A method for evaluating and predicting cutterhead wear of shield machine

Abstract

The present invention belongs to the technical field of mechanical equipment maintenance, and discloses a shield machine cutter head wear assessment prediction method, comprising: obtaining a data sample set Data under the shield machine excavation state; preprocessing the data sample set Data to obtain a data sample set Data1; calculating the equivalent wear amount W of the cutter head when the shield machine changes the cutter each time; m ; Calculate the coefficient vector q based on Lasso regression L ; Get the data sample set Data2 of the shield machine in the state of waiting for excavation; Calculate the cumulative equivalent wear amount W of the shield machine cutter head after the excavation mileage L meters L ; Obtain the evaluation prediction results of the cutter head wear of the shield machine. The present invention uses Lasso regression to establish a prediction model, which has a regularized sparse solution screening function, and removes the features in the input that have a weak contribution to the prediction results, thereby eliminating the multicollinearity problem that is prevalent in the linear regression model. The feature subset used in the final model is more accurate, the model prediction accuracy is higher, and the generalization ability is stronger, which is convenient for evaluating and predicting the overall wear of the cutter head.

Description

Technical Field

[0001] The invention belongs to the technical field of mechanical equipment maintenance, and relates to a shield machine cutter head wear assessment and prediction method, and specifically to a shield machine cutter head wear assessment and prediction method based on Lasso regression, which can be used to assess and predict the cutter head wear of shield machine tools. Background Art

[0002] At present, with the continuous acceleration of my country's urbanization process, the development and utilization of urban underground space has developed rapidly. The shield method has become the main construction method for tunnel engineering due to its safe construction technology, strong adaptability to soil layers, and little impact on the environment along the line. The shield machine drives the cutter to break the rock through the cutter head, and brings the crushed stone and sand into the soil bin through the screw machine, so as to continuously excavate underground. The cutter head system is the core component of the shield machine to achieve rock breaking and excavation. Due to the complex and harsh construction environment and long-term operation under low speed, heavy load and dynamic variable working conditions, excessive wear and abnormal damage of the cutter have become the main reasons for the failure and shutdown of the shield machine, and it is also one of the main difficulties in safe and efficient excavation of the shield. During shield construction, once the cutter is excessively worn, if it cannot be discovered and replaced in time, it will greatly accelerate the wear rate of other cutters on the cutter head. On the other hand, the tool replacement process is complicated, difficult and costly. Opening the bin to change the cutter is likely to cause engineering accidents such as instability of the excavation face and tunnel collapse. The problem of tool wear and its impact is more prominent in complex formations. If not handled properly, it will threaten life and property safety. In order to avoid these problems, the tool must be opened frequently to check the tool. However, due to the harsh working environment of the shield machine and the complex mechanical structure of the cutterhead system, it is difficult to directly obtain the tool wear situation. The degradation state of the cutterhead system cannot be perceived in real time during the actual construction process. As a result, the timing of tool change can only be determined by manual experience, or a regular opening and changing strategy is adopted, resulting in a large number of omissions or misjudgments, causing unnecessary risks and waste of resources. Especially under complex geological conditions, such as crossing rivers and undersea tunnels, where frequent opening is not possible, the problem of tool wear failure is more prominent. Once a major accident occurs, it often results in the serious consequences of the entire shield machine being scrapped and the project being completely stagnant.

[0003] At present, the methods for evaluating the wear of shield machine cutters are mainly divided into wear mechanism research and sensor monitoring. Since the shield machine is a complex equipment with thermal and electrical coupling, and the cutter is the core component of the shield machine, its wear degradation correlation characteristics are affected by many factors. In addition, due to the complex operating conditions of the shield machine, the number of parameters in the operation of the shield machine is very large. Theoretical analysis cannot consider all parameters, and the establishment of the analysis model often relies on expert experience and stress analysis under ideal conditions. Expert experience is summarized by business experts in long-term shield practice, and there is no rigorous mathematical proof. In addition, due to too many assumptions in the stress analysis scenario, the mechanism model is very one-sided, so it is often only applicable to qualitative analysis of specific working conditions. For example, in the article "Shield Cutter Wear Mechanism and Prediction Analysis" published by Wu Jun et al. in Volume 30, Issue 8 of China Journal of Highway and Transportation in August 2017, the wear mechanism of shield cutters was analyzed based on the theory of metal tribology, considering the combined effects of abrasive wear, adhesive wear and fatigue wear, combining the calculation formulas of Rabinowicz abrasive wear, Archard adhesive wear and fatigue wear, introducing the Colorado University of Mines rock-breaking mechanical model of the cutter and the Central South University rock-breaking mechanical model of the cutter, and derived a general calculation model for predicting the wear of shield cutters and cutters. The shortcoming of this method is that it does not consider the specificity of different types of shield machines under different working conditions. Although the general calculation model is widely applicable, it does not have high accuracy. Sensor-based wear assessment directly measures and obtains tool wear information by installing special sensors. According to the detection principle, the current detection methods can be divided into three types: electrical detection, hydraulic detection, and gas detection. In November 2017, Liu Zhaowei et al. published an article titled "Study on Wireless Detection System of Shield Cutter Wear Based on Ultrasonic Waves" in Tunnel Construction (Chinese and English) Vol. 37, No. 11. Based on ultrasonic detection technology, they used 433MHz wireless data transmission technology to study the wear detection technology of shield cutters and completed a real-time wear detection device. The disadvantage of this method is that the sensor is not tested under harsh working conditions, and the accuracy of the sensor cannot be guaranteed after leaving the laboratory environment.

