Shield cutter cutting load real-time monitoring and adjusting method and system

Abstract

The invention discloses a shield cutter cutting load real-time monitoring and adjusting method and system. The method comprises the following steps: constructing a shield cutter distributed sensing network, and collecting strain, vibration and temperature of a cutter; a wireless transmission network is established, and the collected strain, vibration and temperature of the cutter are transmitted in real time; according to the strain, vibration and temperature, received in real time, of the cutter, cutting load parameters are obtained in a multi-source data fusion mode; based on the obtained cutting load parameters, stratum parameter identification is carried out in combination with a deep learning algorithm; in combination with the stratum parameters and the cutting load parameters, a wear state evaluation model based on fuzzy reasoning is established, and the tool wear state is evaluated; and designing a multi-objective optimization function according to the wear state of the cutter and the stratum parameters, solving the multi-objective optimization function, obtaining optimal shield tunneling machine operation parameters, and performing dynamic adjustment. The cutting load of each cutter in shield construction can be monitored in real time, and the cutter parameters can be adaptively adjusted.

Description

Technical Field

[0001] This application relates to the field of engineering construction monitoring and adjustment technology, specifically to a method and system for real-time monitoring and adjustment of the cutting load of tunnel boring machine tools. Background Technology

[0002] Tunnel boring machine (TBM) technology plays a crucial role in underground engineering construction. It enables efficient and safe tunnel excavation and is widely used in urban subways, water conservancy projects, and transportation tunnels. With the acceleration of urbanization and the increasing demand for infrastructure construction, the development of TBM technology is of great significance for improving project efficiency, ensuring project quality, and reducing construction risks.

[0003] In existing tunnel boring machine (TBM) construction processes, monitoring the cutting load on the cutter head typically employs post-construction analysis or indirect measurement methods. Post-construction analysis generally involves analyzing cutter wear and TBM operating parameters after the TBM is completed to infer the stress on the cutter head during construction. Indirect measurement involves measuring macroscopic parameters such as the TBM's total thrust and torque to estimate the stress on the cutter head. While these conventional methods can provide some reference for construction, they have limitations. They cannot obtain real-time data on the actual cutting force when the cutter head contacts the ground. Furthermore, operators often rely solely on experience when adjusting cutter parameters, lacking scientific basis and precise control methods.

[0004] However, existing monitoring and adjustment methods have significant drawbacks. The inability to accurately monitor the cutting load of each tool in real time makes it difficult to distinguish the stress differences between different tools, leading to uneven tool wear and untimely tool replacement, which severely impacts construction efficiency and safety. Furthermore, the lack of an ability to automatically adjust tool parameters based on real-time cutting load data makes it difficult for operators to meet the precise control requirements under complex geological conditions and hinders the dynamic optimization and adjustment of tunnel boring machine operating parameters. Summary of the Invention

[0005] In order to monitor the cutting load of each tool in shield tunneling in real time and adaptively adjust the tool parameters, this application provides a method and system for real-time monitoring and adjustment of the cutting load of shield tunneling tools.

[0006] In a first aspect, this application provides a method for real-time monitoring and adjustment of cutting load on a tool, including: Construct a distributed sensor network for tunnel boring machine cutters to collect data on the cutter's strain, vibration, and temperature; Establish a wireless transmission network to transmit the collected strain, vibration, and temperature of the cutting tool in real time; Based on the real-time received data on the tool's strain, vibration, and temperature, cutting load parameters are obtained through multi-source data fusion. Based on the acquired cutting load parameters, formation parameters are identified using deep learning algorithms. By combining formation parameters and cutting load parameters, a wear state assessment model based on fuzzy inference is established to evaluate the tool wear state. Based on the cutter wear state and formation parameters, a multi-objective optimization function is designed, the multi-objective optimization function is solved, the optimal shield machine operating parameters are obtained, and dynamic adjustments are made.

[0007] By adopting the above scheme, a distributed sensor network for tunnel boring machine (TBM) cutters is constructed to accurately collect cutter status data; a wireless transmission network is established to transmit the collected data in real time; cutting load parameters are obtained through multi-source data fusion to improve the accuracy of cutting load parameter acquisition; formation parameters are identified by combining deep learning algorithms to achieve intelligent identification of formation parameters; a wear state assessment model based on fuzzy reasoning is established to assess the cutter wear state and accurately evaluate the cutter wear situation; and a multi-objective optimization function is designed to solve and dynamically adjust the optimal TBM operating parameters, realizing the automatic optimization of TBM operating parameters.

[0008] Preferably, the construction of the shield tunneling cutter distributed sensing network includes: Multiple types of sensors, including strain sensors, acceleration sensors, vibration sensors, and temperature sensors, are arranged at the installation positions of the cutterhead of the tunnel boring machine. Among them, a non-uniform distribution strategy is adopted when arranging sensors on the cutterhead of the tunnel boring machine. Sensor arrays with different densities are set in the central area, middle area, and edge area of ​​the cutterhead, and at least one set of multiple types of sensors is arranged on each cutter.

[0009] By adopting the above scheme, a non-uniform distribution strategy is used to set up sensor arrays of different densities in different areas of the cutter head, which can focus on monitoring easily worn parts; at least one set of multi-type sensors is arranged on each tool to ensure that relevant data of each tool is effectively collected, thereby realizing accurate monitoring of the tool stress state.

[0010] Preferably, real-time data transmission via a wireless transmission network includes: Multi-channel synchronous data acquisition is performed; a dual-band wireless transmission method with a primary frequency band and a backup frequency band is adopted, and automatic power adjustment and frequency band switching operations are performed during wireless transmission; the automatic power adjustment refers to dynamically adjusting the transmission power according to the signal strength of the acquired data; the frequency band switching operation refers to dynamically switching the frequency band according to the signal quality of the acquired data.

[0011] By adopting the above scheme, multi-channel synchronous data acquisition ensures data synchronization and integrity. By combining dual-band wireless transmission with automatic power adjustment and frequency band switching, the reliability and stability of data transmission are improved, ensuring real-time and accurate data transmission.

