CN116909339B

CN116909339B - Intelligent household safety early warning method and system based on artificial intelligence

Info

Publication number: CN116909339B
Application number: CN202311186251.9A
Authority: CN
Inventors: 葛云生; 王坤; 蔡斌
Original assignee: Li Zhuang Information Technology Suzhou Co ltd
Current assignee: Li Zhuang Information Technology Suzhou Co ltd
Priority date: 2023-09-14
Filing date: 2023-09-14
Publication date: 2024-01-09
Anticipated expiration: 2043-09-14
Also published as: CN116909339A

Abstract

The invention provides an intelligent home safety early warning method and system based on artificial intelligence, which belong to the field of intelligent home, and the method comprises the following steps: when the first monitoring period is met, activating an environment monitoring task of the intelligent home; performing sensitive source analysis on the intelligent home to generate an associated object group; traversing the associated object group, collecting the home running state, and sending the home running state to an index prediction processing node to obtain index prediction characteristics; traversing the associated object group, collecting home distribution positions, sending the home distribution positions to an influence range analysis node, and obtaining a range calibration result; partitioning the intelligent home according to the acquired content, and acquiring a partitioning result and an index early warning threshold; and when the index early warning is met, carrying out danger early warning on the partition result. The technical problems of poor accuracy and effectiveness of household environment early warning in the prior art are solved, fine management and dynamic management and control of household space are realized, and the technical effects of improving the effectiveness and accuracy of household early warning are achieved.

Description

Intelligent household safety early warning method and system based on artificial intelligence

Technical Field

The invention relates to the field of intelligent home, in particular to an intelligent home safety early warning method and system based on artificial intelligence.

Background

With the development of intelligent home, home safety early warning technology becomes a hotspot and difficulty. The existing household safety early warning method mainly depends on the traditional smoke alarm, security system and the like, the threshold value is usually set according to experience, the threshold value is fixed, accurate monitoring and dynamic management of the household environment cannot be realized, the situation of false alarm or no alarm exists, various threats in the household space are difficult to accurately identify, and the early warning accuracy and effectiveness are poor.

Disclosure of Invention

The application aims to solve the technical problems of poor accuracy and effectiveness of household environment early warning in the prior art by providing the intelligent household safety early warning method and system based on artificial intelligence.

In view of the above problems, the application provides an intelligent home safety early warning method and system based on artificial intelligence.

In a first aspect of the disclosure, an intelligent home security early warning method based on artificial intelligence is provided, and the method includes: activating an environmental monitoring task of the smart home when the first monitoring period is met, wherein the environmental monitoring task comprises one or more of temperature monitoring and humidity monitoring; performing sensitive source analysis on the intelligent home according to the environment monitoring task to generate an associated object group; traversing the associated object group, acquiring an object running state through an intelligent home sensor, and sending the object running state to an index prediction processing node deployed in a cloud data center to acquire index prediction characteristics; traversing the associated object group, acquiring object distribution positions through an intelligent home sensor, and transmitting the object distribution positions to an influence range analysis node deployed in a cloud data center by combining index prediction features to acquire a range calibration result; partitioning the intelligent home according to the index prediction characteristics and the range calibration results to obtain partitioning results and index early warning thresholds; and when the index early warning is met, carrying out danger early warning on the partition result.

In another aspect of the disclosure, an artificial intelligence based smart home security early warning system is provided, the system comprising: the environment monitoring task module is used for activating an environment monitoring task of the smart home when the first monitoring period is met, wherein the environment monitoring task comprises one or more of temperature monitoring and humidity monitoring; the sensitive source analysis module is used for carrying out sensitive source analysis on the intelligent home according to the environment monitoring task and generating an associated object group; the index prediction feature module is used for traversing the associated object group, acquiring the running state of the object through the intelligent home sensor, sending the running state to an index prediction processing node deployed in the cloud data center, and acquiring index prediction features; the range calibration result module is used for traversing the associated object group, acquiring object distribution positions through the intelligent home sensor, and transmitting the object distribution positions to an influence range analysis node deployed in the cloud data center by combining the index prediction characteristics to acquire a range calibration result; the indoor household partition module is used for partitioning the intelligent household according to the index prediction characteristics and the range calibration results to obtain partition results and index early warning thresholds; and the danger early warning module is used for carrying out danger early warning on the partition result when the index early warning is met.

One or more technical solutions provided in the present application have at least the following technical effects or advantages:

according to the technical scheme, the technical problems of poor accuracy and effectiveness of household environment early warning in the prior art are solved, the refined management and dynamic management and control of the household space are realized, and the technical effects of improving the effectiveness and accuracy of early warning are achieved.

The foregoing description is only an overview of the technical solutions of the present application, and may be implemented according to the content of the specification in order to make the technical means of the present application more clearly understood, and in order to make the above-mentioned and other objects, features and advantages of the present application more clearly understood, the following detailed description of the present application will be given.

Drawings

Fig. 1 is a schematic flow chart of a possible smart home security early warning method based on artificial intelligence according to an embodiment of the present application;

fig. 2 is a schematic flow chart of a possible process of obtaining an index prediction feature in an intelligent home safety pre-warning method based on artificial intelligence according to an embodiment of the present application;

fig. 3 is a schematic flow chart of a possible process of obtaining a range calibration result in an intelligent home security early warning method based on artificial intelligence according to an embodiment of the present application;

fig. 4 is a schematic diagram of a possible structure of an intelligent home security early warning system based on artificial intelligence according to an embodiment of the present application.

Reference numerals illustrate: the system comprises an environment monitoring task module 11, a sensitive source analysis module 12, an index prediction feature module 13, a range calibration result module 14, an indoor household partition module 15 and a danger early warning module 16.

Detailed Description

The technical scheme provided by the application has the following overall thought:

the embodiment of the application provides an intelligent home safety early warning method based on artificial intelligence, which is used for monitoring key indexes such as temperature, humidity and the like in a home environment in real time by activating an environment monitoring task of an intelligent home, identifying and correlating various sensitive sources and influence objects thereof in the home environment according to monitoring results, and acquiring the running state and position distribution characteristics of each object. And combining the index prediction processing node and the influence range analysis node of cloud deployment to realize fine division of the home space and index threshold setting. When the index exceeds the threshold value, relevant space or objects are early-warned pertinently, so that the accurate identification and positioning of household security threats are realized, and the accuracy and effectiveness of early warning are improved.

Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.

