A method for improving the homogeneity of a soil stabilization process by segmented mixing

Abstract

The application relates to the field of automatic control technology and discloses a method for improving the segment stirring and uniformity of a solidified soil mixing process, which comprises the following steps: step 1, detecting the water content of raw soil and the proportion of over-limit particle groups, determining and executing a pretreatment scheme according to the detection results; step 2, after material feeding to form mixed materials, sequentially executing a dry stirring section, a semi-wet stirring section and a wet stirring section, extracting trend level, fluctuation amplitude and order degree index as three features from a stirring shaft torque signal in each judgment period, judging the mixed materials as a granular state, a liquid bridge transition state or a plastic flow state, and driving the segment switching according to the confirmed material state; and step 3, collecting a mixing surface image and dividing the image into sub-regions according to spraying partition, judging a dry region according to a gray scale deviation, only opening the atomizing nozzle group covering the dry region to spray supplementary water for mixing and supplementing stirring until the mixing uniformity reaches the standard and then discharging. The application aligns the segment switching with the actual material state, concentrates the water supply in the dry region, and improves the mixing uniformity and batch consistency.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology, and specifically relates to a method for segmented mixing and uniformity improvement in the solidified soil mixing process. Background Technology

[0002] Solidified soil is an engineering material made primarily from local or transported soil, mixed with cement or lime and water to form a binder. It is widely used in foundation pit backfilling, trench backfilling, road base courses, and site leveling. The consistency of the binder and water dispersion in the soil directly determines the strength and durability of the finished product, making the mixing process crucial for quality control. Currently, fixed-duration, segmented mixing processes are commonly used in engineering projects. The durations of dry mixing, water addition, and wet mixing are set based on experience, and water is added at a single point. Existing practices also include monitoring the mixing power or torque curve on the mixing equipment to help determine the degree of mixing, and automatically calculating the amount of water added based on online moisture content monitoring results. These methods are generally usable when the raw materials are stable, but they still have the following problems: First, the particle size distribution and moisture content of the original soil fluctuate significantly depending on the location of the soil extraction and the weather. Fixed-duration segmented mixing is disconnected from the actual state of the material, resulting in both under-mixing and over-mixing. Torque or power curves can only reflect the overall resistance trend, and both dry and wet mixing are represented by a curve plateau, making it difficult to distinguish between the two based solely on the curve. Second, concentrated water addition at a single point forms overly wet mud clumps near the water droplet, exacerbating agglomeration. Third, if there are areas of localized dryness or concentrated hardener after mixing, conventional methods only allow for extended mixing of the entire batch or overall water replenishment, resulting in low replenishment efficiency, passively increasing the overall moisture content, and disrupting the predetermined mix proportions. Therefore, a mixing method is needed that can switch mixing segments according to the actual state of the material and implement targeted treatment for locally imbalanced areas. Summary of the Invention

[0003] Therefore, the main objective of this invention is to provide a segmented mixing and uniformity improvement method for the solidified soil mixing process. The method is driven by the physical state recognition results of the mixed materials to switch between dry mixing, semi-wet mixing and wet mixing sections, and to perform zoned and directional water replenishment on the dry areas located on the mixing surface. This solves the problems of fixed-duration segmented switching being disconnected from the actual state of the materials and low overall mixing efficiency in local imbalance areas.

[0004] The technical solution adopted in this invention is as follows: A method for staged mixing and homogeneity improvement in the solidification soil mixing process includes the following steps: Step 1: Detect the moisture content and percentage of excess particles in the original soil, and determine and implement the pretreatment plan based on the test results; Step 2: The raw soil and solidifying agent treated in Step 1 are added to the mixing chamber of the mixer to form a mixed material. The dry mixing section, semi-wet mixing section, and wet mixing section are executed sequentially. The torque signal of the mixing shaft is collected, and three features—trend level, fluctuation amplitude, and orderliness index—are extracted in each consecutive discrimination cycle. Based on the three features, the mixed material is classified into a granular state, a liquid bridge transition state, or a plastic flow state. When the discrimination result of the material state is continuously valid within the preset confirmation time, the mixed material is confirmed to be in the corresponding material state. When the mixed material is confirmed to be in a granular state, the dry mixing section is switched to the semi-wet mixing section. In the semi-wet mixing section, mixing water is sprayed onto the mixed material through multiple atomizing nozzle groups covering different spray zones. When the mixed material is successively confirmed to be in a liquid bridge transition state and a plastic flow state, the wet mixing section is switched to. In the wet mixing section, homogenization and mixing are carried out. When the discrimination result of the plastic flow state is continuously valid for a preset homogenization time, the process proceeds to Step 3. Step 3: Collect images of the mixing surface and divide them into corresponding image sub-regions according to the coverage position of each spray zone. Calculate the grayscale mean of each image sub-region and the overall average of all grayscale mean values. Image sub-regions whose grayscale mean exceeds the overall average by a amount greater than the preset grayscale deviation threshold are identified as dry areas. Based on the determination of dry areas, perform uniformity determination. If the mixing uniformity is determined to be in need of improvement, trigger the supplementary mixing strategy: only turn on the atomizing nozzle group covering the dry area to spray supplementary mixing water to the dry area, then perform supplementary mixing and re-execute uniformity determination until the mixing uniformity is determined to meet the standard before discharging.

[0005] Further: The pretreatment scheme includes pre-crushing treatment, pre-wetting treatment, and pre-dry mixing treatment; in step 1, the raw soil is conveyed to the mixer via a feeding belt, and an image of the raw soil is acquired by a detection camera set above the feeding belt. The raw soil image is then processed by grayscale and threshold segmentation to extract the connected regions corresponding to each particle cluster. The equivalent circle diameter of the connected region is taken as the equivalent particle size of the particle cluster. The proportion of the area of ​​particle clusters with an equivalent particle size greater than the preset upper limit of particle size to the total area of ​​all particle clusters in the raw soil image is counted as the proportion of over-limit particle clusters. The moisture content of the raw soil is detected by a moisture content detection device set at the discharge end of the feeding belt. When the proportion of over-limit particle clusters is greater than the preset proportion threshold, pre-crushing treatment is performed. The raw soil is fed into a toothed roller pre-crushing device for crushing. The crushed raw soil is then conveyed to the mixer via the feeding belt. The soil is re-contained and re-tested until the proportion of excess particles drops below the preset threshold. When the moisture content is below the preset lower limit, pre-wetting treatment is performed by spraying atomized water onto the surface of the original soil and turning it over. The amount of atomized water sprayed is included in the total amount of mixing water until the moisture content reaches the preset lower limit. When the moisture content is above the preset upper limit, pre-dry mixing treatment is performed by turning and dissipating moisture from the original soil until the moisture content drops below the preset upper limit. When the detection results of excess particle proportion and moisture content trigger multiple pretreatment schemes simultaneously, pre-crushing treatment is performed before pre-wetting treatment and pre-dry mixing treatment. When the excess particle proportion is within the preset threshold, the moisture content reaches the preset lower limit, and the moisture content is within the preset upper limit, the original soil proceeds to step 2.

[0006] Further: In step 2, the acquisition of torque signals, extraction of three features, and discrimination and confirmation of the state of matter are continuously performed from the start of the dry mixing section; the torque signal is continuously acquired by a torque sensor installed at the drive end of the stirring shaft at a fixed sampling period; the slowly changing part of the torque signal retained after low-pass filtering is taken as the trend component, and the rapidly changing part of the torque signal after filtering out the trend component is taken as the fluctuation component; the cutoff frequency of the low-pass filter is set to be lower than the rotation frequency of the stirring shaft; a preset discrimination time is taken as one discrimination period, and the preset discrimination time covers multiple sampling periods; the trend level is the average value of the trend component within the discrimination period, and the fluctuation amplitude is the difference between the maximum and minimum values ​​of the fluctuation component within the discrimination period.