[0012] Preferably, the step of obtaining cutting load parameters through multi-source data fusion based on the real-time received tool strain, vibration, and temperature includes: According to preset data quality assessment rules, the data quality of the real-time received tool strain, vibration and temperature is assessed; the preset data quality assessment rules include: the signal-to-noise ratio of tool strain, the dynamic fluctuation parameters of tool vibration signal, and the stability parameters of tool temperature. According to preset decision rules, multi-source data fusion methods are matched based on data quality assessment results, including: a first multi-source data fusion method based on Kalman filtering, a second multi-source data fusion method based on neural networks, and a joint multi-source data fusion method based on Kalman filtering and neural networks. The preset decision rules include: if the tool strain has a low signal-to-noise ratio, the second multi-source data fusion method is directly matched; if the tool strain has a medium signal-to-noise ratio, and the vibration is judged to be low fluctuation, the first multi-source data fusion method is matched, and the joint multi-source data fusion method is replaced when the temperature is low and stable; if the vibration is judged to be medium fluctuation, the joint multi-source data fusion method is matched; if the vibration is judged to be high fluctuation, the second multi-source data fusion method is matched; if the tool strain has a high signal-to-noise ratio, and the vibration is judged to be low fluctuation, the first multi-source data fusion method is matched, and the joint multi-source data fusion method is replaced when the temperature is low and stable; if the vibration is judged to be medium fluctuation, the joint multi-source data fusion method is matched; if the vibration is judged to be high fluctuation, the second multi-source data fusion method is matched, and the joint multi-source data fusion method is replaced when the temperature is high and stable.

[0013] By adopting the above scheme, the data quality of tool strain, vibration and temperature is evaluated according to preset rules, and different multi-source data fusion methods are matched according to the evaluation results. The most suitable fusion method is selected for different data quality conditions, thereby obtaining cutting load parameters more accurately and improving the accuracy and pertinence of data fusion.

[0014] Preferably, the formation parameter identification based on the acquired cutting load parameters and combined with a deep learning algorithm includes: Based on the real-time acquired cutting load parameters, the torque and power consumption are calculated; An improved formation parameter identification model is constructed using deep learning algorithms. The model structure includes convolutional layers, pooling layers, fully connected layers, and long short-term memory network layers. The inputs of the improved formation parameter identification model are real-time cutting load data, historical load data, real-time calculated torque and power consumption, and known formation parameters. The output is the current formation parameters, including formation type, strength parameters, and corrosion parameters.

[0015] By adopting the above scheme, considering torque and power consumption, and combining deep learning algorithms to construct a formation parameter identification model, the accurate identification of the current formation parameters can be achieved.

[0016] Preferably, the step of establishing a wear state assessment model based on fuzzy inference by combining formation parameters and cutting load parameters, and assessing the tool wear state includes: By combining formation parameters and cutting load parameters, key features are extracted, including: mean cutting force and vibration spectrum characteristics; A wear state assessment model based on fuzzy reasoning is established, defining key features as input variables and the membership function of wear state as output variable; Design a rule base for reasoning, including rules for non-mixed strata and rules for mixed strata; Based on the identified current stratigraphic parameters, the wear amount is obtained by adaptively matching non-mixed stratigraphic rules or mixed stratigraphic rules in the rule base, performing fuzzy reasoning and defuzzification. To identify the current formation parameters as a mixed formation type, a nonlinear correction mechanism for mixed formation parameters is introduced. The formation parameter correction factor is calculated, and the obtained wear amount is multiplied by the formation parameter correction factor to obtain the final wear amount.

[0017] By adopting the above scheme, combining formation parameters and cutting load parameters, and using a fuzzy inference-based wear state assessment model, the wear state of tools under different formation types can be accurately assessed. In particular, nonlinear corrections are made for mixed formations to improve the accuracy and reliability of tool wear assessment.

[0018] Preferably, the step of designing a multi-objective optimization function based on the cutter wear state and formation parameters, solving the multi-objective optimization function, obtaining the optimal shield machine operating parameters, and dynamically adjusting them includes: A mathematical model of the operating parameters and performance indicators of a tunnel boring machine (TBM) is constructed by combining response surface methodology and neural networks to obtain the mapping relationship between the operating parameters and performance indicators. The operating parameters include cutterhead rotation speed, propulsion speed, cutterhead opening ratio, and grouting parameters; the performance indicators include tunneling efficiency and energy consumption. The design includes multiple objective functions, such as maximizing tunneling efficiency, minimizing energy consumption, minimizing wear rate, and maintaining stability; design constraints, such as physical constraints (cutterhead rotation speed range, feed speed range, and opening ratio adjustment range); safety constraints (cutter normal force range and cutterhead torque range); and formation type adaptation constraints (adapting cutterhead rotation speed range, feed speed range, and opening ratio adjustment range to different formation types). The weights of the multi-objective function are dynamically adjusted based on the real-time identified cutter wear status and geological type to complete the multi-objective optimization solution and obtain the optimal shield machine operating parameters.

[0019] By adopting the above scheme, the mapping relationship between the operating parameters and performance indicators of the tunnel boring machine is accurately established. Multi-objective functions and constraints are designed by comprehensively considering various factors, and the weights are dynamically adjusted according to the real-time situation to complete the optimization solution, thereby improving tunneling efficiency, reducing energy consumption, reducing cutter wear, and ensuring the stability of tunnel boring machine operation.

[0020] Preferred options also include: The system monitors and acquires cutting load parameters, tool wear status, and formation parameters in real time. It then determines whether these parameters exceed the corresponding preset threshold or whether there are any changes in the preset parameters that exceed the corresponding preset threshold. If any of these parameters exceed the corresponding preset threshold or change threshold, a warning message is generated for the corresponding parameter.

[0021] By adopting the above scheme, cutting load parameters, tool wear status and formation parameters can be monitored in real time. This allows for timely detection of situations where parameters exceed the corresponding preset threshold or parameter changes exceed the corresponding preset change threshold, generating corresponding early warning prompts and providing early warnings of abnormal conditions for tunnel boring machine construction.

[0022] Preferred options also include: A digital twin model of the tunnel boring machine (TBM) cutter is constructed, and the real-time acquired cutting load parameters, cutter wear status, and formation parameters are mapped into the digital twin model to achieve visualization and predictive analysis of the TBM's operating status.

[0023] By adopting the above scheme, a digital twin model of the tunnel boring machine cutter is constructed and real-time parameters are mapped into it, which enables an intuitive presentation of the tunnel boring machine's operating status and predictive analysis of its future status.