Example 1

As shown in fig. 1, the embodiment of the application provides an intelligent home security early warning method based on artificial intelligence, which is applied to an early warning client, the early warning client is in communication connection with a cloud data center, the cloud data center is in communication connection with an intelligent home sensor,

specifically, the early warning client is used for receiving and prompting early warning information from the cloud data center. The early warning client can be a mobile terminal, such as a smart phone or a tablet computer, or a fixed terminal, such as an intelligent display screen, an intelligent sound box and the like. The early warning client side is in communication connection with the cloud data center through a wireless network or a wired network and is used for data transmission and interaction. The cloud data center is used for collecting monitoring data sent by the intelligent home sensor, analyzing, calculating and judging the data, and pushing the early warning information to the early warning client when the judging result reaches the early warning condition. The cloud data center can be in the form of a server cluster, and comprises an application server, a database server, an analysis server and the like, so that the cloud data center has both computing capacity and storage capacity. The cloud data center is in communication connection with the intelligent home sensor through a network and is used for receiving monitoring data and issuing control instructions. The intelligent home sensor is used for collecting various physical quantity parameters in the home, such as temperature, humidity, smoke concentration and the like, and sending monitoring data to the cloud data center. The intelligent home sensor can be in communication connection with the cloud data center by adopting a special communication network or a home local area network. The intelligent home network can be built by the intelligent home sensors, so that the home environment is monitored in all directions.

The intelligent household safety early warning method comprises the following steps:

activating an environment monitoring task of the smart home when a first monitoring period is met, wherein the environment monitoring task comprises one or more of temperature monitoring and humidity monitoring;

specifically, the environmental monitoring task refers to monitoring environmental parameters, such as temperature, humidity, etc., inside a home by using an intelligent home sensor. The first monitoring period may be a time period or an event period. Wherein, the time period refers to activating the environment monitoring task according to a fixed time period, such as 08:00 activation every day; the event period is activated according to the occurrence event, such as when the household access control system detects a valid door opening signal.

When the first monitoring period is met, the cloud data center sends an environment monitoring task activation command to the intelligent home sensor, and the environment monitoring task can only comprise one of temperature monitoring or humidity monitoring, and can also comprise both temperature monitoring and humidity monitoring. The cloud data center can select and activate a single type or multiple types of environment monitoring tasks according to actual needs of the home environment. After receiving the environmental monitoring task activation command, the intelligent home sensor starts corresponding sensing elements such as a temperature sensor and a humidity sensor, samples environmental parameters in home in real time, and sends monitoring data to a cloud data center.

By setting the first monitoring period, the environment monitoring task of the smart home is activated under the condition that the first monitoring period is met, so that the environment monitoring task is periodically activated and executed, and basic data support is provided for subsequent environment parameter analysis and early warning.

Performing sensitive source analysis on the intelligent home according to the environment monitoring task to generate an associated object group;

specifically, the sensitive source analysis refers to analyzing which devices or objects in the home have a larger influence or higher sensitivity on the monitoring index by using the monitoring data acquired by the environmental monitoring task. For example, the environmental monitoring task includes temperature monitoring, and the cloud data center analyzes objects with great influence on the home temperature, such as heating equipment and cooking equipment in the home. The environment monitoring task comprises humidity monitoring, and the cloud data center can analyze objects with larger influence on the household humidity, such as a humidifier, a dehumidifier, a washing machine and the like in the household.

The associated object group refers to a collection of in-home devices or objects with larger influence or higher sensitivity to monitoring environmental parameters. In the sensitive source analysis process, the cloud data center extracts characteristic parameters in the monitoring data, such as the working state, the working strength, the working time and the like of the objects, judges the sensitive influence of the objects represented by the characteristic parameters on the monitoring index through correlation analysis, a machine learning algorithm and other methods, sorts and screens the sensitive influence evaluation results of the objects, and selects the more sensitive objects to form a related object group.

By carrying out sensitive source analysis on the smart home, selecting an object with high sensitive influence on the monitoring index to construct an associated object group, and providing an object basis for subsequent early warning judgment and control.

Traversing the associated object group, acquiring an object running state through an intelligent home sensor, and sending the object running state to an index prediction processing node deployed in a cloud data center to acquire index prediction characteristics;

specifically, the cloud data center sends an acquisition command of an object running state to the intelligent home sensor, wherein the objects are objects in the associated object group, and the command contains identification information of each object in the associated object group. After receiving the command, the intelligent home sensor collects the running state of each object, such as the information of the working state, the working mode, the working strength and the like of the object, and sends the collection result to the cloud data center.

And after receiving the object running state acquisition result, the cloud data center inputs the object running state acquisition result to the deployed index prediction processing node. The index prediction processing node may be a machine learning model, such as a linear regression model, a neural network model, etc., which has learned the mapping relationship between the object operation state and the environmental parameter change through training. And inputting an object running state acquisition result, and obtaining the predicted variation or probability of the environmental parameter, namely the index prediction characteristic of the environmental monitoring task, by the index prediction processing node according to the acquisition result. For example, the input air conditioner operating state is "high Wen Moshi", and the index prediction processing node can predict the characteristic that the home temperature rises by 3 ℃; the input gas stove working state is 3 stoves working, and the index prediction processing node can predict the characteristic that the household humidity is increased by 10%.

The running states of all objects in the associated object group are collected, the running states are input into trained index prediction processing nodes for analysis and judgment, the index prediction characteristics of the environment parameters corresponding to the running states of the current objects are obtained, basic reference basis is provided for subsequent influence range analysis and early warning judgment, accurate prediction of the parameter change of the home environment is achieved, and the accuracy of intelligent home safety early warning is improved.

Traversing the associated object group, acquiring object distribution positions through the intelligent home sensor, and transmitting the object distribution positions to an influence range analysis node deployed in the cloud data center by combining the index prediction characteristics to acquire a range calibration result;

specifically, the cloud data center sends an object position information acquisition command to the intelligent home sensor, wherein the objects are objects in the associated object group, and after receiving the command, the intelligent home sensor acquires position information of each object, such as a spatial position coordinate, and sends an acquisition result to the cloud data center. The cloud data center inputs the received object position information and the obtained index prediction characteristics to an influence range analysis node deployed in the cloud data center. The influence range analysis node can adopt a machine learning algorithm, such as a linear regression model, a neural network model and the like, and the node has learned the mapping relation of the object position, the running state and the influence range of the environmental parameters.

