Flow-state solidified soil detection method for backfilling

Abstract

The invention discloses a flow-state solidified soil detection method for backfilling, and belongs to the technical field of flow-state solidified soil detection.The flow-state solidified soil detection method comprises the following steps that S1, a modular sensing array is constructed, and the modular sensing array comprises a construction parameter module, a mechanical property module and an environment-friendly module; sensor calibration is completed through two processes of factory calibration and on-site secondary calibration; an AI model of CNN quality prediction is constructed; s2, detecting environmental protection indexes of the raw materials through a green environmental protection module; calculating a carbon footprint and setting an emission reduction reference value; inputting an AI model to pre-judge the rationality of the ratio; s3, layered monitoring and dynamic verification in construction; s4, AI prediction and long-acting tracking after construction; and S5, carrying out complete-cycle data fusion and traceability. According to the invention, the problems of one-sided detection dimension, fragmented process, insufficient precision, low intelligent degree, poor scene adaptability and the like in the prior art are solved through the modular sensor array, the full-period closed-loop process and multi-technology fusion.

Description

Technical Field

[0001] This invention relates to the field of fluidized solidified soil testing technology, specifically to a method for testing fluidized solidified soil used in backfilling. Background Technology

[0002] Fluidized solidified soil is a mixture made by sourcing soil locally, adding a solidifying agent and water, and possessing the core characteristics of high fluidity and good self-compactment. After hardening, it can form a geotechnical engineering material with stable strength and is widely used in municipal and construction engineering fields such as foundation pit backfilling, roadbed reinforcement, and mine pit remediation. Its engineering quality directly determines the bearing capacity, stability, and long-term service performance of the backfill. As a core means of quality control, testing technology needs to accurately assess key indicators such as fluidity, suspension, density, and strength, which is an important link in ensuring engineering safety.

[0003] With the upgrading of environmental protection requirements, the testing of fluidized solidified soil is no longer limited to traditional mechanical performance indicators, but is gradually developing towards green, environmentally friendly, intelligent and efficient directions. It is not only necessary to ensure that the engineering quality meets the standards, but also to take into account the control of the release of harmful substances and the evaluation of carbon emission reduction effectiveness. The formation of a full-cycle, multi-dimensional testing system has become an inevitable trend in the development of the industry.

[0004] Currently, multiple implementation paths have emerged for the detection technology of fluidized solidified soil: In terms of single index detection, a suspension detection method based on the propagation characteristics of ultrasonic signals has appeared, which constructs a suspension coefficient by calculating the horizontal and vertical fluctuation coefficients to achieve the initial detection of segregation phenomena; for fluidity detection, a special device based on a vertical outer tube and a scale rod has been developed, which quickly calculates the fluidity through the principle of permeability; in the field of intelligent monitoring, there are already detection equipment based on the theory of full-field elastic wave testing, which can achieve high-precision perspective of pile foundation integrity, and realize anomaly early warning and dynamic calculation of carbon sinks when combined with an intelligent monitoring platform; at the same time, there is also a compaction detection scheme that relies on robotic soil sampling combined with sand filling method, which realizes single quality index accounting through on-site sampling.

[0005] The aforementioned technologies have achieved breakthroughs in the accuracy of single indicators, detection efficiency, or level of intelligence, forming the mainstream technical system for the current detection of fluidized solidified soil and providing basic support for engineering quality control.

[0006] Although existing technologies have met some testing requirements, they still have significant limitations in practical engineering applications, making it difficult to meet the needs of dual quality and environmental protection control and full-cycle risk prevention and control in complex scenarios.

[0007] 1. One-sided testing dimensions and lack of green environmental assessment: Existing technologies mostly focus on the testing of single mechanical indicators, such as only conducting specific monitoring of suspension, flowability, or density, without incorporating the release of harmful substances such as heavy metals and VOCs, as well as carbon footprint, into the testing system, making it unsuitable for current environmental acceptance requirements for engineering projects. A few technologies involving carbon sink calculations lack a full-process traceability mechanism, resulting in a lack of verifiability for environmental data.

[0008] 2. Fragmented testing process, lacking a closed-loop logic throughout the entire lifecycle: Existing technologies are mostly single-point testing, lacking a linkage mechanism for pre-construction prediction, in-construction monitoring, and post-construction verification. For example, ultrasonic levitation testing can only detect abnormalities during construction, and elastic wave testing focuses on post-construction integrity assessment. Neither can predict the rationality of the mix design or the trend of strength development in advance, resulting in passive quality risk prevention and control.

[0009] 3. Insufficient data accuracy and reliability: Single detection methods are prone to errors. For example, robotic soil sampling suffers from sampling disturbances, and directly buried sensors are susceptible to construction damage. Furthermore, current technologies lack multi-source data fusion and correction mechanisms, resulting in detection deviations generally exceeding ±6%. Additionally, a dynamic calibration system for changes in ambient temperature and water-cement ratio has not been established, making data accuracy highly susceptible to variations in the application environment.

[0010] 4. Low level of intelligence and insufficient decision-making efficiency: Existing technologies mostly rely on manual data calculation or post-event intensity measurement, lacking AI-driven quality prediction capabilities. Even if some technologies have early warning functions, they have not established a correlation model between indicators and long-term performance, making it impossible to predict intensity in advance. This results in non-conforming areas needing rework and rectification, increasing project costs.

[0011] 5. Poor scenario adaptability and high implementation costs: Dedicated detection devices are mostly designed for specific scenarios. For example, flow detection devices are only suitable for surface detection, and robotic detection cannot be adapted to deep backfill scenarios. Moreover, existing technologies have not optimized equipment protection and power supply solutions, resulting in a high sensor failure rate and high installation and maintenance costs, making it difficult to meet the high-efficiency detection needs of large-scale projects.

[0012] The present invention aims to provide a method for detecting fluidized solidified soil for backfilling, in order to solve the problems of the existing technology, such as one-sided detection dimensions, fragmented process, insufficient accuracy, low level of intelligence and poor adaptability to scenarios. Summary of the Invention

[0013] This invention addresses the problem of overly simplistic solutions in existing technologies by providing a significantly different approach. It primarily offers a method for detecting fluidized solidified soil used in backfilling, constructing a dual-dimensional testing system for quality and environmental protection. This fills the gap in existing technologies regarding the lack of monitoring for hazardous substances and carbon footprint tracing, enabling simultaneous control of mechanical properties and green attributes. It establishes a closed-loop process encompassing pre-construction prediction, in-construction monitoring, post-construction verification, and full-cycle traceability, resolving the fragmented testing and passive risk control issues of existing technologies. By integrating multi-source testing data with AI prediction models and combining dynamic calibration and data fusion correction mechanisms, it improves testing accuracy and long-term performance prediction capabilities, reducing reliance on manual labor and decision-making time. Furthermore, it optimizes modular design and risk control mechanisms, enhancing the scenario adaptability and reliability of the testing equipment, reducing sensor damage rates and installation and maintenance costs, and meeting the efficient testing needs of various backfilling scenarios.

[0014] For those skilled in the art, based on existing patents and related technologies, it would be difficult to conceive of the technical solution in this application, which suffers from the following technical pain points and technical biases:

[0015] 1. The fragmentation of testing dimensions: Existing technologies have resulted in two independent directions: quality testing and environmental assessment.

