Tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance

Abstract

The application discloses a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance, in particular to a shield cutterhead and related equipment operation and maintenance system, and relates to the technical field of tunnel construction equipment operation and maintenance.The system comprises four modules: a model construction module imports a cutterhead BIM model, binds cutter parameters and a unique identifier, and simultaneously deploys sensors to synchronize data and perform processing; a collaborative analysis module extracts cutter residual life, determines maintenance time by correlating construction nodes and a spare parts system; an intelligent allocation module obtains optimal operation and maintenance personnel through two screenings and in combination with distance; and a model optimization module feeds back maintenance deviation to optimize a prediction algorithm.The system realizes equipment operation and maintenance digitization and precision, improves collaborative efficiency, and provides support for stable operation of tunnel construction equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel construction equipment operation and maintenance, in particular to a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance. BACKGROUND

[0002] In the field of tunnel construction, the cutter head of a shield machine is a core operating component, and its operation and maintenance quality directly affects construction efficiency and safety. However, the existing tunnel construction equipment operation and maintenance system has the following defects:

[0003] Firstly, the existing system mostly only builds a basic BIM model, and does not realize precise binding of cutter unique identification and parameters, real-time data, resulting in disconnection between cutter wear data and the model.

[0004] Secondly, the existing prediction algorithm mostly ignores the influence of stratum hardness on wear rate, and only calculates life based on general data, which may lead to inconsistency between the prediction result and the actual working condition of the construction section, and does not establish a time alignment mechanism between life data and construction nodes, resulting in poor matching between maintenance time and construction progress.

[0005] Thirdly, the existing deployment only takes the idle time of the shift as the screening standard, does not consider the actual performance of the personnel such as recent maintenance frequency, time consumption, and maintenance interval of the same cutter, resulting in low matching degree of skills and tasks, and does not consider the distance factor between personnel and construction surface, resulting in low deployment efficiency.

[0006] Therefore, there is an urgent need for a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance. SUMMARY

[0007] In view of the deficiencies of the prior art, the present application provides a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance, which solves the problems of weak digital correlation and poor prediction coordination of the existing tunnel construction equipment operation and maintenance.

[0008] To achieve the above purpose, the present application is implemented by the following technical scheme: a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance, comprising:

[0009] A model construction module imports a three-dimensional BIM model of a cutter head of a shield tunneling machine, binds basic parameters of the cutter head and each cutter, and assigns a unique identifier to each cutter, constructs a digital twin model of the cutter head of the shield tunneling machine, deploys exclusive sensors at key positions of the cutter head, and synchronizes collected cutter head vibration, temperature and cutter head torque data to the BIM model, and the specific operation of constructing the digital twin model of the cutter head of the shield tunneling machine is as follows: the three-dimensional BIM model of the cutter head is obtained from the shield tunneling machine manufacturer, loaded to the system through the model import module of the operation and maintenance system, and geometrically verified; the basic parameters of the cutter head and the cutters are extracted according to the equipment manual: for the cutter head, the diameter, driving power and maximum torque are extracted; for the cutters, the model, material hardness, blade size and rated life are extracted; the extracted basic parameters are associated with the geometric entities of the cutter head and the corresponding cutters in the BIM model through the system parameter binding function, a unique ID is generated according to the rule of "cutter head number-position-number", and the ID is bound to the three-dimensional entity of the corresponding cutter in the BIM model through the system ID mapping function;

[0010] A collaborative analysis module extracts the residual life data of the target cutter from the digital twin model of the cutter head, associates the tunnel construction nodes in the BIM model, judges the matching of the maintenance requirement and the construction progress to determine the reasonable maintenance time, queries the inventory of special spare parts matching the target cutter through the model docking spare parts warehouse system, and

[0011] An intelligent deployment module obtains the maintenance time period of the target cutter, converts it into a standardized time format, and stores the standard maintenance time period to the deployment database, extracts the skill qualifications and current scheduling data of each operation and maintenance personnel from the operation and maintenance personnel management system, performs a screening according to the standard maintenance time period and the current scheduling data, marks potential available personnel, performs a secondary screening based on the maintenance frequency, single maintenance time consumption and cutter maintenance interval, marks actual available personnel, calculates the distance between their real-time regions and construction surfaces, sorts them in ascending order, and preferentially selects the operation and maintenance personnel corresponding to the smaller distance.

[0012] A model optimization module associates the target cutter identifier through the mobile terminal tool, uploads the maintenance process data to the digital twin model in real time, and feeds back the deviation data of the maintenance effect and the predicted result to the digital twin model to optimize the prediction algorithm parameters.

