Building comprehensive early warning method and system based on database

Abstract

The invention discloses a building comprehensive early warning method and system based on a database, and relates to the technical field of building safety early warning, and the method comprises the following steps: S1, installing distributed monitoring sensors at each floor in a building, and building a coordinate system through the plane where each floor is located; s2, dividing node grids in each layer of coordinate system, and analyzing associated monitoring data of unknown nodes based on actual monitoring data acquired by known nodes; s3, periodically updating the monitoring data of all floors in the building database, and identifying potential risks in each floor based on the comprehensive early warning model; and S4, verifying the accuracy of the safety early warning notification, and outputting a building management decision after the verification is passed. According to the invention, a building database is constructed through the distributed monitoring sensors and building of a floor coordinate system, and effective storage and management of monitoring data of each floor are realized; risk identification and safety early warning notification are carried out based on the comprehensive early warning model, and timely decision support is provided for building managers.

Description

Technical Field

[0001] The present invention relates to the technical field of building safety early warning, and in particular to a database-based comprehensive building early warning method and system. Background Art

[0002] Building is usually used to describe high-rise or multi-story buildings, especially buildings for commercial, office or residential purposes. In the urban environment, buildings, as an important part of the city, play the role of places for people to work, live and play. With the acceleration of urbanization and the continuous development of construction technology, the scale and complexity of buildings are also increasing. However, with this comes a variety of potential safety hazards and operational risks, such as fire, gas leaks, water leaks, electrical failures, etc. In order to ensure the safety and normal operation of the interior of the building, building early warning technology came into being.

[0003] Building early warning technology is a system that uses advanced sensors, monitoring equipment and intelligent systems to conduct real-time monitoring and early warning of various safety and operating conditions inside buildings. It aims to improve the safety and operating efficiency of buildings and reduce the occurrence of accidents; to protect the lives and property of residents and staff, and to promptly detect and respond to potential risks. The functions of building early warning technology include real-time monitoring, rapid early warning, remote management and emergency response, in order to improve the perception and response capabilities of building managers and related personnel to potential dangers and ensure a safe environment inside the building.

[0004] Buildings are places where people work and live, with dense populations and frequent activities. Once an accident occurs, the consequences may be very serious. Secondly, there are various potential safety hazards and operational risks inside buildings, such as fire, gas leakage, water leakage, etc., which need to be discovered and responded to in a timely manner through early warning systems. Thirdly, with the acceleration of urbanization, the scale of buildings continues to expand, and the difficulty of management increases. It is necessary to use early warning technology to improve management efficiency and level.

[0005] Although building early warning technology has played an important role in improving building safety, there are still some defects and challenges. First, there is a high false alarm rate in the existing early warning technology, which makes it difficult to meet the safety needs of comprehensive building monitoring and early warning. Secondly, the maintenance cost of large-scale early warning equipment is high, and regular inspection and replacement are required, which increases management costs; and a small number of monitoring equipment cannot guarantee the comprehensiveness of monitoring data in the building, and the monitoring points cannot be effectively linked to form early warning coverage of the building.

[0006] Therefore, in order to overcome these defects and challenges, further technological innovation and improvement are needed. Summary of the invention

[0007] Based on this, it is necessary to provide a database-based comprehensive building early warning method and system to address the above technical issues.

[0008] In a first aspect, the present invention provides a database-based comprehensive building early warning method, the method comprising the following steps: S1. Install distributed monitoring sensors on each floor of the building, establish a coordinate system based on the plane where each floor is located, and record the position coordinates of each monitoring sensor on the floor to which it belongs; S2, divide the node grid in each layer coordinate system, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database; S3. Regularly update the monitoring data of all floors in the building database, identify potential risks on each floor based on the comprehensive early warning model, and output safety warning notifications; S4. Verify the accuracy of the safety warning notification, and after verification, output the building management decision.

[0009] Furthermore, node grids are divided in each layer of coordinate system, and based on the actual monitoring data collected by known nodes, the associated monitoring data of unknown nodes are analyzed to construct a building database, which includes the following steps: S21. Arrange the coordinate system numbers according to the order of the floors, and divide the node network in the coordinate system, with the monitoring sensors as known nodes and the nodes without monitoring sensors installed as unknown nodes; S22, obtaining actual monitoring data of known nodes, establishing a spatial autoregressive model, predicting associated monitoring data of unknown nodes, and verifying the legitimacy of the associated monitoring data; S23. Integrate the monitoring data of all nodes in a single floor to form a floor data table, set data monitoring strategy and data storage structure, and build a building database.