[0007] Furthermore, the extraction process of the orderliness index is as follows: within the discrimination period, along the time sequence, a permutation group consisting of a preset number of consecutive sampling points from the torque signal is extracted point by point with a step of 1 sampling point. The sampling points in each permutation group are sorted in ascending order of value, and the sampling points with equal values ​​are sorted in chronological order. The sequential positional relationship of each sampling point in the original permutation group after sorting is recorded as a permutation pattern. The occurrence frequency of each permutation pattern in all permutation groups is counted, and the proportion of the occurrence frequency of the permutation pattern with the most occurrences to the total number of permutation groups is used as the orderliness index.

[0008] Furthermore, the rules for determining the physical state of the mixed material based on three characteristics are as follows: when the change in trend level between adjacent determination periods is less than a preset stability threshold, the fluctuation amplitude is greater than a preset fluctuation threshold, and the orderliness index is less than a preset orderliness threshold, the mixed material is determined to be in a granular state; when the trend level increases with each determination period, the increase between adjacent determination periods is greater than a preset stability threshold, and the fluctuation amplitude is greater than a preset fluctuation threshold, the mixed material is determined to be in a liquid bridge transition state; when the change in trend level between adjacent determination periods is less than a preset stability threshold, the fluctuation amplitude is less than a preset fluctuation threshold, and the orderliness index is greater than a preset orderliness threshold, the mixed material is determined to be in a plastic flow dynamic state.

[0009] Further: In step 2, the curing agent is added to the mixing chamber according to a preset dosage ratio via a spiral metering conveyor; multiple atomizing nozzle groups are arranged along the axial direction of the mixing shaft at the top of the mixing chamber, each atomizing nozzle group corresponding to one spraying zone on the top surface of the mixing chamber, and all spraying zones are combined to cover the entire mixing surface; in the dry mixing section, the mixing shaft operates at a dry mixing speed, and all atomizing nozzle groups remain closed to dry mix the original soil and curing agent; in the semi-wet mixing section, the mixing shaft operates at a semi-wet mixing speed, and all atomizing nozzle groups spray synchronously in the form of spray pulses, each spray pulse is the opening of the atomizing nozzle group and the spraying time is preset before it closes, and the interval between adjacent spray pulses is preset interval time, the mixing water is evenly distributed according to all spraying zones and added to the mixing material in the form of droplets, the target amount of mixing water is determined according to the target moisture content of the mixing material and the moisture content detected in step 1, and the spraying continues until the target amount is added; in the wet mixing section, the mixing shaft operates at a wet mixing speed, which is higher than the semi-wet mixing speed.

[0010] Furthermore: In the semi-wet mixing section, when the target amount of water is added and the state of the mixed material is most recently confirmed as granular or liquid bridge transition after a preset waiting time since the completion of the addition, all atomizing nozzle groups are controlled to add additional spray in the form of spray pulses. The amount of water added by the additional spray is included in the total amount of water used for mixing, and the total amount of water used for mixing is constrained by the preset upper limit of the total amount of water added. After each additional spray, the preset waiting time is recalculated, and the above judgment and additional spray are repeated. When the total amount of water used for mixing reaches the preset upper limit of the total amount of water added, and the state of the mixed material is most recently confirmed as granular or liquid bridge transition, a manual intervention prompt is output and the machine is stopped for inspection.

[0011] Further: In step 3, before acquiring the image of the mixing surface, the stirring shaft speed is reduced to the circulating mixing speed, which is lower than the dry mixing speed, semi-wet mixing speed, and wet mixing speed, so that the mixing surface slowly turns over and remains within the field of view of the industrial camera set at the top of the mixing chamber; the mixing surface image is acquired by the industrial camera under the illumination of the supplementary light source, and after grayscale processing, it is used to divide the image sub-regions; the fluctuation amplitude of the last discrimination cycle before the stirring shaft speed is reduced to the circulating mixing speed is used as the evaluation fluctuation amplitude; the uniformity judgment rule is: when the number of dry areas is 0, the difference between the maximum and minimum values ​​of the grayscale average of all image sub-regions is less than the preset uniformity threshold, and the evaluation fluctuation amplitude is less than the preset fluctuation threshold, the mixing uniformity is judged to meet the standard, the stirring shaft speed is restored to the wet mixing speed, and the material is discharged; when there are dry areas, or the difference between the maximum and minimum values ​​reaches or exceeds the preset uniformity threshold, or the evaluation fluctuation amplitude reaches or exceeds the preset fluctuation threshold, the mixing uniformity is judged to need to be improved, and the supplementary mixing strategy is triggered.

[0012] Furthermore: In the supplementary mixing strategy, the stirring shaft is kept running at a constant stirring speed, and the mixing surface image is re-acquired according to a preset refresh cycle. In each refresh cycle, the image sub-regions are re-divided based on the latest acquired mixing surface image, the grayscale mean and the overall average value are recalculated, and the dry area is re-determined. The determination of the dry area is performed independently for each image sub-region. The spray zone corresponding to the image sub-region where the dry area is located in this determination is taken as the spray zone where the dry area is currently located. When spraying supplementary mixing water to the dry area, the spraying is performed in the form of spray pulses. The number of spray pulses corresponding to a single dry area is set according to the preset grading interval into which the grayscale mean of the dry area falls above the overall average value. The preset grading intervals are divided in ascending order of the amount above the overall average value. The higher the preset grading interval, the more spray pulses are set.

[0013] Furthermore: In the supplementary mixing strategy, the direction of tumbling of the mixed materials on the mixing surface is determined by the rotation of the stirring shaft, and the dry zone moves along the tumbling direction as the mixed materials tumble; the spray zone where the dry zone is currently located is recorded as the original spray zone, and the spray zone that the dry zone is about to enter along the tumbling direction is recorded as the subsequent spray zone; when the dry zone moves to the boundary area between the original spray zone and the subsequent spray zone, the atomizing nozzle group corresponding to the subsequent spray zone is first turned on, and after the dry zone as a whole enters the image sub-area corresponding to the subsequent spray zone, the nozzle group corresponding to the original spray zone is then turned off. The corresponding atomizing nozzle group ensures that the spray point continuously follows and covers the moving dry area; the cumulative spray volume of the supplementary mixing water is included in the total mixing water volume and is subject to the preset upper limit of the total water volume; after the spray pulses corresponding to all dry areas have been executed, the stirring shaft speed is restored to the wet mixing speed for supplementary mixing. After the supplementary mixing time reaches the preset supplementary mixing time, the stirring shaft speed is reduced to the patrol mixing speed again, and the uniformity judgment is re-executed; when the cumulative number of times the supplementary mixing strategy is executed reaches the preset upper limit, a manual intervention prompt is output and the machine is stopped for inspection.

[0014] By adopting the above technical solution, the present invention has produced the following beneficial effects: The present invention uses the stirring shaft torque signal as the information source, combining three characteristics—trend level, fluctuation amplitude, and orderliness index—to determine the physical state of the mixed material. Particle state and plastic flow dynamics both exhibit a stable platform in terms of trend level, but can be clearly distinguished by the differences in fluctuation amplitude and orderliness index. Segmented switching thus aligns with the actual physical state of the material, and the switching timing automatically adapts to the original soil properties, simultaneously eliminating under-mixing and over-mixing. The confirmation mechanism requires the judgment result to be continuously valid for it to take effect, filtering out occasional disturbances and serving as the endpoint for the solidifier dispersion in the dry mixing section. In the semi-wet mixing section, spraying… The spray pulse adds mixing water evenly according to the spray zones, suppressing the tendency of concentrated water droplets to form overly wet mud clumps. Additional sprays replenish water in a closed loop based on the physical state confirmation results, and the moisture content detection deviation is absorbed by the mixing process itself. In the uniformity evaluation, the mixing surface image reflects the surface moisture distribution, and the evaluation fluctuation amplitude reflects the internal rheological state. The two channels complement each other. The replenishing water is only applied to the positioned dry area and the amount is graded according to the degree of gray scale deviation. The spray landing point switches in relay as the dry area moves, and the water replenishment is completed in one go, so that the overall moisture content of the mixed material is maintained with the predetermined mix ratio. The total amount of water added and the number of times of replenishment are both set with upper limits, and the process will inevitably terminate, which significantly improves the consistency of mixing quality between batches. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the layout of the solidified soil mixing system provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of torque signal filtering and decomposition provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the evolution curves of three characteristics throughout the mixing process provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the principle of orderliness index extraction provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of image sub-region division and dry region determination provided in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the principle of relay spraying in dry areas, as provided in an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. The following content is used to illustrate the present invention and provides one feasible implementation and several optional embodiments; the implementation of the present invention is not limited thereto.

