Optimization control method and system for traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction

Abstract

The present invention discloses a traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction optimization control method and system, which specifically relates to the technical field of underground engineering construction, including: collecting consistency time curves and thickness change trajectories during traction and transportation; analyzing consistency gradient and thickness fluctuation indicators, and entering the analysis stage when they exceed preset thresholds; constructing a flow homogeneity index and a backfill uniformity index; importing the two indices as input into a machine learning model trained with construction data, and outputting a coordination control coefficient; executing a traction adjustment strategy based on the coefficient and its changing trend to achieve dynamic optimization of the backfill construction status; the present invention realizes full-process status monitoring of material transportation and backfilling by continuously collecting consistency time curves and thickness change trajectories; constructing a flow homogeneity index and a backfill uniformity index to quantitatively evaluate structural stability and thickness control level; driving dynamic adjustment of the traction rhythm based on the coordination control coefficient to form a data-driven intelligent construction closed-loop control system.

Description

Technical Field

[0001] The present invention relates to the technical field of underground engineering construction, and more particularly to a method and system for optimizing control of backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit. Background Art

[0002] In underground construction projects such as urban rail transit, municipal pipeline corridors, deep foundation pit support, and pipeline maintenance, rapid, stable, and environmentally friendly backfilling of foundation pits has become a crucial technical step in the construction process. Fluidized solidified soil, a material that combines fluidity with solidified strength, is widely used in foundation pit backfill, trench backfill, and infill repair. This material, typically formulated in specific proportions of cement, fly ash, solid waste admixtures, additives, and water, exhibits excellent pumpability, self-compacting properties, and early strength growth, making it a key development direction for environmentally friendly geotechnical materials.

[0003] During application, fluidized soil is typically transported from a centralized preparation area to the construction site via traction pumping, where it is then backfilled into the target area using paving equipment or manual methods. Due to the generally long traction transport routes and the complex terrain of the construction area, fluidized soil is susceptible to factors such as gravity settling, water separation, and changes in viscous resistance in long-distance pipelines. This can cause unpredictable changes in its internal consistency over time and location, impacting its structural stability and subsequent paving results.

[0004] Existing traction conveying control methods typically rely on fixed-rate propulsion or manual adjustments based on experience, lacking the ability to perceive the actual material state. This is especially true when material consistency fluctuates dramatically or construction on-site is inconsistent, making it difficult to detect conveying anomalies in a timely manner. This can lead to material accumulation at the end of the pipe, uneven backfill thickness, and loose local structures, seriously impacting construction quality and schedule stability.

[0005] At the same time, controlling the backfill thickness on-site presents significant technical difficulties. In actual construction, due to factors such as terrain undulations, equipment accuracy, and human intervention, the backfill thickness fluctuates spatially, with high frequency and rapid changes. Existing quality assessment methods rely primarily on manual spot checks or post-inspection feedback, lacking a real-time feedback mechanism and making it difficult to form an effective basis for dynamic adjustments.

[0006] Furthermore, a unified, quantifiable standard has yet to be established in the current construction process to measure the degree of coordination between "material delivery status" and "on-site backfill behavior." Most control methods lack a data-driven mechanism, making it impossible to achieve closed-loop regulation between material status, thickness changes, and construction cadence. This lack of data leads to frequent "too fast" or "too slow" construction cadences, which not only affects material utilization but can also pose structural safety risks. Therefore, a method and system for optimizing the control of traction-type, environmentally friendly, fluidized solidified soil foundation pit backfill construction is proposed to address the aforementioned issues. Summary of the Invention

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] The optimization control method for the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit includes the following steps:

[0009] Continuous data collection is performed during the traction and transportation of fluidized solidified soil. The data is divided into two categories: one is the consistency-time curve of the material consistency changing with time along the transportation path, and the other is the thickness variation trajectory of the backfill thickness in the paving area changing with position and time. These two data serve as the parameter basis for subsequent processing.

[0010] After completing data collection, the gradient value of the consistency time curve in the current construction cycle is calculated, and the maximum and minimum value differences of the thickness change trajectory are extracted and the deviation trend is analyzed. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met and the next step of analysis and processing is entered;

[0011] After triggering the analysis, the consistency-time curve is used to construct the flow homogeneity index, which is used to determine the structural stability of the material during long-distance transportation. The thickness change trajectory is used to construct the backfill uniformity index, which reflects the thickness control level of the paving area.

[0012] The flow homogeneity index and backfill uniformity index are combined as inputs into a machine learning model trained on construction data. The model then calculates and outputs a coordination control coefficient that reflects the degree of coordination between the current material delivery status and the on-site backfill behavior.

[0013] According to the value of the coordination control coefficient and its changing trend during the continuous construction process, the corresponding traction adjustment strategy is executed.

[0014] In a preferred embodiment, the consistency time curve continuously obtains the real-time change value of the material consistency through evenly spaced measurement points during the data acquisition process, and uses interpolation processing to construct a continuous curve, wherein each consistency change value corresponds to a unique combination of conveying path position and time node, which is used for the subsequent calculation of the gradient value of the consistency time curve. At the same time, the consistency time curve is limited to the construction effective path interval range, the construction effective path interval range is consistent with the traction conveying direction, and the curve continuity is maintained between adjacent conveying path segments;

[0015] During the data collection process, the thickness change trajectory obtains continuous thickness measurements at equidistant measurement points distributed along the construction direction in the paving area, and double-labels the position and time of each measurement point data in combination with time information to generate a complete thickness change trajectory. The thickness change trajectory maintains data continuity within a single backfill operation cycle and is comparable between adjacent operation cycles.

[0016] In a preferred embodiment, after completing data collection, the gradient value of the consistency time curve is obtained by calculating the slope between continuous consistency change values, and the overall consistency stability is measured by the average slope change rate during the current construction cycle. If the average slope change rate exceeds the corresponding construction preset threshold, that is, the gradient change threshold, it is judged that the consistency time curve does not meet the construction stability requirements of the current operation section.

[0017] In a preferred embodiment, the two independent data items contained in the thickness change trajectory are processed separately. One is the maximum and minimum value difference calculated by the extreme value difference in the measured thickness values ​​of the measuring points within the same construction period, and the other is the number of trend change direction switches of the thickness values ​​of consecutive measuring points in the position sequence. The maximum and minimum value difference is used to reflect the fluctuation amplitude of the thickness data at the numerical level, and the number of direction switches is used to reflect the fluctuation frequency of the thickness data in the spatial distribution. Respective construction preset thresholds, namely the numerical difference threshold and the switching number threshold, are set. When the maximum and minimum value difference exceeds the numerical difference threshold, or the number of direction switches exceeds the switching number threshold, it is judged that the thickness change trajectory does not meet the construction stability requirements of the current working section.

