Indoor environment detection method and system

Abstract

The invention belongs to the field of environment detection, and provides an indoor environment detection method and system.The indoor environment data are collected through a plurality of sensor networks distributed at different positions, the sensors are used for detecting a plurality of environment elements, and the environment elements comprise the temperature, the humidity, the air quality and the illumination intensity; preprocessing the environmental data, wherein the preprocessing comprises data cleaning, data normalization and data synchronization; determining relevance among the environment data, and establishing an undirected graph according to the relevance; and carrying out collaborative correction, data fusion and inference and environment state evaluation on real-time indoor environment detection data according to the undirected graph. According to the scheme, indoor environment detection can be efficiently carried out.

Description

Technical Field

[0001] The present invention belongs to the field of environmental detection, and particularly relates to an indoor environmental detection method and system. Background Art

[0002] In indoor environmental detection, the wide application of sensor networks makes it possible to collect and monitor key elements of the indoor environment (such as temperature, humidity, air quality, light intensity, etc.) in real time. However, in the prior art, sensor networks often face some problems in practical applications, especially sensor drift, data loss, and measurement errors, which affect the accuracy of environmental data and the stability of the system.

[0003] With the long-term use of sensors, the measurement accuracy of sensors often changes, and this phenomenon is called sensor drift. The drift problem is particularly common in physical quantity sensors such as temperature and humidity sensors. Sensor drift may be affected by various factors, such as the aging of sensor elements, the increase of electromagnetic interference in the environment, and the circuit loss of the sensor itself. The result of drift is that there is a significant deviation between the measured value output by the sensor and the actual value, which in turn affects the overall accuracy of the environmental monitoring system. For example, when a temperature sensor continuously measures a too high or too low temperature due to drift, it may trigger incorrect environmental control measures, thereby affecting the comfort and safety of the indoor environment.

[0004] In an actual sensor network, due to unstable network transmission, sensor failures, insufficient battery power, or other unpredictable external factors, data loss problems occur from time to time. Data loss may be caused by the inability of the sensor to transmit data for a short time or the long-term stop of operation. In indoor environmental monitoring, data loss will cause the system to be unable to obtain continuous environmental data, thus affecting the accurate assessment of the environmental state. For example, during air quality monitoring, if the air quality sensor fails and key pollutant concentration data cannot be obtained for a period of time, the system cannot timely determine whether the air is at a healthy level, thereby affecting the health of users.

[0005] Existing sensor arrangements are usually based on fixed position planning and cannot dynamically adapt to changes in the indoor environment. This fixed layout may lead to large measurement errors in environmental data in some areas, especially when the sensor is affected by external environmental interference (such as direct sunlight, direct cold air blowing), the measured data may deviate significantly from the true value. In addition, the accuracy of sensors is limited by cost and performance. Although high-precision sensors can improve measurement accuracy, they are often costly in actual deployment and may be unstable in harsh environments.

[0006] To solve the above problems, the prior art usually adopts methods such as calibration and data filling for processing, but these methods rely on fixed rules and models and are difficult to cope with the complex and changeable indoor environment. Summary of the Invention

[0007] To solve the problems in the prior art, the present invention provides an indoor environment detection method, and the method includes the following steps: Collect indoor environment data through multiple sensor networks distributed at different positions. The sensors are used to detect multiple environmental factors, and the environmental factors include temperature, humidity, air quality, and light intensity; Preprocess the environmental data, and the preprocessing includes data cleaning, data normalization, and data synchronization; Determine the correlation between the environmental data, and establish an undirected graph according to the correlation; Perform collaborative correction, data fusion and inference, and environmental state evaluation on the real-time indoor environment detection data according to the undirected graph.

