Cement mixing pile construction parameter self-adaptive regulation and control system based on edge computing

Abstract

The present application relates to the technical field of geotechnical engineering construction intelligent control and edge computing, in particular to a cement mixing pile construction parameter self-adaptive regulation and control system based on edge computing, which comprises an edge computing node deployed at a construction site; mechanical time sequence parameters, fluid time sequence parameters and spatial position parameters of construction equipment are collected; theoretical torque and theoretical pump pressure are generated based on an ideal fluid-structure coupling model and current drill bit depth; a theoretical residual template set is generated by injecting geological distortion parameters and equipment distortion parameters; the measured torque and measured pump pressure are aligned and differenced with the theoretical values to generate a real residual sequence and perform similarity comparison; according to the comparison result, normal construction, formation change, equipment abnormality or composite risk state are output, and construction parameter regulation and control instructions or anti-blocking disposal instructions are issued to the main shaft driving mechanism, lifting mechanism and grouting pump control loop, so as to realize the differentiation of underground formation change and equipment abnormality and real-time self-adaptive closed-loop regulation and control on site.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and edge computing technology in geotechnical engineering construction, specifically to an adaptive control system for cement mixing pile construction parameters based on edge computing. Background Technology

[0002] Cement mixing pile construction parameter control refers to judging the operating status of construction equipment and adjusting construction parameters based on parameters collected during drilling and grouting construction, such as torque, rotation speed, lifting speed, pump pressure, grout flow rate, and drill bit depth. Current cement mixing pile construction parameter control methods mainly include manual experience adjustment, alarm control based on fixed thresholds, and data analysis methods relying on remote platforms.

[0003] However, when adjusting construction parameters based on existing technologies, it is difficult to accurately distinguish between changes in underground strata and equipment anomalies such as blockage of grouting circuits and drill bit adhesion from superficially similar parameter fluctuations. On the other hand, the accuracy and timeliness of status identification and real-time control are also reduced due to problems such as time lag, misalignment and network latency in field data. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an adaptive control system for cement mixing pile construction parameters based on edge computing. Specifically, the technical solution of this invention includes:

[0005] The parameter acquisition module collects mechanical time-series parameters from the construction equipment, including measured drill rod torque, measured spindle speed, and measured drill bit lifting rate; fluid time-series parameters, including measured pump pressure and measured slurry flow rate; and spatial position parameters, including the current drill bit depth.

[0006] The benchmark reconstruction module generates theoretical benchmark parameters containing theoretical torque and theoretical pump pressure based on a preset ideal fluid-structure interaction model and spatial position parameters. This model characterizes the coupling relationship between the drilling cutting load, which is the theoretical torque, and the grouting load, which is the theoretical pump pressure, as they change with the current drill bit depth, measured spindle speed, measured drill bit lifting rate, and measured grout flow rate.

[0007] The parameter injection module injects geological distortion parameters that characterize abrupt changes in formation mechanical parameters and equipment distortion parameters that characterize blockages in the grouting circuit or abnormalities in the drill bit structure into the theoretical reference parameters. It also calculates the difference between the simulation reference curve after the distortion parameters are injected and the original theoretical reference parameters, thereby generating a set of theoretical residual templates in the form of data vectors with specific time steps or depth steps.

[0008] The residual calculation module, according to a unified sampling period and / or a unified depth range, aligns the measured drill rod torque and measured pump pressure with the theoretical torque and theoretical pump pressure respectively to generate a real residual sequence, and compares it with the theoretical residual template set to output the comparison result;

[0009] The control judgment module determines the status assessment result based on the comparison result. The status assessment result is selected from one of the following: normal construction status, geological change status, equipment abnormal status, or composite risk status.

[0010] The control and execution module communicates with the construction equipment controller and issues construction parameter control commands or anti-blocking commands to the controller based on the status assessment results, so as to control the main shaft drive mechanism, lifting mechanism and grouting pump control circuit of the construction equipment.

[0011] Optionally, the parameter acquisition module includes:

[0012] The mechanical parameter acquisition unit is used to acquire measured drill rod torque, measured spindle speed and measured drill bit lifting rate.

[0013] The fluid parameter acquisition unit is used to acquire measured pump pressure and measured slurry flow rate;

[0014] The position parameter acquisition unit is used to acquire the current drill bit depth and inclination angle;

[0015] The mechanical timing parameters are derived from the mechanical parameter acquisition unit, the fluid timing parameters are derived from the fluid parameter acquisition unit, and the spatial position parameters are derived from the position parameter acquisition unit.

[0016] Optionally, the reference reconstruction module includes:

[0017] The soil mechanics modeling unit is used to generate theoretical cutting loads based on preset soil strength parameters and the current drill bit depth.

[0018] The fluid dynamics modeling unit is used to generate theoretical slurry delivery load based on preset slurry rheological parameters and the measured slurry flow rate;

[0019] The benchmark calculation unit, connected to the soil mechanics modeling unit and the fluid mechanics modeling unit, is used to generate the theoretical benchmark parameters by combining the theoretical cutting load, the theoretical grouting load, the measured spindle speed, and the measured drill bit lifting rate.

[0020] The theoretical benchmark parameters include the theoretical torque and the theoretical pump pressure; the preset soil strength parameters and the preset slurry rheological parameters are entered by the operator before construction through a human-machine interface deployed at the construction site, or downloaded by the edge computing node from the engineering survey database and construction organization design database in the cloud via a wireless network.

[0021] Optionally, the parameter injection module includes:

[0022] The geological distortion injection unit is used to inject a segmented step increment relative to the preset soil cohesion benchmark value into the theoretical benchmark parameter in order to generate a stratigraphic abrupt change residual template.

[0023] The equipment distortion injection unit is used to inject preset pipe cross-sectional area attenuation parameters and drill bit equivalent diameter increase parameters used to characterize the increase in cutting profile caused by drill bit adhesion blockage into the theoretical reference parameters, so as to generate a blockage abnormality residual template.

[0024] The template library unit, connected to the geological distortion injection unit and the equipment distortion injection unit, is used to store the formation abrupt change residual template and the blockage anomaly residual template.

[0025] The theoretical residual template set is derived from the template library unit.

[0026] Optionally, the residual calculation module includes:

[0027] The real residual generation unit is used to align and differentially process the measured drill rod torque and the measured pump pressure with the theoretical torque and the theoretical pump pressure at the same sampling time and / or the same depth interval through timestamp matching or depth index interpolation resampling, so as to generate the real residual sequence.

[0028] A temporal similarity comparison unit is connected to the actual residual generation unit and the template library unit for storing the theoretical residual template set. It is used to perform dynamic time warping and / or cross-correlation operations on the actual residual sequence and the theoretical residual template set to generate the comparison result. When cross-correlation operation is used, the temporal similarity comparison unit extracts the normalized maximum peak value in the cross-correlation function sequence and uses it as the similarity score between the actual residual sequence and the corresponding theoretical residual template.

[0029] Optionally, the control judgment module is further configured to:

[0030] When the comparison result indicates that the similarity score between the actual residual sequence and the formation mutation residual template is greater than or equal to a preset formation threshold, and the similarity score between the actual residual sequence and the formation mutation residual template is less than a preset blockage threshold, the formation change status is output.

[0031] When the comparison result indicates that the similarity score between the actual residual sequence and the blockage anomaly residual template is greater than or equal to the blockage threshold, and the similarity score between the actual residual sequence and the formation mutation residual template is less than the formation threshold, the device is output as an abnormal state.

[0032] When the comparison results indicate that the similarity score between the actual residual sequence and the formation mutation residual template and the similarity score between the actual residual sequence and the blockage anomaly residual template are both less than the corresponding threshold, the normal construction status is output.

[0033] When the comparison results indicate that the similarity score between the actual residual sequence and the formation mutation residual template, as well as the similarity score between the actual residual sequence and the blockage anomaly residual template, are both greater than or equal to the corresponding thresholds, a composite risk status is output.

[0034] Optionally, the control execution module is further configured to:

[0035] When the state assessment result indicates a change in the formation state, a construction parameter control command is issued to reduce the drill bit lifting rate, increase the spindle allowable output torque limit set by the controller, and increase the grouting volume.

[0036] When the status assessment result indicates that the equipment is in an abnormal state, an anti-blocking control command is issued to stop the pump, run the in-situ dry water wash, and trigger an alarm.