[0016] Patents related to quality testing (such as CN119757115B and CN118168503A) only focus on mechanical indicators such as density, strain, and strength, without considering the needs for carbon footprint accounting and monitoring of the release of hazardous substances.

[0017] Environmental protection technologies are mostly focused on industrial pollution control and have not been integrated with the quality control process for fluidized bed solidification soil backfilling. This current technological situation has led to a preconceived notion among those skilled in the art that quality and environmental protection are unrelated, making it difficult to conceive of integrating the two into a dual-dimensional detection system. Combining the two requires simultaneously solving two major technical challenges: modular compatibility between mechanical and environmental sensors, and the linkage and traceability of carbon footprint data and construction data. Existing technologies have not provided any precedent for such integration, creating a natural capability barrier.

[0018] 2. Fragmentation of the testing process: Existing technologies are all single-point testing, which cannot form a coordinated process from pre-construction to construction, construction, and post-construction.

[0019] Pre-construction testing (such as CN116678786A) can only verify the suspension of the slurry and cannot predict the later strength;

[0020] Monitoring during construction (such as CN120507431B) can only detect real-time anomalies and cannot trace the source of mixing problems;

[0021] Post-construction assessments (such as CN119757115B) can only provide actual measurement results and cannot prevent the risk of rework in advance.

[0022] This fragmentation makes it difficult for technical personnel to conceive of a closed-loop logic of "prediction-monitoring-verification-tracing". This closed-loop logic requires overcoming multiple technical bottlenecks, such as cross-stage data linkage, real-time interaction between AI models and on-site sensors, and multi-source data fusion and correction. However, existing technologies lack experience in integrating relevant technical modules, making it impossible to find a feasible integration path even if the fragmentation problem is recognized.

[0023] 3. The contradiction in implementation reliability: Existing technologies always face the contradiction between high precision and high adaptability.

[0024] High-precision detection (such as robotic soil sampling and ultrasonic detection) relies on specialized equipment and is only suitable for specific scenarios (such as surface detection and wall perimeter detection), and cannot be adapted to complex deep backfilling scenarios such as foundation pits and mine pits.

[0025] Highly adaptable detection methods (such as simple sensor pre-embedded devices) are easily affected by construction interference, have a high sensor damage rate (such as the wired transmission of CN118168503A which is easily squeezed), and have insufficient data accuracy.

[0026] This contradiction puts engineers in a dilemma: either sacrifice accuracy or limit application scenarios, making it difficult to devise a solution that combines modular design with end-to-end risk control. This solution needs to address multiple sub-problems simultaneously, such as sensor collision avoidance, dual power supply, and dynamic calibration. However, existing technologies only optimize for single problems (such as improving sensor accuracy or expanding applicable scenarios) without providing a systematic approach to risk control.

[0027] 4. Technical positioning bias: Those skilled in the art have long regarded fluidized solidified soil testing as a post-construction quality verification tool, and have a bias that the core value of testing is to confirm whether the construction results are qualified, while ignoring the core need to predict risks in advance and optimize the construction process.

[0028] Existing patents (such as CN119757115B and CN117647467A) all aim to obtain real-time data or actual measurement results, and do not involve the linkage between AI prediction models and construction parameters.

[0029] Technicians subjectively believe that performance indicators such as strength can only be measured after curing, and find it hard to think of using indirect indicators such as shear wave velocity and wet density, combined with AI models, to predict the strength at 28 days in advance, thus falling into a mindset of passively accepting the results.

[0030] 5. Technology Combination Bias: For a long time, existing technologies have adopted an improvement path that focuses on a single detection method. This has led to a bias among technical personnel that optimizing the accuracy of a single method equals technological progress, thus rejecting the integration of multiple technologies across dimensions.

[0031] Ultrasonic detection technologies (such as CN120507431B) only optimize ultrasonic signal analysis algorithms and do not consider integration with radar detection or sensor arrays;

[0032] Robotic inspection technologies (such as CN119757115B) only improve the automation level of soil sampling and sand filling, without involving data fusion and AI prediction;

[0033] Technicians subjectively believe that combining multiple technologies will lead to structural complexity and increased costs, ignoring the dual benefits of improved accuracy and reduced risk brought about by the complementarity of multiple technologies. They find it difficult to conceive of a cross-dimensional combination solution of sensor arrays, dual-modal detection, and AI prediction.

[0034] 6. Bias in Indicator Perception: There is a common bias among those skilled in the art that testing of fluidized solidified soil requires only attention to its mechanical properties, and that environmental indicators are merely an additional requirement for project acceptance and do not need to be included in the testing during construction.

[0035] Existing patents do not include indicators such as carbon footprint, VOCs, and heavy metals in the testing system. Technical personnel subjectively believe that environmental protection indicators should be tested separately by environmental protection departments and are unrelated to quality control during the construction process.

[0036] This bias has led technicians to never attempt to modularly combine environmental sensors with quality testing sensors, nor have they considered the linkage and traceability of carbon footprint data with construction parameters (such as raw material transportation and equipment energy consumption), making it difficult to conceive of a dual-dimensional testing system for quality and environmental protection.

[0037] In summary, the aforementioned technical pain points 1-3 prevent existing technologies from providing a multi-dimensional, closed-loop technological foundation. Even if technical personnel wish to make breakthroughs, they lack feasible technical modules to support them. Furthermore, the aforementioned technical biases 4-6 cause technical personnel, even when exposed to cross-dimensional technologies, to subjectively reject innovative directions such as the integration of quality and environmental protection, or the combination of detection and prediction, thus falling into fixed mindsets. These two factors reinforce each other, ensuring that those skilled in the art remain within a framework of single-dimensional, passive monitoring, and single-point optimization, making it difficult to transcend the cognitive boundaries of existing technologies and ultimately preventing them from conceiving the innovative technical solution proposed in this application that integrates quality and environmental protection in a dual-dimensional, closed-loop, and multi-technology manner.

[0038] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0039] A method for detecting fluidized solidified soil used in backfilling includes the following steps:

[0040] S1. Prefabrication and Calibration of the Detection System: Construct a modular sensor array, which includes a construction parameter module, a mechanical performance module, and a green and environmentally friendly module; complete sensor calibration through a dual process of factory calibration and on-site secondary calibration; and construct an AI model for CNN quality prediction.

[0041] S2. Green traceability and prediction before construction: The green environmental protection module is used to detect the environmental indicators of raw materials; the carbon footprint is calculated and emission reduction benchmark values ​​are set; the raw material parameters and construction environment parameters are input into the AI ​​model to predict the rationality of the ratio;

[0042] S3. Layered monitoring and dynamic verification during construction: Sensor arrays are arranged in layers according to the thickness of the backfill layer, and data is collected in real time through wireless transmission. Cross-verification is carried out in combination with dual-modal non-destructive testing equipment.

[0043] S4. Post-construction AI prediction and long-term tracking: Predict 28-day strength using CNN quality prediction model and strength prediction formula; evaluate green indicators; continuously monitor the long-term performance of solidified soil.

[0044] S5. Full-cycle data fusion and traceability: The cloud platform integrates multi-source data to generate visual reports; a unique engineering traceability code is established to link the entire process data.