[0008] Further, establishing an undirected graph according to the correlation includes: Use historical data to analyze the correlation between different types of sensor data through statistical correlation analysis; According to the correlation analysis, calculate the correlation coefficient between different environmental factors, generate a correlation matrix. If the absolute value of the correlation coefficient of two environmental factors exceeds a set threshold, it is considered that there is a significant correlation between them; Construct an undirected graph according to the determined correlation. Each node in the undirected graph represents an environmental factor, and the edge represents the correlation between environmental factors; If there is a significant correlation between two environmental factors, add an undirected edge in the undirected graph to connect the corresponding nodes.

[0009] Further, the collaborative correction includes: During the real-time monitoring process, the system automatically detects whether the data of a certain sensor is abnormal or drifts based on the change trend of historical data or the deviation between the current measurement value and the data of other nodes in the undirected graph; Determine the sensor nodes associated with the abnormal or drifted data through the undirected graph, and extract the data of these associated sensors; According to the data of the associated sensors and the weights of the edges in the undirected graph, use the collaborative correction model to correct the abnormal data, and the correction formula performs weighted summation according to the contribution degree of the associated data.

[0010] Further, data fusion and inference include: During the real-time monitoring process, the system detects the data loss or sensor failure of a certain sensor, and identifies the environmental factor of the missing data and its corresponding node; According to the edges in the undirected graph, extract the data of other sensors associated with the missing data, and the weight of the edge represents the influence degree of the associated sensor on the missing data; Utilize associated sensor data to fuse multi-source data according to the edge weights in the undirected graph; Calculate the inferred values of the missing data through the data fusion model, and adjust the results according to the weights of each sensor in the undirected graph to ensure that the inferred missing data is reasonable and conforms to the current environmental state.

[0011] Furthermore, the environmental state assessment includes: During the environmental monitoring process, the system collects data from each sensor in real time, monitors the status and change trends of each node in the undirected graph, and the system also monitors the edge weights and data correlations in the undirected graph; The system detects whether the edges in the undirected graph have changed significantly, and detects whether the correlation strength between environmental elements deviates from the normal range. When the weight of a certain correlation relationship increases or decreases abnormally, it indicates that there may be abnormalities or sudden changes in the relevant environmental elements; According to the detected changes in edge weights, the system analyzes which environmental elements have abnormal correlations and outputs them. On the other hand, the present invention also provides an indoor environmental detection system, which includes the following modules: A prediction module for predicting environmental parameters using a first machine learning model, where the environmental parameters include temperature, humidity, and duct pressure; An acquisition module for collecting indoor environmental data through a plurality of sensor networks distributed at different positions, where the sensors are used to detect a plurality of environmental elements, and the environmental elements include temperature, humidity, air quality, and light intensity; A preprocessing module for preprocessing the environmental data, where the preprocessing includes data cleaning, data normalization, and data synchronization; An association module for determining the correlation between the environmental data and establishing an undirected graph according to the correlation; A processing module for performing collaborative correction, data fusion and inference, and environmental state assessment on the real-time indoor environmental detection data according to the undirected graph.

[0012] By constructing a sensor data association model based on an undirected graph, the present invention realizes dynamic collaborative correction, data fusion, and missing data inference of various environmental data in the sensor network, effectively solving problems such as sensor drift, data missing, and measurement errors in the prior art, and having the following beneficial effects: Through the correlation between sensor nodes in the undirected graph, the present invention can dynamically monitor the abnormalities and drift conditions of sensor data, and automatically extract the data of relevant sensors for collaborative correction to ensure the accuracy and stability of the measurement data, and avoid long-term measurement deviations caused by sensor drift.

[0013] In the case of a failure or data loss of a certain sensor, the present invention uses the data of other sensors associated with the missing sensor for inference, generates reasonable alternative data, ensures the continuity and integrity of the data stream, and avoids the environmental monitoring blind area caused by data loss.

[0014] The present invention dynamically fuses the data of multiple sensors through an undirected graph model, and allocates the data contribution degree according to their correlation weights, ensuring that the fused data more conforms to the actual environmental conditions and improving the perception ability of the system in complex environments.