[0018] In a preferred embodiment, in the process of constructing the flow homogeneity index using the consistency time curve, the consistency time curve in the current construction period is first divided into multiple adjacent sub-segments at fixed time intervals. All consistency change values ​​are extracted in each sub-segment, and the difference between the maximum and minimum consistency change values ​​in the sub-segment is calculated and recorded as the consistency fluctuation amplitude of the sub-segment. The consistency fluctuation amplitudes of all sub-segments are further arranged in sequence to construct a complete fluctuation sequence. Then, based on the fluctuation sequence, the sign change frequency of the differences between each pair is calculated and used as the disturbance direction index of the fluctuation sequence. The viscosity fluctuation amplitudes of all sub-segments are then multiplied by their corresponding disturbance direction index to obtain a set of directional disturbance correction values. The difference between the maximum and minimum absolute values ​​in the correction value sequence is used as the consistency consistency response value. Finally, the consistency consistency response value is divided by the total number of sub-segments to define the flow homogeneity index in the current construction period. This index is used to judge the degree of fluctuation concentration and directional continuity of the current consistency time curve under segment division.

[0019] In a preferred embodiment, in the process of constructing the backfill uniformity index using the thickness variation trajectory, first, an equally spaced position sequence is selected on the thickness variation trajectory of the current construction period, and the corresponding thickness measurement value is extracted for each position node to form a thickness position pair sequence; the thickness position pair sequence is divided into two continuous equal-length regions, the average value of the thickness measurement in each region is calculated respectively, and the square of the difference between the average values ​​of the two regions is calculated as the regional difference factor, and then each region is divided into subgroups again, and the maximum and minimum difference of the internal thickness values ​​of each subgroup is calculated, which is recorded as the internal fluctuation amplitude; the fluctuation amplitudes of all subgroups in the two regions are arranged in sequence, and the average square deviation of all fluctuation amplitudes is calculated as the local fluctuation factor; the regional difference factor is multiplied by the local fluctuation factor to obtain the fluctuation structure coupling value of the current construction period, and then the ratio of the fluctuation structure coupling value to the square root of the regional length of the thickness position pair sequence is defined as the backfill uniformity index. This index is used to characterize the degree of consistency of the thickness variation trajectory at the local and global scales, and has a basis for numerical comparison.

[0020] In a preferred embodiment, the machine learning model is any one of the following three: a convolutional neural network model, a recurrent neural network model, or a gradient boosting tree model. All three models use the flow homogeneity index and the backfill uniformity index as input, and use construction sample data as the training basis to output the coordination control coefficient.

[0021] In a preferred embodiment, after obtaining the coordination control coefficient, a traction adjustment strategy is executed based on the value of the coordination control coefficient and its changing trend formed in continuous construction cycles. When the value fluctuation range of the coordination control coefficient in multiple continuous construction cycles is within a preset stable range, the existing conveying rhythm and traction advancement rhythm are maintained unchanged; when the coordination control coefficient shows a continuous increasing trend, it is determined that the material structure state and the backfill thickness fluctuation are both tending to be unstable, and a rhythm slowing strategy is executed in the control process to reduce the conveying rate by extending the time used for traction per unit length; when the coordination control coefficient shows a continuous decreasing trend, it is determined that the coordination between conveying and backfilling is improved, and a rhythm acceleration strategy is executed.

[0022] In a preferred embodiment, the traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction optimization control system includes:

[0023] The data acquisition unit is used to continuously collect data during the traction and transportation of fluidized solidified soil. The data is divided into two categories: one is the consistency time curve formed by the change of material consistency over time along the transportation path, and the other is the thickness variation trajectory formed by the change of backfill thickness in the paving area as it changes with position and time. Both types of data serve as the parameter basis for subsequent processing;

[0024] The state assessment unit is used to calculate the gradient value of the consistency time curve within the current construction cycle after completing data collection, extract the maximum and minimum value differences of the thickness change trajectory, and analyze the deviation trend. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met, and the processing flow is advanced to subsequent analysis;

[0025] The feature extraction unit is used to construct a flow homogeneity index based on the consistency time curve after triggering analysis, which is used to determine the structural stability of the material during long-distance transportation; and a backfill uniformity index based on the thickness change trajectory, which is used to reflect the thickness control level of the paving area;

[0026] The model building unit is used to take the flow homogeneity index and backfill uniformity index as joint inputs and import them into the machine learning model trained based on construction data. The output is a coordination control coefficient that reflects the degree of coordination between the current material delivery status and the on-site construction backfill behavior. This coefficient is a continuous value and is used for subsequent control path judgment.

[0027] The strategy execution unit is used to execute the corresponding traction adjustment strategy according to the value of the coordination control coefficient and its changing trend in the continuous construction cycle.

[0028] Technical effects and advantages of the present invention:

[0029] This invention enables full-process monitoring and coordinated assessment of material delivery and construction backfill status, enhancing the controllability and responsiveness of the construction process. By continuously collecting data during the traction and transport of fluidized solidified soil and categorizing the data into two types: consistency-time curves and thickness-change trajectories, this system structures previously fragmented and scattered construction monitoring information into a dynamic parameter system with temporal continuity and spatial characteristics. The consistency-time curve captures the material's consistency evolution at different locations and time points along the delivery path, while the thickness-change trajectories record the thickness variations within the paving area over time and position. By continuously acquiring full-process data from both curves, it is possible to accurately characterize the structural state trends throughout the material's transport from source to placement, as well as their actual feedback during backfill control. Compared to existing construction status assessment methods that rely primarily on point-by-point sampling and empirical observation, this invention establishes a full-process structural analysis framework based on real-world construction behavior at the source data collection stage. This provides greater data continuity and timely response, provides a highly consistent numerical basis for subsequent assessment of construction stability, and improves the controllability of the overall process.

