Control method for adjusting tunneling posture of shield tunneling machine

Abstract

The invention discloses a control method for adjusting the tunneling attitude of a shield tunneling machine, particularly relates to the field of attitude control of the shield tunneling machine, and aims to solve the problem that attitude deviation caused by underground environment change cannot be timely and effectively corrected in the tunneling process of the shield tunneling machine. The method comprises the following steps of: calculating membership degrees of horizontal deviation, vertical deviation and change rates of the horizontal deviation and the vertical deviation, extracting high-frequency and low-frequency characteristics, generating independent phase signals, and calculating a judgment coefficient by using a multi-dimensional data fusion and high-dimensional exponential integration method so as to dynamically adjust the weight of a fuzzy control rule base and realize adaptive optimization of PID (Proportion Integration Differentiation) gain parameters; a master control output signal is generated based on the updated PID parameters, the thrust of the partition oil cylinders is accurately adjusted, it is ensured that the posture of the shield tunneling machine is rapidly, accurately and stably adjusted under the complex environment condition, the control precision of the tunneling direction and the construction safety are improved, and the working efficiency is improved. And meanwhile, the overall posture correction capability of the shield tunneling machine is further enhanced through dynamic adjustment of the rolling angle.

Description

Technical Field

[0001] The present invention relates to the field of shield machine posture control, and more specifically, to a control method for adjusting the tunneling posture of a shield machine. Background Art

[0002] During the underground tunnel excavation process of the shield machine, it is crucial to accurately control the excavation direction of the shield machine along the designed axis. Due to the complexity of the underground rock formations and the existence of obstacles such as underground pipelines, the shield machine often faces the challenge of adjusting its posture in real time during the excavation process. At present, the shield machine uses a total station installed on the fuselage and a prism at the center of the shield head and other positioning and guidance systems to measure and monitor key posture parameters such as shield head deviation, shield tail deviation, excavation trend, roll angle and pitch angle to ensure accuracy and safety during the construction process.

[0003] However, the traditional posture adjustment method is mainly based on PID control, which achieves precise control within a certain range by adjusting the proportional, integral and differential gain parameters. However, under the nonlinear, multivariable and complex environmental conditions of the shield machine, the response speed and control effect of the traditional PID control are often difficult to meet the needs, especially in scenarios where geological conditions change frequently or the frequency of tunneling trajectory adjustment is high. It is difficult for the system to dynamically adapt to environmental changes, resulting in delayed posture adjustment or insufficient control accuracy. In this case, how to achieve dynamic adjustment and real-time optimization of parameters has become a key issue that needs to be urgently solved in posture control technology.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a control method for adjusting the tunneling posture of a shield machine. Through an adaptive control method based on fuzzy PID, combined with the multi-dimensional posture data obtained by the laser positioning and guidance system and the total station, the precise control of the posture of the shield machine during the tunneling process is achieved, and the adaptability to complex nonlinear and multi-variable environments is significantly improved. Through membership calculation, frequency domain feature extraction, phase signal processing and multi-dimensional data fusion, a judgment coefficient for dynamically adjusting the PID gain is generated, and the weight distribution of the fuzzy control rule base is further optimized to ensure that the proportional, integral and differential gain parameters are updated in real time according to the actual tunneling state. At the same time, the fuzzy control output is converted into an accurate gain adjustment amount through defuzzification processing, and then the thrust distribution of the propulsion cylinder and the cutter head reversal adjustment process are dynamically optimized to achieve high-precision correction of the roll angle, pitch angle and tunneling deviation of the shield machine, meet the requirements of high precision and high reliability of tunneling, and solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A control method for adjusting the tunneling posture of a shield machine, comprising the steps of: S1, using the laser positioning guidance system and total station to measure the various operating data of the shield machine in real time and convert the data into the original numerical set; S2, divide the horizontal deviation and vertical deviation and their change rate into seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and calculate the membership degree; S3, performs a fifth-order fast Fourier transform on the horizontal error, extracts high-frequency and low-frequency features, processes the vertical error to generate a phase signal, fuses the two-directional signals and calculates the judgment coefficient; S4, adjusting the weight of the fuzzy control rule base based on the judgment coefficient to generate a PID gain adjustment amount; S5, adding the defuzzified gain adjustment amount to the current PID gain to dynamically update the PID parameters; S6, using the updated PID parameters to calculate the total control output, and transmit it to the partition cylinder actuator to adjust the posture; S7, adjust the cylinder thrust according to the total control output.

[0007] In a preferred embodiment, step S1 includes the following contents: A laser scanner is used to monitor the positions of the shield head and tail of the shield machine in real time. The omnidirectional measurement capability of the total station is used to obtain the tunneling trend angle, roll angle and pitch angle of the shield machine in three-dimensional space. The three-dimensional spatial posture data is converted into a complex domain representation. Wavelet transform is used to perform multi-resolution analysis on the complex domain data to extract posture change features at different frequency levels. The extracted high-dimensional feature data is standardized. The standardized data is mapped to the [0,1] interval. The processed standardized and normalized data is organized into a multi-dimensional original numerical set.