After the collected object position information and index prediction characteristics are input into an influence range analysis node, the influence range analysis node obtains the influence range of the environmental parameters in space according to the input content, namely, the influence range calibration result of the environmental monitoring task. For example, the input air conditioner is located in a living room (5,3,1), the index prediction feature is' raise by 3 ℃, and the influence range analysis node can predict the influence range calibration result that the temperature of the living room is raised by 1-3 ℃ within the range of 3 meters.

The position information of each object in the associated object group is collected, the obtained index prediction characteristics are combined, the influence range analysis node is input for judgment, the space range of the current running state of the object influencing the household environment is obtained, the basis is provided for the subsequent early warning area division, the accurate positioning and calibration of the influence range of the household environment are realized, and the accuracy of intelligent household safety early warning is improved.

Partitioning the intelligent home according to the index prediction characteristics and the range calibration results to obtain partitioning results and index early warning thresholds;

specifically, the cloud data center judges the space range of the influence of the object on the environmental parameters according to the range calibration result output by the influence range analysis node, and then divides the indoor space of the home according to the space range. And setting a higher early warning threshold value in a space outside the range calibration result, and setting a lower early warning threshold value in a space inside the range calibration result according to the index prediction characteristics. For example, if the influence range analysis node outputs that the influence range of the air conditioner is within 3 meters, the air conditioner is divided into one block within a radius of 3 meters by taking the air conditioner position as the center, and a space beyond the radius of 3 meters is divided into another block. According to the air conditioner index prediction characteristic of raising the temperature by 3 ℃, the block within the air conditioner influence range is set with an early warning threshold value of 28 ℃, and the block outside the influence range is set with an early warning threshold value of 30 ℃. Because the influence ranges of different objects may overlap or influence each other, the cloud data center calculates influence factors of a plurality of objects at the same time when dividing blocks and setting an early warning threshold. For example, the influence of the gas stove may exist in the influence range of the air conditioner, and the early warning threshold value is not only dependent on the index prediction characteristics of the air conditioner, but also needs to be combined to determine the final early warning threshold value of the block.

The household indoor space is distinguished and divided according to the obtained index prediction characteristics and the obtained range calibration results, so that a partition result and an index early warning threshold value of each block are obtained, and when the monitoring data exceeds the early warning threshold value in the block, early warning information is pushed into the block, so that the effectiveness of early warning is improved.

And when the index early warning is met, carrying out danger early warning on the partition result.

Specifically, the cloud data center receives the environmental monitoring data uploaded by the intelligent home sensor in real time, and compares the environmental monitoring data with the obtained partition result and the early warning threshold value of the corresponding block. If the monitoring data exceeds the early warning threshold value of a certain block, the block is in a dangerous state, and early warning is needed.

According to the judging result, the cloud data center immediately locks a block needing dangerous early warning, corresponding early warning information is generated, the block position, monitoring data of environmental parameters, recommended measures and the like are included, and then the cloud data center pushes the early warning information to an early warning client, such as a mobile phone terminal of an resident and the like. After the early warning client receives the early warning information, the early warning client informs the resident of the early warning information in the form of an acousto-optic prompt, for example, the mobile phone pops up an early warning interface, sends out an alarm sound and displays the early warning information. The cloud data center pushes early warning information to the early warning client side and simultaneously sends an automatic control instruction to the home system, so that dangerous states are intelligently eliminated. For example, when the temperature of the dangerous area exceeds the early warning threshold, the cloud data center can directly send a control instruction for reducing the temperature setting to the air conditioning system.

By comparing the monitoring environment data with the early warning threshold values in each partition in real time, when the monitoring data exceeds the threshold values, the corresponding dangerous blocks are early warned, real-time monitoring and early warning control of the household environment are realized, and the technical effects of improving the effectiveness and accuracy of the household environment early warning are achieved.

Further, the embodiment of the application further includes:

setting a first retrieval attribute, a second retrieval attribute, a third retrieval attribute and a fourth retrieval attribute, wherein the first retrieval attribute represents a monitoring distance, the second retrieval attribute represents the number of objects, the third retrieval attribute represents a monitoring characteristic value, and the fourth retrieval attribute represents an application scene characteristic of the objects;

according to the first retrieval attribute, the second retrieval attribute, the third retrieval attribute and the fourth retrieval attribute, performing data mining on a first object to be analyzed based on the environment monitoring task to obtain M data retrieval results;

screening N data retrieval results with fixed second retrieval attributes from the M data retrieval results, and carrying out serialization adjustment on the third retrieval attributes according to the first retrieval attributes from small to large to generate a first adjustment result;

Screening the L data retrieval results with the fixed first retrieval attribute from the M data retrieval results, and carrying out serialization adjustment on the third retrieval attribute according to the second retrieval attribute from small to large to generate a second adjustment result;

performing sensitive source analysis on the first object to be analyzed according to the first adjustment result and the second adjustment result to generate a sensitivity analysis result;

and when the sensitivity analysis result meets a sensitivity threshold, storing the first object to be analyzed and the fourth retrieval attribute in a correlated way, and adding the first object to be analyzed and the fourth retrieval attribute into the correlated object group.

Specifically, the first retrieval attribute refers to a spatial distance between the object and the environment monitoring task, namely, a characterization monitoring distance; the second retrieval attribute refers to the number of objects and the corresponding working state, namely the number of characterization objects; the third retrieval attribute is an influence value of the object on the monitoring index during working, namely a characteristic monitoring characteristic value; the fourth retrieval attribute refers to a usage scene of the object, namely, a characteristic of an application scene of the object. The first search attribute is set according to a monitoring range of an environmental monitoring task, for example, the temperature monitoring task can be set to three value ranges of 0-5 meters, 5-10 meters and more than 10 meters, and the closer the monitoring distance is, the greater the influence of an object on the monitoring index is, and the higher the correlation with the task is. The second search attribute is set according to the working mode of the object, for example, four air conditioners can be set to be four values of shutdown, startup 1, startup 2 and startup 3, and the more the number of the objects is, the greater the influence on the monitoring index is, and the higher the task correlation is. And extracting the influence range of the object on the monitoring index during working according to the historical monitoring data by the third retrieval attribute, taking the intermediate value as a monitoring characteristic value, wherein the larger the monitoring characteristic value is, the larger the influence of the object on the monitoring index is, and the higher the correlation with the task is. The third retrieval attribute is set according to matching of the usage scene of the object and the environment monitoring task, for example, temperature monitoring can be set as a 'refrigeration', 'heating', 'dehumidification' scene of the air conditioner, and the more matching of the application scene of the object and the environment monitoring task is, the higher the correlation is.