[0015] The present invention can dynamically adjust the weights of the edges in the undirected graph according to the historical trends of sensor data and the current environmental changes, ensuring that the system can adapt to the correlation relationships under different environmental conditions, and enhancing the robustness and adaptability of the system. Brief Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0017] Figure 1 is a flowchart of the method of the present invention. Detailed Embodiments

[0018] Next, in combination with the drawings and specific embodiments, a preferred description of the invention will be given.

[0019] This embodiment solves the above problems through the following steps: In one embodiment, referring to Figure 1 , the present invention provides an indoor environment detection method, which is a technical process of collecting and monitoring data of various physical or chemical factors affecting indoor environmental health, comfort and safety in real time through a sensor network deployed inside a building, and evaluating and regulating the state of the indoor environment through data processing, fusion and analysis technologies. This method mainly aims at key elements affecting the indoor environment such as temperature, humidity, air quality (such as concentrations of pollutants such as PM2.5, CO 2 , VOC, etc.), light intensity, noise level, etc., and uses Internet of Things technology, data fusion technology, etc. to form a complete indoor environment dynamic monitoring and feedback mechanism.

[0020] In the present invention, the indoor environment refers to elements such as climate, air, light, sound, etc. inside a building, mainly including temperature, humidity, air quality, noise, light intensity, etc., which directly affect the comfort, health and work efficiency of building users.

[0021] Collect indoor environmental data through multiple sensor networks distributed at different locations. The sensors are used to detect multiple environmental factors, including temperature, humidity, air quality, and light intensity.

[0022] In the present invention, the sensor network achieves comprehensive monitoring of environmental factors in each area of the indoor by multi-point layout, ensuring the extensiveness of the monitoring range and the accuracy of data collection.

[0023] The temperature sensors of the present invention are used to collect temperature data at different locations. By setting multiple temperature sensors to cover different areas of the indoor, including but not limited to the center of the room, corners, areas near windows or doors, etc., an accurate perception of the indoor temperature distribution can be achieved. The humidity sensors are used to monitor the relative humidity of the indoor air. By arranging multiple humidity sensors in the environment, the humidity levels of each area can be accurately reflected, especially in areas with large humidity changes such as kitchens, bathrooms, etc., ensuring the reliability of humidity monitoring.

[0024] The air quality sensors in the present invention are used to detect the pollutant concentration in the indoor air. The air quality detection includes but not limited to the real-time monitoring of multiple pollutants such as PM2.5, carbon dioxide (CO 2 ), volatile organic compounds (VOC), formaldehyde, etc. Through the accurate detection of these pollutants, the healthy level of the indoor air quality can be effectively guaranteed. The light intensity sensors are used to detect the indoor lighting conditions, including the monitoring of the light intensity of natural light and artificial light sources. By arranging light sensors at different heights and different locations, the dynamic monitoring of the light distribution and changes can be achieved.

[0025] Perform preprocessing on the environmental data. The preprocessing includes data cleaning, data normalization, and data synchronization.

[0026] The preprocessing steps of the present invention aim to improve the accuracy, consistency, and timeliness of the environmental data collected from multiple sensor networks distributed at different locations through a series of processing operations, ensuring the effectiveness and stability of subsequent data analysis, fusion, and regulation processes.

[0027] Among them, data cleaning is an important preprocessing step in the present invention. The data cleaning step is used to identify and remove abnormal data, invalid data, or noise data generated during the acquisition process. Specifically, the abnormal data may include abnormal readings caused by sensor failures, data transmission interruptions, or extreme environmental conditions. For example, during the sensor acquisition process, the temperature sensor may generate extremely deviated temperature values due to external interference in a short period of time. Such data will cause misjudgments in the overall system. Therefore, the present invention detects such extreme outliers through a data cleaning algorithm and deletes or corrects them. In addition, if some sensors have data missing due to network failures or power consumption problems, the cleaning step of the present invention will fill in the data through interpolation algorithms or data from neighboring sensors to maintain the continuity of the data. Through this step, the impact of abnormal situations on the overall quality of the data can be effectively reduced, ensuring the stability of the system input data.