[0024] Secondly, this application provides a real-time monitoring and adjustment system for the cutting load of a tunnel boring machine cutter, comprising: The tool data acquisition module is used to build a distributed sensor network for tunnel boring machine tools to collect the strain, vibration and temperature of the tools. The tool data transmission module is used to establish a wireless transmission network to transmit the collected tool strain, vibration and temperature in real time. The tool data fusion module is used to obtain cutting load parameters by fusing multi-source data based on the real-time received tool strain, vibration and temperature. The formation parameter identification module is used to identify formation parameters based on the acquired cutting load parameters and combined with deep learning algorithms. The tool wear assessment module is used to combine formation parameters and cutting load parameters to establish a wear state assessment model based on fuzzy inference and assess the tool wear state. The cutter parameter adjustment module is used to design a multi-objective optimization function based on the cutter wear state and formation parameters, solve the multi-objective optimization function, obtain the optimal shield machine operating parameters, and dynamically adjust them. By adopting the above scheme, real-time and accurate monitoring of the cutting load of the tunnel boring machine cutter can be achieved, intelligent identification of geological parameters can be achieved, accurate assessment of the wear status of the cutter can be achieved, and automatic optimization and adjustment of the operating parameters of the tunnel boring machine can be achieved. This reduces uneven wear of the cutter and untimely replacement, lowers construction risks, and improves the level of intelligence in tunnel boring construction.

[0025] In summary, this application has the following beneficial effects: By constructing a distributed sensor network for the tunnel boring machine (TBM) cutters and employing a non-uniform distribution strategy to deploy multiple types of sensors, the status data of each cutter is collected in real time and accurately. By establishing a wireless transmission network and utilizing a dual-band design, automatic switching mechanism, and automatic power adjustment mechanism, real-time and efficient data transmission from the cutterhead is achieved. Based on the real-time collected strain, vibration, and temperature data, the cutting load parameters of the cutters are obtained through multi-source data fusion. Combined with deep learning algorithms, geological parameters are identified, and the wear state of the cutters is evaluated based on fuzzy inference. Finally, a multi-objective optimization function is designed to dynamically adjust the operating parameters of the TBM, thus realizing the automatic optimization of the TBM operating parameters. Attached Figure Description

[0026] Figure 1 This is a flowchart of the method for real-time monitoring and adjustment of the cutting load of the tunnel boring machine cutter as described in a specific embodiment; Figure 2 This is a schematic diagram of the real-time monitoring and adjustment system for the cutting load of the tunnel boring machine cutter described in a specific embodiment. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0028] like Figure 1 As shown in the figure, this application discloses a method for real-time monitoring and adjustment of cutting load on tunnel boring machine cutters, and the specific steps are as follows: S1. Construct a distributed sensor network for the tunnel boring machine cutter to collect the operating status data of the cutter.

[0029] Specifically, constructing a distributed sensor network for tunnel boring machine (TBM) cutters involves deploying various types of sensors, such as strain sensors, accelerometers, vibration sensors, and temperature sensors, at the cutterhead mounting locations to collect operational status data of the cutters, including strain, vibration, and temperature. The strain sensors are fiber optic strain sensors, selectively installed at the connection between the cutterhead and the cutter holder, secured with a dedicated high-strength alloy clamp to ensure a tight fit between the sensor and the cutter holder. The sensor surface is coated with a special protective layer to prevent mud erosion. The accelerometers are triaxial piezoelectric accelerometers, bolted to the non-cutting surface of the cutter body. The sensor housing is made of titanium alloy, and the internal circuitry is sealed with epoxy resin to ensure stable operation under high pressure and high humidity environments. Vibration sensors, such as accelerometers, monitor the vibration frequency and amplitude of the cutter and are selectively placed on the cutter base. Temperature sensors, such as thermocouples or fiber optic temperature sensors, can be placed inside or on the surface of the cutter.

[0030] To focus on monitoring wear-prone areas, a non-uniform distribution strategy is adopted for the sensor layout during the construction of the distributed sensor network for the tunnel boring machine cutterhead. Sensor arrays of different densities are set in the central, middle, and edge regions of the cutterhead, respectively. Considering the gradual decrease in wear levels in the edge, middle, and center regions, sensor arrays with progressively lower densities are used. To ensure effective data acquisition from each cutter, at least one set of multi-type sensors (strain sensors, acceleration sensors, vibration sensors, and temperature sensors) is installed on each cutter; typically, two sets are used, with one set of multi-type sensors designed as a redundancy. Each sensor is equipped with independent signal conditioning circuitry, including a preamplifier, low-pass filter, and analog-to-digital converter, to perform preliminary signal processing.

[0031] S2. Establish a wireless transmission network to transmit the collected shield cutter operating status data in real time.

[0032] Specifically, based on the sensor network constructed in step S1, a wireless data transmission network is designed, including setting up and installing several data transmission devices. Each data transmission device includes a data acquisition unit and a data transmission unit. The data acquisition unit adopts a modular design, including a multi-channel synchronous acquisition unit, a digital signal processing unit, and a data buffer unit, and is installed in a protective compartment on the back of the cutterhead. The data acquisition unit connects to each sensor via waterproof connectors, using a star topology to reduce signal transmission distance. The data transmission unit adopts a dual-band design, with the primary band operating in the industrial-grade communication band and the backup band operating in a different band. An automatic switching mechanism ensures reliable signal transmission. The data transmission unit uses a directional antenna pointing towards the tunnel boring machine control center and features an automatic power adjustment mechanism that dynamically adjusts the transmission power based on signal strength. The tunnel boring machine control center includes a high-gain receiving antenna, a signal demodulation module, and a high-performance computing processor, receiving and processing the transmitted data. Furthermore, the data transmission device can be loaded with a data compression algorithm to perform lossless compression of the acquired data, improving transmission efficiency; it also implements encrypted data transmission to prevent external interference and data theft.

[0033] Real-time data transmission via wireless transmission network includes: multi-channel synchronous data acquisition to obtain tool strain, vibration, and temperature data; employing a dual-band wireless transmission method with a main band and a backup band; switching to the backup band if the signal quality of the main band transmission data is lower than a preset signal quality, i.e., performing a band switching operation during wireless transmission; correspondingly, adjusting the transmission power of the enhanced data transmission unit if the signal strength of the transmitted data is lower than a preset signal strength during wireless transmission; the automatic power adjustment refers to dynamically adjusting the transmission power based on the signal strength of the acquired data; the band switching operation refers to dynamically switching the band based on the signal quality of the acquired data.