[0037] When the status assessment result is a normal construction status, a maintenance command to maintain the current parameters is issued;

[0038] When the status assessment result is a composite risk state, a combined protection command of speed reduction, torque limitation, pump shutdown and alarm is issued.

[0039] Optionally, the edge computing node serves as a local processing center, used to generate the theoretical benchmark parameters, the theoretical residual template set, and the actual residual sequence in real time at the construction site, and output the comparison results and issue the instructions to reduce communication latency.

[0040] Optionally, the system further includes:

[0041] An adaptive update module, which is communicatively connected to the control judgment module and the edge computing node, is used to correct the model parameters used to generate the theoretical benchmark parameters in the ideal fluid-structure interaction model based on the actual residual sequence within the current construction cycle when the state evaluation result is a normal construction state.

[0042] Used to freeze the model parameter update when the status assessment result is an abnormal equipment status or a complex risk status;

[0043] Used to update the value range of the geological distortion parameter when the state assessment result is a state of stratigraphic change.

[0044] Optionally, the construction equipment is a cement mixing pile drilling and grouting construction equipment;

[0045] The mechanical timing parameters are the operating data of the actuator of the cement mixing pile drilling and grouting construction equipment, the fluid timing parameters are the operating data of the grouting circuit of the cement mixing pile drilling and grouting construction equipment, and the state assessment results are used to characterize the geological change state, abnormal equipment state, normal construction state, or complex risk state of the cement mixing pile drilling and grouting construction equipment during underground operations.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] 1. This system actively injects geological distortion parameters and equipment distortion parameters into the theoretical benchmark parameters generated based on the ideal fluid-structure interaction model by setting up a benchmark reconstruction module, a parameter injection module, and a residual calculation module. This generates a theoretical residual template set, which is then compared with the real residual sequence generated by the alignment and difference of measured parameters. This mechanism successfully separates underground stratum changes from surface parameter fluctuations such as grouting circuit blockage or drill bit structure abnormalities. This allows the control judgment module to accurately output normal construction status, stratum change status, equipment abnormal status, or composite risk status, enabling the control execution module to issue targeted construction parameter control commands or anti-blockage handling commands. This solves the problem of erroneous control caused by the difficulty in distinguishing the root cause of abnormalities in existing technologies.

[0048] 2. This system uses edge computing nodes as local processing centers deployed on the construction site, enabling the generation and similarity comparison of theoretical benchmark parameters, theoretical residual template sets, and actual residual sequences to be completed locally in real time, reducing communication latency caused by reliance on remote platforms. Simultaneously, the residual calculation module aligns the data through timestamp matching or deep index interpolation resampling, and extracts similarity scores using dynamic time warping and / or cross-correlation operations. This structure not only effectively tolerates time delays and misalignments during on-site data acquisition but also ensures that control commands meet real-time requirements, improving the accuracy and timeliness of construction equipment status identification and construction parameter control.

[0049] 3. This system establishes a model parameter update mechanism deeply coupled with the state assessment results by configuring an adaptive update module: under normal construction conditions, the model parameters of the ideal fluid-structure interaction model are corrected based on the actual residual sequence; under stratum change conditions, the range of values ​​for geological distortion parameters is updated; and when an abnormal equipment state or a complex risk state is identified, the model parameter update is strictly frozen. This mechanism ensures that the system can dynamically correct itself by following the slow changes in the actual stratum and equipment state over a long period of time, and prevents fault characteristics or uncontrolled anomalies from being incorrectly absorbed into the normal theoretical benchmark parameters, thereby ensuring the long-term identification accuracy and closed-loop control stability of the cement mixing pile drilling and grouting construction equipment throughout the complex underground operation process. Attached Figure Description

[0050] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0051] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0053] Example 1:

[0054] Please see Figure 1 An edge computing-based adaptive control system for cement mixing pile construction parameters includes edge computing nodes deployed at the construction site. These edge computing nodes include:

[0055] The parameter acquisition module collects mechanical time-series parameters from the construction equipment, including measured drill rod torque, measured spindle speed, and measured drill bit lifting rate; fluid time-series parameters, including measured pump pressure and measured slurry flow rate; and spatial position parameters, including the current drill bit depth.

[0056] The benchmark reconstruction module generates theoretical benchmark parameters containing theoretical torque and theoretical pump pressure based on a preset ideal fluid-structure interaction model and spatial position parameters. This model characterizes the coupling relationship between the drilling cutting load, which is the theoretical torque, and the grouting load, which is the theoretical pump pressure, as they change with the current drill bit depth, measured spindle speed, measured drill bit lifting rate, and measured grout flow rate.

[0057] The parameter injection module injects geological distortion parameters that characterize abrupt changes in formation mechanical parameters and equipment distortion parameters that characterize blockages in the grouting circuit or abnormalities in the drill bit structure into the theoretical reference parameters. It also calculates the difference between the simulation reference curve after the distortion parameters are injected and the original theoretical reference parameters, thereby generating a set of theoretical residual templates in the form of data vectors with specific time steps or depth steps.

[0058] The residual calculation module, according to a unified sampling period and / or a unified depth range, aligns the measured drill rod torque and measured pump pressure with the theoretical torque and theoretical pump pressure respectively to generate a real residual sequence, and compares it with the theoretical residual template set to output the comparison result;

[0059] The control judgment module determines the status assessment result based on the comparison results. The status assessment result is selected from one of the following: normal construction status, geological change status, equipment abnormal status, or composite risk status.

[0060] The control and execution module communicates with the construction equipment controller and issues construction parameter control commands or anti-blocking commands to the controller based on the status assessment results, so as to control the main shaft drive mechanism, lifting mechanism and grouting pump control circuit of the construction equipment.

[0061] This embodiment provides an adaptive control mechanism for cement mixing pile construction parameters based on edge computing. Specifically, when cement mixing piles are being constructed in a deep soft soil site, the construction equipment drills vertically along the pile location and grouts simultaneously. An industrial-grade edge computing node is set up on the site side. This node is connected to a torque sensor, a speed encoder, a lifting displacement sensor, a pump pressure sensor, a flow meter, and a drill bit depth detection device via a fieldbus, and communicates bidirectionally with the construction equipment controller.

[0062] Specifically, the parameter acquisition module collects mechanical timing parameters, fluid timing parameters, and spatial position parameters according to a preset cycle. For ease of explanation, the uniform sampling cycle is set to 0.1 seconds. Within a certain consecutive 5 sampling points, the measured drill rod torque is 82, 85, 87, 110, and 112 N·m, the measured spindle speed is maintained at 38 rpm, the measured drill bit lifting rate is 0.42, 0.42, 0.40, 0.38, and 0.35 m / min, the measured pump pressure is 0.92, 0.95, 0.96, 0.98, and 0.99 MPa, the measured slurry flow rate is 118, 118, 117, 117, and 116 L / min, and the current drill bit depth corresponds to 6.0, 6.1, 6.2, 6.3, and 6.4 meters. The above data is written to the local buffer of the edge node, along with a uniform timestamp and depth index.

[0063] The benchmark reconstruction module calls a preset ideal fluid-structure interaction model to generate corresponding theoretical benchmark parameters based on the current drill bit depth, spindle speed, drill bit lifting rate, and slurry flow rate. This model is used to jointly calculate cutting load and slurry load under the assumptions of homogeneous soil, intact equipment, and unobstructed slurry delivery circuit. Taking the above 5 sampling points as an example, the corresponding theoretical torques generated can be 80, 82, 84, 86, and 88 N·m, and the theoretical pump pressures can be 0.91, 0.93, 0.95, 0.96, and 0.98 MPa. This forms an ideal benchmark curve that changes synchronously with the field conditions, rather than a fixed single threshold.

[0064] Based on this, the parameter injection module does not directly perform a simple out-of-limit judgment on the measured data. Instead, it actively injects distortion factors into the theoretical benchmark parameters to obtain multiple theoretical residual templates. For example, if a geological distortion of a sudden increase in soil strength is injected into the theoretical torque, template A can be obtained, with torque residuals possibly being 0, 2, 4, 20, and 22 N·m, while pump pressure residuals are only 0, 0.01, 0.01, 0.02, and 0.01 MPa. If an equipment distortion of pipeline blockage and drill bit adhesion thickening is injected into the theoretical pump pressure and theoretical torque, template B can be obtained, with torque residuals possibly being 1, 3, 8, 15, and 28 N·m, and pump pressure residuals being 0.02, 0.05, 0.12, 0.25, and 0.40 MPa. The template set formed in this way can be continuously called locally for subsequent matching.

