An intelligent real-time monitoring and early warning device for ancient buildings based on multi-sensor heterogeneity

Abstract

The present invention discloses an intelligent real-time monitoring and early warning device for heterogeneous ancient buildings based on multi-sensors, which includes a collection terminal, a transmission unit and a controller. The collection terminal is installed at the corresponding monitoring positions of the ancient building, and it includes an acceleration sensor, an inclination sensor, a displacement sensor, a crack sensor and a multi-channel timing control module. The acceleration sensor, the inclination sensor, the displacement sensor and the crack sensor form a multi-mode sensor group to realize the measurement and monitoring of the ancient building. The multi-channel timing control module ensures the synchronization of each electrical pulse signal collected by the multi-mode sensor group, and transmits each electrical pulse signal collected to the controller through the transmission unit. The present invention realizes the measurement and monitoring of the ancient building through the multi-mode sensor group, and improves the perception accuracy through the excellent performance of the sensor in processing time series data and pattern recognition, and can provide support for the health monitoring of the ancient building more efficiently and intelligently.

Description

Technical Field

[0001] The present invention relates to the field of measurement and monitoring, and particularly to an intelligent real-time monitoring and early warning device for ancient buildings based on multi-sensor heterogeneity. Background Art

[0002] The value of ancient buildings is multi-dimensional, not only reflected in the inheritance of history and culture, but also having a profound impact on art, science, society, economy, environment and spirit. Protecting and rationally utilizing ancient buildings is of great significance for inheriting human civilization, promoting social development and enhancing cultural confidence. Using modern scientific and technological means to monitor ancient buildings can timely discover structural problems, prevent small problems from evolving into major damages, and at the same time, more scientific maintenance and management of ancient buildings can be carried out based on this.

[0003] In the field of ancient building monitoring, most of the current monitoring is relatively macroscopic or the monitoring methods are single, making it difficult to achieve comprehensive and in-depth monitoring. In addition, there are the following problems: the health risk problems presented by ancient buildings are often multi-dimensional and complex. The current monitoring means are often based on single parameter indicators for independent analysis, making it difficult to obtain accurate evaluation and early warning results; during the monitoring process of ancient buildings, due to the tiny changes in their own structures, the attenuation during signal transmission, and external environmental factors (wind influence, temperature change, etc.), the monitored signals are often weak, resulting in an increase in the difficulty of subsequent signal analysis and processing. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent real-time monitoring and early warning device for ancient buildings based on multi-sensor heterogeneity to solve the above problems.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions:

[0006] An intelligent real-time monitoring and early warning device for heterogeneous ancient buildings based on multi-sensors, comprising a collection terminal, a transmission unit and a controller. The collection terminal is installed at the corresponding monitoring positions of the ancient building, and it includes an acceleration sensor, an inclination sensor, a displacement sensor, a crack sensor and a multi-channel timing control module. The acceleration sensor is used to detect the vibration of the ancient building, the displacement sensor is used to detect the displacement of the ancient building, the inclination sensor is used to detect the inclination change of the ancient building, and the crack sensor is used to detect the change in the crack width existing in the ancient building. The acceleration sensor, the inclination sensor, the displacement sensor and the crack sensor constitute a multi-mode sensor group to realize the measurement and monitoring of the ancient building. The input end of the multi-channel timing control module is connected to the signal output end of the multi-mode sensor group, and ensures the synchronization of each electrical pulse signal collected by the multi-mode sensor group. The multi-channel timing control module transmits each collected electrical pulse signal to the controller through the transmission unit. The controller includes a time calibrator, a time-amplitude converter, a signal conversion module and a data analysis module. The time calibrator is used to record the instantaneous value and time interval of the electrical pulse signal. The signal conversion module converts the electrical pulse signal into a group of discrete pulse time series based on the instantaneous value and time interval of the electrical pulse signal. The time-amplitude converter is used to convert the time interval of the electrical pulse signal into a pulse amplitude, and output an analog pulse proportional to time, and then generate a group of time-amplitude pulse series with time information through the signal conversion module. The signal conversion module sends the pulse time series and the time-amplitude pulse series to the data analysis module for the analysis and evaluation of the health state of the ancient building. The data analysis module includes a multi-mode perceptron and a pulse neural network system. The multi-mode perceptron is used to perform binary fitting on the pulse time series. The multi-mode perceptron is a plurality of perceptrons connected together to perform exclusive OR Boolean operations. The pulse neural network system includes a building state feature extraction layer and a building state detection layer. The building state feature extraction layer obtains a building state feature pulse series based on the binary-fitted pulse time series. The building state detection layer performs building state detection based on the building state feature pulse series and obtains a building state evaluation result.