[0017] Figure 1The arrangement of the solidified soil mixing system used in this embodiment is shown. The original soil is fed into the receiving hopper of the toothed roller pre-crushing device by a loader. The toothed roller pre-crushing device adopts a double-roller structure with a roller diameter of 400 mm. The two toothed rollers rotate in opposite directions, and the roller gap is adjusted to 40 mm. The raw soil, broken up by shearing and compression, falls to the starting end of the feeding conveyor belt. The feeding conveyor belt has a width of 800 mm and a belt speed of 0.8 m / s. A detection camera is installed 1.2 meters above the conveyor belt. The detection camera is a color area array camera with a resolution of 2448 pixels × 2048 pixels, a lens focal length of 8 mm, and a field of view covering the entire width of the conveyor belt. A moisture content detection device is installed at the discharge end of the feeding conveyor belt. In this embodiment, a microwave transmission type is selected, with a range of 4% to 40% and a detection accuracy of ±0.5 percentage points. The dielectric loss of microwave energy by water molecules is much higher than that of soil particles. The attenuation of microwaves after passing through the soil layer on the conveyor belt increases monotonically with the moisture content. The moisture content of the original soil can be continuously output after calibration curve conversion. The raw soil enters the mixing chamber of the mixer via a discharge chute. The mixer is a twin-shaft forced mixer with an effective volume of 2 cubic meters and an axial length of 2.4 meters. Two mixing shafts are arranged in parallel and driven by a drive unit via a reduction gear. The impellers are arranged spirally along the mixing shafts with a 15-degree installation angle. A flange-type torque sensor is connected in series between the drive end of the mixing shaft and the output flange of the reduction gear, with a range of 0 to 5000 N·m and an accuracy of 0.2% of full scale. The curing agent is stored in a curing agent silo and is added to the mixing chamber according to a preset dosage ratio via a screw metering conveyor. The screw metering conveyor is driven by a variable frequency motor with a metering accuracy of ±1%.

[0018] Six atomizing nozzle groups are evenly spaced along the axial direction of the mixing shaft at the top of the mixing chamber. Each atomizing nozzle group includes three solid conical atomizing nozzles arranged along the width of the mixing chamber. The working pressure is 0.4 MPa, and the rated flow rate of a single nozzle is 0.9 cubic meters per hour. Each atomizing nozzle group is independently controlled by a high-speed solenoid valve, with a response time of less than 30 milliseconds. The mist field of each atomizing nozzle group corresponds to a spray zone covering the top surface of the mixing chamber with an axial length of 0.4 meters. The six spray zones are combined to cover the entire mixing surface. Figure 1 The locations of each spray zone are marked with numbers 1 to 6. An observation window is located in the center of the top plate of the mixing chamber. An industrial camera and supplementary lighting are installed above the observation window: the industrial camera is a color area array camera with a resolution of 5 megapixels and a frame rate of 10 frames per second, its field of view covering the entire mixing surface through the observation window; the supplementary lighting consists of two strip-shaped LED light sources with a color temperature of 5600 Kelvin, creating an illuminance of approximately 2000 lux on the mixing surface, thus eliminating the influence of ambient light fluctuations in the workshop on the brightness of the mixed surface image. All the above-mentioned detection and execution components are uniformly connected to a programmable logic controller (PLC) in the field. Signal processing, material state determination, and the opening and closing of the solenoid valves are all completed by the PLC.

[0019] Step 1 involves detecting the state of the original soil on the conveyor belt. The detection camera acquires images of the original soil at a frequency of 2 frames per second. The images are first converted to grayscale, and the grayscale value of each pixel is obtained by weighted summation according to the weights of the red component (0.299), green component (0.587), and blue component (0.114). Then, the segmentation threshold is determined using the maximum inter-class variance method, and threshold segmentation is performed. The conveyor belt surface is dark and uniformly reflective, and the grayscale distribution of the particle cluster area and the belt background exhibits a clear bimodal structure. The grayscale value that maximizes the variance between the foreground and background classes is taken as the segmentation threshold. The threshold automatically follows the slow change in illumination. The segmented foreground is labeled with 8-neighborhood connectivity to form several connected regions, each corresponding to one particle cluster. Let S be the number of pixels contained in a connected region, and u be the actual side length of each pixel after calibration by the detection camera. In this embodiment, u is 0.55 mm per pixel. The equivalent particle size d of the particle cluster is calculated using the following formula:

[0020] In the formula, d is the equivalent particle size of the aggregate, in millimeters; u is the actual side length corresponding to each pixel, in millimeters per pixel; S is the number of pixels contained in the connected region; and π is pi. The equivalent particle size is the diameter of a circle with the same area as the projected area of ​​the aggregate. The projected shape of the aggregate on the belt is irregular; the width reading of elongated aggregates varies depending on their orientation. Using the equivalent circle diameter as a size metric avoids interference from orientation. A preset upper limit for the particle size is 60 millimeters. The proportion of aggregates with an equivalent particle size greater than 60 millimeters is calculated as the percentage of aggregates exceeding the limit in the original soil image. A preset threshold for this proportion is 8%. Regarding moisture content, a preset lower limit of 12% and a preset upper limit of 22% are used. The target moisture content of the solidified soil in this embodiment is 18%.

[0021] The detection cycle is coordinated with the belt movement: belt speed is 0.8 meters per second, image acquisition frequency is 2 frames per second, the belt advances 0.4 meters between two adjacent frames, the field of view of the detection camera along the belt direction is about 0.9 meters, the coverage areas of adjacent frames overlap, and the particles on the belt are completely captured one by one, thus avoiding missed detections; the proportion of particles exceeding the limit and the moisture content are both used for discrimination after the sliding average of the most recent 10 readings, and accidental occlusion or reflection in a single frame is smoothed out. The re-inspection after pre-processing follows the same path: the pre-crushed raw soil passes through the detection camera and moisture content detection device again, and the moisture content is re-measured after each round of mixing after pre-wetting and pre-drying. After all three threshold conditions are met, the soil is released, thus keeping the state of the raw soil entering the mixing chamber under control, and the various thresholds for subsequent physical state recognition can be kept stable near a set of calibration values ​​for a long time. When soil properties change systematically due to changes in soil source or seasonal alternation, repeating a trial mix of 5 batches once can update the various threshold values ​​and the relationship between ash content and moisture content calibration. The calibration process can be completed within half a working day.

[0022] The pretreatment plan was determined and implemented based on two test results. When the proportion of excessive particle size exceeded 8%, pre-crushing was performed. The original soil was fed into a toothed roller pre-crushing device for crushing. The crushed soil was then re-conveyed via a conveyor belt and re-tested until the proportion of excessive particle size decreased to below 8%. Large particles of high-plasticity clay are difficult to crush by impact; the shearing and squeezing action of the toothed rollers is more effective for cohesive soils. The 40mm roller gap ensures that the particle size of the crushed particles is generally within the preset upper limit. When the moisture content was below 12%, pre-wetting was performed. Atomized water was sprayed onto the surface of the original soil and it was agitated. The amount of water sprayed was included in the total mixing water until the moisture content reached 12%. Excessively dry soil produces a lot of dust and absorbs water slowly upon entering the mixing chamber, delaying the formation of liquid bridges. Adding bottom water around the conveyor belt beforehand can significantly shorten the semi-wet mixing stage. When the moisture content is higher than 22%, pre-dry mixing is performed. The original soil is turned and moistened using a turner, with the moisture content measured after each turning, until the moisture content drops below 22%. Excessively wet soil will clump together during the dry mixing stage, making it difficult for the solidifying agent to coat the soil effectively. When the percentage of excess particles and the moisture content simultaneously trigger multiple pretreatment procedures, pre-crushing is performed first. This is because the internal moisture content of large particles cannot be read by surface detection methods, and wetting and moisture dissipation only affect the particle surface. Crushing first, followed by moisture content adjustment, ensures that the moisture content reading represents the entire batch of original soil. Pre-wetting and pre-dry mixing will only trigger one of them depending on the direction of moisture content deviation. When the percentage of excess particles is below 8% and the moisture content reaches 12% but is below 22%, the original soil proceeds to step 2.