[0010] Furthermore, obtaining the actual monitoring data of known nodes, establishing a spatial autoregressive model, predicting the associated monitoring data of unknown nodes, and verifying the legitimacy of the associated monitoring data include the following steps: S221, define the spatial distance weight coefficient component and the self-fitting coefficient between any two nodes in the node network, if any two nodes are different nodes, the self-fitting coefficient is 1, if any two nodes are the same nodes, the self-fitting coefficient is 0; S222, using the spatial distance between nodes as network input and the spatial weight coefficient component as network output, establishing a spatial neural network, and performing nonlinear fitting on the spatial weight coefficient component; S223, taking the product of the spatial weight coefficient component after nonlinear fitting and the self-fitting coefficient as the spatial weight coefficient, and establishing a spatial weight matrix; S224, multiplying the monitoring data of the known nodes by the spatial weight matrix, constructing a spatial autoregressive model, and setting a training set and a validation set to perform anti-fitting training on the spatial autoregressive model; S225. Based on the spatial autoregressive model, known nodes are used to predict the associated monitoring data of adjacent unknown nodes, and the legitimacy of all associated monitoring data of unknown nodes is verified.

[0011] Furthermore, setting the training set and the validation set to perform anti-fitting training on the spatial autoregressive model includes the following steps: S2241, dividing the actual monitoring data into a training set and a validation set; S2242, monitoring the mean square error of the validation set, and terminating the training in advance when the mean square error of the training set continues to decrease and the mean square error of the validation set begins to increase; S2243. Using regularization techniques, randomly hide some neurons in the spatial neural network and introduce a learnable activation function for each layer in the spatial neural network. S2244, use batch normalization technology to accelerate the model training process and improve stability.

[0012] Furthermore, using known nodes to predict the associated monitoring data of adjacent unknown nodes and verifying the legitimacy of all associated monitoring data of unknown nodes includes the following steps: S2251, setting the correlation radius of the known nodes, drawing a circle with the correlation radius as the prediction range of the known nodes, and using the spatial regression model to calculate the correlation monitoring data of the unknown nodes within the prediction range; S2252. Count all associated monitoring data of each unknown node predicted by adjacent known nodes, calculate the mean difference of all associated monitoring data, if the mean difference is less than the preset mean difference threshold, determine that the associated monitoring data is legal, and calculate the average value as the only associated monitoring data; if the mean difference is greater than the preset mean difference threshold, determine that the associated monitoring data is illegal, feedback the illegal information, eliminate the outliers in the associated monitoring data, and then calculate the average value as the only associated monitoring data.

[0013] Furthermore, the monitoring data of all nodes in a single floor are integrated to form a floor data table, and the data monitoring strategy and data storage structure are set. The construction of the building database includes the following steps: S231, configuring a dynamic performance view of monitoring data, and establishing floor data tables and relationships between tables for storing different types of monitoring data to form a monitoring data dictionary; S232, for each floor data table corresponding to each floor, a customized monitoring method is performed to monitor different operations of different floor data tables; S233. Construct floor data tables and table structures for different floors in the building to form a building database.

[0014] Furthermore, the monitoring data of all floors in the building database are regularly updated, and based on the comprehensive early warning model, the potential risks in each floor are identified, and the output of the safety early warning notification includes the following steps: S31. Based on the set monitoring frequency, obtain the monitoring data of each floor in the building database; S32. Set up an early warning system based on the type of building safety risks and establish a comprehensive early warning model; S33. Based on the comprehensive early warning model, analyze the potential risks of all nodes on a single floor, and locate the risks according to the risk level and location information of each node; S34. Based on the node risk identification result, output a safety warning notification including location information.

[0015] Furthermore, according to the building safety risk type, an early warning system is set up, and a comprehensive early warning model is established, including the following steps: S321. Set the types of security risks existing in the building, select corresponding types of monitoring data as risk warning indicators of the security risk types, and establish a building risk warning indicator system and warning level; S322. Set the warning level critical points of the risk warning indicators, and establish a risk degree calculation formula based on the warning level critical points to calculate the risk degree of each risk warning indicator; S323. Obtain the risk calculated from historical monitoring data as sample data to form training samples and test samples, build a comprehensive early warning model, and conduct training and testing to obtain optimal model parameters.

