A tunnel equipment and facility status monitoring and early warning system based on machine learning

Abstract

The present invention discloses a tunnel equipment and facility status monitoring and early warning system based on machine learning, comprising: a performance data acquisition module that acquires historical performance data of tunnel operation data monitoring equipment; an operation data monitoring module that acquires historical signal parameters of tunnel operation data monitoring equipment within a preset time period; an analysis module that determines whether the historical performance data of the operation monitoring equipment meets the requirements: if it does not meet the requirements, a first early warning signal is generated; if it meets the requirements, a risk analysis is performed on the historical signal parameters: when the risk analysis meets the requirements, an early warning analysis is performed; otherwise, a second early warning signal is generated; when the early warning analysis meets the requirements, a patrol and maintenance analysis is performed; otherwise, a third early warning signal is generated; a model construction module that inputs historical performance data into a data-driven model for training to construct a data-driven prediction model, and outputs prediction signal parameters; a patrol and maintenance module performs patrol and maintenance analysis, and obtains patrol and maintenance strategies based on the patrol and maintenance analysis results.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment status early warning, and in particular to a tunnel equipment facility status monitoring and early warning system based on machine learning. Background Art

[0002] The research on the accuracy of operational monitoring data collected by electromechanical system collection equipment in highway tunnels is carried out through the digitization and partial intelligence of equipment control and data monitoring management in tunnels. The main purpose is to solve the problems of discovering operational data monitoring faults of various tunnel equipment and facilities, predicting faulty equipment problems in advance and optimizing operational data.

[0003] At present, in order to improve the data accuracy of tunnel equipment and facility status monitoring, intelligent data collection and processing methods are used to realize the overall selection process of monitoring data types, characteristics and collection methods. However, most of the existing algorithms use a single detector to mark abnormal data and provide early warning feedback. This type of single early warning method is difficult to detect specific data problems, and it is difficult to provide effective solution signals when multiple operating monitoring equipment fails. In addition, if accurate early warning information is to be provided, it relies on the existing single early warning data analysis method based on machine learning technology, and a large amount of type data needs to be provided for fitting training at the same time, which can easily lead to problems such as overfitting and low efficiency.

[0004] The existing tunnel equipment and facility status monitoring and early warning system based on machine learning is difficult to provide timely and efficient accurate early warning feedback of equipment monitoring and operation data during highway tunnel operation, and also lacks feedback on the safety issues of the operating equipment itself, which limits the efficiency and level of highway tunnel operation management and control. Summary of the invention

[0005] The purpose of the present invention is to provide a tunnel equipment and facility status monitoring and early warning system based on machine learning to solve the following technical problems:

[0006] How to achieve efficient prediction and accurate feedback of equipment detection and operation data;

[0007] How to provide a timely feedback mechanism for operating equipment safety issues to improve the efficiency and level of highway tunnel operation management and control.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] A tunnel equipment and facility status monitoring and early warning system based on machine learning includes:

[0010] Performance data collection module, used to collect historical performance data of tunnel operation data monitoring equipment;

[0011] Historical performance data includes equipment failure records, equipment maintenance logs, and equipment monitoring data;

[0012] An operation data monitoring module is used to collect historical signal parameters of tunnel operation data monitoring equipment within a preset time period;

[0013] The historical signal parameters include temperature signal parameters and vibration signal parameters;

[0014] The analysis module is used to determine whether the historical performance data of the operation monitoring equipment meets the requirements:

[0015] If the requirements are not met, a first warning signal is generated;

[0016] If the requirements are met, risk analysis is performed on the historical signal parameters:

[0017] When the risk analysis meets the requirements, early warning analysis is conducted;

[0018] Otherwise, a second warning signal is generated;

[0019] When the early warning analysis meets the requirements, conduct patrol and maintenance analysis;

[0020] Otherwise, a third warning signal is generated;

[0021] A model building module is used to input historical performance data into a data-driven model for training to build a data-driven prediction model and output prediction signal parameters;

[0022] The patrol module is used to obtain the dynamic changes of the predicted signal parameters and the dynamic changes of the historical signal parameters for patrol analysis, and obtain the patrol strategy based on the analysis results;

[0023] The fault warning module is used to issue a first warning message, a second warning message and a third warning message.