[0023] In step 2, the wet weight of the original soil measured by the belt scale is 3300 kg, and the moisture content detection device indicates a moisture content of 14.2%. Ordinary Portland cement with a strength grade of 42.5 is used as the curing agent, with a preset dosage ratio of 8% of the dry soil mass, corresponding to 231 kg. This cement is simultaneously added to the mixing chamber along with the original soil via a screw conveyor to form the mixed material. The mixing shaft speeds for the dry mixing section, semi-wet mixing section, and wet mixing section are 28 rpm for dry mixing, 22 rpm for semi-wet mixing, and 35 rpm for wet mixing, respectively. In the dry mixing section, a medium speed is used for scattering and turning, allowing cement particles to coat the surface of the soil particles through repeated scattering. In the semi-wet mixing section, the speed is reduced because the material resistance is greatest during the liquid bridge formation period. The lower speed reduces peak torque, avoids overloading the drive device, and extends the shear-kneading time for moisture to penetrate into the particle clusters. In the wet mixing section, the speed is increased, relying on a higher shear rate to achieve overall homogenization; therefore, the wet mixing speed is higher than the semi-wet mixing speed.

[0024] The state recognition process is continuously executed from the start of the dry mixing section, and continues throughout the water addition process, the additional spraying process, and the subsequent mixing process. The torque sensor continuously outputs torque signals with a fixed sampling period of 10 milliseconds; the torque signal at time t is denoted as M(t). A second-order Butterworth low-pass filter is applied to M(t), with a cutoff frequency of 0.2 Hz. The slowly varying portion retained by the filter is the trend component, and the rapidly varying portion remaining after filtering out the trend component is the fluctuation component.

[0025] In the formula, M(t) is the torque signal at time t; ML(t) is the trend component at time t; MH(t) is the fluctuation component at time t; all three are in Newton-meters (N·m). The key to the valid division is that the cutoff frequency is lower than the rotational frequency of the stirring shaft: the rotational frequency corresponding to the 3rd gear operating speed of 22 to 35 revolutions per minute is 0.37 Hz to 0.58 Hz, and the cutoff frequency of 0.2 Hz is below it. The periodic disturbances caused by the impeller sweeping the material per revolution of the stirring shaft, as well as the rapid changes caused by particle collisions, all fall into the fluctuation component, while the trend component only retains the slow change in the overall rheological resistance of the material. A preset discrimination period of 5 seconds is used as one discrimination cycle, and the discrimination cycles are sequentially connected in chronological order, with each discrimination cycle covering 500 sampling cycles. The trend level of the kth discrimination cycle is the average value of the trend component within the discrimination cycle.

[0026] In the formula, Tk is the trend level of the k-th discrimination period, in Newton-meters (N·m); Ns is the number of sampling points in one discrimination period, which is 500 in this embodiment; ti is the time of the i-th sampling point in the discrimination period; ML(ti) is the trend component at time ti. The fluctuation amplitude is the difference between the maximum and minimum values ​​of the fluctuation component in the discrimination period.

[0027] In the formula, Ak is the fluctuation amplitude of the k-th discrimination period, in Newton-meters; MH,max and MH,min are the maximum and minimum values ​​of the fluctuation component within the discrimination period, respectively. Figure 2 The measured signal for one batch of mixing is given: Figure 2 In (a), the thin line represents the torque signal, and the thick line represents the trend component. Water is added starting around 55 seconds, and the trend component then rises along the slope and stabilizes around 200 seconds. Figure 2 In the middle (b), the corresponding wave component is shown. The envelope of the granular stage is close to ±260 N·m, and it narrows to around ±75 N·m after entering the plastic flow dynamics.

[0028] The orderliness index is extracted directly from the torque signal. Within each discrimination period, the sampling point is slid sequentially along the time axis, incrementing by one sampling point, to extract permutations of four consecutive sampling points from the torque signal. In this embodiment, the preset number of consecutive sampling points is 4; 500 sampling points yield 497 permutations. The four sampling points within each permutation are sorted by value from smallest to largest, and sampling points with equal values ​​are sorted by time sequence. The sequential positional relationship of each sampling point within the original permutation is recorded as one permutation pattern. There are 24 possible permutation patterns for the sequential position of the four sampling points. Figure 4 (a) illustrates the truncation and encoding process using 8 consecutive sampling points as an example: Permutation group 1 consists of sampling points 1 to 4, and their order of appearance after being sorted from smallest to largest is 2-4-1-3, meaning that the smallest value is the 2nd sampling point in the permutation group, followed by the 4th, 1st, and 3rd sampling points in sequence; Permutation group 2 consists of sampling points 2 to 5, and their order of appearance is 1-3-4-2. The frequency of each permutation pattern in all permutation groups within the statistical discrimination period is calculated using the following formula:

[0029] In the formula, Pk is the orderliness index of the k-th discrimination period, with a value ranging from 0 to 1; nm is the number of occurrences of the most frequent arrangement pattern; Np is the total number of arrangement groups within the discrimination period, and in this embodiment, Np is 497. The reason for setting the number of consecutive presets to 4 is as follows: when the number is 3, there are only 6 arrangement patterns, and the frequency distribution difference between granular state and plastic flow dynamics is small, resulting in insufficient discrimination; when the number is 5, there are as many as 120 arrangement patterns, and the average number of occurrences of each arrangement pattern within the 5-second discrimination period is less than 5, and statistical fluctuations mask the distribution differences. If the discrimination period is extended, the response of the state recognition will be slowed down; when the number is 4, the 24 arrangement patterns are statistically sufficient with a sample size of 497 arrangement groups, balancing discrimination and real-time performance. The permutation group is directly taken from the torque signal without first subtracting the trend component because of the time scale: the time span of one permutation group is only 40 milliseconds, and the rate of change of the trend component at the steepest point of the liquid bridge transition is about 12 N·m / s. The trend increment within 40 milliseconds is less than 0.5 N·m, which is negligible compared to the instantaneous amplitude of the fluctuation component, which is tens to hundreds of N·m. The positional relationship within the permutation group is actually determined by the rapid change component. In the granular stage, the collision and impact between the blade and loose particles are irregular, and no single permutation pattern is dominant. The frequency of the 24 permutation patterns tends to be uniform, and the orderliness index is low. In the plastic flow dynamic stage, the material flows with the blade as a whole, and the torque signal exhibits smooth quasi-periodic fluctuations. Monotonically rising and monotonically falling permutation patterns dominate, and the orderliness index increases significantly. Figure 4In (b), the frequency distribution of the arrangement patterns of the two states of matter in one discrimination period was compared. The mode of the granular state was about 0.09, while the mode of the plastic flow state reached 0.47. The two can be stably separated with 0.30 as the boundary.