[0016] Furthermore, the risk calculation formula is: In the formula, y i Indicates i The risk level of each risk warning indicator; x i Indicates i The value of a risk warning indicator; k Indicates the warning level number; x i(k) and x i(k+1) Indicates i The critical point of the warning level of each risk warning indicator; y i(k) and y i(k+1) Indicates x i(k) and x i(k+1) The warning level vector of the risk warning indicators.

[0017] In a second aspect, the present invention further provides a database-based comprehensive building early warning system, the system comprising: The monitoring module is used to install distributed monitoring sensors on each floor of the building, and to establish a coordinate system based on the plane where each floor is located, and to record the position coordinates of each monitoring sensor on the floor to which it belongs; The database module is used to divide the node grids in the coordinate systems of each layer, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database; The early warning identification module is used to regularly update the monitoring data of all floors in the building database, identify potential risks on each floor based on the comprehensive early warning model, and output safety early warning notifications; Verification decision module, used to verify the accuracy of safety warning notifications, and output building management decisions after verification; Among them, the monitoring module, database module, early warning identification module and verification decision module are connected in sequence.

[0018] The beneficial effects of the present invention are: 1. Through the establishment of distributed monitoring sensors and floor coordinate systems, an overall comprehensive building database is constructed, realizing the effective storage and management of monitoring data of each floor; by dividing the node grid and the spatial autoregressive model based on known node data, the monitoring data of unknown nodes can be predicted, and the legitimacy of the data can be verified to ensure the accuracy and reliability of the data; combined with real-time monitoring to update the database, and based on the comprehensive early warning model to identify risks and provide safety early warning notifications, it provides timely decision-making support for building managers and ensures the safe operation and management efficiency of the building.

[0019] 2. By acquiring monitoring data from each floor and updating it based on the set monitoring frequency, a comprehensive early warning model was established, and an early warning system was set up according to the type of building safety risks, which can comprehensively identify potential risks on each floor; the comprehensive early warning model is used to analyze risks, and the risks are located in combination with node location information, thereby achieving accurate identification and location of risks; finally, by outputting safety warning notifications containing location information, timely and accurate risk warning information is provided to building managers, which helps to take targeted management measures to ensure the safe operation of the building. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings: Figure 1 is a flow chart of a database-based comprehensive building early warning method according to an embodiment of the present invention; Figure 2 The present invention is a system principle block diagram of a database-based comprehensive building early warning system according to an embodiment of the present invention.

[0021] Figure numbers: 1. Monitoring module; 2. Database module; 3. Early warning identification module; 4. Verification and decision module. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0023] See also Figure 1 , provides a database-based building comprehensive early warning method and system, the method comprising the following steps: S1. Install distributed monitoring sensors on each floor of the building, and establish a coordinate system based on the plane where each floor is located to record the position coordinates of each monitoring sensor on its corresponding floor.

[0024] Specifically, the location of the monitoring point on each floor is determined first. Considering the coverage and monitoring needs of the monitoring point, it is usually chosen to arrange the monitoring sensors in the center of the floor or in key areas, or to set a number of monitoring sensors at equal intervals according to the building area. Sensors include various types, such as temperature sensors, humidity sensors, smoke sensors, piezoelectric sensors, and power sensors, etc., which are selected according to the specific conditions and safety needs of the building.

[0025] For each floor, establish a plane coordinate system to describe the location of the sensor. Usually a Cartesian coordinate system is used, where the center point of the floor is used as the origin, the horizontal and vertical axes are established, and the positive direction and unit of the coordinate axis are determined. In the established coordinate system, the location coordinates of each monitoring sensor are recorded. The location of each sensor can be represented by a two-dimensional coordinate, such as (x, y), where x and y represent the horizontal and vertical positions of the sensor, respectively. The node coordinates are used for subsequent monitoring data analysis and location information query.

[0026] S2. Divide the node grid in each layer coordinate system, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database.

[0027] In the description of the present invention, node grids are divided in each layer coordinate system, and based on the actual monitoring data collected by known nodes, the associated monitoring data of unknown nodes are analyzed to construct a building database, which includes the following steps: S21. Arrange the coordinate system numbers according to the order of the floors, and divide the node network in the coordinate system, with monitoring sensors as known nodes and nodes without monitoring sensors installed as unknown nodes.