[0024] Preferably, the risk analysis process of the analysis module includes:

[0025] Statistical history of signal parameters over time , ... ;

[0026] Will , ... Compare with the corresponding standard threshold ranges respectively:

[0027] like ∈ , then the risk analysis is judged to meet the requirements;

[0028] Otherwise, the risk analysis is judged to be unsatisfactory;

[0029] Where n is the number of signal parameters; is the curve of the i-th signal parameter, is the standard interval corresponding to the i-th signal parameter.

[0030] Preferably, the process of early warning analysis of the analysis module is:

[0031] Perform multimodal data feature extraction on each historical signal parameter to obtain the state monitoring coefficient of the corresponding item;

[0032] Analyze the time variation curve of each condition monitoring coefficient , ... ;

[0033] Will , ... Compare with the corresponding standard thresholds respectively:

[0034] like < , then it is judged that the early warning analysis meets the requirements;

[0035] Otherwise, it is judged that the early warning analysis does not meet the requirements;

[0036] in, is the curve of the i-th condition monitoring coefficient; is the standard threshold of the i-th condition monitoring coefficient.

[0037] Preferably, the process of constructing the prediction model by the model building module is:

[0038] Collect historical performance data of tunnel operation data monitoring equipment at several time points and signal parameters of performance-related items corresponding to the corresponding time points to generate training samples;

[0039] A data-driven model is built with a neural network structure, and the training samples of various signal parameters are input into the data-driven model for training, the predicted signal parameters are output, and the data-driven prediction model is updated.

[0040] Preferably, the patrol analysis process of the patrol module is: through the formula

[0041] Calculate the patrol and maintenance risk value ;

[0042] in, is the preset time period, and 0≤ < ; is the first conversion function; is the second conversion function; To predict the real-time changing parameter value of the temperature signal; To predict the real-time changing parameter values ​​of vibration signals; , Respectively represent the preset weight coefficients related to the influence of temperature signal and vibration signal; and , All are greater than 0; To predict the temperature signal; To predict vibration signals; is the standard temperature signal; is the standard vibration signal; is the patrol position coefficient;

[0043] is the judgment function, judging when ≥ When = ;otherwise, = ;

[0044] The allowed threshold value corresponding to the preset time period; It is the third warning value.

[0045] Preferably, the patrol and maintenance strategy is:

[0046] when < hour:

[0047] The patrol risk value The preset standard threshold range To compare:

[0048] like ∈ , it is judged that the current operation data monitoring equipment is operating normally;

[0049] like < , it is judged that the current operation data monitoring equipment operation and maintenance is abnormally low, and the inspection and maintenance frequency is increased;

[0050] like > , it is judged that the current operation data monitoring equipment operation and maintenance abnormality is high, and the patrol and maintenance frequency is reduced.

[0051] Preferably, it also includes:

[0052] The detection module is used to analyze and determine the health status of the equipment based on the historical performance data of the operation data monitoring equipment within a preset time period.

[0053] Preferably, the analysis process of the device health status of the detection module is:

[0054] By formula

[0055] Calculate the equipment health factor ;

[0056] in, Record parameters for device faults, Maintain log parameters for the device, Monitor data parameters for equipment; Record standard parameters for equipment failures, Maintain log standard parameters for the device, Provides standard parameters for equipment monitoring data; Record parameter deviation values ​​for preset equipment faults; Maintain log parameter deviation values ​​for preset equipment; It is the deviation value of the preset equipment monitoring data parameter; Signal environmental factors impact function recorded for equipment failure; Signal environment factors affecting functions of equipment maintenance logs; is the signal environment factor influence function of the equipment monitoring data; among them, is the ambient temperature signal; It is the environmental vibration signal.

[0057] Preferably, the equipment health factor Compare with preset standard threshold:

[0058] If the equipment health factor is greater than the preset standard threshold, it is determined that the operation data monitoring equipment needs to be repaired or replaced.