[0008] In a preferred embodiment, step S2 includes the following contents: Horizontal deviation and vertical deviation Defined as the angular deviation between the current posture of the shield machine and the design axis; horizontal deviation change rate and vertical deviation rate of change Defined as the change in horizontal deviation and vertical deviation per unit time; Set the range of horizontal deviation, vertical deviation, horizontal deviation change rate and vertical deviation change rate to [-90°, +90°]; Divide the interval into seven subsets: negative large , Negative , negative small ,zero , Zhengxiao ,middle , Zhengda ; A triangular membership function is designed for each subset, with its peak at the center of the subset and the basis covering the boundaries of adjacent subsets; the center point of each subset is defined as and the boundary points are ; For any input value , which is in the subset The membership degree in Determined by the following triangular membership functions: ; Specific formula description: When the input value is in the left half of the subset When , the membership increases linearly from 0 to 1; When the input value is in the right half of the subset When , the membership decreases linearly from 1 to 0; When the input value is not within the interval of the subset, the membership is 0; For each input value, a vectorized calculation method is used to calculate its membership in the seven subsets: ; Use nonlinear weighting factors The membership degree of each subset is adjusted as follows: ; Among them, the weighting factor Determined by the following composite function: ;parameter , , is a preset constant; To ensure that the sum of all subset memberships is 1, the adjusted memberships are normalized, and then organized into a membership matrix, where each row corresponds to an input value and each column corresponds to a subset.

[0009] In a preferred embodiment, step S3 includes the following contents: S31. Obtain the membership values ​​of the horizontal deviation and vertical deviation and their change rates in each membership subset; combine the corresponding membership information with the original error value, and perform a weighted transformation on each error value: , ;in and is the final membership value obtained after normalizing the membership of the horizontal deviation and the vertical deviation in step S2, and Indicates the index of the belonging subset; S32. Perform a fifth-order fast Fourier transform decomposition on the weighted horizontal error, map it from the time domain to the frequency domain, and recursively perform a five-layer decomposition; set the first layer fast Fourier transform output to ,right After frequency band separation and nonlinear phase correction, a new time domain signal is obtained, and fast Fourier transform is performed again until the fifth-order decomposition is completed; S33. Implementing dual nonlinear phase modulation on the weighted vertical error; first mapping the weighted vertical error to the complex domain: ;in is a nonlinear phase function dynamically calculated based on actual geological parameters. , , , is a preset constant; the same type of nonlinear phase modulation is applied to the phase modulated signal to obtain a more complex complex phase signal ; S34. Perform multi-dimensional data interleaving and fusion on the complex plane of the horizontal error composite spectrum extracted by the fifth-order fast Fourier transform and the complex phase signal obtained by the double nonlinear phase modulation of the vertical error; the process uses multi-dimensional complex space interpolation to decompose the horizontal error composite spectrum and the complex phase signal into several sub-bands and sub-phase units, and cross-arrange them on the complex plane; Perform high-dimensional exponential integral calculation on the fusion matrix to obtain the judgment coefficient for judgment .

[0010] In a preferred embodiment, step S4 includes the following contents: Assume that the rule base contains rules, each rule is as follows: ;in is the error, is the error change rate, and It is a subset of the seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large. , , They are the fuzzy output subsets of proportional, integral and differential gain adjustment quantities; Define each rule The weight factor as follows: ; In the formula is a preset constant used to reflect the rules Base sensitivity to gain adjustments; It is a nonlinear index used to enhance the weighting effect at high complexity; Based on the known error and error change rate membership, each rule is matched with its antecedents, and the minimum operation is used to achieve the fuzzy synthesis of the rule antecedents: ;in and are the membership degrees of error and error change rate under the corresponding subsets respectively; Then, As the cut-set height of the subsequent output fuzzy set, the fuzzy subsets of the proportional, integral, and differential gain adjustment are cut, and the weight factor is combined Vertically enlarge the membership function of the corresponding cut-set posterior: ;in is the original membership function of the corresponding rule output subset; Perform maximum aggregation on the weighted output fuzzy sets of all rules, and synthesize the corresponding gain adjustment fuzzy outputs into three final fuzzy sets; The maximum membership method is used to defuzzify the final fuzzy set; the point with the maximum membership is found in the domain of the gain adjustment amount, and the value of this point is output as the final PID gain adjustment amount.

[0011] In a preferred embodiment, step S5 includes the following contents: Use nonlinear mapping functions to perform complex nonlinear amplification or compression on the gain adjustment amount; The following nonlinear mapping methods are defined: (1) Update of proportional gain: ; (2) Update of integral gain: ; (3) Update of differential gain: ; in , , They are the proportional, integral, and differential gain adjustment values ​​respectively; Through nonlinear mapping, the proportional, integral, and differential gain adjustments are integrated with the judgment coefficients and their respective function mappings to finally obtain new PID parameters. , , .

[0012] In a preferred embodiment, step S6 includes the following contents: According to the PID control logic, three correction quantities are calculated in sequence: The proportional correction is achieved by multiplying the current error value by the updated proportional gain, reflecting the direct impact of the current deviation; The integral correction is based on the historical error accumulation and combined with the nonlinear kernel function to amplify the impact of the continuous deviation. The differential correction depends on the error change rate and is combined with a smoothing function to limit the risk of over-adjustment caused by sudden changes. After calculating the basic control output, the attitude compensation parameters are added and generated through the five-dimensional exponential convolution formula to dynamically balance the asymmetric distribution of the thrust between the cylinder groups; finally, the total control output is obtained.

[0013] In a preferred embodiment, step S7 includes the following contents: The thrust of the four groups of cylinders of the shield machine is adjusted by using the obtained total control output signal. The thrust adjustment amount of each group of cylinders is allocated by calculating the posture adjustment information in the total control output signal, and it is ensured that the extension amount of each group of cylinders is always within the preset safety threshold range.

