GBDT-based fault monitoring method for screw conveyor shaft of shield machine

By using a torque prediction model based on GBDT to monitor the operating status of the tunnel boring machine's screw conveyor shaft, the problem of difficulty in timely monitoring of screw conveyor shaft faults in existing technologies has been solved, enabling efficient fault early warning and equipment maintenance.

CN114837686BActive Publication Date: 2024-11-26WUXI XUELANG DIGITAL TECH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210529744.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-16
Publication Date
2024-11-26
Estimated Expiration
2042-05-16

AI Technical Summary

Technical Problem

Existing technologies make it difficult to monitor the operating status of the tunnel boring machine's screw conveyor shaft in a timely and accurate manner, leading to potential equipment malfunctions and impacting construction progress.

Method used

A torque prediction model based on GBDT is adopted, which uses real-time operating data of the target working parameters of the tunnel boring machine to identify the fault of the screw conveyor shaft by the deviation between the predicted and actual torque values, and combines the health index for real-time monitoring and alarm.

Benefits of technology

It achieves high stability and real-time monitoring of the screw conveyor shaft, provides timely alarms, avoids serious equipment failures, and reduces construction risks and costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114837686B_ABST
    Figure CN114837686B_ABST
Patent Text Reader

Abstract

The present invention discloses a GBDT-based shield machine screw conveyor shaft fault monitoring method, which relates to the shield machine field. The method uses a pre-trained torque prediction model combined with real-time operation data of the target working parameters of the shield machine. According to the deviation between the real-time torque actual value and the torque prediction value of the screw conveyor shaft, the health of the running state of the screw conveyor shaft can be monitored in real time during the shield machine construction process. The method has the advantages of high stability and high real-time performance, and can generate an alarm for the failure of the screw conveyor shaft, and even generate an alarm when an early failure occurs. Through the monitoring function of the system of the present invention, on-site construction personnel can be assisted to judge the current state, and serious equipment failures can be avoided to a certain extent to cause significant losses.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of shield machines, and in particular to a GBDT-based shield machine screw conveyor shaft fault monitoring method. Background Art

[0002] The shield machine is a tunnel construction engineering machine. Compared with the original drilling and blasting method, the shield machine is fast and efficient, safe and reliable, and has better construction quality. It has now become the first choice for tunnel construction. The screw conveyor shaft is an important part of the shield machine. The screw conveyor shaft is the earth discharge device of the shield machine. Its main function is to discharge the soil cut by the cutter head from the earth bin of the shield machine.

[0003] During the use of the shield machine, the screw conveyor shaft needs to bear great stress and impact loads due to working conditions, fatigue, improper manual operation and other reasons, and the material is prone to fracture and shaft breakage. At present, workers usually judge whether the screw conveyor shaft has a broken shaft by on-site observation. It is difficult for workers to pay attention to the status of the screw conveyor shaft in time around the clock. When the screw conveyor shaft is found to be abnormal, serious consequences have often occurred, which not only affects the advancement of the on-site construction process, but may also pose certain safety hazards. Summary of the invention

[0004] In view of the above problems and technical requirements, the inventors have proposed a GBDT-based screw conveyor shaft fault monitoring method for a shield machine. The technical solution of the present invention is as follows:

[0005] A GBDT-based shield machine screw conveyor shaft fault monitoring method, the method comprising:

[0006] Acquire real-time operating data of the target operating parameters of the shield machine, where the target operating parameters are the operating parameters that reflect the operating status of the screw conveyor shaft among all the operating parameters of the shield machine;

[0007] The real-time operation data of the acquired target working parameters are input into the torque prediction model, and the output of the torque prediction model is used as the torque prediction value of the screw conveyor shaft; the torque prediction model is obtained by training based on the GBDT model using the historical operation data of the shield machine without faults in the stable operation state of the screw conveyor shaft, and the torque prediction model is used to reflect the relationship between the target working parameters of the shield machine without faults in the stable operation state of the screw conveyor shaft and the actual value of the torque of the screw conveyor shaft;

[0008] The fault of the screw conveyor is identified according to the deviation between the real-time actual torque value and the torque predicted value of the screw conveyor shaft.

[0009] A further technical solution is that the shield machine includes several supporting components, the screw conveyor shaft is one of the supporting components, and the target working parameters include the working parameters of the screw conveyor shaft and the working parameters of other supporting components in the shield machine except the screw conveyor shaft.

[0010] A further technical solution is that the working parameters of the screw conveyor shaft include the speed, pressure, oil replenishment pressure, oil leakage temperature, inlet soil pressure, outlet soil pressure, upper unloading door opening and lower unloading door opening of the screw conveyor shaft.