[0034] S3. Based on the real-time received tool strain, vibration and temperature, obtain cutting load parameters through multi-source data fusion.

[0035] Specifically, the tunnel boring machine control center calculates cutting load parameters based on the real-time cutter strain, vibration, and temperature data obtained in step S2 above, using a multi-level data processing workflow. First, the raw signals are preprocessed, including noise reduction, drift correction, and outlier detection, using a combination of wavelet transform and adaptive filtering to improve signal quality. Second, a multi-source data fusion method is selected to fuse and calculate the cutter's cutting load parameters, including normal force, tangential force, lateral force, and resultant force, thereby calculating torque and power consumption. The multi-source data fusion method can be a first multi-source data fusion method based on Kalman filtering, a second multi-source data fusion method based on neural networks, or a joint multi-source data fusion method based on Kalman filtering and neural networks.

[0036] The first multi-source data fusion method based on Kalman filtering includes: First, establishing a strain-force conversion matrix for different tool types, obtained through a combination of finite element analysis and experimental calibration; considering factors such as tool geometry, mounting angle, and material properties, establishing dedicated conversion models for different tool types (e.g., hobs, scrapers, center cutters); Second, using the conversion matrix to convert strain data into force measurements; Then, designing a Kalman filtering framework, such as a simple or extended Kalman filtering algorithm, setting the states as forces (normal, tangential, lateral); observations as strain (converted to force through the conversion matrix as observed values) and vibration and temperature (used to adjust observation noise or state transition noise), constructing state transition equations and observation equations, dynamically adjusting the process noise covariance Q and observation noise covariance R, performing Kalman filtering prediction and updates, thereby achieving multi-source information fusion of strain data, vibration data, and temperature through the Kalman filter to obtain the output state values.

[0037] The second multi-source data fusion method based on neural networks includes: First, constructing a neural network model, inputting the strain, vibration and temperature of the real-time cutting tool, performing feature extraction (such as using a CNN-LSTM hybrid network), feature fusion, and outputting the cutting load parameters of the real-time (shield tunneling) cutting tool; through historical cutting load parameters; the model is generated by training using the strain, vibration and temperature of historical cutting tools and the corresponding historical cutting load parameters.

[0038] The joint multi-source data fusion method based on Kalman filtering and neural networks includes: obtaining the cutting load parameters of the tool by weighted fusion of the two methods to obtain the final cutting load parameters.

[0039] S4. Based on the acquired cutting load parameters, formation parameters are identified using a deep learning algorithm.

[0040] Specifically, a formation parameter identification model is constructed using deep learning algorithms. The model structure includes convolutional layers, pooling layers, fully connected layers, and long short-term memory network layers. The inputs to the formation parameter identification model are real-time cutting load data, historical load data, and known formation parameters (such as bottom layer parameters obtained from previous periods). The outputs are the current formation parameters, including formation type (such as single formation types like sand, clay, and gravel, or mixed formation types), strength parameters (such as cohesion and internal friction angle), and corrosivity parameters.

[0041] In the formation parameter identification model, the convolutional layer is responsible for extracting local features, and the long short-term memory network layer captures temporal dependencies. The convolutional layer is designed with a feature extraction module, which includes extracting statistical features (mean, variance, peak factor, etc.), frequency domain features (power spectral density, main frequency, etc.) and time-frequency features (wavelet coefficients, etc.) from the cutting load time series data.

[0042] In addition, a transfer learning strategy is adopted to initialize model parameters using existing engineering experience and continuously optimize them through online learning to achieve rapid adaptation to new geological conditions; an uncertainty quantification module is designed to evaluate the credibility of the identification results and trigger a manual intervention mechanism for areas with low credibility.

[0043] S5. By combining formation parameters and cutting load parameters, establish a wear state assessment model to evaluate the tool wear state.

[0044] Specifically, by combining formation parameters and cutting load parameters, key features are extracted, including: mean cutting force, tangential force fluctuation amplitude, vibration spectrum characteristics (frequency band energy ratio, vibration kurtosis), hard rock ratio, and interface switching frequency.

[0045] A wear state assessment model based on fuzzy inference is established. First, membership functions are defined for key features as input variables and wear states (light wear, moderate wear, severe wear) as output variables. The membership functions are defined using triangular or trapezoidal functions. The membership functions for some key features are designed as follows: mean normal force: fuzzy subset: low (L, 0-20kN), medium (M, 15-40kN), high (H, 30-60kN); vibration kurtosis: fuzzy subset: low (L, 0-3), medium (M, 2-5), high (H, 4-8); hard rock percentage: low (L, <30%), medium (M, 30%-70%), high (H, >70%). Secondly, a reasoning rule base is designed, including rule number, conditions, and conclusions. For example: Rule number: R1, conditions: under the combination of high mean normal force, high vibration kurtosis, and high hard rock ratio, conclusion: severe wear. Fuzzy reasoning is performed, including fuzzification (mapping the measured values ​​of input variables to the membership degrees of each fuzzy subset), rule matching and activation (using the smaller operator to calculate the activation strength of each rule), aggregation (using the larger operator to merge the output membership degrees of all rules), and defuzzification (using the centroid method), to derive the fuzzy evaluation of the tool wear degree and output the wear amount.

[0046] Alternatively, a deep learning-based wear condition assessment model can be used to construct a wear condition assessment neural network model. The inputs are formation parameters and cutting load parameters, and the output is the wear condition assessment result, including wear condition and wear amount.

[0047] In addition, a remaining service life prediction model based on support vector regression was constructed, taking into account historical wear rates, current formation parameters, and future tunneling paths, to predict the remaining service life of each tool. A tool wear mode recognition module was established to distinguish between normal wear, abnormal wear, and catastrophic wear, and to provide corresponding processing suggestions for different wear modes. The wear assessment results were presented through a visualization interface, using a cutterhead topology diagram and color coding to intuitively display the wear status of each tool, and providing wear trend analysis charts to assist in maintenance decisions.

