阅读说明：本技术 一种具备校自动准能力的古建筑受力监测预警系统及方法 (Ancient building stress monitoring and early warning system and method with calibration and automatic calibration capabilities ) 是由 陈传东 张建华 罗海波 于 2021-02-22 设计创作，主要内容包括：本发明专利涉及古建筑监测技术领域,尤其为一种具备校自动准能力的古建筑受力监测预警系统,包括MCU、压力传感器、通信模块、校准模块、控制模块、按键模块、警报模块和显示模块；本发明专利的传感器乙平时处于关闭状态极少使用,能够避免各种因素对其造成的影响,能够根据传感器乙示数作为标定数据重新训练BP神经网络,得到新的校准网络,修正了现有监测预警系统中的压力传感器由于设备老化、环境变化等因素造成的误差,使得读数更加精准,能够更加真实准确地反应古建筑的实际受力情况,从而更好地达到监测预警的作用；针对压力传感器输出数据的非线性特征,采用了拟合能力较强的BP神经网络作为校准算法,提高了校准精度。(The invention relates to the technical field of historic building monitoring, in particular to a historic building stress monitoring and early warning system with calibration and automatic calibration capabilities, which comprises an MCU (microprogrammed control unit), a pressure sensor, a communication module, a calibration module, a control module, a key module, an alarm module and a display module; the sensor B is in a closed state for extremely little use at ordinary times, so that the influence of various factors on the sensor B can be avoided, the BP neural network can be retrained according to the second index of the sensor as calibration data to obtain a new calibration network, the error of a pressure sensor in the existing monitoring and early warning system caused by factors such as equipment aging and environmental change is corrected, the reading is more accurate, the actual stress condition of an ancient building can be reflected more truly and accurately, and the monitoring and early warning effect is achieved better; aiming at the nonlinear characteristics of the output data of the pressure sensor, the BP neural network with strong fitting capability is adopted as a calibration algorithm, and the calibration precision is improved.)

1. The utility model provides an ancient building atress monitoring and early warning system that possesses automatic calibration ability which characterized in that: the device comprises an MCU (1), a pressure sensor (2), a communication module (3), a calibration module (4), a control module (5), a key module (6), an alarm module (7) and a display module (8);

the pressure sensor (2) comprises a first sensor (21) and a second sensor (22), the first sensor (21) is in a working state as a daily monitoring sensor in a working mode, the second sensor (22) is in a closed state as a standby sensor at ordinary times, errors caused by factors such as equipment aging and environmental changes are avoided, and the first sensor (21) and the second sensor (22) are in working states in a calibration mode;

the communication module (3) plays a role in sending and receiving data, the MCU (1) is used as a host, the sensor A (21) and the sensor B (22) are used as slaves, the sensor A (21) and the sensor B (22) have different addresses, the host sends a start signal, a corresponding slave is searched immediately after an address signal is transmitted, a reading or writing instruction is sent, after the slaves answer, the host and the slaves can transmit data, after each operation is finished, the slaves can generate an answer signal to judge whether the communication is successful, and after the data transmission is finished, the host sends an end signal;

the calibration module (4) is used for calibrating the data sent by the control module (5), the calibration module (4) is realized by adopting a BP neural network, the weight and the bias parameters of the BP neural network are updated according to the training result of the control module (5), the BP neural network consists of an input layer, a hidden layer and an output layer, the data is input from the input layer, is subjected to linear transformation processing of a weight value and a bias item, and is subjected to an activation layer to obtain the output of the hidden layer;

the key module (6) is provided with a switch key and a calibration key, the switch key controls the start and the stop of the system, the calibration key is pressed down, the control module (5) can receive a calibration request, and the system enters a calibration mode;

the alarm module (7) adopts an active buzzer to send out an alarm;

the display module (8) adopts an LCD screen, and when the LCD screen is in a calibration mode, the LCD screen displays: in system calibration; when the building is in the working mode, the LCD screen displays the number displayed by the pressure sensor (2) and the judgment result (the health condition of the historic building) of the control module (5);