[0059] Beneficial effects of the present invention:

[0060] (1) The present invention uses an analysis module to perform an efficient prediction and accurate feedback process of equipment detection operation data, and judges whether the historical performance data of the operation monitoring equipment meets the requirements: if it does not meet the requirements, a first warning signal is generated; if it meets the requirements, a risk analysis is performed on the historical signal parameters: further, when the risk analysis meets the requirements, a warning analysis is performed; otherwise, a second warning signal is generated; in addition, when the warning analysis meets the requirements, a patrol analysis is performed; otherwise, a third warning signal is generated; risk analysis is helpful to make a preliminary risk judgment on the normal operation of the operation monitoring equipment in advance, so as to eliminate the risk conditions of equipment operation in advance.

[0061] (2) The present invention also sets a model building module based on machine learning to input historical performance data into a data-driven model for training and constructing a data-driven prediction model, and outputs prediction signal parameters; through early warning analysis, it is beneficial to conduct a specific analysis of the detailed description of historical signal parameters, so as to predict potential abnormal changes in the monitoring status and eliminate the occurrence of potential operating status risks in advance.

[0062] (3) The patrol module is used to obtain the dynamic changes of the predicted signal parameters and the dynamic changes of the historical signal parameters for patrol analysis, and the patrol strategy is obtained based on the analysis results; the patrol strategy is to judge the patrol situation at the current location based on the size of the real-time difference between the two signals, and further analyze the status of the patrol equipment to ensure that the abnormal operation and maintenance of the current operation data monitoring equipment is predicted in advance and a new patrol solution is provided or the patrol route is updated.

[0063] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0065] Figure 1 This is a module diagram of a tunnel equipment and facility status monitoring and early warning system based on machine learning according to the present invention. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] The existing tunnel equipment and facility status monitoring and early warning system based on machine learning is difficult to provide timely and efficient accurate early warning feedback of equipment monitoring and operation data during the operation of highway tunnels, and also lacks feedback on the safety issues of the operating equipment itself, which limits the efficiency and level of highway tunnel operation management and control.

[0068] To solve the above technical problems, please refer to Figure 1 As shown, the present invention is a tunnel equipment and facility status monitoring and early warning system based on machine learning, comprising:

[0069] Performance data collection module, used to collect historical performance data of tunnel operation data monitoring equipment;

[0070] Historical performance data includes equipment failure records, equipment maintenance logs, and equipment monitoring data;

[0071] An operation data monitoring module is used to collect historical signal parameters of tunnel operation data monitoring equipment within a preset time period;

[0072] The historical signal parameters include temperature signal parameters and vibration signal parameters;

[0073] The analysis module is used to determine whether the historical performance data of the operation monitoring equipment meets the requirements:

[0074] If the requirements are not met, a first warning signal is generated;

[0075] If the requirements are met, risk analysis is performed on the historical signal parameters:

[0076] When the risk analysis meets the requirements, early warning analysis is conducted;

[0077] Otherwise, a second warning signal is generated;

[0078] When the early warning analysis meets the requirements, conduct patrol and maintenance analysis;

[0079] Otherwise, a third warning signal is generated;

[0080] A model building module is used to input historical performance data into a data-driven model for training to build a data-driven prediction model and output prediction signal parameters;

[0081] The patrol module is used to obtain the dynamic changes of the predicted signal parameters and the dynamic changes of the historical signal parameters for patrol analysis, and obtain the patrol strategy based on the analysis results;

[0082] The fault warning module is used to issue a first warning message, a second warning message and a third warning message.

[0083] In the above technical scheme, the present invention provides six operation modules, which are: first, a performance data acquisition module is used to collect historical performance data, equipment fault records, equipment maintenance logs and equipment monitoring data of tunnel operation data monitoring equipment; second, an operation data monitoring module is used to collect historical signal parameters of tunnel operation data monitoring equipment within a preset time period; wherein, according to the equipment signal analysis system mainly including two aspects of temperature monitoring and vibration monitoring, therefore, by collecting the temperature data signal of the current equipment for transmission and collecting the vibration data obtained by connecting the current vibration monitoring equipment with the vibration measuring optical cable for transmission, as the main transmission operation characteristics of the equipment, therefore, the signal parameters include temperature signal parameters and vibration signal parameters.