[0030] Each discrimination cycle provides a state determination based on three characteristics: when the change in trend level between adjacent discrimination cycles is less than a preset stability threshold, the fluctuation amplitude is greater than a preset fluctuation threshold, and the orderliness index is less than a preset orderliness threshold, the mixed material is determined to be in a granular state; when the trend level increases with each discrimination cycle, the increase between adjacent discrimination cycles is greater than a preset stability threshold, and the fluctuation amplitude is greater than a preset fluctuation threshold, the mixed material is determined to be in a liquid bridge transition state; when the change in trend level between adjacent discrimination cycles is less than a preset stability threshold, the fluctuation amplitude is less than a preset fluctuation threshold, and the orderliness index is greater than a preset orderliness threshold, the mixed material is determined to be in a plastic flow state. In this embodiment, the preset stability threshold is 40 N·m, the preset fluctuation threshold is 300 N·m, the preset orderliness threshold is 0.30, and the preset confirmation time is 15 seconds. When the discrimination result of a certain state is continuously true for three consecutive discrimination cycles, the mixed material is confirmed to be in the corresponding state. Thresholds were calibrated through trial mixing: Five batches of representative soil samples were mixed, and the value ranges of each characteristic in the stable phase of the three physical states were recorded. The preset stability threshold was set at 1.5 times the maximum observed value of the trend level change in the adjacent discrimination period of the granular state plateau. The preset fluctuation threshold and preset orderliness threshold were set at the midpoints between the two distributions of the granular state and plastic flow dynamics, respectively. The trend levels of the granular state and plastic flow dynamics also showed stable plateaus. The trend curves alone could not distinguish between the completion of dry mixing and the completion of wet mixing. After incorporating the fluctuation amplitude and orderliness index into the criteria, the granular state showed high fluctuation and low orderliness, while the plastic flow dynamics showed low fluctuation and high orderliness. The two types of plateaus were thus clearly separated, which is also the basis for using a combination of three characteristics for discrimination. There is one constraint in the selection of the preset confirmation time: the confirmation time should be shorter than the actual duration of the liquid bridge transition state. If the liquid bridge transition state ends before confirmation is obtained, the switching logic will lose its landing point. In this embodiment, the liquid bridge transition segment is measured to last from 90 seconds to 150 seconds, with 15 seconds leaving sufficient margin. The confirmation time should also cover at least 2 discrimination cycles, so that occasional disturbances in a single discrimination cycle are blocked from the segmentation condition. Figure 3 This illustrates the characteristic evolution of a complete mixing process in one pan: Figure 3 In the middle (a), the trend curve is horizontal. The dry mixing platform is about 950 N·m, and it rises after adding water. The plastic platform is about 2400 N·m. The two vertical dotted lines are the switching moments of the segments. Figure 3 In (b), the fluctuation range narrowed from about 520 N·m to about 150 N·m, and the order index increased from about 0.12 to 0.47. The two horizontal dashed lines are the preset fluctuation threshold and the preset order threshold, respectively. The three material state intervals from left to right are granular state, liquid bridge transition state and plastic flow dynamics.

[0031] In the dry mixing section, all atomizing nozzles remain closed, and the mixed material is dry-mixed at a speed of 28 revolutions per minute. Initially, cement-rich clumps are intermittently broken up by the paddles, resulting in a significant abrupt change in resistance in the torque signal, and the trend is not yet stable; the determination of the granular state is intermittent. As cement gradually coats the soil particles, the resistance abrupt changes subside, and the three conditions become consistently met. The granular state is confirmed after a 15-second confirmation period, and the mixed material switches from the dry mixing section to the semi-wet mixing section. This confirmation mechanism also serves as the criterion for the dry mixing endpoint: the confirmation of the granular state means that the curing agent has been dispersed properly, and further dry mixing will no longer be effective. In this embodiment, the actual duration of the dry mixing section is approximately 55 seconds. The dry mixing time for the same batch of soil under a fixed-duration process is uniformly set to 90 seconds, and the adaptive switching shortens the dry mixing section by about one-third.

[0032] After entering the semi-wet mixing section, the rotation speed is reduced to 22 revolutions per minute, and the six atomizing nozzle groups spray synchronously in a pulse spray manner. The target injection volume of mixing water is determined by the following formula:

[0033] In the formula, W is the target injection amount in kilograms; m is the wet weight of the original soil added to the mixing chamber, which is 3300 kilograms in this embodiment; w0 is the moisture content detected in step 1, which is 14.2% in this embodiment; wt is the target moisture content of the mixed material, which is 18% in this embodiment. Substituting these values, W is approximately 110 kilograms. Each spray pulse consists of the atomizing nozzle group being turned on and spraying for a preset duration of 2 seconds before turning off. The interval between adjacent spray pulses is a preset interval of 2 seconds. Each spray pulse from a single atomizing nozzle group sprays approximately 1.5 kilograms of water. When 6 groups are executed simultaneously, each round of spray pulses adds approximately 9 kilograms of water. The target injection amount is completed in approximately 13 rounds, lasting more than 50 seconds. The mixing water is evenly distributed across all spray zones and added to the mixed material in the form of droplets. The interval duration is used to allow time for the droplets to settle and be captured by the turned material; if the spraying is continuous, the droplets suspended in the cavity will merge into large droplets and fall, which is no different from concentrated water falling; the pulse combined with the intermittent rhythm, plus the even distribution of each spray zone, the mixing water is evenly spread on the constantly turned mixing surface, and the conditions for the formation of overly wet mud clumps are lost.

[0034] After the target amount of water has been added and a preset waiting time of 20 seconds has elapsed since the addition, the most recently confirmed physical state of the mixed material is checked. If it is in a granular state or a liquid bridge transition state, it indicates that the detected moisture content is too high or the soil sample absorbs water more strongly than expected, and the actual amount of water has not yet brought the material into a plastic flow state. All atomizing nozzle groups are then controlled to perform an additional spray in a pulse spray manner. The amount of water used in this additional spray is included in the total mixing water volume. The waiting time is then reset to 20 seconds, and the above judgment and additional spray are repeated. The total mixing water volume is constrained by a preset upper limit for the total water volume, which is calculated by increasing the target moisture content by 2 percentage points. In this example, it is approximately 168 kg. The amount of water sprayed during the pre-wetting treatment is also included. When the total amount of water used for mixing reaches the preset upper limit for total water addition, and the most recently confirmed state of the mixed material is still in the granular state or liquid bridge transition state, the system will output a manual intervention prompt via an audible and visual alarm and a human-machine interface, and will stop the machine for inspection, prompting the operator to check the soil sample properties or the status of the testing device. The 20-second waiting period covers four judgment cycles, which is sufficient for the additional water added in one round to disperse and be reflected in the torque characteristics. In the closed loop of addition and waiting, the trend level passes the climbing section, the fluctuation amplitude decreases, the orderliness index increases, and the liquid bridge transition state and plastic flow dynamics are confirmed one after another. The mixed material switches to the wet mixing section. The switching is conditional on the confirmation of the two states one after the other. The climbing process of the trend level is thus included in the confirmation chain. Occasional stability caused by short-term disturbances will not be misjudged as wet mixing completion. In the wet mixing section, the stirring shaft rotates at 35 revolutions per minute to homogenize and stir the mixed material. When the judgment result of plastic flow dynamics is continuously valid for a period of time that reaches the preset homogenization time of 30 seconds, the segmented mixing ends and proceeds to step 3.

[0035] Step 3 begins by reducing the stirring shaft speed to a circulating speed of 6 revolutions per minute, which is lower than the dry mixing speed, semi-wet mixing speed, and wet mixing speed. This speed reduction is based on two considerations: First, at the operating speed, the mixing surface tumbles violently, resulting in motion blur in the image. The dry area is difficult to stabilize due to the rapid agitation of the blades. At the circulating speed, the mixing surface slowly tumbles axially at approximately 0.03 meters per second, allowing a 10 frames per second industrial camera to continuously track it. Second, maintaining slow tumbling without continuous rotation ensures continuous updating of the mixing surface, and the image evaluation reflects the overall material rather than a static surface layer. The fluctuation amplitude of the last judgment cycle before the stirring shaft speed is reduced to the circulating speed is used as the evaluation fluctuation amplitude. The fluctuation amplitude is strongly correlated with the rotation speed. During circulating mixing, the torque generally decreases, and the value taken during circulating mixing will mismatch with the preset fluctuation threshold. Taking the value of the last judgment cycle before the speed reduction is both close to the discharge time and maintains comparability with the threshold.