the control module (5) is mainly responsible for normalization processing of data and coordination and operation of the whole system, when a preset time is reached or a calibration request is received, a calibration mode is entered, the control module (5) activates the sensor B (22) and receives the data all the time until enough data meeting certain requirements are obtained, then normalization processing is carried out on the data, the data obtained by the sensor A (21) is used as training data, the data obtained by the sensor B (22) is used as a training label, reverse training is carried out by utilizing a gradient descent method to obtain new network parameters and update a calibration network, then the sensor B (22) is closed, when the control module (5) is in a working mode, the control module (3) needs to receive the data of the sensor A (21) through the communication module (3), the data are sent to the calibration module (4) after normalization processing is carried out, the calibrated data are received and reverse normalization is carried out after calibration is finished, and finally, judging whether the data is normal or not, if the data is abnormal, controlling a display module (8) to display corresponding information, and giving an alarm by an alarm module (7).

2. The ancient building stress monitoring and early warning system with the calibration and automatic calibration capability of claim 1, wherein: the pressure sensor is characterized in that the MCU (1) is in communication connection with the communication module (3), the communication module (3) is in communication connection with the pressure sensor (2), and the MCU (1) is in communication connection with the pressure sensor (2) through the communication module (3).

3. The ancient building stress monitoring and early warning system with the calibration and automatic calibration capability of claim 1, wherein: the pressure sensor is characterized in that the MCU (1) adopts an STM32F103 series single chip microcomputer, and the pressure sensor (2) is divided into a working mode and a calibration mode.

4. The ancient building stress monitoring and early warning system with the calibration and automatic calibration capability of claim 1, wherein: the communication module (3) adopts an I2C bus and a serial bus consisting of a data line SDA and a clock SCL.

5. The ancient building stress monitoring and early warning system with the calibration and automatic calibration capability of claim 1, wherein: the structure from the hidden layer to the output layer is very similar to that from the input layer to the hidden layer, and a four-in-one structure BP neural network is adopted.

6. The ancient building stress monitoring and early warning system with the calibration and automatic calibration capability of claim 1, wherein: the control module (5) is electrically connected with the alarm module (7), and the key module (6) is electrically connected with the control module (5).

7. The utility model provides an ancient building atress monitoring and early warning method that possesses automatic calibration ability which characterized in that: the method comprises the following steps:

in the working mode:

the method comprises the following steps: the sensor A (21) sends data to the control module (5) through the communication module (3);

step two: the control module (5) normalizes the received data and sends the normalized data to the calibration module (4);

step three: the calibration module (4) inputs data into the trained neural network module for forward calculation to obtain an output result of a neural network output layer, and then the output result is sent into the control module (5);

step four: the control module (5) performs inverse normalization processing on the BP neural network output data to obtain calibrated output data;

step five: the control module (5) judges whether the data is normal or not, if the data is abnormal, the control display module (8) displays corresponding information, and the alarm module (7) gives an alarm.

In the calibration mode:

the method comprises the following steps: reaching a preset time or receiving a calibration request;

step two: the sensor A (21) and the sensor B (22) send data to the control module (5) through the communication module (3);

step three: the control module (5) normalizes the received data;

step four: the control module (5) takes the data obtained by the sensor A (21) as training data, takes the data obtained by the sensor B (22) as a training label, and carries out reverse training by using a gradient descent method to obtain new network parameters;

step five: and the calibration module (4) updates the weight and the offset value of the BP neural network and closes the sensor B (22).

Technical Field

The invention relates to the technical field of historic building monitoring, in particular to a historic building stress monitoring and early warning system and method with automatic calibration capability.

