Bridge road section monitoring early warning and prevention and control system

Abstract

The invention relates to a bridge road section monitoring early warning and prevention and control system, and belongs to the technical field of traffic safety. The system comprises a data acquisition end, a data processing end, a first dual-channel power supply module and a second dual-channel power supply module, the data acquisition end comprises an image processing device, a sensor, a video sensing device and an unvarnished transmission data sending device; the data processing end comprises a core control unit, an early warning screen controller, a bridge abnormal state early warning screen and a transparent transmission data receiving device; according to the invention, real-time monitoring of the health state of the bridge in the bridge section can be realized, the method is suitable for in-service highway bridges, the traffic safety capability of the bridge section under severe weather conditions is improved, the defects of traditional bridge monitoring are overcome, and monitoring blind spots of the highway bridges are eliminated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of traffic safety, and particularly relates to a bridge section monitoring, early warning and prevention and control system. BACKGROUND

[0002] In a traffic network, a bridge has the role of connecting traffic lines and is the throat of traffic, and has important strategic significance in economy, military and social life. Due to internal and external factors such as traffic load changes, sudden traffic incidents, wire aging, storms, earthquakes and the like, the aging and damage of the bridge structure are accelerated, and if there is a lack of monitoring, maintenance and safety early warning of the bridge, it can lead to huge economic losses and serious catastrophic events, and even cause significant casualties. In the face of the urgent demand for highway bridge section monitoring, the safety monitoring technology of the highway bridge section gradually becomes a direction that road safety operation and management units and scientific researchers urgently need to research.

[0003] Retrieving existing bridge early warning and active prevention and control technologies, it is found that at present, in the field of highway bridge section monitoring, a filtering algorithm is mainly used to remove abnormal signals. However, in the actual monitoring process, there is still a problem that the monitoring data is mistakenly deleted by the algorithm due to the inability of the system to effectively identify.

[0004] Secondly, the current monitoring of the bridge still uses a wired communication scheme to complete the data transmission of the sensor, and lacks a sensor data transmission scheme using a transparent transmission mode. The sensor wiring needs to consume a large amount of manpower and material resources, and it is particularly difficult to troubleshoot the aging of the wire and the failure of the sensor.

[0005] Thirdly, due to the lack of bridge section abnormal event sensing capability, some highway bridge sections lack effective vehicle abnormal driving behavior (stopping, accident) identification, sensing, early warning plan, and the operation safety risk of the highway bridge section is high.

[0006] Furthermore, in the traditional scheme of bridge early warning and active prevention and control using only the LSTM algorithm, since

[0007] The LSTM model directly uses the original data, lacks a special abnormal detection mechanism, and thus may misjudge noise as a normal pattern or fail to accurately capture short-term or sudden abnormalities. In addition, LSTM is good at capturing long-term dependencies in time series data, but its ability to distinguish complex abnormalities is limited. For example, LSTM often cannot make accurate judgments when dealing with persistent faults (such as data deviating from normal values for a long time) and occasional disturbances (such as data suddenly deviating at a certain time).

[0008] Finally, in the traditional scheme of only using the Isolation Forest algorithm to bridge early warning and active prevention, since the Isolation Forest is suitable for large-scale data and has high training efficiency, the abnormality detection capability of the data processing relying only on the Isolation Forest algorithm is limited by the data structure.

[0009] Therefore, it is necessary to further study the monitoring technology suitable for the highway bridge section, so as to improve the accuracy and data integrity of data monitoring, and fundamentally reduce the use and maintenance cost. SUMMARY

[0010] The purpose of the present application is to solve the problems of the prior art and provide a bridge section monitoring, early warning and prevention system.

[0011] To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0012] A bridge section monitoring, early warning and prevention system comprises a data acquisition end, a data processing end, a first double-channel power supply module and a second double-channel power supply module.

[0013] The first double-channel power supply module is connected to the data processing end and used to supply power to the data processing end.

[0014] The second double-channel power supply module is connected to the data acquisition end and used to supply power to the data acquisition end.

[0015] The data acquisition end comprises an image processing device, a sensor, a video sensing device and a transparent data sending device.

[0016] The data processing end comprises a core control unit, an early warning screen controller, a bridge abnormal state early warning screen and a transparent data receiving device.

[0017] The image processing device is connected to the transparent data receiving device and the video sensing device.

[0018] The video sensing device is used to automatically intercept the abnormal event video stream after detecting the bridge section vehicle abnormal event and the bridge section meteorological disaster event, and transmit the video stream to the image processing device.

[0019] The image processing device is used to identify and classify the abnormal event video stream into two categories of bridge section vehicle abnormal event and bridge section meteorological disaster event, identify the severity of the bridge section meteorological disaster, and transmit the classification and identification results to the transparent data receiving device.

[0020] The transparent data sending device comprises a sensing data connection port, a data format converter and a transparent data transmitter; the data format converter is connected to the sensing data connection port, the transparent data transmitter and the second double-channel power supply module; and the sensor is connected to the sensing data connection port.

[0021] The sensor is used to collect displacement data, strain data and settlement data obtained by detecting the bridge, and then transmit the original message data of the collected data to the data format converter through the sensor data connection port;

[0022] The data format converter is used to analyze and convert the original message data collected by the sensor;

[0023] The transparent data transmitter is used to send the data analyzed and converted by the data format converter to the transparent data receiving device;

[0024] The core control unit is connected with the early warning screen controller and the transparent data receiving device respectively; the early warning screen controller is also connected with the bridge abnormal state early warning screen;

[0025] The core control unit is used to obtain the data received by the transparent data receiving device, and calculate the severity of the bridge according to the data received by the transparent data receiving device;

[0026] The core control unit outputs the severity of the bridge and the meteorological severity of the bridge section to the early warning screen controller, and then the early warning screen controller displays the early warning information to the passing vehicles through the bridge abnormal state early warning screen according to the severity of the bridge and the meteorological severity of the bridge section.

[0027] Further, preferably, the abnormal driving behavior of the vehicle includes parking and accident; and the meteorological disaster of the bridge section includes fog and icing;

[0028] The transparent data sending device and the transparent data receiving device communicate with each other by at least one of RS485, RS232, NB-IOT and 433MHz communication mode.