According to the set four attributes, the cloud data center performs data retrieval on the first object to be analyzed in the historical monitoring data based on the activated environment monitoring task to obtain M data results meeting the attribute conditions, wherein the M data results comprise the working state, the working mode and the position information of the object, the influence value of the object on the monitoring index and the like. The first object to be analyzed refers to an object to be subjected to data retrieval, such as an air conditioner, a refrigerator, a washing machine and the like.

And screening N pieces of data with fixed object quantity from the M pieces of data retrieval results by the cloud data center. And sequencing according to the monitoring distance from small to large, and carrying out serialization adjustment according to the monitoring characteristic value to generate a first adjustment result. For example, assuming that N pieces of data are in the M data retrieval results, the number of objects of the gas stove is "2 on", the cloud data center will screen out the N pieces of data first, then the cloud data center will sort from small to large according to the monitoring distances of the N pieces of data, if the monitoring distances of the N pieces of data are respectively 0-5 meters, 5-10 meters and more than 10 meters, the sorted order is 0-5 meters, 5-10 meters and more than 10 meters. Then, the cloud data center carries out serialization adjustment on the monitoring characteristic values according to the monitoring distance according to the N ordered data, if the monitoring characteristic values corresponding to the N data of the gas stove are 23 ℃, 30 ℃ and 27 ℃, the adjusted sequence is as follows: monitoring distance: 0-5 m, monitoring characteristic value: 30 ℃; monitoring distance: 5-10 m, monitoring characteristic values: 27 ℃; monitoring distance: more than 10 meters, monitoring characteristic values: 23 ℃; thereby obtaining a first adjustment result. Similarly to the first adjustment result, the cloud data center screens out L pieces of data with fixed monitoring distance from M pieces of data retrieval results, orders the L pieces of data according to the number of objects from small to large, and then carries out serialization adjustment on the monitoring characteristic values to generate a second adjustment result.

After the first adjustment result and the second adjustment result are obtained, the cloud data center comprehensively analyzes the first adjustment result and the second adjustment result, and considers the influence degree of the objects reflected by the two adjustment results on the monitoring index under different monitoring distances and different object numbers as a sensitivity analysis result. When the sensitivity analysis result meets a preset sensitivity threshold, the cloud data center needs to store the first object to be analyzed in association with the application scene characteristics of the object and add the first object to be analyzed to the associated object group. For example, setting the temperature sensitivity influence degree to be higher than 2 ℃ and the sensitivity threshold to be higher, and calculating the sensitivity source analysis result of the state of 2 "start-up" of the air conditioner to meet the sensitivity threshold. At this time, the cloud data center performs association storage on the air conditioner and the application scene characteristics of the air conditioner, such as "refrigeration", where the storage content may include: object name-air conditioner; object application scene feature-refrigeration; sensitive source analysis results-highly sensitive, then storage is added to the associated object group.

By setting the attribute for data retrieval and analysis and carrying out multi-angle sensitive source analysis on the object according to the attribute, the sensitive influence of the object on the monitoring index is accurately judged, and the associated object group highly related to the current environment monitoring task is constructed, so that early warning judgment errors are avoided, and the accuracy of early warning of the home environment is improved.

Further, the embodiment of the application further includes:

based on the first adjustment result, carrying out correlation analysis on the first retrieval attribute and the third retrieval attribute to obtain a first correlation analysis result;

based on the second adjustment result, performing correlation analysis on the second retrieval attribute and the third retrieval attribute to obtain a second correlation analysis result;

when the first correlation analysis result is a negative correlation and the second correlation analysis result is a positive correlation, setting the first object to be analyzed as a quasi-sensitive source;

when the first object to be analyzed is set as the quasi-sensitive source, sensitivity grading is carried out on the first correlation analysis result, and a first sensitivity grading result is obtained;

carrying out sensitivity classification on the second correlation analysis result to obtain a second sensitivity classification result;

and carrying out hierarchical fusion on the first sensitivity grading result and the second sensitivity grading result to generate the sensitivity analysis result.

Specifically, the cloud data center calculates pearson correlation coefficients of a first search attribute (monitoring distance) and a third search attribute (monitoring feature value) based on a first adjustment result of a first analysis object, judges the correlation type and degree between the first search attribute and the third search attribute, and obtains a first correlation analysis result. And then, based on the second adjustment result, calculating the pearson correlation coefficient of the second retrieval attribute (the number of objects) and the third retrieval attribute (the monitoring characteristic value), and obtaining a second correlation analysis result.

If the first correlation analysis result is negative correlation, namely the Pearson correlation coefficient is between-1 and-0.2, the monitoring characteristic value is reduced along with the increase of the monitoring distance; and the second correlation analysis result is positive correlation, namely the Pearson correlation coefficient is between 0.2 and 1, which shows that the first object to be analyzed is judged to be a quasi-sensitive source as the number of objects increases and the monitoring characteristic value increases. And then, the cloud data center carries out sensitivity division according to the correlation coefficient of the first correlation analysis result and the size of the correlation coefficient in the correlation analysis result, wherein the higher the correlation coefficient is, the higher the sensitivity level is. Sensitivity can be divided into: the method comprises the steps of extremely high sensitivity-extremely strong correlation (0.8-1), high sensitivity-strong correlation (0.6-0.8), medium sensitivity-medium correlation (0.4-0.6), low sensitivity-weak correlation (0.2-0.4) and extremely low sensitivity without correlation (0-0.2), so that a first sensitivity grading result is obtained. And carrying out corresponding sensitivity classification according to the correlation coefficient of the second correlation analysis result to obtain a second sensitivity classification result. And finally, the cloud data center considers the first sensitivity grading result and the second sensitivity grading result simultaneously to acquire a sensitivity analysis result, and if the two results are strongly correlated, the final sensitivity analysis result is strongly correlated.

The sensitivity influence degree of the first object to be analyzed is judged in a mode of correlation analysis, sensitivity grading and grading fusion, a theoretical basis is provided for sensitivity judgment, meanwhile, limitation of single correlation analysis result judgment is avoided, and judgment accuracy is improved.

Further, as shown in fig. 2, the embodiment of the present application further includes:

judging whether the object running state of a first associated object of the associated object group meets the fourth retrieval attribute of the first associated object;

if yes, inputting the object running state into the index prediction processing node to acquire the index prediction characteristics of the environment monitoring task;

and if the first association object does not meet the initial threshold value, monitoring the first association object according to the index of the environment monitoring task.