Background

The ancient Chinese architecture has a long history and deep cultural background and has higher artistic and cultural values. However, ancient buildings are increasingly damaged due to various reasons such as structural aging, environmental corrosion, artificial damage and the like. Therefore, the health state of the historic building is monitored according to the stress condition of the historic building, and the monitoring method is particularly important. However, the existing ancient building detection mainly uses professional manual survey as a main part, the workload is large, the efficiency is low, the problems are difficult to find timely and accurately, part of patents propose that related sensors are used for arranging the ancient building stress monitoring and early warning system to monitor the health condition of the ancient building, and part of patents propose that a calibration module is added to calibrate the sensors to improve the precision, but the existing scheme does not consider the reduction of various factors such as self equipment aging and external environment influence on the precision of the sensors, and due to the particularity of the ancient building, the related equipment is difficult to frequently update so as to avoid damaging the ancient building, the precision of the schemes can be gradually reduced along with the time, and even wrong judgment is given, so that the ancient building stress monitoring and early warning system and the method with the automatic calibration capability are provided to solve the problems.

SUMMARY OF THE PATENT FOR INVENTION

The invention aims to provide a historic building stress monitoring and early warning system with calibration and automatic calibration capabilities and a method thereof, which solve the problems that the existing historic building detection mainly takes professional manual survey as a main part, has large workload and low efficiency and is difficult to find problems timely and accurately, partial patents propose that related sensors are used for arranging the historic building stress monitoring and early warning system to monitor the health condition of the historic building, partial patents also propose that a calibration module is added to calibrate the sensors so as to improve the precision, but the existing proposal does not consider the reduction of the sensor precision caused by various factors such as self equipment aging, external environment influence and the like, due to the particularity of the historic building, the related equipment is difficult to be frequently updated so as to avoid damaging the historic building, the accuracy of these schemes can gradually decrease over time, even giving the problem of false positives.

In order to achieve the purpose, the invention provides the following technical scheme: an ancient building stress monitoring and early warning system with calibration and automatic calibration capabilities comprises an MCU (microprogrammed control unit), a pressure sensor, a communication module, a calibration module, a control module, a key module, an alarm module and a display module;

the pressure sensor comprises a sensor A and a sensor B, the sensor A is in a working state as a daily monitoring sensor in a working mode, the sensor B is in a closed state as a standby sensor at ordinary times, errors caused by factors such as equipment aging and environmental changes are avoided, and the sensor A and the sensor B are in working states in a calibration mode;

the communication module plays a role in sending and receiving data, the MCU is used as a host, the sensor A and the sensor B are used as slaves, the sensor A and the sensor B both have different addresses, the host sends an initial signal, the corresponding slave is searched next to a transmitted address signal, and a reading or writing instruction is sent, after the slaves respond, the host and the slaves can transmit data, after each operation is finished, the slaves can generate a response signal to judge whether the communication is successful, and after the data transmission is finished, the host sends an end signal;

the calibration module is used for calibrating the data sent by the control module, the calibration module is realized by adopting a BP neural network, the weight and the offset parameter of the calibration module are updated according to the training result of the control module, the BP neural network consists of an input layer, a hidden layer and an output layer, the data is input from the input layer, the linear transformation processing of a weighted value and an offset item is carried out, and the output of the hidden layer is obtained through an activation layer;

the key module is provided with a switch key and a calibration key, the switch key controls the start and the stop of the system, and the control module can receive a calibration request by pressing the calibration key, so that the system enters a calibration mode;

the alarm module adopts an active buzzer to send out an alarm;

the display module adopts the LCD screen, and when being in the calibration mode, the LCD screen shows: in system calibration; when the system is in a working mode, the LCD screen displays the reading of the pressure sensor and the control module judges the health condition of the historic building;

the control module is mainly responsible for the normalization processing of data and the coordination and operation of the whole system, when a preset time is reached or a calibration request is received, the control module enters a calibration mode, activates the sensor B and receives the data all the time until enough data meeting certain requirements are obtained, then carries out the normalization processing on the data, takes the data obtained by the sensor A as training data, takes the data obtained by the sensor B as a training label, carries out reverse training by using a gradient descent method to obtain new network parameters and update a calibration network, then closes the sensor B, when the control module is in a working mode, the control module needs to receive the data of the sensor A through the communication module, sends the data to the calibration module after the normalization processing, receives the calibrated data after the calibration is finished and carries out reverse normalization, finally judges whether the data is normal or not, if the data is abnormal, the control display module displays corresponding information, and the alarm module gives an alarm.

