental monitoring data set, the anomaly index is calculated to identify abnormal points in the data. The calculation formula is: ; in, is the abnormal index of the ith data point, is the fth measurement value of the ith data point, is the average value of the fth measurement, and F is the number of types of measurement values; Based on the anomaly index, labels and weights are assigned to the identified abnormal events to form an event identification set.

[0018] Specifically, by installing sensors in the breeding house, the environmental monitoring data set contains data on ambient temperature, humidity and ventilation. The real-time collection of data depends on sensors distributed in different locations. These sensors send data to the central database at regular intervals. The composition of the environmental monitoring data set includes temperature readings, humidity levels and ventilation conditions collected from each sensor. Each data point has a timestamp and location label so that the specific measurement environment can be tracked in subsequent analysis. After the data set is collected, it is stored in the central server for subsequent analysis and processing to obtain the environmental monitoring data set.

[0019] The benefit of the formula is that it can evaluate the degree of abnormality of the data by the sum of the squares of the deviations of each data point from the mean. Such a calculation not only takes into account the abnormality of a single variable, but also provides a comprehensive abnormality indicator by summarizing the deviations of all related variables, making anomaly detection more comprehensive and accurate. Calculation process: An environmental monitoring data set, in which only two types of monitoring are temperature and humidity (F=2). For a specific data point, the temperature and humidity readings are and The average temperature and humidity are and , enter the formula to calculate: ; The result shows that the abnormal index of this data point is 25, which means that the temperature and humidity of this point are significantly different from the average value. If the value is greater than a set threshold, such as 50, then the point may require further inspection and analysis.

[0020] Based on the abnormal index The process of assigning labels and weights to each abnormal event involves calculating The value is compared with the preset standard. According to the comparison result, abnormal events above a certain threshold are marked as high priority. The labels of these events usually include "urgent", "attention", etc., and the weight is based on The value is assigned, the higher The larger the value, the higher the weight. This ensures that more serious events are given priority in subsequent processing, resulting in a set of event identifiers arranged according to severity.

[0021] The steps to obtain the event hierarchy are: By analyzing the data in the event identification set, the causal relationship between events is identified, a directed connection graph is constructed, and a correlation network of events is formed to obtain a directed correlation network; Based on the directed association network, the impact degree of events is analyzed, and each event node is hierarchical according to its influence to form an event hierarchy.

[0022] Specifically, by analyzing the data recorded in the event identification set, identifying the direct and indirect relationships between each event, comparing the timestamp, duration and intensity of each event in the data set, and constructing a mapping of nodes and edges, each node represents an event, and each edge represents the direct causal relationship between events. The direction and strength of the edge are determined by the probability and influence of the interaction between events, and the influence score is calculated as follows: influence score = weight + anomaly index. Through this method, the direction of the event and the strength of the relationship can be clarified, so that after integrating all related events, a directed event association network is formed, and a directed association network is obtained.

[0023] Based on the constructed directed association network, the influence of each node in the network is analyzed in detail. Based on the node connectivity, node centrality and the shortest path between nodes in the network, the influence of each event node is quantitatively evaluated. The event nodes are sorted according to their influence through numerical sorting to ensure the accuracy and operability of the hierarchical division. In the hierarchical process, each node is assigned a hierarchical index, and nodes with high influence are placed in the upper layer, while nodes with lower influence are placed in the lower layer. Such a hierarchical structure helps to identify key events and minor events and form a clear event hierarchy.

[0024] The steps to obtain the initial trigger path are: Traverse the event hierarchy, extract node data one by one, verify and record the attributes of each node, and form a node attribute data set; Based on the node attribute data set, each node is scored according to the data difference with other nodes. The scoring calculation formula is: ; in, Representative Node The comprehensive score of Is a node With Node The correlation coefficient of Is a node The priority of is the base of natural logarithms, is the total number of nodes; Based on the comprehensive score, the urgency of the node is determined according to the comprehensive score, and the node path that needs to be processed first is identified and marked as the initial trigger path.

[0025] Specifically, in the process of traversing the event hierarchy, the key attributes of each node such as node ID, directly connected nodes and node level are identified and recorded. These data are stored in a temporary database for further processing. Through this method, it is possible to ensure that the collected node data is not only complete but also accurate, thus forming a node attribute dataset.

[0026] The benefit of the formula is that it adjusts the influence of the correlation coefficient by introducing an exponential function, so that the sensitivity of the score is proportional to the correlation between the nodes; The steps for obtaining the parameters are as follows: first, analyze the interaction frequency and intensity between each node i and node j through historical data, and then convert them into correlation coefficients through standardization. The steps of obtaining parameters are as follows: weights are assigned according to the centrality of node j in the network or the previously set priority, and the total number of nodes J is the actual number of nodes obtained from statistics in the database; Calculation process: Let the correlation coefficient between node i and node j be , the priority of node j is 2, and the total number of nodes J is 10, and the calculation is: ; The result shows that the comprehensive score of node i is 6.22459, which reflects its urgency and processing priority. The higher the score, the more priority the node should be given.