[0007] Preferably, the multi-channel timing control module adopts a data acquisition unit card, and the data acquisition unit card adopts an NI PCI-6229 multi-functional data acquisition unit card. The data acquisition unit card is integrated with LabVIEW software, and the LabVIEW software is used to perform synchronous configuration of the signal acquisition unit to ensure that the signal acquisition units of each sensor start to act at the same moment.

[0008] Preferably, the transmission unit includes an electro-optic modulator, a wavelength division multiplexer, a demultiplexer, and an optoelectronic converter. The electro-optic modulator, the wavelength division multiplexer, and the demultiplexer are sequentially connected in series through optical fibers on the same optical path. The input end of the electro-optic modulator is connected to the output end of the multi-channel timing control module, and is used to convert multiple electrical pulse signals output by the multi-channel timing control module into multiple optical pulse signals. The wavelength division multiplexer is used to multiplex multiple optical pulse signals and transmit them through optical fibers. The demultiplexer is used to separate the multiplexed optical pulse signals to obtain multiple optical pulse signals. The optoelectronic converter is used to convert multiple optical pulse signals into multiple electrical pulse signals.

[0009] Preferably, the controller further includes a signal fusion module, and the signal fusion module is used to sum the pulse time sequence and the time-amplitude pulse sequence as the input of the multi-mode sensor.

[0010] Preferably, the acquisition terminal further includes a configuration module, and the configuration module is used to configure the acquisition period and the range of acquisition signal values.

[0011] Preferably, the binary fitting of the multi-mode sensor is realized by the following formula:

[0012]

[0013] Wherein, is the output value of n inputs, represents the input vector, represents the number of input vectors, represents the on-value weight, represents the adjustment coefficient, is the output value of all inputs at the i moment, is used as the input of the activation function, and the output of the activation function is the result of binary fitting.

[0014] Preferably, the activation function adopts a step function, which is expressed as:

[0015]

[0016] Wherein, is the result of binary fitting, is used as the input of the activation function, is the output threshold.

[0017] Preferably, the network structure of the spiking neural network includes an input layer, a hidden layer, and an output layer. Each layer is composed of a group of LIF neuron models. The input layer and the hidden layer form a memory generation module. The binarized and fitted spike time series of the input layer are introduced into the spiking neural network, causing the corresponding input neurons to generate spikes and transmitting the spikes to the hidden layer for information processing and conversion.

[0018] Preferably, the building state feature pulse sequence is a pulse sequence representing building risk factors, and the building state evaluation result is a result representing possible risks of ancient buildings.

[0019] The building state detection layer performs building state detection based on the building state feature pulse sequence, which is implemented through a Pearson correlation analysis model. The Pearson correlation analysis model is:

[0020]

[0021] Where and are the observed values of two variables, which are determined based on the building state feature pulse sequence. and are the observed values of two variables. k is the sample size.

[0022] After adopting the above technical solution, compared with the background technology, the present invention has the following beneficial effects:

[0023] 1. The present invention uses an acceleration sensor, an inclination sensor, a displacement sensor, and a crack sensor to perform multi-dimensional measurement and monitoring on ancient buildings, and utilizes the excellent performance of the perceptron in processing time series data and pattern recognition to improve the perception accuracy. By setting a multi-channel timing control module, it is possible to ensure that the signal acquisition units of each sensor start to act at the same moment, facilitating the multi-mode perceptron to monitor events at the same moment.

