Normal pressure cutterhead flushing pipeline blockage early warning monitoring system of shield machine

Abstract

The application relates to the technical field of shield machine monitoring, and particularly discloses a normal-pressure cutterhead flushing pipeline blockage early warning monitoring system for a shield machine, which compares real-time flow with a minimum threshold value in real time, establishes a quick response mechanism to lock the limit working condition of complete blockage of the pipeline in time, and serves as the first logical defense line. Furthermore, under the condition that the pipeline is in a non-blockage flow state, the system uses a sliding window technology to deeply mine the dynamic evolution characteristics of historical flow data, accurately captures early gradual change signals of pipeline diameter reduction by extracting flow decay slopes, and effectively overcomes the hysteresis of traditional single-threshold alarm. More importantly, in view of the misjudgment risk under variable working conditions, a hydraulic conductance normalization and time-weighted regression mechanism is introduced, the rotating speed of a booster pump is taken as an active driving variable to completely decouple the booster pump from passive response flow data, and recent data is given a higher weight, so that false flow fluctuations caused by active speed adjustment of an operator are effectively eliminated.

Description

Technical Field

[0001] This application relates to the field of tunnel boring machine monitoring technology, and more specifically, to a monitoring system for early warning of blockage in the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine. Background Technology

[0002] When tunneling with a tunnel boring machine (TBM) through complex geological conditions, especially water-rich sand layers or highly viscous strata, the reliability of the atmospheric pressure cutterhead flushing system directly impacts construction safety and efficiency. The atmospheric pressure cutterhead sprays high-pressure slurry into the cutterhead panel and soil chamber through its central and peripheral flushing nozzles. This aims to clean and cool the cutters and critical structures, preventing the formation of mud cakes from the adhering excavated soil. If the flushing pipeline becomes blocked, not only will the core component of the cutterhead rapidly form mud cakes due to localized lack of flushing, but this will also lead to abnormally high excavation torque, excessive cutter wear, and even uneven wear failure. In severe cases, manual cleaning via high-risk pressurized entry into the chamber is necessary. Therefore, developing a monitoring system capable of real-time sensing of pipeline status and early warning of blockage risks is crucial for ensuring the safe long-distance tunneling of the TBM.

[0003] However, existing tunnel boring machine (TBM) pipeline monitoring methods typically employ a crude single-threshold alarm model, generally suffering from delayed response and high false alarm rates. Traditional monitoring systems often only focus on whether the pipeline is completely cut off, triggering an alarm only when the flow sensor reading drops to an extremely low level. By this time, the pipeline has often already experienced severe physical blockage, and operators have missed the optimal window for pulse flushing or parameter adjustments, leading to significant difficulties in post-incident handling. Furthermore, existing technologies lack the ability to analyze the deeper causes of flow changes, failing to distinguish whether the flow drop is caused by internal pipeline narrowing (i.e., a reduction in effective flow area due to gradual blockage on the pipe wall) or by operators actively reducing the booster pump speed. This singular monitoring dimension makes it difficult for the system to accurately identify true signs of blockage when facing complex variable-condition flushing operations, failing to meet the needs of refined construction.

[0004] Therefore, an optimized early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline in tunnel boring machines is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. An embodiment of this application provides an early warning and monitoring system for blockage in the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine.

[0006] According to one aspect of this application, a blockage warning and monitoring system for the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine is provided, comprising:

[0007] The raw signal acquisition module is used to acquire the raw current signal and booster pump operation signal on the flushing pipeline based on a preset sampling frequency;

[0008] The signal preprocessing module is used to discretize, sample, and filter the raw current signal to obtain real-time flow data.

[0009] The historical traffic time sequence arrangement module is used to inject real-time traffic data into the first-in-first-out buffer in timestamp order to obtain a historical traffic time sequence window containing traffic values ​​of multiple consecutive time periods.

[0010] The slurry flowability instantaneous state determination module is used to extract the current flow value from the real-time flow data and determine the pipeline slurry flowability instantaneous state based on the preset minimum flow threshold to obtain the current instantaneous flow state of the pipeline.

[0011] The pipeline diameter reduction prediction module is used to extract and predict the flow rate reduction trend characteristics of the data in the historical flow time window when the instantaneous flow status is displayed as non-blocking, so as to obtain the flow rate reduction trend value that characterizes the pipeline diameter reduction.

[0012] The traffic decay trend graded early warning module is used to make graded early warning decisions based on preset alarm gradients to obtain monitoring and early warning instructions.

[0013] Compared with existing technologies, this application provides a shield tunneling machine's atmospheric pressure cutterhead scour pipeline blockage early warning and monitoring system. It establishes a rapid response mechanism by instantaneously comparing real-time flow with a minimum threshold to immediately lock onto the extreme condition of complete pipeline blockage, serving as the first logical line of defense. Furthermore, when the pipeline is in a non-blocked flow state, the system utilizes sliding window technology to deeply mine the dynamic evolution characteristics of historical flow data. By extracting the flow attenuation slope, it accurately captures early gradual changes in pipeline diameter, effectively overcoming the lag of traditional single-threshold alarms. More importantly, to address the risk of misjudgment under varying operating conditions, the solution introduces hydraulic admittance normalization and time-weighted regression mechanisms. It completely decouples the booster pump speed as an active driving variable from the passively responding flow data and assigns higher weight to recent data. This effectively eliminates false flow fluctuations caused by operator-initiated speed adjustments, accurately quantifies the actual physical unobstructedness attenuation rate of the pipeline, and achieves a qualitative leap from post-event alarms to pre-event trend warnings under complex operating conditions. Attached Figure Description

[0014] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1This is a system block diagram of a shield machine's atmospheric pressure cutterhead flushing pipeline blockage early warning and monitoring system according to an embodiment of this application.

[0016] Figure 2 This is a flowchart of a shield machine's atmospheric pressure cutterhead flushing pipeline blockage early warning monitoring system according to an embodiment of this application.

[0017] Figure 3 This is a block diagram of the instantaneous state determination module for slurry flowability in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to an embodiment of this application.

[0018] Figure 4 This is a block diagram of the pipeline diameter reduction prediction module in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead scour pipeline of a tunnel boring machine according to an embodiment of this application.

[0019] Figure 5 This is a block diagram of the flow attenuation trend determination unit in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to an embodiment of this application. Detailed Implementation

[0020] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0021] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0022] While this application makes various references to certain modules of the systems according to embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0023] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0024] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0025] To address the shortcomings of existing monitoring methods for atmospheric pressure cutterhead scour pipelines, which rely on absolute threshold alarms leading to severe response delays and cannot effectively distinguish whether flow rate drops are caused by physical blockage in the pipeline or manual speed adjustment of the booster pump, thus making it difficult to accurately identify early signs of blockage under varying operating conditions, this solution adopts an interference-resistant trend rheology early warning monitoring strategy. Specifically, this application proposes an early warning monitoring system for blockage in the atmospheric pressure cutterhead scour pipeline of a tunnel boring machine. This system firstly identifies and eliminates extreme conditions of complete physical blockage in the pipeline by filtering and comparing the collected signals with instantaneous logic thresholds. Based on this, for slurry in a flowing state, the solution further constructs a historical flow time-series sliding window, using a hydraulic admittance normalization algorithm to decouple the passive response of the flow rate from the active drive of the booster pump speed, eliminating spurious fluctuations caused by operational adjustments. Finally, it combines a weighted least squares method based on a forgetting factor to perform trend regression calculations on the normalized data, assigning higher weights to recent data. In this way, the system can accurately extract the flow attenuation trend value that truly represents the degree of pipe diameter reduction from dynamically changing fluid data, and generate hierarchical monitoring and early warning instructions accordingly, thereby achieving accurate perception and proactive intervention in the early stage of mud cake formation.

