Curing agent batching system and method based on multi-source mud fingerprint

Abstract

The present application relates to a kind of based on multi-source mud fingerprint curing agent batching system and method, comprising: mud feature recognition unit extracts feature vector, forms unique fingerprint;Decision control unit is built-in multi-source mud fingerprint database and formula inversion model, matching generation solidification formula;Metering unit is according to formula and is divided into two stages of physical granulation and chemical solidification and feeds;Feedback type stirring unit is monitored by torque sensor torque and change rate;Decision control unit compares measured torque with preset target interval, dynamically controls additional material and stirring time.The present application is characterized in that: for extreme mud, such as ultra-high water content, high organic matter, high clay, a kind of can realize " one mud one strategy " automatic batching, two-stage solidification of sequence and torque feedback closed-loop control complete system and method are proposed.

Description

Technical Field

[0001] This invention relates to the intersection of solid waste resource utilization and intelligent manufacturing, specifically to a solidifying agent formulation system and method for complex muds from multiple sources, such as river and lake dredging sludge, subway tunnel slurry, and building pile foundation slurry. Background Technology

[0002] With the acceleration of urbanization, engineering activities such as river dredging, subway tunneling, and building pile foundation construction generate a large amount of engineering mud. This type of mud typically exhibits "three highs and one low" engineering characteristics: high water content, high clay content, high organic matter content, and low bearing capacity. Furthermore, its sources are complex and its properties fluctuate significantly. For example, river and lake sediments often contain organic components such as humic acid, which significantly inhibits cement hydration; tunneling mud is affected by geological changes, resulting in large fluctuations in sand content and clay ratio; building pile foundation mud is highly viscous, exhibits significant thixotropy, and its water content often exceeds the liquid limit. Current mud solidification methods mostly employ fixed formulas or minor adjustments based on experience using cement and other cementing materials. Because the properties of the incoming mud fluctuate in real time with varying operating conditions, fixed formulas are difficult to match the current mud characteristics. This can easily lead to insufficient solidifier dosage, resulting in substandard strength, rework, or even abandonment; or excessive addition to ensure compliance, causing a surge in costs, increased carbon emissions, and a higher risk of high-alkalinity leaching. Meanwhile, traditional mixing equipment lacks real-time perception of material states and cannot quantify the transformation of slurry from a fluid state to a formable state and the extent of solidification reaction. Therefore, there is an urgent need for an intelligent batching system and method that can identify slurry characteristics online, automatically generate the optimal formula, and implement closed-loop correction based on the mixing state, in order to achieve stable and low-carbon utilization of slurry resources. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a solidifier batching system and method based on multi-source mud fingerprint spectrum to solve the problems of "difficulty in perceiving properties, difficulty in matching formulas, and difficulty in controlling processes" in existing mud solidification treatment. In particular, for extreme muds with ultra-high water content, high organic matter, and high viscosity, a complete system and method is proposed that can realize automatic batching of "one mud, one policy", two-stage sequential solidification and torque feedback closed-loop control.

[0004] To achieve the above objectives, a solidifier batching system based on multi-source mud fingerprinting is designed, comprising: a mud feature identification unit, located in the detection section of the mud inlet pipe, integrating a near-infrared spectroscopy probe and preprocessing components, used to extract feature vectors of the mud to form a unique fingerprint spectrum of the mud, forming an input variable in the solidifier batching system; and a decision control unit, signal-connected to the mud feature identification unit, containing a multi-source mud fingerprint database and a formulation inversion model. The multi-source mud fingerprint database stores fingerprint features of mud from different sources, historical optimal formulations, and target performance indicators. The formulation inversion model is used for... A solidification formula for the current mud is generated by matching the unique fingerprint spectrum of the mud with a multi-source mud fingerprint database; a metering and batching unit, connected to the decision control unit, is used to add solidification materials in two stages, physical granulation and chemical solidification, according to the solidification formula; a feedback stirring unit includes a stirring device and a torque sensor, the torque sensor being used to monitor the torque and torque change rate of the stirring shaft in the stirring device in real time; the decision control unit is also configured to compare the torque and torque change rate with a preset target torque range and evaluation time window, and control the addition of solidification materials and adjust the stirring time according to the comparison results.

[0005] Preferably, the present invention further includes the following technical solution: the near-infrared spectral probe is used to scan the slurry flowing through the detection section in the 900nm to 2500nm band of the infrared spectrum to obtain the raw spectral data of the slurry; the preprocessing component is signal-connected to the near-infrared spectral probe and is used to preprocess the raw spectral data to extract feature vectors, and to analyze multiple property parameters of the slurry based on the feature vectors; the multiple property parameters include absolute water content, organic matter content, and particle composition parameters.