Preferably, the MCU is in communication connection with a communication module, the communication module is in communication connection with the pressure sensor, and the MCU is in communication connection with the pressure sensor through the communication module.

Preferably, the MCU adopts STM32F103 series single-chip microcomputer, and the pressure sensor is divided into a working mode and a calibration mode.

Preferably, the communication module adopts an I2C bus, a serial bus consisting of a data line SDA and a clock SCL.

Preferably, the structure from the hidden layer to the output layer is very similar to that from the input layer to the hidden layer, and a four-in-one structure BP neural network is adopted.

Preferably, the control module is electrically connected with the alarm module, and the key module is electrically connected with the control module.

An ancient building stress monitoring and early warning method with calibration and automatic calibration capabilities comprises the following steps:

in the working mode:

the method comprises the following steps: the sensor A sends data to the control module through the communication module;

step two: the control module normalizes the received data and sends the normalized data to the calibration module;

step three: the calibration module inputs data into the trained neural network module for forward calculation to obtain an output result of a neural network output layer and then sends the output result into the control module;

step four: the control module performs inverse normalization processing on the BP neural network output data to obtain calibrated output data;

step five: the control module judges whether the data are normal or not, if the data are abnormal, the display module is controlled to display corresponding information, and the alarm module gives an alarm.

In the calibration mode:

the method comprises the following steps: reaching a preset time or receiving a calibration request;

step two: the sensor A and the sensor B send data to the control module through the communication module;

step three: the control module carries out normalization processing on the received data;

step four: the control module takes the data obtained by the sensor A as training data and the data obtained by the sensor B as a training label, and performs reverse training by using a gradient descent method to obtain new network parameters;

step five: and the calibration module updates the weight and the offset value of the BP neural network and closes the sensor B.

Compared with the prior art, the invention has the following beneficial effects:

1. the system has the automatic calibration capability by utilizing the unique arrangement of the double sensors and the calibration mode, the sensor B is in a closed state at ordinary times and is rarely used, the influence of various factors on the sensor B can be avoided, and the reading is accurate, so that the BP neural network can be retrained according to the reading of the sensor B as calibration data to obtain a new calibration network, the error of a pressure sensor in the existing monitoring and early warning system caused by factors such as equipment aging and environmental change is corrected, the reading is more accurate, the actual stress condition of the ancient building can be reflected more truly and accurately, and the monitoring and early warning effect is achieved better;

2. the invention aims at the nonlinear characteristic of the output data of the pressure sensor, adopts the BP neural network with stronger fitting capability as a calibration algorithm, and improves the calibration precision.

Drawings

FIG. 1 is a block diagram of a patented structural system of the present invention;

FIG. 2 is a flow chart of the working mode of the present invention;

FIG. 3 is a flow chart of the patented calibration mode of the invention.

In the figure: 1. MCU; 2. a pressure sensor; 21. a sensor A; 22. a sensor B; 3. a communication module; 4. a calibration module; 5. a control module; 6. a key module; 7. an alarm module; 8. and a display module.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by a person skilled in the art based on the embodiments in the patent of the invention without any inventive work belong to the protection scope of the patent of the invention.

Referring to fig. 1-3, an ancient building stress monitoring and early warning system with calibration and automatic calibration capabilities comprises an MCU 1, a pressure sensor 2, a communication module 3, a calibration module 4, a control module 5, a key module 6, an alarm module 7 and a display module 8;

the pressure sensor 2 comprises a sensor A21 and a sensor B22, the sensor A21 is in a working state as a daily monitoring sensor in a working mode, the sensor B22 is in a closed state as a standby sensor at ordinary times, errors caused by factors such as equipment aging and environmental changes are avoided, and the sensor A21 and the sensor B22 are in working states in a calibration mode;

the communication module 3 plays a role in sending and receiving data, the MCU 1 is used as a host, the sensor A21 and the sensor B22 are used as slaves, the sensor A21 and the sensor B22 both have different addresses, the host sends a start signal, a corresponding slave is searched next to a transmission address signal, and a reading or writing instruction is sent, after the slaves respond, the host and the slaves can transmit data, each time operation is finished, the slaves can generate a response signal to judge whether communication is successful, and after data transmission is finished, the host sends an end signal;