[0026] Figure 1 This is a system block diagram of a shield machine's atmospheric pressure cutterhead flushing pipeline blockage early warning and monitoring system according to an embodiment of this application. Figure 2 This is a flowchart of a shield tunneling machine's atmospheric pressure cutterhead scour pipeline blockage early warning and monitoring system according to an embodiment of this application. Figure 1 and Figure 2As shown, the shield machine's atmospheric pressure cutterhead flushing pipeline blockage early warning monitoring system 100 according to an embodiment of this application includes: a raw signal acquisition module 110, used to acquire raw current signals and booster pump operation signals on the flushing pipeline based on a preset sampling frequency; a signal preprocessing module 120, used to discretize and filter the raw current signals to obtain real-time flow data; a historical flow time sequence arrangement module 130, used to inject the real-time flow data into a first-in-first-out buffer according to the timestamp order to obtain a historical flow time sequence window containing flow values ​​of multiple consecutive time periods; and a slurry fluidity instantaneous state determination module. 140 is used to extract the current flow value from the real-time flow data and determine the instantaneous state of the pipeline slurry flowability based on a preset minimum flow threshold to obtain the current instantaneous flow state position of the pipeline; 150 is used to extract and predict the flow attenuation trend characteristics of the data in the historical flow time series window when the instantaneous flow state position is displayed as non-blocking to obtain the flow attenuation trend value characterizing the degree of pipeline attenuation; 160 is used to make graded early warning decisions on the flow attenuation trend value based on a preset alarm gradient to obtain monitoring and early warning instructions.

[0027] In the aforementioned shield tunneling machine's atmospheric pressure cutterhead flushing pipeline blockage early warning monitoring system 100, the raw signal acquisition module 110 is used to acquire the raw current signal and booster pump operation signal on the flushing pipeline based on a preset sampling frequency. It should be understood that the shield tunneling site is filled with high-frequency electromagnetic interference generated by frequency converters and high-power motors, and a single flow reading lacking equipment power status constraints cannot logically distinguish between zero flow caused by system shutdown and flow interruption caused by physical blockage of the pipeline. Therefore, in the technical solution of this application, the raw current signal and booster pump operation signal on the flushing pipeline are acquired based on a preset sampling frequency to simultaneously obtain low-level fluid monitoring data with both high time-domain resolution and equipment operating status correlation attributes. This provides a precise data source containing complete physical characteristics for subsequent signal filtering and cleaning, and full-process effectiveness threshold filtering, ensuring that the entire early warning analysis logic is based on real and causally related working conditions.

[0028] Specifically, in one example of this application, the system utilizes electromagnetic flow sensors installed at the inlet and outlet of the atmospheric pressure cutter head flushing pipeline and an industrial bus module connected to the main controller to collaboratively perform the data acquisition task. First, the system receives the raw current signal in analog current form output by the flow sensor in real time through a high-precision analog input channel. The amplitude of this signal linearly corresponds to the instantaneous flow rate change of the flushing slurry within the pipeline. Simultaneously, the system reads the logic level of the contactor auxiliary contacts or the inverter status word in the booster pump control circuit in parallel through a digital input interface, directly acquiring the booster pump operation signal characterizing the immediate start / stop status of the booster pump. Second, the built-in high-speed analog-to-digital converter strictly follows the clock interrupt cycle generated by the system's pre-set sampling frequency to perform equal-time discretization, hold, and quantization encoding operations on the continuously changing raw current signal, converting it into a discrete numerical sequence recognizable by the digital processor. It ensures that each sampling point in this sequence is strictly aligned in time with the synchronously acquired booster pump operation signal, ultimately generating a raw mixed signal set containing fluid flow rate characteristic data and power source status characteristic data, which is then transmitted to the subsequent signal preprocessing module.

[0029] In the aforementioned shield tunneling machine's atmospheric pressure cutterhead flushing pipeline blockage early warning monitoring system 100, the signal preprocessing module 120 is used to discretize and filter the original current signal to obtain real-time flow data. It should be understood that analog signals collected in industrial settings are not only mixed with high-frequency electromagnetic noise generated by frequency converters and generator sets, making them unusable for real-time digital logic operations without processing, but also, during non-flushing operation periods, the zero-point drift of the sensors themselves often produces meaningless false readings. Therefore, in the technical solution of this application, the original current signal is further discretized and filtered to obtain real-time flow data, thereby achieving a high-fidelity conversion from the analog signal domain to the digital signal domain, and simultaneously filtering out random noise and invalid background data from non-operational states. This ensures that subsequent time-series trend analysis algorithms are based on pure, stable flow values ​​with real physical fluid properties, preventing noise errors from eroding the accuracy of early warnings at the source.

[0030] Specifically, in this embodiment, the signal preprocessing module includes: a current signal quantization and encoding unit, used to perform time-domain discretization acquisition and quantization encoding on the acquired raw current signal based on a preset sampling frequency to obtain a discrete flow sequence; a denoising and smoothing processing unit, used to perform denoising and smoothing processing on the discrete flow sequence based on a sliding window to obtain a smoothed flow value; and an effectiveness filtering unit, used to perform effectiveness logic threshold filtering on the smoothed flow value based on the booster pump operation signal to obtain real-time flow data.

[0031] More specifically, the current signal quantization and encoding unit is used to perform time-domain discretization and quantization encoding on the acquired raw current signal based on a preset sampling frequency to obtain a discrete flow sequence. It should be understood that since the raw current signal is essentially an analog physical quantity of electrical signal that continuously fluctuates with the pipeline flow velocity, and the core controller carrying the early warning algorithm can only recognize and process discrete binary digital logic, and the transient minute changes in pipeline flow velocity require sufficiently high time resolution to be completely recorded, the technical solution of this application uses a preset sampling frequency to perform time-domain discretization and quantization encoding on the acquired raw current signal to build a bridge connecting the continuous analog signal domain and the digital computing domain. This allows the continuous waveform representing the physical operating condition to be accurately converted into a standardized digital dataset that is readable and computable by a computer, establishing the most basic data structure for subsequent algorithm analysis.

[0032] Accordingly, in a specific example of this application, the system's data acquisition front end uses a built-in high-precision hardware clock source to generate stable periodic trigger pulses. The frequency of these pulses is strictly locked to a preset sampling frequency that conforms to the Shannon sampling theorem. Whenever a clock pulse edge is triggered, the sample-and-hold circuit inside the analog-to-digital conversion module immediately locks and holds the instantaneous amplitude of the current signal at the input terminal. Then, the quantizer is activated to map the locked analog level value into a binary digital code corresponding to the range. Subsequently, the system writes the digital codes generated by this series of continuous conversions into the data buffer register in a strict chronological order, thereby reconstructing a discrete flow sequence in the digital domain that is equidistantly distributed on the time axis and can restore the details of the fluid's dynamic changes.