[0006] Preferably, the present invention further includes the following technical solutions: the mud feature identification unit further includes: a moisture fingerprint analysis component, configured to calculate the absolute moisture content of the mud by identifying the hydroxyl bond overtone absorption peaks of water molecules at wavelengths of 1450 nm and 1940 nm in the infrared spectrum of the mud, and combining it with a partial least squares regression model or a neural network regression model; an organic matter fingerprint analysis component, configured to calculate the organic matter content in the mud by extracting the characteristic absorption information in the 1600 nm to 1800 nm band of the infrared spectrum of the mud; and a particulate component fingerprint analysis component, configured to invert the particulate component parameters in the mud by analyzing the spectral scattering characteristics and baseline drift in the infrared spectrum of the mud, wherein the particulate component parameters include clay content and / or cation exchange capacity.

[0007] Preferably, the present invention further includes the following technical solution: the formula inversion model of the decision control unit adopts a deep neural network, and establishes a nonlinear mapping between "mud fingerprint - curing agent formula - solidified soil performance" through supervised learning, and outputs the superabsorbent resin content S_sap, the total content of composite cementitious material S_b and the mass ratio of magnesium oxide to granulated blast furnace slag powder R_mg. The curing formula includes the superabsorbent resin content, the ratio of magnesium oxide to granulated blast furnace slag powder and the total content of cementitious material.

[0008] Preferably, the present invention further includes the following technical solutions: The formulation inversion model specifically includes the following configuration: Adaptive calculation of the superabsorbent resin dosage S_sap: Based on the current water content W and clay content of the slurry, calculate the minimum water absorption threshold required for the slurry to change from a fluid state to a plastic state or a granular state; When the water content W is higher than the liquid limit or the target water content W_target, output the superabsorbent resin addition command, wherein the calculation logic of the superabsorbent resin dosage S_sap is expressed as: S_sap = α×(W-W_target)+β, where α and β are coefficients related to soil quality; Dynamic proportioning of the magnesium oxide-granulated blast furnace slag powder cementing system: When the system identifies that the organic matter content is higher than the threshold, increase the proportion of the added magnesium oxide in the magnesium oxide-granulated blast furnace slag powder mass ratio R_mg, and use the magnesium oxide hydration to generate magnesium hydroxide to provide continuous alkalinity and neutralize organic acid interference, ensuring the depolymerization of granulated blast furnace slag powder and the gelation of hydrated calcium silicate gel and hydrated calcium aluminosilicate gel.

[0009] Preferably, the present invention further includes the following technical solution: the metering and batching unit includes a superabsorbent resin silo, a magnesium oxide silo, and a granulated blast furnace slag powder silo, as well as a variable frequency screw conveyor metering pump and injection valve. The metering and batching unit adopts a two-stage sequential batching process, including: a first stage of physical granulation, in which superabsorbent resin is added to the fluid slurry to transform the slurry from a fluid state into a loose granular or plastic mixture; a second stage of chemical curing, in which magnesium oxide and granulated blast furnace slag powder are added to the mixture after the mixture has a stirable skeleton; the granulated blast furnace slag powder is activated in an alkaline environment to generate hydrated calcium silicate gel and hydrated calcium aluminosilicate gel to provide cementation; the magnesium hydroxide generated by magnesium oxide hydration maintains alkalinity and resists organic matter buffering, and on the other hand, generates micro-expansion to compensate for shrinkage and promote pore filling; the residual superabsorbent resin acts as an "internal curing" carrier to slowly release water in the later stage, promote the continued hydration and improve water stability.

[0010] Preferably, the present invention further includes the following technical solution: the feedback stirring unit is configured to: acquire the torque T and torque change rate dT / dt of the stirring shaft in real time through a torque sensor; preset a target torque range [T_min, T_max], an evaluation time window Δt, a torque change rate threshold, a stage stirring setting time, and a torque stability duration threshold, wherein T_min is the minimum torque setting value, and T_max is the maximum torque setting value; when the stirring time reaches the stage stirring setting time, if the torque T is less than the minimum torque setting value T_min, or the torque change rate dT / dt is lower than the torque change rate threshold within the evaluation time window Δt, then it is determined that granulation or curing is insufficient, triggering the addition of superabsorbent resin and / or gelling material, and extending the stirring time; when the torque T is stable within the target torque range [T_min, T_max] and the duration reaches the torque stability duration threshold, it is determined to be qualified and discharged.