[0004] Through the above analysis, the problems and defects of the prior art are as follows:

[0005] (1) The existing solutions that rely on manual experience to determine the timing of tool change or adopt a strategy of regularly opening positions for tool change will cause a large number of omissions or misjudgments, leading to unnecessary risks and waste of resources.

[0006] (2) The existing shield machine tool wear assessment method based on wear mechanism research cannot take all parameters into account. Expert experience does not have rigorous mathematical proof and is only applicable to qualitative analysis under specific working conditions.

[0007] (3) The existing methods do not take into account the specificity of different types of shield machines under different working conditions. Although the general calculation model is widely applicable, it does not have high accuracy.

[0008] (4) In the existing sensor-based wear assessment methods, the sensors are not tested under harsh working conditions, and the accuracy of the sensors cannot be guaranteed after leaving the laboratory environment.

[0009] The difficulty in solving the above problems and defects is: relying on manual experience to determine the timing of cutter change mainly relies on the work experience accumulated by construction personnel over the years, and it is impossible to form a clear method to provide to new personnel; different shield machine cutter head materials, wear degrees in different working environments and complex working environments make it impossible to establish a simple mechanical mathematical relationship between the factors affecting the shield machine cutter head wear and the shield machine cutter head wear amount.

[0010] The significance of solving the above problems and defects is that the proposed shield machine cutter wear assessment and prediction method can maintain high accuracy under different shield machine models and different working environments of the shield machine, and can accurately help construction personnel determine the replacement time of the shield machine cutter head, which can not only ensure that the shield machine cutter head can be utilized to the maximum extent, but also avoid unsafe accidents caused by late cutter replacement, which is of great significance when the shield machine is working. Summary of the invention

[0011] In view of the problems existing in the prior art, the present invention provides a method for evaluating and predicting the cutter disc wear of a shield machine, and in particular relates to a method for evaluating and predicting the cutter disc wear of a shield machine based on Lasso regression, which aims to solve the problems of the prior art such as narrow application scope, poor accuracy and low reliability.

[0012] The present invention is implemented as follows: a shield machine cutter head wear assessment prediction method, the shield machine cutter head wear assessment prediction method comprising:

[0013] From the shield construction tool change record data, select the time point t when the number of tool changes was less than P 0 The time t at which the number of tool changes is greater than Q 1 The resulting tool degradation interval [t 0 ,t 1 ], each tool change record data includes 16 kinds of characteristic data, including the rock and soil bearing capacity characteristic data of the shield machine to be constructed area, the shield machine excavation control characteristic data collected by the sensor during the shield machine construction process, the shield machine PLC state characteristic data and the shield machine excavation mileage data; the tool change record data with less than or equal to 0 in the total propulsion force data of the shield machine excavation control data and less than or equal to 0 in the cutter head speed data in the H tool change records are eliminated to form the data sample set Data under the shield machine excavation state;

[0014] The data are comprehensively preprocessed to obtain a data sample set Data1; the equivalent wear of the cutterhead of the shield machine is calculated each time the cutter is changed, and it is used as a label; the multivariate linear regression model-Lasso regression model is applied, and the Lasso regression model is used to estimate the parameters, and according to the values ​​of the estimated parameters, the insignificant influencing factors are eliminated, and finally a calculation formula for the equivalent wear of the cutterhead is established; the formula is used to evaluate and predict the wear degree of the cutterhead after calculating the equivalent wear of the cutterhead after a certain distance.

[0015] Furthermore, the shield machine cutter head wear assessment prediction method comprises the following steps:

[0016] Step 1: Obtain the data sample set Data under the tunneling state of the shield machine;

[0017] Provide original data sample set for model establishment

[0018] Step 2: preprocess the data sample set Data;

[0019] Remove outliers from the data sample set, normalize the outlier processing results, and then perform noise reduction on the normalized results to improve the quality of the data sample set and the accuracy of the model.

[0020] Step 3: Calculate the equivalent wear amount W of the cutter head every time the shield machine changes the cutter head. m ;

[0021] Providing labels for model training

[0022] Step 4: Calculate the coefficient vector q based on Lasso regression L ;

[0023] Calculate the coefficient vector and remove the features that have little effect on the wear of the shield machine cutter head according to the value of the coefficient vector

[0024] Step 5, obtaining a data sample set Data2 of the shield machine in the state of waiting for excavation;

[0025] Get the test set to test the accuracy of the model; use the same data set to estimate the wear of the shield machine cutterhead in the future

[0026] Step 6: Calculate the cumulative equivalent wear W of the shield machine cutterhead after tunneling mileage L meters L ;

[0027] The cumulative equivalent wear amount W of the shield machine cutterhead after the excavation mileage L meters is obtained L

[0028] Step seven, obtain the evaluation and prediction results of the shield machine cutter head wear.

[0029] Based on the evaluation and prediction results of the shield machine cutterhead wear, construction personnel can accurately determine the timing of cutter change, which can not only ensure that the shield machine cutterhead can be utilized to the maximum extent, but also avoid unsafe accidents caused by changing the cutter too late.

[0030] Furthermore, the data sample set Data obtained in the shield machine excavation state in step 1 includes:

[0031] (1) Select the time point t when the number of tool changes was less than P from the shield construction tool change record data 0 The time t at which the number of tool changes is greater than Q 1 The resulting tool degradation interval [t 0 ,t 1 ], each tool change record data includes 16 kinds of characteristic data, including the rock and soil bearing capacity characteristic data of the shield machine to be constructed area, the shield machine excavation control characteristic data collected by sensors during the shield machine construction process, the shield machine PLC status characteristic data and the shield machine excavation mileage data; among them, the shield machine excavation control data includes the total thrust of the shield machine and the cutter head speed; the shield machine PLC status characteristic data includes the thrust speed, cutter head torque, penetration, screw machine speed, cutter head speed, lower left soil bin pressure, left middle soil bin pressure, upper left soil bin pressure, group A thrust pressure, group B thrust pressure, group C thrust pressure, group D thrust pressure, P≤4, Q≥30, H≥20.