[0014] The technical effects and advantages of a control method for adjusting the tunneling posture of a shield machine according to the present invention are as follows: The present invention realizes precise control of the posture of the shield machine during the excavation process by combining the laser positioning guidance system and the multi-dimensional posture data obtained by the total station through an adaptive control method based on fuzzy PID, and significantly improves the adaptability to complex nonlinear and multi-variable environments. Through membership calculation, frequency domain feature extraction, phase signal processing and multi-dimensional data fusion, a judgment coefficient for dynamically adjusting the PID gain is generated, and the weight distribution of the fuzzy control rule base is further optimized to ensure that the proportional, integral and differential gain parameters are updated in real time according to the actual excavation state. At the same time, the fuzzy control output is converted into an accurate gain adjustment amount through defuzzification processing, and then the thrust distribution of the propulsion cylinder and the cutterhead reversal adjustment process are dynamically optimized, so as to realize high-precision correction of the roll angle, pitch angle and excavation deviation of the shield machine, and meet the requirements of high precision and high reliability of tunnel excavation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flow chart of a control method for adjusting the tunneling posture of a shield machine. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Embodiment 1: Figure 1 The present invention provides a control method for adjusting the tunneling posture of a shield machine, comprising: S1, using the laser positioning guidance system and total station to measure the various operating data of the shield machine in real time and convert the data into the original numerical set; S2, divide the horizontal deviation and vertical deviation and their change rate into seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and calculate the membership degree; S3, performs a fifth-order fast Fourier transform on the horizontal error, extracts high-frequency and low-frequency features, processes the vertical error to generate a phase signal, fuses the two-directional signals and calculates the judgment coefficient; S4, adjusting the weight of the fuzzy control rule base based on the judgment coefficient to generate a PID gain adjustment amount; S5, adding the defuzzified gain adjustment amount to the current PID gain to dynamically update the PID parameters; S6, using the updated PID parameters to calculate the total control output, and transmit it to the partition cylinder actuator to adjust the posture; S7, adjust the cylinder thrust according to the total control output.

[0018] During the tunneling process of the shield machine, real-time and accurate measurement and acquisition of the shield machine's multi-dimensional posture data is the basis for achieving automatic posture adjustment. The data obtained by the high-precision sensor system not only reflects the current operating status of the shield machine, but also provides necessary information support for subsequent posture adjustments. The core of step S1 is to use the laser positioning guidance system and total station technology to collect the key posture parameters of the shield machine in real time and convert them into original numerical sets suitable for subsequent processing. This process involves complex data acquisition and conversion technology to ensure the accuracy and real-time nature of the data, laying a solid foundation for the efficient operation of the overall control system.

[0019] Step S1 includes the following contents: 1. Multi-sensor data acquisition: Laser positioning and guidance system: A high-precision laser scanner is used to monitor the position of the shield head and tail of the shield machine in real time. The laser positioning and guidance system transmits a laser beam, receives the reflected signal, and calculates the deviation of the shield head and tail relative to the predetermined design axis.

[0020] Total station: The total station's all-round measurement capability is used to obtain the tunneling trend angle, roll angle, and pitch angle of the shield machine in three-dimensional space. The total station updates the shield machine's spatial position and attitude changes in real time through precise angle and distance measurements.

[0021] 2. Data synchronization and timestamp: To ensure data synchronization, all sensor data are transmitted to the central processing unit via a high-frequency data bus (such as EtherCAT or CAN bus).

[0022] Each set of data comes with an accurate timestamp, so that different sensor data can be time-aligned during subsequent data processing to eliminate errors caused by transmission delays.

[0023] 3. High-dimensional data fusion and preprocessing: Data fusion algorithm: Adaptive Kalman filter is used to fuse the multi-source data from the laser positioning guidance system and the total station. The Kalman filter filters out sensor noise and improves the signal-to-noise ratio of the data through prediction and update steps.

[0024] Outlier detection and correction: An outlier detection method based on principal component analysis is used to identify and eliminate possible sensor abnormal data to ensure data reliability.

[0025] 4. Complex number field conversion and feature extraction: Complex number domain conversion: Convert the three-dimensional spatial posture data into complex number domain representation for subsequent frequency domain analysis and phase processing. Mainly, the shield head deviation and shield tail deviation are represented as complex numbers.

[0026] Feature extraction: Wavelet transform is applied to perform multi-resolution analysis on complex domain data to extract posture change features at different frequency levels. These features include low-frequency trend components and high-frequency fluctuation components, which correspond to the large-scale posture adjustment and subtle posture correction requirements of the shield machine.

[0027] 5. Data standardization and normalization processing: In order to meet the input requirements of the subsequent fuzzy control algorithm, the extracted high-dimensional feature data is standardized. The Z-score standardization method is used to convert the data of each dimension into a standard normal distribution with a mean of zero and a standard deviation of one, eliminating the influence of different dimensions and numerical ranges on the control system.

[0028] Min-Max normalization is further applied to map the standardized data to the interval [0,1] to ensure that the data is within a uniform scale, which is convenient for the membership calculation and rule matching of the fuzzy controller.

[0029] 6. Generation and output of original value sets: The processed standardized and normalized data are organized into a multi-dimensional original numerical set, including key parameters such as shield head deviation, shield tail deviation, tunneling trend angle, roll angle, pitch angle, etc. These parameters are stored in the form of vectors as input for subsequent fuzzy control algorithms.

[0030] The generated raw value set is transmitted to the central control unit via a high-speed data bus interface (such as PCIe or Ethernet interface) to ensure the real-time and integrity of the data.

[0031] Step S1 generates a high-quality original value set of multi-dimensional posture data through high-precision real-time measurement of the laser positioning guidance system and the total station, combined with data fusion and processing technology. Through complex number domain conversion, feature extraction, standardization and normalization processing, the accuracy and consistency of the data are ensured, providing a reliable data foundation for the subsequent fuzzy PID control algorithm. This process not only improves the accuracy and efficiency of data acquisition, but also provides solid technical support for the automatic posture adjustment of the shield machine in complex geological environments.

