Building collapse risk monitoring system for fire rescue

Abstract

A building collapse risk monitoring system for fire rescue belongs to the technical field of fire rescue, and comprises a front-end sensing unit and a rear-end control device which are connected with each other, the front-end sensing unit is used for collecting inclination angle / acceleration, surface strain and displacement data of a building, and the rear-end control device is used for controlling the front-end sensing unit. The rear-end control device comprises a central control module, a display module, a wireless transceiver module, a protocol conversion module, an interaction assembly and a power management module. According to the invention, the collected building collapse risk data is processed, displayed and alarmed, so that the personal safety level of a fire rescue team during task execution is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of fire fighting and rescue, and particularly relates to a building collapse risk monitoring system for fire fighting and rescue. Background Art

[0002] During the process of fire fighting and rescue, building collapse accidents may occur. Building collapses can be divided into various types according to the causes, including collapses caused by fires, earthquakes, explosions, etc. The characteristics of building collapses in fire fighting and rescue include strong suddenness, urgent time, large destructiveness, and complex on-site environments. Building collapses usually have no obvious omens. After the collapse occurs, rescue personnel need to respond within an extremely short time, otherwise serious consequences may be caused. Collapse incidents often result in a large number of casualties and property losses. There are usually a large amount of ruins and sundries at the scene, increasing the difficulty of rescue. Due to the suddenness and unpredictability of this accident, it also poses a great threat to the personal safety of the fire fighting and rescue teams.

[0003] At present, when fire fighting and rescue teams perform fire fighting and rescue tasks, the monitoring of building collapse risks mainly relies on visual observation by personnel, and there are no scientific and effective technical means to scientifically monitor building collapse risks. Summary of the Invention

[0004] In view of the above-mentioned disadvantages and deficiencies of the prior art, the present invention provides a building collapse risk monitoring system for fire fighting and rescue, which displays and alarms after processing the collected data on building collapse risks, and improves the personal safety level of fire fighting and rescue teams when performing tasks.

[0005] To achieve the above object, the main technical solutions adopted by the present invention include:

[0006] A building collapse risk monitoring system for fire fighting and rescue, including a front-end sensing unit and a back-end control device connected to each other. The front-end sensing unit is used to collect building inclination / acceleration, surface strain, and displacement data. The back-end control device includes a central control module, a display module, a wireless transceiver module, a protocol conversion module, an interaction component, and a power management module. The central control module is connected to the front-end sensing unit through the wireless transceiver module, and is used to obtain and process the data of the front-end sensing unit and send instructions to the front-end sensing unit. The central control module is connected to the protocol conversion module, and the protocol conversion module is connected to the interaction component. The central control module is used to control the protocol conversion module to parse the instructions of the interaction component, and the interaction component is used for human-computer interaction. The central control module is connected to the display module and is used to control the data display of the display module. The power management module is connected to the display module, the wireless transceiver module, the protocol conversion module, and the interaction component, and is used to supply power to each module.

[0007] Further, the front-end sensing unit includes an inclination / acceleration sensing module, a surface strain sensing module, and a displacement sensing module.

[0008] Further, the data processing of the central control module for the front-end sensing unit includes time monitoring and space monitoring; the time monitoring includes monitoring according to the relationships between the parameters of the building inclination (I), acceleration (A), surface strain (S), and displacement (D) and time:

[0009] Inclination rate of change with time F I (t) = ΔI / Δt;

[0010] Acceleration rate of change with time F A (t) = ΔA / Δt;

[0011] Surface strain rate of change with time F S (t) = ΔS / Δt;

[0012] Displacement rate of change with time F D (t) = ΔD / Δt;

[0013] Where: ΔI represents the change in the building inclination within the time interval Δt; ΔA is the change in acceleration within the time Δt; ΔS is the change in surface strain within the time Δt; ΔD is the change in displacement within the time Δt;

[0014] Set data membership degrees: small rate of change, medium rate of change, large rate of change, where small rate of change is 0 ≤ F(t) ≤ 20%, medium rate of change is 20% < F(t) ≤ 80%, large rate of change is F(t) > 80%, where F(t) is the data rate of change; when the rate of change of any one or more of the four parameters of inclination, acceleration, surface strain, and displacement is within the membership degree of "small rate of change", the building collapse risk monitoring is in a "stable" state and no state transition occurs; when the rate of change of any one or more of the four parameters is within the membership degree of "medium rate of change", the building collapse risk monitoring is in a "warning" state and the state transfers to the space monitoring step; when the rate of change of any two or more of the four parameters is within the membership degree of "large rate of change", the building collapse risk monitoring is in a "collapse" state and an alarm is issued;

[0015] Space monitoring includes:

[0016] 1) Construct a self-organizing neural network: The input vector of the data input layer is input = (I, A, S, D) T , the number of neurons in the feature extraction layer is 16, and the decision output layer has three neurons, corresponding to the outputs of these three states respectively; for each neuron in the feature extraction layer and the decision output layer, randomly initialize its weight vector W j, the dimension of the weight vector is the same as that of the input vector, and its value range is within the interval [-0.1, 0.1], and the network connection is constructed;