[0011] This invention also provides a method for preparing a curing agent based on a multi-source mud fingerprint spectrum, comprising the following steps: S1 Extracting feature vectors of the mud using a near-infrared spectral probe integrated into the mud inlet pipe detection section in conjunction with a pretreatment component to form a unique fingerprint spectrum of the mud; S2 Generating a curing formula for the current mud based on the unique fingerprint spectrum of the mud using a built-in multi-source mud fingerprint database and a formula inversion model, wherein the multi-source mud fingerprint database stores fingerprint characteristics, historical optimal formulas, and target performance indicators of mud from different sources; S3 Adding curing materials according to the curing formula in two stages: physical granulation and chemical curing; S4 Monitoring the torque and torque change rate of the stirring shaft in real time during the stirring process using a torque sensor; S5 Comparing the torque and torque change rate with a preset target torque range and evaluation time window, and controlling the addition of curing materials and adjusting the stirring time based on the comparison results.

[0012] Preferably, the present invention further includes the following technical solution: the near-infrared spectral probe scans the slurry flowing through the detection section in the 900nm to 2500nm band of the infrared spectrum to obtain the raw spectral data of the slurry; the preprocessing component preprocesses the raw spectral data to extract feature vectors, and analyzes multiple property parameters of the slurry based on the feature vectors; the multiple property parameters include absolute water content, organic matter content, and particle composition parameters.

[0013] Preferably, the present invention further includes the following technical solution: the step of comparing the torque and torque change rate with the preset target torque range includes: a preset target torque range [T_min, T_max], an evaluation time window Δt, a torque change rate threshold, a stage stirring setting time, and a torque stability duration threshold, wherein T_min is the minimum torque setting value, and T_max is the maximum torque setting value; when the stirring time reaches the stage stirring setting time, if the torque T is less than the minimum torque setting value T_min, or the torque change rate dT / dt is lower than the torque change rate threshold within the evaluation time window Δt, then it is determined that granulation or curing is insufficient, triggering the addition of superabsorbent resin and / or gelling material, and extending the stirring time; when the torque T is stable within the target torque range [T_min, T_max] and the duration reaches the torque stability duration threshold, it is determined to be qualified and discharged.

[0014] Compared with the prior art, the advantages of this invention are: Compared with the prior art, the present invention has at least the following beneficial effects: (1) Online identification and precise formulation: by constructing a mud fingerprint spectrum through near-infrared spectroscopy, the moisture content, organic matter and clay content can be rapidly detected in non-contact mode, and the formulation can be calculated in milliseconds through database and neural network model, eliminating the need for empirical batching. (2) Two-stage sequential solidification and adaptation to extreme mud: first, SAP is used to quickly absorb free water to achieve "physical granulation", so that the fluid mud is transformed into a formable plastic / granular state, and then MgO-GGBS is used for "chemical solidification", which significantly improves the mixing uniformity and solidification efficiency of ultra-high moisture content mud. (3) Dual closed-loop process control: the material state is characterized by torque feedback, and the closed-loop correction of "model calculation - physical feedback" can be realized, which can cope with the fluctuation of mud properties, improve the stability of finished product performance, and avoid excessive addition of solidifying agent. (4) Low carbon and environmentally friendly: the use of MgO and slag solid waste to replace part of the cement can significantly reduce carbon emissions and alkaline leaching risk; the solidified product has high water stability and is suitable for the resource utilization of roadbed filler and backfill materials. Attached Figure Description

[0015] Figure 1 This is a block diagram of the overall structure logic of the system of the present invention; Figure 2 This is a flowchart of adaptive batching control based on fingerprint mapping. Detailed Implementation

[0016] To make the purpose, principle and structure of the present invention clearer, the following description is provided in conjunction with the accompanying drawings and specific embodiments.

[0017] See Figure 1 , Figure 2This invention provides a solidifier formulation system and method based on multi-source mud fingerprint spectrum, comprising: (1) a mud feature identification unit, which is set at the mud inlet pipe and uses online detection methods such as near-infrared spectroscopy to obtain key features of the mud such as water content W, organic matter content OM, and clay content Clay, and constructs a mud fingerprint spectrum; (2) an intelligent decision control unit, which is connected to the mud feature identification unit and has a built-in multi-source mud fingerprint database and formulation inversion model; the formulation inversion model is obtained based on deep neural network training, and establishes a nonlinear mapping between "mud fingerprint - solidifier formulation - solidified soil performance", and outputs the water-absorbing material dosage S_sap and the total dosage of composite cementitious material S_sap. _b and MgO / GGBS mass ratio R_mg; (3) Precision metering and batching unit, including independent silos for raw materials such as SAP, MgO and GGBS, variable frequency screw conveyor / metering pump and injection valve, and adopts a two-stage sequential feeding structure: the first stage adds SAP in the premixing zone for water absorption and granulation, and the second stage adds MgO-GGBS in the strong stirring zone for solidification; (4) Feedback stirring unit, including stirring host and torque sensor, the torque sensor collects the stirring torque T and its change rate dT / dt with time in real time, as a physical characterization of the consistency of the mixture and the granulation / solidification process, and feeds it back to the intelligent decision control unit to trigger additional feeding and process parameter correction.