[0048] S6. Based on the cutter wear state and formation parameters, design a multi-objective optimization function, solve the multi-objective optimization function, obtain the optimal shield machine operating parameters, and make dynamic adjustments.

[0049] The following data should be acquired: Status parameters: tool wear condition (W), formation type (G), current cutting load (normal force Fn, tangential force Ft, resultant force F_total), vibration data, temperature; Operating parameters: cutter head rotation speed (N), feed rate (v), cutter head opening ratio (…). Grouting parameters (grouting pressure (P), grouting volume (Q); performance indicators: tunneling efficiency (E = v / 1000×60), energy consumption (C = total power / tunneling distance), tool wear rate ( = Wear increment / tunneling time), cutterhead stability index (S = torque fluctuation rate + vibration kurtosis / preset threshold); Establish the operating parameters of the tunnel boring machine (N, v, ... P, Q) and performance indicators (E, C, Mathematical models for parameters such as cutting load, tunneling efficiency, and energy consumption are constructed using a combination of response surface methodology and neural networks, considering the interactions between parameters. For example, for each performance index, a second-order response surface model is constructed to characterize the mathematical relationship between the performance index and the operating parameters; a neural network structure is constructed, including an input layer (5 control parameters), a hidden layer (ReLU activation), and an output layer (4 performance indices, linear activation); the predicted values ​​of the response surface model are used as initial labels, 1000 sets of field measured data are fused, and the model is trained using the Adam optimizer.

[0050] Design a model predictive control algorithm and define the control objective functions, including maximizing the tunneling speed (maxE), minimizing energy consumption (minC), and minimizing tool wear rate (min). ) and maintain the stability of the cutter head (the function formula is: 1-S / When S≤ The function remains at its maximum value of 1 when S> When the function decreases linearly with the increase of S, multiple objectives are set, including: physical constraints of the equipment (N, v, ...). The constraints (P, Q) are set accordingly, and different parameter constraint ranges are set for different strata types. The safety constraints (normal force Fn, cutterhead torque T, stability index S) are set accordingly. The improved NSGA-II algorithm (algorithm initialization, fitness evaluation, selection and evolution and optimal solution screening) is used to solve the constrained multi-objective optimization problem and output the optimal solution set, that is, the optimal shield machine operation parameters.

[0051] Specifically, the system predicts the objective function value over a future period under simulated or given control operation parameter sets (which can be obtained based on expert experience) and current conditions (tool wear, formation parameters), solves for the optimal operation parameters, and performs real-time operation adjustments. These adjustments include cutterhead rotation speed, feed speed, cutterhead opening ratio, and grouting parameters. A hierarchical structure can be adopted, with the upper layer responsible for strategy optimization and the lower layer responsible for execution control, coordinating the two layers through a feedback mechanism. A parameter smoothing transition mechanism is designed during operation adjustments to avoid shocks and vibrations caused by sudden parameter changes. Simultaneously, a rule-based safety monitoring mechanism is implemented, automatically adjusting the operation strategy when approaching safety boundaries to ensure system safety. A user-friendly interface allows operators to set control preferences and manually intervene.

[0052] Furthermore, to ensure timely early warning, a multi-level early warning mechanism is established. Specifically, cutting load parameters, tool wear status, and formation parameters are monitored in real time. It is determined whether these parameters exceed corresponding preset thresholds or whether there are changes in preset parameters exceeding their respective thresholds. If any of these thresholds are exceeded, an early warning message is generated for that parameter. Further processing strategies are provided for different parameter warning messages, such as parameter adjustment plans, shutdown and inspection suggestions, and tool replacement plans. The preset parameter thresholds and preset parameter change thresholds can be multi-level, and the corresponding early warning messages can also be multi-level, such as three levels: attention level, warning level, and emergency level. Historical early warning data is continuously recorded and analyzed, and machine learning methods are used to continuously optimize the preset parameter thresholds and preset parameter change thresholds to improve the accuracy and timeliness of early warnings.

[0053] To continuously optimize the real-time monitoring and adjustment of the cutting load of the tunnel boring machine cutter, all data generated in the above steps can be integrated into a unified data storage and analysis platform. A distributed database architecture is designed, including a time-series database, a relational database, and a knowledge graph database, which respectively store the raw sensor data, structured analysis results, and knowledge rules. Then, a data mining module is developed, using methods such as association rule mining, time-series pattern mining, and cluster analysis to discover hidden patterns and knowledge from historical data, such as typical formation parameters and the wear state corresponding to typical formation parameters and cutting load parameters.

[0054] Furthermore, a digital twin model of the tunnel boring machine (TBM) cutterhead is constructed, mapping real-time acquired cutting load parameters, cutter wear status, and geological parameters into the digital twin model to achieve visualization and predictive analysis of the TBM's operating status. Simultaneously, an algorithm optimization module is developed, continuously adjusting the algorithm parameters of each model through genetic algorithms and Bayesian optimization methods. Alternatively, a model performance evaluation mechanism is designed to periodically assess the accuracy, stability, and adaptability of each model and generate optimization suggestions. Integration with other TBM systems is achieved through API interfaces, enabling data sharing and collaborative optimization. The platform supports multi-level permission management and remote access, facilitating use and maintenance by personnel with different roles.

[0055] In a specific embodiment, to obtain cutting load parameters more accurately and improve the accuracy and relevance of data fusion, the method further includes: obtaining cutting load parameters through multi-source data fusion based on real-time received tool strain, vibration, and temperature, which further includes: According to the preset data quality assessment rules, the data quality of the strain, vibration and temperature of the cutter received in real time is assessed; the cutter referred to below refers to the shield cutter; the preset data quality assessment rules include: the signal-to-noise ratio of the cutter strain, the dynamic fluctuation parameters of the cutter vibration signal, and the stable parameters of the cutter temperature; among them, the cutter strain signal-to-noise ratio is less than the first signal-to-noise ratio (10), which is identified as low signal-to-noise ratio; it is between the first signal-to-noise ratio and the second signal-to-noise ratio (10-20), which is identified as medium signal-to-noise ratio; and it is greater than the second signal-to-noise ratio (20), which is identified as high signal-to-noise ratio. The dynamic fluctuation parameters of the tool vibration signal include kurtosis K and peak factor CF. When K < first kurtosis value (3) and CF < first peak factor (3.5), it is considered low vibration fluctuation; when first kurtosis value (3) ≤ K ≤ second kurtosis value (4) or first peak factor (3.5) ≤ CF ≤ second peak factor (5), it is considered medium vibration fluctuation; when K > second kurtosis value (4) and CF > second peak factor (5), it is considered high vibration fluctuation. The temperature stability parameters include the temperature change rate (…). ) and temperature standard deviation ( ),when First rate of temperature change and temperature standard deviation The first temperature standard deviation is considered as low temperature stability; when the second temperature change rate... First temperature change rate or second temperature standard deviation The first temperature standard deviation is considered stable within the temperature range; when the second temperature change rate... And the first temperature standard deviation It was determined to be highly stable at high temperatures.