[0032] (2) Eliminate the tool change record data whose total thrust data of the shield machine excavation control data is less than or equal to 0 and whose cutter head speed data is less than or equal to 0 in the H tool change records. The remaining M tool change record data constitute the data sample set Data under the shield machine excavation state.

[0033] The preprocessing of the data sample set Data in step 2 includes:

[0034] Perform outlier processing on each feature data in the data sample set Data, normalize the outlier processing result, and then perform noise reduction processing on the normalized result to obtain the preprocessed data sample set Data1;

[0035] The preprocessing of the data sample set Data specifically includes:

[0036] (1) Replace the value of 0 in each feature data contained in the data sample set Data with 0.00000001 to obtain the data sample set Data′ that processes the abnormal values ​​of the data sample set Data;

[0037] (2) Each feature data contained in the data sample set Data′ is normalized to its maximum value by the following formula to obtain the data sample set Data″ after maximum value normalization;

[0038]

[0039] Among them, z * represents the normalized result of the maximum value of each feature data z in the data sample set Data′, z max and z min Represent the maximum and minimum values ​​of z respectively;

[0040] (3) Sliding window sampling is performed on each feature data contained in the data sample set Data″, and the root mean square value of the feature data contained in the data sample set Data″ in each window is calculated, and the calculation result is used as the replacement value of the feature data contained in the data sample set Data″ in the window to achieve noise reduction for each feature data of Data″; wherein the data noise reduction includes:

[0041] 1) Set the sliding window length to m 1 , the step size of the window movement is s 1 ; Among them, m 1 Take any positive integer in the interval [30, 60], s 1 The value of m 1 equal;

[0042] 2) Calculate the RMS value of all data in the window after each slide according to the following formula:

[0043]

[0044] Among them, d i represents the root mean square value of all data in the i-th sliding window, y a represents the ath data in the window after the i-th sliding, and ∑ represents the summation operation.

[0045] The data sample set Data is preprocessed to obtain the data sample set Data1.

[0046] Further, the equivalent wear amount W of the cutter head of the shield machine is calculated each time the cutter is changed in step 3. m include:

[0047] (1) Obtain the wear amount of each tool at each tool change:

[0048] 1) Assume that the center area, front area and edge area of ​​the shield machine cutter head contain N cutters in total, N ≥ 46. According to the records of the cutter maintenance log, determine whether each cutter has been replaced each time the cutter is changed. If so, the wear amount of the replaced cutter is 0, otherwise, execute step 2);

[0049] 2) Determine whether the wear amount of the unreplaced tool is normal according to the tool wear record in the tool maintenance log. If so, execute step 3); otherwise, use the replacement threshold of the unreplaced tool with abnormal wear amount as the wear amount of the tool each time the tool is changed; wherein, the replacement threshold of the edge area tool is 5mm, and the replacement threshold of the front area tool and the center area tool is 15mm;

[0050] 3) Determine whether the unreplaced tool with normal wear has a wear measurement value. If so, use the wear measurement value as the wear amount of the tool each time the tool is changed; otherwise, use the interpolation result of the two wear values ​​of the tool in the tool maintenance log as the wear amount of the tool each time the tool is changed.

[0051] (2) The weighted sum of the wear of the N tools on the cutter head during each tool change is calculated to obtain the equivalent wear of the cutter head during the mth tool change: m :

[0052]

[0053] Among them, α represents the wear coefficient of the tool. The wear coefficient of the tool in the center area and the front area is 1, and the wear coefficient of the tool in the edge area is 3. mn Indicates the wear amount of the nth tool during the mth tool change.

[0054] Further, the coefficient vector q is calculated based on Lasso regression in step 4. L include:

[0055] Calculate the equivalent wear amount W of the cutter head during the mth tool change m The logarithm of y m , and use the Lasso regression method, through the data sample set Data1 and W m The logarithm of y m Calculate the coefficient vector q of the logarithmic calculation formula of the shield machine cutterhead cumulative equivalent wear L :

[0056]

[0057] y m =lnW m ;

[0058] Among them, q represents the regression coefficient vector, R 16 The regression coefficient vector q is a column vector containing 16 elements. 16 represents the 16 characteristic data that affect the equivalent wear of the cutter head contained in the data sample set Data. ∑ represents the summation operation. λ represents the adjustment coefficient. m The logarithmic vector of factors influencing the cumulative equivalent wear of the cutter disc during the mth tool change.

[0059] The data sample set Data2 obtained in the step 5 when the shield machine is in the state of waiting for excavation includes:

[0060] The Lasso regression regularized sparse solution screening method is used to screen out the 16 characteristic data affecting the equivalent wear of the cutterhead contained in the data sample set Data1, and the final characteristic data affecting the equivalent wear of the cutterhead are screened out: cutterhead torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, screw machine speed, and excavation mileage; assuming that the future excavation mileage is L meters, the characteristic data values ​​of cutterhead torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, and screw machine speed are the same as the characteristic data values ​​of the last cutter change record in the data sample set Data1, forming the data sample set Data2 of the shield machine in the state of waiting for excavation.