Specifically, the cloud data center needs to determine an operation state of a first associated object in the associated object group, determine whether an index prediction feature of the environmental monitoring task needs to be predicted by using the index prediction processing node, or continue to monitor the first associated object according to an initial threshold value. Judging whether the first associated object meets the corresponding sensitive application scene or not, and monitoring the pertinence of the area when the first associated object meets the corresponding sensitive application scene, otherwise, performing ordinary monitoring.

The cloud data center firstly judges whether the running state of the first associated object meets the fourth retrieval attribute, namely the object application scene feature. For example, if the air conditioner is in an operation state and in a cooling mode, the object application scene characteristics are satisfied; and when the gas stove is in an operating state, the object application scene characteristics are met. If the running state of the first associated object meets the object application scene characteristics, inputting the running state of the first associated object into an index prediction processing node to acquire the index prediction characteristics of the environment monitoring task. The index prediction processing node may predict a change trend of each index according to the specific parameter of the first association object. If the running state of the first associated object does not meet the object application scene characteristics, the cloud data center continues to monitor the first associated object according to the index initial threshold value of the environment monitoring task.

For example, assume that the indoor temperature of one room needs to be monitored, the initial threshold value is 25 ℃. A refrigerating air conditioner with 3KW refrigerating capacity and 2KW indoor heat load is arranged in a room. And judging whether the running state of the air conditioner is the running state and is in a refrigerating mode. If the air conditioner is running and is refrigerating, the application scene characteristics of the air conditioner are met, and the running parameters of the air conditioner, such as the refrigerating capacity of 3KW and the indoor heat load of 2KW, are input into a temperature prediction model. The model can predict that if the air conditioner is continuously operated, the indoor temperature will be reduced to 22 ℃ after 30 minutes; if the air conditioner is in a closed state or is in a ventilation mode only, the characteristics of the application scene of the air conditioner are not met, and the indoor temperature is continuously monitored according to the initial threshold value of 25 ℃.

Judging the application scene of the related object through the running state of the first related object, and selecting a corresponding detection mode according to specific conditions, so as to realize more intelligent and flexible environment monitoring and realize active detection of the home environment; the target running state parameters can be utilized to predict through the index prediction processing node, so that the detection response speed is improved, the monitoring efficiency is improved, and the effectiveness of home early warning is improved.

Further, the embodiment of the application further includes:

taking the first associated object as a scene constraint condition, and collecting an index prediction processing node construction data set, wherein the index prediction processing node construction data set comprises first associated object running state record data and environment index change value identification data;

the first association object running state record data is used as input data, the environment index change value identification data is used as output supervision data to carry out forward propagation training on the BP neural network, and a first forward propagation error is obtained;

based on the first forward propagation error, combining the environmental index change value identification data and the first associated object running state record data to perform backward propagation adjustment on the BP neural network;

And repeating iteration, and acquiring the index prediction processing node when the mean square error of the BP neural network is smaller than or equal to a mean square error threshold value.

Specifically, the cloud data center firstly takes a first associated object as a scene constraint condition, collects index prediction processing nodes to construct a data set, wherein the data set comprises running state record data of the first associated object and environment index change value identification data, such as whether running, a running mode, a room temperature change value and the like, and the environment index change value identification data is calibrated by an expert. Then, the cloud data center takes the first associated object running state record data as input data of the BP neural network, takes the environmental index change value identification data as output supervision data, performs forward propagation training, namely supervision training, and acquires a first forward propagation error, such as mean square error MSE and the like. And then, according to the first forward propagation error, combining the environmental index change value identification data and the air conditioner running state record data, and performing backward propagation adjustment on the BP neural network, namely adjusting the weight and the bias of the BP neural network according to the obtained error. Finally, the cloud data center repeats the iterative training process, and when the mean square error MSE of the BP neural network is smaller than or equal to a preset error threshold value, index prediction processing nodes are obtained.

Taking an air conditioner as an example, the running state record data of the air conditioner is collected, and the running state record data of the air conditioner is that a certain air conditioner is as follows: 00-12: 00 operation refrigeration mode; 12: 00-17: 00 is closed; 17: 00-21: 00 operation refrigeration mode; the expert-calibrated room temperature change value identification data is that the room temperature rising rate is 0.5 ℃ per hour under the non-running state of the air conditioner. And (3) taking the air conditioner running state record data as BP neural network input, taking the room temperature change value identification data as supervision output, and performing forward propagation training to obtain a first training error MSE=0.8 ℃. And back propagation adjustment is carried out on the weight and bias of the BP neural network according to the error of 0.8 ℃ and the acquired data. And repeating the training process, and obtaining the room temperature index prediction processing node when the BP neural network continuously outputs the error for 3 times to be less than or equal to 0.5 ℃.

By collecting object operation data corresponding to an object application scene and combining environment index change value identification data, a specific index prediction processing node is constructed based on a BP neural network, so that an intelligent home safety early warning system can judge in advance according to a prediction result of a model, and dynamically adjust in advance, thereby realizing fine monitoring and improving early warning accuracy.

Further, the embodiment of the application further includes:

Taking the first association object as a scene constraint condition, and collecting an index prediction processing node verification data set, wherein the index prediction processing node verification data set and the index prediction processing node construction data set are different;

verifying the index prediction processing node according to the index prediction processing node verification data set, obtaining the index prediction processing node verification data set with the first forward propagation error of the index prediction processing node being greater than or equal to a propagation error threshold value, and setting the index prediction processing node verification data set as a first loss data set;

when the data volume of the first lost data set is larger than or equal to a data volume threshold value, the first lost data set is added to the index prediction processing node to construct a data set, the weight of the index prediction processing node to construct the data set is adjusted, and a first auxiliary predictor is trained based on a BP neural network, wherein the weight of the first lost data set is larger than that of other data sets;

and repeating training until the data volume of the Mth lost data set is smaller than the data volume threshold value, and performing fusion adjustment on the index prediction processing nodes according to the first auxiliary predictor until the Mth auxiliary predictor.