[0045] 1. The core limitation of existing technologies is the independent separation of quality testing and environmental assessment, which cannot meet the dual control requirements under environmental protection regulations. This application achieves a breakthrough through three key design features:

[0046] (1) Modular compatible design: Innovatively create a magnetic sensor array that integrates construction parameter module, mechanical performance module and green environmental protection module, and physically compatible the flow, density and other quality sensors with environmental protection sensors such as VOCs, heavy metals and carbon footprint chips, solve the technical problem that different types of sensors are difficult to work together, and realize one-time deployment and simultaneous collection of multiple indicators.

[0047] (2) Data linkage and traceability: By recording raw material transportation and production energy consumption data through carbon footprint recording chips, and combining them with quality parameters in the construction process, a raw material-construction-environmental protection data chain is constructed, so that carbon footprint accounting and quality testing data are linked together, solving the problem of disconnect between environmental protection data and construction process.

[0048] (3) Dual-dimensional evaluation criteria: Establish dual qualification thresholds for quality indicators (strength, density, etc.) and environmental protection indicators (heavy metal content, carbon emission reduction). For example, the quality requirement is ≥3MPa compressive strength at 28 days, and the environmental protection requirement is Pb. 2+ ≤0.05mg / L, forming a complete two-dimensional acceptance system, filling the gap in existing environmental assessment technologies.

[0049] 2. Existing technologies lack a coordinated mechanism for pre-construction, construction, and post-construction, leading to passive risk control. This application achieves a closed loop through a four-stage process design:

[0050] (1) Pre-construction prediction: Input the raw material parameters and environmental conditions into the trained CNN model, and use the suspension comprehensive evaluation coefficient formula to predict the rationality of the ratio, adjust the parameters in advance, avoid quality risks from the source, and solve the problem of no prediction before construction.

[0051] (2) Monitoring during construction: Sensor arrays are arranged in layers, and data is transmitted wirelessly every 10 minutes. Combined with radar-elastic wave dual-mode detection cross-verification, the accuracy is improved by multi-dimensional data fusion correction formula, so as to realize real-time monitoring and dynamic adjustment and solve the problem of isolated monitoring during construction.

[0052] (3) Post-construction prediction: Using the CNN model and the 28-day intensity prediction formula, the long-term intensity can be predicted 7-14 days in advance, the influencing factors can be accurately located and the rectification plan can be generated, thus solving the passive dilemma of only being able to measure after construction.

[0053] (4) Full-cycle traceability: Establish a unique engineering traceability code to link data from raw material testing, construction monitoring, and environmental assessment throughout the entire process, so as to make problems traceable and responsibilities accountable, and solve the traceability difficulties caused by process fragmentation.

[0054] 3. Existing technologies struggle to balance detection accuracy and scenario adaptability. This application overcomes this contradiction through a dual design approach combining hardware optimization and data processing.

[0055] (1) Hardware protection adaptation: The sensor array shell is made of modified polypropylene material (IP68 waterproof), with added anti-collision protrusions and flexible cables. Deep nodes are equipped with anti-settlement fixing brackets to solve the problem of equipment damage caused by construction squeezing. It adopts dual power supply of solar energy and lithium battery, and signal repeaters are arranged at the edge to adapt to the power supply and signal transmission needs of deep and complex backfilling scenarios.

[0056] (2) Data accuracy assurance: The innovative factory calibration + on-site secondary calibration dual process corrects the deviation caused by water-cement ratio and ambient temperature through the sensor dynamic calibration formula; it integrates three types of data: sensor array, radar and elastic wave, and reduces the error of a single detection method through a weighted fusion formula, so that the density detection deviation is ≤±2%, taking into account both accuracy and adaptability.

[0057] (3) High-efficiency detection design: The dual-modal detection equipment scans at a grid of 1.5m×1.5m, with a detection efficiency of 30㎡ / minute. Combined with the layered and dense point layout strategy, it not only meets the high-efficiency detection needs of large-scale projects, but also ensures the detection accuracy of key parts by encrypting key areas.

[0058] 4. Those skilled in the art have long regarded testing as a post-event verification tool; this application reshapes the positioning of testing through technological innovation:

[0059] (1) Prediction model construction: Collect more than 1,000 sets of engineering data, use BorderlineSMOTE technology to balance the dataset, construct a CNN quality prediction model, combine the SHAP algorithm to analyze the influence weight of each indicator on quality, and achieve a prediction accuracy of ≥92%, realizing the transformation from post-verification to early prediction.

[0060] (2) Indirect index correlation: Break through the mindset that strength can only be measured, establish the correlation between shear wave velocity, wet density, water-cement ratio and 28-day strength, and realize the accurate conversion of indirect index to long-term strength through strength prediction formula, predict the results 7-14 days in advance, and reserve sufficient time for construction adjustment.

[0061] (3) Proactive prevention and control mechanism: Set multi-level early warning thresholds, trigger an early warning within 30 seconds of abnormal data, and automatically push handling guidelines in case of emergency (such as initiating grouting reinforcement when settlement exceeds the standard), so that the detection is upgraded from confirming the result to preventing and controlling the risk, and completely breaking the inherent positioning bias.

[0062] 5. Technical personnel often rely on a single method to achieve optimal results. This application achieves a breakthrough through the complementary use of multiple technologies:

[0063] (1) Fusion of sensing and detection: Combining the real-time performance of in-situ sensing arrays with the deep penetration of radar-elastic wave detection, the sensing array captures real-time surface data, and the dual-mode device detects deep defects, thus solving the limitations of single technology that is either shallow or slow.

[0064] (2) Hardware and algorithm integration: The hardware adopts a modular sensor array, and the software is equipped with a CNN prediction model and multiple sets of quantization formulas. The hardware provides the data foundation, and the algorithm realizes data value-added, such as reducing errors through data fusion formulas and realizing prediction through AI models, thus solving the problem of hardware and algorithm disconnect.

[0065] (3) Cost and benefit balance: The complexity of equipment is reduced by modular design, and the labor cost is reduced by wireless transmission and automatic calculation. Although multiple technologies are combined, the overall installation and maintenance cost is reduced by 40%, breaking the prejudice that multiple technologies are necessarily costly.

[0066] 6. Technical personnel generally believe that environmental protection indicators are additional requirements; this application achieves deep integration through technical design:

[0067] (1) Process-oriented environmental monitoring: The green environmental protection module monitors the release of VOCs and heavy metals in real time during the construction process, rather than only during the acceptance stage. It can promptly detect and deal with environmental protection issues, making environmental monitoring the core link of construction testing.

[0068] (2) Carbon footprint full-process accounting: Through the carbon footprint quantification formula, carbon emissions from raw material production, transportation, construction energy consumption and other links are included in the accounting, and a carbon emission reduction trend map is generated, so that the carbon footprint becomes a monitorable and controllable indicator in the construction process, rather than an independent environmental assessment item.

[0069] (3) Acceptance closed-loop design: The environmental protection acceptance pass rate and the quality qualification rate are used as the project acceptance standard. If the environmental protection indicators do not meet the standards, the project is unqualified. The environmental protection indicators are established as the core position of construction testing from the system design, breaking the prejudice that environmental protection is unrelated to construction.