[0032] During the tunneling process of the shield machine, real-time monitoring and accurate evaluation of its posture deviation are crucial to ensure construction quality and safety. Horizontal deviation, vertical deviation and their rate of change are important indicators for measuring the posture of the shield machine. In order to achieve precise posture control, these deviation data must be systematically divided and refined. The core of step S2 is to subdivide the horizontal deviation, vertical deviation and their rate of change into seven subsets in the range of -90 degrees to 90 degrees, and through a complex membership calculation method, ensure that the membership calculation of each input value in each subset is accurate and innovative. This process provides a high-quality input basis for subsequent fuzzy control and enhances the response sensitivity and adjustment accuracy of the control system.

[0033] Step S2 includes the following contents: 1. Definition and preparation of deviation and its rate of change: Horizontal deviation and vertical deviation Defined as the angular deviation between the current posture of the shield machine and the design axis, in degrees.

[0034] Horizontal deviation change rate and vertical deviation rate of change It is defined as the change in horizontal deviation and vertical deviation per unit time, in degrees per second.

[0035] 2. Interval division: The value ranges of horizontal deviation, vertical deviation, horizontal deviation change rate and vertical deviation change rate are set to [-90°, +90°].

[0036] Divide the interval into seven subsets: negative large , Negative , negative small ,zero , Zhengxiao ,middle , Zhengda .

[0037] 3. Triangular membership function design: A unique triangular membership function is designed for each subset, with its peak located at the subset center point and the basis covering the boundaries of adjacent subsets.

[0038] Define the center point of each subset and boundary points as follows:

[0039] 4. Membership calculation formula: For any input value (represents one of the horizontal deviation, vertical deviation, horizontal deviation change rate, and vertical deviation change rate), which is in the subset The membership degree in Determined by the following triangular membership functions: ; Specific formula description: When the input value is in the left half of the subset , the membership increases linearly from 0 to 1.

[0040] When the input value is in the right half of the subset , the membership decreases linearly from 1 to 0.

[0041] When the input value is not within the interval of the subset, the membership is 0.

[0042] 5. Membership calculation formula: For each input value, calculate its membership in the seven subsets: ; Use vectorized calculation method to ensure efficient processing of membership calculation of multiple input values.

[0043] 6. Membership adjustment mechanism: To improve the accuracy of membership calculation, a nonlinear weighting factor is used The membership degree of each subset is adjusted as follows: ; Among them, the weighting factor Determined by the following composite function: ;parameter , , It is a preset constant used to adjust the nonlinear change of the membership degree to ensure flexible adjustment of the membership degree under extreme deviation conditions.

[0044] 7. Membership adjustment mechanism: To ensure that the sum of all subset memberships is 1, the adjusted memberships are normalized: ; This processing ensures that the relative proportional relationship of the membership degree remains unchanged, while meeting the requirements of the fuzzy control algorithm for input data.

[0045] 8. Output membership matrix: The normalized membership Organized into a membership matrix where each row corresponds to an input value and each column corresponds to a subset.

[0046] The membership matrix serves as the basic input for subsequent fuzzy reasoning and control rule matching, ensuring that the control system can make decisions based on accurate membership information.

[0047] Step S2 divides the horizontal deviation and vertical deviation and their change rate into seven precise subsets in the range of -90 degrees to 90 degrees, and uses a complex triangular membership function combined with a nonlinear weighting mechanism to calculate the membership of each input value in each subset. This process not only improves the accuracy and adaptability of membership calculation, but also ensures that the control system can make accurate attitude adjustment decisions based on high-quality input data through complex mathematical formulas and multi-dimensional data processing methods. Through this innovative processing logic, the system shows higher sensitivity and response speed in complex geological environments, significantly improving the level of automation and intelligent control during the tunneling process of the shield machine.

[0048] In the automatic attitude control system of the shield machine, accurately judging whether it is necessary to adjust the weights of the proportional gain, integral gain and differential gain is the key link to achieve precise control. Through the processing of the first two steps, the membership information and corresponding error values ​​of the multi-dimensional posture data have been obtained. On this basis, step S3 interweaves and fuses the two-directional information in the multi-dimensional space by performing high-order frequency domain analysis on the horizontal error and implementing nonlinear phase modulation on the vertical error, and then performs high-dimensional exponential integration operation on the fused signal, and finally obtains the judgment coefficient that can quantify the complexity of the attitude adjustment. This judgment coefficient provides a direct and accurate basis for subsequent PID parameter optimization and dynamic weight adjustment.

[0049] Step S3 includes the following contents: S31. Input data and preprocessing: After completing step S2, the membership values ​​of the obtained horizontal deviation and vertical deviation and their change rates in each membership subset are combined with the original error values ​​(including horizontal error and vertical error) obtained in step S1, and each error value is weighted transformed: , ;in and is the final membership value obtained after normalizing the membership of the horizontal deviation and the vertical deviation in step S2, and the weighted value obtained by using these memberships and Provide more accurate input data basis for subsequent frequency domain analysis and phase modulation. and Indicates the index of the member subset.

[0050] S32. Fifth-order Fast Fourier Transform Analysis of Horizontal Error: The weighted horizontal error is decomposed by a fifth-order fast Fourier transform, mapped from the time domain to the frequency domain, and recursively decomposed into five layers. The output of the first layer of fast Fourier transform is set to ,right After frequency band separation and nonlinear phase correction, a new time domain signal is obtained, and fast Fourier transform is performed again until the fifth-order decomposition is completed. The process can be expressed as: ;in represents the fast Fourier transform operator, is the initial time domain signal, It is the signal restored by the last inverse fast Fourier transform. Through five-order iterations, each order performs nonlinear mapping on the phase and amplitude characteristics (for example, using exponentially nested nonlinear phase correction on the frequency components), thereby obtaining horizontal error composite spectrum data containing multi-level frequency domain characteristics. .

