Large shield main drive real-time health state monitoring method and system

Abstract

The invention provides a large shield main drive real-time health state monitoring method and system, and relates to the technical field of shield main drive state monitoring, and the method comprises the steps: obtaining an input signal comprising a stress wave signal SH0 in the horizontal direction of a main drive bearing, a stress wave signal SV0 in the vertical direction of the main drive bearing, and a current signal I0 of a shield machine PLC system; and the monitoring input signal is input into the diagnosis model, and the health state of the main bearing is judged in real time according to the prediction result of the diagnosis model. According to the invention, the stress wave sensor is innovatively combined to obtain the stress wave signal SH in the horizontal direction and the stress wave signal SV in the vertical direction, and the stress wave sensor is matched with the current signal I of the main drive, so that the comprehensive coverage monitoring of the main drive system of the shield tunneling machine is realized, the problem of insufficiency of the traditional single sensor monitoring means is solved, and the PCA is utilized to carry out multi-modal fusion. Comprehensive features are extracted, the accuracy of a fault diagnosis result is improved through an integrated learning diagnosis model, and the accuracy and stability of the diagnosis result are ensured.

Description

Technical Field

[0001] The invention relates to the technical field of shield main drive status monitoring, and in particular to a real-time health status monitoring method and system for a large shield main drive. Background Art

[0002] As a key equipment in modern tunnel engineering, the main drive system of a large shield machine is responsible for driving the cutterhead to rotate and overcoming the ground resistance. The health of the main drive system is directly related to the construction efficiency and engineering safety of the shield machine. However, due to the complex shield construction environment, the main drive system is in a high-load, low-speed and high-noise working state for a long time, which is prone to problems such as bearing wear, gear failure and motor failure. If these problems are not discovered in time, they may cause equipment shutdown or even cause serious engineering accidents.

[0003] At present, the monitoring technology for the health status of the main drive of large shield machines mainly includes single sensor monitoring and fault diagnosis technology based on traditional signal processing methods. For example, vibration signals are collected by acceleration sensors, and fault characteristics are identified through spectrum analysis or wavelet transform. However, this type of method has the following shortcomings: 1. Single signal limitation: The signal collected by a single sensor is difficult to fully reflect the health status of the main drive system, especially under the superposition of multiple fault modes or complex environmental noise interference, the diagnostic accuracy is significantly reduced; 2. Imperfect feature extraction method: Some feature extraction methods may be overly simplified, resulting in the loss of some important information in the original data, thereby affecting the accuracy of subsequent tasks; 3. Insufficient real-time performance: Traditional methods usually require offline processing and cannot achieve real-time monitoring and early warning of the health status of the main drive.

[0004] In recent years, the development of multi-sensor fusion technology and artificial intelligence algorithms has provided new ideas for main drive health status monitoring; by collecting multi-modal data such as vibration, temperature, and current through multiple sensors, and using machine learning algorithms to achieve feature fusion and intelligent diagnosis, the comprehensiveness and accuracy of monitoring can be effectively improved. However, the existing technology still has the following problems in application: data fusion difficulties: the fusion of multi-modal data needs to solve the synchronization, correlation and redundancy problems of different sensor data, and the robustness and efficiency of the fusion algorithm still need to be improved; model performance is limited: the existing diagnostic model has insufficient performance when processing non-stationary signals and small sample data, and it is difficult to adapt to the real-time monitoring needs under complex working conditions; lack of optimization strategy: there is a lack of effective weight optimization strategy for the decision fusion of multi-model diagnostic results, which leads to the accuracy and stability of the final diagnostic results. Therefore, there is an urgent need for a real-time health monitoring method based on multi-sensor and multi-modal fusion, combined with an integrated learning algorithm, to solve the problems of data fusion difficulties, insufficient diagnostic model performance and lack of decision optimization in the existing technology, so as to comprehensively improve the accuracy and real-time performance of the health monitoring of the main drive system of the large shield machine. Summary of the invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a real-time health status monitoring method and system for the main drive of a large shield machine, which can solve the problems of data fusion difficulty, insufficient diagnostic model performance and lack of decision optimization in the existing technology, so as to comprehensively improve the accuracy and real-time performance of the health monitoring of the main drive system of a large shield machine.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A real-time health status monitoring method for a main drive of a large shield machine comprises the following steps:

[0007] S1. Obtain the horizontal stress wave signal S at the main drive bearing H0 , vertical stress wave signal S V0 and the current signal I of the shield machine PLC system 0 ;

[0008] S2, inputting each signal obtained in step S1 into the diagnosis model, and judging the health status of the main bearing in real time according to the prediction result of the diagnosis model;

[0009] The construction of the diagnostic model includes the following steps:

[0010] Step 1: Obtain the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V and the current signal I of the shield machine PLC system, and the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V Perform minimum-maximum normalization processing on the current signal I of the shield machine PLC system;

[0011] Step 2: PCA (abbreviation of Principal Component Analysis) is used to analyze the horizontal stress wave signal S at the main drive bearing. H , vertical stress wave signal S V Dimension reduction is performed to convert the horizontal stress wave signal S H and the vertical stress wave signal S V Fusion into a single stress wave signal S F ;

[0012] Step 3: The fused stress wave signal S F Based on the sampling frequency and the defined time interval, it is divided into manageable sub-signal segments, generating a signal consisting of multiple sub-S F Dataset composed of fragments , and use the continuous wavelet packet method to Top Pair S F Transform to generate a two-dimensional time-frequency diagram;

[0013] Step 4: Train the generated two-dimensional time-frequency graph dataset on the Resnet-50 model to generate a ResNet model;

[0014] Step 5: Use the time-frequency characteristic index formula to calculate the time-frequency characteristic index from each stress wave signal S F Calculate 27 feature indexes in the data segment, use feature indexes to train the SVM model, and adopt a one-to-many strategy to generate the SVM model;

[0015] Step 6: Using the time-frequency characteristic index formula, 27 characteristic indexes 2 are extracted from each current signal I data segment, and the characteristic indexes 2 are used to train the decision tree model. The decision tree model gradually divides the characteristic space by constructing multiple branch nodes, thereby generating a decision tree model.

[0016] Step 7: Integrated learning diagnosis model: Use integrated learning to combine the above three models through decision-level fusion to improve the accuracy of fault diagnosis results; the prediction formula is as follows:

[0017] ①

[0018] In formula ①, P(x) is the final predicted category; is the class probability predicted by the ResNet model; is the class probability predicted by the SVM model; is the category probability predicted by the decision tree model; a is the weight coefficient assigned to the ResNet model; b is the weight coefficient assigned to the SVM model; and c is the weight coefficient assigned to the decision tree model.

[0019] A further technical solution of the present invention is as follows: In step 7 of S2, the values ​​of a, b and c are optimized by a genetic algorithm so that the diagnostic performance of the integrated model is optimized. The specific optimization model is as follows:

[0020] ②

[0021] ③

[0022] in, is the optimization objective function;

[0023] Defined as:

[0024] ④

[0025] Based on the predicted category Calculate the classification accuracy;

[0026] The definition of accuracy is:

[0027] ⑤

[0028] : Based on predicted category Calculated F1 score;

[0029] is defined as:

[0030] ⑥

[0031] in, ;

[0032] TP is the number of true positives; TN is the number of true negatives; FP is the number of false positives; FN is the number of false negatives; α and β are weight coefficients used to balance the importance of accuracy and F1 score;

[0033] The fusion process is optimized through genetic algorithm to determine the optimal weight coefficients a, b and c of the basic model.

[0034] Preferably, the step 1 also includes measuring the horizontal stress wave signal S at the main drive bearing. H , vertical stress wave signal S V And the current signal I of the shield machine PLC system is filtered.

[0035] Preferably, the cross entropy loss function used in the Resnet-50 model in step 4 is defined as follows:

[0036] ⑦

[0037] In formula ⑦, N is the total number of samples in the dataset; M is the total number of categories in the classification task; Indicator function, The value is 0 or 1. When the true category of sample i is c, is 1, otherwise is 0; pic is the predicted probability that sample i belongs to category c, provided by the softmax output.

[0038] Preferably, the 27 feature indicators 1 in step five include 14 time domain features and 13 frequency domain features, and the 27 feature indicators 2 in step six include 14 time domain features and 13 frequency domain features.