[0017] 2) Preliminary machine learning

[0018] Calculate the Euclidean distance: For each neuron in the mapping layer, calculate the Euclidean distance Distance between its weight vector W j and the input vector input j = ||input - W j ||;

[0019] Determine the winning neuron: Find the neuron with the shortest Euclidean distance as the winning neuron, and mark its adjacent neurons;

[0020] Update and adjust the weights of the adjacent neurons using the neighborhood function; Update the weights using the following formula:

[0021] w ij (t + 1) = w ij (t) + η(t)·h j* (t)·(input i (t) - w ij (t))

[0022] where, is the connection weight from the input node i to the neuron j at the moment of w ij (t + 1); w ij (t) is the connection weight at time t; η(t) is the learning rate; input i (t) is the i-th component of the input vector at time t; h j* (t) is the neighborhood function value;

[0023] Adjust the radius of the neighborhood function using the following formula:

[0024]

[0025] where, σ(t) is the radius of the neighborhood function; σ0 is the initial neighborhood radius; t is the current training step; τ is the time constant;

[0026] Select the data of the building in the normal state, warning state and collapsed state for training to obtain a self-organizing neural network for judging the collapse of the building;

[0027] 3) Input the inclination angle (I), acceleration (A), surface strain (S), and displacement (D) data of the time monitoring in the warning state into the trained self-organizing neural network, and output one of the three states of stable, warning and collapsed.

[0028] Further, the central control module includes an industrial control computer U1, and the display module includes a video data protocol conversion chip U3 and a liquid crystal display module U4. The video data protocol conversion chip U3 is connected to the liquid crystal display module U4 for video data protocol conversion; the industrial control computer U1 is connected to the video data protocol conversion chip U3 for the transmission of display data; the industrial control computer U1 is connected to the liquid crystal display module U4 for instruction interaction with the central control module. The industrial control computer U1, the video data protocol conversion chip U3, and the liquid crystal display module U4 are connected to display the data transmitted by the central control module.

[0029] Further, the wireless transceiver module includes a USB bus transfer chip U8, an RS-485 transceiver chip U9, and a Lora wireless module U10. The USB bus transfer chip U8 is connected to the industrial control computer U1 of the central control module for USB protocol data transmission; the USB bus transfer chip U8 is connected to the RS-485 transceiver chip U9 for data exchange; the RS-485 transceiver chip U9 is connected to the Lora wireless module U10 for RS-485 data transmission; the Lora wireless module U10 is used for wireless communication with the front-end sensing unit.

[0030] Further, the protocol conversion module includes a microcontroller chip U11, and the interaction components include relay modules U14, U13, and a push-button switch S3. The microcontroller chip U11 is connected to the industrial control computer U1 of the central control module for external input / output instruction data interaction; the microcontroller chip U11 is connected to the relay module U14 for early warning sound signal data transmission; the microcontroller chip U11 is connected to the relay module U13 for normal and early warning status indication data transmission; the microcontroller chip U11 is connected to the connection switch S3 for start signal input; the microcontroller chip U11 is connected to a start configuration circuit composed of a straight plug connector U12 and a resistor R3.

[0031] Further, the microcontroller chip U11 is also connected to a power supply indication circuit, a clock circuit, a reset circuit, and filter capacitors.

[0032] Further, the relay module U13 is also connected to a normal status indicator light and an early warning status indicator light; the relay module U14 is connected to a buzzer.

[0033] Furthermore, the power management module includes voltage conversion chips U2, U6, U7, and an external power access connector U5. The voltage conversion chip U2 is connected to the external power access connector U5 via a power switch S1 for switching the power supply. The voltage conversion chip U2 is used to convert 24V voltage to 12V voltage; the voltage conversion chip U6 is used to convert 12V voltage to 5V voltage; the voltage conversion chip U7 is used to convert 12V voltage to 3.3V voltage; the voltage powers the display module, wireless transceiver module, protocol conversion module, and interaction component after passing through filter capacitors.

[0034] The beneficial effects of the present invention are:

[0035] The building collapse risk monitoring system for fire rescue of the present invention can evaluate the building collapse risk by means of technical data monitoring. The method proposed by the present invention can analyze the building collapse risk by relying on the data of the front-end building collapse risk monitoring sensing unit. At the same time, the device proposed by the present invention can parse and display data, monitor and give early warnings, and control commands. Compared with the traditional method of personnel observation, the technical effects achieved by the device proposed by the present invention are more scientific, with rich functions and convenient use, providing technical support for the fire rescue team to monitor the building collapse risk during the execution of rescue tasks and improving the personal safety level of the fire rescue team during the execution of tasks. Description of the Drawings

[0036] Figure 1 It is a schematic diagram of the building collapse risk monitoring system for fire rescue of the present invention;

[0037] Figure 2 It is a schematic diagram of the inclination / acceleration sensing module;

[0038] Figure 3 It is a schematic diagram of the surface strain sensing module;

[0039] Figure 4 It is a schematic diagram of the displacement sensing module;

[0040] Figure 5 It is a schematic diagram of the central control module;

[0041] Figure 6 It is a schematic diagram of the display module;

[0042] Figure 7 It is a schematic diagram of the wireless transceiver module;

[0043] Figure 8 It is a schematic diagram of the protocol conversion module;

[0044] Figure 9 It is a schematic diagram of the interaction component module;

[0045] Figure 10It is a schematic diagram of the power management module. Detailed implementation manners

[0046] For better explaining the present invention for easy understanding, the present invention will be described in detail below in conjunction with the accompanying drawings through specific implementation manners.