[0018] Among them, SAP (Super Absorbent Polymer) is officially known as high water-absorbing resin, MgO (Magnesium Oxide) is officially known as magnesium oxide, GGBS (Ground Granulated Blast-furnace Slag) is officially known as granulated blast furnace slag powder, and DNN (Deep Neural Network) is officially known as "deep neural network". The complete description of the MgO-GGBS cementing system is "magnesium oxide-granulated blast furnace slag powder cementing system". OM refers to organic matter content, where "organic matter" is different from the broad definition of "organic matter", specifically referring to the total amount of organic components in the soil / mud system that have engineering properties such as cementation and buffering. The official Chinese name for “C-(A)-SH gel” is “hydrated calcium silicate gel (CSH) and hydrated calcium aluminosilicate gel (CASH)”. CSH gel (hydrated calcium silicate gel) is the core product of cement-based material hydration. It is an amorphous nanoscale gel that constitutes the main cementing skeleton of the solidified body. CASH gel (hydrated calcium aluminosilicate gel) is an aluminum-containing variant formed when some silicon ions are replaced by chloride ions in a system containing an aluminum source (such as Al2O3 in slag). It has a similar structure to CSH but contains aluminum, which enhances the cementitious stability and durability. Together, they provide strength development and microstructure density, which is the basis for the formation of mechanical properties in the chemical solidification stage of mud.

[0019] The proper Chinese name for Clay is clay content. In the fields of geotechnical engineering, soil science, and mud property analysis, it specifically refers to the mass percentage of fine particles with a particle size of less than 2μm in soil or mud. This parameter directly affects the plasticity and solidification behavior of mud.

[0020] (I) Construction and identification of mud fingerprint patterns based on near-infrared spectroscopy.

[0021] This invention preferably employs online near-infrared spectroscopy (NIR) technology to address the problem of the inability to perceive mud properties in real time. The mud feature recognition unit is integrated into the detection section of the mud inlet pipe. The NIR probe scans the 900nm~2500nm wavelength band, and feature vectors are extracted using preprocessing algorithms (such as smoothing, baseline correction, and scattering correction). Moisture fingerprint analysis: Utilizing the overtone absorption peaks of the OH bonds of water molecules at approximately 1450nm and 1940nm, combined with partial least squares (PLS) or neural network regression models, the absolute moisture content W is analyzed in real time, covering a detection range of 30%~200%, with a response time of less than 2 seconds. Organic matter fingerprint analysis: For organic components such as humic acid / fulvic acid, characteristic absorption information in the 1600nm~1800nm ​​wavelength band is extracted, and the organic matter content OM is estimated in real time. Particulate matter fingerprint analysis: Parameters such as clay content Clay and / or cation exchange capacity CEC in the mud are retrieved through spectral scattering characteristics and baseline drift inversion. These features constitute a unique fingerprint spectrum of the mud, providing input variables for subsequent intelligent batching.

[0022] Here, 1450nm and 1940nm represent two specific wavelength values ​​in the near-infrared spectrum, measured in nanometers (nm). These values ​​indicate the characteristic absorption peak positions of the OH chemical bonds (hydrogen-oxygen bonds) in water molecules (H₂O) in the near-infrared region. Near 1450nm: corresponds to the first overtone absorption peak of the OH bond stretching vibration (2νOH); near 1940nm: corresponds to the combined overtone absorption peak of the OH stretching vibration and the HOH bending vibration (νOH+δHOH). In this invention, these two wavelength bands serve as the "spectral fingerprint" for moisture detection. Because water molecules selectively absorb light of specific wavelengths, the absorption intensity is quantitatively related to the water concentration (conforming to the Lambert-Beer law). By real-time monitoring of the absorption intensity of the mud spectrum at 1450nm and 1940nm, combined with PLS or neural network models, the absolute water content can be accurately inverted (range 30%~200%, response <2s). The word "approximately" is used in this specific embodiment because the actual peak position may be slightly shifted due to temperature, matrix interference, etc. (e.g., 1440-1460nm, 1930-1950nm), but this wavelength range has a clear water identification specificity.

[0023] The preprocessing algorithm can be used to smooth and reduce noise in the raw obtained mud infrared spectral data, perform baseline correction to eliminate drift, and perform scattering correction to suppress physical interference. Specifically, the appropriate methods can be selected based on the obtained infrared spectral data to improve the quality of the spectral data and ensure that the subsequently extracted feature vectors accurately reflect the intrinsic chemical and physical properties of the mud. Partial least squares (PLS) or neural network regression models are used to establish a quantitative mapping relationship between the near-infrared spectral feature vectors of the mud and the absolute water content, converting the spectral data into quantifiable water content values, enabling rapid and accurate real-time analysis.