[0033] More specifically, the denoising and smoothing unit is used to perform denoising and smoothing processing on the discrete flow sequence based on a sliding window to obtain smoothed flow values. It should be understood that due to the complex electromagnetic environment of the tunnel boring machine combined with the mechanical vibration of the scouring pipe itself, the original digital signal obtained directly through analog-to-digital conversion inevitably contains non-physical random white noise and instantaneous spikes and glitches generated by high-frequency electromagnetic interference and measurement jitter. If these high-frequency noise points are not processed and directly participate in subsequent precise trend calculations, it will lead to serious algorithm oscillations and false alarms. Therefore, in the technical solution of this application, the discrete flow sequence is further denoised and smoothed based on a sliding window to obtain smoothed flow values. This utilizes the statistical mean principle to filter out random interference components superimposed on the real signal, while fully preserving the low-frequency effective characteristics reflecting the slurry flow pattern. This significantly smooths out data fluctuations and outputs a smooth curve with a high signal-to-noise ratio that truly reflects the fluid movement, providing a stable data foundation for subsequent state determination and trend prediction.

[0034] Accordingly, in a specific example of this application, the denoising and smoothing processing unit first opens a first-in-first-out circular buffer with a predetermined depth in the system's volatile memory as a sliding window for data observation. The depth parameter of this window is finely set according to the hydraulic response time constant of the pipeline fluid. As the sampling process continues, the processor receives the latest sampled data points from the discrete flow sequence from upstream in real time and pushes them into the head of the buffer. At the same time, following the timing logic, it removes the earliest historical data point stored at the tail of the buffer, always maintaining a constant total amount of effective data within the window. Then, it performs fast accumulation summation and division operations on all discrete flow values ​​residing in the sliding window at the current moment to obtain the arithmetic mean. The calculation result is output as the smoothed flow value at the current moment after digital filtering and cleaning to the next-level processing unit, thereby realizing real-time dynamic denoising of the flow data.

[0035] More specifically, the effectiveness filtering unit is used to perform effectiveness logic threshold filtering on the smoothed flow rate value based on the booster pump operating signal to obtain real-time flow data. It should be understood that flow sensors often experience unavoidable zero-point drift or produce non-zero false readings due to environmental interference in a static state without flow. Furthermore, any numerical fluctuations during the shutdown state of the flushing system have no practical physical reference value for predicting pipeline blockage. Including such invalid data in subsequent calculations will inevitably lead to inaccuracies and misjudgments in the trend prediction model. Therefore, in the technical solution of this application, the smoothed flow rate value is further filtered based on the booster pump operating signal using effectiveness logic threshold filtering based on the equipment status to obtain real-time flow data, thereby constructing a strict logical hard-lock relationship between the data flow and the equipment's power status. This completely cuts off and eliminates background noise data generated during equipment shutdown or standby periods, ensuring that subsequent algorithms operate only under effective flushing conditions and with real physical driving force.

[0036] Accordingly, in the embodiments of this application, the validity filtering unit is used to: read the logic level of the booster pump operation signal in real time; and use logical AND operation to perform validity logic threshold filtering on the smooth flow value, retaining the value only when the booster pump is on, so as to obtain real-time flow data.

[0037] Specifically, in a specific example of this application, the validity filtering unit first reads the logic level of the booster pump operating signal, which represents the start-stop state of the power source, in real time by scanning the industrial fieldbus at high frequency or directly reading the I / O register of the controller. This level accurately corresponds to the physical operating conditions of the booster pump's operation and shutdown in binary high or low bits. Subsequently, the system uses a logical AND operation mechanism to perform validity logic threshold filtering on the smoothed flow rate value input from the front end. That is, the logic state value of the booster pump operating signal is used as a gate coefficient and Boolean multiplication is performed with the smoothed flow rate value. The value is retained as is and transmitted downstream only when the booster pump is in the on state, i.e., when the signal verification passes. When the pump is off, the data stream is directly forced to be reset to zero, thereby accurately eliminating drift data during the shutdown period and finally outputting real-time flow rate data that has been strictly verified by the operating condition.

[0038] In the aforementioned shield tunneling machine's atmospheric pressure cutterhead scour pipeline blockage early warning monitoring system 100, the historical flow time series arrangement module 130 is used to inject real-time flow data into a first-in-first-out (FIFO) buffer in timestamp order to obtain a historical flow time series window containing flow values ​​for multiple consecutive time periods. It should be understood that since the instantaneous flow value at a single moment can only characterize the current static physical state of the pipeline, it lacks time dimension information reflecting the evolution of fluid dynamics and cannot be directly used to calculate high-order trend indicators that depend on continuously changing characteristics, such as the flow attenuation slope. Therefore, in the technical solution of this application, real-time flow data is further injected into the FIFO buffer in timestamp order to obtain a historical flow time series window containing flow values ​​for multiple consecutive time periods, thereby constructing a data observation domain with temporal continuity and dynamic updates. This allows discrete and isolated monitoring points to be reconstructed into a time series set containing rheological trend characteristics, providing the necessary complete computational samples for subsequent accurate capture of the pipeline's gradual diameter reduction process.

[0039] Specifically, in this embodiment, the historical traffic time sequence arrangement module is used to: read the current system clock, bind the real-time traffic data with the current time value using time sequence tags and encapsulate the data to obtain time-stamped traffic data packets; append the time-stamped traffic data packets to the tail of the current cache queue, and remove the data at the head of the queue based on a preset window depth to obtain an updated cache queue; and perform component extraction and matrix reorganization on the updated cache queue to obtain a historical traffic time sequence window.

[0040] More specifically, in a concrete example of this application, the system first reads the high-precision current system clock by calling the timing interface of the underlying operating system. The received real-time traffic data is then strictly bound to the current time value using time-series tags and encapsulated into data packets, generating standardized time-stamped traffic data packets containing traffic values ​​and corresponding time indices. Subsequently, following the first-in-first-out queue maintenance principle, the system appends the newly generated time-stamped traffic data packets to the tail of the current cache queue and immediately verifies the queue length. Based on a preset window depth, it strictly removes outdated data from the excessively long queue head, thereby achieving synchronous sliding of the observation window along the time axis and obtaining an updated cache queue while maintaining a constant total data volume. Finally, batch parsing is performed on all data packets stored in the updated cache queue, extracting vectorized components and recombining matrices for the time index and traffic values, ultimately constructing a historical traffic time-series window that meets the requirements of least squares regression operations and contains traffic values ​​from multiple consecutive time periods.

[0041] In the aforementioned shield machine's atmospheric pressure cutterhead scour pipeline blockage early warning monitoring system 100, the slurry flowability instantaneous state determination module 140 is used to extract the current flow value from the real-time flow data and determine the pipeline slurry flowability instantaneous state based on a preset minimum flow threshold to obtain the current instantaneous flow state of the pipeline. It should be understood that, since complex flow trend prediction algorithms typically involve high-load matrix operations, and in extreme conditions where the pipeline experiences physical hard blockage due to the solidification of the transported medium, leading to complete flow interruption, continuing to perform continuous rheological trend analysis on almost zero static data not only lacks physical meaning but is also highly prone to algorithm collapse and misjudgment due to value division to zero or singularities. Therefore, in the technical solution of this application, the current flow value is further extracted from the real-time flow data, and the pipeline slurry flowability instantaneous state is determined based on a preset minimum flow threshold. This establishes a pre-emptive fast-cut logic gate in the hierarchical monitoring system, quickly filtering out blocking samples with no analytical value before entering higher-order operations. This ensures that subsequent trend calculation resources are concentrated only on effective operating conditions with actual liquidity and potential risk of deterioration, while achieving millisecond-level instant locking and response to a fatal failure such as complete blockage.