[0084] Third, the analysis module is used to determine whether the historical performance data of the operation monitoring equipment meets the requirements: if it does not meet the requirements, a first warning signal is generated; if it meets the requirements, a risk analysis is performed on the historical signal parameters: further, when the risk analysis meets the requirements, a warning analysis is performed; otherwise, a second warning signal is generated; in addition, when the warning analysis meets the requirements, a patrol and maintenance analysis is performed; otherwise, a third warning signal is generated; fourth, the model construction module is used to input the historical performance data into the data-driven model for training to build a data-driven prediction model, and output the prediction signal parameters; fifth, the patrol and maintenance module is used to obtain the dynamic changes of the prediction signal parameters and the dynamic changes of the historical signal parameters for patrol and maintenance analysis, and obtain the patrol and maintenance strategy according to the analysis results; sixth, the fault warning module is used to issue the first warning information, the second warning information and the third warning information.

[0085] As an implementation of the present invention, the risk analysis process of the analysis module includes:

[0086] Statistical history of signal parameters over time , ... ;

[0087] Will , ... Compare with the corresponding standard threshold ranges respectively:

[0088] like ∈ , then the risk analysis is judged to meet the requirements;

[0089] Otherwise, the risk analysis is judged to be unsatisfactory;

[0090] Where n is the number of signal parameters; is the curve of the i-th signal parameter, is the standard interval corresponding to the i-th signal parameter.

[0091] In the above technical solution, the risk analysis of the analysis module in this embodiment determines the risk change trend by judging the time change curve of the historical signal parameters, thereby ensuring timely signal warning and the next step of analysis process. Specifically, the historical signal parameter set is obtained by statistically analyzing the signal changes at different time points of the same device or the time changes of the facility signal monitored at the same location, and the historical signal parameter change curve over time is obtained. , ... ;Will , ... Compare the sizes with the corresponding standard threshold intervals and determine if ∈ , then the risk analysis meets the requirements; otherwise, it is judged that the risk analysis does not meet the requirements; where n is the number of signal parameters; is the curve of the i-th signal parameter, is the standard interval corresponding to the i-th signal parameter. Risk analysis is helpful to make a preliminary risk judgment on the normal operation of the operation monitoring equipment in advance, so as to eliminate the risk of equipment operation in advance. And the standard threshold interval here is also obtained in advance based on the normal operation signal of the historical tunnel monitoring equipment, as a reference to the normal range, which is the existing technology and will not be described in detail here.

[0092] As an implementation mode of the present invention, the process of early warning analysis of the analysis module is as follows:

[0093] Perform multimodal data feature extraction on each historical signal parameter to obtain the state monitoring coefficient of the corresponding item;

[0094] Analyze the time variation curve of each condition monitoring coefficient , ... ;

[0095] Will , ... Compare with the corresponding standard thresholds respectively:

[0096] like < , then it is judged that the early warning analysis meets the requirements;

[0097] Otherwise, it is judged that the early warning analysis does not meet the requirements;

[0098] in, is the curve of the i-th condition monitoring coefficient; is the standard threshold of the i-th condition monitoring coefficient.

[0099] In the above technical solution, the early warning analysis of the analysis module in this embodiment is a process of further analysis after the reasonable interval analysis of the normal range of risk analysis. Specifically, the multimodal data feature extraction is performed on each historical signal parameter to obtain the state monitoring coefficient of the corresponding item; then, the time variation curve of each state monitoring coefficient is analyzed. , ... ; Finally, , ... Compare the sizes with the corresponding standard thresholds respectively to determine if < , then the early warning analysis is judged to meet the requirements; otherwise, the early warning analysis is judged not to meet the requirements; among them, is the curve of the i-th condition monitoring coefficient; is the standard threshold of the i-th condition monitoring coefficient; early warning analysis is conducive to a specific analysis of the historical signal parameter details, which is convenient for predicting potential abnormal changes in the monitoring status and eliminating the occurrence of potential operating status risks in advance. The standard threshold here It is the result of fitting based on empirical data and will not be described in detail here.