[0013] As a further scheme of the application, the specific steps of deploying exclusive sensors at key positions of the cutter head and synchronizing the collected data to the BIM model are as follows:

[0014] A micro piezoelectric vibration sensor is installed at a position d1 of the cutter handle from the blade, and d1 is in the range of [5cm, 10cm];

[0015] An infrared temperature sensor is installed at a position corresponding to the rear of the cutter d2 supported by the cutter holder, and d2 is in the range of [10cm, 15cm];

[0016] The torque data is directly read from the shield machine main drive PLC system without additional deployment of sensors;

[0017] The micro piezoelectric vibration sensor and the infrared temperature sensor transmit and collect data through an industrial Ethernet, and the torque data of the main drive PLC system is aligned through a time stamp.

[0018] As a further scheme of the application, the original vibration data and temperature data are respectively subjected to median filtering, sliding average filtering, and an abnormal value threshold is set to filter abnormal data, and the processed data are stored according to the cutter ID.

[0019] As a further scheme of the application, a grey prediction GM(1,1) algorithm is integrated in the cutter digital twin model, and the residual life of each cutter is predicted by the algorithm and associated with the corresponding cutter identifier.

[0020] As a further scheme of the application, the matching of the maintenance requirement and the construction progress is determined to determine the specific operation process of the reasonable maintenance time:

[0021] The following trigger conditions are configured in the cutter digital twin model: when the residual life of any cutter is lower than the configurable threshold, the system automatically extracts the unique identifier, the current residual life and the wear rate data of the cutter, and synchronously records the extraction timestamp;

[0022] The tunnel construction node plan divided by mileage is exported from the BIM model, a visual time axis is constructed in the system, the extracted cutter residual life data is aligned and associated with the construction node in the time dimension, and the tunnel construction node plan includes the start time of each node plan and the corresponding stratum type.

[0023] The residual life is corrected by the stratum hardness coefficient to obtain the actual available life of the cutter in the target construction section, and then compared with the construction node time to obtain the required advance time for maintenance.

[0024] As a further scheme of the application, the specific steps of correcting the residual life by the stratum hardness coefficient are:

[0025] The stratum is divided into 5 levels according to hardness, including soft soil = 1.0, silty clay = 1.2, sandstone = 1.5, granite = 1.8, and basalt = 2.0, to form a standardized hardness coefficient table;

[0026] Through the geological information module of the BIM model, the construction section mileage is bound with the corresponding stratum hardness coefficient to ensure that each construction node can directly call the corresponding coefficient;

[0027] The modified formula is obtained: actual available life = residual life / formation hardness coefficient.

[0028] As a further scheme of the present application, the specific steps for determining whether early maintenance is needed and the required early duration are as follows:

[0029] The target construction section plan start time and the interval time of the current cutter to the construction section start are extracted from the BIM model construction plan, and the time parameters are converted into a unified unit;

[0030] The determination rule is set: if the actual available life is greater than or equal to the construction interval time, early maintenance is not needed, otherwise, early maintenance is necessary;

[0031] The calculation formula for the early maintenance duration is: early duration = construction interval time - actual available life + safety redundancy days, wherein the safety redundancy days are set according to the importance of the construction section, if it is a key section, 2 days are taken, if it is a non-key section, 1 day is taken, and the obtained calculation result needs to be rounded up.

[0032] As a further scheme of the present application, the specific process for querying the inventory situation of the special spare parts matched with the target cutter is as follows:

[0033] The corresponding spare part model is bound to the unique ID of each cutter, and then the warehouse code is associated;

[0034] A standardized interface is developed between the digital twin model and the spare parts warehouse management system, and the OPC UA protocol is adopted, the triggering logic of the interface is: when the maintenance demand is determined, the interface is automatically called through the cutter ID to query the real-time inventory quantity and the warehousing time of the corresponding spare parts;

[0035] The inventory quantity is compared with the required quantity to determine whether the inventory quantity is sufficient, the transportation time of the spare parts from the warehouse to the construction surface is calculated to confirm whether it can be delivered before the planned maintenance time, the cutter installation size is called through the BIM model and compared with the spare part parameters to determine the size compatibility;

[0036] If the inventory is insufficient, the system automatically triggers an early warning: the "spare parts shortage" is displayed in the BIM model, and at the same time, the compatible models are called from the alternative scheme library, and the inventory and transportation time of the alternative models are calculated; if there is no alternative model, the procurement application is automatically pushed to the material department.