[0051] The five-layer fast Fourier transform decomposition of horizontal errors in the attitude control system of the shield machine is based on the need for in-depth analysis of the signal frequency domain characteristics. The five-layer decomposition can effectively capture multi-level frequency components from low frequency to high frequency, ensuring that the system can not only identify slowly changing trend errors, but also detect rapidly changing transient disturbances. In addition, the five-layer decomposition achieves the best balance between spectral resolution and computational complexity, providing sufficient frequency domain details to support accurate judgment coefficient calculations while avoiding the consumption of computing resources and real-time processing delays caused by too many layers. Through this multi-level frequency domain analysis method, the system can fully and meticulously understand and evaluate the dynamic behavior of the shield machine in a complex geological environment, thereby providing high-quality input data for subsequent phase modulation and data fusion, and ultimately achieving accurate dynamic optimization of PID controller parameters.

[0052] S33. Double nonlinear phase modulation of vertical error: Double nonlinear phase modulation is performed on the weighted vertical error. First, the weighted vertical error is mapped to the complex domain: ;in is a nonlinear phase function dynamically calculated based on actual geological parameters. , , , is a preset constant. This phase function depends nonlinearly on the square term of the vertical error value and the sine transform, ensuring that different phase offset characteristics are presented under different error ranges.

[0053] Repeat the process of step S33 twice, that is, apply the same type of nonlinear phase modulation to the phase modulated signal again to obtain a more complex complex phase signal. .

[0054] S34. Multi-dimensional data interweaving and fusion: The horizontal error composite spectrum extracted by the fifth-order fast Fourier transform and the complex phase signal obtained by the double nonlinear phase modulation of the vertical error are interleaved and fused in the complex plane. This process uses multi-dimensional complex space interpolation to decompose the horizontal error composite spectrum and the complex phase signal into several sub-bands and sub-phase units, and cross-arrange them in the complex plane: ;in is the multidimensional frequency and phase joint domain, represents tensor concatenation operations over complex fields, is a nonlinear weighting factor to ensure that the two-directional data form a complementary structure in the multidimensional space, so that the final fusion matrix At the same time, the frequency domain and phase characteristics of horizontal and vertical errors are retained.

[0055] in, It is the frequency component index used to index the frequency domain features extracted from the weighted horizontal error after the fifth-order fast Fourier transform; It is a phase component index used to index the independent phase signal generated by the weighted vertical error after being processed by the dual nonlinear phase modulation algorithm.

[0056] 5. High-dimensional exponential integral calculation judgment coefficient: Perform high-dimensional exponential integral calculation on the fusion matrix to obtain the judgment coefficient for judgment .

[0057] The high-dimensional exponential integral is defined as: ;in is the complex modulus value of the corresponding unit in the fusion matrix, represents the natural logarithm, is a nonlinear distribution coefficient related to geological parameters, cutterhead speed, and propulsion speed, which emphasizes the influence of specific frequency and phase components in a weighted manner. is a normalization factor used to map the exponential integration result to a reasonable interval (e.g., 0 to 1).

[0058] Step S3 performs multi-level transformation and fusion processing on the horizontal error, vertical error and their membership relationship based on the existing data in the first two steps. The multi-frequency domain characteristics of the horizontal error are extracted by the fifth-order fast Fourier transform, and the phase characteristics of the vertical error signal are enhanced by double nonlinear phase modulation. The two are intertwined and fused in multi-dimensional space, and finally the judgment coefficient is calculated by high-dimensional exponential integration. This judgment coefficient accurately depicts the multi-dimensional complexity of the attitude error, and provides solid data and decision-making support for the subsequent fuzzy rule weight adjustment and PID gain adaptive adjustment.

[0059] After the processing of the previous steps, the system has obtained the accurate membership and judgment coefficient of the multi-dimensional posture data. As the core indicator reflecting the complexity of the shield machine posture adjustment, the judgment coefficient has a key impact on the subsequent PID gain parameter weight allocation. The goal of step S4 is to make full use of the judgment coefficient and redistribute the weights of the outputs of each rule in the Mandelbrot fuzzy control rule base, so that under conditions with higher complexity, the influence of specific rules on the gain adjustment process is more prominent; under conditions with lower complexity, a relatively balanced rule effect is maintained. Through this weight adjustment and fuzzy reasoning process, the fuzzy output is finally defuzzified into a clear PID gain adjustment amount using the maximum membership method.

[0060] Step S4 includes the following contents: 1. Mandelbrot type fuzzy control rule base structure: Assume that the rule base contains rules, each rule is as follows: ;in is the error, is the error change rate, and It is a subset of the seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large. , , They are the fuzzy output subsets for proportional, integral, and differential gain adjustments respectively.

[0061] 2. Calculation of weight adjustment factor: In general, each fuzzy rule has the same weight in Manderly-type reasoning. To achieve adaptive adjustment, define each rule The weight factor as follows: ; In the formula is a preset constant used to reflect the rules Base sensitivity to gain adjustments; is a nonlinear index used to enhance the weighting effect at high complexity. When the judgment coefficient is small, The value is lower, When it is close to 1, it means that each rule basically maintains its initial weight; when the judgment coefficient increases and reaches a higher level, the rule weight factor increases rapidly, allowing specific rules to play a greater role under high complexity conditions.

[0062] 3. Mandermey type fuzzy reasoning process: Based on the known error and error change rate membership (from step S2), antecedent matching is performed on each rule, and the fuzzy synthesis of rule antecedents is realized by using the minimum operation: ;in and are the membership degrees of error and error change rate in the corresponding subsets respectively.