[0056] According to preset decision rules, multi-source data fusion methods are matched based on data quality assessment results, including: a first multi-source data fusion method based on Kalman filtering, a second multi-source data fusion method based on neural networks, and a joint multi-source data fusion method based on Kalman filtering (KF) and neural networks (NN). Specifically, the first multi-source data fusion method based on Kalman filtering is suitable for data fusion in scenarios where the data is relatively stable, has small fluctuations, and a clear linear relationship; the second multi-source data fusion method based on neural networks is suitable for data fusion in scenarios where the data noise is high, fluctuations are severe, and nonlinearity is prominent; the joint multi-source data fusion method is suitable for data fusion in scenarios with moderate noise, moderate fluctuations, and a mixture of linear and nonlinear relationships. The corresponding preset decision rules include: If the tool strain has a low signal-to-noise ratio, the second multi-source data fusion method is directly matched. Under low signal-to-noise ratio conditions, the linear Kalman filter (KF) is easily affected by noise, and the nonlinear fitting capability of the neural network (NN) can effectively suppress noise. If the tool strain has a medium signal-to-noise ratio, and the vibration is judged to be low-fluctuation, then the linear relationship dominates, and Kalman filtering is selected, matching the first multi-source data fusion method. Furthermore, under low temperature stability... If nonlinear interference is enhanced, replace the matching joint multi-source data fusion method; if the vibration is judged to be moderately volatile, then there is some nonlinearity, and the matching joint multi-source data fusion method is used again; if the vibration is judged to be highly volatile, then the nonlinearity is prominent, and the matching second multi-source data fusion method is used; if the tool strain has a high signal-to-noise ratio, and the vibration is judged to be low, then the signal quality is good and the dynamics are stable, and the linear model is sufficiently accurate, and the matching first multi-source data fusion method is used, and the matching joint multi-source data fusion method is replaced when the temperature is low and stable; if the vibration is judged to be moderately volatile, then there is slight nonlinearity, and nonlinearity correction needs to be added, and the matching joint multi-source data fusion method is used again; if the vibration is judged to be highly volatile, then the nonlinearity is significant, and the matching second multi-source data fusion method is used, and the linear relationship is more reliable when the temperature is high and stable, and the matching joint multi-source data fusion method is replaced.

[0057] In a specific embodiment, to further achieve accurate identification of geological parameters, provide a basis for subsequent adjustment of tunnel boring machine (TBM) operating parameters, and improve the efficiency and safety of TBM construction, the method further includes: the identification of geological parameters based on the acquired cutting load parameters and combined with a deep learning algorithm further includes: Based on the real-time acquired cutting load parameters, the torque and power consumption are calculated; Based on the formation parameter identification model constructed using deep learning algorithms, the model structure, including the parameters of the convolutional layer, is further modified to extract cutting load data, torque, power consumption, and feature data of known formation parameters, thereby generating an improved formation parameter identification model. The input of the improved formation parameter identification model is modified to real-time cutting load data, historical load data, real-time calculated torque and power consumption, and known formation parameters, and the output is the current formation parameters, including formation type, strength parameters, and corrosion parameters.

[0058] In addition, to further verify the accuracy of the identification results based on the improved formation parameter identification model, based on the known formation parameters and the spatial distribution knowledge of formation parameters, the interpolation method is used to predict the formation change area in front of the cutterhead, and the similarity between the predicted formation parameters and the identified formation parameters is judged. If the similarity is lower than the preset similarity, it is determined that there may be an error in the identification, and the identification is re-performed or the model parameters are adjusted and re-identified.

[0059] In a specific embodiment, to improve the accuracy and reliability of tool wear assessment and provide a more precise basis for tool maintenance and replacement, the method further includes: establishing a wear state assessment model based on fuzzy inference by combining formation parameters and cutting load parameters; the assessment of tool wear state further includes: Based on the extraction of key features, the definition of key features as input variables, and the membership function of wear state as output variables, and combined with rule settings under different formation types, nonlinear correction is introduced for mixed formation types to highlight the special wear mechanism of mixed formations, thereby further improving the accuracy of tool wear state assessment. Specifically, this includes designing a reasoning rule base, including rules for non-mixed formations and rules for mixed formations; where hard rock accounts for ≤30%. If the hard rock percentage is ≥80%, corresponding to the non-mixed strata rule, wear is mainly due to a single mechanism. Some rules are defined as follows: R1: Conditions: If the mean normal force is low, the tangential force fluctuation amplitude is low, the proportion of hard rock characteristic frequency band energy is low, and the hard rock percentage is low, conclusion: slight wear; R2: Conditions: If the mean normal force is high, the tangential force fluctuation amplitude is high, the proportion of hard rock characteristic frequency band energy is high, and the hard rock percentage is high, conclusion: severe wear. When the hard rock percentage is 30% < hard rock percentage < 80%, corresponding to the mixed strata rule, collaborative wear will occur. Some rules are defined as follows: R3: If the mean normal force is medium, the tangential force fluctuation amplitude is medium, the vibration kurtosis is medium, and the hard rock percentage is medium, conclusion: moderate wear; R4: If the mean normal force is medium, the tangential force fluctuation amplitude is high, the vibration kurtosis is high, and the hard rock percentage is high, conclusion: severe wear.