[0039] Another object of the present invention is to provide a real-time health status monitoring system for a large shield main drive, comprising: a stress wave sensor, the stress wave sensor being fixedly mounted on the outer ring end surface of the main bearing;

[0040] A dynamic data acquisition module, which is used to collect detection signals of stress wave sensors and current signals of a PLC system of a shield machine;

[0041] An edge computing device is communicatively connected to the dynamic data acquisition module, and the edge computing device is used to execute the above-mentioned diagnostic model.

[0042] Preferably, the stress wave sensor is connected to the dynamic data acquisition module via a shielded cable.

[0043] The present invention provides a method and system for real-time health status monitoring of a main drive of a large shield machine. It has the following beneficial effects:

[0044] 1. The present invention obtains the stress wave signal S in the horizontal direction by innovatively combining the stress wave sensor H , vertical stress wave signal S V The stress wave sensor is used to monitor the stress wave signal of the main bearing. Together with the current signal I of the main drive, it realizes comprehensive coverage monitoring of the main drive system of the shield machine, making up for the shortcomings of the traditional single sensor monitoring method; and the horizontal stress wave signal S H , vertical stress wave signal S V PCA is used for multi-modal fusion to extract comprehensive features; and through real-time online monitoring, early warning can be given before a fault occurs, avoiding downtime for maintenance due to sudden faults, greatly reducing unplanned downtime, thereby improving the efficiency and continuity of shield construction; the present invention uses an integrated learning diagnosis model to combine three models through decision-level fusion to improve the accuracy of fault diagnosis results. The fusion process is optimized through a genetic algorithm to determine the optimal weight coefficients a, b and c of the basic model, thereby ensuring the accuracy and stability of the diagnosis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A distribution diagram of three-way vibration sensors and stress wave sensors in a real-time health status monitoring system for a large shield main drive proposed by the present invention;

[0046] Figure 2 A schematic diagram of the distribution of three-way vibration sensors and stress wave sensors on the main bearing in a real-time health status monitoring system for the main drive of a large shield machine proposed by the present invention;

[0047] Figure 3A schematic diagram of PCA fusion in a real-time health status monitoring method for a main drive of a large shield machine proposed in the present invention;

[0048] Figure 4 The present invention proposes a method for real-time health status monitoring of the main drive of a large shield machine using a continuous wavelet packet method in the data set. Top Pair S F Transform to generate a two-dimensional time-frequency diagram;

[0049] Figure 5 This is a schematic diagram of the Resnet-50 model architecture in a real-time health status monitoring method for a large shield main drive proposed in the present invention;

[0050] Figure 6 A schematic diagram of an SVM model in a real-time health status monitoring method for a main drive of a large shield machine proposed in the present invention;

[0051] Figure 7 This is a schematic diagram of a decision tree model in a real-time health status monitoring method for a main drive of a large shield machine proposed in the present invention. DETAILED DESCRIPTION

[0052] 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.

[0053] Embodiment 1:

[0054] like Figure 1-Figure 7 As shown, an embodiment of the present invention provides a method for real-time health status monitoring of a main drive of a large shield machine, comprising the following steps:

[0055] S1. Obtain the horizontal stress wave signal S at the main drive bearing H0 , vertical stress wave signal S V0 and the current signal I of the shield machine PLC system 0 , horizontal stress wave signal S at the main drive bearing H0 , vertical stress wave signal S V0 and the current signal I of the shield machine PLC system 0 Constitute monitoring input signal;

[0056] S2. Input the input signal into the diagnostic model, and determine the health status of the main bearing in real time according to the prediction results of the diagnostic model.

[0057] The diagnostic model is executed by the edge computing device. After the input signal is input into the diagnostic model, the diagnostic model outputs the prediction result, which covers various health conditions of the main drive bearing (such as normal, slight damage, moderate damage, severe damage, and failure).

[0058] The construction of the diagnostic model includes the following steps:

[0059] Step 1: Obtain the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V and the current signal I of the shield machine PLC system, and the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V The current signal I of the shield machine PLC system is normalized to the minimum and maximum value.

[0060] ⑧

[0061] In formula 8, Describe the output signal after minimum-maximum normalization, Less than 1 and greater than 0, X represents the input signal, X max , X min Respectively express the maximum and minimum of the input signal X.