[0024] The CEC parameter, also known as cation exchange capacity, refers to the ability of a unit mass of mud or soil to adsorb and exchange cations on the surface of a colloidal substance. The unit is cmol(+) / kg, and it is a key parameter characterizing the activity of clay minerals and the chemical behavior of mud.

[0025] (II) Intelligent decision-making and the formula inversion algorithm for "one mud, one policy".

[0026] The intelligent decision control unit incorporates a multi-source mud fingerprint database and a formulation inversion model. The database records the fingerprint characteristics, historical optimal formulations, and target performance indicators of mud from different sources, such as 7-day / 28-day unconfined compressive strength (UCS), California Bearing Ratio (CBR), and water stability coefficient.

[0027] The optimal formula inversion model employs a deep neural network (DNN) to establish a nonlinear mapping between fingerprint features and the optimal formula through supervised learning. Its core control mechanisms include the following:

[0028] (1) Adaptive calculation of the dosage of superabsorbent polymer (SAP) resin S_sap in the curing agent formulation: Based on the detected moisture content W and clay content, calculate the minimum water absorption threshold required to change the mud from "fluid state" to "plastic / granular state"; when W is higher than the liquid limit or the target moisture content W_target, output the SAP addition command. Its approximate logic can be expressed as S_sap=α×(W-W_target)+β, where α and β are coefficients related to soil quality and can be selected according to the specific soil conditions.

[0029] (2) Dynamic proportion of MgO-GGBS gel system: When the system identifies that the organic matter content OM is higher than the threshold (e.g., OM>3%), the proportion of magnesium oxide MgO is increased (e.g., MgO / GGBS is adjusted from 1:9 to 1:3~1:4). The hydration of MgO generates magnesium hydroxide Mg(OH)2 to provide continuous alkalinity and neutralize the interference of organic acids, ensuring the depolymerization of slag glass and the formation of gels such as C-(A)-SH, thereby improving the strength development.

[0030] The principle of calculating the organic matter content (OM) in mud in this invention can be summarized as follows: For typical organic components such as humic acid and fulvic acid in mud, characteristic absorption information (i.e., "organic matter fingerprint") of their near-infrared spectra is extracted in the 1600nm~1800nm ​​band. The spectral response in this band has a quantifiable correlation with the organic matter content (OM). The system workflow is as follows: the near-infrared probe scans to acquire the raw spectrum, and the feature vector is extracted through preprocessing algorithms (smoothing, baseline correction, scattering correction). The characteristic absorption information is analyzed by focusing on the 1600~1800nm ​​band, and real-time estimation is completed through the data processing module.

[0031] The "estimation" function relies on the quantitative analysis logic already defined in the overall system architecture. The moisture analysis section provides a solution combining partial least squares (PLS) or neural network regression models. Furthermore, by "extracting feature vectors," time-quantified data can be processed uniformly, mapping characteristic absorption information to organic matter content values. The selection of this wavelength band is based on the specific spectral response of components such as humic acid / fulvic acid in this region. It requires no additional chemical reagents, responds rapidly, and the results serve as a key component of the mud fingerprint spectrum, directly supporting intelligent batching decisions.

[0032] (III) The two-stage sequential synergistic mechanism of "physical granulation - chemical solidification".

[0033] This invention abandons the "one-time mixing" process and proposes a two-stage sequential batching to optimize the curing path: The first stage (physical granulation): Highly absorbent polymeric acid (SAP) is added preferentially, allowing its polymer network to rapidly capture free water and form hydrogel cores. This promotes the agglomeration of slurry particles, transforming the fluid slurry into a loose granular or plastic mixture, thereby significantly improving the dispersion uniformity and mixing efficiency of subsequent cementitious materials. The second stage (chemical curing): After the mixture has developed a stirable framework, MgO and granulated blast furnace slag powder (GGBS, also known as ground granulated blast furnace slag) are added. GGBS generates C-(A)-SH gel in an alkaline environment, providing binding; the Mg(OH)2 generated from MgO hydration maintains alkalinity and resists organic matter buffering, while also producing micro-expansion to compensate for shrinkage and promote pore filling. Furthermore, residual SAP acts as an "internal curing" carrier, slowly releasing moisture in the later stages, promoting continued hydration and improving water stability.