[0065] The residual calculation module aligns and differs the measured values ​​with the theoretical values ​​to obtain the actual residual sequence. Taking the aforementioned data as an example, the actual torque residuals are 2, 3, 3, 24, and 24 N·m, and the actual pump pressure residuals are 0.01, 0.02, 0.01, 0.02, and 0.01 MPa. Then, the actual residual sequence is compared with the template set for similarity. If the overall similarity with template A is 0.91, while the overall similarity with template B is only 0.34, it can be determined that the current waveform is more consistent with the characteristics of formation changes than the characteristics of equipment blockage.

[0066] After receiving the comparison results, the control judgment module outputs the status evaluation results, and the regulation execution module sends instructions to the construction equipment controller accordingly. Taking the similarity results as an example, the system can determine that the formation is in a state of change, and send instructions to the spindle drive mechanism to increase the upper limit of the allowable torque, to the lifting mechanism to reduce the lifting speed, and to the grouting pump control circuit to increase the grouting volume by a preset step size, so that the drilling process can adapt to the transition of hard layers.

[0067] In abnormal handling scenarios, if sensor data transmission loss, timestamp misalignment, or abnormal depth value jumps occur within a certain sampling period, the residual calculation module prioritizes data alignment verification. If alignment fails, the window does not enter the status judgment process; instead, the previous valid control command is used and an anomaly marker is recorded. Furthermore, only when the window is valid and the data quality participating in the comparison meets the requirements will the control judgment module output the normal construction status, stratum change status, equipment abnormal status, or composite risk status based on the similarity threshold. If the measured data quality is insufficient, the effective sample length is insufficient, or the alignment verification fails, the pending confirmation status is output and the current parameters are maintained for manual review or waiting for the next valid window to be judged. Thus, the pending confirmation status corresponds to the temporary handling of invalid windows or low-confidence windows, rather than a replacement classification of valid comparison results. If the sampled value suddenly exceeds the equipment safety limit, for example, the torque suddenly increases to more than twice the normal average, the protection logic is directly entered without going through the conventional comparison process.

[0068] During the construction of a cement mixing pile with a designed pile length of 12 meters, when the drill bit reached a depth of about 6.3 meters, the soil layer at the site transitioned from silty soil to stiff plastic clay, and the measured torque increased rapidly, but the pump pressure remained stable. After the edge node completed the ideal benchmark reconstruction, distortion template call and actual residual comparison locally, it was determined to be a change in the stratum state rather than a pipe blockage. Therefore, a combination of control measures was issued to reduce the lifting speed, increase the torque limit, and increase the grouting volume, so as to avoid the system misjudging it as a blockage and executing the pump stop, thus ensuring the continuity of pile construction.

[0069] The purpose of this step is to separate underground stratum changes and equipment anomalies from surface-similar torque fluctuations by using physical benchmark reconstruction and residual template matching, thereby achieving an adaptive control closed loop suitable for real-time on-site conditions.

[0070] To further clarify, the aforementioned pending confirmation status is only used to indicate window handling flags when data alignment fails, sample length is insufficient, or data quality is insufficient. Its function is to indicate that the current window will not trigger formal classification and control based on similarity comparison results. Under this handling flag, the control judgment module will not include it in any of the categories of normal construction status, stratum change status, equipment abnormal status, or composite risk status, nor will it write it as a new valid status assessment result into the subsequent adaptive update process. Instead, it will maintain the status assessment result corresponding to the previous valid window or simply record it as an invalid window event.

[0071] Therefore, the status assessment results remain consistent with the subsequent four-state determination in the effective comparison scenario, and the data quality handling and the working condition status determination are separated from each other. In order to avoid unclear terminology boundaries, in this embodiment, the status assessment results refer specifically to the normal construction state, the stratum change state, the equipment abnormal state, or the composite risk state when the effective comparison conditions are met. The state to be confirmed is only used as a window validity management mark and is not used as a formal state category in subsequent adaptive updates, control strategy switching, or working condition statistical analysis.

[0072] The parameter acquisition module includes:

[0073] The mechanical parameter acquisition unit is used to acquire measured drill rod torque, measured spindle speed and measured drill bit lifting rate.

[0074] The fluid parameter acquisition unit is used to acquire measured pump pressure and measured slurry flow rate;

[0075] The position parameter acquisition unit is used to acquire the current drill bit depth and inclination angle;

[0076] Among them, the mechanical timing parameters are obtained from the mechanical parameter acquisition unit, the fluid timing parameters are obtained from the fluid parameter acquisition unit, and the spatial position parameters are obtained from the position parameter acquisition unit.

[0077] This embodiment provides a unit-based acquisition mechanism for multi-source heterogeneous signals in the field. Specifically, in the previous implementation scenario, although the original parameters could be obtained by relying solely on a unified data entry point, under conditions of high vibration, high dust, and pumping pulsation, the sampling frequencies, filtering methods, and fault characteristics of different types of signals are not the same. If all signals are acquired in a mixed manner, problems such as time reference confusion and data quality degradation may easily occur. Therefore, the parameter acquisition module is subdivided into mechanical parameter acquisition unit, fluid parameter acquisition unit, and position parameter acquisition unit.

[0078] Specifically, the mechanical parameter acquisition unit prioritizes the rotary and lifting actuators. For example, the drill rod torque can be obtained from a torque sensor installed at the spindle output end, the spindle speed can be obtained from an encoder, and the drill bit lifting rate can be calculated from a hydraulic cylinder displacement sensor or a wire rope displacement sensor. For ease of explanation, suppose that within a 30-second window, the displacement sensor measures a change in drill bit depth from 6.00 meters to 6.21 meters, then the corresponding average lifting rate is approximately 0.42 meters per minute. If the sampling values ​​of two adjacent encoders correspond to the spindle rotating 0.063 revolutions within 0.1 seconds, then the converted spindle speed is approximately 37.8 revolutions per minute. Furthermore, if the on-site control cycle is shorter, the unit can first sample at the second level, and then output the average lifting rate at a 30-second or other set duration window to avoid incorrectly recording instantaneous jitter as process speed.

[0079] The fluid parameter acquisition unit prioritizes the slurry delivery circuit; pump pressure is acquired by a pump outlet pressure sensor, and slurry flow rate is acquired by an electromagnetic flowmeter or mass flowmeter; considering the periodic pulsation of pumping, a short window smoothing can be performed within this unit; for example, three consecutive original pump pressure values ​​are 0.90, 1.02, and 0.93 MPa, respectively, and after short window processing, 0.95 MPa can be output, avoiding the influence of a single pulse value on the accuracy of subsequent residual calculations;

[0080] In addition to acquiring the current drill bit depth, the position parameter acquisition unit can also acquire the drill rod inclination angle. The inclination angle is introduced because the same depth does not mean the same force. If the pile frame is slightly tilted, the cutting resistance of the drill bit and the slurry backflow state will be different. For example, at the same depth of 6.3 meters, if the inclination angle is 0.3 degrees, it can be regarded as approximately vertical. If the inclination angle increases to 2.1 degrees, the subsequent model can use a compensation coefficient to correct the theoretical torque.

[0081] Under the anomaly protection mechanism, if the mechanical parameter acquisition unit is normal but the fluid parameter acquisition unit experiences a short-term communication interruption, the system can temporarily retain the mechanical residual path, reduce the weight of equipment anomaly judgment, and not directly interrupt construction; if the position parameter acquisition unit fails and cannot obtain accurate depth, the theoretical benchmark reconstruction will lose key independent variables. At this time, the edge node freezes and automatically adjusts, retaining only over-limit protection; if the tilt angle value exceeds the allowable construction range, such as being greater than the preset 3 degrees, the control judgment module can prioritize outputting attitude anomaly prompts, so that subsequent parameter adjustments are not based on distorted pose.

[0082] During the same construction shift, when the equipment was switched from the first pile to the second pile, the flow meter experienced a two-second fluctuation due to the short-term adhesion of particles in the slurry, while the mechanical parameter acquisition unit and the position parameter acquisition unit remained normal. Since each unit works independently and has a local quality mark, the edge node does not mistakenly map the abnormal noise of the flow meter as an overall machine malfunction, but instead temporarily reduces the weight of the fluid data and continues to perform safety monitoring.

[0083] The purpose of this mechanism is to acquire, verify and preprocess mechanical, fluid and spatial pose data from different sources to achieve the data consistency and reliability required for subsequent modeling and comparison.