[0063] Then, As the cut-set height of the fuzzy set output as the consequent, the fuzzy subsets of the proportional, integral, and differential gain adjustment quantities are cut (i.e., the fuzzy reasoning steps in the Mandelbrot-type reasoning), and the weight factor is combined Vertically enlarge the membership function of the corresponding cut-set posterior: ;in is the original membership function of the corresponding rule output subset. Through this operation, the output fuzzy set of each rule is weighted and deformed according to the judgment coefficient and its own matching degree.

[0064] 4. Fuzzy output synthesis and aggregation: The weighted output fuzzy sets of all rules are maximized and the corresponding gain adjustment fuzzy outputs are synthesized into three final fuzzy sets: , , ; Through maximum aggregation, the weighted influence of all rule outputs is integrated into the overall fuzzy result.

[0065] 5. Defuzzification to obtain PID gain adjustment: The maximum membership method is used to calculate the final fuzzy set , , This method searches for the point with the largest membership degree in the domain of the gain adjustment, and outputs the value of this point as the final PID gain adjustment: , , ; obtained , , They are the precise values ​​of proportional, integral, and differential gain adjustments respectively.

[0066] Step S4, under the guidance of the judgment coefficient, redistributes the weights of the Mandermi fuzzy control rule base, generates the fuzzy output of the gain adjustment amount through the Mandermi fuzzy reasoning process, and then converts the fuzzy output into a clear PID gain adjustment value using the maximum membership method. Through the processing of this step, the control system can adaptively adjust the relative weights of proportional, integral and differential gains under complex working conditions, improve the accuracy and real-time performance of attitude adjustment, and provide a highly reliable data basis for the dynamic update and execution of PID controller parameters in subsequent steps.

[0067] In the previous steps, the system has extracted accurate membership distribution and judgment coefficients from the multi-dimensional posture data, and obtained the corresponding proportional gain adjustment, integral gain adjustment and differential gain adjustment through fuzzy reasoning and defuzzification. At this time, the basic parameters of the PID controller (proportional gain, integral gain, differential gain) have not been actually updated, and these adjustments need to be used to dynamically correct the existing PID parameters. The core of step S5 is to nonlinearly superimpose the gain adjustment obtained from the previous defuzzification step with the current PID parameters to achieve adaptive update of the PID parameters, and by introducing complex mapping and transformation methods, the gain update response to posture control is more flexible and targeted, thereby achieving more accurate posture adjustment under complex working conditions.

[0068] Step S5 includes the following contents: 1. Nonlinear mapping function design: Although simply adding the gain adjustment amount to the old parameters can achieve the update, it cannot reflect the system's flexible adaptation to complex environments. Therefore, a nonlinear mapping function is used to perform complex nonlinear amplification or compression on the gain adjustment amount to achieve flexible control of parameter adjustment.

[0069] The following nonlinear mapping methods are defined: (1) Update of proportional gain: ; This formula means: When the judgment coefficient is close to 1, it is amplified exponentially, and when the judgment coefficient is low, it changes smoothly, so that the adjustment range of the proportional gain is significantly enhanced under highly complex working conditions.

[0070] The gain adjustment is mapped to a sinusoidal relationship in the interval [-1,1] to ensure a sensitive response to small adjustments and an oscillation suppression effect for excessive adjustments.

[0071] (2) Update of integral gain: ; This formula means: is a Sigmoid function. When is positive and large, the update amplitude approaches 1, otherwise it approaches 0, so the adjustment of the integral gain has asymmetric and nonlinear characteristics. When it is positive, the integral gain increases faster; when the judgment coefficient is low or When negative, the update amplitude is limited.

[0072] (3) Update of differential gain: ; This formula means: When the judgment coefficient is higher, it will have a stronger influence on the differential gain, thereby suppressing rapid error changes more timely under complex working conditions.

[0073] when It may have a slight inhibitory effect when it becomes larger to ensure that the adjustment of the differential gain is not over-amplified, so that the system maintains stability while suppressing high-frequency oscillations.

[0074] 2. Output after parameter update: Through the above nonlinear mapping, the proportional, integral, and differential gain adjustment amounts are integrated with the judgment coefficients and their respective complex function mappings to finally obtain the new PID parameters. , , .

[0075] These new parameters will replace the original PID parameters and be used in the proportional, integral and differential calculation process in the subsequent step S6.

[0076] Explanation of parameter changes: In complex geological environments or working conditions with drastic attitude changes, the judgment coefficient is too large, and the compound operation of exponential and trigonometric functions will moderately amplify the gain parameter, thereby increasing the sensitivity of the controller to sudden deviations. When the working conditions are relatively simple and the judgment coefficient is relatively small, the gain update is compressed and passivated, so that the system maintains a relatively stable control strategy.

[0077] Step S5 achieves fine dynamic adjustment of proportional, integral and differential gains by performing complex nonlinear superposition of the gain adjustment amount obtained by defuzzification and the existing PID parameters, and introducing the judgment coefficient as an elastic amplification or suppression factor. Through this process, the system can automatically adapt and optimize its own control strategy when facing tunnel construction environments of different complexities, so that the PID controller can show higher accuracy and flexibility in subsequent work, laying a solid foundation for realizing high-precision automatic control of the shield machine posture.

[0078] During the tunneling process, the precise control of the shield machine's posture is of key significance to ensure the construction quality and equipment stability. Posture deviation may cause problems such as directional deviation and uneven force, thus affecting the tunneling efficiency and construction safety. Therefore, adjusting the proportional gain, integral gain and differential gain values ​​through the PID controller to dynamically generate control outputs, adjusting the thrust of the partition cylinder and correcting the posture deviation become the core means to solve this problem. The goal of step S6 is to use the updated PID parameters to perform comprehensive error calculations, generate high-precision total control outputs, and directly drive the partition cylinder actuators to complete posture correction.