[0028] Specifically, the order of the floors in the building is determined, and they are numbered from the bottom floor to the top floor or from the top floor to the bottom floor, ensuring that each floor has a unique serial number. For each floor, the serial number of the coordinate system is arranged according to the determined order of the floors, and the order of the floors is selected as the serial number of the coordinate system, ensuring that each floor has a unique serial number of the coordinate system.

[0029] In the coordinate system of each floor, the node network is divided, and the monitoring sensors are regarded as known nodes, and the nodes without monitoring sensors are regarded as unknown nodes. The density and layout of the node network are determined according to the actual situation to ensure the monitoring coverage and data collection efficiency. For known nodes (i.e. nodes with monitoring sensors installed), their location coordinates are recorded. For unknown nodes, their location coordinates are determined according to the division of the node network and recorded in the coordinate system together with the known nodes.

[0030] S22. Obtain the actual monitoring data of known nodes, establish a spatial autoregressive model, predict the associated monitoring data of unknown nodes, and verify the legitimacy of the associated monitoring data.

[0031] In the description of the present invention, obtaining actual monitoring data of known nodes, establishing a spatial autoregressive model, predicting associated monitoring data of unknown nodes, and verifying the legitimacy of the associated monitoring data include the following steps: S221. Define the spatial distance weight coefficient component and the self-fitting coefficient between any two nodes in the node network. If any two nodes are different nodes, the self-fitting coefficient is 1; if any two nodes are the same nodes, the self-fitting coefficient is 0.

[0032] Among them, the self-fitting coefficient refers to the coefficient used to measure the relationship between the node itself and itself in the spatial autoregressive model. When defining the spatial distance weight coefficient component and the self-fitting coefficient between any two nodes in the node network, if any two nodes are the same node, the self-fitting coefficient is 0, indicating that there is no spatial distance weight between the node itself and the node itself, because there is no distance between the node itself and itself.

[0033] S222. Using the spatial distance between nodes as network input and the spatial weight coefficient component as network output, a spatial neural network is established to perform nonlinear fitting on the spatial weight coefficient component.

[0034] Specifically, the spatial distance between nodes is used as the input of the spatial neural network, and the spatial weight coefficient component is used as the output of the network to reflect the relationship between nodes, which needs to be predicted and fitted by the neural network.

[0035] The spatial neural network includes: an input layer that accepts the spatial distance between nodes as input; an intermediate hidden layer that extracts features and performs nonlinear transformations through multiple hidden layers, and uses various activation functions (such as ReLU, Sigmoid, Tanh, etc.) to increase the nonlinear ability of the network; an output layer, where the number of neurons in the output layer should match the dimension of the spatial weight coefficient component to output the predicted spatial weight coefficient component.

[0036] S223. The product of the spatial weight coefficient component after nonlinear fitting and the self-fitting coefficient is used as the spatial weight coefficient to establish a spatial weight matrix.

[0037] Specifically, the expression of the spatial weight matrix is: In the formula, λ represents the spatial weight coefficient; ρ represents the spatial weight coefficient component; h represents the self-fitting coefficient; n Indicates the number of nodes.

[0038] S224, multiplying the monitoring data of the known nodes by the spatial weight matrix, constructing a spatial autoregressive model, and setting a training set and a validation set to perform anti-fitting training on the spatial autoregressive model.

[0039] Among them, the expression of the spatial autoregressive model is: In the formula, s represents the actual monitoring data of the known node; Represents the associated monitoring data of unknown nodes.

[0040] In the description of the present invention, setting a training set and a validation set to perform anti-fitting training on a spatial autoregressive model includes the following steps: S2241. Divide the actual monitoring data into a training set and a validation set.

[0041] S2242. Monitor the mean square error of the validation set. When the mean square error of the training set continues to decrease and the mean square error of the validation set starts to increase, terminate the training early.

[0042] Among them, Mean Squared Error (MSE) is a commonly used indicator to measure the prediction error of the model, which can be used to evaluate the performance of the model on the training set and the validation set. The idea of ​​early termination of training is to monitor the performance on the validation set during the training process and stop training when the performance of the model on the validation set begins to decline, so as to prevent the model from overfitting the training data.

[0043] The specific steps are as follows: Train the model and calculate the mean squared error (MSE) on the training and validation sets.

[0044] When the MSE of the training set continues to decrease and the MSE of the validation set begins to increase, it means that the model may have begun to overfit the training data, that is, it performs well on the training set but poorly on the validation set. In this case, the training can be terminated early to avoid further decline in the performance of the model on the validation set, thereby improving the generalization ability of the model.