[0034] The principle of this process lies in achieving synergistic efficiency through physical state regulation and chemical reaction timing optimization: In the first stage, superabsorbent polymer (SAP) is added to preferentially fix free water and induce particle aggregation, converting the fluid mud into a loose granular or plastic mixture with a stirrable skeleton, fundamentally improving the dispersion conditions and mixing efficiency of the cementitious material; In the second stage, magnesium oxide and granulated blast furnace slag powder are added based on the formed physical skeleton, enabling them to efficiently hydrate to form cementitious products in an optimized contact environment. Meanwhile, the residual SAP serves as an internal curing carrier to slowly release water to maintain the continuous progress of the hydration reaction. The two work in sequence and协同解决了高含水率泥浆中胶凝材料分散不均、有机质干扰及水化水分不足等关键问题,最终提升固化体的结构均匀性、强度发展与水稳定性。

[0035] (4) Closed-loop control strategy for torque feedback.

[0036] The feedback stirring unit obtains the torque T of the stirring shaft and its change rate dT / dt in real time through a torque sensor. In the water absorption granulation stage, the rapid increase of the torque from a low value indicates the fixation of free water and the formation of the particle skeleton; in the cementitious curing stage, the further increase and stabilization of the torque indicate the thickening of the slurry and the establishment of the structure. The system sets the target torque range [T_min, T_max] and the evaluation time window Δt. When T < T_min after stirring reaches the set time or dT / dt is lower than the threshold within Δt, it is determined that the granulation / curing is insufficient, triggering the additional addition of SAP and / or cementitious material and extending the stirring; when T stabilizes within the target range and the duration reaches the preset value, it is determined to be qualified and the material is discharged. This strategy can cope with on-site disturbances such as sudden increase in the water content of the incoming mud and local uneven mixing, ensuring the consistency of the discharged material state and performance.

[0037] The principle of the closed-loop control strategy of this invention is based on the dynamic mapping relationship between the stirring torque and the physical state of the mud. The torque value and its change rate reflect the progress of the granulation stage (fixation of free water, formation of the particle skeleton) and the curing stage (thickening of the slurry, establishment of the structure) in real time; the system compares the measured torque with the preset target range and the change rate threshold, automatically determining the process state - triggering supplementary feeding and extended stirring when insufficient, and allowing discharging when reaching the standard, forming a closed-loop regulation mechanism with torque as the feedback signal, thereby dynamically adapting to on-site disturbances such as sudden change in the water content of the incoming mud and uneven mixing, ensuring the consistency of the physical state and engineering performance of each batch of discharged material.

[0038] The following is a comparison through experimental examples.

[0039] It should be noted that there seems to be an incomplete sentence in the translation of . Please check and correct it if necessary.Examples and Comparative Examples: To verify the effectiveness of this system, three typical extreme engineering muds were selected for field pilot testing: Mud A (ultra-high water content dredged mud, W=125%), Mud B (high organic matter lake mud, W=85%, OM=5.8%), and Mud C (high-viscosity shield tunnel mud, W=65%, Clay=65%). All examples employed the fully automated processing of the present invention's system: online fingerprint feature detection → intelligent formula generation → SAP priority granulation → MgO and GGBS solidification → torque-compliant discharge. For comparison, a fixed cement formula was used, with manual adjustment versus a fixed MgO / GGBS ratio.

[0040] Table 1. Summary of intelligent batching parameters and solidified soil performance indicators under different mud sources.

[0041]

[0042] Example 1: For mud with ultra-high water content.

[0043] The system detected that the water content of mud A was as high as 125%. The intelligent decision control unit matched the SAP dosage to 0.55% and added it for granulation. After addition, the stirring torque rapidly increased to 1650 N·m, indicating that the mud changed from a fluid state to a loose granular state and could be compacted. The final solidified soil CBR reached 12.5%, meeting the requirements of first-class highway subgrade fill. In contrast, Comparative Example 1 used a 20% cement fixed formula. Although the amount of cementitious material was higher, it could not effectively treat free water, and the output was in a fluid plastic state. The 7-day unconfined compressive strength UCS was only 120 kPa, the CBR was only 1.2%, and the water stability was significantly insufficient.

[0044] Example 2: For high organic matter mud.

[0045] Slurry B, with an organic matter content of 5.8%, has a buffering and inhibitory effect on alkaline activation and hydration reactions. After recognizing this fingerprint, the system automatically adjusted the MgO / GGBS mass ratio to 1:3 and increased the cementitious material content to 18%. Utilizing the high reactivity and alkalinity maintenance ability of MgO to resist organic matter interference, the unconfined compressive strength (UCS) reached 1950 kPa after 28 days. Comparative Example 2 (cement curing) and Comparative Example 5 (fixed MgO / GGBS = 1:9) both showed limited strength development, indicating the crucial role of dynamic mix proportioning in the chemical curing stage.

[0046] Example 3 Analysis: For high-viscosity mud.