[0047] The present invention provides a building collapse risk monitoring system for fire rescue, as Figure 1 shown. The system includes a front-end sensing unit and a back-end control device. The back-end control device of the building collapse risk monitoring system of the present invention can collect data of the front-end building collapse risk monitoring sensing unit through a wireless data transmission method. At the same time, the device can analyze and display the data, conduct building collapse monitoring and early warning. At the same time, the device can also perform command control on the building collapse risk monitoring sensing unit; the front-end building collapse risk monitoring sensing unit is mainly responsible for collecting physical parameters such as building inclination, acceleration, surface strain, and displacement; the central control module running in the back-end control device processes and analyzes the physical parameters such as building inclination, acceleration, surface strain, and displacement collected by the front-end building collapse risk monitoring sensing unit, and then monitors and warns of the building collapse risk.

[0048] The front-end sensing unit includes an inclination / acceleration sensing module, a surface strain sensing module, and a displacement sensing module; the inclination / acceleration sensing module mainly measures building inclination and acceleration data information and sends it to the back-end control device through a wireless signal; the surface strain sensing module mainly measures building surface strain data information and sends it to the back-end control device through a wireless signal; the displacement sensing module mainly measures building displacement data information and sends it to the back-end control device through a wireless signal.

[0049] The back-end control device includes a central control module, a display module, a wireless transceiver module, a protocol conversion module, an interaction component, and a power management module; the central control module is the control center of the entire device, controlling the wireless transceiver module to perform wireless data transceiver in the air, controlling the protocol conversion module to perform command parsing of the interaction component, and controlling the display module to display data; the display module is the display part of the device, and displays the data received by the wireless transceiver module after being processed by the central control module; the wireless transceiver module is the wireless communication part of the device, responsible for receiving the wireless data sent by the front-field building collapse risk monitoring sensing unit into the air, and sending the commands of the central control module to the front-field building collapse risk monitoring sensing unit; the protocol conversion module is the protocol processing part of the device, responsible for protocol conversion between the external interaction component and the central control module; the interaction component is the external input / output part, used for the man-machine interaction function of the system; the power management module is responsible for the change of different voltages of the device and supplies power to each module of the device.

[0050] The tilt / acceleration sensing module of the front-end sensing unit, such as Figure 2 shown, includes U21, U15, and U18. U21 is a tilt / acceleration composite sensor, model SINCT-485. Its pin 1 is connected to pin 3 of U15 and pin 1 of U18, which is the power supply pin with a supply voltage of 12V. Pins 2 and 3 of U21 are connected to pins 1 and 2 of U15, which are RS-485 data pins for data transmission. Pin 4 of U21 is connected to pin 4 of U15 and pin 2 of U18, which is the GND signal. U15 is a Lora wireless module, model TAS-Lora-189F. Its pins 1 and 2 are connected to pins 2 and 3 of U21, its pin 3 is connected to pin 1 of U21 and pin 1 of U18, and its pin 4 is connected to pin 4 of U21 and pin 2 of U18. U18 is a lithium-ion battery pack, XH2.54, with a supply voltage of 12V. Its pin 1 is connected to pin 1 of U21 and pin 3 of U15, its pin 2 is connected to pin 4 of U21 and pin 4 of U15, and U18 supplies power to U21 and U15.

[0051] The surface strain sensing module of the front-end sensing unit, such as Figure 3 shown, includes U22, U16, and U19. U22 is a surface strain sensor, model RS-ZX / ZXT-BMYB-1. Its pin 1 is connected to pin 3 of U16 and pin 1 of U19, which is the power supply pin with a supply voltage of 12V. Pins 2 and 3 of U22 are connected to pins 1 and 2 of U16, which are RS-485 data pins for data transmission. Pin 4 of U22 is connected to pin 4 of U16 and pin 2 of U19, which is the GND signal. U16 is a Lora wireless module, model TAS-Lora-189F. Its pins 1 and 2 are connected to pins 2 and 3 of U22, its pin 3 is connected to pin 1 of U22 and pin 1 of U19, and its pin 4 is connected to pin 4 of U22 and pin 2 of U19. U19 is a lithium-ion battery pack, XH2.54, with a supply voltage of 12V. Its pin 1 is connected to pin 1 of U22 and pin 3 of U16, its pin 2 is connected to pin 4 of U22 and pin 4 of U16, and U19 supplies power to U22 and U16.