Specifically, the cloud data center firstly takes a first associated object as a scene constraint condition, and collects a verification data set of the index prediction processing node, wherein the data set is different from a model construction data set and is used for verifying the specific performance of the obtained index prediction processing node. And then, verifying the index prediction processing node according to the verification data set, obtaining a verification data set with the model output error being greater than or equal to the propagation error threshold value, setting the verification data set as a first loss data set, wherein the trained index prediction processing node cannot reach the preset performance, for example, the propagation error threshold value can be set to be 0.5 ℃, and when the verification result deviation is greater than 0.5, the prediction side accuracy of the representative node is low, and adding the corresponding verification data set into the first loss data set. If the data volume of the first loss data set is greater than or equal to the data volume threshold, adding the first loss data set into the model construction data set, adjusting the weight of the construction data set, training the first auxiliary predictor based on the BP neural network, wherein the weight of the first loss data set is greater than the weight of other data sets, performing training on the first loss data set, and the data volume threshold can be set to be 100. And repeating the training process, training the auxiliary predictors for the obtained M loss data sets until the data quantity of the Mth loss data set is smaller than the data quantity threshold value, and carrying out fusion adjustment on the index prediction processing node according to the obtained M auxiliary predictors to improve the processing performance of the index prediction processing node.

And (3) finding out the condition of inaccurate model prediction through verifying the index prediction processing node, and then performing important training on the part of data to obtain an auxiliary predictor and fusing the auxiliary predictor with the original model, so that the prediction accuracy of the index prediction processing node is improved, and the accuracy of home early warning is further improved.

Further, as shown in fig. 3, the embodiment of the present application further includes:

the influence range analysis node comprises a data mining module and a data statistics module;

taking the index prediction feature and the ventilation area as scene constraint features, taking the monitoring distance and the index feature value as retrieval targets, and collecting an influence range analysis data set through the data mining module;

determining, by the data statistics module, the monitored distance at which the indicator feature value is greater than or equal to an indicator initial threshold value;

and constructing the range calibration result by taking the object distribution position as a central position and the monitoring distance as a radius.

Specifically, the influence range analysis node comprises a data mining module and a data statistics module. The data mining module is used for mining and analyzing historical data, and the data statistics module is used for counting and analyzing a data set. And taking the index prediction feature and the ventilation area as scene constraint features, taking the monitoring distance and the index feature value as retrieval targets, and collecting an influence range analysis data set through a data mining module. And then, the data statistics module performs statistical analysis on the influence range analysis data set to determine the monitoring distance of which the index characteristic value is higher than or equal to the index initial threshold value. And setting a range calibration result by taking the home object as a center and taking the monitoring distance as a radius.

For example, taking air conditioning and room temperature as examples, the data mining module is used for mining historical temperature monitoring data, and key factors affecting the room temperature, such as room area, air conditioning position and the like, are found. And then, the influence of the air conditioner on the temperature change of different positions of the room is counted by the data statistics module, and the change rule of the temperature along with the distance is obtained. Such as cooling temperature of 24 ℃ and ventilation area 40And for scene characteristics, taking the monitoring points with a distance of 1-7 m from the air conditioner and the temperature of the monitoring points higher than or equal to 24 ℃ as retrieval targets, and collecting an influence range analysis data set through a data mining module. Then, the data set is analyzed for the range by the data statistics module to determine that the monitoring distance is 5m. The air conditioner is arranged in a radius of 5m by taking the position of the air conditioner as the centerAnd (5) calibrating the result for the influence range. The analysis result of the influence range shows that the influence range of the temperature under the air conditioner refrigeration is 5m radius with the air conditioner position as the center under the refrigeration mode of the air conditioner.

And excavating and counting historical data through the influence range analysis node, and determining a range calibration result of the environmental index. The method provides a basis for the follow-up classification monitoring and different early warning strategies in the influence range, realizes the fine detection, improves the pertinence of the monitoring, and ensures that the safety early warning is more accurate and efficient.

Further, the embodiment of the application further includes:

outside the range calibration result, taking the index initial threshold value as a first early warning threshold value;

and taking the index prediction characteristic and the preset early warning duration as a second early warning threshold in the range calibration result, wherein the method comprises the following steps:

when the index prediction characteristics are larger than or equal to the index prediction characteristics, safety precaution is carried out;

when the index prediction characteristic is smaller than the index prediction characteristic and is larger than or equal to the preset early warning duration, safety early warning is carried out;

determining the partition result according to the range calibration result;

and determining the index early warning threshold according to the first early warning threshold and the second early warning threshold.

Specifically, if the index to be detected exceeds the influence range of the household object outside the range calibration result, the cloud data center takes the initial threshold value of the index of the environment monitoring task as a first early warning threshold value. And in the range calibration result, taking the index prediction characteristic and the preset early warning time length as a second early warning threshold value, and combining the first early warning threshold value and the second early warning threshold value as index early warning threshold values to carry out regional monitoring early warning on the household environment.

For example, a certain range of calibration results for a certain air conditioner are: setting 24 ℃ in a refrigeration mode, wherein the influence radius is 5m, and the obtained partitioning result is as follows: zone 1: the air conditioner is used as a circle center, the radius is 2m, and the sampling frequency and the early warning duration are 1min; the early warning threshold value is more than or equal to 24 ℃; zone 2: the radius is 2-5 m, the sampling frequency and the early warning time length are 3min, and the early warning threshold value is more than or equal to 24.5 ℃; zone 3: and in the range out of 5m, the sampling frequency and the early warning time length are 5min, and the early warning threshold is an initial threshold value which is more than or equal to 28 ℃.

If the acquired temperature is more than or equal to 24 ℃, safety early warning is carried out on the index prediction characteristics, such as monitoring points in the range of the zone 1; for monitoring points in the range of the zone 2, if the acquisition temperature is more than or equal to 24.5 ℃, early warning is carried out; and (3) for monitoring points outside the range of the zone 3, if the acquisition temperature is more than or equal to 28 ℃, early warning is carried out. When the preset early warning time is less than the index prediction characteristic and is greater than or equal to the preset early warning time, carrying out safety early warning; if the monitoring points in the area 2 are detected, the continuous detection temperature is 22 ℃ in 3min, and then safety early warning is carried out.

By setting the early warning threshold according to the range calibration result and dividing the household environment according to different early warning thresholds, the classified monitoring of the environment indexes is realized, and different early warning strategies are adopted under different classifications, so that the monitoring of the household environment is finer and has pertinence. By combining the index prediction characteristics with the duration for early warning, the abnormal change of the environmental index can be found earlier, the response speed of monitoring is improved, the monitoring precision and efficiency are improved, and the technical effects of improving the accuracy and effectiveness of household early warning are achieved.