[0028] Data normalization is another important preprocessing step in the present invention. The normalization process is used to address the differences in dimension and range of heterogeneous data obtained from multiple different types of sensors. For example, the sensors in the present invention may simultaneously collect data such as temperature, humidity, air quality (such as PM2.5 concentration), and light intensity. The numerical ranges and units of these data are not the same. Among them, temperature is expressed in degrees Celsius (°C), humidity is expressed as a percentage of relative humidity (%), and the concentration of PM2.5 in air quality is usually expressed in micrograms per cubic meter (µg / m³). To ensure that subsequent fusion algorithms and analysis algorithms can process these data on a unified data scale, the present invention introduces data normalization technology to convert different types of data into dimensionless data within the same dimension. Specifically, the normalization method can adopt min-max normalization to normalize each data value to the range of 0 to 1.

[0029] Data synchronization is a crucial part of the preprocessing steps in the present invention. The data synchronization is used to address the time deviation problems caused by different sampling frequencies or transmission delays of different sensors. The sensor network of the present invention may include multiple sensors with different sampling frequencies. For example, the temperature sensor collects data once a minute, while the air quality sensor may collect data once every 5 seconds. This inconsistency in sampling frequencies will cause the data to be unable to be directly compared in terms of time. The present invention aligns and interpolates the data at different time points through data synchronization technology to ensure that the data of all sensors are processed on the same time axis. For example, if the latest data of the air quality sensor at a certain moment is before or after the sampling data of the temperature sensor, the system will generate the air quality data at the corresponding time point through time interpolation methods to ensure that the environmental data is synchronized and consistent at each moment. The data synchronization process also includes unifying the time bases of sensors from different sources through timestamp matching technology to eliminate data deviations caused by transmission delays or sampling time asynchronization.

[0030] Through the above cleaning, normalization, and synchronization processes, the preprocessing step of the present invention can significantly improve the quality and consistency of environmental data. For example, for temperature and light data collected from different locations, if the light sensor experiences high-intensity interference caused by direct sunlight, the cleaning process of the present invention can detect this abnormal situation and correct the abnormal value using historical data or data from neighboring sensors. The normalization step uniformly processes data of different environmental factors, making temperature, humidity, air quality, etc. comparable, and ensuring that the accuracy of the model is not affected by differences in the data dimensions in subsequent algorithms. Data synchronization ensures that data from various sensors can be precisely matched in time, avoiding misjudgments caused by time lags.

[0031] Determine the correlation between the environmental data, and establish an undirected graph based on the correlation. The undirected graph is used to represent the association relationships between different types of environmental data, and analyze and process the data through a graph theory model to optimize the collaborative calibration and data fusion of various sensors in the sensor network. Based on the internal correlations between multiple environmental factors (such as temperature, humidity, light intensity, air quality, etc.), the present invention automatically establishes a correlation model between various environmental factors through historical data and algorithm analysis, and expresses it in the form of an undirected graph, thereby providing a structured basis for subsequent environmental monitoring and data processing.

[0032] The present invention first automatically identifies the correlation between the environmental data through historical data analysis and algorithm calculation. The correlation is determined by analyzing the statistical correlation, physical dependence relationship, or machine learning-based association model between environmental factor data (such as temperature, humidity, light intensity, air quality, etc.) collected by multiple sensors.

[0033] To this end, the present invention can use correlation coefficient calculation (such as Pearson correlation coefficient or Spearman correlation coefficient), multivariate regression analysis, and machine learning models (such as support vector machines, random forests, etc.) to analyze different types of sensor data and extract the dependence relationships between different environmental factors. For example, if a high correlation is found between temperature and humidity in historical data, or a strong physical relationship exists between light intensity and temperature, the system can identify them as strongly correlated data pairs.