[0062] Step 2: Use PCA to measure the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V Dimension reduction is performed to convert the horizontal stress wave signal S H and the vertical stress wave signal S V Fusion into a single stress wave signal S F , which realizes the multimodal fusion of data and can extract comprehensive features.

[0063] Step 3: The fused stress wave signal S F Based on the sampling frequency and the defined time interval, it is divided into manageable sub-signal segments, generating a signal consisting of multiple sub-S F Dataset composed of fragments , and use the continuous wavelet packet method to Top Pair S F Transform to generate a two-dimensional time-frequency diagram.

[0064] Step 4: Train the generated two-dimensional time-frequency graph dataset on the Resnet-50 model to generate a ResNet model.

[0065] The cross entropy loss function used in the Resnet-50 model is defined as follows:

[0066] ⑦

[0067] In formula ⑦, N is the total number of samples in the dataset; M is the total number of categories in the classification task; Indicator function, The value is 0 or 1. When the true category of sample i is c, is 1, otherwise is 0; pic is the predicted probability that sample i belongs to category c, provided by the softmax output.

[0068] Based on historical operation data and expert experience, a database containing 27 characteristic values ​​(time domain, frequency domain, time-frequency domain characteristics) can be constructed, covering different health states of the main drive bearing (such as normal, slight damage, moderate damage, severe damage, and failure);

[0069] Label definition: The health status of the main bearing is divided into five categories as the output labels of the model.

[0070] The 27 eigenvalues ​​include 14 time domain features and 13 frequency domain features; Table 1 is the time domain index, and Table 2 is the frequency domain index; Table 1 Time domain indicators

[0071] Table 2 Frequency domain indicators

[0072] Step 5: Use the time-frequency characteristic index formula to calculate the time-frequency characteristic index from each stress wave signal S F Calculate 27 feature indexes in the data segment, use feature indexes to train the SVM model, and use a one-to-many strategy to generate the SVM model in order to achieve multi-classification tasks;

[0073] Step 6. Use the time-frequency feature index formula to extract 27 feature indexes 2 from each current signal I data segment, and use feature index 2 to train the decision tree model. To achieve multi-classification tasks, the decision tree model constructs multiple branch nodes and gradually divides the feature space to generate the third model.

[0074] Step 7: Ensemble learning diagnosis model: Use ensemble learning to combine the above three models through decision-level fusion to improve the accuracy of fault diagnosis results;

[0075] ①

[0076] In formula ①, P(x) is the final predicted category; P ResNet (x) is the class probability predicted by the ResNet model; PSVM (x) is the class probability predicted by the SVM model; P DT (x) is the category probability predicted by the decision tree model; a is the weight coefficient assigned to the ResNet model; b is the weight coefficient assigned to the SVM model; c is the weight coefficient assigned to the decision tree model;

[0077] The values ​​of a, b and c are optimized by genetic algorithm to achieve the best diagnostic performance of the integrated model. The specific optimization model is as follows:

[0078] ②

[0079] ③

[0080] in, is the optimization objective function;

[0081] Defined as:

[0082] ④

[0083] Based on the predicted category Calculate the classification accuracy;

[0084] The definition of accuracy is:

[0085] ⑤

[0086] : Based on predicted category Calculated F1 score;

[0087] is defined as:

[0088] ⑥

[0089] in, ;

[0090] TP is the number of true positives; TN is the number of true negatives; FP is the number of false positives; FN is the number of false negatives; α and β are weight coefficients used to balance the importance of accuracy and F1 score;

[0091] The fusion process is optimized through genetic algorithm to determine the optimal weight coefficients a, b and c of the basic model.

[0092] The first step also includes: detecting the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S VAnd the current signal I of the shield machine PLC system is filtered.

[0093] Real-time monitoring and early warning: Deploy the diagnostic model in edge computing devices to achieve real-time monitoring of the health status of the main drive system of the large shield machine and fault early warning.

[0094] The model training process mainly includes: data processing and offline training:

[0095] Data processing

[0096] The main drive main bearing database is constructed using historical operation data, including stress wave signals, current signals and fault labeling information; a training data set is generated through steps such as data cleaning and feature extraction (such as time domain, frequency domain and time-frequency domain features).