[0037] As a further scheme of the present application, the steps for screening according to the standard maintenance time period and the current scheduling data are as follows:

[0038] The idle time period in the current scheduling data is extracted based on the standard maintenance time period, the overlapping duration is calculated through the time axis mapping algorithm, and if the overlapping duration is greater than or equal to the maintenance estimated time consumption + buffer time, it is marked as potential available personnel.

[0039] As a further scheme of the present application, the specific operation of secondary screening based on the maintenance frequency, single maintenance time consumption and the same cutter maintenance interval is as follows:

[0040] For the 12-week maintenance data of potential available personnel, three groups of original sequences are extracted: maintenance frequency sequence F, single time consumption sequence T and the same cutter disc maintenance interval sequence S, wherein the elements in F represent the number of maintenance per week, the elements in T represent the average time consumption per week, and the elements in S represent the newly added interval length per week;

[0041] The maintenance frequency sequence F and the same cutter disc maintenance interval sequence S are directly subjected to min-max normalization processing, and the single time consumption sequence T is subjected to reverse min-max normalization processing;

[0042] Two core morphological features of each standardized curve are extracted, including end point trend score and overall trend score:

[0043] The end point trend score is calculated as the average standardized value of the last 4 weeks of the curve, denoted as Fe', Te' and Se';

[0044] The overall trend score is calculated by linear regression to calculate the slope k of the curve in 12 weeks, if k>=0, the trend score=1; if k<0, the trend score=1+2k, denoted as Fk, Tk and Sk;

[0045] The comprehensive level value C is calculated according to the formula C=(Fe'xFk+Te'Tk+Se'Sk);

[0046] The distribution of the comprehensive level value of the tool maintenance quality qualified personnel in history is counted, and the 70% quantile value thereof is taken as the screening threshold Cth, and the operation and maintenance personnel with C>Cth are selected as the actual available personnel;

[0047] For the actual available personnel, the distance between their real-time region and the construction surface is calculated, and they are sorted in ascending order, and the operation and maintenance personnel corresponding to the small distance are preferentially selected.

[0048] The present application provides a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance, which has the following advantages compared with the prior art:

[0049] (1) The present application realizes the precise binding of cutter unique identification, basic parameters and real-time sensor data by constructing a cutter disc digital twinning model, simultaneously, integrates the grey prediction GM(1,1) algorithm and corrects the remaining life by combining with the formation hardness coefficient, avoids the defects of ignoring the working condition difference in the existing prediction, and makes the life prediction more suitable for the construction practice;

[0050] (2) The application solves the problems of poor coordination and mismatch of personnel skills by automatically checking spare parts inventory, transportation timeliness and size compatibility through a standardized interface for spare parts warehouse system, calling alternative solutions when there is a shortage, and screening personnel based on time matching and maintenance performance quantization combined with distance sorting, thereby improving the smoothness of the maintenance process;

[0051] (3) The application adjusts the gray prediction algorithm parameters and key parameter warning thresholds based on the deviation data between the maintenance effect and the prediction result fed back by the operation and maintenance personnel, forms a closed loop of prediction, maintenance, deviation feedback and parameter optimization, avoids the defects of the existing system algorithm accuracy decline under working conditions, and continuously improves the prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The system principle frame of the application;

[0053] Figure 2 The step flow chart of the application for verifying the feasibility of the target tool spare parts supply;

[0054] Figure 3 The step flow chart of the application for secondary screening of operation and maintenance personnel. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0056] Embodiment 1

[0057] As Figure 1 , the application provides a tunnel construction equipment operation and maintenance system based on digital twinning and predictive maintenance, comprising:

[0058] A model construction module imports a shield machine cutterhead three-dimensional BIM model, binds the basic parameters of the cutterhead and each tool and allocates a unique identifier to each tool, and constructs a shield machine cutterhead digital twinning model.

[0059] The three-dimensional BIM model of the cutterhead is obtained from the shield machine manufacturer, loaded to the system through the model import module (supporting lightweight processing of BIM models) of the operation and maintenance system, and geometrically checked to ensure that the cutterhead diameter, tool mounting hole, support seat structure, etc. are consistent with the actual object;

[0060] According to the equipment manual, the basic parameters of the cutterhead and the tool are extracted: for the cutterhead, the diameter, driving power and maximum torque are extracted; for the tool, the model, material hardness, blade size and rated life are extracted;

[0061] The extracted basic parameters are associated with the geometric entities of the cutter head and the corresponding cutters in the BIM model through the system parameter binding function, such as "12 o'clock direction No. 1 cutter" binding "model TS-2023, material tungsten carbide, rated life 800 hours";