[0052] The displacement sensing module of the front-end sensing unit, such as Figure 4As shown, it includes U23, U17, and U20. U23 is a displacement sensor with the model number JXBS-LGS. Its pin 1 is connected to pin 3 of U17 and pin 1 of U20, which is the power supply pin with a supply voltage of 12V. Pin 2 and pin 3 of U23 are connected to pin 1 and pin 2 of U17, which are RS-485 data pins for data transmission. Pin 4 of U23 is connected to pin 4 of U17 and pin 2 of U20, which is the GND signal. U17 is a Lora wireless module with the model number TAS-Lora-189F. Its pin 1 and pin 2 are connected to pin 2 and pin 3 of U23. Its pin 3 is connected to pin 1 of U23 and pin 1 of U20. Its pin 4 is connected to pin 4 of U23 and pin 2 of U20. U20 is a lithium-ion battery pack, XH2.54, with a supply voltage of 12V. Its pin 1 is connected to pin 1 of U23 and pin 3 of U17. Its pin 2 is connected to pin 4 of U23 and pin 4 of U17. U20 supplies power to U23 and U17.

[0053] The central control module U1 of the backend control device, as Figure 5 shown, is an embedded industrial computer with the model number ARK-6322. Its pin 1 is connected to 12VCC, and pin 2 is connected to GND to supply power to the module. Pins 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 of U1 are respectively connected to pins 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 of U3 correspondingly to achieve the transmission of display data. Pins 25 and 26 of U1 are respectively connected to pins 7 and 8 of U8 correspondingly to achieve the transmission of wireless communication data. Pins 27, 28, 29, 39 of U1 are respectively connected to pins 25, 24, 22, 23 of U4 respectively for instruction interaction with the display module. Pins 31 and 32 of U1 are respectively connected to pins 48 and 47 of U11 correspondingly to achieve data transceiver with the protocol conversion module.

[0054] The display module of the backend control device, as Figure 6As shown in the figure, it includes two parts, U3 and U4. U3 is a video data protocol conversion chip with the model number CS5801. Pins 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 of U3 are respectively connected to pins 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 of U1. Pins 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 of U3 are respectively connected to pins 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 of U4 for video data protocol conversion. U4 is a liquid crystal display module with the model LOBOROBOT13.3. Pins 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 of U4 are respectively connected to pins 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 of U3 to display the data transmitted by the central control module. Pins 25, 24, 22, 23 of U4 are respectively connected to pins 27, 28, 29, 39 of U1 for instruction interaction with the central control module.

[0055] The wireless transceiver module of the backend control device, such as Figure 7As shown in the figure, the wireless transceiver module includes parts U8, U9, and U10. U8 is a USB bus transfer chip with the model number CH341T. Its pins 7 and 8 are respectively connected to pins 25 and 26 of U1 to achieve USB protocol data transmission; pin 1 of U8 is connected in series to 5VCC through LED indicator L1 and resistor R1 as the chip power supply indicator; pins 3 and 4 of U8 are respectively connected to pins 4 and 1 of U9 for data exchange; pin 6 of U8 is connected to 3.3VCC; pins 9 and 10 of U8 are connected to the crystal oscillator Q1, capacitor C20, and capacitor C21 to form a clock circuit; pins 13 and 20 of U8 are connected to 5VCC; pins 11, 12, and 19 of U8 are connected to GND; pin 18 of U8 is connected to pins 2 and 3 of U9 for signal enabling; U9 is an RS-485 transceiver chip with the model number MAX487. Its pins 4 and 1 are respectively connected to pins 3 and 4 of U8, its pins 2 and 3 are connected to pin 18 of U8, its pin 8 is connected to 5VCC, its pin 5 is connected to GND, and its pins 6 and 7 are respectively connected to pins 1 and 2 of U10 for RS-485 data transmission; U10 is a Lora wireless module with the model number TAS-Lora-189F. Its pins 1 and 2 are respectively connected to pins 6 and 7 of U9 for data interaction. Its pin 3 is connected to 12VCC for power supply, and its pin 4 is connected to GND; capacitor C19 is a 3.3VCC filter capacitor, capacitors C22, C24, C34, and C35 are 5VCC filter capacitors, and capacitor C36 is a 12VCC filter capacitor.