the calibration module 4 is used for calibrating the data sent by the control module 5, the calibration module 4 is realized by adopting a BP neural network, the weight and the offset parameter of the BP neural network are updated according to the training result of the control module 5, the BP neural network consists of an input layer, a hidden layer and an output layer, the data is input from the input layer, the data is subjected to linear transformation processing of weight values and offset items, and then the output of the hidden layer is obtained through an activation layer;

the key module 6 is provided with a switch key and a calibration key, the switch key controls the start and the stop of the system, the calibration key is pressed, the control module 5 can receive a calibration request, and the system enters a calibration mode;

the alarm module 7 adopts an active buzzer to send out an alarm;

the display module 8 adopts an LCD screen, and when the LCD screen is in a calibration mode, the LCD screen displays: in system calibration; when the device is in a working mode, the LCD screen displays the reading of the pressure sensor 2 and the control module 5 judges the health condition of the historic building;

the control module 5 is mainly responsible for normalization processing of data and coordination and operation of the whole system, when a preset time is reached or a calibration request is received, a calibration mode is entered, the control module 5 activates the sensor B22 and receives the data all the time until enough data meeting certain requirements are obtained, then normalization processing is carried out on the data, the data obtained by the sensor A21 is used as training data, the data obtained by the sensor B22 is used as a training label, reverse training is carried out by utilizing a gradient descent method to obtain new network parameters and update a calibration network, then the sensor B22 is closed, when the control module 5 is in a working mode, the data of the sensor A21 needs to be received through the communication module 3 and sent to the calibration module 4 after normalization processing is carried out, the calibrated data is received and subjected to reverse normalization after the calibration is finished, and finally whether the data are normal or not is judged, if the data is abnormal, the control display module 8 displays corresponding information, and the alarm module 7 gives an alarm.

In this embodiment, MCU 1 and communication module 3 communication connection, communication module 3 and pressure sensor 2 communication connection, MCU 1 passes through communication module 3 and pressure sensor 2 communication connection, conveniently controls opening of pressure sensor 2 and stops.

In this embodiment, MCU 1 adopts STM32F103 series singlechip, and pressure sensor 2 divide into mode and calibration mode, conveniently controls.

In this embodiment, the communication module 3 adopts an I2C bus and a serial bus formed by a data line SDA and a clock SCL, which facilitates information transmission.

In this embodiment, the structure from the hidden layer to the output layer is very similar to that from the input layer to the hidden layer, and a BP neural network having a four-in-one structure is adopted, which is beneficial to training.

In this embodiment, the control module 5 is electrically connected to the alarm module 7, and the key module 6 is electrically connected to the control module 5, so as to facilitate early warning and information control input.

An ancient building stress monitoring and early warning method with calibration and automatic calibration capabilities comprises the following steps:

in the working mode:

the method comprises the following steps: the sensor A21 sends data to the control module 5 through the communication module 3;

step two: the control module 5 normalizes the received data and sends the normalized data to the calibration module 4;

step three: the calibration module 4 inputs data into the trained neural network module for forward calculation to obtain an output result of a neural network output layer, and then the output result is sent to the control module 5;

step four: the control module 5 performs inverse normalization processing on the BP neural network output data to obtain calibrated output data;

step five: the control module 5 judges whether the data is normal or not, if the data is abnormal, the control display module 8 displays corresponding information, and the alarm module 7 gives an alarm.

In the calibration mode:

the method comprises the following steps: reaching a preset time or receiving a calibration request;

step two: the sensor A21 and the sensor B22 send data to the control module 5 through the communication module 3;

step three: the control module 5 performs normalization processing on the received data;

step four: the control module 5 takes the data obtained by the sensor A21 as training data and the data obtained by the sensor B22 as training labels, and performs reverse training by using a gradient descent method to obtain new network parameters;

step five: the calibration module 4 updates the weight and the offset value of the BP neural network and turns off the sensor B22.

Although embodiments of the present patent have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the present patent, the scope of which is defined in the appended claims and their equivalents.