[0045] S2243. Using regularization technology, some neurons in the spatial neural network are randomly hidden, and a learnable activation function is introduced for each layer in the spatial neural network.

[0046] Among them, Dropout (regularization technology) reduces the complexity of the model and prevents overfitting by randomly hiding some neurons in the network during training. In spatial neural networks, by adding a Dropout layer after each hidden layer, the proportion of discarded neurons is specified to force the model to learn more robust feature representations and improve generalization capabilities.

[0047] In addition, the learnable activation function can be PReLU (Parametric Rectified Linear Unit), which is a modified linear unit activation function with parameters. Compared with the traditional ReLU activation function, PReLU has stronger expressive power. In each layer of the spatial neural network, PReLU is used as the activation function, so that the network can better learn nonlinear relationships and improve the model's fitting ability.

[0048] S2244, use batch normalization technology to accelerate the model training process and improve stability.

[0049] Among them, Batch Normalization (BN) is used to accelerate the model training process and improve the stability of the model. Its main idea is to normalize the intermediate layers of the neural network on each small batch of training data to alleviate problems such as internal covariate shift and gradient disappearance / explosion, thereby accelerating convergence and improving model performance.

[0050] Specifically, batch normalization technology is introduced into the spatial neural network, and the following steps are performed: 1. Add a batch normalization layer after each hidden layer. Add a batch normalization layer after each hidden layer of the spatial neural network. In this way, the output of the hidden layer will be normalized on each small batch of data.

[0051] 2. For the output of each hidden layer, calculate its mean and variance on each mini-batch of data.

[0052] 3. Use the batch mean and variance to normalize the output of the hidden layer so that the mean of the output is close to 0 and the variance is close to 1.

[0053] 4. Scaling and translation operations are performed on the normalized output so that the model can learn more complex transformations. In this step, two learnable parameters gamma and beta are introduced for scaling and translation operations respectively.

[0054] 5. Before applying the activation function, the scaled and translated output is used as the final output of the hidden layer.

[0055] S225. Based on the spatial autoregressive model, known nodes are used to predict the associated monitoring data of adjacent unknown nodes, and the legitimacy of all associated monitoring data of unknown nodes is verified.

[0056] In the description of the present invention, using known nodes to predict the associated monitoring data of adjacent unknown nodes and verifying the legitimacy of all associated monitoring data of the unknown nodes includes the following steps: S2251. Set the association radius of the known nodes, draw a circle with the association radius as the prediction range of the known nodes, and use the spatial regression model to calculate the association monitoring data of the unknown nodes within the prediction range.

[0057] S2252. Count all associated monitoring data of each unknown node predicted by adjacent known nodes, calculate the mean difference of all associated monitoring data, if the mean difference is less than the preset mean difference threshold, determine that the associated monitoring data is legal, and calculate the average value as the only associated monitoring data; if the mean difference is greater than the preset mean difference threshold, determine that the associated monitoring data is illegal, feedback the illegal information, eliminate the outliers in the associated monitoring data, and then calculate the average value as the only associated monitoring data.

[0058] Specifically, for illegal information, it is necessary to eliminate outliers in the associated monitoring data, and it is necessary to use statistical methods to identify outliers, such as the 3σ principle: calculate the mean and standard deviation based on the associated monitoring data predicted by the known nodes, and then define a threshold range, that is, the mean plus or minus a number of standard deviations, such as 3σ, and treat data points outside this range as outliers.

[0059] S23. Integrate the monitoring data of all nodes in a single floor to form a floor data table, set data monitoring strategy and data storage structure, and build a building database.

[0060] In the description of the present invention, the monitoring data of all nodes in a single floor are integrated to form a floor data table, and the data monitoring strategy and data storage structure are set. The construction of the building database includes the following steps: S231. Configure a dynamic performance view of monitoring data, and establish floor data tables for storing different types of monitoring data and relationships between tables to form a monitoring data dictionary.

[0061] Specifically, a dynamic performance view is created in the database to monitor the monitoring data of the floor in real time. According to the type of monitoring data (such as temperature, humidity, air quality, etc.), the corresponding floor data table is created, and the relationship between the tables is established to facilitate data integration and query; a monitoring data dictionary is formed to record the definition, source, storage location and other information of each monitoring data for easy management and maintenance.