[0060] Furthermore, based on the identified current stratigraphic parameters, the system adaptively matches non-mixed stratigraphic rules or mixed stratigraphic rules in the reasoning rule base, performs fuzzy reasoning and defuzzification, and obtains the wear amount. If the current stratigraphic parameters are identified as mixed stratigraphic type, a nonlinear correction mechanism for mixed stratigraphic parameters is introduced, with the formula: K = 1 + × Percentage of hard rock; The correction factor typically ranges from 0.5 to 1.5, depending on the hardness of the rock. For example, if the hardness is less than 60 MPa, a is 0.8; otherwise, it is 1.2. The formation parameter correction factor is calculated, and the obtained wear amount is multiplied by the formation parameter correction factor to obtain the final wear amount.

[0061] In a specific embodiment, to further improve tunneling efficiency, reduce energy consumption, decrease cutter wear, and ensure the operational stability of the tunnel boring machine, the method further includes: designing a multi-objective optimization function based on the cutter wear state and geological parameters, solving the multi-objective optimization function, obtaining the optimal tunnel boring machine operating parameters, and dynamically adjusting them; Based on the mapping relationship between the operating parameters and performance indicators of the tunnel boring machine obtained in the above embodiments and the design of multi-objective functions, the constraints of different strata types are further considered and the constraints are designed, including: strata type adaptation constraints: the range of cutterhead rotation speed, the range of propulsion speed and the range of opening ratio adjustment for different strata types.

[0062] The weights of the multi-objective function are dynamically adjusted based on real-time identified cutter wear status and geological type to achieve multi-objective optimization and obtain the optimal tunnel boring machine (TBM) operating parameters. Each combination of cutter wear status and geological type has a preset weight allocation ratio. For example, if the geological type is mixed and the cutter wear status is severe, the priority is to reduce wear and maintain stability at the expense of efficiency, with corresponding weight allocation ratios of 0.1, 0.1, 0.5, and 0.3 for w1, w2, w3, and w4, respectively. If the geological type is mixed and the cutter wear status is moderate, the priority is to reduce wear and maintain stability, with efficiency as a secondary consideration, with corresponding weight allocation ratios of 0.2, 0.1, 0.3, and 0.4 for w1, w2, w3, and w4, respectively. If the geological type is mixed and the cutter wear status is slight, the priority is to maintain a balance between efficiency and stability, with corresponding weight allocation ratios of 0.3, 0.2, and 0.2 for w1, w2, w3, and w4, respectively. For soft soil layers such as sand or clay, where the tool wear is slight, efficiency is prioritized while energy consumption is considered. The corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.4, 0.3, 0.1, and 0.2, respectively. For soft soil layers such as sand or clay, where the tool wear is moderate, efficiency and wear are balanced. The corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.3, 0.2, 0.3, and 0.2, respectively. For soft soil layers such as sand or clay, where the tool wear is severe, wear is limited, and efficiency is reduced. The corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.2, 0.2, and 0.4, respectively. For hard rock formations like pebble layers, with slight tool wear, stability is prioritized over efficiency, and the corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.3, 0.1, 0.2, and 0.4, respectively. For hard rock formations like pebble layers, with moderate tool wear, both stability and wear are prioritized, and the corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.2, 0.1, 0.3, and 0.4, respectively. For hard rock formations like pebble layers, with severe tool wear, forced speed reduction is implemented to maintain stability and reduce wear, and the corresponding weight allocation ratios for w1, w2, w3, and w4 are 0.1, 0.1, 0.4, and 0.4, respectively.

[0063] like Figure 2 As shown in the figure, this application discloses a real-time monitoring and adjustment system for the cutting load of a tunnel boring machine cutter, comprising: The tool data acquisition module 101 is used to construct a distributed sensing network for the tunnel boring machine tool and collect the strain, vibration and temperature of the tool. The tool data transmission module 102 is used to establish a wireless transmission network to transmit the collected tool strain, vibration and temperature in real time. The tool data fusion module 103 is used to obtain cutting load parameters by multi-source data fusion based on the real-time received tool strain, vibration and temperature. The formation parameter identification module 104 is used to identify formation parameters based on the acquired cutting load parameters and combined with a deep learning algorithm. The tool wear assessment module 105 is used to combine formation parameters and cutting load parameters to establish a wear state assessment model based on fuzzy reasoning and assess the tool wear state. The tool parameter adjustment module 106 is used to design a multi-objective optimization function based on the tool wear state and formation parameters, solve the multi-objective optimization function, obtain the optimal shield machine operation parameters, and make dynamic adjustments.

[0064] In one specific embodiment, the system further includes: The tool operation early warning module 107 is used to monitor the acquired cutting load parameters, tool wear status and formation parameters in real time, and to determine whether the cutting load parameters, tool wear status and formation parameters exceed the corresponding preset parameter thresholds or whether there are preset parameter changes that exceed the corresponding preset parameter change thresholds. Once the corresponding preset parameter thresholds or preset parameter change thresholds are exceeded, an early warning prompt for the corresponding parameter is generated.

[0065] In one specific embodiment, the system further includes: The tool operation simulation module 108 is used to build a digital twin model of the shield machine tool application. It maps the real-time acquired cutting load parameters, tool wear status and formation parameters into the digital twin model to realize the visualization and predictive analysis of the shield machine's operating status.

[0066] This application also discloses a computer-readable storage medium.

[0067] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executed, such as the method for real-time monitoring and adjustment of the cutting load of the tunnel boring machine cutter described above. The computer-readable storage medium includes, for example, various media that can store program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0068] This application also discloses a computer device.

[0069] Specifically, the computer device includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed to perform the aforementioned method for real-time monitoring and adjustment of the cutting load of the tunnel boring machine cutter.

[0070] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for real-time monitoring and adjusting of cutting load of a shield cutter, characterized in that, The application relates to a shield cutter distributed sensing network, a wireless transmission network, a cutting load parameter acquisition method, a stratum parameter identification method, a cutter wear state evaluation model and a shield machine operation parameter optimization method. The shield cutter distributed sensing network comprises the following steps: A plurality of types of sensors are arranged at cutter installation positions of a cutter disc of a shield machine, including strain sensors, acceleration sensors, vibration sensors and temperature sensors; wherein a non-uniform distribution strategy is adopted when the sensors are arranged at the cutter disc of the shield machine, different density sensor arrays are arranged in a central region, a middle region and an edge region of the cutter disc, and at least one group of the plurality of types of sensors is arranged on each cutter. Real-time data transmission through the wireless transmission network comprises the following steps: Multi-channel synchronous data acquisition is performed; a dual-frequency wireless transmission mode of a main frequency band and a backup frequency band is adopted, and automatic power adjustment and frequency band switching operations are performed during the wireless transmission process; the automatic power adjustment refers to dynamically adjusting the transmission power according to the signal strength of the collected data; and the frequency band switching operation refers to dynamically switching the frequency band according to the signal quality of the collected data. The cutting load parameter acquisition method comprises the following steps: Data quality evaluation is performed on the real-time received strain, vibration and temperature of the cutter according to a preset data quality evaluation rule; the preset data quality evaluation rule comprises a signal-to-noise ratio of the strain of the cutter, a dynamic fluctuation parameter of the vibration signal of the cutter and a stability parameter of the temperature of the cutter.