[0097] Model training:

[0098] An ensemble learning method is used to combine multimodal data for offline training and optimize model parameters to improve the accuracy and robustness of fault diagnosis.

[0099] Embodiment 2:

[0100] like Figure 1-Figure 2 As shown, an embodiment of the present invention provides a real-time health status monitoring system for a main drive of a large shield machine, including: a stress wave sensor, a dynamic data acquisition module and an edge computing device.

[0101] The stress wave sensor is fixedly installed on the outer ring end face of the main bearing. The stress wave sensor captures the early fault signal of the main bearing (such as high-frequency signals generated by friction and impact). For example, the stress wave sensor is model 370A, has a high-frequency response (>10 kHz), and a sensitivity of 11.5 mV / g. The dynamic data acquisition module is used to collect the detection signal of the stress wave sensor and the current signal of the shield machine PLC system. The VSE153 vibration analysis module can be used to support 0-12000 Hz signal acquisition, including stress wave and vibration data channels. The shield machine PLC system is equipped with a DC power supply and a control cabinet, which can detect the current signal of the shield machine PLC system and obtain the current signal I of the shield machine PLC system. The edge computing device is connected to the dynamic data acquisition module for communication. The edge computing device is used to execute the diagnostic model in the first embodiment, that is, according to the real-time acquisition of the horizontal stress wave signal S at the main drive bearing H0 , vertical stress wave signal S V0 and the current signal I of the shield machine PLC system 0， It is input into the diagnostic model, and the output of the diagnostic model is used to judge the health status of the main drive of the large shield machine.

[0102] Stress wave sensors are installed on the main bearing and the main drive motor of the shield machine. The stress wave sensors are connected to the dynamic data acquisition module through shielded cables. The signals are converted into digital signals after amplification and filtering. The dynamic acquisition module transmits the data to the PLC controller through the Modbus TCP protocol, and then transmits it to the remote monitoring server through the optical fiber network. The data server and Web server are responsible for real-time data storage and display of analysis results respectively.

[0103] In this solution, stress wave sensors with high durability are selected. For example, they have a temperature resistance range of -40°C to +150°C, moisture resistance of IP67 protection level, and can withstand vibration shocks of more than 10 g. They are suitable for shield construction environments with high noise, high humidity, high temperature and strong vibration. At the same time, the stress wave sensor signal processing module has a background noise suppression ratio of 40 dB, which effectively filters environmental interference and ensures the accuracy and reliability of monitoring data.

[0104] In this solution, the stress wave sensor has high sensitivity (10 mV / g) and wide frequency response range (1 kHz to 1MHz), which is suitable for capturing low-frequency and tiny fault signals of the main bearing; the vibration sensor has a sensitivity of 100 mV / g and a frequency range of 10Hz to 10 kHz, which can accurately monitor the vibration status of the motor. The two work together to increase the fault detection rate to more than 95%, significantly enhancing the accuracy and reliability of diagnosis.

[0105] System operation steps

[0106] A stress wave sensor is arranged on the end face of the main bearing outer ring and connected to the dynamic data acquisition module. After hardware debugging is completed and the system is started, the stress wave sensor continuously collects physical quantity signals and displays the operating status of the main bearing and motor in real time. The system performs multi-factor trend analysis on the collected signals, determines the fault location and type, and generates a diagnostic report. It provides equipment maintenance suggestions based on the monitoring results and regularly updates sensor calibration parameters and system software.