[0047] For mud with high C-clay content (65%) but moderate moisture content (65%), the system automatically reduced SAP to 0.15% and the total cementitious material content to 12% to control costs, while maintaining the MgO / GGBS ratio at 1:6. Results showed that the finished solidified soil achieved a CBR of 22.4% and a 28-day unconfined compressive strength (UCS) of 2400 kPa, demonstrating economic efficiency while meeting performance requirements.

[0048] Environmental and durability analysis: The leachate pH values ​​of Examples 1-3 were 9.5-10.2, significantly lower than the pH value of over 12 in the cement-cured control example, reducing the potential impact of highly alkaline leaching on the surrounding soil and water environment. The water stability coefficients of all examples exceeded 94%, indicating that the internal curing and porosity regulation effects of SAP help maintain the integrity of the hydration product skeleton and improve long-term durability.

[0049] The above description is merely a specific embodiment of the invention, but the scope of protection of the invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the invention, based on the technical solution and concept of the invention, should be covered within the scope of protection of the invention.

Claims

1. A curing agent batching system based on multi-source mud fingerprinting, characterized in that, include: The mud feature recognition unit is set in the detection section of the mud inlet pipe. It integrates a near-infrared spectroscopy probe and pretreatment components to extract the feature vector of the mud, which constitutes the unique fingerprint spectrum of the mud and forms the input variable in the solidifier batching system. The decision control unit is signal-connected to the mud feature recognition unit and has a built-in multi-source mud fingerprint database and a formula inversion model. The multi-source mud fingerprint database stores the fingerprint features, historical best formulas and target performance indicators of mud from different sources. The formula inversion model is used to generate a solidification formula for the current mud by matching the unique fingerprint spectrum of the mud with the multi-source mud fingerprint database. A metering and dispensing unit, connected to the decision control unit, is used to add curing materials in two stages, physical granulation and chemical curing, according to the curing formula. A feedback stirring unit includes a stirring device and a torque sensor, wherein the torque sensor is used to monitor the torque and torque change rate of the stirring shaft in the stirring device in real time; The decision control unit is further configured to compare the torque and torque change rate with a preset target torque range and evaluation time window, and control the addition of curing material and adjust the stirring time based on the comparison results.

2. The curing agent batching system based on multi-source mud fingerprinting as described in claim 1, characterized in that, The near-infrared spectral probe is used to scan the slurry flowing through the detection section in the 900nm to 2500nm band of the infrared spectrum to obtain the raw spectral data of the slurry. The preprocessing component is connected to the near-infrared spectral probe signal and is used to preprocess the raw spectral data to extract feature vectors and analyze multiple property parameters of the mud based on the feature vectors. The multiple property parameters include absolute moisture content, organic matter content, and particle composition parameters.

3. The curing agent batching system based on multi-source mud fingerprinting as described in claim 2, characterized in that, The mud feature recognition unit further includes: The moisture fingerprint analysis component is configured to identify the hydroxyl bond overtone absorption peaks of water molecules at wavelengths of 1450nm and 1940nm in the infrared spectrum of mud, and calculate the absolute moisture content of mud by combining a partial least squares regression model or a neural network regression model. The organic matter fingerprint analysis component is configured to calculate the organic matter content in the mud by extracting characteristic absorption information in the 1600nm to 1800nm ​​band of the infrared spectrum of the mud. The particle component fingerprinting component is configured to retrieve particle component parameters in the mud by analyzing the spectral scattering characteristics and baseline drift in the infrared spectrum of the mud. The particle component parameters include clay content and / or cation exchange capacity.

4. The curing agent batching system based on multi-source mud fingerprinting as described in claim 1, characterized in that, The formula inversion model of the decision control unit adopts a deep neural network. Through supervised learning, a nonlinear mapping between "mud fingerprint - solidifier formula - solidified soil performance" is established, and the outputs are the superabsorbent resin content S_sap, the total content of composite cementitious material S_b, and the mass ratio of magnesium oxide to granulated blast furnace slag powder R_mg. The solidification formula includes the superabsorbent resin content, the ratio of magnesium oxide to granulated blast furnace slag powder, and the total content of cementitious material.

5. The curing agent batching system based on multi-source mud fingerprinting as described in claim 4, characterized in that, The formula inversion model specifically includes the following configuration: Adaptive calculation of superabsorbent polymer (S_sap) dosage: Based on the current water content W and clay content of the mud, calculate the minimum water absorption threshold required for the mud to change from a fluid state to a plastic or granular state; when the water content W is higher than the liquid limit or the target water content W_target, output the superabsorbent polymer addition command. The calculation logic of the superabsorbent polymer dosage S_sap is expressed as: S_sap=α×(W-W_target)+β, where α and β are coefficients related to soil properties. Dynamic proportioning of the magnesium oxide-granulated blast furnace slag powder cementitious system: When the system identifies that the organic matter content is higher than the threshold, the proportion of magnesium oxide in the magnesium oxide-granulated blast furnace slag powder mass ratio R_mg is increased. The magnesium hydroxide generated by magnesium oxide hydration provides continuous alkalinity and neutralizes the interference of organic acids, ensuring the deagglomeration of granulated blast furnace slag powder and the gel formation of hydrated calcium silicate gel and hydrated calcium aluminosilicate gel.