[0029] Further, preferably, the data processing end further comprises a data uploading device;

[0030] The data uploading device is connected with the core control unit;

[0031] The data uploading device is used to transmit the severity of the bridge and the meteorological severity of the bridge section calculated by the core control unit to the remote control center; and is also used to transmit the data received by the transparent data receiving device to the remote control center, including the data collected by the sensor and the abnormal event video stream processed by the image processing device and the result thereof.

[0032] Further, preferably, the data processing end further comprises a fault troubleshooting device; the fault troubleshooting device is connected with the core control unit and the data uploading device respectively;

[0033] When the video sensing device fails, the corresponding fault code is sent to the fault troubleshooting device through the image processing device, the transparent data receiving device and the core control unit in turn;

[0034] When the image processing device fails, the corresponding fault code is sent to the fault troubleshooting device through the transparent data receiving device and the core control unit in turn;

[0035] When the sensor fails, the corresponding fault code is sent to the fault troubleshooting device through the transparent data sending device, the transparent data receiving device and the core control unit in turn;

[0036] When the transparent data sending device fails, the corresponding fault code is sent to the fault troubleshooting device through the transparent data receiving device and the core control unit in turn;

[0037] When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device through the core control unit;

[0038] When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device through the core control unit;

[0039] When the early warning screen controller fails, the corresponding fault code is sent to the fault troubleshooting device through the core control unit;

[0040] When the dual-channel power supply module fails, the corresponding fault code is sent to the fault troubleshooting device;

[0041] The fault troubleshooting device displays the corresponding fault problem description according to the fault code. At the same time, the fault troubleshooting device sends the fault code to the remote control center through the data uploading device.

[0042] Further, preferably, the fault code, the corresponding fault location and the fault problem description are as shown in the following table:

[0043] Fault code Fault location Fault problem description F01-01 Sensor Sensor signal loss F01-02 Sensor Sensor data out of normal range F02-01 Image processing device Image cannot be recognized F03-01 Video sensing device Picture blurred or interfered F03-02 Video sensing device No video signal F04-01 Core control unit Instruction execution timeout or no response F05-01 Transparency data receiving device Incomplete received data F05-02 Transparency data sending device Data packet loss F06-01 Early warning screen controller Screen image display disorder F06-02 Early warning screen controller Early warning information refresh failure F07-01 First dual-channel power supply module No voltage and current data F07-02 Second dual-channel power supply module Voltage and current data abnormal .

[0044] Further, preferably, the core control unit is built-in with a hybrid algorithm of Isolation Forest and LSTM. The algorithm first extracts a feature vector from sensor data, then calculates an anomaly score by using Isolation Forest, and then passes the anomaly score of Isolation Forest to LSTM, taking the anomaly score and the feature vector as the input of LSTM, and taking the classification of persistent fault and occasional interference as the output for training. If the hybrid algorithm of Isolation Forest and LSTM identifies that there is a persistent fault in the bridge monitoring data, the severity of the bridge itself is located according to the sensor position corresponding to the persistent fault data, and finally the bridge itself severity warning information is reported.

[0045] Further, preferably, the infrared light supplementing lamp is used for supplementing light for the video sensing device.

[0046] Further, preferably, the transparent data receiving device comprises a transparent data receiver, a data fusion unit and a wired data transmitter; the transparent data receiver, the data fusion unit and the wired data transmitter are sequentially connected; the wired data transmitter is connected to the core control unit in a wired connection manner.

[0047] The transparent data receiver is used for receiving data sent by the transparent data sending device and the image processing device.

[0048] The data fusion unit is used for pre-processing data received by the transparent data receiver; the pre-processing comprises separation, fusion, cleaning and summarization; when summarizing, the data is sequentially sorted and summarized according to the uplink and downlink directions of the bridge section.

[0049] The wired data transmitter is used for sending data pre-processed by the data fusion unit to the core control unit.

[0050] Further, preferably, the first dual-channel power supply module comprises a first mains interface, a first power supply channel intelligent conversion module, a first energy storage device and a first solar power supply interface.

[0051] The first solar power supply interface is connected to the first energy storage device.

[0052] The first power supply channel intelligent conversion module is connected to the first mains interface and the first energy storage device respectively.

[0053] The first power supply channel intelligent conversion module is used for switching the power supply mode, i.e. selecting solar power supply or mains power supply.

[0054] The first energy storage device is used for storing electric energy converted from solar energy.

[0055] Further, preferably, the second dual-channel power supply module comprises a second mains interface, a second power supply channel intelligent conversion module, a second energy storage device and a second solar power supply interface.

[0056] The second solar power supply interface is connected to the second energy storage device.

[0057] The second power supply channel intelligent conversion module is connected to the second mains interface and the second energy storage device respectively.

[0058] The second power supply channel intelligent conversion module is used for switching the power supply mode, i.e. selecting solar power supply or mains power supply.

[0059] The second energy storage device is used for storing electric energy converted from solar energy.

[0060] In the application, the sensor layout scheme is laid out according to the bridge design drawing.

[0061] In the application, the function of the transparent data sending device is to uniformly transmit different communication protocols of various sensors, and to uniformly access and process the data collected by the sensors.

[0062] In the application, the image processing device is used for identifying and classifying abnormal event video streams, i.e., classifying vehicle abnormal driving behaviors and bridge section meteorological disasters; the vehicle abnormal driving behaviors include parking and accidents; and the bridge section meteorological disasters include fog and icing.

[0063] In the application, the displacement data, the strain data and the settlement data are respectively connected to different serial ports of the core control unit with different numbers.

[0064] Table 1

[0065] Data type Corresponding serial number Displacement data USART1 Strain data USART2 Settlement data USART3

[0066] Displacement data: spatial position change of the bridge structure;

[0067] The sensor corresponding to the displacement data is a bridge displacement meter.

[0068] Numerical type: displacement, angle, acceleration, vibration and temperature;

[0069] Separation step: according to the serial port number (USART1) corresponding to the received sensor data, perform data separation operation.

[0070] Strain data: stress deformation of the beam body;

[0071] The sensor corresponding to the strain data is a vibrating wire strain gauge; the strain measurement numerical range is 0-±1500με.

[0072] Separation step: according to the serial port number (USART2) corresponding to the received sensor data, perform data separation operation.

[0073] Settlement data: vertical subsidence of the foundation and the ground;

[0074] The sensor corresponding to the settlement data is a vibrating wire soil settlement gauge.

[0075] Numerical type: displacement change amount of the structure, unit: mm;

[0076] Separation step: according to the serial port number (USART3) corresponding to the received sensor data, perform data separation operation.