[0062] A unique ID is generated according to the "cutter head number-direction-number" rule, such as "D-12-01" representing the D cutter head, 12 o'clock direction, and No. 1 cutter, and the ID is bound to the three-dimensional entity of the corresponding cutter in the BIM model through the system ID mapping function, ensuring that clicking any cutter in the model can directly call its ID and associated parameters;

[0063] The unique ID can realize accurate mapping of physical cutters, digital models, and parameter data, avoiding confusion between multiple cutters, such as the wear data of the 12 o'clock direction No. 1 cutter and the 3 o'clock direction No. 1 cutter;

[0064] For the bound basic parameters, they can be stored in the relational database of the operation and maintenance system (such as MySQL), and a real-time query association is established between the ID field and the BIM model, making it convenient to directly call the basic parameters in the database through the ID field, such as clicking the cutter ID "D-12-01" in the model, the database automatically returns its model, rated life, and other parameters;

[0065] Deploy dedicated sensors at key positions of the cutter head, and synchronize the collected cutter head vibration, temperature, and cutter head torque data to the BIM model, the specific operation steps are as follows:

[0066] Install a micro piezoelectric vibration sensor at the position of d1 between the cutter handle and the cutting edge, d1 can generally be taken as [5cm, 10cm], which has the smallest vibration transmission loss and can most directly reflect the cutter cutting vibration;

[0067] Install an infrared temperature sensor at the position of d2 behind the corresponding cutter of the cutter head support seat, d2 can generally be taken as [10cm, 15cm], without the need for direct contact with the cutter (to avoid affecting the rotation of the cutter);

[0068] Torque data is directly read from the main drive PLC system of the shield tunneling machine, without the need for additional deployment of sensors (using existing equipment data interface);

[0069] The sensors transmit data through industrial Ethernet, and the torque data of the main drive PLC system is aligned through timestamp, ensuring the correlation of the obtained vibration, temperature, and torque data;

[0070] The original vibration data and temperature data are respectively subjected to median filtering and sliding average filtering, and an abnormal value threshold is set to filter abnormal data, and the processed data are stored according to the cutter ID.

[0071] Cutterhead vibration data is susceptible to instantaneous pulse noise, which is manifested as sudden high-amplitude spikes. The median filter is sensitive to pulse noise and can effectively eliminate such isolated spikes while preserving the true vibration trend and meaningful transient impact signals, avoiding the loss of fault warning information due to excessive filtering.

[0072] Cutterhead temperature changes are characterized by slow variation, and noise is mainly small-amplitude random fluctuations. Moving average filtering can smooth such random fluctuations while preserving the overall temperature trend by calculating the average temperature within a certain time window.

[0073] The gray prediction GM(1,1) algorithm is integrated into the model to predict the remaining life of each tool and associate it with the corresponding tool identifier. Due to the small amount of data samples for shield machine tool wear, prediction algorithms like deep learning require a large amount of data for training and cannot meet actual needs. However, the gray algorithm only needs a small amount of data to model.

[0074] The collaborative analysis module extracts the remaining life data of the target tool from the cutterhead digital twin model and associates it with the tunnel construction nodes in the BIM model to determine the matching of maintenance needs and construction progress to determine the reasonable maintenance time.

[0075] The specific operation process for determining the matching of maintenance needs and construction progress to determine the reasonable maintenance time is as follows:

[0076] The following trigger conditions are configured in the cutterhead digital twin model: when the remaining life of any tool is lower than the configurable threshold, the system automatically extracts the unique identifier, current remaining life, and wear rate data of the tool, and synchronously records the extraction timestamp;

[0077] The tunnel construction node plan divided by mileage (including the start time of each node plan and the corresponding stratum type) is exported from the BIM model, and a visual timeline is constructed in the system to align and associate the extracted tool remaining life data with the construction nodes in the time dimension.

[0078] The timeline visualization solves the information fragmentation problem of traditional text or table comparison, allowing maintenance personnel to visually observe the time overlap relationship between tool life and construction nodes.

[0079] The actual available life of the tool in the target construction section is obtained by correcting the remaining life with the stratum hardness coefficient:

[0080] The stratum is divided into 5 levels according to hardness (reference: "Geotechnical Engineering Investigation Specification"), including soft soil = 1.0, silty clay = 1.2, sandstone = 1.5, granite = 1.8, and basalt = 2.0, forming a standardized hardness coefficient table.