[0042] Figure 3 This is a block diagram of the slurry flowability instantaneous state determination module in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to an embodiment of this application. Figure 3As shown in the embodiments of this application, the instantaneous flow state determination module 140 of the slurry includes: a matrix addressing and numerical decoupling unit 141, used to perform matrix addressing and numerical decoupling on the last row vector of the historical flow time window to obtain the current instantaneous flow value; a threshold logic comparison and discrimination unit 142, used to perform threshold logic comparison and discrimination on the current instantaneous flow value based on a preset minimum flow threshold to obtain the original comparison result; and a status bit encoding and circulation unit 143, used to encode and circulate the status bit of the original comparison result to obtain the instantaneous flow status bit.

[0043] Specifically, in a specific example of this application, the judgment logic begins with the precise extraction of data. The matrix addressing and numerical decoupling unit directly performs a matrix last-row indexing operation on the historical flow time window stored in the cache. This last-row vector corresponds to the most recent sampling period in time. The flow component is extracted from the time-flow composite structure through the data decoupling algorithm, thereby obtaining the current instantaneous flow value without the influence of filtering delay. Subsequently, the threshold logic comparison and discrimination unit compares the obtained current instantaneous flow value with the minimum flow threshold preset by the system. This threshold is set based on the minimum flow rate limit required for the flushing pipeline to maintain basic cleaning capacity. If the flow value is higher than the limit, it is judged as passing; otherwise, it is judged as failing. Based on this, a binary raw comparison result is generated. Finally, the status bit encoding and flow unit receives the raw comparison result and maps it to an enumerated status code that conforms to industrial control standards. That is, when the result is passing, it is set to a flow code; when the result is failing, it is set to a blocking code. Finally, the instantaneous flow status bit for subsequent strategy flow control is output.

[0044] More specifically, the state bit encoding and transfer unit is used to: perform state mapping and control bit latching on the original comparison results, wherein the passing results are mapped to a flowing state code and the failing results are mapped to a blocking state code, thereby generating instantaneous flowing state bits. In other words, since the original logical comparison results are only low-level Boolean levels lacking state semantics, they are difficult to directly use as standard inputs to drive complex early warning strategy machines to perform multi-branch decisions. Furthermore, in a multi-task parallel processing architecture, transient jumps in state signals, if not maintained, can easily cause disorder in downstream control logic. Therefore, in the technical solution of this application, state mapping and control bit latching are performed on the original comparison results to complete the standardized conversion from low-level logic levels to system-level state enumeration values, and to ensure the absolute stability of the state indicator within a single control cycle. This provides a global state index with clear control semantics and resistance to interference for subsequent trend prediction modules or emergency shutdown procedures, thereby generating instantaneous flowing state bits. Accordingly, in a specific example of this application, the state processing logic first performs a lookup table mapping operation. According to the system's predefined finite state machine protocol, when the input original comparison result is logically true (pass), it is mapped to a binary flow status code representing that the pipeline flushing medium is in an effective flow condition. Conversely, when the result is logically false (fail), it is mapped to a blocking status code representing that the pipeline has experienced a physical hard blockage or flow interruption. Then, a control bit latching operation is performed, and the generated status code is immediately written into the global state holding register or a specific data block address inside the controller for locking and maintenance until the judgment result of the next sampling cycle refreshes it, thereby outputting an instantaneous flow status bit for parallel use by the entire system.

[0045] In the aforementioned shield machine's atmospheric pressure cutterhead scouring pipeline blockage early warning monitoring system 100, the pipeline diameter reduction prediction module 150 is used to extract and predictively analyze the flow decay trend characteristics of data within the historical flow time series window when the instantaneous flow status position shows a non-blocking state, in order to obtain a flow decay trend value characterizing the pipeline diameter reduction degree. It should be understood that simply monitoring the absolute value of instantaneous flow can only identify the current physical on / off state of the pipeline, but cannot perceive the dynamic evolution process of pipe diameter reduction caused by the gradual accumulation and thickening of mud cake on the pipe wall. Furthermore, traditional fixed threshold alarm mechanisms often only trigger after complete blockage occurs, exhibiting significant lag. Therefore, in the technical solution of this application, when the instantaneous flow status position shows a non-blocking state, the flow decay trend characteristics of data within the historical flow time series window are further extracted and predicted, thereby capturing the minute change rate of the effective flow cross-sectional area within the pipeline over time through mathematical modeling. This allows for the quantification of the pipeline diameter reduction rate in the initial stage of mud cake formation, thereby identifying the risk of soft blockage before complete fluid interruption and achieving predictive maintenance.

[0046] Specifically, in a specific example of this application, the instantaneous flow status bit is first subjected to logic gating verification. After confirming that the pipeline is in a flow state, the linear regression operation logic is activated. Then, the matrix data within the historical flow time series window is decoupled into column vectors, and the time series vector and flow series vector are extracted respectively. The arithmetic mean of the two is calculated to complete the data centering preprocessing. Next, the system uses the least squares algorithm to calculate the slope of the flow rate change relative to time, that is, by calculating the ratio of the covariance of time and flow rate to the variance of time to fit the optimal trend line. Finally, the system extracts the negative component of the fitted slope, and its absolute value is used as a physical indicator to quantify the rate of pipeline blockage deterioration, i.e., the flow rate attenuation trend value. Specifically, the formula for calculating the fitted slope is as follows:

[0047]

[0048] in, The slope of the fitted change in the output represents the change in flow rate per unit time. The window depth for the historical traffic time series window; and The first and second parts of the time series vector and the flow series vector are respectively the first part of the time series vector and the flow series vector. Data from each sampling point; and These represent the time mean and the flow rate mean, respectively. This formula uses statistical regression to eliminate the interference of single-point measurement noise, revealing macroscopically whether the flow rate of the scouring slurry is in a stable state or in an irreversible, continuous decaying state within a time window. The more negative the value, the more severe the narrowing phenomenon caused by mud cake in the pipeline, and the higher the risk of blockage.