[0077] In the application, the data fusion unit is used for preprocessing the data received by the transparent data receiver; the preprocessing includes separation, fusion, cleaning and summarizing.

[0078] The separation is performed according to the serial port number (USART1, USART2, USART3) corresponding to the sensor data received by the sensor monitoring data collected by the sensor.

[0079] After the separation step is completed, the monitoring data splicing and fusion are completed according to the direction of the bridge section sensor arrangement (in the order of left and right width), and the bridge section state is reflected from point to surface.

[0080] According to the data acquisition range specified in the sensor specification, the data outside the acquisition range is deleted to realize the data cleaning function.

[0081] When summarizing, the left and right widths are summarized separately.

[0082] In the application, the bridge abnormal state early warning screen and the early warning screen controller are communicated in a wireless communication mode, and according to the actual communication distance, a relay node can be dynamically installed.

[0083] The fault locating device can quickly locate the fault point of the system for maintenance personnel, and quickly maintain the system, thereby improving the maintenance efficiency of the system.

[0084] In the application, in view of the problem that abnormal data (external interference factors such as strong wind) and fault data are difficult to classify in the system monitoring process, a mixed algorithm of isolated forest and LSTM is deployed in the core control unit of the system, which can distinguish between incidental interference (short-time noise, environmental fluctuation) and persistent failure (system structure damage) according to the difference of data sources, and provide a more reliable data basis for structure health assessment.

[0085] The bridge abnormal state early warning screen is deployed in the roadbed section 500-1km away from the bridge section, and can send early warning information to the passing vehicles in advance.

[0086] The sensors are arranged on different monitoring points of the highway bridge section to monitor the bridge operation state.

[0087] The transparent data sending device is connected with the sensor, and the monitoring data is sent out through wireless transmission.

[0088] The first double-channel power supply module and the second double-channel power supply module are compatible with two power supply modes of commercial power and solar power, and can automatically switch the power supply mode by the power supply channel intelligent conversion module according to the light intensity around the deployment environment, so as to realize the output power supply of 24V, 12V and 5V.

[0089] In the case of failure of the mains power supply, the power supply channel intelligent conversion module can be powered by the energy storage device, ensuring the normal operation of the system functions.

[0090] The first power supply channel intelligent conversion module in the first dual-channel power supply module can monitor the power supply voltage and current of the first mains interface, the first energy storage device and the first solar power supply interface in real time; when the power supply voltage and current of the first mains interface and the first energy storage device are monitored to have a failure (no data, voltage too low), the power supply channel intelligent conversion module sends a fault code information to the fault troubleshooting device, and the fault troubleshooting device records and uploads to the remote control center.

[0091] The second power supply channel intelligent conversion module in the second dual-channel power supply module can monitor the power supply voltage and current of the second mains interface, the second energy storage device and the second solar power supply interface in real time; when the power supply voltage and current of the second mains interface and the second energy storage device are monitored to have a failure (no data, voltage too low), the power supply channel intelligent conversion module sends a fault code information to the fault troubleshooting device, and the fault troubleshooting device records and uploads to the remote control center.

[0092] The transparent data receiving device can standardize the sensor data and solve the problem of non-uniform data caused by the independent data collected by the sensors at different bridge layout points, providing a data source for subsequent algorithm processing data;

[0093] The infrared light supplementing lamp can provide object recognition capability for the video sensing device in a low lighting environment (night, low visibility). In the actual engineering implementation process, the video sensing device can be added independently according to the distance of the bridge section and the actual monitoring function needs, and the detection capability of the target vehicle is improved.

[0094] In the present application, the wireless communication technology includes LoRa, ZigBee;

[0095] The sensor supports RS485, RS232, 4G-DTU and other communication modes;

[0096] In the present application, the mains interface supports the access of mains 220 AC, and the built-in rectifier filter circuit can reduce the 220V (AC) voltage to the 12V (DC) voltage supported by the system;

[0097] In the present application, the solar power supply interface is connected with the solar panel and outputs a 12V DC power supply;

[0098] In the application, the data uploading device is used to transmit the bridge self severity calculated by the core control unit (1) and the bridge section weather severity to the remote control center; and is also used to transmit the data received by the data receiving device to the remote control center, including the data collected by the sensor and the abnormal event video stream processed by the image processing device and the results thereof. Meanwhile, the fault troubleshooting device sends the fault code to the remote control center through the data uploading device, so as to facilitate and inform the relevant personnel to process.

[0099] The core control unit in the application is built-in with a mixed algorithm of isolated forest and LSTM, and the bridge self severity is calculated.

[0100] In the traditional bridge monitoring process, the detection equipment cannot effectively distinguish between the occasional interference (short-time noise, normal fluctuation of data caused by environmental changes) and the persistent fault (system structure damage) data in the complex scene (many variable interference factors), and is easy to cause false alarm of the system due to the occasional interference (short-time noise, normal fluctuation of data caused by environmental changes). In this regard, the mixed algorithm of isolated forest and LSTM is adopted, according to the advantage of isolated forest in quickly identifying interference data, the abnormal value (noise data) is removed, the training data is purified, the learning of LSTM to normal time sequence mode is avoided, and the prediction stability of the model is improved. The clean data after preprocessing is input into LSTM, and LSTM performs secondary verification on the abnormal value deviating from the mode (such as distinguishing between "real structure abnormal data" and "normal fluctuation of sensor data caused by environmental changes"), according to the identification result, the warning is completed, and the accuracy of abnormal detection is greatly improved.

[0101] The advantages are as follows

[0102] 1. Higher accuracy and robustness:

[0103] By passing the abnormal score of isolated forest to LSTM, the mixed algorithm of isolated forest and LSTM can better distinguish between normal mode and abnormal mode, especially in the presence of complex scenes (such as normal vibration and abnormal vibration interference in bad weather), and the detection effect of this algorithm is excellent.

[0104] 2. Can effectively distinguish between persistent fault and occasional interference:

[0105] The mixed algorithm of isolated forest and LSTM can simultaneously process and distinguish between persistent abnormality and occasional abnormality. In many cases, potential abnormal data is first screened out by isolated forest, and then LSTM is used to verify the time sequence consistency of these data, which greatly improves the accuracy of abnormal detection and avoids unnecessary maintenance or missed detection risk caused by misjudgment of a single algorithm.