In summary, the intelligent home security early warning method based on artificial intelligence provided by the embodiment of the application has the following technical effects:

When the first monitoring period is met, activating an environment monitoring task of the intelligent home, wherein the environment monitoring task comprises one or more of temperature monitoring and humidity monitoring, and simultaneously monitoring parameters such as temperature, humidity and the like in the home environment in real time, so as to provide a data base for safety analysis; performing sensitive source analysis on the intelligent home according to the environment monitoring task, generating an associated object group, and providing an object basis for early warning; traversing the associated object group, acquiring an object running state through an intelligent home sensor, sending the object running state to an index prediction processing node deployed in a cloud data center, acquiring index prediction characteristics, and providing input data for a cloud model; traversing the associated object group, acquiring object distribution positions through an intelligent home sensor, and transmitting the object distribution positions to an influence range analysis node deployed in a cloud data center by combining index prediction features to acquire a range calibration result; partitioning the intelligent home according to the index prediction characteristics and the range calibration results, obtaining partition results and index early warning thresholds, dividing the home space into a plurality of management partitions according to the object influence range, and setting a dynamic threshold for each partition; when the index early warning is met, dangerous early warning is carried out on the partition result, comprehensive monitoring and dynamic management on the threat of the household space are achieved, and the accuracy and effectiveness of the household environment early warning are improved.

Example two

Based on the same inventive concept as the intelligent home safety pre-warning method based on artificial intelligence in the foregoing embodiments, as shown in fig. 4, an embodiment of the present application provides an intelligent home safety pre-warning system based on artificial intelligence, where the system is applied to a pre-warning client, and the pre-warning client is in communication connection with a cloud data center, and the cloud data center is in communication connection with an intelligent home sensor, and includes:

the environment monitoring task module 11 is configured to activate an environment monitoring task of the smart home when the first monitoring period is satisfied, where the environment monitoring task includes one or more of temperature monitoring and humidity monitoring;

the sensitive source analysis module 12 is used for carrying out sensitive source analysis on the smart home according to the environment monitoring task to generate an associated object group;

the index prediction feature module 13 is used for traversing the associated object group, acquiring an object running state through the intelligent home sensor, sending the object running state to an index prediction processing node deployed in the cloud data center, and acquiring index prediction features;

the range calibration result module 14 is configured to traverse the associated object group, collect object distribution positions through the smart home sensor, and send the object distribution positions to an influence range analysis node deployed in the cloud data center in combination with the index prediction features to obtain a range calibration result;

The indoor home partition module 15 is configured to partition an intelligent home according to the index prediction feature and the range calibration result, and obtain a partition result and an index early warning threshold;

and the danger early warning module 16 is used for carrying out danger early warning on the partition result when the index early warning is met.

Further, the embodiment of the application further includes:

the search attribute setting module is used for setting a first search attribute, a second search attribute, a third search attribute and a fourth search attribute, wherein the first search attribute represents a monitoring distance, the second search attribute represents the number of objects, the third search attribute represents a monitoring characteristic value, and the fourth search attribute represents an application scene characteristic of the object;

the data mining module is used for carrying out data mining on a first object to be analyzed based on the environment monitoring task according to the first retrieval attribute, the second retrieval attribute, the third retrieval attribute and the fourth retrieval attribute to obtain M data retrieval results;

the first adjustment result module is used for screening N data retrieval results with fixed second retrieval attributes from the M data retrieval results, carrying out sequential adjustment on the third retrieval attributes according to the first retrieval attributes from small to large, and generating a first adjustment result;

The second adjustment result module is used for screening the L data retrieval results with the fixed first retrieval attributes from the M data retrieval results, carrying out sequential adjustment on the third retrieval attributes according to the second retrieval attributes from small to large, and generating a second adjustment result;

the sensitivity analysis result module is used for carrying out sensitivity source analysis on the first object to be analyzed according to the first adjustment result and the second adjustment result to generate a sensitivity analysis result;

and the association storage module is used for associating and storing the first object to be analyzed and the fourth retrieval attribute when the sensitivity analysis result meets a sensitivity threshold value, and adding the first object to be analyzed and the fourth retrieval attribute into the association object group.

Further, the embodiment of the application includes:

the first correlation analysis module is used for carrying out correlation analysis on the first retrieval attribute and the third retrieval attribute based on the first adjustment result to obtain a first correlation analysis result;

the second correlation analysis module is used for carrying out correlation analysis on the second retrieval attribute and the third retrieval attribute based on the second adjustment result to obtain a second correlation analysis result;

The quasi-sensitive source module is used for setting the first object to be analyzed as a quasi-sensitive source when the first correlation analysis result is negative correlation and the second correlation analysis result is positive correlation;

the first sensitivity grading module is used for carrying out sensitivity grading on the first correlation analysis result after the first object to be analyzed is set as the quasi-sensitive source, so as to obtain a first sensitivity grading result;

the second sensitivity grading module is used for grading the sensitivity of the second correlation analysis result to obtain a second sensitivity grading result;

and the grading result fusion module is used for carrying out grading fusion on the first sensitivity grading result and the second sensitivity grading result to generate the sensitivity analysis result.

Further, the embodiment of the application further includes:

the associated object judging module is used for judging whether the object running state of the first associated object of the associated object group meets the fourth retrieval attribute of the first associated object;

the prediction characteristic acquisition module is used for activating the index prediction processing node to perform the operation state of the object if the index prediction processing node is satisfied, and acquiring the index prediction characteristic of the environment monitoring task;

And the associated object monitoring module is used for monitoring the first associated object according to the index initial threshold value of the environment monitoring task if the first associated object is not satisfied.

Further, the embodiment of the application further includes:

the data set construction module is used for taking the first associated object as a scene constraint condition, and collecting index prediction processing nodes to construct a data set, wherein the index prediction processing nodes construct the data set to comprise first associated object running state record data and environment index change value identification data;

the first forward propagation error module is used for performing forward propagation training on the BP neural network by taking the first associated object running state record data as input data and the environmental index change value identification data as output supervision data to acquire a first forward propagation error;

the back propagation adjustment module is used for performing back propagation adjustment on the BP neural network by combining the environmental index change value identification data and the first associated object running state record data based on the first forward propagation error;

and the prediction model acquisition module is used for repeatedly iterating, and acquiring the index prediction processing node when the mean square error of the BP neural network is smaller than or equal to a mean square error threshold value.