[0049] Specifically, in the aforementioned slope estimation mechanism embodiment, which uses ordinary least squares to perform linear regression on single flow time-series data, while theoretically feasible, it suffers from two significant technical drawbacks in the specific industrial scenario of tunnel boring machine scouring pipelines. First, this mechanism ignores the unique causal coupling relationship between active drive and passive response within the system. Specifically, the change in pipeline flow (passive response variable) is not simply driven by time, but is strongly coupled with the operator-controlled booster pump speed (active drive variable). This leads to the risk of misjudgment: when pipeline resistance increases due to the initial formation of mud cake, if the operator actively increases the pump speed to maintain tunneling efficiency, the original flow rate reading may remain stable or even slightly increase. In this case, the least squares algorithm, which only analyzes the flow-time relationship, will arrive at a risk-free conclusion, resulting in serious false negatives and missing the early warning window. Conversely, if the operator normally reduces the pump speed due to geological changes or energy-saving considerations, the flow rate will naturally decrease. The least squares algorithm will calculate a negative slope, triggering unnecessary blockage warnings (false positives). This not only interferes with normal construction but also reduces the operator's trust in the system. Secondly, this mechanism lacks weighting for the freshness of time-series data. The least squares algorithm treats all historical data points within the sliding window equally, assigning them the same weight. However, in early warning scenarios, data points closer to the current time have a stronger indicative significance for judging the impending blockage risk. This equal-weighting approach results in a significant lag in the algorithm's response to sudden blockages (such as localized mud cakes that form rapidly in cohesive formations), making early warnings insufficient and potentially causing the blockage to develop to an unmanageable stage by the time the system issues an alarm.

[0050] To overcome the aforementioned shortcomings, an improved optimization mechanism is proposed. This mechanism introduces the booster pump's rotational speed data as a core verification dimension and performs weighted optimization on the trend assessment algorithm itself. In other words, by introducing pump speed as a verification dimension and weighting the trend assessment algorithm, proactive immunity to human interference caused by process operations is achieved, and the system's sensitivity to recent data changes is enhanced. This fundamentally improves the robustness and response sensitivity of the early warning logic, outputting a high-fidelity pipeline flow rate attenuation index after pump speed correction and time-weighted adjustment.

[0051] Specifically, in another specific example of this application, Figure 4 This is a block diagram of the pipeline diameter reduction prediction module in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead scour pipeline of a tunnel boring machine according to an embodiment of this application. Figure 4As shown in the embodiments of this application, the pipeline diameter reduction prediction module 150 includes: a vector decoupling unit 151, used to perform vector decoupling on the historical flow time series window in response to the instantaneous flow state being flowing to obtain a time series vector, a flow series vector, a time mean, and a flow mean; a slope estimation unit 152, used to perform slope estimation based on the least squares method on the time series vector, the flow series vector, the time mean, and the flow mean to obtain a fitted change slope; a booster pump speed acquisition unit 153, used to acquire a booster pump speed sequence aligned with the flow series vector time series; and a flow attenuation trend determination unit 154, used to determine the flow attenuation trend value based on the booster pump speed sequence, the flow series vector, and the time series vector.

[0052] More specifically, the vector decoupling unit 151 and the slope estimation unit 152 are used to perform vector decoupling on the historical flow time series window in response to the instantaneous flow state being flowing, to obtain a time series vector, a flow series vector, a time mean, and a flow mean. Furthermore, they perform slope estimation based on the least squares method on the time series vector, flow series vector, time mean, and flow mean to obtain the fitted change slope. It is worth noting that this implementation method is the slope estimation mechanism used in the first embodiment. It should be understood that since the historical flow time series window constructed by the first-in-first-out buffer is itself only a set of raw data points arranged in time sequence, the flow gradual change trend implied within it is implicit and cannot be directly used as a quantitative indicator to determine whether the pipeline is in a slow deterioration process. Therefore, in the technical solution of this application, in response to the instantaneous flow state being in a state of flow, the historical flow time series window is decoupled by vectorization to obtain a time series vector, a flow series vector, a time mean, and a flow mean. The slope of the time series vector, flow series vector, time mean, and flow mean is estimated using the least squares method to obtain the fitted change slope. This slope is then fitted to a trend line that characterizes the macroscopic change pattern through linear regression mathematical modeling. In this way, the scattered flow data points within the observation window can be refined and transformed into a single feature value that accurately characterizes the rate of deterioration or improvement of fluid flow within the pipeline per unit time.

[0053] More specifically, the booster pump speed acquisition unit 153 and the flow rate attenuation trend determination unit 154 are used to acquire a booster pump speed sequence aligned with the flow rate sequence vector time series, and further determine the flow rate attenuation trend value based on the booster pump speed sequence, flow rate sequence vector, and time series vector. It is worth noting that this implementation method is an improved mechanism used in the second embodiment, namely, introducing booster pump speed data as the core verification dimension and performing weighted optimization on the trend evaluation algorithm itself. It should be understood that because the linear regression mechanism of a single flow rate time series ignores the strong causal coupling between pipeline flow as a passive response variable and booster pump speed as an active driving variable, it cannot mathematically isolate the flow fluctuation artifacts caused by operator-initiated speed adjustments, and its equal-weighted processing of historical data also leads to a significant lag in capturing the trend of sudden blockage risks. Therefore, in the technical solution of this application, a booster pump speed sequence aligned with the flow sequence vector is obtained. Based on the booster pump speed sequence, flow sequence vector, and time sequence vector, the flow attenuation trend value is determined. This allows for the introduction of the real-time status of the power source as the core verification dimension and weighted optimization of the trend evaluation algorithm itself, constructing a predictive model that is proactively immune to process operation interference and has higher sensitivity to data timeliness. This fundamentally eliminates false alarms and missed alarms caused by frequent changes in operating conditions, accurately quantifies the intrinsic attenuation rate that truly characterizes the physical smoothness of the pipeline, and outputs a high-fidelity flow attenuation trend value.

[0054] Figure 5 This is a block diagram of the flow attenuation trend determination unit in the early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to an embodiment of this application. Figure 5 As shown in the embodiments of this application, the flow attenuation trend determination unit 154 includes: a hydraulic admittance normalization subunit 1541, used to perform hydraulic admittance normalization on the flow sequence vector and the booster pump speed sequence to obtain a hydraulic admittance sequence; an exponential attenuation weight allocation subunit 1542, used to perform exponential attenuation weight allocation based on the time span for each sampling point in the time series vector based on a preset forgetting factor to obtain a time attenuation weight set; and a least squares trend calculation subunit 1543, used to perform weighted least squares trend calculation on the time series vector and the hydraulic admittance sequence based on the time attenuation weight set to obtain the admittance attenuation weighted slope as the flow attenuation trend value.

[0055] Accordingly, the hydraulic admittance normalization subunit 1541 is used to perform hydraulic admittance normalization on the flow sequence vector and the booster pump speed sequence to obtain a hydraulic admittance sequence. It should be understood that, due to the passive response characteristics of pipeline flow changes, their numerical fluctuations depend not only on the evolution of the internal physical impedance of the pipeline but also directly on the operator's active adjustment of the booster pump speed. This strong coupling relationship means that in the initial stage of mud cake formation, if the operator artificially increases the pump speed to maintain tunneling efficiency, the original flow reading may remain stable, thus masking the true risk of blockage. Conversely, normal energy-saving speed reduction operations may trigger false alarms due to a sudden drop in flow. Therefore, in the technical solution of this application, the flow sequence vector and the booster pump speed sequence are further hydraulically normalized to obtain a hydraulic admittance sequence, thereby performing multi-dimensional data fusion calculations to decouple and isolate the human-induced adjustment interference from the drive end from the original observation data. In this way, the analytical perspective of the algorithm can be transformed from the apparent flow dimension, which is affected by a variety of factors, to the intrinsic unobstructed flow dimension, which simply reflects the efficiency of fluid transmission in the pipeline, ensuring that the monitoring indicators are only sensitive to changes in the physical state of the pipeline itself.