[0100] As an implementation of the present invention, the process of constructing a prediction model by the model construction module is as follows:

[0101] Collect historical performance data of tunnel operation data monitoring equipment at several time points and signal parameters of performance-related items corresponding to the corresponding time points to generate training samples;

[0102] A data-driven model is built with a neural network structure, and the training samples of various signal parameters are input into the data-driven model for training, the predicted signal parameters are output, and the data-driven prediction model is updated.

[0103] In the above technical scheme, in this embodiment, by combining the analysis results of the analysis module, the historical performance data that meets the requirements is extracted, and the signal parameters corresponding to the historical performance data at the corresponding time points are collected. Through machine training, training samples are generated as the input information of the model, and a data-driven model is built with a neural network structure. Then, the training samples of various signal parameters are input as input information into the data-driven model for training through a machine learning-based method. After training, the predicted signal parameters are output, and the updated data-driven prediction model is obtained. The process of obtaining the data-driven prediction model can form the conditions for continuing the patrol and maintenance analysis in the next step. The above-mentioned data-driven model is determined by building a conventional convolutional neural network model and generating training samples, and data training and learning are all implemented by conventional machine learning algorithms in the prior art, which will not be described in detail here.

[0104] As an implementation mode of the present invention, the patrol analysis process of the patrol module is as follows:

[0105] Calculate the patrol and maintenance risk value ;

[0106] in, is the preset time period, and 0≤ < ; is the first conversion function; is the second conversion function; To predict the real-time changing parameter value of the temperature signal; To predict the real-time changing parameter values ​​of vibration signals; , Respectively represent the preset weight coefficients related to the influence of temperature signal and vibration signal; and , All are greater than 0; To predict the temperature signal; To predict vibration signals; is the standard temperature signal; is the standard vibration signal; is the patrol position coefficient; is the judgment function, judging when

[0107] ≥ When = ;

[0108] otherwise, = ;

[0109] The allowed threshold value corresponding to the preset time period; It is the third warning value.

[0110] In the above technical solution, the specific process of patrol analysis performed by the patrol module in this embodiment is to calculate the formula Calculate the patrol and maintenance risk value It is realized, and through the two indicators of signal parameters, namely temperature signal and vibration signal, further data processing and analysis process, the predicted real-time change parameter value of temperature signal and the predicted real-time change parameter value of vibration signal are obtained respectively; and the current position patrol and maintenance situation is judged according to the size of the real-time difference between the two signals, and the patrol and maintenance equipment status is further analyzed to ensure that the operation and maintenance abnormalities of the current operation data monitoring equipment are predicted in advance and new patrol and maintenance solutions are provided or the patrol and maintenance routes are updated.

[0111] It should be noted that the first conversion function in the formula , the second conversion function They are all fitted based on the test data under the normal operation and maintenance status of the current operating equipment. Therefore, the real-time change parameter values ​​of the temperature signal and the real-time change parameter values ​​of the vibration signal are calculated and processed by the corresponding conversion function to ensure the size of the patrol and maintenance risk value obtained for early warning.

[0112] And, the standard temperature signal , Standard vibration signal , patrol position coefficient All settings are selected based on historical experience data. They can be selected and adjusted according to the patrol status. They will not be described in detail here. In addition, during the analysis process, when the third warning value appears When > .

[0113] As an implementation mode of the present invention, the patrol and maintenance strategy is:

[0114] when < hour:

[0115] The patrol risk value The preset standard threshold range To compare:

[0116] like ∈ , it is judged that the current operation data monitoring equipment is operating normally;

[0117] like < , it is judged that the current operation data monitoring equipment operation and maintenance is abnormally low, and the inspection and maintenance frequency is increased;

[0118] like > , it is judged that the current operation data monitoring equipment operation and maintenance abnormality is high, and the patrol and maintenance frequency is reduced.

[0119] In the above technical solution, this embodiment further determines the situation where the third warning signal does not appear: determine the patrol risk value The preset standard threshold range Compare: If ∈ , then it is judged that the current operation data monitoring equipment is operating normally; if < , then the current operation data monitoring equipment operation and maintenance is judged to be abnormally low, and the patrol and maintenance frequency is increased; if > , it is judged that the current operation data monitoring equipment operation and maintenance abnormality is high, and the patrol and maintenance frequency is reduced.