[0061] Further, the calculation of the equivalent wear amount W of the shield machine cutter head after the tunneling mileage L meters in step 6 is L include:

[0062] (1) Perform logarithmic calculation on each characteristic data affecting the equivalent wear of the cutter head contained in the data sample set Data2 of the shield machine in the state of waiting for excavation, and obtain the data sample set Data2′. Then, the logarithmic value y of the cumulative equivalent wear of the cutter head of the shield machine after the excavation mileage L meters is calculated by the data sample set Data2′:

[0063] y=q L (Data2′)+ε;

[0064] Here, ε represents the random error.

[0065] (2) After exponential calculation of the logarithm y of the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters, the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters is obtained: L ;

[0066] W L =e y .

[0067] The evaluation and prediction results of obtaining the cutter head wear of the shield machine in step seven include:

[0068] According to the wear level range determined in the actual construction process, check the cumulative equivalent wear W of the shield machine cutter head after the excavation mileage L meters L The wear level range is used to evaluate the wear degree of the shield machine cutter head after the shield machine has excavated a distance of L meters.

[0069] Combining all the above-mentioned technical solutions, the advantages and positive effects of the present invention are as follows: the shield machine cutter disc wear assessment and prediction method provided by the present invention adopts an industrial big data driven method, which can be used to assess and predict the cutter disc wear of shield machine tools, solving the technical problem of a narrow scope of application in the prior art, while improving the accuracy of tool health status monitoring and life prediction.

[0070] The present invention fully considers the geological characteristics and characteristic data of the shield machine's operating status, and realizes the evaluation and prediction of the overall wear status of the cutter disc by weighted summing up the tool wear in different areas of the cutter disc, rather than targeting a single tool. It is more in line with actual industrial scenarios, has a wide range of applications and high accuracy.

[0071] The present invention is based on the tool wear measurement records. By weighted summing up the wear of single tools in different areas of the cutter disc, an equivalent overall wear value of the cutter disc is constructed. Compared with traditional research that takes the wear of a single tool as the prediction object, the mapping relationship between the observed data and the overall wear of the cutter disc can be established more accurately, which facilitates the evaluation and prediction of the overall wear of the cutter disc.

[0072] The present invention adopts Lasso regression to establish a prediction model, which has a regularized sparse solution screening function and can eliminate the features in the input that have weak contribution to the prediction results, thereby eliminating the multicollinearity problem that is common in linear regression models. The feature subset used in the final model is more accurate, the model prediction accuracy is higher, and the generalization ability is stronger. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0074] Figure 1 It is a flow chart of a shield machine cutter head wear assessment and prediction method provided by an embodiment of the present invention.

[0075] Figure 2 It is a schematic diagram of a shield machine cutter head wear assessment and prediction method provided in an embodiment of the present invention.

[0076] Figure 3 It is a schematic diagram of the prediction result of the wear increment of the 448-ring cutter disc in the future 5 rings in the example verification provided by the embodiment of the present invention.

[0077] Figure 4 It is a schematic diagram of the shield machine cutter head area division provided by an example of the present invention. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0079] In view of the problems existing in the prior art, the present invention provides a method for evaluating and predicting cutter head wear of a shield machine. The present invention is described in detail below in conjunction with the accompanying drawings.

[0080] like Figure 1 As shown, the shield machine cutter head wear assessment prediction method provided by the embodiment of the present invention comprises the following steps:

[0081] S101, obtaining a data sample set Data under the tunneling state of the shield machine;

[0082] S102, preprocessing the data sample set Data;

[0083] S103, calculate the equivalent wear amount W of the cutter head when the shield machine changes the cutter head each time m ;

[0084] S104, calculate coefficient vector q based on Lasso regression L ;

[0085] S105, obtaining a data sample set Data2 of the shield machine in the state of waiting for excavation;

[0086] S106, calculate the cumulative equivalent wear amount W of the shield machine cutterhead after the tunneling mileage L meters L ;

[0087] S107, obtaining the evaluation prediction result of the cutter head wear of the shield machine.

[0088] The principle diagram of the shield machine cutter head wear assessment prediction method provided by the embodiment of the present invention is as follows: Figure 2 shown.

[0089] Figure 4 It is a schematic diagram of the shield machine cutter head area division provided by an embodiment of the present invention.

[0090] The technical solution of the present invention is further described below in conjunction with specific embodiments.

[0091] The purpose of the present invention is to overcome the problems of poor accuracy and low reliability in the above-mentioned prior art. An industrial big data driven method is adopted to provide a shield machine cutter head wear assessment and prediction method based on Lasso regression. The method fully considers the geological characteristics and the characteristic data of the shield machine operation status. The overall wear status of the cutter head is evaluated and predicted by weighted summing the tool wear amounts in different areas of the cutter head, rather than targeting a single tool. This is more in line with actual industrial scenarios and at the same time improves the accuracy of cutter head health status monitoring and life prediction.

[0092] The idea of ​​realizing the purpose of the present invention is to select the time point t when the number of tool changes last time is less than P from the shield construction tool change record data 0 The time t at which the number of tool changes is greater than Q 1 The resulting tool degradation interval [t 0 ,t 1 ], each tool change record data includes 16 kinds of characteristic data, including the rock and soil bearing capacity characteristic data of the shield machine to be constructed area, the shield machine excavation control characteristic data collected by the sensor during the shield machine construction process, the shield machine PLC state characteristic data and the shield machine excavation mileage data. Then, the tool change record data less than or equal to 0 in the total thrust data of the shield machine excavation control data and the cutter head speed data less than or equal to 0 in the H tool change records are eliminated to form the data sample set Data under the shield machine excavation state. Then, the above data are comprehensively preprocessed to obtain the data sample set Data1. Then, the equivalent wear amount of the cutter head at each tool change of the shield machine is calculated and used as a label. Then, a multivariate linear regression model-Lasso regression model is applied. The Lasso regression model is used to estimate the parameters, and then according to the values ​​of the estimated parameters, the insignificant influencing factors are eliminated, and finally the calculation formula for the equivalent wear amount of the cutter head is established. Using this formula, after calculating the equivalent wear amount of the cutter head after a certain distance, the wear degree of the cutter head is evaluated and predicted.