[0106] 3. Automatically process noise and high-dimensional data in data:

[0107] The isolated forest and LSTM hybrid algorithm helps LSTM focus on learning abnormal patterns without considering noise interference (accidental data) by passing the purified clean data to LSTM.

[0108] 4. Shorter training time, especially in the case of more data noise:

[0109] By combining the anomaly score of isolated forest and the sequence modeling ability of LSTM, the isolated forest and LSTM hybrid algorithm can handle more complex patterns, especially in high-dimensional, nonlinear, and different time scale abnormal patterns, and has good data processing ability.

[0110] Compared with the prior art, the advantages of the algorithm are:

[0111] Using LSTM alone: LSTM can learn long-term dependencies in time series, but when there are noise or accidental anomalies in the data, LSTM may mistakenly learn these abnormal points as "normal" patterns. After combining isolated forest, LSTM only needs to focus on the cleaned abnormal data, which can more effectively learn abnormal patterns and distinguish between persistent faults and accidental interference. The comparison process is shown in Figure 4 .

[0112] In summary, the development of the isolated forest and LSTM hybrid algorithm is based on the optimal solution to some limitations of the prior art (such as poor noise processing ability, inability to effectively distinguish between persistent and accidental interference, strong dependence on labeled data, etc. ) and other problems. By combining the advantages of isolated forest and LSTM, it can more efficiently and accurately handle anomaly detection tasks.

[0113] Compared with the prior art, the present application has the beneficial effects that:

[0114] The present application can realize real-time monitoring of the health state of the bridge section of the bridge, and is suitable for in-service highway bridges, but is not limited to such scenarios. The present application overcomes the problems of traditional monitoring techniques, such as difficulty in distinguishing between accidental interference (short-term noise, environmental fluctuations) and persistent fault data, low system integration, etc. Combined with vehicle abnormal driving behavior (stopping, accidents) sensing equipment, it can effectively improve the detection ability of abnormal states such as abnormal driving behavior (stopping, accidents) of vehicles on highways, improve the ability of traffic safety on bridge sections under adverse weather conditions, and complement the "shortcomings" of traditional bridge monitoring, eliminating the blind spots of highway bridge monitoring.

[0115] At the algorithm level, the isolation forest algorithm, as an unsupervised algorithm, can identify abnormal points (such as short-time noise, environmental fluctuations, and transient interference) without relying on labeled noise data, has strong detection performance for occasional impulse noise (such as sensor transient failure and electromagnetic interference), and avoids problems such as excessive signal smoothing that may be caused by traditional filtering methods. In addition, LSTM is good at capturing long-term dependencies of time series and can distinguish between noise and real vibration patterns of bridge structures (such as periodic load and environmental temperature changes). Traditional filtering algorithms (such as low-pass filtering) may filter out high-frequency useful signals (such as bridge resonance characteristics), while LSTM can dynamically reconstruct signals combined with historical data, learn the time domain distribution difference between noise and effective signals through training, and retain key features (such as modal frequency and sudden response) in structural health monitoring (SHM) while reducing noise. BRIEF DESCRIPTION OF DRAWINGS

[0116] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0117] Figure 1 The structural diagram of the bridge section monitoring, early warning and prevention and control system of the present application;

[0118] Figure 2 The execution flowchart of the isolation forest and LSTM hybrid algorithm of the bridge section monitoring, early warning and prevention and control system of the present application;

[0119] Figure 3 The algorithm function comparison test diagram of the bridge section monitoring, early warning and prevention and control system of the present application; wherein, a is to use the isolation forest algorithm to monitor abnormal data points, obtain an abnormal score, and normalize the data; b is to pass the abnormal score as an input feature to the LSTM model to distinguish abnormal patterns (persistent failure and occasional interference data) in the time series; c is to directly use the original data collected by the sensor as input to train the LSTM model to distinguish abnormal patterns;

[0120] Figure 4 The structural diagram of the second dual-channel power supply module of the bridge section monitoring, early warning and prevention and control system of the present application. Figure 5 The structural diagram of the transparent data sending device of the present application;

[0121] Figure 6 The structural diagram of the transparent data receiving device of the present application;

[0122] Figure 7 It is an appearance schematic view of the video sensing and early warning issuing device of the present application;

[0123] 1, core control unit; 2, early warning screen controller; 3, bridge abnormal state early warning screen; 4, image processing device; 5, sensor; 6, first mains power interface; 7, first power supply channel intelligent conversion module; 8, first energy storage device; 9, first solar power supply interface; 10, sensor data connection port; 11, data format converter; 12, transparent data transmitter; 13, transparent data receiver; 14, data fusion unit; 15, wired data transmitter; 16, data uploading device; 17, fault troubleshooting device; 18, video sensing device; 19, second mains power interface; 20, second power supply channel intelligent conversion module; 21, second energy storage device; 22, second solar power supply interface; 23, solar panel; 24, infrared light supplementing lamp; 25, communication antenna; 26, power cord; 27, vertical pole; 28, cross arm. DETAILED DESCRIPTION

[0124] The present application will be further described in conjunction with the following examples.

[0125] Those skilled in the art will appreciate that the following examples are given for illustrative purposes only and are not meant to limit the scope of the present application. Unless otherwise indicated, technical or conditions not specified in the examples were performed according to techniques or conditions described in the literature or according to the manufacturer's instructions. Where the manufacturer of a material or device is not named, a conventional product available from commercial vendors can be used.

[0126] Those skilled in the art will appreciate that the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It will be understood that when an element is referred to as being "connected" to or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements can be present. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0127] In the description of the present application, unless otherwise stated, the meaning of "a plurality of" is two or more. The orientation or positional relationship indicated by the terms "inner", "upper", "lower", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0128] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "provided with" should be understood in a broad sense, for example, can be fixedly connected, can be detachably connected, or integrally connected; can be mechanically connected, can be electrically connected; can be directly connected, or indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in the present application should be understood according to the specific circumstances.

[0129] Those skilled in the art can understand that unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as such.