[0070] 7. Technicians mistakenly believe that accuracy depends on equipment and ignore scene interference. This application achieves a breakthrough through systematic prevention and control:

[0071] (1) Construction interference response: Optimize the timing of sensor placement (bury after initial compaction), add anti-collision protrusions and flexible cables, and equip with anti-settlement brackets to transform construction squeezing, displacement and other interference factors into controllable variables and solve the bias of uncontrollable scene interference.

[0072] (2) Environmental interference correction: The influence of water-glue ratio and ambient temperature on the detection results is corrected by the sensor dynamic calibration formula; a low power consumption working mode and signal repeater are adopted to solve environmental interference such as power interruption and signal attenuation, so as to achieve stable accuracy despite environmental changes.

[0073] (3) Data reliability assurance: Through multi-source data cross-verification (combination of sensing, radar and elastic waves), dual calibration process and quantitative formula correction, even if a single device is interfered with, it can be compensated by other links, breaking the one-sided perception that the accuracy of the device is reliable.

[0074] In summary, the technical solution presented in this application is not a partial improvement of existing technologies, but rather, through innovative design involving modular integration, a closed-loop lifecycle, multi-technology collaboration, and systematic prevention and control, it fundamentally overcomes the core pain points and technological biases of existing technologies. Its core logic is: not limited to optimizing a single indicator or a single link, but rather constructing a complete system that is hardware-compatible, has a closed-loop process, adds value to data, and is risk-controllable, ultimately achieving dual control over quality and environmental protection, and simultaneous improvement in efficiency and precision, completely transcending the mindset and capability limitations of existing technologies.

[0075] Furthermore, in step S1, the construction parameter module integrates a flowability sensor, a suspension sensor, and a moisture content / temperature sensor;

[0076] The mechanical performance module incorporates a miniature compressive strength sensor and a shear wave velocity sensor.

[0077] The green and environmentally friendly module is equipped with a VOCs sensor, a heavy metal ion sensor, and a carbon footprint recording chip.

[0078] Further, in step S1, the on-site secondary calibration is as follows: before on-site construction, test blocks are made from the solidified soil used in the actual project, and the sensor parameters are calibrated simultaneously to dynamically match the detection benchmarks with different mix ratios; and the following sensor dynamic calibration formula is adopted: ; in, Water-to-binder ratio; t is ambient temperature, in °C; , Obtained through laboratory calibration.

[0079] Furthermore, in step S1, the steps for constructing the AI ​​model for CNN quality prediction include: (1) Physical calibration: standard test blocks with the same engineering ratio are made, and core index data of different ages are collected synchronously under standard curing environment to establish a basic index-mechanical performance correlation benchmark library; (2) Collect more than 1,000 sets of historical data from different engineering scenarios, use Borderline SMOTE technology to balance the dataset, construct a CNN quality prediction model, and use the SHAP algorithm to analyze the influence weight of each indicator on quality.

[0080] Furthermore, in step S2, the AI ​​model is used to predict the rationality of the mix ratio specifically by: predicting the probability of the fluidity and suspension of the solidified soil meeting the standards, and adjusting the mix ratio parameters in advance; Furthermore, in step S2, the following formula for the comprehensive evaluation coefficient of levitation is adopted: ; in, The coefficient of variation of wet density. The ratio of aggregate settling rates; Ks≤3.0 is considered acceptable.

[0081] Furthermore, in step S3, the sensor array is arranged as follows: the spacing in the normal area is 2m×2m, and the spacing around the pipeline and the edge of the foundation pit is increased to 1m×1m.

[0082] Furthermore, in step S3, data on fluidity, suspension, moisture content, and initial strength are collected every 10 minutes via LoRa-NB-IoT dual-mode wireless transmission and uploaded to the cloud platform. An automatic warning is issued when the deviation exceeds ±5%.

[0083] Furthermore, in step S3, the dual-mode non-destructive testing device is a radar-elastic wave detection device with a detection frequency of 1-5MHz and a depth of 0-8m, scanning in a 1.5m×1.5m grid.

[0084] Furthermore, in step S3, cross-validation employs a multi-dimensional detection data fusion correction formula: ; in, , , These are the density / compactness data for in-situ sensing array, radar detection, and elastic wave detection, respectively. , , All are weighting coefficients.

[0085] Further, in step S4, the intensity prediction formula is: ; in, This indicates the shear wave velocity, measured in km / s. This indicates wet density, in g / cm³. 3 ; Indicates the water-to-glue ratio; Indicates the curing temperature, in °C; Compressive strength, in MPa; , , All are coefficients, obtained through fitting engineering data.

[0086] Furthermore, in step S4, when evaluating green indicators, a carbon footprint quantification formula is used: ; in, The carbon footprint of a material is expressed by the following formula: f is the carbon emission factor per unit mass, and m is the mass; The carbon footprint of transportation is expressed by the following formula: ,in For material quality, For transportation distance, Carbon intensity of transportation; The energy consumption carbon footprint is represented by the following formula: Where P is the equipment power and t is the working time. It represents the carbon emission intensity per unit of electrical energy.

[0087] Furthermore, in step S4, continuous monitoring of the long-term performance of the solidified soil includes: the sensor array working continuously to focus on tracking the strength decay rate, strain change, and release trend of harmful substances in the solidified soil; setting multi-level early warning thresholds; issuing SMS and / or APP warnings within 30 seconds when abnormalities occur; and automatically triggering on-site audible and visual alarms and simultaneously pushing out disposal guidelines when the emergency threshold is reached.

[0088] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0089] (1) This invention designs a modular sensor array that integrates construction parameter modules, mechanical performance modules, and green environmental protection modules. This achieves physical compatibility and synchronous data acquisition, and establishes a dual-dimensional qualification threshold and a full-process data chain. It links and traces the environmental testing of raw materials, construction quality monitoring, and carbon footprint accounting, forming an integrated quality and environmental protection acceptance standard. This fills the gap in existing environmental assessment technologies and enables full-process monitoring of the release of harmful substances such as heavy metals and VOCs and the effectiveness of carbon emission reduction. At the same time, it avoids the disconnect between quality and environmental testing, and multiple indicators can be obtained with a single deployment, reducing the cost of repeated testing and ensuring high acceptance by third parties.

[0090] (2) Before construction, this invention uses a CNN model and a comprehensive evaluation coefficient formula for suspension to predict the rationality of the mix design, thus avoiding risks at the source. During construction, a layered layout combined with dual-modal non-destructive testing is used for cross-verification, and the formula is improved by multi-dimensional data fusion to correct the error. After construction, the 28-day strength prediction formula is used to predict long-term performance 7-14 days in advance, combined with 2-year long-term monitoring and a multi-level early warning mechanism. A unique engineering traceability code is established, linking the entire process data to achieve traceability. This breaks the limitations of existing single-point detection technology, forming a closed loop of prediction-monitoring-forecasting-traceability, transforming passive remediation into proactive prevention and control. Furthermore, the abnormal response time is ≤30 seconds, and emergency response guidelines are automatically pushed, effectively avoiding engineering accidents such as settlement and cracking in the later stages. It also enables full-process data traceability, meeting the needs of engineering acceptance and accountability, and improving the standardization of project management.