Further, the embodiment of the application further includes:

the data set verification module is used for collecting an index prediction processing node verification data set by taking the first association object as a scene constraint condition, wherein the index prediction processing node verification data set is different from an index prediction processing node construction data set;

the first loss data set module is used for verifying the index prediction processing node according to the index prediction processing node verification data set, acquiring the index prediction processing node verification data set with the first forward propagation error of the index prediction processing node being greater than or equal to a propagation error threshold value, and setting the index prediction processing node verification data set as a first loss data set;

a first auxiliary predictor module for increasing the first lost data set to the index prediction processing node construction data set when the data amount of the first lost data set is greater than or equal to a data amount threshold, and adjusting the weight of the index prediction processing node construction data set, and training a first auxiliary predictor based on a BP neural network, wherein the weight of the first lost data set is greater than the weight of other data sets;

and the fusion adjustment module is used for repeating training until the data volume of the Mth lost data set is smaller than the data volume threshold value, and carrying out fusion adjustment on the index prediction processing nodes according to the first auxiliary predictor until the Mth auxiliary predictor.

Further, the embodiment of the application further includes:

the analysis data set acquisition module is used for acquiring an influence range analysis data set by taking the index prediction feature and the ventilation area as scene constraint features and taking the monitoring distance and the index feature value as retrieval targets;

the monitoring distance determining module is used for determining the monitoring distance that the index characteristic value is greater than or equal to an index initial threshold value through the data statistics module;

and the range calibration result module is used for constructing the range calibration result by taking the object distribution position as a central position and the monitoring distance as a radius.

Further, the embodiment of the application further includes:

the first early warning threshold module is used for taking the index initial threshold value as a first early warning threshold value outside the range calibration result;

the second early warning threshold module is configured to take the index prediction feature and the preset early warning duration as a second early warning threshold in the range calibration result, and includes:

the partition result determining module is used for determining the partition result according to the range calibration result;

and the index early warning threshold module is used for determining the index early warning threshold according to the first early warning threshold and the second early warning threshold.

Any of the steps of the methods described above may be stored as computer instructions or programs in a non-limiting computer memory and may be called by a non-limiting computer processor to identify any of the methods to implement embodiments of the present application, without unnecessary limitations.

Further, the first or second element may not only represent a sequential relationship, but may also represent a particular concept, and/or may be selected individually or in whole among a plurality of elements. It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the present application and the equivalents thereof, the present application is intended to cover such modifications and variations.

Claims

1. The intelligent home safety early warning method based on artificial intelligence is characterized by being applied to an early warning client, wherein the early warning client is in communication connection with a cloud data center, and the cloud data center is in communication connection with an intelligent home sensor and comprises the following steps:

when the index early warning is met, carrying out danger early warning on the partition result;

the smart home is subjected to sensitive source analysis according to the environment monitoring task, and an associated object group is generated, which comprises the following steps:

when the sensitivity analysis result meets a sensitivity threshold, storing the first object to be analyzed and the fourth retrieval attribute in a correlated manner, and adding the first object to be analyzed and the fourth retrieval attribute into the correlated object group;

performing sensitivity source analysis on the first object to be analyzed according to the first adjustment result and the second adjustment result to generate a sensitivity analysis result, including:

performing hierarchical fusion on the first sensitivity grading result and the second sensitivity grading result to generate the sensitivity analysis result;

traversing the associated object group, acquiring an object running state through an intelligent home sensor, sending the object running state to an index prediction processing node deployed in a cloud data center, and acquiring index prediction characteristics, wherein the method comprises the following steps:

If yes, an index prediction processing node is activated to map the running state of the object, and the index prediction characteristics of the environment monitoring task are obtained;

if the first association object does not meet the initial threshold value, monitoring the first association object according to the index of the environment monitoring task;

traversing the associated object group, acquiring object distribution positions through the intelligent home sensor, transmitting the object distribution positions to an influence range analysis node deployed in the cloud data center in combination with the index prediction features, and acquiring a range calibration result, wherein the method comprises the following steps of:

the object distribution position is taken as a central position, the monitoring distance is taken as a radius, and the range calibration result is constructed;

partitioning the intelligent home according to the index prediction characteristics and the range calibration results to obtain partitioning results and index early warning thresholds, wherein the partitioning methods comprise the following steps:

determining the partition result according to the range calibration result;

2. The method of claim 1, wherein traversing the associated object group, collecting an object running state by an intelligent home sensor, sending the object running state to an index prediction processing node deployed in a cloud data center, and obtaining an index prediction feature, further comprising:

the first association object running state record data is used as data to be processed, the environmental index change value identification data is used as training supervision data to conduct forward propagation training on the BP neural network, and a first forward propagation error is obtained;

3. The method as recited in claim 2, further comprising:

4. Intelligent home safety early warning system based on artificial intelligence, its characterized in that is applied to early warning customer end, early warning customer end and high in the clouds data center communication connection, high in the clouds data center and intelligent home sensor communication connection, include:

the environment monitoring task module is used for activating an environment monitoring task of the intelligent home when a first monitoring period is met, wherein the environment monitoring task comprises one or more of temperature monitoring and humidity monitoring;

the sensitive source analysis module is used for carrying out sensitive source analysis on the intelligent home according to the environment monitoring task and generating an associated object group;

the index prediction feature module is used for traversing the associated object group, acquiring an object running state through the intelligent home sensor, sending the object running state to an index prediction processing node deployed in the cloud data center, and acquiring index prediction features;

The range calibration result module is used for traversing the associated object group, acquiring object distribution positions through the intelligent home sensor, and sending the object distribution positions to an influence range analysis node deployed in the cloud data center in combination with the index prediction characteristics to acquire a range calibration result;

the indoor household partitioning module is used for partitioning the intelligent household according to the index prediction characteristics and the range calibration results to obtain partitioning results and index early warning thresholds;

the danger early warning module is used for carrying out danger early warning on the partition result when the index early warning is met;

the association storage module is used for associating and storing the first object to be analyzed and the fourth retrieval attribute when the sensitivity analysis result meets a sensitivity threshold value, and adding the first object to be analyzed and the fourth retrieval attribute into the association object group;

the grading result fusion module is used for carrying out grading fusion on the first sensitivity grading result and the second sensitivity grading result to generate the sensitivity analysis result;

the associated object monitoring module is used for monitoring the first associated object according to the index initial threshold value of the environment monitoring task if the first associated object is not satisfied;

the range calibration result module is used for constructing the range calibration result by taking the object distribution position as a central position and the monitoring distance as a radius;