[0084] The baseline reconstruction module includes:

[0085] The soil mechanics modeling unit is used to generate theoretical cutting loads based on preset soil strength parameters and the current drill bit depth.

[0086] The fluid dynamics modeling unit is used to generate theoretical slurry transport load based on preset slurry rheological parameters and measured slurry flow rate;

[0087] The benchmark solution unit, connected to the soil mechanics modeling unit and the fluid mechanics modeling unit, is used to generate theoretical benchmark parameters by combining theoretical cutting load, theoretical grouting load, measured spindle speed and measured drill bit lifting rate.

[0088] The theoretical benchmark parameters include theoretical torque and theoretical pump pressure; the preset soil strength parameters and preset slurry rheological parameters are entered by the operator before construction through the human-machine interface deployed on the construction site, or downloaded by the edge computing node from the engineering survey database and construction organization design database in the cloud via wireless network.

[0089] This embodiment provides a hierarchical modeling mechanism for generating dynamic theoretical benchmarks. Specifically, in the aforementioned scenario, if fixed torque and pump pressure limits are set based solely on experience, the influence of depth, flow rate, and rotational speed changes on the stress state will be ignored, leading to the same normal working condition being misidentified as abnormal at different depths. Therefore, this embodiment further divides the benchmark reconstruction module into a soil mechanics modeling unit, a fluid mechanics modeling unit, and a benchmark solution unit.

[0090] Specifically, the soil mechanics modeling unit generates theoretical cutting loads based on preset soil strength parameters and the current drill bit depth. The soil strength parameters here can be equivalent strength values ​​set in advance according to the exploration data. For clarity, it is assumed that the equivalent strength coefficient is 1.0 for depths of 0 to 5 meters, 1.3 for depths of 5 to 8 meters, and 1.6 for depths below 8 meters. If the current drilling depth increases from 4.9 meters to 5.1 meters, the theoretical cutting load can increase from 50 units to 65 units, reflecting the influence of strata stratification on cutting resistance.

[0091] The fluid dynamics modeling unit generates a theoretical slurry delivery load based on preset slurry rheological parameters and measured slurry flow rate. For example, if the theoretical slurry delivery load is 0.80 units when the flow rate is 110 liters / minute and 0.92 units when the flow rate is increased to 120 liters / minute, it means that when the pipeline is unobstructed and the slurry properties are stable, the pump pressure should rise smoothly with the change in flow rate, rather than producing a step change.

[0092] The benchmark calculation unit receives the two loads mentioned above and, combined with the spindle speed and drill bit jacking rate, generates theoretical torque and theoretical pump pressure. This can be achieved using lookup tables, piecewise functions, or lightweight numerical calculation methods. For ease of understanding, assuming a depth of 6.3 meters, a theoretical cutting load of 72 units, a theoretical grouting load of 0.96 units, a spindle speed of 38 rpm, and a drill bit jacking rate of 0.38 m / min, the calculated theoretical torque is 86 N·m and the theoretical pump pressure is 0.97 MPa. If the jacking rate is increased to 0.55 m / min while maintaining the same depth, the theoretical torque may rise to 93 N·m, indicating that this benchmark is not a static value but changes synchronously with the construction operation.

[0093] Under the anomaly protection mechanism, if the soil strength parameters are incomplete due to missing survey data, the soil mechanics modeling unit can first use the default layered template and add a low confidence mark to the output results; if the flow meter is inaccurate and the fluid mechanics modeling unit cannot obtain a reliable grouting load, the benchmark solution unit can degenerate to only output the theoretical torque and suspend the theoretical pump pressure from participating in the comparison; if the spindle speed and lifting rate inputs are abnormal at the same time, no new theoretical benchmark will be generated to avoid the erroneous benchmark from increasing the probability of misjudgment.

[0094] When the same pile is being constructed to the 6-7 meter range, the site is still in the normal pile formation stage. However, the operators temporarily increase the drill bit lifting speed based on process adjustments. If a fixed threshold is used, the increase in torque may be misjudged as an increase in ground hardness. In this embodiment, the benchmark calculation unit will synchronously incorporate the change in lifting speed into the theoretical torque calculation, so that the actual value and the theoretical value still maintain a small deviation, thereby avoiding misjudging human process adjustments as underground anomalies.

[0095] The purpose of this mechanism is to construct a theoretical reference trajectory that can be updated in conjunction with changes in depth, rotation speed, lifting speed and flow rate, thereby achieving more accurate subsequent residual extraction.

[0096] The parameter injection module includes:

[0097] The geological distortion injection unit is used to inject piecewise step increments relative to the preset soil cohesion benchmark value into the theoretical benchmark parameters in order to generate a stratigraphic abrupt change residual template.

[0098] The equipment distortion injection unit is used to inject preset pipe cross-sectional area attenuation parameters and drill bit equivalent diameter increase parameters to characterize the increase in cutting profile caused by drill bit adhesion and blockage into the theoretical reference parameters, so as to generate a blockage abnormality residual template.

[0099] The template library unit, connected to the geological distortion injection unit and the equipment distortion injection unit, is used to store the formation abrupt change residual template and the blockage anomaly residual template.

[0100] The theoretical residual template set is derived from the template library unit.

[0101] This embodiment provides a parameterized injection mechanism for typical anomaly mechanisms. Specifically, when only an ideal benchmark exists at the upper level, the system can only obtain the theoretical benchmark under normal operating conditions, and it is still difficult to determine which anomaly characteristic is more consistent after a deviation occurs. Especially in underground, invisible operating environments, hardening of the strata and blockage of pipes and plugs can both cause an increase in torque, which is difficult to distinguish based on the deviation amplitude alone. Therefore, this embodiment introduces a geological distortion injection unit and an equipment distortion injection unit to actively generate anomaly residual templates on the ideal benchmark.

[0102] Specifically, the geological distortion injection unit mainly simulates the segmented step changes in soil mechanical parameters. For ease of explanation, it is assumed that the theoretical torque corresponding to the preset soil cohesion benchmark value in the depth range of 6.0 meters to 6.4 meters is 80, 82, 84, 86, and 88 N·m. If a step increment of 30% cohesion is injected at 6.2 meters, the new simulated torque curves can be changed to 80, 82, 96, 99, and 101 N·m. After subtraction, the residual template of the abrupt change in formation is obtained as 0, 0, 12, 13, and 13 N·m. The corresponding theoretical pump pressure changes from 0.93, 0.95, 0.96, 0.97, and 0.98 to 0.93, 0.95, 0.97, 0.98, and 0.99 MPa, with residuals of only 0, 0, 0.01, 0.01, and 0.01 MPa. This forms a template characteristic of significantly increased torque and basically stable pump pressure.

[0103] The equipment distortion injection unit mainly simulates the attenuation of pipeline cross-sectional area and the increase of the equivalent diameter of the drill bit. The theoretical pump pressure reference is assumed to be 0.91, 0.93, 0.95, 0.96, and 0.98 MPa. If the injected cross-sectional area decreases by 20%, the simulated pump pressure can be changed to 0.94, 0.99, 1.08, 1.20, and 1.36 MPa. If the equivalent diameter of the injected drill bit increases by 10% simultaneously, the simulated torque can change from 80, 82, 84, 86, and 88 to 82, 86, 92, 101, and 115 N·m. After subtraction, the equipment abnormality residual template shows torque residuals of 2, 4, 8, 15, and 27 N·m and pump pressure residuals of 0.03, 0.06, 0.13, 0.24, and 0.38 MPa. Compared with formation abrupt changes, this model has a stronger pump pressure surge characteristic.

[0104] The template library unit is responsible for storing templates of different degrees, depths, and combinations. For example, an index can be created based on tags such as slight formation abrupt change, moderate formation abrupt change, severe formation abrupt change, slight blockage, moderate blockage, and severe blockage. Templates do not necessarily need to cover all continuous values; they only need to be discretized to form a searchable set. For example, the cohesion increment can be 10%, 20%, or 30%, the cross-sectional area decrease can be 10%, 20%, or 40%, and the drill bit equivalent diameter increment can be 5%, 10%, or 15%.

[0105] In cases where no coverage is available, if an anomaly occurs on-site that is not yet included in the template library, such as local grout leakage causing a drop in pump pressure while the torque remains essentially unchanged, the comparison results will show that the similarity to existing templates is lower than the preset matching lower threshold. In this case, the system will classify the residual fragment into the unidentified category and retain the original data for subsequent expansion of the template library. If mild formation hardening and mild blockage exist simultaneously at the same time, the parameter injection module can also superimpose two templates in a certain proportion to form a composite template, avoiding the need to establish only a single anomaly hypothesis.