[0034] Specifically, assume that the temperature data collected in the sensor network , humidity data , light intensity data , air quality data etc., and the following correlation matrix is obtained through correlation analysis: , Among them, represents the correlation coefficient between temperature and humidity, represents the correlation coefficient between temperature and light intensity, and so on. When the correlation coefficient exceeds a certain set threshold in absolute value, such as , it is considered that there is a significant association between the two, and an undirected graph will be established based on this association later.

[0035] According to the determined environmental data associations, the present invention establishes an undirected graph , where each node in the graph represents the environmental factor data (such as temperature, humidity, light intensity, etc.) collected by a sensor, and the edge represents the association between these environmental factors.

[0036] Specifically, if there is a significant association between two environmental factor data, then in the undirected graph, the corresponding nodes are connected by an undirected edge and . For example, if there is a strong association between the data collected by a temperature sensor and a humidity sensor, then in the undirected graph, there will be an undirected edge between the temperature node and the humidity node .

[0037] Assume that the associations between temperature, humidity, light, and air quality are as follows: Temperature and humidity have a strong positive correlation; Light intensity and temperature have a significant positive correlation; The association between air quality and temperature and light intensity is weak; then the corresponding undirected graph can be represented in the following form: , where represent temperature, humidity, light intensity, and air quality respectively, and the edge represents the associations between temperature and humidity, and temperature and light.

[0038] Based on the undirected graph, data collaborative correction, data fusion and inference, and environmental state assessment are performed. The undirected graph is constructed from the associations between different types of environmental data collected by multiple sensors and is used to represent the internal relationships between various environmental factor data. Based on the undirected graph, the present invention applies graph theory algorithms to perform data collaborative correction, data fusion and missing data inference, and environmental state assessment to achieve high-precision monitoring and intelligent control of the indoor environment.

[0039] Data collaborative correction based on the undirected graph: The present invention utilizes the correlation between nodes in the undirected graph to perform collaborative correction on anomalies or drifts in measurement data. Each node in the undirected graph represents the environmental data collected by a certain type of sensor, and the undirected edges between different nodes represent the correlation or dependency between environmental factor data. When it is detected that the data of a certain sensor drifts or is abnormal, the system will perform collaborative correction on it through the data of associated nodes to improve the accuracy of the data.

[0040] In an indoor environmental monitoring system, the correlation between the temperature sensor and the humidity sensor is represented by an edge in the undirected graph. Suppose the measured value of the current temperature sensor is 35°C, while the temperature value predicted by the system according to the correlation of the undirected graph calculated through the humidity value is 30°C. Since there is a large difference between the two, the system determines that the temperature sensor may have drifted, so it is corrected through the following collaborative correction formula: , where and are the weight coefficients automatically assigned by the system, and their magnitudes are determined by the edges of the undirected graph. Suppose , , then the corrected temperature is: , Through this collaborative correction, the present invention effectively reduces the impact of sensor drift on the system accuracy and ensures the reliability of environmental data.

[0041] Data fusion and missing data inference based on undirected graph: The present invention performs fusion processing on data from different sensors through the data association model of the undirected graph. Especially in the case where the data of a certain sensor is missing or faulty, it uses the data of other sensors to infer the missing data. The edges in the undirected graph represent the association strength between sensors, and the system infers the missing data based on this correlation.

[0042] Suppose that at a certain moment, the light intensity sensor breaks down and fails to collect light data. The system infers the light data based on the correlation between light and temperature, humidity in the undirected graph. The association model can be represented by a multiple regression formula as: , where is the current temperature, is the current humidity, , are the regression coefficients, and their magnitudes are determined by the edges of the undirected graph. Suppose the current temperature , humidity , obtained through the association model: , The light intensity inferred by the system through this formula is 72 lux. Even if the light sensor fails, the system can still provide reliable light data.

[0043] Environmental state assessment based on undirected graph: The present invention evaluates the state of the entire indoor environment through an undirected graph. Especially when the correlation between multiple environmental factors changes significantly, it can quickly detect abnormal situations and issue early warnings. The edges of the undirected graph represent the normal correlation relationships between environmental factors. When a certain correlation deviates significantly from the historical pattern, the system will judge that there may be an abnormality in this environmental factor.