[0011] A further technical solution is that the working parameters of other supporting components include advancement speed, penetration rate and mortar injection port pressure.

[0012] Its further technical solution is that the method further comprises:

[0013] Acquire historical operation data of a shield machine without any faults in a stable operation state of the screw conveyor, wherein the historical operation data includes operation data of target working parameters of the shield machine without any faults in a stable operation state of the screw conveyor and an actual torque value of the screw conveyor shaft without any faults;

[0014] The operating data of the target working parameters is used as the input of the GBDT model, and the actual torque value is used as the output of the GBDT model. The torque prediction model is obtained by training the GBDT model using historical operating data.

[0015] A further technical solution is to determine that the screw conveyor is in a stable operating state when the propulsion speed of the shield machine is not 0 and fluctuates within an error range.

[0016] Its further technical solution is that the method further comprises:

[0017] According to a predetermined feature screening method, several working parameters that have the greatest impact on the torque of the screw conveyor shaft are screened out from all the working parameters of the shield machine as target working parameters.

[0018] A further technical solution is that the feature screening method for screening out the target working parameters includes:

[0019] The torque regression model is constructed based on random forest by using the operating data of all working parameters of the screw conveyor shaft of the shield machine without any faults in the stable operating state and the actual torque value of the screw conveyor shaft without any faults.

[0020] The importance of each working parameter in the entire torque regression model is calculated, and several working parameters with the highest importance are selected as target working parameters.

[0021] A further technical solution is to identify the fault of the screw conveyor according to the deviation between the actual torque value and the predicted torque value of the screw conveyor shaft, including:

[0022] Calculate the actual torque value T of the screw conveyor shaft real and torque prediction value T pre The relative residual between And determine the health index of the screw conveyor H = 100-r*100;

[0023] When the health index of the screw conveyor is lower than a predetermined threshold, it is determined that the screw conveyor is faulty.

[0024] Its further technical solution is that the method further comprises:

[0025] The health index of the screw conveyor is displayed in real time in the monitoring center, and an alarm signal is issued when a fault is determined in the screw conveyor.

[0026] The beneficial technical effects of the present invention are:

[0027] The present application discloses a method for monitoring the screw conveyor shaft fault of a shield machine based on GBDT. The method utilizes a torque prediction model combined with the real-time operating data of the target working parameters of the shield machine, and can monitor the health of the operating status of the screw conveyor shaft in real time during the construction of the shield machine. It has the advantages of high stability and high real-time performance, and can generate an alarm for the failure of the screw conveyor shaft, and even generate an alarm when an early failure occurs. The monitoring function of the system of the present invention can assist on-site construction personnel in judging the current status, and to a certain extent avoid serious equipment failures that cause major losses, thereby achieving the role of early maintenance. In addition, the method does not require the addition of new sensors, does not change the existing hardware, and is low-cost and easy to deploy and promote. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The present invention is a flow chart of a method for monitoring a screw conveyor shaft fault of a shield machine in an embodiment.

[0029] Figure 2 It is a method flow chart of a method for monitoring a screw conveyor shaft fault of a shield machine in another embodiment.

[0030] Figure 3 The following is a timing diagram of the actual torque value of the screw conveyor shaft in an example.

[0031] Figure 4 is Figure 3 A time series diagram of the health index of the screw conveyor calculated using the method of the present application in the same example. DETAILED DESCRIPTION

[0032] The specific implementation of the present invention will be further described below in conjunction with the accompanying drawings.

[0033] This application discloses a GBDT-based shield machine screw conveyor shaft fault monitoring method, which includes the following steps, please refer to Figure 1 and Figure 2 The flowchart shown:

[0034] Step 1: Obtain real-time operating data of the target working parameters of the shield machine.

[0035] The shield machine includes several supporting components, the screw conveyor shaft is one of them, and other common supporting components include the cutter head, articulated device, propulsion cylinder, etc. Due to the complex structure, the shield machine contains a large amount of working parameter data. In order to avoid excessive calculation, and to avoid subsequent overfitting and model planning problems, this application does not directly use all the working parameters of the shield machine for subsequent fault monitoring, but only uses the real-time operation data of some of the target working parameters. The target working parameter is the working parameter that reflects the operating status of the screw conveyor shaft among all the working parameters of the shield machine.

[0036] The target operating parameters are predetermined. In one embodiment, several operating parameters that have the greatest impact on the torque of the screw conveyor shaft are selected from all the operating parameters of the shield machine according to a predetermined feature screening method as target operating parameters. The torque of the screw conveyor shaft can well reflect the operating state of the screw conveyor shaft. Therefore, these selected target operating parameters are the operating parameters that have the greatest impact on the operating state of the screw conveyor shaft.