[0062] S232. Customize the monitoring method for the floor data table corresponding to each floor to monitor different operations of different floor data tables; this can prevent excessive occupation of database resources and more standardized management of the large amount of monitoring data generated by the monitoring process.

[0063] Specifically, set data monitoring strategies, including monitoring frequency, outlier detection, data backup, etc., to ensure data integrity and reliability. For different types of monitoring data, formulate corresponding monitoring methods and processing rules, such as setting thresholds and alarm mechanisms.

[0064] In the present invention, an Oracle database is used as the basis for building a building database, and the definition of the Oracle database monitoring method is located in the data dictionary DBA_AUDIT_POLICIES view under the highest authority SYS administrator user. The policy view contains a brief description of some important columns in the monitoring dynamic view. By writing a program script, the addition is performed under the SYSTEM local administrator user. The administrator can add monitoring policies for different data tables by entering different audit policy commands. In the monitoring policy table, each row of data represents a monitoring policy. The PNAME field is the audit policy name, and STMT_TYPE is the different types of monitoring policies. Any combination of the four operation modes of SELECT, IN-SERT, DELETE and UPDATE can be monitored arbitrarily.

[0065] S233. Construct floor data tables and table structures for different floors in the building to form a building database.

[0066] Specifically, create a data table for each floor according to the building structure and monitoring requirements, and design a suitable table structure, including field names, data types, constraints, etc. Ensure the uniformity and standardization of the floor data table to facilitate data management and analysis. Integrate the data tables of each floor into the building database to form a complete database system to provide support for subsequent data analysis and decision-making.

[0067] S3. Regularly update the monitoring data of all floors in the building database, identify potential risks on each floor based on the comprehensive early warning model, and output safety warning notifications.

[0068] In the description of the present invention, the monitoring data of all floors in the building database are regularly updated, and based on the comprehensive early warning model, the potential risks in each floor are identified, and the output of the safety early warning notification includes the following steps: S31. Based on the set monitoring frequency, obtain the monitoring data of each floor in the building database.

[0069] S32. Set up an early warning system based on the type of building safety risks and establish a comprehensive early warning model.

[0070] In the description of the present invention, according to the building safety risk type, setting up an early warning system and establishing a comprehensive early warning model includes the following steps: S321. Set the types of security risks in the building, select corresponding types of monitoring data as risk warning indicators for the security risk types, and establish a building risk warning indicator system and warning level.

[0071] Specifically, classify the possible safety risks that buildings may face, such as fire, water leakage, power failure, wall load-bearing problems, and passage blockage. For each type of safety risk, select appropriate monitoring data as risk warning indicators, such as temperature, humidity, water pressure, and wall pressure. On the basis of each indicator, establish a risk warning indicator system, and divide the warning level according to the risk level to facilitate subsequent risk assessment and warning processing.

[0072] S322. Set the warning level critical points of the risk warning indicators, and based on the warning level critical points, establish a risk calculation formula to calculate the risk level of each risk warning indicator.

[0073] In the description of the present invention, the risk calculation formula is: In the formula, y i Indicates i The risk level of each risk warning indicator; x i Indicates i The value of a risk warning indicator;k Indicates the warning level number; x i(k) and x i(k+1) Indicates i The critical point of the warning level of each risk warning indicator; y i(k) and y i(k+1) Indicates x i(k) and x i(k+1) The warning level vector of the risk warning indicators.

[0074] S323. Obtain the risk calculated from historical monitoring data as sample data to form training samples and test samples, build a comprehensive early warning model, and conduct training and testing to obtain optimal model parameters.

[0075] S33. Based on the comprehensive early warning model, analyze the potential risks of all nodes on a single floor, and locate the risks according to the risk level and location information of each node.

[0076] Specifically, the acquired monitoring data is analyzed and evaluated using the established comprehensive early warning model to calculate the risk level of each node. According to the early warning level set by the building early warning system, the risk level of each node is compared with the early warning level to determine the risk level of the node; then, combined with the location information of the node, the nodes with higher risk are located. That is, according to the floor plan or coordinate system, the position of each node is accurately located within the floor to determine which areas or equipment have higher risks.

[0077] S34. Based on the node risk identification result, output a safety warning notification including location information.

[0078] S4. Verify the accuracy of the safety warning notification, and after verification, output the building management decision.

[0079] See also Figure 2 , and also provides a database-based building comprehensive early warning system, the system comprising: The monitoring module 1 is used to install distributed monitoring sensors on each floor of the building, and to establish a coordinate system based on the plane where each floor is located, and to record the position coordinates of each monitoring sensor on the floor to which it belongs.