[0056] In a specific example of this application, firstly, a booster pump speed sequence strictly aligned with the flow sequence vector on the time axis is acquired in parallel via an industrial bus, ensuring that each flow sampling point has a corresponding drive speed value as a reference. Then, the entire time window is traversed, and a point-to-point ratio calculation is performed for each discrete sampling point, i.e., the quotient of the instantaneous flow rate and the corresponding booster pump drive speed is calculated, with a small regularization parameter introduced into the denominator to prevent numerical overflow errors caused by zero speed. Finally, these calculated ratios are reassembled in their original time order to generate a new set of physical quantitative indicators, namely, the hydraulic admittance sequence. The formula for calculating the normalized hydraulic admittance is as follows:

[0057]

[0058] in, The i-th element in the hydraulic admittance sequence represents the flow output efficiency per unit pump speed. and These represent the i-th element in the input flow rate sequence vector and the booster pump speed sequence, respectively. It is a small regularization term set up to prevent the denominator from being zero. The sequence length is the window depth of the time window. This step, through normalization, transforms the raw, operator-dependent absolute flow rate value into a hydraulic admittance value characterizing the flow output efficiency per unit pump speed. This normalization shifts the analytical benchmark of the early warning algorithm from superficial phenomena to an intrinsic patency indicator that better reflects the pipeline's health status. For example, when pipeline resistance increases due to the initial formation of mud cake, if the operator actively increases the pump speed to maintain tunneling efficiency, the raw flow rate reading will... It may remain stable or even increase slightly, but due to the booster pump speed as the denominator... A significant increase in hydraulic admittance will inevitably lead to a decrease in the calculated hydraulic admittance, thus accurately detecting this hidden risk of blockage and effectively avoiding missed detections. Under this mechanism, regardless of how the operator subsequently adjusts the speed of the booster pump (whether by increasing or decreasing the speed), as long as the physical state inside the pipeline (such as the degree of sludge cake formation) remains unchanged, the calculated hydraulic admittance value should remain theoretically stable. This effectively avoids false alarms and missed detections caused by pump speed adjustments, achieving precise monitoring of the actual physical state of the pipeline.

[0059] Accordingly, the exponential decay weight allocation subunit 1542 is used to perform exponential decay weight allocation based on time span for each sampling point in the time series vector based on a preset forgetting factor to obtain a time decay weight set. It should be understood that the ordinary least squares method based on equal weights not only lacks the ability to identify the freshness of time series data, but also suffers from a significant lag effect in responding to sudden changes in operating conditions due to its indiscriminate treatment of all historical data within the sliding window. This is particularly true in high-risk scenarios such as rapidly forming localized mud cakes in viscous formations, where data points closer to the current time are clearly more indicative of the impending blockage risk than distant historical data. Therefore, in the technical solution of this application, an exponential decay weight allocation based on time span is further performed on each sampling point in the time series vector based on a preset forgetting factor to construct a non-uniform weighted evaluation system that focuses on reflecting recent data changes. This allows the algorithm to simulate the judgment patterns of experienced operators who focus on the latest changes in the dashboard readings, giving more weight to recent data in decision-making, greatly reducing the system's detection delay for rapidly forming congestion risks, and gaining valuable time windows for taking preventive measures such as counter-pulse flushing.

[0060] In a specific example of this application, the weight allocation process first reads a pre-set scalar value strictly between 0 and 1 as a forgetting factor, which determines the system's forgetting rate of historical data. Then, it iterates through each sampling point within the historical traffic time-series window, calculates the time span difference between the index of that sampling point in the time-series vector and the index of the latest time at the end of the window, and applies this difference as an exponential power to the forgetting factor, thereby calculating an independent weight value for each data point within the window that exhibits an exponentially decreasing characteristic with increasing time distance. Finally, the system encapsulates these calculated weight values ​​according to the original time-series index order, generating a time-decay weight set that perfectly matches the dimension of the input vector. Specifically, the calculation formula for the time-decay weight set is as follows:

[0061]

[0062] in, It is an element in the time decay weight set, representing the first... The weight of each data point; It is a preset forgetting factor, with a value between 0 and 1; It is the window depth of the time window, that is, the total number of sampling points contained in the window; It is the index of the data point in the window, with a value range of [0, ..., ...]. The formula constructs an exponential decay model based on time distance, where... Precisely represents the first The historical data points are the distance from the latest time (index is...). The time distance. As this distance increases, the weight... will with The exponential decay of the base means that when the system performs subsequent trend calculations, it will significantly increase the weight of the latest data and rapidly reduce the influence of outdated data. This ensures that the output warning results can keenly capture millisecond-level deterioration trends in pipeline conditions, achieving a high dynamic response. In other words, by giving higher weight to recent data, subsequent trend calculations can respond more quickly to sudden changes in flow or admittance values, greatly shortening the system's detection delay for rapidly forming congestion risks and buying valuable time to take preventative measures.

[0063] Accordingly, the least squares trend calculation subunit 1543 is used to perform weighted least squares trend calculation on the time series vector and the hydraulic admittance sequence based on the time decay weight set to obtain the admittance decay weighted slope as the flow decay trend value. It should be understood that although the preceding steps generate a hydraulic admittance sequence that objectively reflects the physical smoothness of the pipeline and a time decay weight set that reflects the timeliness differences of the data, without organically integrating these two dimensions at the mathematical level through a statistical regression algorithm that can fuse the weights, it is still impossible to generate a single, accurate decision indicator that can quantify the current deterioration rate of the pipeline. Therefore, in the technical solution of this application, a weighted least squares trend calculation is further performed on the time series vector and the hydraulic admittance sequence based on the time decay weight set, thereby deeply coupling the interference-resistant normalized physical characteristics with the highly sensitive timeliness weights to calculate a trend feature value with high fidelity. In this way, the output is no longer a simple apparent flow rate change rate, but a pipeline patency decay rate after pump speed correction and time weighting, thereby achieving robust quantification of the risk of physical blockage under complex operating conditions.

[0064] In a specific example of this application, firstly, using the input time-decay weight set, a weighted average operation is performed on the time series vector and the hydraulic admittance sequence respectively to calculate the weighted time mean and weighted hydraulic admittance mean with the center of gravity biased towards recent data, which are used as the weighted center point for regression analysis. Then, based on the core algorithm of weighted least squares, the module uses the weight set to weighted correct the deviation of each sampling point, calculating the weighted time-admittance covariance as the numerator and the weighted time variance as the denominator. Finally, the system performs a division operation to calculate the optimal fitting slope that accurately reflects the dynamic evolution of hydraulic admittance over time, i.e., the admittance decay weighted slope, and directly outputs it as the flow decay trend value for use by the graded early warning module. Specifically, the formula for calculating the admittance decay weighted slope is as follows:

[0065]

[0066] in, It is the final output admittance attenuation weighted slope (i.e., the flow attenuation trend value). It is the weight of the i-th sampling point in the time decay weight set; and They are respectively through Weighted average time value and hydraulic admittance mean; and These represent the i-th element in the time series vector and the hydraulic admittance sequence, respectively. Thus, the output of this step is... It is no longer a simple rate of change in flow rate, but rather a highly robust and sensitive characteristic indicator generated after pump speed correction and time-weighted adjustment: the rate of decline in pipeline patency. Through weighted least squares calculation, this indicator accurately isolates pump speed fluctuation interference and amplifies recent blockage signals. The magnitude of its negative value directly and accurately corresponds to the severity and development speed of physical blockage risk, enabling the system to issue reliable early warnings in the early stages of cake formation.