[0044] In a certain office environment, there is usually a strong positive correlation between temperature and humidity. Through the undirected graph, the system records the historical correlation between temperature and humidity. When it detects that the temperature rises while the humidity abnormally decreases below the normal range, the system judges that the correlation between temperature and humidity has deviated abnormally according to the edges in the undirected graph and triggers the environmental state abnormal assessment mechanism.

[0045] Based on the overall state of each node and associated edges in the undirected graph, the system judges that there may be problems such as overly dry air or air conditioner failure in the room, and issues corresponding warning signals to prompt users or automatic control devices. For example, automatically start the humidifier or lower the air conditioner cooling intensity to restore the normal environmental state.

[0046] On the other hand, the present invention also provides an indoor environment detection system, including: A collection module for collecting indoor environmental data through a plurality of sensor networks distributed at different positions. The sensors are used to detect multiple environmental factors, and the environmental factors include temperature, humidity, air quality, and light intensity; A preprocessing module for preprocessing the environmental data. The preprocessing includes data cleaning, data normalization, and data synchronization; An association module for determining the correlation between the environmental data and establishing an undirected graph according to the correlation; A processing module for performing collaborative correction, data fusion and inference, and environmental state assessment on the real-time indoor environmental detection data according to the undirected graph.

[0047] For the part of the module structure not specifically defined in the present invention, it shall be subject to the content recorded in the prior art. The prior art mentioned in the foregoing background art part and specific embodiment part of the present invention can be used as a part of the present invention to understand the meaning of some technical features or parameters.

Claims

1. A method for detecting an indoor environment, characterized in that: The method comprises the following steps: Collecting indoor environmental data through a plurality of sensor networks distributed at different locations, wherein the sensors are used to detect a plurality of environmental factors, including temperature, humidity, air quality, and light intensity; Preprocessing the environmental data, wherein the preprocessing includes data cleaning, data normalization, and data synchronization; Determine the correlation between the environmental data, and establish an undirected graph according to the correlation; According to the undirected graph, collaborative correction, data fusion and inference, and environmental status evaluation are performed on the real-time indoor environment detection data.

2. The indoor environment detection method according to claim 1, characterized in that: The determining the correlation between the environmental data and establishing an undirected graph according to the correlation comprises: Use historical data to analyze the correlation between different types of sensor data through statistical correlation; According to the correlation analysis, the correlation coefficients between different environmental factors are calculated to generate a correlation matrix. If the absolute value of the correlation coefficient of two environmental factors exceeds the set threshold, it is considered that there is a significant correlation between them. According to the determined correlation, an undirected graph is constructed, in which each node represents an environmental element, and the edge represents the correlation between environmental elements; If there is a significant correlation between two environmental factors, an undirected edge is added to the undirected graph to connect the corresponding nodes.

3. The indoor environment detection method according to claim 1, characterized in that: The collaborative correction includes: During real-time monitoring, the system automatically detects whether the data of a sensor is abnormal or drifting based on the trend of historical data or the deviation between the current measurement value and the data of other nodes in the undirected graph; Determine sensor nodes associated with abnormal or drifting data through an undirected graph, and extract data of these associated sensors; According to the weights of the associated sensor data and the edges in the undirected graph, a collaborative correction model is used to correct the abnormal data, and the correction formula is weighted summed according to the contribution of the associated data.

4. The indoor environment detection method according to claim 1, characterized in that: The data fusion and inference include: During real-time monitoring, the system detects data loss or sensor failure of a certain sensor and identifies the environmental factors with missing data and their corresponding nodes; According to the edges in the undirected graph, the data of other sensors associated with the missing data are extracted, and the weight of the edge represents the influence of the associated sensor on the missing data; Utilize the associated sensor data and fuse the multi-source data according to the edge weights in the undirected graph; The inferred value of the missing data is calculated through the data fusion model, and the result is adjusted according to the weight of each sensor in the undirected graph to ensure that the inferred missing data is reasonable and consistent with the current environmental status.