[0080] Database module 2 is used to divide the node grid in each layer coordinate system, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database.

[0081] The early warning identification module 3 is used to regularly update the monitoring data of all floors in the building database, and based on the comprehensive early warning model, identify the potential risks in each floor and output safety early warning notifications.

[0082] The decision verification module 4 is used to verify the accuracy of the safety warning notification and output the building management decision after the verification is passed.

[0083] Among them, the monitoring module 1, the database module 2, the early warning identification module 3 and the verification decision module 4 are connected in sequence.

[0084] In summary, with the help of the above technical solution of the present invention, a comprehensive building database is constructed through the establishment of distributed monitoring sensors and floor coordinate systems, and the effective storage and management of monitoring data of each floor is realized; by dividing the node grid and the spatial autoregressive model based on the known node data, the monitoring data of unknown nodes can be predicted, and the legitimacy of the data can be verified to ensure the accuracy and reliability of the data; combined with real-time monitoring to update the database, and based on the comprehensive early warning model to identify risks and safety early warning notifications, timely decision-making support is provided for building managers, and the safe operation and management efficiency of the building are guaranteed. By obtaining the monitoring data of each floor and updating it based on the set monitoring frequency, a comprehensive early warning model is established, and an early warning system is set according to the type of building safety risk, which can comprehensively identify the potential risks in each floor; using the comprehensive early warning model to analyze risks, and combining the node location information to locate risks, accurate identification and location of risks are realized; finally, by outputting the safety early warning notification containing location information, timely and accurate risk early warning information is provided to building managers, which helps to take targeted management measures to ensure the safe operation of the building.

[0085] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

Claims

1. A database-based building comprehensive early warning method, characterized in that: The method comprises the following steps: S1. Install distributed monitoring sensors on each floor of the building, establish a coordinate system based on the plane where each floor is located, and record the position coordinates of each monitoring sensor on its corresponding floor; S2, divide the node grid in each layer coordinate system, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database; S3. Regularly update the monitoring data of all floors in the building database, identify potential risks on each floor based on the comprehensive early warning model, and output safety warning notifications; S4. Verify the accuracy of the safety warning notification, and after verification, output the building management decision.

2. A database-based building comprehensive early warning method according to claim 1, characterized in that: The step of dividing the node grids in the coordinate systems of each layer, analyzing the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes, and constructing a building database includes the following steps: S21. Arrange the coordinate system numbers according to the order of the floors, and divide the node network in the coordinate system, with the monitoring sensors as known nodes and the nodes without monitoring sensors installed as unknown nodes; S22, obtaining actual monitoring data of known nodes, establishing a spatial autoregressive model, predicting associated monitoring data of unknown nodes, and verifying the legitimacy of the associated monitoring data; S23. Integrate the monitoring data of all nodes in a single floor to form a floor data table, set data monitoring strategy and data storage structure, and build a building database.

3. A database-based building comprehensive early warning method according to claim 2, characterized in that: The steps of obtaining actual monitoring data of known nodes, establishing a spatial autoregressive model, predicting associated monitoring data of unknown nodes, and verifying the legitimacy of the associated monitoring data include the following steps: S221, define the spatial distance weight coefficient component and the self-fitting coefficient between any two nodes in the node network, if any two nodes are different nodes, the self-fitting coefficient is 1, if any two nodes are the same nodes, the self-fitting coefficient is 0; S222, using the spatial distance between nodes as network input and the spatial weight coefficient component as network output, establishing a spatial neural network, and performing nonlinear fitting on the spatial weight coefficient component; S223, taking the product of the spatial weight coefficient component after nonlinear fitting and the self-fitting coefficient as the spatial weight coefficient, and establishing a spatial weight matrix; S224, multiplying the monitoring data of the known nodes by the spatial weight matrix, constructing a spatial autoregressive model, and setting a training set and a validation set to perform anti-fitting training on the spatial autoregressive model; S225. Based on the spatial autoregressive model, known nodes are used to predict associated monitoring data of adjacent unknown nodes, and the legitimacy of all associated monitoring data of the unknown nodes is verified.