[0106] After a certain piece of equipment has been used to construct multiple piles, slurry clumps gradually adhere to the outer surface of the drill bit, and slurry also forms on the inner wall of the grouting pipe, causing the diameter to shrink. At the edge nodes, a set of blocking templates with rapidly rising pump pressure and additional torque are constructed in advance using the equipment's distortion injection unit. When the field data enters the similarity comparison stage, such abnormal features can be quickly identified without misinterpreting them as encountering hard strata.

[0107] The purpose of this mechanism is to transform the abnormal mechanisms in expert experience into a set of calculable, callable, and comparable templates, thereby achieving the transition from discovering deviations to identifying the sources of deviations. Furthermore, to make the template calling conditions clearer, the template library unit can simultaneously record the template category, injected parameter combination, applicable depth range, and the corresponding benchmark operating condition range when the template is generated for each template. When calling a template, the residual calculation module prioritizes selecting a subset of templates corresponding to the current depth range, speed range, and flow range for comparison to avoid similarity distortion caused by direct comparison across operating conditions.

[0108] The residual calculation module includes:

[0109] The real residual generation unit is used to align and differentially process the measured drill rod torque and measured pump pressure with the theoretical torque and theoretical pump pressure at the same sampling time and / or the same depth interval through timestamp matching or depth index interpolation resampling, so as to generate a real residual sequence.

[0110] The temporal similarity comparison unit is connected to the real residual generation unit and the template library unit for storing the theoretical residual template set. It is used to perform dynamic time warping and / or cross-correlation operations on the real residual sequence and the theoretical residual template set to generate comparison results. When cross-correlation operation is used, the temporal similarity comparison unit extracts the normalized maximum peak value in the cross-correlation function sequence and uses it as the similarity score between the real residual sequence and the corresponding theoretical residual template.

[0111] This embodiment provides a mechanism for real-world residual extraction and temporal similarity comparison. Specifically, although a theoretical template has been established in the previous layer, if there are slight misalignments between the real-world data and the theoretical data in terms of time or depth, direct point-by-point subtraction may produce non-physical calculation deviations. For example, if the drill bit enters the hard layer 0.2 seconds later due to hydraulic lag, forcibly aligning it with the original sampling points will mistakenly treat the normal lag as an anomaly. Therefore, this embodiment introduces alignment difference and temporal similarity comparison in the residual calculation stage.

[0112] Specifically, the actual residual generation units can be aligned according to a unified sampling time or a unified depth interval. If depth interval alignment is used, assuming the 6.0-meter to 6.5-meter range is divided into five 0.1-meter intervals, the average torque and average pump pressure within each interval are selected and differ from the theoretical values. This can reduce the impact of instantaneous equipment vibration. Taking the above implementation scenario as an example, the measured average torque within the corresponding intervals of 6.0, 6.1, 6.2, 6.3, and 6.4 meters is 82. The measured average pump pressures are 0.92, 0.95, 0.96, 0.98, and 0.99 N·m, respectively. The theoretical pump pressures are 0.91, 0.93, 0.95, 0.96, 0.98, and 0.98 N·m, respectively. The actual residual pump pressures are 0.01, 0.02, 0.01, 0.02, and 0.01, respectively.

[0113] The temporal similarity comparison unit matches the actual residual sequence with each template in the template library. Taking dynamic time warping as an example, if the actual torque residual is [2,3,3,24,24], the formation template torque residual is [0,0,12,13,13], and the equipment template torque residual is [2,4,8,15,27], then by allowing local stretching and compression, we can obtain the result that the actual sequence is closer to the formation template in the pump pressure dimension and significantly deviates from the equipment template in the pump pressure dimension. For ease of explanation, we can calculate the similarity of the two-dimensional residuals separately and then sum them by weight. If the torque similarity weight is 0.6 and the pump pressure similarity weight is 0.4, then the overall similarity with the formation template can be 0.89, and the overall similarity with the equipment template can be 0.41.

[0114] If cross-correlation is used, the consistency of the shape of the actual residual and the template residual under different lags can be examined. For example, the maximum cross-correlation between the actual pump pressure residual and the equipment template pump pressure residual only occurs at a large lag and the peak value is low, indicating that their shapes are not consistent. Even if the actual torque residual and the formation template torque residual are considered to have a high correlation value, it can still be obtained, indicating that the underground formation change is more likely to be the main cause.

[0115] Under the anomaly protection mechanism, if the actual residual sequence length is insufficient, for example, only two sampling points, template matching will not be performed, and window data will continue to be accumulated first; if there are missing values ​​in the residual sequence with a proportion exceeding the set upper limit, interpolation can be used to fill in the missing values ​​when the proportion is less than or equal to the preset threshold, and the window will be deemed invalid when it is higher than the threshold; if the dynamic time warping and cross-correlation methods yield contradictory conclusions, the control judgment module can prioritize the one with higher confidence, or submit both as pending information without immediately executing high-risk actions;

[0116] During the drilling process from 6.2 meters to 6.5 meters, the hydraulic system of the drilling rig experienced a slight response delay, causing the torque to rise sharply one sampling point later than the theoretical template. When using point-by-point rigid comparison, this period might be mistakenly considered as mismatched. However, after using dynamic time warping, the edge nodes allow the sequence to be misaligned in a local range, and the formation characteristics of increased torque and stable pump pressure can still be accurately identified.

[0117] The purpose of this mechanism is to improve the tolerance of residual identification to on-site noise, time delay and beat fluctuation by aligning differential and elastic similarity comparison, thereby achieving more stable state judgment.

[0118] The control and judgment module is also used for:

[0119] When the comparison results indicate that the similarity score between the actual residual sequence and the formation abrupt change residual template is greater than or equal to the preset formation threshold, and the similarity score between the actual residual sequence and the blockage anomaly residual template is less than the preset blockage threshold, the formation change status is output.

[0120] When the comparison results indicate that the similarity score between the actual residual sequence and the blockage anomaly residual template is greater than or equal to the blockage threshold, and the similarity score between the actual residual sequence and the formation mutation residual template is less than the formation threshold, the output device is in an abnormal state.

[0121] When the comparison results indicate that the similarity scores of the actual residual sequence with the formation mutation residual template and with the blockage anomaly residual template are both less than the corresponding thresholds, the normal construction status is output.

[0122] When the comparison results indicate that the similarity scores between the actual residual sequence and the formation mutation residual template, as well as the similarity scores between the actual residual sequence and the blockage anomaly residual template, are both greater than or equal to the corresponding thresholds, the composite risk status is output.

[0123] This embodiment provides a state determination mechanism based on dual thresholds. Specifically, after completing the similarity comparison, if only the template category corresponding to the maximum similarity is taken as the final conclusion, although it is simple to implement, it is easy to cause hard decision distortion when the scores of the two templates are close or both are low. Therefore, this embodiment divides the results into four states: formation change, equipment abnormality, normal construction, and composite risk by setting formation threshold and blockage threshold.

[0124] Specifically, assuming the formation threshold is set to 0.80 and the blockage threshold is set to 0.78; if the formation similarity of a certain window is 0.88 and the blockage similarity is 0.42, the formation change status is output; if the formation similarity is 0.51 and the blockage similarity is 0.85, the equipment abnormal status is output; if both are 0.40 and 0.36 respectively, both below their respective thresholds, it indicates that the actual residual does not have significant formation abrupt change characteristics or typical blockage abnormality characteristics, and the normal construction status is output; if both are 0.83 and 0.81 respectively, both exceeding the thresholds, it indicates that the current residual contains both formation abrupt change and blockage characteristics, and the composite risk status is output.

[0125] This four-part judgment method solves a boundary problem in the previous solution: underground construction is not always a single-factor change; for example, when the drill bit enters the hard layer, the grout return channel begins to narrow, and at this time the torque and pump pressure will rise simultaneously; if classification and matching are based solely on the maximum similarity, the system may only identify one factor, resulting in incomplete control action coverage; therefore, this embodiment defines the situation where the similarity of two types reaches the threshold simultaneously as a composite risk state to retain more complete abnormal information.