[0030] This invention establishes a dual judgment mechanism for structural stability and paving uniformity, enhancing the accuracy and foresight of construction stability assessments. After completing data collection, the invention establishes a dynamic stability judgment mechanism based on time series data by calculating the gradient of the consistency-time curve within the current construction cycle and the difference between the maximum and minimum values ​​in the thickness variation trajectory, as well as the offset trend. When any indicator exceeds a preset construction threshold, the construction stability of the current operation section is deemed to be unsatisfactory, and the next analysis step is entered. This approach not only enables real-time judgment of conveying status and backfill results, but also, by introducing a threshold mechanism, transforms the judgment process from vague manual experience to precise quantitative judgment. Furthermore, a flow homogeneity index and a backfill uniformity index are constructed. The former reflects the structural integrity and flow equilibrium of the material during long-distance conveying, while the latter reflects the degree of thickness control dispersion in the paving area during the backfill process. This dual-index system not only ensures complete coverage of the judgment dimensions, but also enhances the ability to identify the impact of local data anomalies on the overall status, providing a structured data foundation for high-quality input to machine learning models. In this way, the invention shifts construction assessment from coarse-grained to refined, making stability analysis more foresighted and accurate.

[0031] This invention implements a data-driven closed-loop control logic, enabling adaptive adjustment of the construction traction rhythm and intelligent optimization of the construction process. The flow homogeneity index and backfill uniformity index are combined as inputs into a machine learning model trained with historical construction data to generate a coordination control coefficient. Based on the coefficient's value and its changing trend over consecutive construction cycles, corresponding traction adjustment strategies are implemented. This mechanism transcends traditional construction methods that rely on fixed settings or subjective human experience to control the traction rhythm. Instead, it implements a closed-loop response structure that automatically derives control strategies based on model judgment based on multi-source construction status data. The coordination control coefficient serves as a unified evaluation value in the system that reflects the degree of coordination between material delivery status and on-site construction backfill behavior, establishing a complete information path from perception to judgment to decision-making. Based on the value of this coefficient, the original delivery rhythm can be maintained, while a pacing reduction strategy can be implemented when coordination deteriorates or an pacing acceleration strategy can be implemented when coordination improves. This enables the construction control system to adapt to changing site environments and reduces the impact of human misjudgment on construction quality. Furthermore, this mechanism provides the foundation for future expansion to include more control parameters, ensuring traction control with excellent scalability and engineering adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0033] Figure 1 This is a schematic diagram of the optimization control method for the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit in the present invention.

[0034] Figure 2 This is a schematic diagram of the optimization control system for the traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction in the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0036] Reference Figure 1 - Figure 2 The following examples were obtained:

[0037] Example 1: A method for optimizing and controlling the construction of a foundation pit backfill using traction-type, environmentally friendly fluidized solidified soil, including the following steps: Continuously collecting data during the traction and transportation of the fluidized solidified soil. The data is divided into two categories: a consistency-time curve showing the material consistency changing over time along the conveying path, and a thickness variation trajectory showing the backfill thickness in the paving area changing with position and time. These two categories serve as parameter foundations for subsequent processing. This step is used to obtain dynamic information about the material transportation and construction processes. The consistency-time curve reflects the physical state changes experienced by the fluidized solidified soil during traction and transportation, exhibiting both continuity and temporal sequence. The thickness variation trajectory reflects the distribution of thickness data within the paving construction area, which varies with the work position and time. Both types of data are acquired on-site, exhibiting a temporal and spatial correspondence. These data are used to construct parameter expression systems for the material conveying chain and the construction backfill behavior, respectively, providing a complete raw input foundation for subsequent evaluation, calculation, and control.

[0038] After completing data collection, the gradient value of the consistency time curve within the current construction cycle is calculated. The maximum and minimum value differences of the thickness change trajectory are extracted and the deviation trend is analyzed. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met and the next step of analysis and processing is entered. This step is a basic preliminary processing and quantitative evaluation of the collected data. By calculating the gradient value of the consistency time curve, the speed of the material state change on the traction path can be captured to identify whether there are discontinuities or violent fluctuations. The maximum and minimum value differences of the thickness change trajectory are extracted and trend analysis is performed to determine whether the paving results have local fluctuations or continuous deviation problems. The set construction preset threshold serves as the project quality control standard. If the detection value exceeds the range, it indicates that the construction status is unstable and the analysis and processing link is entered to obtain higher-level parameters.

[0039] After trigger analysis, the consistency-time curve is used to construct a flow homogeneity index, which is used to determine the structural stability of the material during long-distance transportation. The thickness variation trajectory is used to construct a backfill uniformity index, which is used to reflect the thickness control level of the paving area. This step elevates the preliminary data structure to a highly expressive indicator form. The flow homogeneity index constructed through sub-segment division and fluctuation amplitude extraction in the consistency-time curve reflects the consistency of the material's physical properties as it propagates along the path, and can be used as a quantitative indicator of the state quality during the transportation stage. The backfill uniformity index, generated by regional mean analysis and local difference extraction of the thickness variation trajectory, can characterize the degree of thickness distribution balance formed in the paving area during backfill operations. The construction of these two indicators compresses the original curve information into a highly comparable and reusable parameter format, providing a clearly structured data foundation for subsequent algorithm judgment and automated processing.

[0040] The flow homogeneity index and backfill uniformity index are combined as inputs into a machine learning model trained on construction data. The model then calculates and outputs a coordination control coefficient that reflects the degree of coordination between the current material delivery status and on-site construction backfill behavior. This step uses the two indicators as feature quantities and inputs them into the trained machine learning model for joint reasoning. Through feature extraction and nonlinear mapping of the relationship between the indicators through the model's internal structure, a continuous-valued coordination control coefficient is output. This coefficient is used to quantify the synergy between the material delivery process and on-site construction behavior. During training, the model incorporates a large number of historical construction samples and quality feedback results, and is capable of learning the matching relationship between input indicators and project status, thus ensuring that current judgments are data-supported and structured decision-making.

[0041] Based on the value of the coordination control coefficient and its changing trend during continuous construction, the corresponding traction adjustment strategy is implemented. This step compares and analyzes the trend of the coordination control coefficient obtained in the previous step across different construction cycles. When the coefficient value is stable, it indicates that the conveying and backfill are well matched, and the current traction rhythm is maintained. When the coefficient shows an increasing trend, the system identifies a decrease in construction coordination and adopts a slowing rhythm to reduce system load. When the coefficient shows a decreasing trend, it is judged that construction continuity is increasing, and efficiency can be further improved by increasing the propulsion speed. This strategy realizes closed-loop feedback regulation between construction status and material behavior, and is the final behavioral link in the dynamic control of this method.