[0056] The protocol conversion module of the back-end control device, such as Figure 8 As shown in the figure, the protocol conversion module includes parts U11, U12, etc. U11 is a microcontroller chip with the model number STM32F103V8. Its pins 47 and 48 are respectively connected to pins 32 and 31 of U1 for external input / output instruction data interaction. Its pin 57 is connected to the power supply indication circuit composed of indicator L2 and resistor R4. Its pin 61 is connected to pin 4 of U14 for early warning sound signal data transmission. Its pin 62 is connected to pin 4 of U13 for normal state indication data transmission. Its pin 66 is connected to pin 3 of U13 for early warning state indication data transmission. Its pin 72 is connected to switch S3 for start signal input. Its pin 94 is connected to the start configuration circuit composed of through-hole connector U12 and resistor R3 for initial bootstrapping. Its pins 12 and 13 are connected to the crystal oscillator X2, capacitor C23, and capacitor C25 to form a clock circuit as the signal clock source. Its pin 14 is connected to the reset circuit composed of resistor R2, switch S2, and capacitor C18. Its pins 11, 21, 22, 28, 50, 75, and 100 are connected to the 3.3VCC signal. Its pins 10, 19, 20, 27, 49, 74, and 99 are connected to the GND signal. Capacitors C26, C27, C28, C29, C30, C31, and C32 are 3.3VCC signal filter capacitors.

[0057] The interaction components of the backend control device, such as Figure 9 shown in the figure. The interaction components include U13, U14, S3, etc. U13 and U14 are relay modules with the model BMZ02R1-E. The pin 4 of U14 is connected to the pin 61 of U11. The pins 7 and 8 of U14 are connected to the buzzer LS1 (model AD101-22SM). The pin 1 of U14 is connected to 12VCC, and the pin 2 of U14 is connected to GND. The pin 4 of U13 is connected to the pin 62 of U11. The pin 3 of U13 is connected to the pin 66 of U11. The pins 5 and 6 of U13 are connected to the normal status indicator D2 (model XA2EVB3LC, green). The pins 7 and 8 of U13 are connected to the warning status indicator D1 (model XA2EVB4LC, red). The pin 1 of U13 is connected to 12VCC, and the pin 2 of U13 is connected to GND. U13 and U14 implement the signal conversion function. S3 is a push-button switch with the model LAY39B-LA38-11BNZSD-G, and its pins are connected to the pin 72 of U11 as the input of the start signal.

[0058] The power management module of the backend control device, such as Figure 10 shown in the figure. The power management module includes U2, U5, U6, U7. U5 is an external power access connector with the model GX20-4 core, mainly implementing the function of accessing the external 24V voltage signal. S1 is a power switch with the model LAY39B-LA38-11BNZSD-R-AC / DC24V. One end of its pin is connected to the pin 2 of U5, and the other end is connected to the pin 3 of U2, realizing the function of the power switch. U2 is a voltage conversion chip with the model URB4812LD-30WR3. Its pins 2 and 5 are connected to GND, the pin 3 is connected to the switch S1, and the pin 4 outputs a 12VCC voltage signal, realizing the function of converting the voltage from 24V to 12V. U6 is a voltage conversion chip with the model URB2405LD-30WR3. Its pins 2 and 5 are connected to GND, the pin 3 is connected to the 12VCC voltage signal, and the pin 4 outputs a 5V voltage signal, realizing the function of converting the voltage from 12V to 5V. U7 is a voltage conversion chip with the model URB2403LD-30WR3. Its pins 2 and 5 are connected to GND, the pin 3 is connected to the 12VCC voltage signal, and the pin 4 outputs a 3.3V voltage signal, realizing the function of converting the voltage from 12V to 3.3V. Capacitors C1, C2, C3, C4 are the filtering capacitors of 24VCC, filtering out the voltage signal clutter. Capacitors C5, C6, C7, C8, C9 are the filtering capacitors of 12VCC, filtering out the voltage signal clutter. Capacitors C10, C11, C12, C13 are the filtering capacitors of 5VCC, filtering out the voltage signal clutter. Capacitors C14, C15, C16, C17 are the filtering capacitors of 3.3VCC, filtering out the voltage signal clutter.

[0059] The building collapse risk monitoring system running in the backend control device adopts a comprehensive monitoring method for building collapse risk based on the comprehensive analysis of spatio-temporal data and machine learning through the fusion of sensing data. This method runs in the backend control device and deeply analyzes the inclination (I), acceleration (A), surface strain (S), and displacement (D) parameters of the building to comprehensively monitor the building collapse risk during the fire rescue process.

[0060] The specific steps of the monitoring method are as follows:

[0061] (I) Data acquisition and preprocessing

[0062] 1. Sensor deployment

[0063] Sensors for measuring the inclination (I), acceleration (A), surface strain (S), and displacement (D) are deployed at key parts of the building, such as load-bearing walls, beam-column joints, and foundations. These sensors can obtain the information of the corresponding physical parameters of the building in real time and accurately, providing basic data for subsequent analysis.

[0064] 2. Data preprocessing

[0065] The raw data collected usually contains noise and interference information and needs to be preprocessed. The Kalman filtering algorithm is used to denoise the data. This method can effectively estimate and correct the measurement error according to the dynamic characteristics of the data, improving the accuracy and reliability of the data.

[0066] (II) Time monitoring steps

[0067] 1. Parameter definition and status setting

[0068] The inclination (I), acceleration (A), surface strain (S), and displacement (D) parameters of the building are used to characterize its collapse risk situation. The state of the building is divided into three types: stable, warning, and collapse. This more detailed state division helps to detect potential dangers earlier and gain time to take corresponding measures.