[0081] Through the geological information module of the BIM model, the construction section mileage is bound with the corresponding stratum hardness coefficient, such as the K2+500-K2+600 section bound coefficient 1.5, to ensure that each construction node can directly call the corresponding coefficient;

[0082] A correction formula is obtained: actual available life = residual life / stratum hardness coefficient, which is logically based on the engineering rule that "the higher the stratum hardness, the faster the tool wear rate";

[0083] Then, compared with the construction node time, it is determined whether early maintenance is needed and the required early time length, and the specific steps are as follows:

[0084] Two core times are extracted from the BIM model construction plan: ① target construction section plan start time; and ② tool current interval time to construction section start, and the time parameters are converted into a unified unit (such as "day", with two decimal places), to ensure consistent comparison dimensions;

[0085] The determination rule is set: if the actual available life ≥ construction interval time, it indicates that the tool can support the completion of the construction section and no early maintenance is needed, otherwise, it indicates that the tool will fail due to excessive wear before the completion of the construction section and must be maintained in advance;

[0086] The calculation formula for the early maintenance time length is: early time length = construction interval time-actual available life + safety redundancy days, wherein the safety redundancy days are set according to the importance of the construction section, if it is a key section, 2 days are taken, if it is a non-key section, 1 day is taken, to cope with sudden delays, such as spare parts transportation delay, on-site sudden failure, and the obtained calculation result needs to be rounded up to ensure sufficient time;

[0087] Through the model docking spare parts warehouse system, the inventory situation of special spare parts matched with the target tool is queried to verify the feasibility of the target tool spare supply.

[0088] The intelligent deployment module obtains the maintenance time period of the target tool and converts it into a standardized time format, such as YYYY-MM-DD HH:MM-HH:MM, and stores the standard maintenance time period to the deployment database, wherein the maintenance time period includes an absolute period and a time window;

[0089] The absolute period refers to a fixed start and end date range covering the entire tool maintenance process determined according to the early time length;

[0090] The time window refers to a daily fixed time period determined according to the construction gap of the shield machine;

[0091] For example, the "D-12-01" cutter of a certain shield machine needs to be replaced before the wear reaches the threshold value, and maintenance needs to be started 4 days in advance to reserve buffer time. If it is predicted that the cutter will reach the maximum allowed wear on May 20, the absolute time period can be determined as May 16-May 19, but the time window can be set as 12:00-14:00 every day, which is the established construction gap;

[0092] From the operation and maintenance personnel management system, two types of key data are extracted: ① skill qualifications, such as cutter maintenance level 1 qualification and hydraulic tool operation certification, only personnel with cutter maintenance qualifications are selected, and personnel without qualifications are excluded; ② current scheduling data, such as X month X day 10:00-12:00 free, 12:00-14:00 already scheduled for tunnel inspection, 14:00-16:00 free, which also needs to be converted to a standardized time format and aligned with the maintenance period data in the field;

[0093] A correlation index table is constructed with maintenance period ID and personnel ID as the double primary keys, ensuring that each maintenance period data can be quickly matched to the personnel scheduling data that meets the skill requirements, avoiding cross-system data query delays;

[0094] According to the standard maintenance time period and the current scheduling data, a screening is performed to mark potential available personnel:

[0095] Based on the standard maintenance time period, the free time period in the current scheduling data is extracted; the overlap duration of the two is calculated through a time axis mapping algorithm; if the overlap duration is greater than or equal to the estimated duration of maintenance + buffer time (buffer time for on-site preparation), it is marked as a potential available personnel;

[0096] For potential available personnel, a secondary screening is performed based on maintenance frequency, single maintenance duration, and cutter maintenance interval.

[0097] The model optimization module allows operation and maintenance personnel to associate target cutter identifiers through mobile tools and upload maintenance process data to the digital twin model in real time;

[0098] The operation and maintenance personnel carry mobile devices with NFC function, approach the NFC chip built-in the target cutter handle, the device automatically reads the ID and triggers the maintenance data collection template to load, the template includes cutter installation completion time, cutter installation completion picture, and cutter installation completion sensor data;

[0099] After the operation and maintenance personnel fill in the data according to the template, the filled data is preferentially uploaded to the digital twin model in real time through UWB positioning + 5G fusion network in the tunnel, if the network is interrupted (such as signal blind area in deep tunnel), the mobile device automatically starts offline caching function, when entering the network area, the device automatically encrypts and supplements the data, which can be encrypted by AES-256 to prevent data tampering;

[0100] The deviation data between the maintenance effect and the predicted result is fed back to the digital twin model to optimize the prediction algorithm parameters, and the specific process is as follows:

[0101] The predicted data Dp before tool maintenance and the actual data Da after maintenance are called from the database;

[0102] Dp includes: residual life Lp output by the grey prediction model, vibration threshold Vp, temperature threshold Tp, and torque threshold Ep;

[0103] Da includes: the life La of the tool actually running to the next maintenance after maintenance, vibration average Va, temperature average Ta, and torque average Ea;

[0104] According to Dp and Da, the life deviation rate , the vibration deviation rate , the temperature deviation rate , and the torque deviation rate are calculated respectively.