[0091] (3) This invention adopts a dual process of factory calibration and on-site secondary calibration, and corrects deviations caused by water-cement ratio and ambient temperature through a dynamic calibration formula for sensors; it integrates three types of data: sensor array, radar, and elastic wave, and reduces single detection errors through a weighted fusion formula, with a density detection deviation of ≤±2%; the sensor array shell is made of modified polypropylene, with added anti-collision protrusions and flexible cables, and deep nodes are equipped with anti-settlement supports, using dual power supply of solar energy and lithium batteries and signal repeaters; the deployment method is adaptable to different scenarios. The detection accuracy of this invention is significantly better than that of the prior art, the data accuracy is not affected by the environment and construction, and the sensor damage rate is 0%; it is suitable for various backfilling scenarios such as foundation pits, pipe trenches, and mine pits, especially meeting the detection needs of deep and complex scenarios, and reducing installation and maintenance costs by 40%; each layer of detection only takes 30 minutes, which is 75% more efficient than the prior art, and is suitable for the efficient management and control needs of large-scale projects.

[0092] (4) This invention trains a CNN quality prediction model based on more than 1,000 sets of engineering data, and combines the SHAP algorithm to analyze the index weights; it quantifies the index through formulas such as the comprehensive evaluation coefficient of suspension, intensity prediction, and carbon footprint quantification, and the cloud platform automatically integrates the data to generate a visual report; it wirelessly transmits data in real time, and automatically warns when the deviation exceeds ±5%, without the need for manual calculation. This invention breaks through the limitation that intensity can only be measured in practice, predicts long-term performance in advance, reserves sufficient time for construction adjustments, and avoids rework losses; it also reduces the workload of manual operation and data processing, reduces decision-making time by 50%, and reduces the risk of human error; the visual report intuitively presents the dual-dimensional status of quality and environmental protection, which facilitates quick and accurate decision-making.

[0093] (5) This invention quantifies carbon emission reduction effectiveness to form a green construction assessment report, which is compatible with environmental protection verification requirements; the standardized implementation process and modular design can flexibly adapt to the testing of fluidized solidified soil with different ratios and scenarios. This invention responds to the trend of environmental protection upgrading, helps the green transformation of engineering projects, and enhances the core competitiveness of projects; it is also mature and reliable, and can be directly applied to key projects such as backfilling and ecological restoration around subway stations. It has strong promotion potential and can effectively solve the core pain points of existing technologies such as fragmentation, passivity, and non-greenness.

[0094] In summary, this invention achieves precise quality and environmental protection management of fluidized solidified soil through modular sensor arrays, a full-cycle closed-loop process, multi-technology integration, and systematic risk control design. It overcomes the limitations of existing technologies in terms of limited detection dimensions and fragmented processes, while also considering environmental requirements. It enables full-process monitoring and traceability of heavy metals, VOCs, and carbon footprint, adapting to various backfilling scenarios, transforming passive remediation into proactive prevention, and comprehensively meeting the needs of engineering quality acceptance and environmental verification, demonstrating significant potential for widespread application.

[0095] The present invention will be explained in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0096] Figure 1 This is a flowchart of the steps of the present invention. Detailed Implementation

[0097] To facilitate understanding of the present invention, a more comprehensive description of the present invention will be given below with reference to the accompanying drawings, which illustrate several embodiments of the present invention. However, the present invention can be implemented in different forms and is not limited to the embodiments described in the text. Rather, these embodiments are provided to make the disclosure of the present invention more thorough and complete.

[0098] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly associated with those skilled in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0099] Please refer to the attached document carefully. Figure 1 A method for detecting fluidized solidified soil used for backfilling includes the following steps:

[0100] I. Prefabrication and Calibration of the Detection System

[0101] 1. Modular sensor array design: Creating a combinable sensor unit, the core of which includes three types of modules:

[0102] (1) Construction parameter module: Integrated: Flowability sensor, measuring range of 50-300mm, accuracy of ±2mm; Suspension sensor, used to monitor aggregate settling rate, accuracy of ±0.01mm / h; Moisture content / temperature sensor;

[0103] (2) Mechanical performance module: Built-in miniature compressive strength sensor and shear wave velocity sensor, used to capture the strength development and density change of solidified soil in real time;

[0104] (3) Green and environmentally friendly module: Equipped with VOCs sensor and heavy metal ion sensor for synchronous monitoring of the release of harmful substances in solidified soil; and integrated with carbon footprint recording chip for storing data such as raw material transportation and energy consumption.

[0105] All the above modules are magnetically attached, and the outer shell is made of corrosion-resistant and wear-resistant modified polypropylene material with a waterproof rating of IP68 to adapt to different backfilling scenarios.

[0106] Considering the differences in the properties of solidified soil with different mix proportions in various projects, which may lead to inaccurate sensor calibration benchmarks, a dual-process approach of factory calibration and on-site secondary calibration is adopted to address this deviation risk. At the factory, standard test blocks are prepared according to different project mix proportions to establish a basic calibration library. Before on-site construction, test blocks are prepared using solidified soil from the actual project, and sensor parameters are calibrated simultaneously to dynamically match the detection benchmarks for different mix proportions. In the on-site calibration stage, the following dynamic sensor calibration formula is used to address calibration deviations caused by mix proportion differences: ; By water-to-glue ratio ( The sensor readings are corrected using ambient temperature (t, in °C) and obtained through laboratory calibration. =0.3、 =0.02.

[0107] Correction logic: Original sensor reading × The actual values ​​on site were obtained.

[0108] 2. Dual-model calibration and training:

[0109] (1) Physical calibration: Prepare standard test blocks with the same engineering ratio, and collect core index data (including fluidity, suspension, strength, wave velocity, etc.) at different ages (e.g., 1d, 3d, 7d, 28d) under standard curing environment, and establish a basic index-mechanical performance correlation benchmark library;

[0110] (2) AI model training: Collect more than 1,000 sets of historical data (including qualified and unqualified samples) from different engineering scenarios, use Borderline SMOTE technology to balance the dataset, construct a CNN quality prediction model, and use the SHAP algorithm to analyze the influence weight of each indicator on quality. The model prediction accuracy is ≥92%.

[0111] II. Green Source Tracing and Prediction Before Construction

[0112] 1. Environmental testing of raw materials: Pre-treatment testing is carried out on raw materials such as curing agents and aggregates. The green environmental protection module of the sensor array is used to detect the heavy metal content and VOCs emission in the curing agent to ensure compliance with GB / T 5085.3-2007 standard; unqualified raw materials are directly rejected to avoid environmental risks from the source.

[0113] 2. Carbon footprint pre-calculation: The carbon footprint recording chip records the raw material transportation distance and production energy consumption data, and calculates the estimated carbon emissions in combination with the construction plan. The carbon emission reduction benchmark value is set with reference to the carbon emission reduction standard of "on-site transformation" technology.

[0114] 3. Mixing ratio prediction: Input raw material parameters and construction environment parameters (temperature, humidity) into the AI ​​model to predict the probability of the fluidity and suspension of the solidified soil meeting the standards, and adjust the mixing ratio parameters in advance to avoid rework after construction.

[0115] The slurry mix ratio was verified before construction using the following formula for the comprehensive evaluation coefficient of suspension: ; Combined with wet density variation coefficient ( The ratio of aggregate settling rate to aggregate settling rate ( A comprehensive evaluation of levitation.