[0067] In summary, the second embodiment utilizes the aforementioned improved mechanism to address the dual challenges of high false alarm rates and slow response times in existing technologies under complex and variable operating conditions. By normalizing hydraulic admittance, it achieves proactive immunity to process operation disturbances, ensuring the robustness of the early warning logic and effectively avoiding false alarms caused by deceleration adjustments. Through time-domain exponential weighting, it achieves millisecond-level tracking of minute deterioration trends in pipeline conditions, significantly improving system sensitivity. Ultimately, this technology achieves the engineering effect of accurately quantifying pipeline diameter reduction rates and significantly reducing the risk of mud cake formation on the tunnel boring machine's atmospheric pressure cutterhead without increasing additional sensor hardware costs, solely through software algorithm upgrades.

[0068] In the aforementioned atmospheric pressure cutterhead scouring pipeline blockage early warning monitoring system 100 for tunnel boring machines, the flow attenuation trend graded early warning module 160 is used to make graded early warning decisions on the flow attenuation trend value based on a preset alarm gradient to obtain monitoring and early warning instructions. It should be understood that since the calculated flow attenuation trend value is merely a mathematical abstract scalar characterizing the rate of pipeline condition deterioration, and the main control system and actuators of the tunnel boring machine require explicit logical states or executable binary control words to drive physical actions, a single numerical output cannot directly trigger precise engineering intervention. Therefore, in the technical solution of this application, a graded early warning decision is further made on the flow attenuation trend value based on a preset alarm gradient to obtain monitoring and early warning instructions, thereby mapping the continuously changing quantitative trend indicator into discrete, operable risk control levels. In this way, differentiated response strategies, ranging from flexible interface prompts to mandatory hardware countermeasures, can be implemented according to the severity and urgency of the blockage development, thereby achieving a logical closed loop from monitoring and perception to control execution.

[0069] Specifically, in this embodiment, the flow attenuation trend graded early warning module is used to: use instantaneous flow state bits as logical gating conditions to perform priority interruption discrimination or interval bucketing calculation on the flow attenuation trend value and a preset graded threshold set to obtain a risk level code; use the risk level code as a retrieval key to perform table lookup and parameter extraction on the system's preset strategy mapping table to obtain an early warning response strategy vector containing text indexes and pump control parameters; and based on a preset industrial communication protocol, perform protocol packaging and displacement splicing on the early warning response strategy vector to obtain a monitoring early warning instruction.

[0070] More specifically, by using instantaneous flow status bits as logical gating conditions, priority interruption judgment or interval bucketing calculation is performed on the flow attenuation trend value and a preset set of graded thresholds to obtain a risk level code. It should be understood that pipeline blockage conditions present two distinct fault modes: sudden physical flow interruption and gradual mud cake diameter reduction. The former is an immediate critical fault; failing to address it first and continuing with complex numerical trend analysis not only wastes computing power but may also lead to erroneous conclusions due to invalid data sources. Furthermore, for gradual diameter reduction risks, a general alarm based on a single indicator cannot guide on-site personnel to differentiate between priorities and take appropriate intervention measures. Therefore, in the technical solution of this application, instantaneous flow status bits are used as logical gating conditions to perform priority interruption judgment or interval bucketing calculation on the flow attenuation trend value and a preset set of graded thresholds. This constructs a comprehensive risk quantification assessment model that integrates qualitative status and quantitative trends and possesses priority arbitration capabilities. This ensures that the system has the highest priority instantaneous response capability when facing extreme conditions such as complete physical blockage, and at the same time, while the pipeline is still in a flowing state, it can output standardized risk level instructions with tiered guidance based on the fine differences in the degree of diameter reduction deterioration.

[0071] Accordingly, in a specific example of this application, the instantaneous flow status bit is first read as the first-level logic gating condition. If the status bit indicates that the pipeline is in a non-flowing, blocked state, the system immediately performs a priority interruption judgment operation, forcibly terminates the subsequent numerical comparison process, and directly locks the output code representing the highest danger level (such as physical hard blockage). Conversely, when it is confirmed that the pipeline is in a flowing state, the numerical evaluation channel is activated, and a set of preset graded thresholds containing multiple preset alarm gradients (e.g., corresponding to the early mud cake warning threshold and the severe diameter reduction alarm threshold, respectively) is retrieved. The real-time input flow attenuation trend value is then compared with these thresholds. Subsequently, the system performs interval binning calculation based on the comparison result to accurately determine whether the current trend value falls into the safe zone, the low-risk warning zone, or the high-risk alarm zone, and accordingly quantifies and maps the evaluation result into the corresponding digital risk level code and outputs it to the downstream strategy module.

[0072] More specifically, the risk level code is used as the retrieval key to perform a lookup and parameter extraction on the system's pre-set strategy mapping table to obtain a warning response strategy vector containing text indexes and pump control parameters. It should be understood that the digital risk level code generated by the upstream algorithm is merely an abstract logical identifier representing the severity of pipeline blockage; it lacks the physical semantics to drive the underlying actuators to adjust motor speed and cannot be directly parsed into operator-readable alarm text by the human-machine interface terminal. Furthermore, different levels of danger require different mechanical countermeasures to achieve optimal blockage relief. Therefore, in this application's technical solution, the risk level code is further used as the retrieval key to perform a lookup and parameter extraction on the system's pre-set strategy mapping table to obtain a warning response strategy vector containing text indexes and pump control parameters. This establishes a standardized mapping mechanism from abstract decision-making logic to concrete execution actions. This enables modular configuration and decoupling of control strategies, ensuring that the system can automatically retrieve and output a combined response strategy containing precise human-machine interaction information and equipment driving logic based on the real-time risk assessment.

[0073] Accordingly, in a specific example of this application, the strategy execution unit pre-builds and maintains a strategy mapping table containing multi-level key-value pair relationships in the non-volatile storage area of ​​the controller. This table defines the association between each possible discrete value of risk level and a complete set of disposal schemes. During runtime, it first receives the real-time risk level code output by the evaluation module and uses it as a unique hash index key or array subscript to perform a fast table lookup operation in the strategy mapping table to accurately locate the corresponding strategy configuration line. Subsequently, it extracts parameters in parallel from the data in this line, reading out the text index (i.e., the character library call address) used to drive the interface display, and the pump control parameters used to directly guide the booster pump to perform prompts, pulse backwashing, or shutdown protection actions. Finally, the system encapsulates and combines these extracted heterogeneous control data in a structured manner to generate an early warning response strategy vector and transmits it to the downstream communication module.