[0105] If , the development coefficient of the grey prediction GM(1,1) model is adjusted according to the rule that for every 10% of , the original development coefficient is adjusted in the decreasing direction by to obtain a new coefficient .

[0106] If , the vibration threshold is updated to , wherein G% is a safety redundancy coefficient to avoid excessive fitting of the threshold to single data.

[0107] According to the vibration threshold updating principle, the updated temperature threshold is , and the torque threshold is .

[0108] The corrected parameters are applied to the life prediction of other tools of the same type, and if the requirements are met, the parameters are fixed, otherwise, the above steps are continued to correct the parameters.

[0109] Embodiment 2

[0110] This embodiment further discloses a method for verifying the feasibility of target tool spare parts supply based on embodiment 1, as shown in Figure 2 , the specific content includes:

[0111] Bind the corresponding spare parts model to the unique ID of each tool, and then associate the warehouse code, such as "W2-05-12", which represents the 2nd warehouse, 5th area, and 12th shelf. Store the mapping relationship in the blockchain database to ensure that the association between the model and the code is tamper-proof.

[0112] Develop a standardized interface between the digital twin model and the spare parts warehouse management system using the OPC UA protocol. The trigger logic for this interface is as follows: when maintenance needs are determined, automatically call the interface through the tool ID to query the real-time inventory quantity and storage time of the corresponding spare parts.

[0113] Verify the adequacy of spare parts supply from three aspects: quantity, timeliness, and adaptability. Compare the required quantity for maintenance with the inventory quantity to determine if the inventory quantity is sufficient. Calculate the transportation time from the warehouse to the construction site to confirm whether it can be delivered before the planned maintenance time. Retrieve the tool installation dimensions from the BIM model and compare them with the spare parts parameters to determine size compatibility.

[0114] If the inventory is insufficient, the system automatically triggers an early warning: display "spare parts shortage" in the BIM model, and simultaneously retrieve compatible models from the alternative solution library, such as "TS-2024 and TS-2023 have the same installation dimensions and can be used as substitutes." Calculate the inventory and transportation time of the alternative models. If there are no alternative models, automatically push a procurement application to the material department.

[0115] Embodiment 3

[0116] This embodiment further discloses a secondary screening method based on maintenance frequency, single maintenance time, and interval between maintenance of the same tool, based on embodiments 1 and 2. As shown in FIG. 12, the specific process is as follows: Figure 3

[0117] Extract three groups of original sequences from the 12-week maintenance data of potential available personnel: maintenance frequency sequence F, single time consumption sequence T, and interval sequence S between maintenance of the same tool. The elements in F represent the number of weekly maintenance, the elements in T represent the average weekly time consumption, and the elements in S represent the newly added interval length per week.

[0118] Directly perform min-max normalization processing on the maintenance frequency sequence F and the interval sequence S between maintenance of the same tool. Since both sequences represent that the larger the sequence value, the better the maintenance quality, the normalization can be directly used for scaling.

[0119] ​The reverse processing of min-max normalization of a single time-consuming sequence T has the core logic that the original time-consuming data is first mapped to the [0, 1] interval through min-max normalization, and then 1 is subtracted from the value to realize reverse mapping. The shorter (the better) the original time-consuming sample is, the closer the standardized value is to 1, and the longer (the worse) the time-consuming sample is, the closer the standardized value is to 0, so that the index is consistent with the evaluation direction of the above F, S and the like which are larger and better.

[0120] Two core morphological features of each standardized curve are extracted, including end point trend score and overall trend score:

[0121] The end point trend score calculates the average standardized value of the last 4 weeks (weeks 9-12) of the curve (reflecting the recent state), denoted as Fe', Te', and Se';

[0122] The overall trend score calculates the slope k of the curve for 12 weeks through linear regression. If k≥0 (rising or stable), the trend score = 1; if k<0 (falling), the trend score = 1+2k (k is closer to 0, the score is higher, and small fluctuations are allowed), denoted as Fk, Tk, and Sk.