[0120] As an embodiment of the present invention, it also includes:

[0121] The detection module is used to analyze and determine the health status of the equipment based on the historical performance data of the operation data monitoring equipment within a preset time period.

[0122] In the above technical scheme, a detection module is also provided in one embodiment of the present invention, which analyzes and determines the health status of the equipment based on the historical performance data of the operation data monitoring equipment within a preset time period, and provides a timely feedback mechanism for the safety issues of the operation equipment, so as to improve the efficiency and level of operation and control of highway tunnels.

[0123] As an implementation mode of the present invention, the analysis process of the device health status of the detection module is as follows:

[0124] By formula

[0125] Calculate the equipment health factor ;

[0126] in, Record parameters for device faults, Maintain log parameters for the device, Monitor data parameters for equipment; Record standard parameters for equipment failures, Maintain log standard parameters for the device, Provides standard parameters for equipment monitoring data; Record parameter deviation values ​​for preset equipment faults; Maintain log parameter deviation values ​​for preset equipment; It is the deviation value of the preset equipment monitoring data parameter; Signal environmental factors impact function recorded for equipment failure; Signal environment factors affecting functions of equipment maintenance logs; is the signal environment factor influence function of the equipment monitoring data; among them, is the ambient temperature signal; It is the environmental vibration signal.

[0127] In the above technical solution, the analysis of the device health status of the detection module in this embodiment is obtained by calculation through the formula: , calculate the equipment health factor ;

[0128] For the historical performance data of the tunnel operation data monitoring equipment, by analyzing the specific parameters contained in the historical performance data: equipment failure records, equipment maintenance logs and equipment monitoring data, we can further determine the influence relationship between the signal parameters of the equipment performance data, and by calculating the influence of the signal parameters of each performance data of the equipment and determining the health status of the equipment. Specifically, the signal environment factor influence function of the equipment failure record , Signal environment factors affecting function of equipment maintenance log , Signal environment factors affecting equipment monitoring data With ambient temperature signal and ambient vibration signals It is obtained based on the average value of each performance data of each device under the influence of different signal environments, and , , are obtained based on empirical data fitting, and , , These are also the results selected based on empirical data and will not be explained in detail here.

[0129] As an implementation mode of the present invention, the equipment health factor Compare with the preset standard threshold: If the equipment health factor is greater than the preset standard threshold, it is judged that the operation data monitoring equipment needs to be repaired or replaced.

[0130] In the above technical scheme, in this embodiment, the degree of equipment use and equipment durability are determined by analyzing the size of the equipment health coefficient. When the calculated equipment health coefficient size exceeds the preset standard threshold range, that is, by judging the operating data, the detection equipment needs to be replaced in time or further repaired and reinforced, so as to provide a feedback mechanism in time to improve the efficiency and level of highway tunnel operation management and control.

[0131] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0132] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined in this application, they shall all fall within the protection scope of the present invention.

Claims

1. A tunnel equipment and facility status monitoring and early warning system based on machine learning, characterized in that: include: Performance data collection module, used to collect historical performance data of tunnel operation data monitoring equipment; The historical performance data includes equipment failure records, equipment maintenance logs and equipment monitoring data; An operation data monitoring module is used to collect historical signal parameters of tunnel operation data monitoring equipment within a preset time period; The historical signal parameters include temperature signal parameters and vibration signal parameters; The analysis module is used to determine whether the historical performance data of the operation monitoring equipment meets the requirements: If the requirements are not met, a first warning signal is generated; If the requirements are met, risk analysis is performed on the historical signal parameters: When the risk analysis meets the requirements, early warning analysis is conducted; Otherwise, a second warning signal is generated; When the early warning analysis meets the requirements, conduct patrol and maintenance analysis; Otherwise, a third warning signal is generated; A model building module is used to input historical performance data into a data-driven model for training to build a data-driven prediction model and output prediction signal parameters; The patrol module is used to obtain the dynamic changes of the predicted signal parameters and the dynamic changes of the historical signal parameters for patrol analysis, and obtain the patrol strategy based on the analysis results; A fault warning module, used for issuing a first warning message, a second warning message and a third warning message; The patrol analysis process of the patrol module is as follows: Calculate the patrol and maintenance risk value ; in, is the preset time period, and 0≤ < ; is the first conversion function; is the second conversion function; To predict the real-time changing parameter value of the temperature signal; To predict the real-time changing parameter values ​​of vibration signals; , Respectively represent the preset weight coefficients related to the influence of temperature signal and vibration signal; and , All are greater than 0; To predict the temperature signal; To predict vibration signals; is the standard temperature signal; is the standard vibration signal; is the patrol position coefficient; is the judgment function, judging when ≥ When = D ; otherwise, = ; is the allowable threshold corresponding to the preset time period; D is the third warning value; The patrol and maintenance strategy is: when < hour: The patrol risk value The preset standard threshold range To compare: like ∈ , it is judged that the current operation data monitoring equipment is operating normally; like < , it is judged that the current operation data monitoring equipment operation and maintenance is abnormally low, and the inspection and maintenance frequency is increased; like > , it is judged that the current operation data monitoring equipment operation and maintenance abnormality is high, and the patrol and maintenance frequency is reduced.