[0036] An industrial camera acquires images of the mixing surface under supplementary lighting with an exposure time of 2 milliseconds. This short exposure further reduces the impact of ambient light fluctuations. After grayscale processing, the images are divided into six sub-regions based on the coverage positions of the six spray zones. The pixel coordinates of the spray zone boundaries in the image are determined once during installation calibration and remain valid long-term. Let the mean grayscale value of the j-th sub-region be Gj, and the overall average grayscale value be Gm. The grayscale deviation is calculated using the following formula:

[0037] In the formula, Dj is the grayscale deviation of the j-th image sub-region; Gj is the mean grayscale value of the j-th image sub-region; Gm is the overall average grayscale value of all image sub-regions; all three are in grayscale levels. Image sub-regions where Dj is greater than the preset grayscale deviation threshold are determined as dry regions, and the determination of dry regions is performed independently for each image sub-region. The physical basis for the determination is the negative correlation between moisture and surface brightness: moist soil surfaces absorb light strongly and reflect it weakly; the surface grayscale of mixed materials with low moisture content is higher than that of mixed materials with high moisture content. For the soil samples of this project, for every 1 percentage point decrease in moisture content, the mean surface grayscale value increases by approximately 5 grayscale levels. The preset grayscale deviation threshold is 18 grayscale levels, which corresponds to a moisture content that is about 3.6 percentage points lower. This level of local water shortage manifests as a low strength after the solidified soil is formed, and must be dealt with. The 18 grayscale levels are also significantly higher than the grayscale fluctuations caused by light and texture (measured at about ±6 grayscale levels). Ordinary texture differences are excluded from the dry area judgment.

[0038] by Figure 5 Taking the first evaluation as an example: the average gray levels of the six image sub-regions are 96, 94, 132, 95, 97, and 93 respectively, with an overall average of 101.2. Sub-region 3 has a gray level deviation of 30.8 gray levels, exceeding the preset gray level deviation threshold, and is therefore judged as a dry region. Figure 5 In section (a), the dry area is marked with dots and the direction of turning is indicated by an arrow; Figure 5(b) The average grayscale value of each image sub-region is given by bar chart. The solid line is the overall average value, and the dashed line is the dry area judgment line obtained by adding the overall average value to the preset grayscale deviation threshold. The bar of sub-region 3 crosses the judgment line. The uniformity judgment is performed according to three conditions: when the number of dry areas is 0, the difference between the maximum and minimum values ​​of the average grayscale values ​​of all image sub-regions is less than the preset uniformity threshold of 25 grayscale levels, and the evaluation fluctuation amplitude is less than the preset fluctuation threshold of 300 N·m, the mixing uniformity is judged to meet the standard, the stirring shaft speed is restored to the wet mixing speed and the material is discharged; when there are dry areas, or the difference between the maximum and minimum values ​​reaches more than 25 grayscale levels, or the evaluation fluctuation amplitude reaches more than 300 N·m, the mixing uniformity is judged to need to be improved, and the supplementary mixing strategy is triggered. The grayscale difference between image sub-regions reflects the surface moisture distribution, and the evaluation fluctuation amplitude reflects the rheological state inside the material. The two channels are complementary: when the surface has just been renovated and covered while the inside is still raw, the evaluation fluctuation amplitude is high; when the inside is generally flowing but there are local dry spots on the surface, the image channel reports a dry area. Figure 5 In the example, the difference between the maximum and minimum values ​​is 39 gray levels. Together with the existence of dry areas, the uniformity assessment concludes that it needs to be improved.

[0039] During the remixing strategy, the mixing shaft maintains a constant rotation speed, and the mixing surface image is re-acquired every 2 seconds according to a preset refresh cycle. Within each refresh cycle, the image sub-regions are re-divided based on the latest image, the average grayscale value and the overall average value are recalculated, and the dry area is re-determined. The spray zone corresponding to the image sub-region where the dry area is located is recorded as the current spray zone of the dry area. The remixing water is sprayed onto the dry area in the form of spray pulses. Only the atomizing nozzle group covering the dry area is activated. The number of spray pulses for a single dry area is set according to the preset range of grayscale deviation: 18 grayscale levels higher than the overall average but within 26 grayscale levels fall into the first range, with 2 spray pulses; 26 grayscale levels higher than the overall average but within 34 grayscale levels fall into the second range, with 3 spray pulses; and 34 grayscale levels higher than the overall average fall into the third range, with 4 spray pulses. The preset range is divided from low to high according to the amount of deviation, with higher ranges having more pulses. The conversion of the tiered water supply is based on two calibration relationships: approximately 5 gray levels correspond to 1 percentage point in the moisture content; the water replenishment amount of a single spray pulse to a single spray zone is equivalent to an increase of approximately 0.8 percentage points in the moisture content of the surface material of the spray zone. Each tier increases the water supply by one pulse, and the water replenishment amount is roughly equivalent to the degree of water shortage, with more water replenished to dry areas with greater water shortage and less water replenished to dry areas with less water shortage. Figure 5 In the example, the grayscale deviation in the dry area is 30.8 grayscale levels, which falls into the second level and corresponds to 3 spray pulses.

[0040] The direction of material tumbling on the mixing surface is determined by the rotation of the stirring shaft. In this embodiment, the impeller has a 15-degree installation tilt angle. Under the direction of the stirring rotation, the material slowly moves from spray zone 1 to spray zone 6 at an axial speed of approximately 0.03 meters per second. The dry zone moves along the tumbling direction with the material. The axial dimension of the dry zone is typically 0.3 to 0.5 meters. It takes about 13 seconds to cross a 0.4-meter-wide spray zone. The preset refresh cycle of 2 seconds is much shorter than the crossing time. The position update of the dry zone in each refresh cycle is only about 0.06 meters, and the tracking will not be interrupted. The spray zone where the dry zone is currently located is recorded as the original spray zone, and the spray zone that it is about to enter along the tumbling direction is recorded as the successor spray zone. The boundary area is defined with the boundary line between the two as the center and 0.15 meters before and after along the tumbling direction. When the dry zone enters the boundary area along the leading edge of the turning direction, the atomizing nozzle group corresponding to the next spray zone is activated first; after all connected pixels of the dry zone enter the image sub-region corresponding to the next spray zone, the atomizing nozzle group corresponding to the original spray zone is then deactivated. This pre-activation and post-deactivation is based on two considerations: the settling of droplets from the nozzle to the mixing surface takes 0.3 to 0.5 seconds, so activating it in advance ensures the fog field is established when the dry zone arrives; during the dry zone's crossing of the boundary line, any single atomizing nozzle group can only cover a portion of the dry zone, and the two overlapping spray groups ensure the spray point continuously follows and covers the moving dry zone. Figure 6 (a) shows a top-down view at two moments: at moment 1, the leading edge of the dry area enters the boundary region, and the atomizing nozzle groups corresponding to spray zone 3 and spray zone 4 are turned on simultaneously; at moment 2, the entire dry area enters the image sub-region corresponding to spray zone 4, and the atomizing nozzle group corresponding to spray zone 3 is turned off. Figure 6 (b) shows the corresponding opening and closing sequence. The atomizing nozzle group of the next spray zone opens first at the 3rd second, and the atomizing nozzle group of the original spray zone closes after the 9th second. The cumulative spray volume of the supplementary water is included in the total mixing water volume, which is also constrained by the preset maximum water addition limit of 168 kg. The total amount of targeted water replenishment is usually only a few kilograms to a dozen kilograms, and the increase in the overall moisture content of the mixed material is compressed to within 0.5 percentage points, thus maintaining the predetermined mix proportion. After water replenishment, the moisture content of the mixed material in the dry zone increases and the surface grayness decreases, and it exits the dry zone judgment in subsequent refresh cycles.