6. The curing agent batching system based on multi-source mud fingerprinting as described in claim 1, characterized in that, The metering and batching unit includes a highly absorbent resin silo, a magnesium oxide silo, and a granulated blast furnace slag powder silo, as well as a variable frequency screw conveyor metering pump and injection valves. The metering and batching unit employs a two-stage sequential batching process, including: The first stage is physical granulation, in which highly absorbent resin is added to the fluid slurry to transform the slurry from a fluid state into a loose granular or plastic mixture. The second stage of chemical solidification involves adding magnesium oxide and granulated blast furnace slag powder to the mixture after the mixture has a stirable framework. Granulated blast furnace slag powder is activated in an alkaline environment to generate hydrated calcium silicate gel and hydrated calcium aluminosilicate gel, which provide cementation; magnesium hydroxide generated by magnesium oxide hydration maintains alkalinity and resists organic matter buffering, and on the other hand, it generates micro-expansion to compensate for shrinkage and promotes pore filling. The residual superabsorbent resin acts as an "internal maintenance" carrier, slowly releasing moisture in the later stages, promoting continued hydration and improving water stability.

7. The curing agent batching system based on multi-source mud fingerprinting as described in claim 1, characterized in that, The feedback stirring unit is configured as follows: The torque T and torque change rate dT / dt of the stirring shaft are obtained in real time by a torque sensor. The preset target torque range [T_min, T_max], evaluation time window Δt, torque change rate threshold, stage stirring setting time, and torque stability duration threshold are set, where T_min is the minimum torque setting value and T_max is the maximum torque setting value. When the stirring time reaches the set stirring time of the stage, if the torque T is less than the minimum torque set value T_min, or the torque change rate dT / dt is lower than the torque change rate threshold within the evaluation time window Δt, it is determined that granulation or curing is insufficient, triggering the addition of superabsorbent resin and / or gelling material, and extending the stirring time. When the torque T stabilizes within the target torque range [T_min, T_max] and the duration reaches the torque stabilization duration threshold, the material is deemed qualified and discharged.

8. A method for preparing a curing agent based on multi-source mud fingerprinting, characterized in that, Includes the following steps: S1 extracts the feature vector of the mud by combining a near-infrared spectral probe integrated into the mud inlet pipe detection section with a pretreatment component, thus forming a unique fingerprint spectrum of the mud. S2 generates a solidified formula for the current mud based on the unique fingerprint spectrum of the mud and using the built-in multi-source mud fingerprint database and formula inversion model. The multi-source mud fingerprint database stores the fingerprint characteristics, historical best formula and target performance indicators of mud from different sources. S3 adds curing material in two stages, physical granulation and chemical curing, according to the curing formula; S4 uses a torque sensor to monitor the torque and torque change rate of the stirring shaft in real time during the stirring process; S5 compares the torque and torque change rate with the preset target torque range and evaluation time window, and controls the addition of curing material and adjusts the stirring time based on the comparison results.

9. The curing agent formulation method based on multi-source mud fingerprinting as described in claim 8, characterized in that, The near-infrared spectral probe scans the slurry flowing through the detection section in the 900nm to 2500nm band of the infrared spectrum to obtain the raw spectral data of the slurry. The preprocessing component preprocesses the raw spectral data to extract feature vectors, and analyzes multiple property parameters of the mud based on the feature vectors; The multiple property parameters include absolute moisture content, organic matter content, and particle composition parameters.

10. The method for preparing a curing agent based on multi-source mud fingerprinting according to claim 8, characterized in that, The step of comparing the torque and the rate of change of torque with the preset target torque range includes: The preset target torque range [T_min, T_max], evaluation time window Δt, torque change rate threshold, stage stirring setting time, and torque stability duration threshold are set, where T_min is the minimum torque setting value and T_max is the maximum torque setting value. When the stirring time reaches the set stirring time of the stage, if the torque T is less than the minimum torque set value T_min, or the torque change rate dT / dt is lower than the torque change rate threshold within the evaluation time window Δt, it is determined that granulation or curing is insufficient, triggering the addition of superabsorbent resin and / or gelling material, and extending the stirring time. When the torque T stabilizes within the target torque range [T_min, T_max] and the duration reaches the torque stabilization duration threshold, the material is deemed qualified and discharged.

Images