[0123] The comprehensive level value C is calculated according to the formula C=(Fe'×Fk+Te'×Tk+Se'×Sk);

[0124] The distribution of the comprehensive level value of the tool maintenance quality qualified personnel in history is counted, and the 70% quantile value thereof is taken as the screening threshold Cth. The C>Cth operation and maintenance personnel are selected as the actual available personnel;

[0125] For the actual available personnel, the distance between their real-time region and the construction surface is calculated, and they are sorted in ascending order, and the operation and maintenance personnel corresponding to the small distance are preferentially selected.

[0126] Some data in the above formula are dimensionless numerical calculations, and the contents not described in detail in the specification are all prior art known to those skilled in the art.

[0127] The above examples are only used to illustrate the technical method of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A tunnel construction equipment operation and maintenance system based on digital twins and predictive maintenance, characterized in that, include: The model building module imports the 3D BIM model of the tunnel boring machine (TBM) cutterhead, binds the basic parameters of the cutterhead and each cutter, assigns a unique identifier to each cutter, and constructs a digital twin model of the TBM cutterhead. Dedicated sensors are deployed at key locations on the cutterhead, and the collected data on cutterhead vibration, temperature, and torque are synchronized to the BIM model. The specific operations for constructing the digital twin model of the TBM cutterhead are as follows: The 3D BIM model of the cutterhead is obtained from the TBM manufacturer, loaded into the system through the model import module of the operation and maintenance system, and geometrically verified. Basic parameters of the cutterhead and cutters are extracted according to the equipment manual: for the cutterhead, diameter, drive power, and maximum torque are extracted; for the cutters, model, material hardness, cutting edge size, and rated life are extracted. Through the system parameter binding function, the extracted basic parameters are associated with the geometric entities of the cutterhead and corresponding cutters in the BIM model, generating a unique ID according to the "cutterhead number-orientation-serial number" rule, and then binding the ID to the 3D entity of the corresponding cutter in the BIM model through the system ID mapping function. The collaborative analysis module extracts the remaining life data of the target tool from the cutterhead digital twin model and associates it with the tunnel construction nodes in the BIM model to determine the matching between maintenance needs and construction progress to determine a reasonable maintenance time. It also connects the model to the spare parts warehouse system to query the inventory of special spare parts that match the target tool. The intelligent allocation module obtains the maintenance time period of the target tool, converts it into a standardized time format, and stores the standard maintenance time period in the allocation database. It extracts the skill qualifications and current shift data of each maintenance personnel from the maintenance personnel management system, performs a first screening based on the standard maintenance time period and the current shift data, marks potential available personnel, and performs a second screening based on maintenance frequency, single maintenance time, and maintenance interval of the same tool to mark actual available personnel. It calculates the distance between their real-time location and the construction surface, sorts them in ascending order, and prioritizes the maintenance personnel corresponding to the smaller distance. In the model optimization module, maintenance personnel can associate target tool identifiers with mobile tools and upload maintenance process data to the digital twin model in real time. They can also feed back the deviation data between the maintenance effect and the prediction result to the digital twin model to optimize the prediction algorithm parameters.

2. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific steps for deploying dedicated sensors at key locations on the cutterhead and synchronizing the collected data to the BIM model are as follows: A miniature piezoelectric vibration sensor is installed at a distance d1 from the cutting edge of the tool holder, where the value of d1 is in the range of [5cm, 10cm]. An infrared temperature sensor is installed on the tool holder at a position d2 directly behind the tool, with the value range of d2 being [10cm, 15cm]. Torque data is read directly from the main drive PLC system of the tunnel boring machine, eliminating the need for additional sensor deployment; The miniature piezoelectric vibration sensor and infrared temperature sensor transmit data via industrial Ethernet and are aligned with the torque data of the main drive PLC system via timestamps.

3. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 2, characterized in that, The original vibration and temperature data were subjected to median filtering and moving average filtering, respectively, and an outlier threshold was set to filter out abnormal data. The processed data were then stored according to tool ID.

4. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The gray prediction GM(1,1) algorithm is integrated into the tool digital twin model to predict the remaining life of each tool and associate it with the corresponding tool identifier.

5. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific operational procedure for determining the match between maintenance needs and construction schedule to ascertain reasonable maintenance time is as follows: Configure the following triggering conditions in the tool head digital twin model: When the remaining life of any tool is lower than the configurable threshold, the system automatically extracts the tool's unique identifier, current remaining life, and wear rate data, and records the extraction timestamp simultaneously; Export the tunnel construction node plan divided by mileage from the BIM model, build a visual time axis in the system, and align and associate the extracted remaining tool life data with the construction nodes according to the time dimension. The tunnel construction node plan includes the start time of each node plan and the corresponding stratum type. By correcting the remaining lifespan using the formation hardness coefficient, the actual usable lifespan of the cutting tool in the target construction section is obtained. Then, by comparing it with the construction node time, the advance maintenance time required is determined.

6. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific steps for correcting the remaining lifetime using the formation hardness coefficient are as follows: The strata are divided into 5 levels according to hardness, specifically including soft soil = 1.0, silty clay = 1.2, sandstone = 1.5, granite = 1.8, and basalt = 2.0, forming a standardized hardness coefficient table; By using the geological information module of the BIM model, the construction section mileage is linked to the corresponding stratum hardness coefficient, ensuring that the corresponding coefficient can be directly retrieved for each construction node. The corrected formula is: Actual usable lifespan = Remaining lifespan / Formation hardness coefficient.

7. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific steps to determine whether advance maintenance is needed and the required lead time are as follows: Extract the planned start time of the target construction section and the interval between the current tool and the start of the construction section from the BIM model construction plan, and convert the time parameters into a unified unit; Set the judgment rule: if the actual usable life is greater than or equal to the construction interval, then no maintenance is required in advance; otherwise, maintenance must be carried out in advance. The formula for calculating advance maintenance time is: Advance maintenance time = construction interval time - actual usable life + safety redundancy days. The safety redundancy days are set according to the importance of the construction section. If it is a critical section, take 2 days; if it is a non-critical section, take 1 day. The calculated result needs to be rounded up.

8. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific procedure for checking the inventory of dedicated spare parts that match the target tool is as follows: Each tool is assigned a unique ID that corresponds to a spare part model, which is then linked to the warehouse code. A standardized interface was developed between the digital twin model and the spare parts warehouse management system, using the OPC UA protocol. The triggering logic of this interface is as follows: when maintenance needs are determined, the interface is automatically called through the tool ID to query the real-time inventory quantity and entry time of the corresponding spare parts. Compare the required maintenance quantity with the inventory quantity to determine if the inventory quantity is sufficient; calculate the transportation time of spare parts from the warehouse to the construction site to confirm whether they can be delivered before the planned maintenance time; retrieve the tool installation dimensions from the BIM model and compare them with the spare parts parameters to determine dimensional compatibility; If the inventory is insufficient, the system will automatically trigger an alert: "Spare parts shortage" will be displayed in the BIM model, and a compatible model will be retrieved from the alternative solution library, and the inventory and delivery time of the alternative model will be calculated. If no alternative model is available, the procurement request will be automatically sent to the materials department.

9. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The steps for filtering based on standard maintenance time periods and current scheduling data are as follows: Based on the standard maintenance time period, extract the idle time period from the current shift data; calculate the overlap time between the two using a time axis mapping algorithm; if the overlap time is greater than or equal to the estimated maintenance time plus the buffer time, then mark it as a potentially available personnel.

10. The tunnel construction equipment operation and maintenance system based on digital twin and predictive maintenance according to claim 1, characterized in that, The specific steps for secondary screening based on maintenance frequency, single maintenance time, and maintenance interval for the same tool are as follows: For the 12-week maintenance data of potential available personnel, three sets of raw sequences are extracted: maintenance frequency sequence F, single time consumption sequence T, and maintenance interval sequence S for the same tool head. The elements in F represent the number of maintenance times per week, the elements in T represent the average time consumption per week, and the elements in S represent the newly added interval duration per week. The maintenance frequency sequence F and the maintenance interval sequence S of the same tool head are directly processed by min-max normalization, and the single time consumption sequence T is processed by min-max normalization in reverse. Extract two core morphological features from each standardized curve, including the endpoint trend score and the overall trend score: The endpoint trend score is calculated by taking the average standardized value of the curve over the last four weeks, denoted as Fe', Te', and Se'. The overall trend score is calculated by linear regression of the slope k of the curve over 12 weeks. If k ≥ 0, the trend score = 1; if k < 0, the trend score = 1 + 2k, denoted as Fk, Tk, and Sk. The comprehensive level value C is calculated using the formula C=(Fe'×Fk+Te'×Tk+Se'×Sk); The historical distribution of the comprehensive skill level of personnel who meet the quality standards for tool maintenance was statistically analyzed, and the 70th percentile value was taken as the screening threshold Cth. Maintenance personnel with C > Cth were selected as actual usable personnel. For available personnel, calculate the distance between their real-time location and the construction site, sort them in ascending order, and prioritize the maintenance personnel with the shorter distance.

Images