2. According to the machine learning-based tunnel equipment and facility status monitoring and early warning system of claim 1, it is characterized in that: The risk analysis process of the analysis module includes: Statistical history of signal parameters over time , ... ; Will , ... Compare with the corresponding standard threshold ranges respectively: like ∈ , then the risk analysis is judged to meet the requirements; Otherwise, the risk analysis is judged to be unsatisfactory; Where n is the number of signal parameters; is the curve of the i-th signal parameter, is the standard interval corresponding to the i-th signal parameter.

3. The tunnel equipment and facility status monitoring and early warning system based on machine learning according to claim 2 is characterized in that: The process of early warning analysis of the analysis module is as follows: Perform multimodal data feature extraction on each historical signal parameter to obtain the state monitoring coefficient of the corresponding item; Analyze the time variation curve of each condition monitoring coefficient , ... ; Will , ... Compare with the corresponding standard thresholds respectively: like < , then it is judged that the early warning analysis meets the requirements; Otherwise, it is judged that the early warning analysis does not meet the requirements; in, is the curve of the i-th condition monitoring coefficient; is the standard threshold of the i-th condition monitoring coefficient.

4. According to the machine learning-based tunnel equipment and facility status monitoring and early warning system of claim 1, it is characterized in that: The process of constructing the prediction model by the model building module is as follows: Collect historical performance data of tunnel operation data monitoring equipment at several time points and signal parameters of performance-related items corresponding to the corresponding time points to generate training samples; A data-driven model is built with a neural network structure, and the training samples of various signal parameters are input into the data-driven model for training, the predicted signal parameters are output, and the data-driven prediction model is updated.

5. The tunnel equipment and facility status monitoring and early warning system based on machine learning according to claim 1 is characterized in that: Also includes: The detection module is used to analyze and determine the health status of the equipment based on the historical performance data of the operation data monitoring equipment within a preset time period.

6. The tunnel equipment and facility status monitoring and early warning system based on machine learning according to claim 5 is characterized in that: The analysis process of the device health status of the detection module is as follows: By formula , calculate the equipment health factor ;in, is the preset time period, and 0≤ < ; Record parameters for device faults, Maintain log parameters for the device, Monitor data parameters for equipment; Record standard parameters for equipment failures, Maintain log standard parameters for the device, Provides standard parameters for equipment monitoring data; Record parameter deviation values ​​for preset equipment faults; Maintain log parameter deviation values ​​for preset equipment; It is the deviation value of the preset equipment monitoring data parameter; Signal environmental factors impact function recorded for equipment failure; Signal environment factors affecting functions of equipment maintenance logs; is the signal environment factor influence function of the equipment monitoring data; among them, is the ambient temperature signal; It is the environmental vibration signal.

7. The tunnel equipment and facility status monitoring and early warning system based on machine learning according to claim 6 is characterized in that: Equipment health factor Compare with preset standard threshold: If the equipment health factor is greater than the preset standard threshold, it is determined that the operation data monitoring equipment needs to be repaired or replaced.

Images