[0027] Based on the obtained comprehensive score, the urgency level of each node is determined. By setting a threshold, it is determined which nodes have reached the level that requires immediate processing. The threshold is set based on actual tests and experience. Then, the high-scoring nodes are traced to the related nodes. The path formed by the high-urgency nodes is marked as the initial trigger path, which marks the beginning of risk assessment and response measures.

[0028] The steps to obtain the update trigger path are: Collect the environment and operation data of each event node in the initial trigger path in real time to obtain a node data set updated in real time; Specifically, during the execution process, the environmental and operational data of each event node in the initial trigger path are first collected in real time through the deployed multiple sensor devices, including physical parameters such as temperature, humidity, pressure, as well as operation logs and system status information, to ensure that the collected data is comprehensive and updated. The real-time performance is guaranteed by the synchronized clock to ensure that the timestamps of all data points are consistent, so that the real-time updated node data set has a high degree of consistency and reliability. These data are then stored in a central database for the next step of processing and analysis to obtain a real-time updated node data set.

[0029] Based on the node data set updated in real time, the urgency and influence of each event node are re-evaluated and re-scored to obtain a re-scored node set; From the re-scored node set, the path with the highest score is analyzed and selected as the update trigger path to generate the update trigger path.

[0030] The steps to obtain the final alarm signal are: Based on the data of each event node in the update trigger path, the total credibility is calculated using the following formula: ; in, Representative Node activity, Is a node The alert level, is the median of all node alert levels, is the total number of nodes, is the total reliability; Based on the calculated total reliability, it is compared with the preset threshold. If the total reliability exceeds the threshold, an alarm is triggered and the alarm frequency is monitored. If it is triggered multiple times within ten seconds, the alarm is triggered again according to the set delay to form a final alarm signal.

[0031] Specifically, the formula is beneficial in that by adjusting the alarm level of each node The contribution to the total credibility can more flexibly reflect the urgency of the node status; The parameter acquisition step is determined by the interaction frequency or event triggering frequency. The parameters are based on the severity score of events in the past 24 hours. is the real-time calculated median of all node alarm levels; Calculation process: It includes 3 nodes, whose activity levels are , , The alarm levels are , , , median , the total reliability is calculated as follows: ; ; ; The result shows that the total confidence is 8.433 based on the current node data and alarm level.

[0032] Based on the calculated total confidence, it is compared with the preset threshold. First, the threshold is set to 8. This threshold is set based on past historical data and safety standards. If the calculated result exceeds the threshold, an alarm is triggered, and the time frequency of the alarm is monitored, and the time and interval of each alarm are recorded. If it is triggered multiple times in a short period of time, the alarm strategy is adjusted according to the set delay parameters. This adjustment is to avoid "alarm fatigue" caused by too frequent alarms and to form and obtain the adjusted final alarm signal.

[0033] The steps to obtain auxiliary event activation analysis are: After receiving the final alarm signal, collect all event node data that are not directly triggered but affected to obtain the potential impact node data set; Based on the potential impact node data set, the correlation between each node and the triggered event is calculated using the following formula: ; in, and Represents nodes and triggered nodes In Properties The value of is the total number of attributes, is the relevance of each node; Based on the correlation, the activity potential of the nodes is evaluated, and it is decided which nodes need to be dynamically activated to deal with potential risks, generating auxiliary event activation analysis.

[0034] Specifically, after receiving the final alarm signal, the affected event nodes must first be determined. Although these nodes did not directly trigger the alarm, they may have a high risk of being potentially affected due to their geographical or functional association with the triggered nodes. In this step, all potentially affected nodes are identified and their relevant data are collected, including but not limited to the operating status of the nodes, historical event records, and interaction data with the triggered nodes. Through this data collection process, a comprehensive data set of potentially affected nodes is constructed; The benefits of the formula are: it adjusts the impact of differences between nodes through an exponential function, balances the sensitivity of node differences, and does not have an excessive impact on the results when the node differences are large, thus improving the stability and accuracy of the evaluation; Calculation process: Consider two nodes i and j, each node has K attribute values, and the difference in attribute values ​​is calculated by Calculate, sum all attribute differences and transform through exponential function, and you can get the correlation between node i and node j For example, if nodes i and j differ greatly in some key attributes, then The sum of will be larger, making Become smaller, becomes larger, indicating that the correlation between the two nodes is low; if , , and the total number of attributes K=1, substitute the specific value into the formula to calculate: ; The result is 0.73, which shows that nodes i and j have a high correlation; this result shows that: When the value is close to 1, it means that the nodes are strongly related and their state or behavior changes are likely to affect each other. If it is smaller, the correlation between the two nodes is weak and the mutual influence is small; Based on the obtained node correlation data, it is necessary to quantitatively evaluate the activity potential of each node. The higher the correlation, the higher the activity potential. At the same time, considering the node's correlation and the role and influence of the node in the network, use machine learning models or statistical analysis to predict the probability of a node becoming active due to changes in network events, including but not limited to the node's communication frequency, data traffic and past activity records. Through comprehensive analysis of these data, it can be determined whether the node needs to be dynamically activated to combat potential risks.