[0069] 2. Construct the time-varying function

[0070] A function of each parameter with respect to time is constructed:

[0071] Function of the rate of change of inclination with time: F I (t) = ΔI / Δt, where ΔI represents the change in the inclination of the building within the time interval Δt. This function reflects the rate of change of the building's inclination with time.

[0072] Function of the rate of change of acceleration with time: F A (t) = ΔA / Δt, where ΔA is the change in acceleration within the time Δt. This function is used to measure the change of acceleration with time.

[0073] Surface strain rate of change function: F S (t) = ΔS / Δt, where ΔS is the change in surface strain within the time period Δt. It reflects the trend of surface strain changing over time.

[0074] Displacement rate of change function: F D (t) = ΔD / Δt, where ΔD is the change in displacement within the time period Δt. This function reflects how fast the displacement changes over time.

[0075] 3. Threshold dynamic judgment and step transfer

[0076] The following is the specific process for evaluating the change trends of various parameters (tilt angle I, acceleration A, surface strain S, displacement D) in the monitoring of building collapse risk:

[0077] (1) Data membership analysis

[0078] Define three data memberships: "small rate of change", "medium rate of change", and "large rate of change". When the data rate of change is between 0 - 20% (including 20%), it is "small rate of change"; when the data rate of change is between 20 - 80% (including 80%), it is "medium rate of change"; when the data rate of change is above 80%, it is "large rate of change".

[0079] (2) Threshold judgment and status setting

[0080] When the rate of change of any one or more of the four parameters is in the membership of "small rate of change", the building collapse risk monitoring is in a "stable" state and there is no state transfer; when the rate of change of any one or more of the four parameters is in the membership of "medium rate of change", the building collapse risk monitoring is in a "warning" state and the state transfers to the space monitoring step; when the rate of change of any two or more of the four parameters (including two) is in the membership of "large rate of change", the building collapse risk monitoring is in a "collapse" state and there is no state transfer.

[0081] (3) Relationship between status and monitoring steps

[0082] When entering the stable state, there is no state transfer; when entering the warning state, it transfers to the space monitoring step to further evaluate the building collapse risk; when it is determined to be in the collapse state, it no longer transfers to the space monitoring step but immediately issues an alarm to prompt relevant personnel to take emergency avoidance and other measures.

[0083] (III) Space monitoring step

[0084] 1. Parameter and status description

[0085] During the space monitoring stage, the collapse risk situation of the building is still characterized by parameters such as the inclination angle (I), acceleration (A), surface strain (S), and displacement (D) of the building. The building state is divided into three types: stable, warning, and collapse.

[0086] 2. Construct a self-organizing neural network

[0087] Construct a self-organizing neural network. The network structure adopts a hierarchical design. The bottom layer is the data input layer, the middle layer is the feature extraction layer, and the top layer is the decision output layer. The specific steps are as follows:

[0088] (1) Data input layer: The input vector of this network is input=(I,A,S,D) T , that is, the four parameters of the inclination angle, acceleration, surface strain, and displacement of the building are input into the network as a vector.

[0089] (2) Feature extraction layer: The main function of this layer is to reduce the dimension and extract features of the input data. First, calculate the covariance matrix of the input data. By solving the eigenvalues and eigenvectors of the covariance matrix, sort them according to the size of the eigenvalues, and select the eigenvectors corresponding to the top 16 larger eigenvalues as the principal components. Project the input data onto these principal components to achieve data dimensionality reduction and obtain more representative features. The number of neurons in the feature extraction layer is 16.

[0090] (3) Decision output layer: The output layer is used to give the final decision result, that is, the state judgment of the building (stable, warning, and collapse three states). Therefore, 3 neurons are set, corresponding to the outputs of these three states respectively.

[0091] (4) Initialize the neuron weights

[0092] For each neuron in the mapping layer (feature extraction layer and decision output layer), randomly initialize its weight vector W j . The dimension of the weight vector is the same as the dimension of the input vector, and the value range is within the interval [-0.1,0.1] to ensure that the network has a certain degree of randomness and plasticity in the initial stage of training.

[0093] (5) Construct network connections

[0094] Establish a full connection between the input layer and the feature extraction layer, that is, each neuron in the input layer is connected to each neuron in the feature extraction layer, and the connection weight is the initialized weight. A full connection is also established between the feature extraction layer and the decision output layer to transmit the data after feature extraction processing, and finally the decision output layer makes a state judgment and output.

[0095] 3. Preliminary machine learning

[0096] (1) Calculate the Euclidean distance: For each neuron in the mapping layer, calculate the Euclidean distance Distance between its weight vector W j and the input vector input j = ||input - W j ||. The Euclidean distance is a commonly used method to measure the similarity between two vectors. The smaller the distance, the more similar the two vectors are.

[0097] Determine the winning neuron: Find the neuron with the shortest Euclidean distance as the winning neuron and mark its adjacent neurons. To improve the robustness of the network, a neighborhood function is introduced to adjust the weight update of adjacent neurons, and the radius of the neighborhood function gradually decreases as the training progresses.