[0079] Step S6 includes the following contents: 1. Determination of input data: Collect the proportional gain, integral gain and differential gain values ​​dynamically updated in step S5. The current error value and error change rate provided by steps S1 and S2 are recorded as and , which serve as the calculation input for this step.

[0080] 2. Decomposition of error calculation: According to the PID control logic, three correction quantities are calculated in sequence: ; Item 1 Indicates the proportional correction amount of the current error, which is used to reflect the magnitude of the error in real time; The proportional correction is achieved by multiplying the current error value by the updated proportional gain, reflecting the direct impact of the current deviation; Item 2 represents the integral correction, where the integral kernel function A third-order nonlinear weight function is used, which is: ; Through nonlinear amplification of the accumulation of historical errors, the ability to compensate for continuous error deviation is enhanced; The integral correction is based on the historical error accumulation and combined with the nonlinear kernel function to amplify the impact of the continuous deviation. Item 3 represents the differential correction, where the differential function is the hyperbolic tangent function: ; Smooth the output at high error change rates to avoid over-regulation caused by sudden changes; The differential correction depends on the error change rate and is combined with a smoothing function to limit the risk of over-adjustment caused by sudden changes.

[0081] 3. High-dimensional error compensation: After calculating the basic control output Then, add the attitude compensation parameters , which is generated by the five-dimensional exponential convolution formula and is used to dynamically balance the asymmetric distribution of thrust between cylinder groups: ; Represents the residual error in five directions; is the direction-related error sensitivity coefficient, which is determined through the previous attitude optimization experiment; is the exponential decay factor; is the normalization constant.

[0082] The final total control output is: ; The control quantity after error correction and compensation is transmitted to the partition cylinder actuator for thrust adjustment.

[0083] By combining the dynamically adjusted PID parameters with nonlinear calculation and high-dimensional error compensation technology, step S6 generates an accurate total control output to ensure the flexibility and efficiency of thrust adjustment. This achieves accurate correction of multi-directional attitude deviations and provides a stable mechanical foundation and optimization guidance for the execution of step S7.

[0084] The core of the shield machine's posture control is to achieve precise correction through dynamic adjustment of the thrust of grouped cylinders. Due to the complex construction environment, the shield machine's posture may deviate due to uneven force or changes in geological conditions. Posture adjustment requires not only real-time perception of error signals, but also joint regulation of parameters such as the cylinder's thrust increment and cutterhead roll angle, while building an experience library to optimize subsequent control processes. S7 is a key link in the entire process, ensuring that posture deviations can be quickly corrected through multi-dimensional joint control when the thrust threshold is limited.

[0085] Step S7 includes the following contents: Step S7 uses the total control output signal obtained from step S6 to accurately adjust the thrust of the four groups of cylinders of the shield machine. By calculating the posture adjustment information in the total control output signal, the thrust adjustment amount of each group of cylinders is allocated, and the extension amount of each group of cylinders is ensured to always be within the preset safety threshold. When the extension amount of a group of cylinders exceeds the set threshold, its incremental output is immediately limited, and negative feedback control is used to apply adjustments to the cylinders in the opposite direction to quickly correct the posture deviation. At the same time, in order to correct the posture problems caused by insufficient cylinder adjustment, the roll angle is optimized by accurately controlling the reversing cutterhead to ensure the stability and accuracy of the shield machine's posture. During each control process, the system stores key data such as thrust adjustment amount, extension amount, and cutterhead reversal angle in the experience database, optimizes the subsequent control strategy with the help of data analysis, and realizes dynamic learning from historical experience to further improve the adaptability and robustness of the control.

[0086] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0087] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0088] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including a..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0089] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A control method for adjusting the tunneling posture of a shield machine, characterized in that: Includes steps: S1, using the laser positioning guidance system and total station to measure the various operating data of the shield machine in real time and convert the data into the original numerical set; S2, divide the horizontal deviation and vertical deviation and their change rate into seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large, and calculate the membership degree; S3, performs a fifth-order fast Fourier transform on the horizontal error, extracts high-frequency and low-frequency features, processes the vertical error to generate a phase signal, fuses the two-directional signals and calculates the judgment coefficient; S4, adjusting the weight of the fuzzy control rule base based on the judgment coefficient to generate a PID gain adjustment amount; S5, adding the defuzzified gain adjustment amount to the current PID gain to dynamically update the PID parameters; S6, using the updated PID parameters to calculate the total control output, and transmit it to the partition cylinder actuator to adjust the posture; S7, adjust the cylinder thrust according to the total control output.

2. A control method for adjusting the tunneling posture of a shield machine according to claim 1, characterized in that: Step S1 includes the following contents: A laser scanner is used to monitor the position of the shield head and tail of the shield machine in real time; the omnidirectional measurement capability of the total station is used to obtain the tunneling trend angle, roll angle and pitch angle of the shield machine in three-dimensional space; Convert the three-dimensional spatial posture data into complex domain representation; apply wavelet transform to perform multi-resolution analysis on the complex domain data to extract posture change features at different frequency levels; The extracted high-dimensional feature data is standardized; the standardized data is mapped to the [0,1] interval; and the processed standardized and normalized data is organized into a multi-dimensional original numerical set.