2. The method of claim 1, wherein, ​ ​ 3. The method of claim 1, wherein the method further comprises: ​ ​ 4. The method of claim 1, wherein, ​ ​ According to the preset decision rule, the multi-source data fusion mode is matched based on the data quality evaluation result, including: a first multi-source data fusion mode based on Kalman filtering, a second multi-source data fusion mode based on a neural network, and a joint multi-source data fusion mode based on Kalman filtering and a neural network; the preset decision rule includes: if the strain of the tool is low signal-to-noise ratio, directly match the second multi-source data fusion mode; if the strain of the tool is medium signal-to-noise ratio, continue to judge the vibration low fluctuation, match the first multi-source data fusion mode, and replace the joint multi-source data fusion mode when the temperature is low and stable; continue to judge the vibration medium fluctuation, match the joint multi-source data fusion mode; continue to judge the vibration high fluctuation, match the second multi-source data fusion mode; if the strain of the tool is high signal-to-noise ratio, continue to judge the vibration low fluctuation, match the first multi-source data fusion mode, and replace the joint multi-source data fusion mode when the temperature is low and stable; when continue to judge the vibration medium fluctuation, match the joint multi-source data fusion mode; continue to judge the vibration high fluctuation, match the second multi-source data fusion mode, and replace the joint multi-source data fusion mode when the temperature is high and stable.

5. The method of claim 1, wherein, The formation parameter identification based on the obtained cutting load parameter includes: Based on the real-time obtained cutting load parameter, the torque and power consumption are calculated; An improved formation parameter identification model is constructed by using a deep learning algorithm, and the model structure includes a convolution layer, a pooling layer, a full connection layer and a long short-term memory network layer; the input of the improved formation parameter identification model is real-time cutting load data, historical load data, real-time calculated torque and power consumption and known formation parameters, and the output is the current formation parameter, including formation type, strength parameter and corrosion parameter.

6. The method of claim 1, wherein, The wear state evaluation model based on fuzzy inference is established by combining the formation parameter and the cutting load parameter, and the tool wear state is evaluated, including: Key features are extracted by combining the formation parameter and the cutting load parameter, including: cutting force average, vibration frequency spectrum feature; A wear state evaluation model based on fuzzy inference is established, and the membership functions of the key features as input variables and the wear state as output variables are defined; A reasoning rule base is designed, including non-mixed formation rules and mixed formation rules; Based on the identified current formation parameter, the non-mixed formation rule or the mixed formation rule in the reasoning rule base is adaptively matched, fuzzy inference and defuzzification are performed, and the wear amount is obtained; When the identified current formation parameter is a mixed formation type, a mixed formation parameter nonlinear correction mechanism is introduced, a formation parameter correction factor is calculated, the obtained wear amount is multiplied by the formation parameter correction factor, and the final wear amount is obtained.

7. The method of claim 1, wherein, According to the tool wear state and the formation parameter, a multi-objective optimization function is designed, the multi-objective optimization function is solved, the optimal shield machine operation parameter is obtained and dynamically adjusted, including: A mathematical model of the operation parameters and performance indicators of the shield machine is constructed by combining response surface method and neural network to obtain the mapping relationship between the operation parameters and performance indicators of the shield machine; the operation parameters include cutterhead speed, pushing speed, cutterhead opening rate, and grouting parameters; the performance indicators include tunneling efficiency and energy consumption; A multi-objective function is designed, including maximizing tunneling efficiency, minimizing energy consumption, minimizing wear rate, and maintaining stability; constraint conditions are designed, including physical constraints of equipment, including cutterhead speed range, pushing speed range, and opening rate adjustment range; safety constraints, including cutter normal force range and cutterhead torque range; and stratum type adaptation constraints, including cutterhead speed range, pushing speed range, and opening rate adjustment range for different stratum types; The weights of the multi-objective function are dynamically adjusted according to the real-time identified cutter wear state and stratum type to complete multi-objective optimization solving and obtain optimal shield machine operation parameters.

8. The method of claim 1, wherein, It also includes: The cutting load parameters, cutter wear state, and stratum parameters obtained by real-time monitoring are judged to determine whether they exceed the corresponding preset parameter threshold or whether there is a preset parameter change exceeding the corresponding preset parameter change threshold; once the corresponding parameter threshold or the corresponding parameter change threshold is exceeded, a corresponding parameter warning prompt is generated.

9. The method of claim 1, wherein, It also includes: A digital twin model of the shield cutter is constructed, and the real-time acquired cutting load parameters, cutter wear state, and stratum parameters are mapped into the digital twin model to realize visualization and predictive analysis of the shield machine operating state.

10. A real-time monitoring and adjusting system for cutting load of a shield cutter, characterized in that, It includes: A cutter data acquisition module is used to construct a distributed sensor network for the shield cutter to acquire strain, vibration, and temperature of the cutter; A cutter data transmission module is used to establish a wireless transmission network to transmit the acquired strain, vibration, and temperature of the cutter in real time; A cutter data fusion module is used to acquire cutting load parameters by multi-source data fusion based on the real-time received strain, vibration, and temperature of the cutter; A stratum parameter identification module is used to identify stratum parameters based on the acquired cutting load parameters and deep learning algorithm; A cutter wear evaluation module is used to establish a wear state evaluation model based on fuzzy reasoning by combining stratum parameters and cutting load parameters to evaluate the cutter wear state; A cutter parameter adjustment module is used to design a multi-objective optimization function based on the cutter wear state and stratum parameters, solve the multi-objective optimization function, obtain optimal shield machine operation parameters, and dynamically adjust them.

Images