[0130] As Figure 1 shown, a bridge section monitoring and early warning and prevention system, comprising a data acquisition end, a data processing end, a first double-channel power supply module and a second double-channel power supply module;

[0131] The first double-channel power supply module is connected with the data processing end, for supplying power to the data processing end;

[0132] The second double-channel power supply module is connected with the data acquisition end, for supplying power to the data acquisition end;

[0133] The data acquisition end comprises an image processing device 4, a sensor 5, a video sensing device 18 and a transparent data sending device;

[0134] The data processing end comprises a core control unit 1, a warning screen controller 2, a bridge abnormal state warning screen 3 and a transparent data receiving device;

[0135] The image processing device 4 is connected with the transparent data receiving device and the video sensing device 18 respectively;

[0136] The video sensing device 18 is used to automatically intercept the abnormal event video stream after detecting the bridge section vehicle abnormal event and the bridge section meteorological disaster event, and transmit it to the image processing device 4;

[0137] The image processing device 4 is used to identify and classify the abnormal event video stream, which is classified into two categories of bridge section vehicle abnormal event and bridge section meteorological disaster event, and identify the severity of the bridge section meteorological disaster, and transmit the classification and identification results to the transparent data receiving device;

[0138] As Figure 5As shown, the transparent data sending device includes a sensor data connection port 10, a data format converter 11 and a transparent data transmitter 12; the data format converter 11 is connected with the sensor data connection port 10, the transparent data transmitter 12 and the second double-channel power supply module respectively; the sensor 5 is connected with the sensor data connection port 10;

[0139] The sensor 5 is used to collect displacement data, strain data and settlement data obtained by detecting the bridge, and then transmit the collected original message data of these data to the data format converter 11 through the sensor data connection port 10;

[0140] The data format converter 11 is used to analyze and convert the original message data collected by the sensor 5;

[0141] The transparent data transmitter 12 is used to send the data analyzed and converted by the data format converter 11 to the transparent data receiving device;

[0142] The core control unit 1 is connected with the early warning screen controller 2 and the transparent data receiving device respectively; the early warning screen controller 2 is also connected with the bridge abnormal state early warning screen 3;

[0143] The core control unit 1 is used to obtain the data received by the transparent data receiving device, and calculate according to the data received by the transparent data receiving device to obtain the severity of the bridge itself;

[0144] The core control unit 1 outputs the severity of the bridge itself and the meteorological severity of the bridge section to the early warning screen controller 2, and then the early warning screen controller 2 displays the early warning information to the passing vehicles through the bridge abnormal state early warning screen 3 according to the severity of the bridge itself and the meteorological severity of the bridge section.

[0145] The vehicle abnormal driving behavior includes parking and accident; the bridge section meteorological disaster includes fog and icing;

[0146] The transparent data sending device and the transparent data receiving device communicate with each other by at least one of RS485, RS232, NB-IOT and 433MHz communication modes.

[0147] The data processing end further includes a data uploading device 16;

[0148] The data uploading device 16 is connected with the core control unit 1;

[0149] The data uploading device 16 is configured to transmit the bridge self severity and the bridge section weather severity calculated by the core control unit 1 to the remote control center, and is also configured to transmit the data received by the transparent data receiving device to the remote control center, including the data collected by the sensor and the abnormal event video stream processed by the image processing device 4 and the result of the processing.

[0150] The data processing end further comprises a fault troubleshooting device 17, which is connected with the core control unit 1 and the data uploading device 16.

[0151] When the video sensing device 18 fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the image processing device 4, the transparent data receiving device and the core control unit 1 in sequence.

[0152] When the image processing device 4 fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the transparent data receiving device and the core control unit 1 in sequence.

[0153] When the sensor 5 fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the transparent data sending device, the transparent data receiving device and the core control unit 1 in sequence.

[0154] When the transparent data sending device fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the transparent data receiving device and the core control unit 1 in sequence.

[0155] When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the core control unit 1.

[0156] When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the core control unit 1.

[0157] When the early warning screen controller 2 fails, the corresponding fault code is sent to the fault troubleshooting device 17 through the core control unit 1.

[0158] When the dual-channel power supply module fails, the corresponding fault code is sent to the fault troubleshooting device 17.

[0159] The fault troubleshooting device 17 displays the corresponding fault problem description according to the fault code, and sends the fault code to the remote control center through the data uploading device 16.

[0160] The fault code, the corresponding fault position and the fault problem description are shown in the following table.

[0161] Fault code Fault location Fault problem description F01-01 Sensor Sensor signal loss F01-02 Sensor Sensor data out of normal range F02-01 Image processing device Image cannot be recognized F03-01 Video sensing device Picture blurred or interfered F03-02 Video sensing device No video signal F04-01 Core control unit Instruction execution timeout or no response F05-01 Transparency data receiving device Incomplete received data F05-02 Transparency data sending device Data packet loss F06-01 Early warning screen controller Screen image display disorder F06-02 Early warning screen controller Early warning information refresh failure F07-01 First dual-channel power supply module No voltage and current data F07-02 Second dual-channel power supply module Voltage and current data abnormal .

[0162] The core control unit 1 is built-in with a mixed algorithm of Isolation Forest and LSTM, which firstly extracts a feature vector from the sensor data, then calculates an anomaly score by using the Isolation Forest, and then transmits the anomaly score of the Isolation Forest to the LSTM, taking the anomaly score and the feature vector as the input of the LSTM, and taking the classification of the persistent fault and the occasional interference as the output for training; if the mixed algorithm of Isolation Forest and LSTM identifies that there is a persistent fault in the bridge monitoring data, the severity of the bridge itself is located according to the sensor position corresponding to the persistent fault data, and finally the bridge itself severity warning information is reported.

[0163] It also includes an infrared fill light 24, which is used to fill light for the video perception device 18.

[0164] As shown in Figure 6 The transparent data receiving device includes a transparent data receiver 13, a data fusion unit 14 and a wired data transmitter 15; the transparent data receiver 13, the data fusion unit 14 and the wired data transmitter 15 are sequentially connected; the wired data transmitter 15 is connected with the core control unit 1 in a wired connection mode;

[0165] The transparent data receiver 13 is used to receive the data sent by the transparent data sending device and the image processing device 4;

[0166] The data fusion unit 14 is used to preprocess the data received by the transparent data receiver 13; the preprocessing includes separation, fusion, cleaning and summarization; when summarizing, the data is sequentially sorted and summarized according to the uplink and downlink directions of the bridge section;

[0167] The wired data transmitter 15 is used to send the data preprocessed by the data fusion unit 14 to the core control unit 1.