[0042] The consistency-time curve is the trajectory of the material consistency of fluidized solidified soil over time during traction transport. It is used to reflect the structural state of the material during transport in pipelines or tracks. During actual data collection, multiple measuring points are set up along the traction transport path. These measuring points are spatially spaced at equal intervals, allowing each measuring point to obtain a set of measured material consistency values ​​at the corresponding time point. Consistency refers to the material's resistance to deformation under external forces per unit volume. It can be obtained using an online densitometer, viscometer, or numerical inversion based on the flow-pressure relationship. The consistency value obtained at each measuring point, together with the timestamp of the sampling moment, constitutes the "consistency change value." These values ​​are sequentially arranged to form a discrete data sequence. To convert this sequence into a continuous representation, mathematical processing is performed using interpolation methods, such as linear interpolation and cubic spline interpolation, to render the consistency change as a continuous, differentiable curve along the time axis, facilitating subsequent mathematical modeling and exponential calculations. In this structure, each consistency change value is related not only to the time point but also to its specific spatial position along the conveying path. Therefore, a complete consistency-time curve is an ordered set of curves composed of the three elements of "position-time-consistency." The gradient of the consistency-time curve is subsequently calculated using the time derivative of the curve, or the change in slope between adjacent measurement points, to measure whether the material state fluctuates significantly within a unit of time.

[0043] The spatial range of the entire consistency time curve is limited to the effective construction path interval, that is, from the starting point of traction to the end of the current construction section, and does not include the conveying section that has not yet been operated or the section where backfilling has been completed. This effective path interval is consistent with the traction and conveying direction, maintaining a unidirectional structure, which is conducive to controlling the rhythm and flow calculation. At the same time, in order to ensure data continuity, data fitting, data bridging or measuring point extension are used between adjacent conveying path sections (such as multiple pipe sections for segmented backfilling) to ensure that the curve connection does not have sudden changes or breaks, thereby maintaining the continuity and comparability of the entire consistency time curve structure. In the embodiment, a measuring point can be set every 10 meters, and the sampling frequency can be set every 10 seconds. A complete consistency time curve is constructed through 128 time series samples, and is automatically archived after each construction section is completed for subsequent analysis.

[0044] A thickness variation trajectory is a data track of actual backfill thickness in the paving area as it changes over time and location during the excavation backfill construction process. It reflects the uniformity of the material placement process and the overall placement trend. This data is collected by setting up equidistant measurement points along the construction progress direction on the construction surface, using, for example, a laser scanning rangefinder or a high-precision ultrasonic thickness probe. The fill thickness value at each surface measurement point is obtained. To ensure the integrity of the trajectory, each thickness measurement point is recorded with its specific spatial location (such as relative starting point coordinates or chain code number) and sampling time, creating a dual-labeled measurement point data point. By concatenating multiple measurement points in the construction progress sequence, a complete thickness variation trajectory for that operation cycle is obtained. This trajectory maintains temporal continuity and positional consistency within a single backfill cycle, with no missed measurements, no cross-sections, and no duplicated measurement points. This allows the entire trajectory to be used for fluctuation analysis, trend analysis, and index calculation. Consistent measurement point layout, coordinate reference, and sampling frequency are used across adjacent cycles (e.g., between multiple backfill sections) to maintain a consistent trajectory structure, enabling inter-cycle comparison and quality assessment. For example, if the paving area is a 50-meter-long strip, with a measuring point every 2 meters, a thickness sequence is collected every 5 minutes after the completion of construction. This creates a five-dimensional array (measuring point number × thickness value × time × location × operation number). This continuous measurement generates a thickness trajectory for that section. This data can be used to construct a backfill uniformity index and also provide input for mix ratio adjustments and pacing control.

[0045] After completing the data collection, the gradient value of the consistency time curve is obtained by calculating the slope between the continuous consistency change values, and the overall consistency stability is measured by the average slope change rate during the current construction period. If the average slope change rate exceeds the corresponding construction preset threshold, that is, the gradient change threshold, it is judged that the consistency time curve does not meet the construction stability requirements of the current operation section. The gradient value of the consistency time curve refers to the slope of the consistency change between any two adjacent sampling points of the curve in the time axis direction. The specific calculation method is to divide the difference between the two adjacent consistency change values ​​by the corresponding time interval. Since it has been set in the previous article that the consistency change values ​​are all collected at uniform time intervals, and the time interval is a constant Δt, the gradient value of each section The calculation can be expressed as: ;in is the density change value of the i-th sampling point. A set of gradient value sequences will be formed in a construction cycle:

[0046] ; n is the number index, and then the average slope change rate of the cycle is calculated by counting the average value of all continuous slope change amplitudes in the sequence. This value is used to measure the stability of the state change of the current material in the entire transportation process. The average value is defined as:

[0047] ; Represents the expected value of the slope change rate (i.e., the "intensity of fluctuation" of the curve). In order to determine whether the construction requirements are met, a construction preset threshold is set in advance based on engineering experience or experimental data, called the gradient change threshold, recorded as ,like: , it is considered that the change of the consistency time curve during the construction period is too drastic, reflecting that the physical stability of the material during transportation is insufficient and the flow continuity cannot be maintained. Therefore, this operation section does not meet the construction stability requirements and needs to enter the subsequent analysis link. In the embodiment, if 128 sampling points are used and the sampling interval is 10 seconds, 127 gradient values ​​are formed. If If the value exceeds the set value of 0.8 Pa / s (unit: consistency change / second), an exception will be triggered.

[0048] The two independent data items contained in the thickness variation trajectory are processed separately. One data item is the difference between the maximum and minimum thickness values ​​among all thickness measurements obtained at each measuring point within a construction cycle. This difference is the maximum-minimum difference and reflects the overall fluctuation in thickness data for that construction section. A larger difference indicates more significant thickness variations between measuring points, poorer paving uniformity, and the presence of significant localized peaks or valleys. The other data item is the thickness values ​​of all consecutive measuring points within the construction cycle. After spatially ordering the thickness values, the direction of the thickness variation between each pair of adjacent measuring points is determined. If the thickness at the current measuring point is greater than the thickness at the previous measuring point, the thickness is considered increasing; otherwise, it is considered decreasing. Within the entire measuring point sequence, a direction switch occurs when the thickness variation direction changes from increasing to decreasing, or vice versa. The number of such direction changes in the entire thickness sequence is counted as the number of trend direction switches. The more times this number occurs, the more frequent the thickness fluctuations are within a shorter distance, and the spatial distribution presents high-frequency fluctuation characteristics, reflecting that the fluctuation frequency of the paving process is large and the control accuracy is low.