[0126] In boundary stability control scenarios, if a judgment result frequently jumps within a continuous window, such as fluctuating around the threshold at 0.79, 0.81, 0.78, and 0.82, the control judgment module can introduce a minimum continuous window constraint. For example, it can require that the same state appears consecutively for 2 or 3 windows before formal confirmation. If the differences between the formation similarity and the blockage similarity and the threshold are both within the preset critical zone but do not exceed it, a normal construction-concern state can be output, and the sampling weight of subsequent windows can be increased. If the data quality label from a certain unit is lower than the preset quality benchmark, even if the similarity reaches the threshold, the output of a strong conclusion can be temporarily suspended.

[0127] When the same pile was constructed to a depth of 8.1 meters, the equipment first entered a relatively hard interlayer, and the grout circuit experienced a narrowing due to the accumulation of coagulated particles; the edge node obtained a formation similarity of 0.84 and 0.86 in two consecutive windows, and a blockage similarity of 0.79 and 0.82, respectively; since both types of scores continuously exceeded their respective thresholds, the system no longer made a single attribution, but output a composite risk status to provide a basis for subsequent joint protection;

[0128] The purpose of this mechanism is to replace single maximum value classification with a dual threshold and four-state decision structure, thereby achieving a more refined distinction between single-factor anomalies and multi-factor superposition anomalies.

[0129] Furthermore, the four-state determination in this embodiment is limited to the fulfillment of the preconditions, namely, the residual calculation module has completed the data alignment, differential processing and similarity comparison of the effective window, and the quality of the data participating in the determination meets the preset requirements; if the window belongs to the aforementioned invalid window, low confidence window or insufficient sample length window, then the window will not enter the four-state determination process of this embodiment, but will instead perform maintenance control, abnormal recording or wait for the next effective window to be supplemented for determination; in other words, the normal construction state in this embodiment only corresponds to the effective comparison results of two types of similarity scores that are both lower than the corresponding thresholds, and is not used to absorb invalid data or postpone the determination, so that the boundaries of the four states of formation change, equipment abnormality, normal construction and composite risk are kept clear and consistent; furthermore, in order to avoid confusion between the normal construction-concern state and the four-state determination boundary, in this embodiment, the normal construction-concern state is only used as an auxiliary marker for the priority of subsequent sampling and review, and its formal state category is still classified as normal construction state, and is not used as an independent working condition state to participate in the control action mapping;

[0130] The control execution module is also used for:

[0131] When the state assessment result indicates a change in the formation, a construction parameter control command is issued to reduce the drill bit lifting rate, increase the spindle allowable output torque limit set by the controller, and increase the grouting volume.

[0132] When the condition assessment result indicates that the equipment is in an abnormal state, an anti-blocking action instruction is issued to stop the pump, run the in-situ dry water wash, and trigger an alarm.

[0133] When the condition assessment result is a normal construction condition, a maintenance instruction to maintain the current parameters is issued;

[0134] When the condition assessment result is a complex risk state, a combined protection command of speed reduction, torque limitation, pump shutdown and alarm is issued.

[0135] This embodiment provides a mapping execution mechanism from state to control action. Specifically, state recognition alone is not enough to form a complete closed loop. If the control action is not adjusted according to the state category, the problem of correct recognition but incorrect handling will occur. Especially in the case of abnormal equipment, if the torque and grouting volume are increased according to the idea of ​​hardening the stratum, it may aggravate the blockage. Therefore, this embodiment clarifies the execution strategy corresponding to different states.

[0136] Specifically, when the state assessment result indicates a change in the formation, the control execution module sends a command to the lifting mechanism to reduce the lifting speed, for example, adjusting it from 0.42 m / min to 0.32 m / min; it sends a command to the main shaft drive mechanism to increase the allowable output torque limit, for example, from 100 N·m to 120 N·m; and it sends a command to the grouting pump control circuit to increase the grouting volume, for example, from 118 L / min to 126 L / min. This combination is suitable for maintaining cutting stability and pile compaction during hard layer transition.

[0137] When the status assessment result indicates an abnormal equipment state, the execution logic is the opposite of the previous state; the control execution module issues a pump stop command to prevent continued pressurization; the drill bit is kept in place or undergoes slight micro-movement idling, and the system is switched to water washing mode to flush the pipeline; at the same time, alarm information is sent to the cab terminal and monitoring platform; for example, if the pump pressure residual rises continuously and rapidly to above 0.25 MPa, and the blockage similarity reaches 0.87, the pump stop and in-situ idling water washing can be initiated within one control cycle;

[0138] When the status assessment result is a normal construction state, the control execution module issues a hold command to maintain the current rotation speed, lifting speed and grouting volume unchanged, so as to avoid new fluctuations caused by excessive adjustment.

[0139] When the status assessment result is a complex risk state, considering the coexistence of formation and equipment problems, the system no longer pursues construction efficiency, but focuses on safety protection. At this time, a joint protection command of speed reduction, torque limitation, pump stop and alarm can be issued at the same time. For example, the spindle speed is reduced from 38 rpm to 25 rpm, the torque limit is tightened from 120 Nm to 90 Nm, the grouting pump is stopped immediately, and the on-site personnel are required to check the drill bit and circuit.

[0140] Under the anomaly protection mechanism, if the controller does not confirm receipt of the instruction, the control execution module will resend it after a preset timeout; if the resending fails repeatedly, the system will enter the communication fault protection state and trigger the issuance of a shutdown or low-speed maintenance strategy; if the feedback from the actuator is inconsistent with the instruction, for example, if the pump is required to be stopped but the pressure continues to rise, it can be determined that there is an additional fault in the control link or the execution component, and the alarm level will be upgraded; if after control under the state of formation change, the similarity score does not decrease but instead turns to the blocking template, the system should allow the strategy to switch instead of locking the original conclusion.

[0141] In the scenario where the ground hardened at the aforementioned 6.3-meter depth, after the system issued a command to reduce the lifting speed, increase the torque limit, and increase the grouting volume, the actual torque residuals in the subsequent three windows gradually decreased from 24 and 24 to 18, 15, and 12, indicating that the adjustment was effective. Conversely, when another pile was being constructed to 7.8 meters, the system detected an equipment malfunction and executed a pump stoppage and water flushing. After the water flushing, the pump pressure residual decreased from 0.31 to 0.05, and the blockage similarity decreased synchronously, indicating that the handling action matched the malfunction mechanism.

[0142] The purpose of this mechanism is to translate the identification results into targeted control actions, avoiding the application of the same regulatory logic to different root causes, thereby achieving a balance between construction continuity and equipment safety.

[0143] Edge computing nodes serve as local processing centers, enabling the generation of theoretical benchmark parameters, theoretical residual template sets, and actual residual sequences in real time at the construction site. They also output comparison results and issue commands to reduce communication latency.

[0144] This embodiment provides a real-time processing mechanism on the field side. Specifically, if all data is uploaded to a remote server before modeling, template matching, and command return are completed, the round-trip latency will increase significantly in construction sites with unstable network conditions, and there may even be a problem that the working conditions have changed when the control results are returned. Therefore, this embodiment uses edge computing nodes as local processing centers, so that the main computing loop occurs on the construction site side.

[0145] Specifically, edge nodes can be implemented using industrial computers or embedded controllers and deployed in electrical control cabinets near the piling machine. After parameter acquisition, the data first enters the local memory buffer, and then the local algorithm completes the generation of theoretical benchmarks, template injection, calculation of actual residuals, and similarity comparison. For ease of explanation, if the fieldbus sampling period is 100 milliseconds, the local modeling and matching takes 40 milliseconds, and the instruction encapsulation and issuance takes 20 milliseconds, then the total closed-loop time from acquisition to execution can be controlled within one sampling period. In contrast, if relying on a remote center, the single-way network latency may reach more than 200 milliseconds.

[0146] This local processing method also solves another defect in the previous layer solution: the construction site network is often accompanied by signal blockage, mobile handover and bandwidth congestion; if the remote communication is interrupted momentarily, and the control logic relies entirely on the cloud, the construction equipment can only lose its intelligent adjustment capability or even be in an unprotected state; by leaving the core decision-making on the site, even if the external network is interrupted, the edge node can still operate independently.

[0147] Under the anomaly protection mechanism, if the load of the edge node exceeds the preset safety limit, causing a backlog in the processing queue, safety-related tasks, such as equipment anomaly identification and joint protection command issuance, can be prioritized, while non-real-time tasks such as template library updates can be postponed. If the local clock and the controller clock deviate from the allowable synchronization range, time synchronization will be performed first, and then closed-loop control will be started. If the edge node fails and stops, the construction equipment controller can switch to a basic safety mode, such as fixed low speed, pressure-limited operation, or manual takeover.