[0041] After all the spray pulses corresponding to the dry areas have been executed, the stirring shaft speed returns to the wet mixing speed of 35 rpm for supplementary mixing. The supplementary mixing water is concentrated on the surface and needs to be homogenized and seeped down again. The preset supplementary mixing time is 20 seconds. It takes about 8 to 10 seconds for the blades to complete one overall flip of the surface material. 20 seconds ensures at least two overall flips. After the supplementary mixing time reaches the preset time, the stirring shaft speed drops back to the patrol mixing speed, and the uniformity judgment is re-executed. If the uniformity of the mixture is determined to be up to standard, the wet mixing speed is restored and the material is discharged. If the cumulative number of times the supplementary mixing strategy is executed reaches the preset limit of 3 times and the uniformity judgment still gives a conclusion that needs improvement, a manual intervention prompt is output and the machine is stopped for inspection: if the three supplementary mixing attempts still fail to converge, it is mostly due to threshold drift, nozzle blockage, or sudden changes in soil properties, which require manual investigation. Continuing automatic supplementary mixing will only continue to inject water into the material. The upper limit of the total amount of water added and the upper limit of the number of times constrain the process from two aspects: water volume and number of cycles. The mixing process must terminate within a limited time.

[0042] Twenty batches of the same soil source were continuously mixed at a solidified soil mixing plant. Another twenty batches were mixed using a fixed-duration process (90 seconds of dry mixing, 60 seconds of concentrated water addition, 120 seconds of wet mixing, and a 30-second delay before discharge) as a control. After discharge, six samples were taken from each batch, and the cement dosage was determined using the disodium EDTA titration method. The average coefficient of variation for dosage in each batch of this invention was 5.1%, compared to 12.4% in the control process. The mixing cycle per batch was shortened from approximately 330 seconds in the control process to approximately 255 seconds. The average unconfined compressive strength at 28 days increased from 1.62 MPa to 1.83 MPa, and the coefficient of variation for strength decreased from 18.7% to 9.5%. Five of the twenty batches triggered the supplementary mixing strategy, all meeting the discharge standard after one supplementary mixing. The remaining 15 batches met the standard upon the first uniformity assessment.

[0043] The parameter values ​​and device selections given above illustrate one feasible implementation. Multiple alternative implementations exist around the same concept. The torque signal can be obtained by converting the output current of the stirring shaft drive motor through a calibration relationship. The analog output of the frequency converter can provide the current signal. During calibration, the current and flange torque are recorded synchronously under several load conditions, and a linear relationship is fitted. This conversion method eliminates the need for a torque sensor, making it suitable for retrofitting existing stirring equipment. The low-pass filter can be replaced by a 2-second moving average, whose attenuation of components above 0.37 Hz is comparable to the cutoff characteristics of this embodiment, making engineering implementation simpler. The number of consecutive presets can be selected between 3 and 5. When 5 is selected, the preset discrimination time is correspondingly extended to more than 8 seconds to ensure the sample size of the permutation group. The preset discrimination time can be adjusted between 3 and 8 seconds; a shorter value is used in cases of uniform soil and minimal disturbance to accelerate the response.

[0044] The moisture content detection device can be replaced with a near-infrared reflective type, which can be installed above the belt for operation and is more convenient to maintain; both the detection camera and the industrial camera can be directly selected as black and white area array cameras, and grayscale processing is degraded to directly reading the grayscale values ​​of the black and white image. The number of atomizing nozzle groups can be set between 4 and 8 groups. The longer the mixing chamber, the more zones should be; the preset interval can be divided into 2 to 5 levels. The more levels, the finer the resolution of directional water replenishment; the width of the boundary area can be selected between 0.1 meters and 0.2 meters. For equipment with fast turning speed, the wider value should be used. The curing agent can be a composite material of lime, slag powder and cement. The dry area is determined by the relationship between moisture and surface brightness, and is not related to the type of curing agent. The toothed roller pre-crushing device can be replaced by a crushing and screening bucket, and the crushing and feeding can be completed by the loader at the material yard. The twin-shaft forced mixer can be replaced by a single-shaft or planetary mixer. The spraying zones are arranged along the main circulation direction of the material, and the turning direction is determined by the rotation of the mixing device and the arrangement of the blades. The whole process of segmented mixing, physical state recognition and positioning water replenishment is consistent.

[0045] The connection method of the discrimination cycle can be changed from sequential connection to half-cycle overlapping sliding, which doubles the response of the physical state recognition, at the cost of increased computation. The preset spraying time and preset intermittent time can be matched between 0.5 seconds and 5 seconds, with short spray and long intermittent for nozzles with high water pressure and fine droplets. The preset homogenization time can be selected between 15 seconds and 60 seconds, and the preset remixing time can be selected between 10 seconds and 40 seconds. The target moisture content is given according to the mix design of the solidified soil. When changing the mix ratio, only the target moisture content and the preset admixture ratio need to be rewritten. In the uniformity evaluation stage, three frames of mixing surface images can be continuously acquired, and the median of the gray value of each image sub-region is used for judgment. Mud spots splashed and adhered to the observation window are thus excluded from the evaluation. An air curtain blowing structure can also be added to the observation window to prevent droplets and dust from adhering and ensure the long-term imaging quality of the industrial camera.

[0046] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A method for segmented mixing and uniformity improvement in the solidification soil mixing process, characterized in that, Includes the following steps: Step 1: Detect the moisture content and percentage of excess particles in the original soil, and determine and implement the pretreatment plan based on the test results; Step 2: The raw soil and solidifying agent treated in Step 1 are added to the mixing chamber of the mixer to form a mixed material. The dry mixing section, semi-wet mixing section, and wet mixing section are executed sequentially. The torque signal of the mixing shaft is collected, and three features—trend level, fluctuation amplitude, and orderliness index—are extracted in each consecutive discrimination cycle. Based on the three features, the mixed material is classified into a granular state, a liquid bridge transition state, or a plastic flow state. When the discrimination result of the material state is continuously valid within the preset confirmation time, the mixed material is confirmed to be in the corresponding material state. When the mixed material is confirmed to be in a granular state, the dry mixing section is switched to the semi-wet mixing section. In the semi-wet mixing section, mixing water is sprayed onto the mixed material through multiple atomizing nozzle groups covering different spray zones. When the mixed material is successively confirmed to be in a liquid bridge transition state and a plastic flow state, the wet mixing section is switched to. In the wet mixing section, homogenization and mixing are carried out. When the discrimination result of the plastic flow state is continuously valid for a preset homogenization time, the process proceeds to Step 3. Step 3: Collect images of the mixing surface and divide them into corresponding image sub-regions according to the coverage position of each spray zone. Calculate the grayscale mean of each image sub-region and the overall average of all grayscale mean values. Image sub-regions whose grayscale mean exceeds the overall average by a amount greater than the preset grayscale deviation threshold are identified as dry areas. Based on the determination of dry areas, perform uniformity determination. If the mixing uniformity is determined to be in need of improvement, trigger the supplementary mixing strategy: only turn on the atomizing nozzle group covering the dry area to spray supplementary mixing water to the dry area, then perform supplementary mixing and re-execute uniformity determination until the mixing uniformity is determined to meet the standard before discharging.

2. The method according to claim 1, characterized in that: The pretreatment scheme includes pre-crushing, pre-wetting, and pre-dry mixing. In step 1, the raw soil is conveyed to the mixer via a feeding belt. A detection camera above the feeding belt captures images of the raw soil. The images are then processed for grayscale and threshold segmentation. The connected regions corresponding to each particle cluster are extracted, and the equivalent circle diameter of the connected region is taken as the equivalent particle size of the particle. The proportion of particle areas with equivalent particle sizes greater than the preset upper limit is calculated as the proportion of oversized particle clusters in the raw soil image. The moisture content of the raw soil is detected by a moisture content detection device located at the discharge end of the feeding belt. When the proportion of oversized particle clusters exceeds the preset threshold, pre-crushing is performed. The raw soil is fed into a toothed roller pre-crushing device for crushing. The crushed raw soil is then fed back to the mixing belt via a feeding belt. The soil is transported and re-tested until the proportion of excess particles drops below the preset threshold. When the moisture content is below the preset lower limit, pre-wetting treatment is performed by spraying atomized water onto the surface of the original soil and mixing it. The amount of atomized water sprayed is included in the total mixing water until the moisture content reaches the preset lower limit. When the moisture content is above the preset upper limit, pre-dry mixing treatment is performed by mixing and dissipating moisture from the original soil until the moisture content drops below the preset upper limit. When the detection results of excess particle proportion and moisture content trigger multiple pretreatment schemes simultaneously, pre-crushing treatment is performed before pre-wetting treatment and pre-dry mixing treatment. When the proportion of excess particles is below the preset threshold, the moisture content reaches the preset lower limit, and the moisture content is below the preset upper limit, the original soil proceeds to step 2.