5. The indoor environment detection method according to claim 1, characterized in that: The environmental status assessment includes: During the environmental monitoring process, the system collects data from each sensor in real time, monitors the status and change trend of each node in the undirected graph, and monitors the weight and data relevance of the edges in the undirected graph at the same time; The system detects whether the edges in the undirected graph have changed significantly, and whether the strength of the association between environmental elements deviates from the normal range. When the weight of a certain association increases or decreases abnormally, it indicates that there is an abnormality or sudden change in the relevant environmental elements. Based on the detected changes in edge weights, the system analyzes which environmental elements have abnormal relationships and outputs them.

6. An indoor environment detection system, characterized in that: The system includes the following modules: A collection module, used to collect indoor environmental data through a plurality of sensor networks distributed at different locations, wherein the sensors are used to detect a plurality of environmental factors, including temperature, humidity, air quality, and light intensity; A preprocessing module, used for preprocessing the environmental data, wherein the preprocessing includes data cleaning, data normalization, and data synchronization; An association module, used to determine the association between the environmental data and establish an undirected graph according to the association; The processing module is used to perform collaborative correction, data fusion and inference, and environmental status evaluation on the real-time indoor environment detection data according to the undirected graph.

7. An indoor environment detection system according to claim 6, characterized in that: The determining the correlation between the environmental data and establishing an undirected graph according to the correlation comprises: Use historical data to analyze the correlation between different types of sensor data through statistical correlation; According to the correlation analysis, the correlation coefficients between different environmental factors are calculated to generate a correlation matrix. If the absolute value of the correlation coefficient of two environmental factors exceeds the set threshold, it is considered that there is a significant correlation between them. According to the determined correlation, an undirected graph is constructed, in which each node represents an environmental element, and the edge represents the correlation between environmental elements; If there is a significant correlation between two environmental factors, an undirected edge is added to the undirected graph to connect the corresponding nodes.

8. An indoor environment detection system according to claim 6, characterized in that: The collaborative correction includes: During real-time monitoring, the system automatically detects whether the data of a sensor is abnormal or drifting based on the trend of historical data or the deviation between the current measurement value and the data of other nodes in the undirected graph; Determine sensor nodes associated with abnormal or drifting data through an undirected graph, and extract data of these associated sensors; According to the weights of the associated sensor data and the edges in the undirected graph, a collaborative correction model is used to correct the abnormal data, and the correction formula is weighted summed according to the contribution of the associated data.

9. The indoor environment detection system according to claim 6, characterized in that: The data fusion and inference include: During real-time monitoring, the system detects data loss or sensor failure of a certain sensor and identifies the environmental factors with missing data and their corresponding nodes; According to the edges in the undirected graph, the data of other sensors associated with the missing data are extracted, and the weight of the edge represents the influence of the associated sensor on the missing data; Utilize the associated sensor data and fuse the multi-source data according to the edge weights in the undirected graph; The inferred value of the missing data is calculated through the data fusion model, and the result is adjusted according to the weight of each sensor in the undirected graph to ensure that the inferred missing data is reasonable and consistent with the current environmental status.

10. The indoor environment detection system according to claim 6, characterized in that: The environmental status assessment includes: During the environmental monitoring process, the system collects data from each sensor in real time, monitors the status and change trend of each node in the undirected graph, and monitors the weight and data relevance of the edges in the undirected graph at the same time; The system detects whether the edges in the undirected graph have changed significantly, and whether the strength of the association between environmental elements deviates from the normal range. When the weight of a certain association increases or decreases abnormally, it indicates that there is an abnormality or sudden change in the relevant environmental elements. Based on the detected changes in edge weights, the system analyzes which environmental elements have abnormal relationships and outputs them.

Images