[0148] At the construction site after rain, 4G or 5G signals weaken temporarily, and the upload delay of the monitoring center increases to more than 1 second. Since the judgment and control in this embodiment are completed within the edge node, when the drill bit experiences abnormal pump pressure surge at 8.1 meters, the system can still immediately stop the pump and issue an alarm without waiting for the results to be returned from the remote platform. The remote platform only undertakes the functions of subsequent recording, statistics, and operation and maintenance analysis.

[0149] The purpose of this mechanism is to move the time-sensitive modeling, comparison, and control actions forward to the on-site side, thereby achieving low-latency and network-fluid-resistant construction process control.

[0150] Example 2:

[0151] The system also includes:

[0152] The adaptive update module communicates with the control judgment module and the edge computing node. It is used to correct the model parameters used to generate theoretical reference parameters in the ideal fluid-structure interaction model based on the actual residual sequence within the current construction cycle when the state assessment result is a normal construction state.

[0153] Used to freeze model parameter updates when the condition assessment result is an abnormal equipment condition or a complex risk condition;

[0154] Used to update the range of values ​​for geological distortion parameters when the state assessment result indicates a state of stratigraphic change.

[0155] This embodiment provides an adaptive update mechanism coupled with state. Specifically, the aforementioned implementation methods can achieve real-time identification and control. However, if the ideal fluid-structure interaction model keeps its initial parameters unchanged for a long time, the theoretical benchmark may gradually deviate from the long-term real level in the field due to equipment wear, fine-tuning of slurry ratio, and seasonal changes in soil moisture content. On the other hand, if the model is updated uncontrolled under abnormal conditions, the fault characteristics will be incorrectly absorbed into the normal model. Therefore, this embodiment adopts a differentiated strategy for model updating according to different states.

[0156] Specifically, when the status assessment result indicates normal construction status, the adaptive update module allows for the correction of model parameters based on the actual residual sequence within the current construction cycle. Here, the construction cycle can be a fixed depth interval, such as 0.5 meters per cycle. For clarity, assuming that the average actual torque residual values ​​are +2, +3, and +2 N·m in three consecutive normal cycles, and the average actual pump pressure residual values ​​are +0.01, +0.01, and +0.02 MPa, this indicates that the theoretical baseline is slightly low overall. In this case, the adaptive update module can increase the soil cutting coefficient by a small step, for example, by 2%, and increase the grouting resistance coefficient by 1%. The updated theoretical torque and theoretical pump pressure will be closer to the long-term actual levels on site.

[0157] When the status assessment result is an abnormal equipment status or a complex risk status, the adaptive update module freezes the model parameters. The purpose is to avoid writing abnormal factors such as blockage, drill bit mud adhesion, and local instability into the ideal model. For example, if an equipment abnormality within a certain window causes the pump pressure residual to reach +0.30 MPa, if updates are still allowed, the grouting resistance coefficient may increase abnormally beyond the preset step size, causing the system to underestimate the new abnormality after the fault is cleared.

[0158] When the state assessment result indicates a change in the strata, this embodiment does not immediately modify the core parameters of the ideal model, but updates the range of values ​​for the geological distortion parameters; that is, it treats such deviations as environmental changes rather than changes in the equipment baseline; for example, if moderate strata abrupt changes are repeatedly detected in the 6-meter to 7-meter range of multiple pile locations, the range of geological distortion increments in this depth zone can be expanded from the original 10% to 20% to 15% to 30%, so that the subsequent templates are closer to the new strata characteristics of the site.

[0159] Under the anomaly protection mechanism, if the duration of normal construction status is less than the preset lower limit of stable period, such as less than two periods, the model will not be updated to avoid correcting parameters based on accidental noise; if the absolute value of the residual in the next window increases after the update, the update will be canceled and the previous version of parameters will be restored; if the geological change occurs only once and the duration is less than the minimum effective identification time, it will be recorded as a candidate change without expanding the range of geological distortion; if the model parameter update frequency exceeds the preset update frequency limit, an update time interval limit can be added to prevent the model from following short-term fluctuations.

[0160] After constructing 20 mixing piles at the same site, due to the rise in temperature and the decrease in slurry viscosity within a preset range, the system observed that the theoretical pump pressure was consistently higher than the measured pump pressure in multiple normal state windows, and the difference was within the preset allowable range. Based on this, the adaptive update module adjusted the slurry delivery resistance-related parameters according to the preset correction step size, so that the subsequent ideal pump pressure was closer to the site. When a significant blockage occurred at another time, the system froze the update to ensure that the fault characteristics were not absorbed into the normal model.

[0161] The purpose of this mechanism is to ensure that, under the premise of preventing abnormal features from being introduced into the normal model, the theoretical benchmark can converge and be corrected at a preset rate as the construction environment and equipment change over a long period of time, thereby achieving long-term operational stability.

[0162] The construction equipment is cement mixing pile drilling and grouting construction equipment;

[0163] Among them, the mechanical timing parameters are the operating data of the actuator of the cement mixing pile drilling and grouting construction equipment, the fluid timing parameters are the operating data of the grouting circuit of the cement mixing pile drilling and grouting construction equipment, and the state assessment results are used to characterize the geological change state, abnormal equipment state, normal construction state or complex risk state of the cement mixing pile drilling and grouting construction equipment during underground operations.

[0164] This embodiment provides an application method for specifically implementing the aforementioned mechanism in cement mixing pile drilling and grouting construction equipment; specifically, the construction equipment in this embodiment is a mixing pile machine equipped with a main shaft drive mechanism, a lifting mechanism and a grouting pump circuit, and its operation process includes aligning the pile position, drilling and sinking, mixing and cutting, grouting and pile formation, lifting and re-mixing and exiting the site; the aforementioned mechanical timing parameters, fluid timing parameters and state evaluation results are all defined around the underground operation process of this equipment;

[0165] Specifically, the mechanical timing parameters correspond to the operating data of the actuator, including at least the drill rod torque output by the spindle drive link, the spindle speed, and the drill bit lifting and lowering rate driven by the hoisting mechanism; the fluid timing parameters correspond to the operating data of the slurry delivery circuit, including at least the pump outlet pressure and the instantaneous flow rate of the slurry; the spatial position parameters are provided by the depth detection device and the inclination detection device to describe the actual position and attitude of the drill bit underground; the comprehensive analysis results of these parameters by the system are used to characterize whether the construction equipment belongs to the category of stratum change, equipment abnormality, normal construction, or combined risk in the current underground operation stage;

[0166] Taking the complete process as an example, in the initial 0-2 meter range when the drill bit begins to sink, the shear strength of the soil layer is lower than the preset hard layer threshold, and the system mostly outputs a normal construction status. In the 2-5 meter range, if a local hard interlayer is encountered, the system may output a geological change status and automatically reduce the lifting speed. In the 5-8 meter range, if slurry accumulation and diameter reduction occur in the slurry delivery circuit, the system can output an abnormal equipment status and execute pump shutdown and water flushing. If blockage is superimposed in the deep hard layer, a composite risk status is output and joint protection is triggered. In this way, the status assessment results are directly mapped to an online representation of the operation process of this type of equipment.

[0167] This specific application method solves the problem of algorithms being detached from equipment objects; that is to say, the aforementioned models, templates and control logic are not abstract data processing flows, but are designed for the actual actuators and grouting systems of the mixing pile machine, and the inputs and outputs can be implemented at the equipment level.

[0168] Under the anomaly protection mechanism, if different construction equipment models lead to differences in sensor installation locations or controller communication protocols, data conversion and command mapping can be completed through the equipment adaptation layer without changing the main state recognition process. If a piece of equipment does not have a tilt sensor, the system can operate its basic functions without this parameter, but the theoretical benchmark accuracy related to attitude compensation will be downgraded. If the construction process changes from single grouting to segmented re-mixing, different model parameter groups and template library subsets can be switched according to the process stage.

[0169] In a municipal road soft soil foundation treatment project, a cement mixing pile drilling and grouting construction equipment was constructing a 12-meter-long pile according to the design requirements. When the 9th pile was constructed to a depth of 6.3 meters, the equipment was identified as being in a state of ground change, and the system automatically reduced speed and increased torque. When the 12th pile was constructed to a depth of 7.8 meters, the equipment was identified as being in an abnormal state, and the system immediately stopped the pump and performed in-situ idle water washing. Both states directly reflect the actual working conditions of the equipment during underground operations, and both can be executed by the equipment controller.