[0116] Value range: Ks≤3.0 is considered acceptable. =0.01mm / h.

[0117] The above formula can also be applied to early warning of segregation risks during construction.

[0118] III. Layered Monitoring and Dynamic Verification during Construction

[0119] 1. Sensor array layout: The sensor array is arranged in layers according to the backfill layer thickness of 30-50cm. The array is combined with key densification method: the ordinary area is arranged at a spacing of 2m×2m, and the key areas such as the pipeline and the edge of the foundation pit are densified to 1m×1m. The array is buried at the same time as the backfill layer to ensure that the sensor is in full contact with the solidified soil.

[0120] Considering the potential damage to the sensor array during backfilling, the timing of the placement was optimized to address this risk. The sensors were installed after each layer of backfill soil was initially compacted. Furthermore, anti-collision protrusions were added to the sensor housing, and flexible connecting cables were used to reduce the probability of crushing damage. Deep monitoring nodes were equipped with anti-settlement fixing brackets to prevent displacement.

[0121] 2. Real-time data acquisition: Through LoRa-NB-IoT dual-mode wireless transmission, data such as fluidity, suspension, moisture content, and initial strength are collected every 10 minutes and uploaded to the cloud platform. The data is compared with the calibration benchmark value in real time, and an automatic warning is issued when the deviation exceeds ±5%.

[0122] To address the challenge of powering sensor arrays in long-term monitoring scenarios, a combination of solar power and high-capacity lithium batteries is employed. Shallow nodes are equipped with small solar panels, while deep nodes utilize lithium batteries with a lifespan of ≥2 years. A low-power operating mode is designed, reducing the acquisition frequency during non-critical periods. To address the issue of weak signals, signal repeaters are deployed at the edges of the backfill area to ensure stable data transmission.

[0123] 3. Dual-modal non-destructive testing: 24 hours after each backfilling layer is completed, a mobile radar-elastic wave detection device is used with a detection frequency of 1-5MHz and a depth of 0-8m. The device scans the backfill in a 1.5m×1.5m grid. The radar detects voids and segregation defects in the backfill, while the elastic wave detects the density and deep strength. The detection efficiency reaches 30㎡ / minute. The data is cross-validated with the sensor array data. The allowable deviation range is ±8%. If the deviation exceeds the range, a secondary detection is initiated to investigate the cause of the defect and take timely measures such as refilling and compaction.

[0124] During cross-verification of compaction during construction, the following multi-dimensional detection data fusion correction formula is used to ensure data accuracy (error ≤ ±5%): ; The above formula is used to fuse in-situ sensing arrays ( ), radar detection ( Elastic wave detection () Density / compactness data; weight settings: =0.5 (real-time performance of in-situ sensing) =0.3 (radar deep detection) =0.2 (elastic wave mechanics verification), which can be dynamically adjusted according to the scenario.

[0125] IV. Post-construction AI Prediction and Long-Term Tracking

[0126] 1. Intelligent quality prediction: Input the real-time data collected during construction into the trained CNN quality prediction model to predict the 28-day strength and long-term stability. If the prediction is unqualified, accurately locate the influencing factors (such as insufficient suspension and excessive moisture content) and generate targeted rectification plans (such as adjusting curing conditions and local reinforcement).

[0127] The CNN-SHAP model outputs the predicted strength of solidified soil at 28 days, and the corresponding formula is as follows: ; By shearing wave velocity ( (unit: km / s), wet density ( (unit: g / cm³), water-cement ratio () ), curing temperature ( Predicted compressive strength (unit: °C) (Unit: MPa)

[0128] Coefficient calibration: obtained by fitting more than 1000 sets of engineering data. =3.2、 =1.8、 =0.05.

[0129] Application scenario: Predict the strength 7-14 days after construction to avoid rework later.

[0130] 2. Green Indicator Assessment: By combining the carbon footprint data recorded by the carbon footprint recording chip with the actual energy consumption during construction, the actual carbon emission reduction is calculated and compared with the estimated benchmark value to generate a green construction assessment report.

[0131] The formula for quantifying carbon footprint is: ;

[0132] Itemized calculations:

[0133] Material carbon footprint: f is the carbon emission factor per unit mass, m is the mass, soil is taken as 0.05 kg CO2 / kg, and solidifying agent is taken as 0.8 kg CO2 / kg;

[0134] Transportation carbon footprint: ,in For material quality, For transportation distance, Carbon intensity of transportation;

[0135] Energy consumption carbon footprint: Where P is the equipment power and t is the working time. The carbon emission intensity per unit of electricity can be expressed as 0.6 kg CO2 / (kWh).

[0136] 3. Long-term health monitoring: The sensor array works continuously, and the monitoring cycle can be extended to 2 years. It focuses on tracking the strength decay rate, strain changes, and release trends of harmful substances in the solidified soil. It sets multi-level early warning thresholds (such as strength decay exceeding 10% or heavy metal ion exceeding the standard). In case of abnormality, it sends SMS and / or APP warnings within 30 seconds. When the emergency threshold is reached, it automatically triggers on-site audible and visual alarms and pushes disposal guidelines simultaneously (such as initiating grouting reinforcement when settlement exceeds the standard).

[0137] V. Full-cycle data fusion and traceability

[0138] 1. Cloud Data Integration: The cloud platform uses a weighted fusion algorithm to integrate three types of data: sensor array, non-destructive verification, and AI prediction, to generate a dual-dimensional visualization report on quality and environmental protection, including real-time monitoring curves, defect distribution heat maps, carbon emission reduction trend maps, etc.

[0139] 2. Full-process traceability management: Establish a unique project traceability code, linking raw material testing data, construction process data, quality assessment reports, carbon footprint records, etc., to support full-process traceability and meet the needs of project acceptance and environmental verification.

[0140] Example: Backfilling project for a foundation pit near a subway station. The backfilling depth is 5m and the backfilling volume is 800m³. 3 The mixing ratio of the fluidized solidified soil is: 15kg soil, 1.8kg solidifying agent, and 7.2kg water. It requires a 28-day compressive strength ≥3MPa and a density ≥95%, while also meeting environmental protection requirements (heavy metal Pb). 2+ ≤0.05mg / L).

[0141] The implementation steps are as follows:

[0142] I. Prefabrication and Calibration of the Detection System

[0143] Modular sensor array: Combining construction parameter modules (flowability, suspension, moisture content / temperature sensors), mechanical property modules (miniature compressive strength, shear wave velocity sensors), and green environmental protection modules (VOCs, Pb). 2+ Sensors, carbon footprint chips).

[0144] Dual calibration process: At the factory, standard test blocks are manufactured according to the engineering mix proportions to establish a basic calibration library; on-site, actual solidified soil is used to prepare test blocks, and the parameters are corrected using the sensor dynamic calibration formula (α=0.3, β=0.02, water-cement ratio 0.4, ambient temperature 25℃, K...). 标定= 1 + 0.3 × (0.4 - 0.5) + 0.02 × (25 - 20) = 1.07).

[0145] AI model training: Using 1200 sets of engineering data (including qualified and unqualified samples), the data was balanced using Borderline SMOTE to build a CNN model. The SHAP algorithm was used to parse the weights, and the prediction accuracy was 93%.