[0049] The maximum acceptable fluctuation range during construction is set for each process, referred to as the corresponding preset construction threshold. The preset threshold corresponding to the difference between the maximum and minimum values ​​is used to determine whether the thickness fluctuation amplitude exceeds the standard, referred to as the numerical difference threshold. The preset threshold corresponding to the number of trend change direction switches is used to determine whether the fluctuation frequency exceeds the limit, referred to as the switch count threshold. When the measured value of any data item within a construction cycle exceeds its corresponding preset construction threshold, the thickness variation trajectory is considered to be unstable and fails to meet the construction consistency and uniformity requirements of the current operation section, requiring subsequent analysis or control processing. For example, in a typical construction section, if twenty measurement points are placed along the paving direction, the maximum thickness is 37 cm and the minimum thickness is 25 cm, then the maximum-minimum difference is 12 cm. If the numerical difference threshold set for construction requirements is 10 cm, the thickness fluctuation in this section has exceeded the acceptable range. At the same time, nine direction switches occurred at these twenty measurement points, while the maximum number of switches allowed for this operation is no more than seven, thus meeting the threshold determination criteria. At this point, it can be judged that the thickness variation trajectory of this section has a large fluctuation amplitude and high frequency, and the construction quality is unstable, so further index extraction and model analysis must be carried out.

[0050] When constructing the flow homogeneity index using the consistency-time curve, the consistency-time curve within the current construction cycle is first divided into multiple adjacent sub-segments at fixed time intervals. This consistency-time curve describes the continuous change in material consistency over time and is used to reflect the dynamic trend of the material's physical state during traction and conveying. A construction cycle is a continuous conveying operation period, such as a conveying section corresponding to ten minutes or fifty meters. The fixed time interval can be set based on the sampling frequency, such as dividing every sixty seconds into a sub-segment. This division results in multiple sub-segments, each of which contains the consistency change values ​​of several sampling points.

[0051] For each subsection, all consistency variations are extracted, and the difference between the maximum and minimum values ​​is calculated. This difference is defined as the consistency fluctuation amplitude for that subsection. This fluctuation amplitude characterizes the severity of consistency variations within the region and is an important indicator of local stability. The consistency fluctuation amplitudes for all subsections are arranged sequentially, forming a complete fluctuation amplitude sequence in chronological order. This sequence describes the consistency variations across subsections throughout the construction cycle. Within this fluctuation amplitude sequence, the direction of fluctuation amplitude changes between two consecutive segments is compared. A directional switch is considered to have occurred when the difference between two adjacent fluctuation amplitudes changes from positive to negative or from negative to positive. The frequency of these directional switches within the sequence is used as a disturbance direction index, which indicates whether the consistency fluctuation exhibits directional stability—that is, whether the consistency steadily increases, steadily decreases, or oscillates repeatedly at high frequencies. The consistency fluctuation amplitude for each subsection is then multiplied by its corresponding disturbance direction index to obtain a set of directional disturbance correction values. This correction value not only takes into account the intensity of fluctuations within a single sub-segment, but also whether the direction of fluctuation changes is consistent, which can more truly reflect the structural nature of material state changes. The largest and smallest absolute values ​​are extracted from the directional disturbance correction value sequence, and the difference between the two is calculated as the consistency response value. The smaller the response value, the more consistent the overall consistency of the material during transportation, and the more concentrated the fluctuations and directional changes in each sub-segment. The larger the response value, the more obvious regional differences or inconsistent direction of change in material transportation, reflecting poor overall stability.

[0052] Finally, the consistency response value is divided by the total number of sub-segments for normalization to avoid the reliability of the evaluation results being affected by the number of segments. The calculation result is defined as the flow homogeneity index within the current construction period, which is used to determine the concentration of fluctuations and directional continuity of the current consistency time curve under segment division. This design integrates the local intensity, global consistency, and directional uniformity of the consistency change, taking into account both the spatial discreteness of physical properties and the stability of the temporal evolution trend. It is a hierarchical and closed-loop comprehensive evaluation method for material stability.

[0053] For example, if a construction cycle is divided into ten sub-segments, five consistency values ​​are collected in each segment, totaling fifty points, resulting in ten fluctuation amplitudes. After identifying the disturbance direction, three directional switches indicate a low directional index. After calculating the directional disturbance correction value, if the difference between the maximum and minimum correction values ​​is thirty-six units, dividing this value by the number of sub-segments, ten, yields a flow homogeneity index of 3.6 units, indicating relatively unstable material conditions in that construction section. If the index drops below one unit in subsequent cycles, it can be inferred that the construction process is improving.

[0054] To construct the backfill uniformity index using the thickness variation trajectory, a sequence of equally spaced positions is first selected along the thickness variation trajectory for the current construction cycle. This thickness variation trajectory represents a data sequence of backfill thickness continuously varying with spatial position within the paving area. It exhibits a monotonically progressive distribution in spatial direction and is used to reflect the spatial uniformity of the material layer during construction. The equally spaced position sequence can be selected by placing measurement points at equal intervals along the paving path, for example, one measurement point every two meters, thus forming a series of sequentially arranged spatial position nodes.

[0055] At each position node, the corresponding actual measured thickness value is extracted, and together with the position, it constitutes a thickness position pair. All thickness position pairs are organized into a thickness position pair sequence in spatial order, forming a data basis that continuously reflects the spatial variation of thickness. Subsequently, the thickness position pair sequence is divided into two continuous spatial regions of equal length. For example, if a total of twenty position nodes are collected, each region contains ten nodes. All thickness measurements in each region are averaged to obtain the average thickness value of the two regions, and the difference between the two average values ​​is calculated. The square of the difference is used as the regional difference factor. The physical meaning of the regional difference factor is that it is used to characterize the overall thickness deviation between different position segments within the same construction period, reflecting whether there is systematic inconsistency in the backfill in different spatial segments, and is a global scale quantitative evaluation of the backfill uniformity.

[0056] Each area is further subdivided, for example, every tenth node is further divided into two subgroups, each containing five nodes. For each subgroup, the difference between the maximum and minimum thickness values ​​is calculated, and this is recorded as the subgroup's internal fluctuation amplitude. This fluctuation amplitude represents the degree of thickness fluctuation within the local spatial range of the subgroup, reflecting the short-term stability of the construction process.

[0057] The fluctuation amplitudes of all subgroups in the two regions are arranged in spatial order to form a local fluctuation amplitude sequence. The variance of all fluctuation amplitude data in this sequence is calculated to obtain its average variance, which is used as the local fluctuation factor. This factor comprehensively reflects whether the fluctuations within each local region are consistent and whether the fluctuation amplitudes are concentrated, and serves as a local-scale evaluation basis for judging backfill stability. The regional difference factor is multiplied by the local fluctuation factor to obtain the fluctuation structure coupling value within the current construction period. This value combines the two dimensions of global segment differences and local fluctuation amplitude to comprehensively evaluate whether there are systematic deviations or severe local fluctuations in the current backfill thickness distribution, providing a multi-scale nested uniformity expression.