[0035] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent alarm method for a breeding house, characterized in that: The following steps are involved: The temperature, humidity and ventilation data in the breeding house are collected through sensors, abnormal events are identified, and labels and weights are established for each event to obtain an event identification set; the events in the event identification set are connected according to causal relationships to form a directed connection graph, and each node is layered according to the degree of influence to generate an event hierarchy; Extract nodes from the event hierarchy, compare and score the data of each node, find out the emergency path, and obtain the initial trigger path; then update and re-score the event nodes in the initial trigger path in real time to generate an updated trigger path; Calculate the total credibility according to the updated trigger path, compare it with the preset threshold, and if it exceeds the threshold, activate the alarm, and perform time detection at the same time. If the alarm is repeatedly activated within the set time, delay the alarm again to obtain the final alarm signal; After receiving the final alarm signal, the event nodes that are not directly triggered but affected are analyzed, the correlation with the triggered events is calculated, and it is determined whether to dynamically activate the nodes to obtain the auxiliary event activation analysis.

2. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The steps for obtaining the event identification set are: By installing sensors in the breeding house, we can collect data on ambient temperature, humidity, and ventilation in real time to obtain an environmental monitoring data set; Based on the environmental monitoring data set, the anomaly index is calculated to identify abnormal points in the data. The calculation formula is: ； in, is the abnormal index of the ith data point, is the fth measurement value of the ith data point, is the average value of the fth measurement, and F is the number of types of measurement values; Based on the anomaly index, labels and weights are assigned to the identified abnormal events to form an event identification set.

3. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The steps for obtaining the event hierarchy are: By analyzing the data in the event identification set, identifying the causal relationship between events, constructing a directed connection graph, forming an event association network, and obtaining a directed association network; Based on the directed association network, an event impact degree analysis is performed, and each event node is hierarchical according to its influence to form an event hierarchy.

4. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The steps for obtaining the initial trigger path are: Traversing the event hierarchy, extracting node data one by one, verifying and recording the attributes of each node, and forming a node attribute data set; Based on the node attribute data set, each node is scored according to the data difference with other nodes. The scoring calculation formula is: ； in, Representative Node The comprehensive score of Is a node With Node The correlation coefficient of Is a node The priority of is the base of natural logarithms, is the total number of nodes; Based on the comprehensive score, the urgency of the node is determined according to the comprehensive score, and the node path that needs to be processed first is identified and marked as the initial trigger path.

5. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The steps for obtaining the update trigger path are: Collecting the environment and operation data of each event node in the initial trigger path in real time to obtain a node data set updated in real time; Based on the real-time updated node data set, re-evaluate the urgency and influence of each event node, and re-score to obtain a re-scored node set; From the re-scored node set, the path with the highest score is analyzed and selected as the update trigger path to generate the update trigger path.

6. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The steps for obtaining the final alarm signal are: Based on the data of each event node in the update trigger path, the total credibility is calculated using the following formula: ； in, Representative Node activity, Is a node The alert level, is the median of all node alert levels, is the total number of nodes, is the total reliability; Based on the calculated total reliability, it is compared with the preset threshold. If the total reliability exceeds the threshold, an alarm is triggered and the alarm frequency is monitored. If it is triggered multiple times within ten seconds, the alarm is triggered again according to the set delay to form a final alarm signal.

7. The intelligent alarm method for breeding houses according to claim 1, characterized in that: The acquisition steps of the auxiliary event activation analysis are: After receiving the final alarm signal, all event node data that are not directly triggered but affected are collected to obtain a potential impact node data set; Based on the potential impact node data set, the correlation between each node and the triggered event is calculated using the following formula: ； in, and Represents nodes and triggered nodes In Properties The value of is the total number of attributes, is the relevance of each node; Based on the correlation, the activity potential of the nodes is evaluated, and it is determined which nodes need to be dynamically activated to deal with potential risks, generating auxiliary event activation analysis.

8. An intelligent alarm system for the intelligent alarm method for a breeding house according to any one of claims 1 to 7, characterized in that: include: The data collection and event identification module collects temperature, humidity and ventilation data of the breeding house through multiple sensors, monitors the collected data in real time, and identifies abnormal events when the data deviates from the normal range. It assigns a unique label and weight to each abnormal event and generates an event identification set. The event association and path analysis module connects events according to causal relationships based on the event identification set, constructs a directed connection graph, hierarchically processes the nodes in the directed connection graph according to the impact of the event, extracts key nodes, compares node data, scores and determines the emergency path, and obtains the initial trigger path; The alarm trigger and response module performs real-time data updates and re-scoring of event nodes in the initial trigger path, generates an updated trigger path, calculates the total credibility and compares it with the set threshold. If the threshold is exceeded, the alarm is activated, the final alarm signal is obtained, and based on the final alarm signal, the event nodes that are not directly triggered but affected are analyzed and the correlation is calculated.

Images