[0093] In order to achieve the above object, the technical solution adopted by the present invention includes the following steps:

[0094] (1) Obtain the data sample set Data under the shield machine excavation state:

[0095] (1a) Select the time point t when the number of tool changes was less than P from the shield construction tool change record data 0 The time t at which the number of tool changes is greater than Q 1 The resulting tool degradation interval [t 0 ,t 1], each tool change record data includes 16 kinds of characteristic data, including the rock and soil bearing capacity characteristic data of the shield machine to be constructed area, the shield machine excavation control characteristic data collected by the sensor during the shield machine construction process, the shield machine PLC state characteristic data and the shield machine excavation mileage data; among them, the shield machine excavation control data includes the total thrust of the shield machine and the cutter head speed; the shield machine PLC state characteristic data includes the thrust speed, cutter head torque, penetration, screw machine speed, cutter head speed, lower left soil bin pressure, left middle soil bin pressure, upper left soil bin pressure, group A thrust pressure, group B thrust pressure, group C thrust pressure, group D thrust pressure, P≤4, Q≥30, H≥20;

[0096] (1b) Eliminate the tool change record data whose total thrust data of shield machine excavation control data is less than or equal to 0 and whose cutter head speed data is less than or equal to 0 in the H tool change records, and the remaining M tool change record data constitute the data sample set Data under the shield machine excavation state;

[0097] (2) Preprocess the data sample set Data:

[0098] Perform outlier processing on each feature data in the data sample set Data, normalize the outlier processing result, and then perform noise reduction processing on the normalized result to obtain the preprocessed data sample set Data1;

[0099] (3) Calculate the equivalent wear amount W of the cutterhead each time the shield machine changes the cutter m :

[0100] (3a) Obtain the wear of each tool at each tool change:

[0101] (3a1) Assume that the center area, front area and edge area of ​​the shield machine cutter head contain N cutters in total, N ≥ 46. According to the records in the cutter maintenance log, determine whether each cutter has been replaced each time the cutter is changed. If so, the wear amount of the replaced cutter is 0. Otherwise, execute step (3a2);

[0102] (3a2) judging whether the wear amount of the unreplaced tool is normal according to the wear record of the tool in the tool maintenance log, and if so, executing step (3a3); otherwise, taking the replacement threshold of the unreplaced tool with abnormal wear amount as the wear amount of the tool each time the tool is changed, wherein the replacement threshold of the edge area tool is 5 mm, and the replacement threshold of the front area tool and the center area tool is 15 mm;

[0103] (3a3) determining whether a tool that has not been replaced and has normal wear has a wear measurement value; if so, using the wear measurement value as the wear amount of the tool each time the tool is changed; otherwise, using the interpolation result of the two wear values ​​of the tool in the tool maintenance log as the wear amount of the tool each time the tool is changed;

[0104] (3b) The weighted sum of the wear of the N tools on the cutter head during each tool change is obtained to obtain the equivalent wear of the cutter head during the mth tool change: m :

[0105]

[0106] Among them, α represents the wear coefficient of the tool. The wear coefficient of the tool in the center area and the front area is 1, and the wear coefficient of the tool in the edge area is 3. mn Indicates the wear amount of the nth tool during the mth tool change;

[0107] (4) Calculate the coefficient vector q based on Lasso regression L :

[0108] Calculate the equivalent wear amount W of the cutter head during the mth tool change m The logarithm of y m , and use the Lasso regression method, through the data sample set Data1 and W m The logarithm of y m Calculate the coefficient vector q of the logarithmic calculation formula of the shield machine cutterhead cumulative equivalent wear L :

[0109]

[0110] y m =lnW m

[0111] Among them, q represents the regression coefficient vector, R 16 The regression coefficient vector q is a column vector containing 16 elements. 16 represents the 16 characteristic data that affect the equivalent wear of the cutter head contained in the data sample set Data. ∑ represents the summation operation. λ represents the adjustment coefficient. m The logarithmic vector of factors influencing the cumulative equivalent wear of the cutter head during the mth tool change;

[0112] (5) Obtain the data sample set Data2 of the shield machine in the state of waiting for excavation:

[0113] The Lasso regression regularized sparse solution screening method is used to screen out the 16 characteristic data affecting the equivalent wear of the cutterhead from the data sample set Data1, including cutterhead torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, screw machine speed, and excavation mileage. Assuming that the future excavation mileage is L meters, the characteristic data values ​​of cutterhead torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, and screw machine speed are the same as the characteristic data values ​​of the last cutter change record in the data sample set Data1, forming the data sample set Data2 of the shield machine in the state of waiting for excavation.

[0114] (6) Calculate the cumulative equivalent wear W of the shield machine cutterhead after tunneling mileage L meters L :

[0115] (6a) Logarithmic calculation is performed on each characteristic data affecting the equivalent wear of the cutter head contained in the data sample set Data2 of the shield machine in the state of waiting for excavation, and the data sample set Data2′ is obtained. The logarithmic value y of the cumulative equivalent wear of the cutter head of the shield machine after the excavation mileage L meters is calculated through the data sample set Data2′:

[0116] y=q L (Data2′)+ε

[0117] Among them, ε represents random error;

[0118] (6b) After exponential calculation of the logarithm y of the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters, the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters is obtained: L ;

[0119] W L =e y ;

[0120] (7) Obtain the evaluation and prediction results of shield machine cutterhead wear:

[0121] According to the wear level range determined in the actual construction process, check the cumulative equivalent wear W of the shield machine cutter head after the excavation mileage L meters L The wear level range is used to evaluate the wear degree of the shield machine cutter head after the shield machine has excavated a distance of L meters.