[0058] Finally, the fluctuation structure coupling value is divided by the square root of the length of the area covered by the thickness position sequence to obtain the backfill uniformity index of the current construction period. This index is a numerical indicator with unit consistency and structural closed-loop properties, which can be used for quantitative comparison across periods or regions. The smaller the index value, the more consistent the thickness distribution, the concentrated fluctuation structure, and the consistency between the local and global states; the larger the value, the uneven thickness distribution, the presence of significant unbalanced areas or repeated jump characteristics. In the process of constructing the index, this design scheme adopts a segmented mean and sub-segment fluctuation linkage calculation structure. On the one hand, it can retain the actual physical form of the thickness sequence and avoid the concealment of anomalies caused by global averaging. On the other hand, it introduces a multi-scale nested analysis method, so that the backfill uniformity can be jointly judged from the two aspects of overall consistency and local fluctuations, thereby providing a more discriminative construction quality quantification basis.

[0059] For example, during a certain construction cycle, the measurement location covered an area of ​​40 meters in length, with 20 measuring points set, divided into two main areas, each containing 10 points, which were further divided into two subgroups. The average thickness of the two main areas was 35 centimeters and 32 centimeters, respectively, and the square of the difference was 9 square centimeters, which served as the regional difference factor. The fluctuation amplitudes of the four subgroups were 5 centimeters, 4 centimeters, 6 centimeters, and 5 centimeters, respectively, and the average variance was calculated to be 1 square centimeter, which served as the local fluctuation factor. Multiplying the regional difference factor by the local fluctuation factor yielded a fluctuation structure coupling value of 9 square centimeters. Dividing this value by the square root of the 40-meter area length is approximately 6.3 meters, resulting in a backfill uniformity index of approximately 1.43 units. This value can be used as an evaluation indicator for horizontal comparison with other construction cycles to determine whether the thickness control strategy during construction needs to be optimized.

[0060] The machine learning model is any one of the following three: a convolutional neural network model, a recurrent neural network model, or a gradient boosting tree model. All three models use the flow homogeneity index and the backfill uniformity index as input, and use construction sample data as the training basis to output the coordination control coefficient. Among them, the convolutional neural network model is an artificial intelligence algorithm structure derived from the field of image recognition. Its basic principle is to extract spatial features in local areas by sliding multiple convolution kernels on the input data, and then realize the abstraction of global features through the step-by-step combination of multiple layers of neural nodes. In this technical solution, the flow homogeneity index and the backfill uniformity index have certain spatial structural characteristics in the time dimension and the operation section dimension. Therefore, the index values ​​of multiple cycles can be input as a two-dimensional data array, and the convolution structure is used to extract its local correlation features. The pooling operation and the fully connected structure are combined to output the coordination control coefficient in the form of a continuous numerical value. The core advantage of this model is that it can automatically extract high-dimensional interactive features, which is suitable for identifying complex nonlinear construction response patterns.

[0061] The recurrent neural network model is a machine learning structure that excels at processing time series data. Its hallmark is its internal memory mechanism, which allows it to retain the state of historical inputs and feed them back into current judgments. In this solution, the flow homogeneity index and backfill uniformity index of multiple consecutive construction cycles form a bivariate time series data sequence. The model dynamically outputs the coordination control coefficient by receiving the input of the current cycle at each time step and combining it with the hidden state of the previous step. This structure is particularly suitable for handling situations where dependencies exist between multiple cycles during the construction process, and can accurately capture the cumulative effects or hysteresis changes in the evolution of the construction rhythm.

[0062] The gradient boosting tree model is a structural regression method based on ensemble learning. Its basic principle is to construct multiple basic tree models with decision-making functions. Each tree is optimized for the error of the previous round, gradually improving the overall prediction accuracy. In this technical solution, the flow homogeneity index and the backfill uniformity index are used as a set of input features. The corresponding coordination control coefficient is fitted through the series connection and weighted combination of multiple basic regression trees. The gradient boosting tree model belongs to the category of structured data mining widely used in the existing technology. It has the advantages of clear structure, strong interpretability, and adaptability to small sample training. It is particularly suitable for control logic judgment tasks in construction scenarios where limited historical data is available but high model accuracy requirements are required.

[0063] In actual application, all three models are trained based on data from historical construction cycles. The training data includes the flow homogeneity index, backfill uniformity index, actual control feedback actions, and backfill quality results. All three models are mature migration applications of existing technologies, so the training process will not be described in detail. After training, any model can accept the newly input flow homogeneity index and backfill uniformity index as joint inputs. Through internal parameter calculation, it outputs a continuity indicator that reflects the degree of coordination between the current material delivery status and construction backfill behavior. This indicator is the coordination control coefficient. As a single-valued result, the coordination control coefficient is used for subsequent traction adjustment path determination and is the key technical foundation of the dynamic feedback control link in this solution.

[0064] For example, during a typical construction project, data is collected for ten consecutive construction cycles. Two input indices are calculated for each cycle, and the traction strategy used in each cycle and the corresponding paving performance score are recorded. By feeding these samples into the training process, a set of input-output pairs containing 200 data points is constructed. Through iterative model training, a stable prediction function is ultimately obtained, which can be used to convert any new index input into a coordination control coefficient.

[0065] After obtaining the coordination control coefficient, the traction adjustment strategy is executed according to the value of the coordination control coefficient and its changing trend formed in the continuous construction cycle. The coordination control coefficient is the result of the comprehensive judgment of the current construction status by the aforementioned machine learning model. Its output is a single continuous value, which is used to reflect the degree of coordination between the material transportation status and the on-site construction backfill behavior. It is the core decision-making basis of the feedback control closed loop in the present invention. This coefficient is recorded in real time during the construction process and forms a continuous sequence that evolves over time. This coefficient is updated once in each construction cycle. By performing trend analysis on the coordination control coefficients of multiple consecutive cycles, the state evolution direction of the system operation can be determined, which serves as the basic logic for adjusting the traction strategy.