[0170] The purpose of this application method is to clarify the correspondence between the aforementioned system and the cement mixing pile drilling and grouting construction equipment, so that the source of input data, state semantics and control action objects all have clear engineering landing points;

[0171] To further clarify, the status assessment results in this embodiment specifically refer to four types of results used to characterize the underground operation conditions: stratum change status, equipment abnormal status, normal construction status, and composite risk status. If the preceding processing stage only triggers a temporary suspension due to data packet loss, timestamp misalignment, depth bounce, or insufficient samples, this type of information belongs to data quality management or window validity management information and is not part of the status assessment results in this embodiment. Therefore, the semantics of the operating condition status received by the equipment layer remain consistent, while data abnormality handling is used as independent auxiliary information in conjunction with it.

[0172] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An adaptive control system for construction parameters of cement mixing piles based on edge computing, comprising edge computing nodes deployed at the construction site, characterized in that, The edge computing nodes include: The parameter acquisition module collects mechanical time-series parameters from the construction equipment, including measured drill rod torque, measured spindle speed, and measured drill bit lifting rate; fluid time-series parameters, including measured pump pressure and measured slurry flow rate; and spatial position parameters, including the current drill bit depth. The benchmark reconstruction module generates theoretical benchmark parameters containing theoretical torque and theoretical pump pressure based on a preset ideal fluid-structure interaction model and spatial position parameters. This model characterizes the coupling relationship between the drilling cutting load, which is the theoretical torque, and the grouting load, which is the theoretical pump pressure, as they change with the current drill bit depth, measured spindle speed, measured drill bit lifting rate, and measured grout flow rate. The parameter injection module injects geological distortion parameters that characterize abrupt changes in formation mechanical parameters and equipment distortion parameters that characterize blockages in the grouting circuit or abnormalities in the drill bit structure into the theoretical reference parameters. It also calculates the difference between the simulation reference curve after the distortion parameters are injected and the original theoretical reference parameters, thereby generating a set of theoretical residual templates in the form of data vectors with specific time steps or depth steps. The residual calculation module, according to a unified sampling period and / or a unified depth range, aligns the measured drill rod torque and measured pump pressure with the theoretical torque and theoretical pump pressure respectively to generate a real residual sequence, and compares it with the theoretical residual template set to output the comparison result; The control judgment module determines the status assessment result based on the comparison result. The status assessment result is selected from one of the following: normal construction status, geological change status, equipment abnormal status, or composite risk status. The control and execution module communicates with the construction equipment controller and issues construction parameter control commands or anti-blocking commands to the controller based on the status assessment results, so as to control the main shaft drive mechanism, lifting mechanism and grouting pump control circuit of the construction equipment.

2. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 1, characterized in that, The parameter acquisition module includes: The mechanical parameter acquisition unit is used to acquire measured drill rod torque, measured spindle speed and measured drill bit lifting rate. The fluid parameter acquisition unit is used to acquire measured pump pressure and measured slurry flow rate; The position parameter acquisition unit is used to acquire the current drill bit depth and inclination angle; The mechanical timing parameters are derived from the mechanical parameter acquisition unit, the fluid timing parameters are derived from the fluid parameter acquisition unit, and the spatial position parameters are derived from the position parameter acquisition unit.

3. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 2, characterized in that, The benchmark reconstruction module includes: The soil mechanics modeling unit is used to generate theoretical cutting loads based on preset soil strength parameters and the current drill bit depth. The fluid dynamics modeling unit is used to generate theoretical slurry delivery load based on preset slurry rheological parameters and the measured slurry flow rate; The benchmark calculation unit, connected to the soil mechanics modeling unit and the fluid mechanics modeling unit, is used to generate the theoretical benchmark parameters by combining the theoretical cutting load, the theoretical grouting load, the measured spindle speed, and the measured drill bit lifting rate. The theoretical benchmark parameters include the theoretical torque and the theoretical pump pressure; the preset soil strength parameters and the preset slurry rheological parameters are entered by the operator before construction through a human-machine interface deployed at the construction site, or downloaded by the edge computing node from the engineering survey database and construction organization design database in the cloud via a wireless network.

4. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 3, characterized in that, The parameter injection module includes: The geological distortion injection unit is used to inject a segmented step increment relative to the preset soil cohesion benchmark value into the theoretical benchmark parameter in order to generate a stratigraphic abrupt change residual template. The equipment distortion injection unit is used to inject preset pipe cross-sectional area attenuation parameters and drill bit equivalent diameter increase parameters used to characterize the increase in cutting profile caused by drill bit adhesion blockage into the theoretical reference parameters, so as to generate a blockage abnormality residual template. The template library unit, connected to the geological distortion injection unit and the equipment distortion injection unit, is used to store the formation abrupt change residual template and the blockage anomaly residual template. The theoretical residual template set is derived from the template library unit.

5. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 4, characterized in that, The residual calculation module includes: The real residual generation unit is used to align and differentially process the measured drill rod torque and the measured pump pressure with the theoretical torque and the theoretical pump pressure at the same sampling time and / or the same depth interval through timestamp matching or depth index interpolation resampling, so as to generate the real residual sequence. A temporal similarity comparison unit is connected to the actual residual generation unit and the template library unit for storing the theoretical residual template set. It is used to perform dynamic time warping and / or cross-correlation operations on the actual residual sequence and the theoretical residual template set to generate the comparison result. When cross-correlation operation is used, the temporal similarity comparison unit extracts the normalized maximum peak value in the cross-correlation function sequence and uses it as the similarity score between the actual residual sequence and the corresponding theoretical residual template.

6. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 5, characterized in that, The control judgment module is also used for: When the comparison result indicates that the similarity score between the actual residual sequence and the formation mutation residual template is greater than or equal to a preset formation threshold, and the similarity score between the actual residual sequence and the formation mutation residual template is less than a preset blockage threshold, the formation change status is output. When the comparison result indicates that the similarity score between the actual residual sequence and the blockage anomaly residual template is greater than or equal to the blockage threshold, and the similarity score between the actual residual sequence and the formation mutation residual template is less than the formation threshold, the device is output as an abnormal state. When the comparison results indicate that the similarity score between the actual residual sequence and the formation mutation residual template and the similarity score between the actual residual sequence and the blockage anomaly residual template are both less than the corresponding threshold, the normal construction status is output. When the comparison results indicate that the similarity score between the actual residual sequence and the formation mutation residual template, as well as the similarity score between the actual residual sequence and the blockage anomaly residual template, are both greater than or equal to the corresponding thresholds, a composite risk status is output.

7. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 6, characterized in that, The control execution module is also used for: When the state assessment result indicates a change in the formation state, a construction parameter control command is issued to reduce the drill bit lifting rate, increase the spindle allowable output torque limit set by the controller, and increase the grouting volume. When the status assessment result indicates that the equipment is in an abnormal state, an anti-blocking control command is issued to stop the pump, run the in-situ dry water wash, and trigger an alarm. When the status assessment result is a normal construction status, a maintenance command to maintain the current parameters is issued; When the status assessment result is a composite risk state, a combined protection command of speed reduction, torque limitation, pump shutdown and alarm is issued.

8. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 7, characterized in that, The edge computing node serves as a local processing center, used to generate the theoretical benchmark parameters, the theoretical residual template set, and the actual residual sequence in real time at the construction site, and output the comparison results and issue the instructions to reduce communication latency.

9. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 8, characterized in that, The system also includes: An adaptive update module, which is communicatively connected to the control judgment module and the edge computing node, is used to correct the model parameters used to generate the theoretical benchmark parameters in the ideal fluid-structure interaction model based on the actual residual sequence within the current construction cycle when the state evaluation result is a normal construction state. Used to freeze the model parameter update when the status assessment result is an abnormal equipment status or a complex risk status; Used to update the value range of the geological distortion parameter when the state assessment result is a state of stratigraphic change.

10. The edge computing-based adaptive control system for cement mixing pile construction parameters as described in claim 9, characterized in that, The construction equipment is a cement mixing pile drilling and grouting construction equipment. The mechanical timing parameters are the operating data of the actuator of the cement mixing pile drilling and grouting construction equipment, the fluid timing parameters are the operating data of the grouting circuit of the cement mixing pile drilling and grouting construction equipment, and the state assessment results are used to characterize the geological change state, abnormal equipment state, normal construction state, or complex risk state of the cement mixing pile drilling and grouting construction equipment during underground operations.

Images