[0074] More specifically, based on a preset industrial communication protocol, the early warning response strategy vector is packaged and shifted to obtain monitoring and early warning instructions. It should be understood that since the early warning response strategy vector generated by the algorithm layer is essentially an abstract data structure defined internally by the software, it cannot be directly recognized and parsed by heterogeneous hardware terminals distributed on the industrial fieldbus, and the data interaction between master and slave stations must strictly adhere to the frame format and timing specifications defined by the underlying physical link. Therefore, in the technical solution of this application, the early warning response strategy vector is further packaged and shifted based on a preset industrial communication protocol to compile the logical control parameters into a physical bit stream conforming to the fieldbus transmission standard. This ensures that early warning information and control actions are distributed losslessly to the human-machine interface and actuator controller, achieving the final connection of the software and hardware links.

[0075] Accordingly, in a specific example of this application, the system first loads a pre-set communication register mapping table compatible with industrial standards such as Profibus-DP, Modbus TCP, or Profinet to determine the target address offset of each component data. Then, the input warning response strategy vector is unpacked, and the text index, pump control parameters, and interface color code are shifted left or right to the high or low bit range defined by the protocol word using binary shift operation instructions. These are then merged into a complete control word through logical OR operation and shift concatenation operation. Finally, the system strictly follows the frame structure definition of the preset industrial communication protocol to add a start character, function code, and cyclic redundancy check code to the control word for protocol packaging, ultimately generating a monitoring and warning instruction that can be directly sent by the physical layer.

[0076] In summary, the shield machine's atmospheric pressure cutterhead scour pipeline blockage early warning monitoring system according to the embodiments of this application is explained. It establishes a rapid response mechanism by instantaneously comparing real-time flow with a minimum threshold to immediately lock onto the extreme condition of complete pipeline blockage, serving as the first logical line of defense. Furthermore, when the pipeline is in a non-blocked flow state, the system utilizes sliding window technology to deeply mine the dynamic evolution characteristics of historical flow data. By extracting the flow attenuation slope, it accurately captures the early gradual change signal of pipeline diameter reduction, effectively overcoming the lag of traditional single-threshold alarms. More importantly, to address the risk of misjudgment under varying operating conditions, the solution introduces hydraulic admittance normalization and time-weighted regression mechanisms. It completely decouples the booster pump speed as an active driving variable from the passively responded flow data and assigns higher weight to recent data, thereby effectively eliminating false flow fluctuations caused by operator-initiated speed adjustments. This accurately quantifies the actual physical unobstructedness attenuation rate of the pipeline, achieving a qualitative leap from post-event alarms to pre-event trend warnings under complex operating conditions.

[0077] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A monitoring and early warning system for blockage in the scour pipeline of a tunnel boring machine's cutterhead under atmospheric pressure, characterized in that, include: The raw signal acquisition module is used to acquire the raw current signal and booster pump operation signal on the flushing pipeline based on a preset sampling frequency; The signal preprocessing module is used to discretize, sample, and filter the raw current signal to obtain real-time flow data. The historical traffic time sequence arrangement module is used to inject real-time traffic data into the first-in-first-out buffer in timestamp order to obtain a historical traffic time sequence window containing traffic values ​​of multiple consecutive time periods. The slurry flowability instantaneous state determination module is used to extract the current flow value from the real-time flow data and determine the pipeline slurry flowability instantaneous state based on the preset minimum flow threshold to obtain the current instantaneous flow state of the pipeline. The pipeline diameter reduction prediction module is used to extract and predict flow attenuation trend features from historical flow time series data when the instantaneous flow status is non-blocking, in order to obtain a flow attenuation trend value characterizing the pipeline diameter reduction. It includes: a vector decoupling unit, used to perform vector decoupling on the historical flow time series window in response to the instantaneous flow status being open, to obtain a time series vector, flow series vector, time mean, and flow mean; a slope estimation unit, used to perform slope estimation based on the least squares method on the time series vector, flow series vector, time mean, and flow mean to obtain a fitted change slope; a booster pump speed acquisition unit, used to acquire a booster pump speed sequence aligned with the flow series vector time series; and a flow attenuation trend determination unit, used to determine the flow attenuation trend value based on the booster pump speed sequence, flow series vector, and time series vector. The flow attenuation trend graded early warning module is used to: use instantaneous flow status bits as logical gating conditions to perform priority interruption judgment or interval bucketing calculation on the flow attenuation trend value and the preset graded threshold set to obtain the risk level code; use the risk level code as the retrieval key value to perform table lookup and parameter extraction on the system's preset strategy mapping table to obtain an early warning response strategy vector containing text index and pump control parameters; and based on the preset industrial communication protocol, perform protocol packaging and displacement splicing on the early warning response strategy vector to obtain the monitoring early warning instruction. The flow attenuation trend determination unit includes: a hydraulic admittance normalization subunit, used to perform hydraulic admittance normalization on the flow sequence vector and the booster pump speed sequence to obtain a hydraulic admittance sequence; an exponential attenuation weight allocation subunit, used to perform exponential attenuation weight allocation based on a preset forgetting factor on each sampling point in the time series vector to obtain a time attenuation weight set; and a least squares trend calculation subunit, used to perform weighted least squares trend calculation on the time series vector and the hydraulic admittance sequence based on the time attenuation weight set to obtain the admittance attenuation weighted slope as the flow attenuation trend value.

2. The early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to claim 1, characterized in that, The signal preprocessing module includes: The current signal quantization and encoding unit is used to perform time-domain discretization and quantization encoding on the acquired raw current signal based on a preset sampling frequency to obtain a discrete flow sequence. The denoising and smoothing processing unit is used to perform denoising and smoothing processing on the discrete flow sequence based on a sliding window to obtain smooth flow values. The effectiveness filtering unit is used to filter smoothed flow values ​​based on equipment status-based effectiveness logic thresholds according to the booster pump operation signal to obtain real-time flow data.

3. The early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to claim 2, characterized in that, The validity filtering unit is used for: Real-time reading of the logic level of the booster pump's operating signal; The smoothed flow rate values ​​are filtered by a validity logic threshold using a logical AND operation, retaining the values ​​only when the booster pump is on, in order to obtain real-time flow rate data.

4. The early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to claim 1, characterized in that, The historical traffic time sequence arrangement module is used for: Read the current system clock, bind the real-time traffic data with the current time value using time-series tags, and encapsulate the data to obtain time-stamped traffic data packets; The time-stamped traffic packets are appended to the end of the current cache queue, and the data at the head of the queue is removed based on the preset window depth to obtain the updated cache queue. The updated cache queue is subjected to component extraction and matrix reconstruction to obtain the historical traffic time window.

5. The early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to claim 1, characterized in that, The instantaneous state determination module for slurry fluidity includes: The matrix addressing and numerical decoupling unit is used to perform matrix addressing and numerical decoupling on the last row vector of the historical traffic time series window to obtain the current instantaneous traffic value. The threshold logic comparison and discrimination unit is used to perform threshold logic comparison and discrimination on the current instantaneous flow value based on a preset minimum flow threshold to obtain the original comparison result; The status bit encoding and transfer unit is used to encode and transfer the status bits of the original comparison results to obtain the instantaneous flow status bits.

6. The early warning and monitoring system for blockage of the atmospheric pressure cutterhead flushing pipeline of a tunnel boring machine according to claim 5, characterized in that, The status bit encoding and transfer unit is used to: perform status mapping and control bit latching on the original comparison results, wherein the passing result is mapped to a flow status code and the failing result is mapped to a blocking status code to generate an instantaneous flow status bit.

Images