[0066] After the generation of the coordination control coefficient is completed, the advancement rhythm of the traction and conveying process is dynamically adjusted according to its value itself and the changing trend formed during the continuous construction cycle. The adjustment strategy not only includes "maintaining the original rhythm", "slowing down the rhythm", and "speeding up the rhythm", but is further expanded to deal with the fourth type of working condition where the index fluctuation has no obvious direction but continues to deviate from the stable range, namely "index fluctuation adjustment". The coordination control coefficient is a numerical indicator that reflects the degree of coordination between the material transportation status and the backfill construction status. The range is usually between zero and ten, where the higher the value, the worse the coordination, and the lower the value, the higher the coordination. The system sets an upper and lower limit range based on historical construction data to form a preset stable interval, for example, it is set to between three and four. In the actual control process, the application of the coordination control coefficient is divided into the following three typical states and a chaotic state (that is, a state that cannot be clearly and directly judged):

[0067] Typical Status 1: Stable Values ​​and Stable Trends (Maintaining the Original Rhythm): When the fluctuation range of the coordination control coefficient remains within the preset stable range over multiple consecutive construction cycles, with no apparent upward or downward trend, the current construction coordination is considered good. The existing propulsion speed and conveying frequency of the traction system are maintained, and no adjustments to the control parameters are made.

[0068] Typical State 2: Continuously Increasing Values ​​(Slowing the Pace): When the coordination control coefficient shows a continuous upward trend in multiple adjacent construction cycles, and the current value approaches or exceeds the upper limit of the stable range, it indicates that the material fluidity has decreased, the structural stability has deteriorated, or there are systematic fluctuations in the thickness control, reflecting a decline in construction coordination. At this time, the control system implements the "slowing the pace" strategy. Specifically, the pace adjustment ratio is calculated based on the increase in the coordination control coefficient. For example, if the current cycle coefficient increases by more than 20% compared to the average of the previous three cycles, the traction and propulsion speed will be reduced proportionally, for example, by 10% to 20%; if it increases three times in a row and the increase continues to expand, a graded decreasing method will be used to implement a stronger pace buffer, extending the propulsion time per unit distance.

[0069] Typical State 3: Continuously Decreasing Value (Accelerated Execution Pace): When the coordination control coefficient shows a continuous downward trend and the current value gradually approaches or falls below the lower limit of the stable range, it indicates that the coordination between conveying and backfilling during construction is improving, and coordination is continuously improving. At this time, the system can implement the "accelerated pace" strategy. Similarly, the speed increase action is executed based on the proportion of the current value decreasing relative to the previous average. For example, if the coordination control coefficient has decreased for three consecutive cycles, with an overall decrease of more than 30%, the advancement speed can be increased by 10% and subsequent changes can be observed. If the index continues to decline after acceleration or remains in the optimal range, the speed can be continuously increased slightly to improve construction efficiency.

[0070] Chaotic state, values ​​fluctuate but do not enter the stable range for a long time (performing rhythm fluctuation adjustment): When the coordination control coefficient fluctuates greatly in consecutive construction cycles, for example, it rises rapidly in one cycle, drops significantly in the next cycle, and then rises again, the value change amplitude exceeds the stable range, but there is no obvious single direction trend. This state is called the "high-frequency fluctuation unstable zone". At this time, the control system executes the "rhythm fluctuation adjustment" strategy. Unlike trend control, this strategy determines whether there is feedback lag or operational disturbance by calculating the maximum fluctuation amplitude and the degree of difference between the current cycle and the average value of the previous two cycles. The specific operation is: if the standard deviation of the coordination control coefficient of the last three cycles exceeds the set threshold (for example, a unit value), it is considered that the system has obvious unstable fluctuations.

[0071] If the current cycle value is higher than the average of the previous two cycles by a certain percentage (e.g., 15 percent), a temporary slowdown strategy is implemented, reducing the delivery rate by 5-10 percent. If the current cycle value is lower than the average by a smaller percentage (e.g., 5 percent), the current pace is maintained. If the changes are disordered but persistently high (e.g., exceeding the upper limit of the stable range three times within five cycles), the "upward trend" logic is used. This strategy ensures that the system can still adaptively adjust under short-term disturbances, preventing oscillation caused by excessive adjustments and ensuring construction continuity.

[0072] For example: Assume that the system stability range is 3.0 to 4.0: Cycle one: 3.2, cycle two: 3.3, cycle three: 3.25, small changes, within the range → maintain the original rhythm; Cycle one: 4.1, cycle two: 4.5, cycle three: 4.8 → the rhythm slows down; Cycle one: 4.2, cycle two: 3.8, cycle three: 3.5 → the rhythm speeds up; Cycle one: 4.7, cycle two: 3.4, cycle three: 4.5 → fluctuation adjustment, fine-tuning and deceleration.

[0073] Example 2: A traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction optimization control system, comprising:

[0074] The data acquisition unit is used to continuously collect data during the traction and transportation of fluidized solidified soil. The data is divided into two categories: one is the consistency time curve formed by the change of material consistency over time along the transportation path, and the other is the thickness variation trajectory formed by the change of backfill thickness in the paving area as it changes with position and time. Both types of data serve as the parameter basis for subsequent processing;

[0075] The state assessment unit is used to calculate the gradient value of the consistency time curve within the current construction cycle after completing data collection, extract the maximum and minimum value differences of the thickness change trajectory, and analyze the deviation trend. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met, and the processing flow is advanced to subsequent analysis;

[0076] The feature extraction unit is used to construct a flow homogeneity index based on the consistency time curve after triggering analysis, which is used to determine the structural stability of the material during long-distance transportation; and a backfill uniformity index based on the thickness change trajectory, which is used to reflect the thickness control level of the paving area;

[0077] The model building unit is used to take the flow homogeneity index and backfill uniformity index as joint inputs and import them into the machine learning model trained based on construction data. The output is a coordination control coefficient that reflects the degree of coordination between the current material delivery status and the on-site construction backfill behavior. This coefficient is a continuous value and is used for subsequent control path judgment.

[0078] The strategy execution unit is used to execute the corresponding traction adjustment strategy according to the value of the coordination control coefficient and its changing trend in the continuous construction cycle.