[0146] II. Green Source Tracing and Prediction Before Construction

[0147] Raw material testing: Testing for Pb in the curing agent 2+ The content was 0.03 mg / L, and the VOCs release was 0.02 mg / m³. 3 It conforms to the GB / T5085.3-2007 standard.

[0148] Carbon footprint pre-calculation: CF 材料 =15×0.05+1.8×0.8=2.29kgCO2; CF 运输 =(15×10×0.002)+(1.8×50×0.002)=0.36kgCO2; CF 能耗 = 5kW × 8h × 0.6 = 24kgCO2; Estimated CF 总 =26.65kgCO2, with a carbon emission reduction baseline of 26kgCO2.

[0149] Proportion prediction: Input into the AI ​​model to predict the levitation K. s =2.1 (≤3.0, qualified), fluidity is 180mm (meets requirements), no adjustment of ratio is required.

[0150] III. Layered Monitoring and Dynamic Verification during Construction

[0151] Sensor array layout: layered at 30cm intervals, with 2m×2m points in ordinary areas and 1m×1m densification around pipelines. After initial compaction, the sensors are buried with anti-collision protrusions on the outer shell and flexible cable connections.

[0152] Real-time data acquisition: LoRa-NB-IoT dual-mode transmission, data is collected once every 10 minutes, with a moisture content of 18% and an initial intensity of 0.8MPa, with a deviation of ≤±3% and no warning; powered by solar energy and lithium batteries, with signal repeaters deployed at the edge, ensuring stable transmission.

[0153] Dual-modal non-destructive testing: 24 hours after each backfill layer, a radar-elastic wave detection device scans at a depth of 5m (1.5m × 1.5m). The density data is then fused and corrected with the sensor array. =0.5, =0.3, =0.2, = (0.5×1.75+0.3×1.73+0.2×1.74)=1.74g / cm 3 The deviation is ±2%, which meets the requirements.

[0154] IV. Post-construction AI Prediction and Long-Term Tracking

[0155] Quality prediction: 10 days after construction, input data into the CNN quality prediction model, and calculate using the 28-day intensity prediction formula. =1.2km / s =1.74g / cm 3 , =0.4、t 养护 = 20℃, =3.2 × 1.2 2 ×1.74+1.8×0.4-0.05×20=3.5MPa,≥3MPa), the prediction is qualified.

[0156] Green Indicator Assessment: Actual CF 总 =25.8kgCO2, carbon emission reduction of 0.85kgCO2, which meets the green requirements.

[0157] Long-term monitoring: Continuous monitoring for 2 years, intensity decreased by 8%, Pb 2+ The concentration was 0.028 mg / L, and no abnormality warning was issued.

[0158] V. Full-cycle data fusion and traceability

[0159] The cloud generates quality and environmental protection reports, and the defect distribution heat map shows no anomalies; a traceability code is established to link data from the entire process of raw materials, construction, and monitoring.

[0160] The implementation results are as follows:

[0161] 28. The actual compressive strength is 3.6 MPa, the density is 96.2%, the heavy metal content meets the standard, the carbon emission reduction target is achieved, the project is accepted and there are no problems with subsequent settlement or cracking.

[0162] Comparative Example 1: Referring to the existing technology disclosed in CN119757115B, a fluidized solidified soil compaction detection robot is used, which only detects compaction and has no suspension, environmental protection, or AI prediction functions; data is obtained through soil sampling and sand filling method, and the results are manually calculated.

[0163] Implementation process: The robot took soil samples at 2m×2m intervals, and the density was calculated by sand cone method. Each layer took 2 hours to test. Data was recorded manually, and there was no real-time warning. Heavy metals and carbon footprint were not tested. The strength was measured 30 days after construction.

[0164] Comparative Example 2: Compared to the example, it lacks the AI ​​prediction and formula quantification module.

[0165] The modular sensor array, dual-modal detection, and green traceability are retained, but the CNN-SHAP AI model is removed. Formulas such as intensity prediction and data fusion are not used. The qualification is judged only by real-time sensor data.

[0166] Implementation process: During construction, data such as fluidity and density are monitored, and an alert is issued when the deviation exceeds ±5%; there is no AI prediction function, and the strength is measured 28 days after construction; the data is output directly without integration and correction.

[0167] Comparative Example 3: Compared with the embodiment, it lacks a green and environmentally friendly module and a risk prevention and control mechanism.

[0168] The quality testing modules (sensor array, AI prediction, dual-modal detection) are retained, while the green and environmentally friendly modules (VOCs-free, heavy metal-free sensors and carbon footprint chips) are removed. Risk control measures such as sensor anti-collision, dual power supply, and dual calibration are not implemented.

[0169] Implementation process: No heavy metal release was detected during construction; sensors were directly buried (without initial compaction), and 3 sets of sensors were damaged due to pressure during construction; deep nodes were powered only by lithium batteries, and 2 sets stopped working due to power depletion; no carbon footprint accounting was performed.

[0170] The core indicators of the above embodiments and comparative examples one to three are shown in Table 1:

[0171] Table 1. Comparison of core indicators between the examples and comparative examples one to three.

[0172] Comparison indicators Example Comparative Example 1 Comparative Example 2 Comparative Example 3 28-day strength pass rate 100% 85% 90% 0% rework Density detection deviation ±2% ±6% ±4.5% ±3% (data missing) Time required for each layer of detection 30min 2h 1h 45min Sensor failure rate 0% 5% 0% 25% Environmental protection acceptance rate 100% 0% (No environmental data available) 0% (No environmental data available) 0% (Heavy metals exceed standards) Carbon emission reduction achievement rate 100% - (Not calculated) - (Not calculated) - (Not calculated)

[0173] As can be seen from Table 1:

[0174] Differences between Comparative Example 1 and the Implementation Example:

[0175] Efficiency: Each layer takes 2 hours to test, which is 75% lower than the example (30 minutes / layer);

[0176] Accuracy: Density detection deviation ±6%, higher than the ±2% of the example;

[0177] Risk: Intensity not predicted; insufficient levitation in a certain area (K) s =3.8), resulting in a strength of 2.7MPa (unacceptable), requiring rework;

[0178] Environmental protection: The heavy metal and carbon footprint were not assessed, and the environmental protection acceptance requirements could not be met.

[0179] Differences between Comparative Example 2 and the Implementation Example:

[0180] Predictive capability: Unable to predict strength in advance; in a certain area, due to low curing temperature (15℃), the 28-day strength was 2.8MPa (unacceptable), requiring local reinforcement, which increased the additional cost by 15%.

[0181] Data accuracy: Without data fusion correction, the density detection deviation was ±4.5%, higher than the ±2% of the previous example;

[0182] Decision-making efficiency: There is no formula for quantitative evaluation, and floating and green indicators need to be manually calculated, which doubles the decision-making time.

[0183] Differences between Comparative Example 3 and the Implementation Example:

[0184] Environmental risk: Pb testing of solidified soil during acceptance. 2+ The content was 0.06 mg / L (exceeding the standard), requiring all work to be reworked, resulting in significant losses;

[0185] Equipment reliability: Sensor failure rate 25%, data missing rate 18%, insufficient monitoring integrity;

[0186] Green assessment: Unable to provide carbon footprint data, project acceptance is delayed.

[0187] The comparison shows that the present invention has the following advantages:

[0188] 1. Comprehensive testing system, balancing quality and environmental friendliness.