[0168] The first dual-channel power supply module includes a first mains interface 6, a first power supply channel intelligent conversion module 7, a first energy storage device 8 and a first solar power supply interface 9;

[0169] The first solar power supply interface 9 is connected with the first energy storage device 8;

[0170] The first power supply channel intelligent conversion module 7 is connected with the first mains interface 6 and the first energy storage device 8 respectively;

[0171] The first power supply channel intelligent conversion module 7 is used to switch the power supply mode, that is, to select whether to use solar power supply or mains power supply;

[0172] The first energy storage device 8 is used to store the electric energy converted from solar energy.

[0173] The second dual-channel power supply module comprises a second mains interface 19, a second power supply channel intelligent conversion module 20, a second energy storage device 21 and a second solar power supply interface 22.

[0174] The second solar power supply interface 22 is connected with the second energy storage device 21.

[0175] The second power supply channel intelligent conversion module 20 is connected with the second mains interface 19 and the second energy storage device 21 respectively.

[0176] The second power supply channel intelligent conversion module 20 is used for switching the power supply mode, i.e., selecting whether to use solar power supply or mains power supply.

[0177] The second energy storage device 21 is used for storing the electric energy converted from solar energy.

[0178] The infrared light supplement lamp 24 is installed on the horizontal arm 28, and the infrared light supplement lamp 24 is used for improving the monitoring ability of the highway bridge section in the case of sudden conditions in a dark environment.

[0179] The solar panel 23 is arranged in the direction of the sun, and in actual engineering application, adjustment is made according to the actual application scene.The solar panel 23 is connected with the first solar power supply interface 9.

[0180] The data processing end, the bridge abnormal state early warning screen 3, the solar panel 23, the infrared light supplement lamp 24, the horizontal arm 28, the video sensing device 18, the communication antenna 25, the power line 26 and the stand 27 are assembled into an integrated whole and are arranged on a roadbed section 500-1km away from the bridge section, as shown in Figure 7 The bridge abnormal state early warning screen 3 and the early warning screen controller 2 are in wireless communication.

[0181] In terms of data processing algorithms, the core control unit adopts a kind of isolated forest and LSTM hybrid algorithm.

[0182] The isolated forest and LSTM hybrid algorithm execution flow chart is shown in Figure 2

[0183] Specifically: the severity of the bridge itself is determined by a hybrid algorithm of Isolation Forest and LSTM, which collects data from sensors and consists of two parts of Isolation Forest algorithm and LSTM algorithm in the core control unit.

[0184] To quickly identify single-point or short-time anomalies (sporadic interference), the Isolation Forest algorithm is first used to detect transient anomalies, and statistical features (mean, variance, extreme value, etc.) are extracted from sensor data (vibration, stress, etc.) to form a feature vector x t ;

[0185] The feature vector x t is composed of statistical features and time series features, and the specific combination steps are as follows:

[0186] 1) According to the received sensor data, the feature vector x t is constructed according to the characteristics of data collected by different sensors (data usually contains time domain features and frequency domain features), which can be represented as:

[0187]

[0188] Where d is the feature dimension (i.e. the number of features extracted); x t is the t-th sample (time step feature vector), which is composed of d features These features can be various features (such as statistical features and frequency domain features) extracted from sensor data (such as vibration, stress, etc.); is the value of the d-th statistical feature at time t;

[0189] All the data sets to be detected form a sample set X:

[0190]

[0191] Where n is the number of historical samples;

[0192] After the feature vector x t is constructed, an Isolation Forest model is established, and φ samples are randomly selected to build μ isolated trees (this is done to increase the difference between each tree, avoid overfitting, and also improve the efficiency of the algorithm to handle large-scale data sets. In this algorithm, the sample number φ is selected as 256).

[0193] For each isolated tree T j (there are μ in total): randomly sample samples (without replacement) to form a sub-sample set Initialize the current node data set as X', the tree height h = 0, and the maximum height

[0194] If any of the termination conditions are met:

[0195] a) the current node sample size |X'| ≤ 1, or

[0196] b) the current tree height h ≥ l,

[0197] then create a leaf node and return.

[0198] Otherwise:

[0199] Randomly select a feature dimension q ∈ {1, 2, …, d} and a split value p ∈ [min(x q ), max(x q )] on that feature, where x q is the value of the q-th feature in the current sample set;

[0200] Split the samples:

[0201]

[0202] Recursively build the left and right subtrees with tree height h = h + 1; T left is the left subtree and T right is the right subtree;

[0203] Termination conditions:

[0204] The current node sample size |X| ≤ 1, or the current tree height h ≥ l;

[0205] Repeat the above process μ times to obtain μ independent isolated trees.

[0206] For a sample x t , substitute it into each isolated tree to simulate its traversal path from the root node to the leaf node, record the path length (edge number or node jump number) as h j (x t ), and finally take the average:

[0207]

[0208] The average path length c(φ) and the anomaly score s t are respectively

[0209]

[0210] Where H(k) ≈ lnk + γ, γ ≈ 0.5772;

[0211] s t ≈ 1: Highly abnormal (path too short)

[0212] s t ≈ 0.5: Close to normal;

[0213] s t <0.5: likely normal.

[0214] Set the minimum path distance as l min , the shorter the path length (less than l min ), the greater the degree of deviation, and vice versa. If h j (x t ) < l min , the anomaly score sequence {s t} and the sample x t are transmitted to the LSTM; otherwise, they are not transmitted.

[0215] After the above calculation is completed, the time duration of the abnormal interference is further captured by the LSTM algorithm, and then it is determined whether it is a persistent fault or an occasional interference.

[0216] The anomaly score sequence {s t} and the sample x t output by the previous isolation forest algorithm are spliced to form the input vector of the LSTM algorithm:

[0217] z t = [x t , s t ]

[0218] The input is divided into a time window τ

[0219] Z = [z t-τ , ···, z t ]

[0220] The forward calculation of the LSTM algorithm consists of three parts of the gate control unit, which are the forgetting gate, the input gate and the output gate. The calculation process of the forgetting gate satisfies the following formula:

[0221] f t = σ (W f · [h t-1 , z t ] + b f )

[0222] In the formula, σ is the activation function (sigmoid function); W f is the weight matrix of the forgetting gate; [h t-1 , z t ] is the connection of the two vectors to form a new vector; b f is the bias term of the forgetting gate; h t-1 represents the hidden state output at the previous time step (t-1).