[0122] The shield machine cutterhead area of ​​the present invention is divided as follows Figure 4 shown.

[0123] The present invention is based on the tool wear measurement records and a data-driven method. By weighted summing the wear of a single tool in different areas of the cutter disc, the equivalent overall wear value of the cutter disc is constructed. Compared with the traditional research that takes the wear of a single tool as the prediction object, it can more accurately establish the mapping relationship between the observed data and the overall wear of the cutter disc, which is convenient for evaluating and predicting the overall wear of the cutter disc. The present invention uses Lasso regression to establish a prediction model, which has a regularized sparse solution screening function, and can eliminate the features in the input that have a weak contribution to the prediction result, thereby eliminating the multicollinearity problem that is common in linear regression models. The feature subset used in the final model is more accurate, the model prediction accuracy is higher, and the generalization ability is stronger.

[0124] The effects of the present invention will be further described below in conjunction with the embodiments of the present invention.

[0125] The data set used in the example of the present invention is the data of the left line shield section from Liuwudian Station to Dongjie Station of Xiamen Metro Line 3 collected by China Railway First Bureau Group Co., Ltd. from July 3, 2017 to April 10, 2018.

[0126] The shield machine cutter head wear assessment and prediction method provided by the embodiment of the present invention comprises the following steps:

[0127] (1) Obtain the data sample set Data under the shield machine excavation state:

[0128] (1a) From the construction records of the shield section from Liuwudian Station to Dongjie Station, two tool change time points were selected, with the number of tool changes greater than 30 and the number of the last tool change less than 4, namely the 230th and 130th rings. The 130th to 230th rings were the complete tool degradation intervals, with a total of 38 tool change data, including 16 types of feature data, including the rock and soil bearing capacity feature data of the shield machine construction area, the shield machine excavation control feature data collected by sensors during the shield machine construction process, the shield machine status feature data, and the shield machine excavation mileage data;

[0129] (1b) Extract the data in the tunneling state from the above data set, and remove the data in which the total thrust of the shield machine is less than or equal to 0 and the cutter head speed is less than or equal to 0 from the data set, as the data sample set Data of the shield machine in the tunneling state;

[0130] (2) Preprocess the data sample set Data to obtain the data sample set Data1:

[0131] Perform outlier processing on each feature data in the data sample set Data, normalize the outlier processing results, and then perform noise reduction processing on the normalized results to obtain the preprocessed data sample set Data1. The specific operations are:

[0132] (2a) replacing the value of 0 in each feature data contained in the data sample set Data with 0.00000001, and obtaining the data sample set Data′ for processing the abnormal values ​​of the data sample set Data;

[0133] (2b) Each feature data contained in the data sample set Data′ is normalized to its maximum value by the following formula to obtain the data sample set Data″ after maximum value normalization;

[0134]

[0135] Among them, z * represents the normalized result of the maximum value of each feature data z in the data sample set Data′, z max and z min Represent the maximum and minimum values ​​of z respectively;

[0136] (2c) Sliding window sampling is performed on each feature data contained in the data sample set Data″, and the root mean square value of the feature data contained in the data sample set Data″ in each window is calculated. The calculation result is used as the replacement value of the feature data contained in the data sample set Data″ in the window, so as to realize noise reduction of each feature data of Data″. The specific steps of data noise reduction are:

[0137] The first step is to set the sliding window length to m 1 , the step size of the window movement is s 1 , where m 1 Take any positive integer in the interval [30, 60], s 1 The value of m 1 equal.

[0138] The second step is to calculate the RMS value of all data in the window after each slide according to the following formula:

[0139]

[0140] Among them, d i represents the root mean square value of all data in the i-th sliding window, y a represents the ath data in the window after the i-th sliding, and ∑ represents the summation operation.

[0141] The data sample set Data is preprocessed to obtain the data sample set Data1.

[0142] (3) Calculate the equivalent wear amount W of the cutterhead each time the shield machine changes the cutter m :

[0143] (3a) Obtain the wear of each tool at each tool change:

[0144] ① For normally worn tools with measured values, use their measured values ​​directly;

[0145] ② For normal wear tools without measurement values, interpolate the two adjacent measurement records to approximately calculate their wear values;

[0146] ③ For tools with abnormal wear, such as eccentric wear and edge curling, the wear value will be calculated based on the tool

[0147] The position on the cutter disc is approximately changed, the cutter in the outer ring area is 5mm, and the cutter in the front and center area is 15mm.

[0148] ④ All replacement records in the cutter maintenance log will be regarded as replacement of new cutters without any wear.

[0149] (3b) For different areas of the shield machine cutterhead, the wear coefficients α of the center cutter, front cutter, and edge cutter are different. The wear coefficient of the cutter in the outer ring area is 3, and the wear coefficient of the cutter in the center area and the front area is 1.

[0150] The weighted sum of the wear of all tools is used to obtain the equivalent overall wear of the cutter head, which is used as the label of the data sample set Data1:

[0151]

[0152] (4) Calculate the coefficient vector q based on Lasso regression L :

[0153] Calculate the equivalent wear W of the cutter head m The logarithm of y m , use the data sample set to establish the Lasso regression model and estimate the coefficient q L The implementation steps are:

[0154] (4a) For a given shield machine, the total wear of the cutterhead is affected by the original total wear value of the cutterhead, the excavation length, geological characteristics and operating parameters. Therefore, the cumulative wear value of the cutterhead can be expressed by formula (1).