[0189] (1) Construct a dual-dimensional testing system for quality and environmental protection, which covers core quality indicators such as fluidity, suspension, density and strength, and also includes environmental protection indicators such as heavy metals, VOCs and carbon footprint, so as to fully meet the requirements of project acceptance and policy.

[0190] (2) The indicators are quantified through targeted calculation formulas, avoiding qualitative and vague judgments. The data is traceable and verifiable, and the acceptance rate of third-party acceptance is high.

[0191] 2. Intelligent prediction and accurate risk control

[0192] (1) By integrating the CNN-SHAP AI model with the intensity prediction formula, the intensity at 28 days of age can be predicted 7-14 days in advance with a prediction accuracy of ≥92%. Parameters can be adjusted in time during the construction phase to avoid rework in the later stage, and the rectification cost is 0.

[0193] (2) Establish a multi-level early warning mechanism with an abnormal response time of ≤30 seconds, shifting from passive remediation to proactive prevention and control, and significantly improving the reliability of project quality.

[0194] 3. Highly efficient and reliable implementation with wide adaptability

[0195] (1) The modular sensor array combined with dual-mode non-destructive detection requires only 30 minutes to detect each layer, which is 75% more efficient than existing technologies, and the sensor failure rate is 0, and the data transmission is stable.

[0196] (2) Embedded with a full-process risk control mechanism such as dual calibration, anti-collision protection, and dual power supply, the density detection deviation is ≤±2%, which is suitable for various backfilling scenarios such as foundation pit, pipe trench, and mine pit, and the installation and maintenance cost is reduced by 40%.

[0197] 4. Green and environmentally friendly, with high promotional value.

[0198] (1) Quantify the carbon emission reduction effect, meet the environmental protection verification requirements, and conform to the green development trend of the industry.

[0199] (2) The full-process data traceability and standardized implementation process can be directly applied to key projects, has strong promotion potential, and can effectively solve the core pain points of existing technologies such as fragmentation, passivity and non-greenness.

[0200] The present invention has been described by way of example in conjunction with the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvement made by adopting the inventive concept and technical solution of the present invention, or the direct application of the inventive concept and technical solution of the present invention to other occasions without modification, shall be within the protection scope of the present invention.

Claims

1. A method for detecting fluidized solidified soil used in backfilling, characterized in that: Includes the following steps: S1. Prefabrication and Calibration of the Detection System: Construct a modular sensor array, which includes a construction parameter module, a mechanical performance module, and a green and environmentally friendly module; complete sensor calibration through a dual process of factory calibration and on-site secondary calibration; and construct an AI model for CNN quality prediction. S2. Green traceability and prediction before construction: The green environmental protection module is used to detect the environmental indicators of raw materials; the carbon footprint is calculated and emission reduction benchmark values ​​are set; the raw material parameters and construction environment parameters are input into the AI ​​model to predict the rationality of the ratio; S3. Layered monitoring and dynamic verification during construction: Sensor arrays are arranged in layers according to the thickness of the backfill layer, and data is collected in real time through wireless transmission. Cross-verification is carried out in combination with dual-modal non-destructive testing equipment. S4. Post-construction AI prediction and long-term tracking: Predict 28-day strength using CNN quality prediction model and strength prediction formula; evaluate green indicators; continuously monitor the long-term performance of solidified soil. S5. Full-cycle data fusion and traceability: The cloud platform integrates multi-source data to generate visual reports; a unique engineering traceability code is established to link the entire process data.

2. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S1, the construction parameter module integrates a flowability sensor, a suspension sensor, and a moisture content / temperature sensor; And / or, the mechanical performance module incorporates a miniature compressive strength sensor and a shear wave velocity sensor; And / or, the green and environmentally friendly module is equipped with a VOCs sensor, a heavy metal ion sensor, and a carbon footprint recording chip.

3. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S1, the on-site secondary calibration is as follows: Before on-site construction, test blocks are made from the solidified soil used in the actual project, and the sensor parameters are calibrated simultaneously to dynamically match the detection benchmarks with different mix ratios; and the following sensor dynamic calibration formula is used: ； in, Water-to-binder ratio; t is ambient temperature, in °C; , Obtained through laboratory calibration.

4. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: Step S1, the steps for constructing the AI ​​model for CNN quality prediction include: (1) Physical calibration: standard test blocks with the same engineering ratio are made, and core index data of different ages are collected synchronously under standard curing environment to establish a basic index-mechanical performance correlation benchmark library; (2) Collect more than 1,000 sets of historical data from different engineering scenarios, use Borderline SMOTE technology to balance the dataset, construct a CNN quality prediction model, and use the SHAP algorithm to analyze the influence weight of each indicator on quality.

5. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S2, the AI ​​model is used to predict the rationality of the mix ratio. Specifically, the mix ratio parameters are adjusted in advance by predicting the probability that the fluidity and suspension of the solidified soil will meet the standards. And / or, in step S2, the following formula for the comprehensive evaluation coefficient of levitation is used: ； in, The coefficient of variation of wet density. The ratio of aggregate settling rates; Ks≤3.0 is considered acceptable.

6. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S3, the sensor array is arranged as follows: the spacing in the normal area is 2m×2m, and the spacing around the pipeline and the edge of the foundation pit is increased to 1m×1m. And / or, in step S3, data on fluidity, suspension, moisture content, and initial strength are collected every 10 minutes via LoRa-NB-IoT dual-mode wireless transmission and uploaded to the cloud platform. An automatic warning is issued when the deviation exceeds ±5%.

7. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S3, the dual-mode non-destructive testing device is a radar-elastic wave detection device with a detection frequency of 1-5MHz and a depth of 0-8m, scanning in a 1.5m×1.5m grid. And / or, cross-validation uses a multi-dimensional detection data fusion correction formula: ； in, , , These are the density / compactness data for in-situ sensing array, radar detection, and elastic wave detection, respectively. , , All are weighting coefficients.

8. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S4, the intensity prediction formula is: ； in, This indicates the shear wave velocity, measured in km / s. This indicates wet density, in g / cm³. 3 ; Indicates the water-to-glue ratio; Indicates the curing temperature, in °C; Compressive strength, in MPa; , , All are coefficients, obtained through fitting engineering data.

9. The method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S4, when evaluating green indicators, a carbon footprint quantification formula is used: ； in, The carbon footprint of a material is expressed by the following formula: f is the carbon emission factor per unit mass, and m is the mass; The carbon footprint of transportation is expressed by the following formula: ,in For material quality, For transportation distance, Carbon intensity of transportation; The energy consumption carbon footprint is represented by the following formula: Where P is the equipment power and t is the working time. It represents the carbon emission intensity per unit of electrical energy.

10. A method for detecting fluidized solidified soil for backfilling according to claim 1, characterized in that: In step S4, continuous monitoring of the long-term performance of solidified soil includes: continuous operation of the sensor array, focusing on tracking the strength decay rate, strain change, and release trend of harmful substances in solidified soil, setting multi-level early warning thresholds, and issuing SMS and / or APP warnings within 30 seconds when abnormalities occur; when the emergency threshold is reached, on-site audible and visual alarms are automatically triggered, and disposal guidelines are pushed out simultaneously.

Images