4. A database-based building comprehensive early warning method according to claim 3, characterized in that: The setting of the training set and the validation set to perform anti-fitting training on the spatial autoregressive model comprises the following steps: S2241, dividing the actual monitoring data into a training set and a verification set; S2242, monitoring the mean square error of the validation set, and terminating the training in advance when the mean square error of the training set continues to decrease and the mean square error of the validation set begins to increase; S2243. Using regularization technology, randomly hiding some neurons in the spatial neural network, and introducing a learnable activation function for each layer in the spatial neural network; S2244, use batch normalization technology to accelerate the model training process and improve stability.

5. The database-based comprehensive building early warning method according to claim 3 is characterized in that: The method of using known nodes to predict the associated monitoring data of adjacent unknown nodes and verifying the legitimacy of all associated monitoring data of the unknown nodes includes the following steps: S2251, setting the association radius of the known nodes, drawing a circle with the association radius as the prediction range of the known nodes, and using the spatial regression model to calculate the association monitoring data of the unknown nodes within the prediction range; S2252. Count all associated monitoring data of each unknown node predicted by adjacent known nodes, calculate the mean difference of all associated monitoring data, if the mean difference is less than the preset mean difference threshold, determine that the associated monitoring data is legal, and calculate the average value as the only associated monitoring data; if the mean difference is greater than the preset mean difference threshold, determine that the associated monitoring data is illegal, feedback the illegal information, eliminate the outliers in the associated monitoring data, and then calculate the average value as the only associated monitoring data.

6. A database-based building comprehensive early warning method according to claim 2, characterized in that: The integration of monitoring data of all nodes in a single floor to form a floor data table, and setting data monitoring strategy and data storage structure to construct a building database includes the following steps: S231, configuring a dynamic performance view of monitoring data, and establishing floor data tables and relationships between tables for storing different types of monitoring data to form a monitoring data dictionary; S232, for each floor data table corresponding to each floor, a customized monitoring method is performed to monitor different operations of different floor data tables; S233. Construct floor data tables and table structures for different floors in the building to form a building database.

7. A database-based building comprehensive early warning method according to claim 2, characterized in that: The method of regularly updating the monitoring data of all floors in the building database, identifying potential risks in each floor based on the comprehensive early warning model, and outputting safety early warning notifications includes the following steps: S31. Based on the set monitoring frequency, obtain the monitoring data of each floor in the building database; S32. Set up an early warning system based on the type of building safety risks and establish a comprehensive early warning model; S33, based on the comprehensive early warning model, analyzing the potential risks of all nodes in a single floor, and locating the risks according to the risk level and location information of each node; S34. Based on the node risk identification result, output a safety warning notification including location information.

8. The database-based comprehensive building early warning method according to claim 7 is characterized in that: The method of setting up an early warning system and establishing a comprehensive early warning model based on the building safety risk type includes the following steps: S321. Set the types of security risks existing in the building, select corresponding types of monitoring data as risk warning indicators of the security risk types, and establish a building risk warning indicator system and warning level; S322, setting the warning level critical point of the risk warning indicator, and establishing a risk degree calculation formula based on the warning level critical point to calculate the risk degree of each risk warning indicator; S323. Obtain the risk calculated from historical monitoring data as sample data to form training samples and test samples, build a comprehensive early warning model, and conduct training and testing to obtain optimal model parameters.

9. A database-based building comprehensive early warning method according to claim 8, characterized in that: The risk calculation formula is: In the formula, y i Indicates i The risk level of each risk warning indicator; x i Indicates i The value of a risk warning indicator; k Indicates the warning level number; x i(k) and x i(k+1) Indicates i The critical point of the warning level of each risk warning indicator; y i(k) and y i(k+1) Indicates x i(k) and x i(k+1) The warning level vector of the risk warning indicators.

10. A database-based building comprehensive early warning system, used to implement the database-based building comprehensive early warning method according to any one of claims 1 to 9, characterized in that: The system includes: The monitoring module is used to install distributed monitoring sensors on each floor of the building, and to establish a coordinate system based on the plane where each floor is located, and to record the position coordinates of each monitoring sensor on the floor to which it belongs; The database module is used to divide the node grids in the coordinate systems of each layer, and analyze the associated monitoring data of unknown nodes based on the actual monitoring data collected by known nodes to build a building database; The early warning identification module is used to regularly update the monitoring data of all floors in the building database, identify potential risks on each floor based on the comprehensive early warning model, and output safety early warning notifications; Verification decision module, used to verify the accuracy of safety warning notifications, and output building management decisions after verification; Among them, the monitoring module, the database module, the early warning identification module and the verification decision module are connected in sequence.

Images