[0155] ΔW=W t -W 0 =f(L,g 1 ,…,g i ,s 1 ,…,s j ,e 1 ,…,e k ) (1)

[0156] Where W t is the total wear value of the cutter head at time t, W 0 is the original total wear value of the cutterhead, L is the incremental length of excavation, g i is the ith geological feature, s j is the jth state characteristic of the shield machine, e k is the kth tunneling operation parameter, and f is a nonlinear function.

[0157] (4b) The relationship between the cumulative wear value of the cutter head and various influencing factors can be approximated as an exponential function, that is,

[0158]

[0159] Among them, α, β i , γ j , p k are exponential powers respectively, and ε is the random error.

[0160] (4c) Apply the linear regression model, perform logarithmic operations on both sides of equation (2), and then

[0161] lnΔW=αlnL+β 1 lng 1 +…+β i lng i +γ 1 lns 1 +…+γ j lns j +p 1 lne 1 +…+p k lne k +ε

[0162] (4d) By variable transformation, we define y = lnΔW, x = (lnL, lng 1 ,...,lng i ,lns 1 ,...,lns j ,lne 1 ,...,lne k ) T , q=(α,β 1 ,...,β i ,γ 1 ,...,γ j ,p 1 ,...,p k ), formula (3) can be transformed into formula (4):

[0163] y i =qx i +ε (4)

[0164] (4e) Add L to equation (4) 1 The penalty term, for formula (4), Lasso estimation is:

[0165]

[0166] Among them, q represents the regression coefficient vector, R 16 The regression coefficient vector q is a column vector containing 16 elements. 16 represents the 16 characteristic data that affect the equivalent wear of the cutter head contained in the data sample set Data. ∑ represents the summation operation. λ represents the adjustment coefficient. m The logarithmic vector of factors influencing the cumulative equivalent wear of the cutter disc during the mth tool change.

[0167] After the coefficient vector is calculated in the embodiment of the present invention, it is assumed that the other excavation parameter characteristics of the future excavation distance remain unchanged, and the excavation mileage of each ring is 1.2 meters, and the characteristic data values ​​of the future 5 rings are obtained. Figure 3 The polyline shown. Figure 3 The horizontal axis represents the number of rings, and the vertical axis represents the wear amount of the cutter disc.

[0168] Table 1 Assumptions of the future 5 rings

[0169]

[0170]

[0171] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When the use is implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (e.g., infrared, wireless, microwave, etc.) mode) to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0172] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A shield machine cutter head wear assessment prediction method, characterized in that: The shield machine cutter wear assessment prediction method selects a data sample set Data under the shield machine excavation state in the tool degradation interval [t0, t1] formed by the time point t0 when the number of tool changes was less than P and the time point t1 when the number of tool changes was greater than Q from the shield construction tool change record data; The data sample set Data is preprocessed comprehensively to obtain the data sample set Data1; the equivalent wear of the cutterhead of the shield machine is calculated each time the cutter is changed, and it is used as a label; the multivariate linear regression model-Lasso regression model is used to estimate the parameters, and according to the parameter values, the insignificant influencing factors are eliminated, and finally the calculation formula for the equivalent wear of the cutterhead is established; the formula is used to evaluate and predict the wear degree of the cutterhead after calculating the equivalent wear of the cutterhead after a certain distance; The shield machine cutterhead wear assessment and prediction method includes the following steps: Step 1: Obtain the data sample set Data under the tunneling state of the shield machine; Step 2: preprocess the data sample set Data to obtain the data sample set Data1; Step 3: Calculate the equivalent wear amount W of the cutter head every time the shield machine changes the cutter head. m ; Step 4: Calculate the coefficient vector q based on Lasso regression L ; Step 5, obtaining a data sample set Data2 of the shield machine in the state of waiting for excavation; Step 6: Calculate the cumulative equivalent wear W of the shield machine cutterhead after tunneling mileage L meters L ; Step 7, obtaining the evaluation prediction results of the shield machine cutter head wear; The coefficient vector q calculated based on Lasso regression in step 4 L include: Calculate the equivalent wear amount W of the cutter head during the mth tool change m The logarithm of y m , and use the Lasso regression method, through the data sample set Data1 and W m The logarithm of y m , calculate the coefficient vector q of the numerical calculation formula of the cumulative equivalent wear of the shield machine cutter head L : ym=lnWm; Among them, q represents the regression coefficient vector, R 16 The regression coefficient vector q is a column vector containing 16 elements. 16 represents the 16 characteristic data that affect the equivalent wear of the cutter head contained in the data sample set Data. ∑ represents the summation operation. λ represents the adjustment coefficient. m It represents the logarithmic vector of factors influencing the cumulative equivalent wear of the cutter head during the mth tool change, and M represents the number of tool change record data; The step 5 of obtaining the data sample set Data2 of the shield machine in the state of waiting for excavation includes: using the Lasso regression regularized sparse solution screening method to screen out the final characteristic data affecting the equivalent wear of the cutter disc from the 16 characteristic data affecting the equivalent wear of the cutter disc contained in the data sample set Data1: cutter disc torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, screw machine speed, and excavation mileage; assuming that the future excavation mileage is L meters, the characteristic data values ​​of the cutter disc torque, rock and soil bearing capacity, total thrust, lower left soil bin pressure, and screw machine speed are the same as the characteristic data values ​​of the last cutter change record in the data sample set Data1, forming the data sample set Data2 of the shield machine in the state of waiting for excavation.