[0098] In the self-organizing neural network, after determining the winning neuron, the weight of adjacent neurons is updated and adjusted through the following steps:

[0099] (2) Selection of neighborhood function

[0100] Adopt the Gaussian function as the neighborhood function because the Gaussian function has good smoothness and locality characteristics and is very suitable for describing the influence degree of neurons within the neighborhood. The general form of the Gaussian neighborhood function is:

[0101]

[0102] where h j* (t) is the neighborhood function value of the j-th neuron relative to the winning neuron (denoted as j*) at time t; d j* is the distance between the j-th neuron and the winning neuron; σ(t) is the radius of the neighborhood function.

[0103] (3) Weight update formula

[0104] During the weight correction process, not only the weight of the winning neuron is updated, but the weights of its adjacent neurons are also adjusted according to the neighborhood function. The weight update formula is as follows:

[0105] w ij (t + 1) = w ij (t) + η(t)·h j* (t)·(input i (t) - w ij (t))

[0106] where w ij (t + 1) is the connection weight from the input node i to the neuron j at time t + 1; w ij(t) is the connection weight at time t; η(t) is the learning rate, which is a parameter varying with the training time, with a value range between (0, 1) and gradually decreasing as the training progresses, used to control the step size of each weight update; input i (t) is the i-th component of the input vector at time t; h j* (t) is the value of the neighborhood function.

[0107] (4) The radius is adjusted during training

[0108] As the training progresses, the radius σ(t) of the neighborhood function gradually decreases, which is to enable the network to transition from a coarse-grained adjustment of a larger range of neurons to a fine-grained adjustment of the winning neuron and its very close neighboring neurons. The exponential decay method is used to adjust the radius, and the formula is as follows:

[0109]

[0110] Among them, σ0 is the initial neighborhood radius; t is the current training step; τ is the time constant.

[0111] In the above way, 200 pieces of data of the building in the normal state, warning state, and collapse state are selected for training. During the self-organizing neural network training process, the weights of adjacent neurons are updated and adjusted according to the neighborhood function, gradually optimizing the performance of the network and improving its accuracy and robustness in monitoring the building collapse risk.

[0112] 4. Spatial monitoring judgment output

[0113] The inclination (I), acceleration (A), surface strain (S), and displacement (D) data in the warning state of time monitoring are transferred to the self-organizing neural network after prior machine learning. After analysis and judgment by the self-organizing neural network, one of the three states of stable, warning, and collapse is output.

[0114] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Any modifications, alterations, substitutions, and variations made by those of ordinary skill in the art to the above embodiments fall within the scope of the present invention.

Claims

1. A building collapse risk monitoring system for fire rescue, characterized in that: It includes a front-end sensing unit and a back-end control device which are interconnected. The front-end sensing unit is used to collect building inclination / acceleration, surface strain, and displacement data. The back-end control device includes a central control module, a display module, a wireless transceiver module, a protocol conversion module, an interaction component, and a power management module. The central control module is connected to the front-end sensing unit through the wireless transceiver module and is used to obtain and process the data of the front-end sensing unit and send instructions to the front-end sensing unit. The central control module is connected to the protocol conversion module, and the protocol conversion module is connected to the interaction component. The central control module is used to control the protocol conversion module to parse the instructions of the interaction component, and the interaction component is used for human-computer interaction. The central control module is connected to the display module and is used to control the data display of the display module. The power management module is connected to the display module, the wireless transceiver module, the protocol conversion module, and the interaction component and is used to supply power to each module.

2. The building collapse risk monitoring system for fire rescue according to claim 1, wherein: The front-end sensing unit includes an inclination / acceleration sensing module, a surface strain sensing module, and a displacement sensing module.