[0037] There are many methods for pre-defined feature screening, such as the common principal component analysis. In one embodiment of the present application, the torque regression model is constructed based on random forests using the operating data of all working parameters of the screw conveyor shaft of a shield machine that has not failed and the actual torque value of the screw conveyor shaft that has not failed. Then, the importance of each working parameter in the entire torque regression model is calculated, and several working parameters with the highest importance are selected as target working parameters, so as to achieve dimensionality reduction of all original working parameters and screen out the target working parameters.

[0038] A shield machine that has not failed means that not only the screw conveyor shaft has not failed, but also the regional supporting components of the shield machine have not failed, and the entire shield machine is in normal working condition. The screw conveyor mainly has a shutdown state, a startup state and a stable operating state during operation. According to the propulsion speed of the shield machine, the current state of the screw conveyor can be determined, and the propulsion speed of the shield machine is filtered and denoised: when the propulsion speed is 0, the screw conveyor is determined to be in a shutdown state. When the propulsion speed increases monotonically, the screw conveyor is determined to be in a startup state. When the propulsion speed of the shield machine is not 0 and fluctuates within the error range, that is, when the propulsion speed no longer rises and is basically stable, it is determined that the screw conveyor is in a stable operating state. This application uses the operating data of the shield machine when the screw conveyor is in a stable operating state to construct a torque regression model. When calculating the importance of each working parameter in the entire torque regression model, the importance of each parameter at each node in the entire random forest is added.

[0039] In addition to using the above method to screen and determine the target working parameters from all working parameters, it can also be determined in combination with the manual experience related to the railway construction. Finally, the target working parameters determined by this application include the working parameters of the screw conveyor shaft and the working parameters of other supporting components in the shield machine except the screw conveyor shaft. Specifically, the working parameters of the screw conveyor shaft include the speed, pressure, oil replenishment pressure, oil leakage temperature, inlet soil pressure, outlet soil pressure, upper unloading door opening and lower unloading door opening of the screw conveyor shaft. The working parameters of other supporting components include propulsion speed, penetration rate and mortar injection port pressure.

[0040] Step 2: input the acquired real-time operation data of the target working parameters into the torque prediction model, and use the output of the torque prediction model as the torque prediction value of the screw conveyor shaft.

[0041] The torque prediction model is trained based on the GBDT model using the historical operating data of the shield machine without faults in the stable operating state of the screw conveyor shaft. The torque prediction model is used to reflect the relationship between the target operating parameters of the shield machine without faults in the stable operating state of the screw conveyor shaft and the actual torque value of the screw conveyor shaft.

[0042] The method also includes the step of training the torque prediction model before using the torque prediction model: obtaining the historical operation data of the shield machine without faults in the stable operation state of the screw conveyor, the historical operation data including the operation data of the target working parameters of the shield machine without faults in the stable operation state of the screw conveyor and the actual torque value of the screw conveyor shaft without faults. This step is similar to the above-mentioned method of screening the target working parameters, but after screening the target working parameters, the operation data of the target working parameters can be directly obtained during the modeling in this step without obtaining the operation data of all working parameters. Then, the operation data of the target working parameters in the historical operation data is used as the input of the GBDT model, and the corresponding actual torque value is used as the output of the GBDT model. The torque prediction model is obtained by training based on the GBDT model using the historical operation data.

[0043] By tuning the model parameters, the mean absolute percentage error (MAPE) of the torque prediction model on the test set can reach 11.6%, and the mean square error (MSE) is about 0.607, and the model prediction ability is relatively reliable.

[0044] Step 3, identifying the fault of the screw conveyor according to the deviation between the real-time actual torque value and the torque prediction value of the screw conveyor shaft.

[0045] Since the torque prediction model reflects the relationship between the target working parameters of the shield machine under the stable operation state of the screw conveyor shaft and the actual torque value of the screw conveyor shaft, after the real-time operation data of the target working parameters are input into the torque prediction model, the torque prediction value obtained is the ideal torque value assuming that the screw conveyor shaft has not failed. Therefore, the greater the deviation between the real-time torque actual value and the torque prediction value of the screw conveyor shaft, the higher the probability of the screw conveyor shaft failure, so the deviation between the two can be used to identify the failure of the screw conveyor.

[0046] Specifically, calculate the actual torque value T of the screw conveyor shaft real and torque prediction value T pre The relative residual between And determine the health index H of the screw conveyor = 100-r*100. When the health index H of the screw conveyor is lower than the predetermined threshold, it is determined that the screw conveyor has a fault, thereby the operating state of the screw conveyor shaft can be monitored in real time, and the screw conveyor shaft fault can be detected in time.