3. A control method for adjusting the tunneling posture of a shield machine according to claim 2, characterized in that: Step S2 includes the following contents: Horizontal deviation and vertical deviation Defined as the angular deviation between the current posture of the shield machine and the design axis; horizontal deviation change rate and vertical deviation rate of change Defined as the change in horizontal deviation and vertical deviation per unit time; Set the range of horizontal deviation, vertical deviation, horizontal deviation change rate and vertical deviation change rate to [-90°, +90°]; Divide the interval into seven subsets: negative large , Negative , negative small ,zero , Zhengxiao ,middle , Zhengda ; A triangular membership function is designed for each subset, with its peak at the center of the subset and the basis covering the boundaries of adjacent subsets; the center point of each subset is defined as and the boundary points are ; For any input value , which is in the subset The membership degree in Determined by the following triangular membership functions: ； Specific formula description: When the input value is in the left half of the subset When , the membership increases linearly from 0 to 1; When the input value is in the right half of the subset When , the membership decreases linearly from 1 to 0; When the input value is not within the interval of the subset, the membership is 0; For each input value, a vectorized calculation method is used to calculate its membership in the seven subsets: ; Use nonlinear weighting factors The membership degree of each subset is adjusted as follows: ; Among them, the weighting factor Determined by the following composite function: ;parameter , , is a preset constant; To ensure that the sum of all subset memberships is 1, the adjusted memberships are normalized, and then organized into a membership matrix, where each row corresponds to an input value and each column corresponds to a subset.

4. A control method for adjusting the tunneling posture of a shield machine according to claim 3, characterized in that: Step S3 includes the following contents: S31. Obtain the membership values ​​of the horizontal deviation and vertical deviation and their change rates in each membership subset; combine the corresponding membership information with the original error value, and perform a weighted transformation on each error value: , ;in and is the final membership value obtained after normalizing the membership of the horizontal deviation and the vertical deviation in step S2, and Indicates the index of the belonging subset; S32. Perform a fifth-order fast Fourier transform decomposition on the weighted horizontal error, map it from the time domain to the frequency domain, and recursively perform a five-layer decomposition; set the first layer fast Fourier transform output to ,right After frequency band separation and nonlinear phase correction, a new time domain signal is obtained, and fast Fourier transform is performed again until the fifth-order decomposition is completed; S33. Implementing dual nonlinear phase modulation on the weighted vertical error; first mapping the weighted vertical error to the complex domain: ;in is a nonlinear phase function dynamically calculated based on actual geological parameters. , , , is a preset constant; the same type of nonlinear phase modulation is applied to the phase modulated signal to obtain a more complex complex phase signal ; S34. Perform multi-dimensional data interleaving and fusion on the complex plane of the horizontal error composite spectrum extracted by the fifth-order fast Fourier transform and the complex phase signal obtained by the double nonlinear phase modulation of the vertical error; the process uses multi-dimensional complex space interpolation to decompose the horizontal error composite spectrum and the complex phase signal into several sub-bands and sub-phase units, and cross-arrange them on the complex plane; Perform high-dimensional exponential integral calculation on the fusion matrix to obtain the judgment coefficient for judgment .

5. A control method for adjusting the tunneling posture of a shield machine according to claim 4, characterized in that: Step S4 includes the following contents: Assume that the rule base contains rules, each rule is as follows: ;in is the error, is the error change rate, and It is a subset of the seven subsets: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large. , , They are the fuzzy output subsets of proportional, integral and differential gain adjustment quantities; Define each rule The weight factor as follows: ; In the formula is a preset constant used to reflect the rules Base sensitivity to gain adjustments; It is a nonlinear index used to enhance the weighting effect at high complexity; Based on the known error and error change rate membership, each rule is matched with its antecedents, and the minimum operation is used to achieve the fuzzy synthesis of the rule antecedents: ;in and are the membership degrees of error and error change rate under the corresponding subsets respectively; Then, As the cut-set height of the subsequent output fuzzy set, the fuzzy subsets of the proportional, integral, and differential gain adjustment are cut, and the weight factor is combined Vertically enlarge the membership function of the corresponding cut-set posterior: ;in is the original membership function of the corresponding rule output subset; Perform maximum aggregation on the weighted output fuzzy sets of all rules, and synthesize the corresponding gain adjustment fuzzy outputs into three final fuzzy sets; The maximum membership method is used to defuzzify the final fuzzy set; the point with the maximum membership is found in the domain of the gain adjustment amount, and the value of this point is output as the final PID gain adjustment amount.

6. A control method for adjusting the tunneling posture of a shield machine according to claim 5, characterized in that: Step S5 includes the following contents: Use nonlinear mapping functions to perform complex nonlinear amplification or compression on the gain adjustment amount; The following nonlinear mapping methods are defined: (1) Update of proportional gain: ; (2) Update of integral gain: ; (3) Update of differential gain: ; in , , They are the proportional, integral, and differential gain adjustment values ​​respectively; Through nonlinear mapping, the proportional, integral, and differential gain adjustments are integrated with the judgment coefficients and their respective function mappings to finally obtain new PID parameters. , , .

7. A control method for adjusting the tunneling posture of a shield machine according to claim 6, characterized in that: Step S6 includes the following contents: According to the PID control logic, three correction quantities are calculated in sequence: The proportional correction is achieved by multiplying the current error value by the updated proportional gain, reflecting the direct impact of the current deviation; The integral correction is based on the historical error accumulation and combined with the nonlinear kernel function to amplify the impact of the continuous deviation. The differential correction depends on the error change rate and is combined with a smoothing function to limit the risk of over-adjustment caused by sudden changes. After calculating the basic control output, the attitude compensation parameters are added and generated through the five-dimensional exponential convolution formula to dynamically balance the asymmetric distribution of the thrust between the cylinder groups; finally, the total control output is obtained.

8. A control method for adjusting the tunneling posture of a shield machine according to claim 7, characterized in that: Step S7 includes the following contents: The thrust of the four groups of cylinders of the shield machine is adjusted by using the obtained total control output signal. The thrust adjustment amount of each group of cylinders is allocated by calculating the posture adjustment information in the total control output signal, and it is ensured that the extension amount of each group of cylinders is always within the preset safety threshold range.

Images