[0079] The above numerical calculations are all dimensionless numerical calculations. The formulas involved in the numerical calculations are a formula for the latest real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0080] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0081] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0082] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0083] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit, characterized in that: The following steps are involved: Continuous data collection is performed during the traction and transportation of fluidized solidified soil. The data is divided into two categories: one is the consistency time curve of the material consistency changing with time along the transportation path, and the other is the thickness variation trajectory of the backfill thickness in the paving area changing with position and time. After completing data collection, the gradient value of the consistency time curve in the current construction cycle is calculated, and the maximum and minimum value differences of the thickness change trajectory are extracted and the deviation trend is analyzed. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met and the next step of analysis and processing is entered; After triggering the analysis, the consistency-time curve is used to construct the flow homogeneity index, which is used to determine the structural stability of the material during long-distance transportation. The thickness change trajectory is used to construct the backfill uniformity index, which reflects the thickness control level of the paving area. The flow homogeneity index and backfill uniformity index are combined as inputs into a machine learning model trained on construction data. The model then calculates and outputs a coordination control coefficient that reflects the degree of coordination between the current material delivery status and the on-site backfill behavior. According to the value of the coordination control coefficient and its changing trend during the continuous construction process, the corresponding traction adjustment strategy is implemented; During the data acquisition process, the consistency-time curve continuously obtains the real-time change value of the material consistency through evenly spaced measurement points. Interpolation processing is used to construct a continuous curve, in which each consistency change value corresponds to a unique combination of conveying path position and time node, which is used to subsequently calculate the gradient value of the consistency-time curve. At the same time, the consistency-time curve is confined to the range of the effective construction path interval. The effective construction path interval range is consistent with the traction conveying direction, and the curve continuity is maintained between adjacent conveying path sections. During the data collection process, the thickness change trajectory obtains continuous thickness measurement values ​​by equidistant measurement points distributed along the construction direction in the paving area, and double-marks the position and time of each measurement point data in combination with time information to generate a complete thickness change trajectory.

2. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 1 is characterized in that: After completing data collection, the gradient value of the consistency time curve is obtained by calculating the slope between continuous consistency change values, and the overall consistency stability is measured by the average slope change rate during the current construction cycle. If the average slope change rate exceeds the corresponding construction preset threshold, that is, the gradient change threshold, it is judged that the consistency time curve does not meet the construction stability requirements of the current operation section.

3. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 2 is characterized in that: The two independent data items contained in the thickness change trajectory are processed separately. One is the maximum and minimum value difference calculated by the extreme value difference in the measured thickness values ​​of the measuring points within the same construction period, and the other is the number of direction switching of the trend change of the thickness values ​​of consecutive measuring points in the position sequence. They are respectively set with their own preset construction thresholds, namely the numerical difference threshold and the switching number threshold. When the maximum and minimum value difference exceeds the numerical difference threshold, or the number of direction switching exceeds the switching number threshold, it is judged that the thickness change trajectory does not meet the construction stability requirements of the current operation section.

4. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 3 is characterized in that: In the process of constructing the flow homogeneity index using the consistency time curve, the consistency time curve in the current construction period is first divided into multiple adjacent sub-segments according to fixed time intervals. All consistency change values ​​are extracted in each sub-segment, and the difference between the maximum and minimum consistency change values ​​in the sub-segment is calculated and recorded as the consistency fluctuation amplitude of the sub-segment; the consistency fluctuation amplitudes of all sub-segments are further arranged in sequence to construct a complete fluctuation sequence, and then the sign change frequency of the difference between each two is calculated based on the fluctuation sequence, and this frequency is used as the disturbance direction index of the fluctuation sequence. The viscosity fluctuation amplitudes of all sub-segments are then multiplied by their corresponding disturbance direction index to obtain a set of directional disturbance correction values. The difference between the largest and smallest absolute values ​​in the correction value sequence is used as the consistency consistency response value. Finally, the consistency consistency response value is divided by the total number of sub-segments to define the flow homogeneity index in the current construction period.

5. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 4 is characterized in that: In the process of constructing the backfill uniformity index using the thickness variation trajectory, first, an equally spaced position sequence is selected on the thickness variation trajectory of the current construction period, and the corresponding thickness measurement value is extracted for each position node to form a thickness position pair sequence; the thickness position pair sequence is divided into two consecutive equal-length regions, and the average value of the thickness measurement in each region is calculated respectively. The square of the difference between the average values ​​of the two regions is calculated as the regional difference factor. Subsequently, each region is further divided into subgroups, and the maximum and minimum difference of the internal thickness values ​​of each subgroup is calculated, which is recorded as the internal fluctuation amplitude; the fluctuation amplitudes of all subgroups in the two regions are arranged in sequence, and the average square deviation of all fluctuation amplitudes is calculated as the local fluctuation factor; the regional difference factor is multiplied by the local fluctuation factor to obtain the fluctuation structure coupling value of the current construction period, and then the backfill uniformity index is defined as the ratio of this fluctuation structure coupling value to the square root of the region length of the thickness position pair sequence.

6. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 5 is characterized in that: The machine learning model is any one of the following three: a convolutional neural network model, a recurrent neural network model, or a gradient boosting tree model. All three models take the flow homogeneity index and the backfill uniformity index as input and output a coordination control coefficient.

7. The method for optimizing and controlling the backfill construction of a traction-type environmentally friendly fluidized solidified soil foundation pit according to claim 6 is characterized in that: After obtaining the coordination control coefficient, the traction adjustment strategy is executed according to the value of the coordination control coefficient and its changing trend during the continuous construction cycle. When the fluctuation range of the coordination control coefficient is within the preset stable range during multiple consecutive construction cycles, the existing transportation rhythm and traction propulsion rhythm are maintained unchanged. When the coordination control coefficient shows a continuous increasing trend, it is determined that the material structure state and the backfill thickness fluctuation are both tending to be unstable, and a rhythm slowdown strategy is executed in the control process; when the coordination control coefficient shows a continuous decreasing trend, it is determined that the coordination between transportation and backfilling is improving, and a rhythm acceleration strategy is executed.

8. A traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction optimization control system, used to implement the traction-type environmentally friendly fluidized solidified soil foundation pit backfill construction optimization control method according to any one of claims 1 to 7, characterized in that: include: The data acquisition unit is used to continuously collect data during the traction and transportation of fluidized solidified soil. The data are divided into two categories: one is the consistency time curve formed by the change of material consistency over time on the transportation path, and the other is the thickness change trajectory formed by the change of backfill thickness in the paving area over position and time. The state assessment unit is used to calculate the gradient value of the consistency time curve within the current construction cycle after completing data collection, extract the maximum and minimum value differences of the thickness change trajectory, and analyze the deviation trend. If any indicator exceeds the corresponding construction preset threshold, it is determined that the construction stability requirements of the current operation section are not met, and the processing flow is advanced to subsequent analysis; The feature extraction unit is used to construct a flow homogeneity index based on the consistency time curve after triggering analysis, which is used to determine the structural stability of the material during long-distance transportation; and a backfill uniformity index based on the thickness change trajectory, which is used to reflect the thickness control level of the paving area; A model building unit is used to take the flow homogeneity index and backfill uniformity index as joint inputs, import them into a machine learning model trained based on construction data, and output a coordination control coefficient that reflects the degree of coordination between the current material delivery status and the on-site construction backfill behavior; The strategy execution unit is used to execute the corresponding traction adjustment strategy according to the value of the coordination control coefficient and its changing trend in the continuous construction cycle.

Images