2. The shield machine cutter head wear assessment and prediction method according to claim 1, characterized in that: The data sample set Data obtained in the step 1 under the tunneling state of the shield machine includes: (1) From the tool change record data of shield construction, select H tool change record data within the tool degradation interval [t0, t1] formed by the time point t0 when the number of tool changes was less than P and the time point t1 when the number of tool changes was greater than Q. Each tool change record data includes 16 types of characteristic data, including the rock and soil bearing capacity characteristic data of the shield machine construction area, the shield machine excavation control characteristic data collected by sensors during the shield machine construction process, the shield machine PLC state characteristic data and the shield machine excavation mileage data; among them, the shield machine excavation control characteristic data includes the total thrust of the shield machine and the cutter head speed; the shield machine PLC state characteristic data includes the thrust speed, cutter head torque, penetration, screw machine speed, cutter head speed, lower left soil bin pressure, left middle soil bin pressure, upper left soil bin pressure, group A thrust pressure, group B thrust pressure, group C thrust pressure, group D thrust pressure, P≤4, Q≥30, H≥20; (2) Eliminate the tool change record data whose total thrust data of the shield machine excavation control data is less than or equal to 0 and whose cutter head speed data is less than or equal to 0 in the H tool change records, and the remaining M tool change record data constitute the data sample set Data under the shield machine excavation state; The preprocessing of the data sample set Data in step 2 includes: Perform outlier processing on each feature data in the data sample set Data, normalize the outlier processing result, and then perform noise reduction processing on the normalized result to obtain the preprocessed data sample set Data1; The preprocessing of the data sample set Data specifically includes: (1) Replace the value of 0 in each feature data contained in the data sample set Data with 0.00000001 to obtain the data sample set Data′ that processes the abnormal values ​​of the data sample set Data; (2) Each feature data contained in the data sample set Data′ is normalized to its maximum value by the following formula to obtain the data sample set Data″ after maximum value normalization; Among them, z * represents the normalized result of the maximum value of each feature data z in the data sample set Data′, z max and z min Represent the maximum and minimum values ​​of z respectively; (3) Sliding window sampling is performed on each feature data contained in the data sample set Data″, and the root mean square value of the feature data contained in the data sample set Data″ in each window is calculated, and the calculation result is used as the replacement value of the feature data contained in the data sample set Data″ in the window to achieve noise reduction for each feature data of Data″; wherein the data noise reduction includes: 1) Set the sliding window length to m1, and the step length of the window movement to s1; m1 is any positive integer in the interval [30,60], and the value of s1 is equal to m1; 2) Calculate the RMS value of all data in the window after each slide according to the following formula: Among them, d i represents the root mean square value of all data in the i-th sliding window, y a represents the ath data in the window after the i-th sliding, ∑ represents the sum operation; The data sample set Data is preprocessed to obtain the data sample set Data1.

3. The shield machine cutter head wear assessment and prediction method according to claim 1, characterized in that: The calculation of the equivalent wear amount W of the cutter head when the shield machine changes the cutter each time in step 3 m include: (1) Obtain the wear amount of each tool at each tool change: 1) Assume that the center area, front area and edge area of ​​the shield machine cutter head contain N cutters in total, N ≥ 46. According to the records of the cutter maintenance log, determine whether each cutter has been replaced each time the cutter is changed. If so, the wear amount of the replaced cutter is 0, otherwise, execute step 2); 2) Determine whether the wear amount of the unreplaced tool is normal according to the tool wear record in the tool maintenance log. If so, execute step 3); otherwise, use the replacement threshold of the unreplaced tool with abnormal wear amount as the wear amount of the tool each time the tool is changed; wherein, the replacement threshold of the edge area tool is 5mm, and the replacement threshold of the front area tool and the center area tool is 15mm; 3) Determine whether the unreplaced tool with normal wear has a wear measurement value. If so, use the wear measurement value as the wear amount of the tool each time the tool is changed; otherwise, use the interpolation result of the two wear values ​​of the tool in the tool maintenance log as the wear amount of the tool each time the tool is changed; (2) The weighted sum of the wear of the N tools on the cutter head during each tool change is calculated to obtain the equivalent wear of the cutter head during the mth tool change: m : Among them, α represents the wear coefficient of the tool. The wear coefficient of the tool in the center area and the front area is 1, and the wear coefficient of the tool in the edge area is 3. mn Indicates the wear amount of the nth tool during the mth tool change.

4. The shield machine cutter head wear assessment prediction method according to claim 1, characterized in that: The calculation of the cumulative equivalent wear amount W of the shield machine cutter head after the excavation mileage L meters in step 6 L include: (1) Perform logarithmic calculation on each characteristic data affecting the equivalent wear of the cutter head contained in the data sample set Data2 of the shield machine in the state of waiting for excavation, and obtain the data sample set Data2′. Then, the logarithmic value y of the cumulative equivalent wear of the cutter head of the shield machine after the excavation mileage L meters is calculated by the data sample set Data2′: y=q L (Data2′)+ε; Among them, ε represents random error; (2) After exponential calculation of the logarithm y of the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters, the cumulative equivalent wear of the shield machine cutter head after the excavation mileage L meters is obtained: L ; W L =e y ; The evaluation and prediction results of obtaining the cutter head wear of the shield machine in step seven include: According to the wear level range determined in the actual construction process, check the cumulative equivalent wear W of the shield machine cutter head after the excavation mileage L meters L The wear level range is used to evaluate the wear degree of the shield machine cutter head after the shield machine has excavated a distance of L meters.

Images