[0024] 2. The present invention obtains a spike time series with time information and a time-amplitude pulse sequence by setting a time calibrator and a time-amplitude converter, enabling multiple perceptrons to obtain accurate time intervals and the order of perceived events. At the same time, based on the time-amplitude pulse sequence, weak pulses can be better perceived by the perceptron.

[0025] 3. The present invention sets a signal fusion module to sum the spike time series and the time-amplitude pulse sequence, and then uses it as the input of the multi-mode perceptron, further ensuring an effective and stable perception data basis for subsequent operations.

[0026] 4. The intelligent real-time monitoring and warning device for ancient buildings of the present invention has high energy efficiency, precise time processing, and strong fault tolerance, and can provide support for the health monitoring of ancient buildings more efficiently and intelligently. Brief Description of the Drawings

[0027] Figure 1 It is a schematic diagram of the system architecture of the present invention. Detailed Embodiment

[0028] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. Embodiment

[0029] The present invention discloses an intelligent real-time monitoring and warning device for ancient buildings based on multi-sensor heterogeneity, including a collection terminal 100, a transmission unit 200, and a controller 300, wherein:

[0030] The collection terminal 100 is installed at the corresponding monitoring positions of the ancient building, and it includes an acceleration sensor 110, an inclination sensor 120, a displacement sensor 130, a crack sensor 140, and a multi-channel timing control module 150. The acceleration sensor 110 is used to detect the vibration of the ancient building, the displacement sensor 130 is used to detect the displacement of the ancient building, the inclination sensor 120 is used to detect the inclination change of the ancient building, and the crack sensor 140 is used to detect the change in the crack width existing in the ancient building. The acceleration sensor 110, the displacement sensor 130, the inclination sensor 120, and the crack sensor 140 form a multi-mode sensor group to realize the measurement and monitoring of the ancient building. The input end of the multi-channel timing control module 150 is connected to the signal output end of the multi-mode sensor group, and ensures that the collected electrical pulse signals of the multi-mode sensor group are synchronized. The multi-channel timing control module 150 transmits the collected electrical pulse signals to the controller 300 through the transmission unit 200. In this embodiment, the multi-channel timing control module 150 adopts a data acquisition unit card, and the data acquisition unit card adopts a NI PCI-6229 multi-functional data acquisition unit card. The data acquisition unit card is integrated with LabVIEW software, and the LabVIEW software is used for synchronous configuration of the signal acquisition unit to ensure that the signal acquisition units of each sensor start to act at the same moment. NI PCI-6229 is a multi-functional data acquisition unit card produced by National Instruments (NI), which has 16 or 32 analog input channels, 4 analog output channels, 48 I / O channels, 2 32-bit timers, and supports LabVIEW software at the same time. The present invention realizes that "the signal acquisition units of each sensor start to act at the same moment" by using the real-time signal acquisition unit function of the LabVIEW software.

[0031] The acquisition terminal 100 further includes a configuration module 160, which is used to configure the acquisition period and the range of the acquired signal values. The acquisition period can be a continuous time period or an intermittent segmented period. The range of the acquired signal values can be set according to prior knowledge and experience, and should match the acquisition unit range of the sensor at the same time.

[0032] The transmission unit 200 includes an electro-optic modulator 210, a wavelength division multiplexer 220, a demultiplexer 230, and an optoelectronic converter 240. The electro-optic modulator 210, the wavelength division multiplexer 220, and the demultiplexer 230 are sequentially connected in series through an optical fiber on the same optical path. The input end of the electro-optic modulator 210 is connected to the output end of the multi-channel timing control module 150, and is used to convert multiple electrical pulse signals output by the multi-channel timing control module 150 into multiple optical pulse signals. The wavelength division multiplexer 220 is used to multiplex multiple optical pulse signals and transmit them through an optical fiber. The demultiplexer 230 is used to separate the multiplexed optical pulse signals to obtain multiple optical pulse signals. The optoelectronic converter 240 is used to convert multiple optical pulse signals into multiple electrical pulse signals.