[0223] Splicing operation:

[0224] The current input z t is an m-dimensional vector (feature number m) ;

[0225] The hidden state h t-1 of the last time is an n-dimensional vector (LSTM unit number n)

[0226] The linear transformation of the spliced vector is

[0227]

[0228] The dimension is (n+m) x 1, and the spliced vector is linearly transformed by the weight matrix, and then the calculation of the input gate is as follows:

[0229] i t = σ (W i · [h t-1 , z t ] + b i )

[0230] In the formula, W i is the weight matrix of the input gate; b i is the bias term of the input gate.

[0231] The calculation of the output gate is as follows:

[0232] o t = σ (W o · [h t-1 , z t ] + b o )

[0233] In the formula, W o is the weight matrix of the output gate; b o is the bias term of the output gate.

[0234] In addition, there is also a candidate memory state Z t in the LSTM algorithm, which is obtained according to the previous output and the current input, and satisfies the formula:

[0235]

[0236] In the formula, W z and b Z are the weight matrix and bias term of the output gate. Then the memory state z t is obtained, and satisfies the following formula:

[0237]

[0238] In the formula, f t is the forget gate; z t-1 is the unit state of the last time; i tis the input gate; and is the element-wise multiplication symbol.

[0239] The final output of the LSTM is determined by the output gate and the cell state, and the following formula can be obtained:

[0240] h t = o t tanh z t

[0241] In the formula, h t is the hidden state, and the output gate o t is responsible for determining whether the memory state is output.

[0242] After the above steps are completed, the method of "full connection layer + softmax output probability" is used to output the result, and the model is migrated from "feature extraction" to "classification decision". The process is as follows:

[0243] P t = softmax(W p h t +b p )

[0244] In the formula, P t is the probability; W p is the weight matrix; and b p is the bias. Then the formula is obtained:

[0245]

[0246] In the formula, y t is data in the interval range of {0, 1}, which represents sporadic interference or continuous interference data. The threshold data in actual use can be dynamically modified according to the actual application object. (For example: when the threshold is set to 0.5, if the final value is greater than 0.5, it is determined as a persistent fault, otherwise it is sporadic interference).

[0247] After the persistent fault is judged and recognized, the severity of the bridge itself is judged through the corresponding data of the sensor, and finally the bridge itself severity warning information is sent upward.

[0248] In order to verify the effect of the isolated forest and LSTM hybrid algorithm in data anomaly detection, the simulation data of the sensor (5) is verified as shown in Figure 3 .

[0249] After the data receiving device of the transparent transmission receives the data of the sensor, the data preprocessing is performed by the data fusion unit according to the data acquisition threshold interval of different sensors. After the above operation is completed, the data cleaning is completed by the isolated forest algorithm, and the identification of abnormal values (noise data) and interference data is realized; Figure 3(a) Isolation Forest algorithm is used to identify the data collected by the sensor. The vertical coordinate 0-10 represents the time step, and the horizontal coordinate 0-1000 represents the data number. Between 300-350 data points and 700-705 data points, the Isolation Forest algorithm calculates the anomaly degree of each data point through the “isolation” of the training data. The higher the anomaly score, the weaker the relationship between the data point and other points (i.e. more likely to be abnormal).

[0250] After completing Figure 3 (a) After data cleaning process, there are no outliers (noise signals) in the data, which provides a guarantee for the training accuracy of the data by the LSTM algorithm. Figure 3 (b) LSTM model prediction of abnormal patterns. The vertical coordinate (0.0-1.0) represents the threshold data interval (0.5 is taken as the threshold dividing point, the closer to 0 represents no abnormal situation, and the closer to 1 represents greater possibility of abnormality). The horizontal coordinate represents the number of 0-1000 data. From the training results in figure (b), it can be seen that there are real structural abnormal data between 200-400 data points, and the LSTM model correctly captures and distinguishes the continuous interference data between 600-800 data points.

[0251] Figure (c) is the control group, which only uses a single LSTM to train the data processed by the data fusion unit (14). The vertical coordinate (0.0-1.0) represents the threshold data interval (0.5 is taken as the threshold dividing point, the closer to 0 represents no abnormal situation, and the closer to 1 represents greater possibility of abnormality). The horizontal coordinate represents the number of 0-1000 data. From the figure, it can be seen that the

[0252] Summary: Figure 3 (a) Use Isolation Forest algorithm to monitor abnormal data points, get anomaly score, and normalize data;

[0253] Figure 3 (b) Pass the anomaly score as an input feature to the LSTM model to distinguish the abnormal patterns (continuous interference data and occasional interference data) in the time series;

[0254] In order to further verify the advantages of the Isolation Forest and LSTM hybrid algorithm in processing occasional interference data, the Isolation Forest and LSTM hybrid algorithm and the single LSTM algorithm are compared and quantitatively analyzed:

[0255] Figure 3(c) Directly use the raw data collected by the sensor (5) as input to train the LSTM model to distinguish abnormal patterns. The vertical coordinate is represented as a threshold data interval (the closer to 0 represents no abnormal situation, the closer to 1 represents the greater the possibility of abnormality), and the horizontal coordinate represents the data points of the sensor (5).

[0256] Results analysis:

[0257] In Figure 3 In (c), the single LSTM has a weak response to the sporadic abnormal data (700-705 sensor data points). The reason is that a single LSTM is better at processing long-term trend information, and the recognition efficiency for sporadic states is often not high. For persistent fault data, the learning ability of the LSTM on the persistent data (300-350 sensor data points) is often disturbed by external noise, especially in the case where the data is not clearly marked, the prediction performance will be affected.