[0047] During the operation of the shield machine, the target working parameters of the shield machine can be displayed in real time in the monitoring center. In addition, the health index of the screw conveyor can be displayed in real time in the monitoring center, and an alarm signal will be issued when a fault is determined in the screw conveyor, providing corresponding fault warnings and operation and maintenance references for equipment operation and maintenance and construction personnel.

[0048] In one example, the timing diagram of the actual torque value of the screw conveying shaft is as follows: Figure 3 As shown, the horizontal axis is time, and the vertical axis is the actual torque value, and the unit can be N·m. Using the method provided by the present application, the time series diagram of the health index of the screw conveyor is as follows: Figure 4 As shown, assuming that the predetermined threshold value of the health index is 60, it can be determined by calculation that according to the method of the present application, Figure 4 The screw conveyor shaft was detected to have a fault on November 3, 2018 as marked by the dotted line. In actual engineering applications, when the method of the present application was not used, the screw conveyor shaft broke on November 4, 2018, which is consistent with the theoretical calculation results of the present application, which can fully illustrate the effectiveness of the method of the present application.

[0049] The above is only a preferred embodiment of the present application, and the present invention is not limited to the above embodiments. It is understood that other improvements and changes directly derived or associated by those skilled in the art without departing from the spirit and concept of the present invention should be considered to be included in the protection scope of the present invention.

Claims

1. A GBDT-based method for monitoring the screw conveyor shaft fault of a shield machine. It is characterized in that The method comprises: Acquire real-time operation data of target working parameters of the shield machine, wherein the target working parameters are working parameters that reflect the operation status of the screw conveyor shaft among all working parameters of the shield machine; the shield machine includes several supporting components, the screw conveyor shaft is one of the supporting components, and the target working parameters include working parameters of the screw conveyor shaft and working parameters of other supporting components of the shield machine except the screw conveyor shaft; the working parameters of the screw conveyor shaft include speed, pressure, oil replenishment pressure, oil leakage temperature, inlet soil pressure, outlet soil pressure, upper discharge door opening and lower discharge door opening of the screw conveyor shaft; the working parameters of other supporting components include advancement speed, penetration and mortar injection port pressure; The real-time operation data of the acquired target working parameters are input into the torque prediction model, and the output of the torque prediction model is used as the torque prediction value of the screw conveyor shaft; the torque prediction model is obtained by training based on the GBDT model using the historical operation data of the shield machine without faults in the stable operation state of the screw conveyor shaft, and the torque prediction model is used to reflect the relationship between the target working parameters of the shield machine without faults in the stable operation state of the screw conveyor shaft and the actual value of the torque of the screw conveyor shaft; Identify screw conveyor faults based on the deviation between the real-time torque actual value and the torque prediction value of the screw conveyor shaft; The method further comprises: selecting several working parameters with the greatest influence on the torque of the screw conveyor shaft from all working parameters of the shield machine according to a predetermined feature screening method as the target working parameters; the feature screening method for screening the target working parameters comprises: using the operating data of all working parameters of the shield machine under the stable operating state of the screw conveyor shaft without any fault and the actual value of the torque of the screw conveyor shaft without any fault, constructing a torque regression model based on random forest; calculating the importance of each working parameter in the entire torque regression model, and selecting several working parameters with the highest importance as the target working parameters; The method also includes: obtaining historical operating data of a shield machine without any faults in a stable operating state of the screw conveyor, the historical operating data including operating data of target operating parameters of the shield machine without any faults in the stable operating state of the screw conveyor and an actual value of the torque of the screw conveyor shaft without any faults, and determining that the screw conveyor is in a stable operating state when the propulsion speed of the shield machine is not 0 and fluctuates within an error range; using the operating data of the target operating parameters as the input of the GBDT model, using the actual value of the torque as the output of the GBDT model, and obtaining the torque prediction model based on the GBDT model training using the historical operating data.

2. The method according to claim 1, It is characterized in that The method of identifying the fault of the screw conveyor according to the deviation between the actual torque value and the predicted torque value of the screw conveyor shaft comprises: Calculate the actual torque value T of the screw conveyor shaft real and torque prediction value T pre The relative residual between And determine the health index of the screw conveyor H = 100-r*100; When the health index of the screw conveyor is lower than a predetermined threshold, it is determined that the screw conveyor is faulty.

3. The method according to claim 2, It is characterized in that The method further comprises: The health index of the screw conveyor is displayed in real time at the monitoring center, and an alarm signal is issued when it is determined that the screw conveyor has a fault.