[0033] The controller 300 includes a signal controller 300 including a time calibrator 310, a time-amplitude converter 320, a signal conversion module 330, and a data analysis module, where:

[0034] The time calibrator 310 is used to record the instantaneous value and the time interval of the electrical pulse signal. The signal conversion module 330 converts the electrical pulse signal into a set of discrete pulse time series based on the instantaneous value and the time interval of the electrical pulse signal. The time-amplitude converter 320 is used to convert the time interval of the electrical pulse signal into a pulse amplitude, and output an analog pulse proportional to the time, and then generate a set of time-amplitude pulse series with time information through the signal conversion module 330. The signal conversion module 330 sends the pulse time series and the time-amplitude pulse series to the data analysis module for analyzing and evaluating the health status of the ancient building.

[0035] The data analysis module includes a multi-mode perceptron 350 and a pulse neural network system 360. The multi-mode perceptron 350 is used to perform binary fitting on the pulse time series. The multi-mode perceptron 350 is a plurality of perceptrons connected together to perform exclusive OR Boolean operations.

[0036] A perceptron can have one input and one output, or multiple inputs and one output, that is, multiple perceptions. This not only allows the acquisition unit to collect more data, but importantly, it can play a role in verification. Multi-modal perception can be said to be a very complex electro-chemical reaction acquisition unit, and the entire working process is dynamic. Therefore, we need to select the effective, that is, the most stable, perception data as the base value for calculation. One output can be used as the input for other units. A perceptron has a certain fitting ability and can perform binary classification on the input, that is, divide the input data into two categories, namely, for a given input, output 0 (belonging to class 0) or 1 (belonging to class 1). A perceptron can use operations such as Boolean operation truth values for simulation. For more than two placed weight parameters, that is, M0..., we can adjust the values of the parameters so that the perceptron can simulate Boolean operations. For example, let M0 = 0.3, M1 = 0.3,...M n = 0.5, at this time the perceptron can simulate Boolean operations.

[0037] The binary fitting of the multi-modal perceptron 350 is achieved through the following formula:

[0038]

[0039] Among them, is the output value of n inputs, represents the input vector, represents the number of input vectors, represents the value weight, represents the adjustment coefficient, is the output value of all inputs at the i th moment, is used as the input of the activation function, and the output of the activation function is the result of binary fitting.

[0040] The activation function uses a step function, which is expressed as:

[0041]

[0042] Among them, is the result of binary fitting, is used as the input of the activation function, is the output threshold.

[0043] A pulse is an instantaneous and perceptual stimulus. It may succeed or fail. That is, a perceptron has a threshold. If it receives the stimulus, it is a success; if not, it is a failure. If it fails, no perturbation value is generated. If it succeeds, the perturbation will form a certain standard form and be conducted along the link, which is defined as a mathematical value. Now, let's convert the pulse into computer binary. When it fails (no pulse), it is represented by a value "0", and when it succeeds (there is a pulse), it is represented by a value "1". In theory, a perceptron can have multi-modal computing capabilities, but in reality, it is very difficult for a single perceptron to imitate complex multi-modal logical operations. At this time, multiple perceptrons need to be connected together to make multiple perceptrons become a multi-modal perceptron 350 with the ability to imitate complex logical operations. The multi-modal perceptron 350 is a perception system and can be regarded as the most basic monitoring neural network. As long as the micro-energy activates the multi-modal perceptron 350, it will have a more powerful ability to simulate neurons, and thus can handle more complex problems.