3. The method according to claim 1, characterized in that: In step 2, the acquisition of torque signals, the extraction of three features, and the determination and confirmation of the state of matter are continuously performed from the start of the dry mixing section. The torque signal is continuously collected by a torque sensor installed at the drive end of the stirring shaft at a fixed sampling period. The slowly changing part of the torque signal retained after low-pass filtering is taken as the trend component, and the fast changing part of the torque signal after filtering out the trend component is taken as the fluctuation component. The cutoff frequency of the low-pass filter is set to be lower than the rotation frequency of the stirring shaft. A preset discrimination time is defined as one discrimination cycle, and the preset discrimination time covers multiple sampling cycles; The trend level is the average value of the trend component within the discrimination period, and the fluctuation amplitude is the difference between the maximum and minimum values ​​of the fluctuation component within the discrimination period.

4. The method according to claim 3, characterized in that: The process of extracting the orderliness index is as follows: within the discrimination period, slide and extract a permutation group consisting of a preset number of consecutive sampling points from the torque signal in chronological order, with each sampling point as a step. Sort the sampling points in each permutation group according to their values ​​from smallest to largest, and sort the sampling points with the same values ​​according to their time order. Record the sequential position relationship of each sampling point in the original permutation group as a permutation pattern. Count the number of occurrences of each permutation pattern in all permutation groups. The proportion of the occurrences of the permutation pattern with the most occurrences to the total number of permutation groups is used as the orderliness index.

5. The method according to claim 1, characterized in that: The rule for determining the physical state of the mixed material based on three characteristics is as follows: when the change in trend level between adjacent determination periods is less than the preset stability threshold, the fluctuation amplitude is greater than the preset fluctuation threshold, and the order index is less than the preset order threshold, the mixed material is determined to be in a granular state. When the trend level increases with each discrimination period, the increase between adjacent discrimination periods is greater than the preset stability threshold, and the fluctuation amplitude is greater than the preset fluctuation threshold, the mixed material is judged to be in a liquid bridge transition state; when the change in the trend level between adjacent discrimination periods is less than the preset stability threshold, the fluctuation amplitude is less than the preset fluctuation threshold, and the order index is greater than the preset order threshold, the mixed material is judged to be in a plastic flow dynamic.

6. The method according to claim 1, characterized in that: In step 2, the curing agent is added to the mixing chamber according to a preset dosage ratio via a spiral metering conveyor. Multiple atomizing nozzle groups are arranged along the axial direction of the mixing shaft at the top of the mixing chamber. Each atomizing nozzle group covers one spraying zone on the top surface of the mixing chamber, and all spraying zones are combined to cover the entire mixing surface. In the dry mixing section, the mixing shaft operates at a dry mixing speed, and all atomizing nozzle groups remain closed to dry mix the original soil and curing agent. In the semi-wet mixing section, the mixing shaft operates at a semi-wet mixing speed, and all atomizing nozzle groups spray synchronously in the form of spray pulses. Each spray pulse is when the atomizing nozzle group is turned on and sprays for a preset duration before turning off. There is a preset interval between adjacent spray pulses. The mixing water is evenly distributed in all spraying zones and added to the mixing material in the form of droplets. The target amount of mixing water is determined based on the target moisture content of the mixing material and the moisture content detected in step 1. Spraying continues until the target amount is added. In the wet mixing section, the stirring shaft operates at the wet mixing speed, which is higher than the semi-wet mixing speed.

7. The method according to claim 6, characterized in that: In the semi-wet mixing section, when the target amount of water is added and the material being mixed is confirmed to be in a granular state or a liquid bridge transition state after a preset waiting time since the addition is completed, all atomizing nozzle groups are controlled to add additional spray in the form of spray pulses. The amount of water added is included in the total mixing water, which is subject to the preset upper limit of the total water added. After each additional spray, the preset waiting time is recalculated, and the above judgment and additional spray are repeated. When the total mixing water reaches the preset upper limit of the total water added, and the material being mixed is confirmed to be in a granular state or a liquid bridge transition state, a manual intervention prompt is output and the machine is stopped for inspection.

8. The method according to claim 6, characterized in that: In step 3, before acquiring the image of the mixing surface, the stirring shaft speed is reduced to the circulating mixing speed, which is lower than the dry mixing speed, semi-wet mixing speed, and wet mixing speed. This allows the mixing surface to slowly tumble and remain within the field of view of the industrial camera located at the top of the mixing chamber. The mixing surface image is acquired by the industrial camera under supplementary lighting and is then processed into grayscale to divide the image into sub-regions. The fluctuation amplitude of the last judgment cycle before the stirring shaft speed is reduced to the circulating mixing speed is used as the evaluation fluctuation amplitude. The uniformity judgment rule is as follows: when the number of dry areas is 0, the difference between the maximum and minimum values ​​of the grayscale average of all image sub-regions is less than the preset uniformity threshold, and the evaluation fluctuation amplitude is less than the preset fluctuation threshold, the mixing uniformity is judged to meet the standard, the stirring shaft speed is restored to the wet mixing speed, and the material is discharged. When there are dry areas, or the difference between the maximum and minimum values ​​reaches or exceeds the preset uniformity threshold, or the evaluation fluctuation amplitude reaches or exceeds the preset fluctuation threshold, the mixing uniformity is judged to need improvement, and a supplementary mixing strategy is triggered.

9. The method according to claim 8, characterized in that: In the supplementary mixing strategy, the stirring shaft is kept running at a constant stirring speed, and the mixing surface image is re-acquired according to a preset refresh cycle. In each refresh cycle, the image sub-regions are re-divided based on the latest acquired mixing surface image, the grayscale mean and the overall average value are recalculated, and the dry area is re-determined. The determination of the dry area is performed independently for each image sub-region. The spray zone corresponding to the image sub-region where the dry area is located in this determination is taken as the spray zone where the dry area is currently located. When spraying supplementary mixing water to the dry area, the spraying is performed in the form of spray pulses. The number of spray pulses corresponding to a single dry area is set according to the preset grading interval into which the grayscale mean of the dry area falls above the overall average value. The preset grading intervals are divided in ascending order of the amount above the overall average value. The higher the preset grading interval, the more spray pulses are set.

10. The method according to claim 9, characterized in that: In the supplementary mixing strategy, the direction of tumbling of the mixed materials on the mixing surface is determined by the rotation of the stirring shaft. The dry zone moves along the tumbling direction as the mixed materials tumble. The spray zone where the dry zone is currently located is recorded as the original spray zone, and the spray zone that the dry zone is about to enter along the tumbling direction is recorded as the subsequent spray zone. When the dry zone moves to the boundary area between the original spray zone and the subsequent spray zone, the atomizing nozzle group corresponding to the subsequent spray zone is turned on first. After the dry zone as a whole enters the image sub-area corresponding to the subsequent spray zone, the atomizing nozzle group corresponding to the original spray zone is turned off. The nozzle group is configured to continuously follow and cover the moving dry area with the spray. The cumulative spray volume of the supplementary mixing water is included in the total mixing water volume and is subject to the preset upper limit of the total water volume. After the spray pulses corresponding to all dry areas are completed, the stirring shaft speed is restored to the wet mixing speed for supplementary mixing. After the supplementary mixing time reaches the preset supplementary mixing time, the stirring shaft speed is reduced to the patrol mixing speed again, and the uniformity judgment is re-executed. When the cumulative number of times the supplementary mixing strategy is executed reaches the preset upper limit, a manual intervention prompt is output and the machine is stopped for inspection.

Images