3. A building collapse risk monitoring system for fire rescue according to claim 1, characterized in that: The data processing of the central control module for the front-end sensing unit includes time monitoring and space monitoring. Time monitoring includes monitoring according to the relationships between the building inclination (I), acceleration (A), surface strain (S), displacement (D) parameters and time: Tilt rate of change with time F I (t) = ΔI / Δt; Acceleration change rate with respect to time F A (t) = ΔA / Δt; Surface strain rate of change with time F S (t) = ΔS / Δt; Rate of change of displacement F with respect to time D (t) = ΔD / Δt; Where: ΔI represents the change in building inclination within the time interval Δt; ΔA is the change in acceleration within the time Δt; ΔS is the change in surface strain within the time Δt; ΔD is the change in displacement within the time Δt; Set data membership degrees: small change rate, medium change rate, large change rate. Among them, the small change rate is 0≤F(t)≤20%, the medium change rate is 20%<F(t)≤80%, and the large change rate is F(t)>80%, where F(t) is the data change rate. When the change rate of any one or more of the four parameters of inclination, acceleration, surface strain, and displacement is in the membership degree of "small change rate", the building collapse risk monitoring is in the "stable" state and no state transition occurs. When the change rate of any one or more of the four parameters is in the membership degree of "medium change rate", the building collapse risk monitoring is in the "early warning" state and the state transfers to the space monitoring step. When the change rate of any two or more of the four parameters is in the membership degree of "large change rate", the building collapse risk monitoring is in the "collapse" state and an alarm is issued; Space monitoring includes: 1) Construct a self-organizing neural network: The input vector of the data input layer is input=(I, A, S, D) T , the number of neurons in the feature extraction layer is 16, and three neurons are set in the decision output layer, corresponding to the outputs of these three states respectively; for each neuron in the feature extraction layer and the decision output layer, randomly initialize its weight vector W j , the dimension of the weight vector is the same as that of the input vector, and the value range is Build a network connection within the interval [-0.1, 0.1]; 2) Preliminary machine learning Calculate the Euclidean distance: For each neuron in the mapping layer, calculate the Euclidean distance Distance between its weight vector W j and the input vector input j = ||input - W j ||; Determine the winning neuron: Find the neuron with the shortest Euclidean distance as the winning neuron and mark its adjacent neurons; Use the neighborhood function to update and adjust the weights of the adjacent neurons. Update the weights using the following formula: w ij (t + 1)= w ij (t)+ η(t)·h j* (t)·(input i (t)- w ij (t)) Among them, it is at w ij (t + 1) is the connection weight from input node i to neuron j at time; w ij (t) is the connection weight at time t; η(t) is the learning rate; input i (t) is the i-th component of the input vector at time t; h j* (t) is the neighborhood function value; Adjust the radius of the neighborhood function using the following formula: Where, σ(t) is the radius of the neighborhood function; σ0 is the initial neighborhood radius; t is the current training step; τ is the time constant; Select data of the building in the normal state, early warning state, and collapse state for training to obtain a self-organizing neural network for judging building collapse; 3) Input the inclination (I), acceleration (A), surface strain (S), and displacement (D) data in the warning state of time monitoring into the trained self-organizing neural network, and output one of the three states: stable, warning, and collapse.

4. The building collapse risk monitoring system for fire rescue according to claim 1, characterized in that: The central control module includes an industrial computer U1. The display module includes a video data protocol conversion chip U3 and a liquid crystal display module U4. The video data protocol conversion chip U3 is connected to the liquid crystal display module U4 for video data protocol conversion. The industrial computer U1 is connected to the video data protocol conversion chip U3 for the transmission of display data. The industrial computer U1 is connected to the liquid crystal display module U4 for the instruction interaction with the central control module. The industrial computer U1, the video data protocol conversion chip U3, and the liquid crystal display module U4 are connected to display the data transmitted by the central control module.

5. The building collapse risk monitoring system for fire rescue according to claim 1, characterized in that: The wireless transceiver module includes a USB bus transfer chip U8, an RS-485 transceiver chip U9, and a Lora wireless module U10. The USB bus transfer chip U8 is connected to the industrial computer U1 of the central control module for USB protocol data transmission. The USB bus transfer chip U8 is connected to the RS-485 transceiver chip U9 for data exchange. The RS-485 transceiver chip U9 is connected to the Lora wireless module U10 for RS-485 data transmission. The Lora wireless module U10 is used for wireless communication with the front-end sensing unit.

6. The building collapse risk monitoring system for fire rescue according to claim 1, characterized in that: The protocol conversion module includes a microcontroller chip U11. The interaction components include relay modules U14, U13, and a push-button switch S3. The microcontroller chip U11 is connected to the industrial computer U1 of the central control module for external input / output instruction data interaction. The microcontroller chip U11 is connected to the relay module U14 for warning sound signal data transmission. The microcontroller chip U11 is connected to the relay module U13 for normal and warning state indication data transmission. The microcontroller chip U11 is connected to the connection switch S3 for start signal input. The microcontroller chip U11 is connected to the start configuration circuit composed of a direct plug connector U12 and a resistor R3.

7. The building collapse risk monitoring system for fire rescue according to claim 6, characterized in that: The microcontroller chip U11 is also connected to a power supply indication circuit, a clock circuit, a reset circuit, and filter capacitors.

8. The building collapse risk monitoring system for fire rescue according to claim 6, characterized in that: The relay module U13 is also connected to a normal state indicator light and a warning state indicator light. The relay module U14 is connected to a buzzer.

9. The building collapse risk monitoring system for fire rescue according to claim 6, characterized in that: The power management module includes voltage conversion chips U2, U6, U7, and an external power access connector U5. The voltage conversion chip U2 is connected to the external power access connector U5 through a power switch S1 for switching the power supply. The voltage conversion chip U2 is used to convert 24V voltage to 12V voltage. The voltage conversion chip U6 is used to convert 12V voltage to 5V voltage. The voltage conversion chip U7 is used to convert 12V voltage to 3.3V voltage. The voltage supplies power to the display module, the wireless transceiver module, the protocol conversion module, and the interaction components after passing through the filter capacitors.

Images