[0044] The pulsed neural network system 360 includes a building state feature extraction layer 361 and a building state detection layer 362. The building state feature extraction layer 361 obtains a building state feature pulse sequence based on the binarized and fitted pulse time series. The building state detection layer 362 performs building state detection based on the building state feature pulse sequence and obtains a building state evaluation result. The building state feature pulse sequence is a pulse sequence representing building risk factors, and the building state evaluation result is a result representing possible risks of ancient buildings. The above-mentioned building state detection layer 362 performs building state detection based on the building state feature pulse sequence, which is achieved through a Pearson correlation analysis model. The Pearson correlation analysis model is:

[0045]

[0046] Where, and are the observed values of two variables, which are determined based on the building state feature pulse sequence. and are the observed values of two variables. k is the sample size.

[0047] The network structure of the pulsed neural network includes an input layer, a hidden layer, and an output layer. Each layer is composed of a group of LIF neuron models. The input layer and the hidden layer form a memory generation module. The binarized and fitted pulse time series of the input layer is introduced into the pulsed neural network, causing the corresponding input neurons to generate pulses and transmit the pulses to the hidden layer for information processing and conversion.

[0048] The controller 300 further includes a signal fusion module 340, which is configured to sum the pulse time series and the time-amplitude pulse series as the input of the multi-mode sensor 350. Before summing the pulse time series and the time-amplitude pulse series, in order to ensure the unified processing of the subsequent multi-mode sensor 350, it is necessary to use the signal fusion module 340 to first convert the pulse time series and the time-amplitude pulse series into a unified period, that is, a unified pulse frequency. The "summing of the pulse time series and the time-amplitude pulse series" involved here is for the pulse series derived from the pulse signals collected by the same sensor acquisition unit, especially for the processing of weak sensor pulse information.

[0049] As can be seen from the above description, the present invention uses the acceleration sensor 110, the displacement sensor 130, the tilt sensor 120, and the crack sensor 140 to perform multi-dimensional measurement and monitoring of ancient buildings, and uses the excellent performance of the sensor in processing time series data and pattern recognition to improve the perception accuracy. By setting the multi-channel timing control module 150, it is possible to ensure that the signal acquisition units of each sensor start to act at the same moment, which is convenient for the multi-mode sensor 350 to monitor events at the same moment. By setting the time calibrator 310 and the time-amplitude converter 320, the present invention obtains the pulse time series with time information and the time-amplitude pulse series, which can enable multiple sensors to obtain accurate time intervals and the order of sensed events. At the same time, based on the time-amplitude pulse series, weak pulses can be better sensed by the sensor. By setting the signal fusion module 340, the present invention sums the pulse time series and the time-amplitude pulse series, and then uses it as the input of the multi-mode sensor 350, further ensuring an effective and stable perception data basis for subsequent operations. The intelligent real-time monitoring and warning device for ancient buildings of the present invention has high energy efficiency, precise time processing, and strong fault tolerance, and can provide more efficient and intelligent support for the health monitoring of ancient buildings.