[0258] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A bridge section monitoring, early warning and prevention system, characterized in that, The data acquisition end, the data processing end, the first double-channel power supply module and the second double-channel power supply module are included. The first double-channel power supply module is connected with the data processing end, and is used for supplying power for the data processing end. The second double-channel power supply module is connected with the data acquisition end, and is used for supplying power for the data acquisition end. The data acquisition end includes an image processing device (4), a sensor (5), a video sensing device (18) and a transparent data sending device. The data processing end includes a core control unit (1), a pre-warning screen controller (2), a bridge abnormal state pre-warning screen (3) and a transparent data receiving device. The image processing device (4) is connected with the transparent data receiving device and the video sensing device (18) respectively. The video sensing device (18) is used for automatically intercepting the video stream of the abnormal event after detecting the bridge section vehicle abnormal event and the bridge section meteorological disaster event, and transmitting the video stream to the image processing device (4). The image processing device (4) is used for identifying and classifying the abnormal event video stream, classifying the abnormal event video stream into two categories of bridge section vehicle abnormal event and bridge section meteorological disaster event, identifying the bridge section meteorological severity, and transmitting the classification and identification results to the transparent data receiving device. The transparent data sending device includes a sensing data connection port (10), a data format converter (11) and a transparent data sender (12); the data format converter (11) is connected with the sensing data connection port (10), the transparent data sender (12) and the second double-channel power supply module respectively; and the sensor (5) is connected with the sensing data connection port (10). The sensor (5) is used for collecting displacement data, strain data and settlement data obtained by detecting the bridge, and then transmitting the collected original message data of the data to the data format converter (11) through the sensing data connection port (10). The data format converter (11) is used for analyzing and converting the original message data collected by the sensor (5). The transparent data sender (12) is used for sending the data analyzed and converted by the data format converter (11) to the transparent data receiving device. The core control unit (1) is connected with the pre-warning screen controller (2) and the transparent data receiving device respectively; and the pre-warning screen controller (2) is also connected with the bridge abnormal state pre-warning screen (3). The core control unit (1) is used for obtaining the data received by the transparent data receiving device, and calculating according to the data received by the transparent data receiving device to obtain the bridge self severity. The core control unit (1) outputs the bridge self severity and the bridge section meteorological severity to the pre-warning screen controller (2), and then the pre-warning screen controller (2) displays the pre-warning information to the passing vehicles through the bridge abnormal state pre-warning screen (3) according to the bridge self severity and the bridge section meteorological severity.

2. The bridge section monitoring, early warning and prevention system according to claim 1, characterized in that, The vehicle abnormal driving behavior includes parking and accident; and the bridge section meteorological disaster includes fog and icing. At least one of RS485, RS232, NB-IOT and 433MHz communication modes is used for communication between the transparent data sending device and the transparent data receiving device.

3. The bridge section monitoring, early warning and prevention system of claim 1, wherein, The data processing end further includes a data uploading device (16). The data uploading device (16) is connected with the core control unit (1); The data uploading device (16) is used for transmitting the bridge self severity calculated by the core control unit (1) and the bridge section weather severity to the remote control center; and is also used for transmitting the data received by the transparent data receiving device to the remote control center, including the data collected by the sensor and the abnormal event video stream processed by the image processing device (4) and the result of the processing.

4. The bridge section monitoring, early warning and prevention system of claim 3, characterized in that, The data processing end further comprises a fault troubleshooting device (17); the fault troubleshooting device (17) is connected with the core control unit (1) and the data uploading device (16) respectively; When the video sensing device (18) fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the image processing device (4), the transparent data receiving device and the core control unit (1) in sequence; When the image processing device (4) fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the transparent data receiving device and the core control unit (1) in sequence; When the sensor (5) fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the transparent data sending device, the transparent data receiving device and the core control unit (1) in sequence; When the transparent data sending device fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the transparent data receiving device and the core control unit (1) in sequence; When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the core control unit (1); When the transparent data receiving device fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the core control unit (1); When the early warning screen controller (2) fails, the corresponding fault code is sent to the fault troubleshooting device (17) through the core control unit (1); When the double-channel power supply module fails, the corresponding fault code is sent to the fault troubleshooting device (17); The fault troubleshooting device (17) displays the corresponding fault problem description according to the fault code; meanwhile, the fault troubleshooting device (17) sends the fault code to the remote control center through the data uploading device (16).

5. The bridge section monitoring, early warning and prevention system according to claim 4, characterized in that, The fault code, the corresponding fault position and the fault problem description are shown in the following table: 。 6. The bridge section monitoring, early warning and prevention system according to claim 1, characterized in that, The core control unit (1) is internally provided with a mixed algorithm of isolated forest and LSTM, which firstly extracts a feature vector by using sensor data, then calculates an anomaly score by using isolated forest, and then transmits the anomaly score of isolated forest to LSTM, taking the anomaly score and the feature vector as the input of LSTM and taking the classification of persistent fault and occasional interference as the output for training; If the mixed algorithm of isolated forest and LSTM identifies that there is a persistent fault in the bridge monitoring data, the bridge self severity is located according to the sensor position corresponding to the persistent fault data, and finally the bridge self severity warning information is reported.

7. The bridge section monitoring, early warning and prevention system of claim 1, wherein, Further comprising an infrared fill light (24), which is used for light filling for the video sensing device (18).

8. The bridge section monitoring, early warning and prevention system of claim 1, wherein, The transparent data receiving device comprises a transparent data receiver (13), a data fusion unit (14) and a wired data transmitter (15); the transparent data receiver (13), the data fusion unit (14) and the wired data transmitter (15) are sequentially connected; the wired data transmitter (15) is connected to the core control unit (1) in a wired connection mode; The transparent data receiver (13) is used for receiving data sent by the transparent data sending device and the image processing device (4); The data fusion unit (14) is used for pre-processing data received by the transparent data receiver (13); the pre-processing includes separation, fusion, cleaning and summarization; when summarizing, the data is sequentially sorted and summarized according to the uplink and downlink directions of the bridge section; The wired data transmitter (15) is used for sending the pre-processed data of the data fusion unit (14) to the core control unit (1).

9. The bridge section monitoring, early warning and prevention system of claim 1, wherein, The first double-channel power supply module comprises a first mains interface (6), a first power supply channel intelligent conversion module (7), a first energy storage device (8) and a first solar power supply interface (9); The first solar power supply interface (9) is connected to the first energy storage device (8); The first power supply channel intelligent conversion module (7) is connected to the first mains interface (6) and the first energy storage device (8) respectively; The first power supply channel intelligent conversion module (7) is used for switching the power supply mode, that is, selecting solar power supply or mains power supply; The first energy storage device (8) is used for storing the electric energy converted from solar energy.

10. The bridge section monitoring, early warning and prevention system of claim 1, wherein, The second double-channel power supply module comprises a second mains interface (19), a second power supply channel intelligent conversion module (20), a second energy storage device (21) and a second solar power supply interface (22); The second solar power supply interface (22) is connected to the second energy storage device (21); The second power supply channel intelligent conversion module (20) is connected to the second mains interface (19) and the second energy storage device (21) respectively; The second power supply channel intelligent conversion module (20) is used for switching the power supply mode, that is, selecting solar power supply or mains power supply; The second energy storage device (21) is used for storing the electric energy converted from solar energy.

Images