[0107] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A real-time health status monitoring method for a large shield main drive, characterized in that: The following steps are involved: S1. Obtain the horizontal stress wave signal S at the main drive bearing H0 , vertical stress wave signal S V0 and the current signal I0 of the shield machine PLC system; S2, inputting each signal obtained in step S1 into the diagnosis model, and judging the health status of the main bearing in real time according to the prediction result of the diagnosis model; The construction of the diagnostic model includes the following steps: Step 1: Obtain the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V and the current signal I of the shield machine PLC system, and the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V Perform minimum-maximum normalization processing on the current signal I of the shield machine PLC system; Step 2: Use PCA to measure the horizontal stress wave signal S at the main drive bearing H , vertical stress wave signal S V Dimension reduction is performed to convert the horizontal stress wave signal S H and the vertical stress wave signal S V Fusion into a single stress wave signal S F ; Step 3: The fused stress wave signal S F Based on the sampling frequency and the defined time interval, it is divided into manageable sub-signal segments, generating a signal consisting of multiple sub-S F Dataset composed of fragments , and use the continuous wavelet packet method to Top Pair S F Transform to generate a two-dimensional time-frequency diagram; Step 4: Train the generated two-dimensional time-frequency graph dataset on the Resnet-50 model to generate a ResNet model; Step 5: Use the time-frequency characteristic index formula to calculate the time-frequency characteristic index from each stress wave signal S F Calculate 27 feature indexes in the data segment, use feature indexes to train the SVM model, and adopt a one-to-many strategy to generate the SVM model; Step 6: Using the time-frequency characteristic index formula, 27 characteristic indexes 2 are extracted from each current signal I data segment, and the characteristic indexes 2 are used to train the decision tree model. The decision tree model gradually divides the characteristic space by constructing multiple branch nodes, thereby generating a decision tree model. Step 7: Ensemble learning diagnosis model: Ensemble learning is used to combine the above three models through decision-level fusion to improve the accuracy of fault diagnosis results; its prediction formula is as follows: ① In formula ①, P(x) is the final predicted category; is the class probability predicted by the ResNet model; is the class probability predicted by the SVM model; is the category probability predicted by the decision tree model; a is the weight coefficient assigned to the ResNet model; b is the weight coefficient assigned to the SVM model; and c is the weight coefficient assigned to the decision tree model.

2. A method for real-time health status monitoring of a large shield main drive according to claim 1, characterized in that: In step 7 of S2, the values ​​of a, b and c are optimized by genetic algorithm to optimize the diagnostic performance of the integrated model. The specific optimization model is as follows: The values ​​of a, b and c are optimized by genetic algorithm to optimize the diagnostic performance of the integrated model. The specific optimization model is as follows: ② ③ in, is the optimization objective function; Defined as: ④ Based on the predicted category Calculate the classification accuracy; The definition of accuracy is: ⑤ : Based on predicted category Calculated F1 score; is defined as: ⑥ in, ; TP is the number of true positives; TN is the number of true negatives; FP is the number of false positives; FN is the number of false negatives; α and β are weight coefficients used to balance the importance of accuracy and F1 score; The fusion process is optimized through genetic algorithm to determine the optimal weight coefficients a, b and c of the basic model.

3. A method for real-time health status monitoring of a main drive of a large shield machine according to claim 1 or 2, characterized in that: The step 1 also includes a stress wave signal S in the horizontal direction at the main drive bearing. H , vertical stress wave signal S V And the current signal I of the shield machine PLC system is filtered.

4. A method for real-time health status monitoring of a main drive of a large shield machine according to claim 1 or 2, characterized in that: The cross entropy loss function used in the Resnet-50 model in step 4 is defined as follows: ⑦ In formula ⑦, N is the total number of samples in the dataset; M is the total number of categories in the classification task; Indicator function, The value is 0 or 1. When the true category of sample i is c, is 1, otherwise is 0; pic is the predicted probability that sample i belongs to category c, provided by the softmax output.

5. A method for real-time health status monitoring of a main drive of a large shield machine according to claim 1 or 2, characterized in that: The 27 feature indicators 1 in step 5 include 14 time domain features and 13 frequency domain features, and the 27 feature indicators 2 in step 6 include 14 time domain features and 13 frequency domain features.

6. A real-time health status monitoring system for the main drive of a large shield machine, characterized in that: include: A stress wave sensor, wherein the stress wave sensor is fixedly mounted on the outer ring end surface of the main bearing; A dynamic data acquisition module, which is used to collect detection signals of stress wave sensors and current signals of a PLC system of a shield machine; An edge computing device, wherein the edge computing device is communicatively connected to the dynamic data acquisition module, and the edge computing device is used to execute the diagnostic model described in any one of claims 1-5.

7. A real-time health status monitoring system for main drive of a large shield machine according to claim 6, characterized in that: The stress wave sensor is connected to the dynamic data acquisition module via a shielded cable.

Images