[0050] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An intelligent real-time monitoring and early warning device for heterogeneous ancient buildings based on multiple sensors, characterized by: The invention comprises a collection terminal, a transmission unit and a controller. The collection terminal is installed at a corresponding monitoring position of the ancient building. The collection terminal comprises an acceleration sensor, a tilt sensor, a displacement sensor, a crack sensor and a multi-channel timing control module. The acceleration sensor is used to detect the vibration of the ancient building, the displacement sensor is used to detect the displacement of the ancient building, the tilt sensor is used to detect the tilt change of the ancient building, and the crack sensor is used to detect the change in the width of the crack in the ancient building. The acceleration sensor, the tilt sensor, the displacement sensor and the crack sensor constitute a multi-mode sensor group to realize the measurement and monitoring of the ancient building. The input end of the multi-channel timing control module is connected to the signal output end of the multi-mode sensor group, and ensures that the collected electric pulse signals of the multi-mode sensor group are synchronized. The multi-channel timing control module transmits the collected electric pulse signals to the controller through the transmission unit. The controller comprises a signal. The controller comprises a time calibrator, a time amplitude converter, a signal conversion module and a data analysis module. The time calibrator is used to record the instantaneous value and time of the electric pulse signal. interval, the signal conversion module converts the electric pulse signal into a set of discrete pulse time series based on the instantaneous value and time interval of the electric pulse signal, the time amplitude converter is used to convert the time interval of the electric pulse signal into a pulse amplitude, and output an analog pulse proportional to the time, and then a set of time amplitude pulse sequences with time information is generated through the signal conversion module, the signal conversion module sends the pulse time sequence and the time amplitude pulse sequence to the data analysis module, and performs health status analysis and evaluation of the ancient building, the data analysis module includes a multi-mode sensor and a pulse neural network system, the multi-mode sensor is used to perform binary fitting on the pulse time sequence, the multi-mode sensor is a plurality of sensors connected together to perform XOR Boolean operation, the pulse neural network system includes a building state feature extraction layer and a building state detection layer, the building state feature extraction layer obtains a building state feature pulse sequence based on the pulse time sequence after binary fitting, the building state detection layer performs building state detection based on the building state feature pulse sequence, and obtains a building state evaluation result; The transmission unit includes an electro-optical modulator, a wavelength division multiplexer, a wave splitter and a photoelectric converter. The electro-optical modulator, the wavelength division multiplexer and the wave splitter are sequentially connected in series on the same optical path through optical fibers. The input end of the electro-optical modulator is connected to the output end of the multi-channel timing control module, and is used to convert multiple electrical pulse signals output by the multi-channel timing control module into multiple optical pulse signals. The wavelength division multiplexer is used to combine multiple optical pulse signals and transmit them through optical fibers. The wave splitter is used to separate the combined optical pulse signal to obtain multiple optical pulse signals. The photoelectric converter is used to convert multiple optical pulse signals into multiple electrical pulse signals. The controller further comprises a signal fusion module, the signal fusion module being used to sum the pulse time sequence and the time amplitude pulse sequence as an input of the multi-mode sensor; The binary fitting of the multi-mode perceptron is achieved by the following formula: in, is the output value of n inputs, represents the input vector, represents the number of input vectors, Represents the on-duty weight, represents the adjustment factor, For the i The output value of all inputs at the moment, As the input of the activation function, the output of the activation function is the result of the binary fitting; The activation function adopts a step function, which is expressed as: in, The result of binarization fitting is As the input of the activation function, is the output threshold; The network structure of the spiking neural network includes an input layer, a hidden layer, and an output layer, each layer is composed of a group of LIF neuron models, and the input layer and the hidden layer constitute a memory generation module; the pulse time series after binarization fitting of the input layer is introduced into the spiking neural network, so that the corresponding input neurons generate pulses, and the pulses are transmitted to the hidden layer for information processing and conversion; The building state characteristic pulse sequence is a pulse sequence representing a building risk factor, and the building state assessment result is a result representing a possible risk of the ancient building; The building status detection layer performs building status detection based on the building status characteristic pulse sequence, which is realized by the Pearson correlation analysis model. The Pearson correlation analysis model is: in, and The observed values ​​of the two variables are determined based on the characteristic pulse sequence of the building state, and are the observed values ​​of two variables, k is the sample size.

2. The intelligent real-time monitoring and early warning device for heterogeneous ancient buildings based on multiple sensors as claimed in claim 1 is characterized in that: The multi-channel timing control module adopts a data acquisition unit card, and the data acquisition unit card adopts a NI PCI-6229 multi-function data acquisition unit card. The data acquisition unit card is integrated with LabVIEW software, and the LabVIEW software is used to perform synchronous configuration of the signal acquisition unit to ensure that the signal acquisition unit actions of each sensor start at the same time.

3. The intelligent real-time monitoring and early warning device for heterogeneous ancient buildings based on multiple sensors as claimed in claim 2 is characterized in that: The acquisition terminal also includes a configuration module, and the configuration module is used to configure the acquisition period and the range of the acquisition signal value.

Images