Early warning and adjusting method for mixing material ratio fluctuation of flow-state solidified soil

Abstract

The invention provides a flow-state solidified soil mixing material ratio fluctuation early warning adjustment method, and belongs to the technical field of flow-state solidified soil construction.The method includes the steps that the input amount of each component is monitored in real time by building a material ratio fluctuation curve model, a material ratio stable interval is set to form an early warning value, material ratio abnormal points are recorded, and the material ratio fluctuation trend is calculated; applying a material ratio dynamic prediction equation set to analyze a historical change rule and predict a future trend, analyzing material ratio abnormal point relevance to carry out time sequence clustering, constructing a material ratio abnormal point and equipment start-stop relation model to reveal a fluctuation mechanism, calculating a material ratio fluctuation influence area to determine an influenced material range, and calculating the material ratio fluctuation influence area. And executing a material ratio fluctuation compensation program to automatically adjust the input amount of each component and guide fluctuating materials into a temporary storage area for treatment, and finally implementing material ratio recovery adjustment to gradually recover the input amount of each component to a standard value, so that the technical problem that the material ratio fluctuation is difficult to predict in real time and accurately regulate and control in the mixing process of the flow-state solidified soil in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of flow state solidified soil construction, and in particular relates to a flow state solidified soil mixing material ratio fluctuation early warning adjustment method. BACKGROUND

[0002] As an important foundation treatment material, flow state solidified soil is widely used in soft foundation reinforcement, underground structure backfilling and other engineering fields. The traditional flow state solidified soil mixing technology mainly relies on fixed proportion design and simple feeding control system to realize slurry mixing by presetting the input amount of each component. In large-scale engineering applications, continuous mixing equipment such as disc feeder, screw conveying system and other equipment are usually used for material metering and conveying.

[0003] However, the traditional flow state solidified soil mixing technology has obvious defects. Due to factors such as equipment start-stop, material property fluctuation, conveying system delay, etc., the feeding ratio of each component often fluctuates during the mixing process, resulting in quality problems such as uneven strength and unstable flowability of the final product. The existing technology mainly uses passive methods such as periodic sampling detection and manual intervention adjustment to cope with it, which is difficult to realize real-time monitoring and early warning of material ratio fluctuation.

[0004] At present, there is still a lack of a technical solution in the field of flow state solidified soil mixing that can effectively predict the fluctuation trend of the material ratio, analyze the influence range of the fluctuation and perform accurate compensation adjustment, making it difficult to solve the technical problem of real-time prediction and accurate control of material ratio fluctuation in the flow state solidified soil mixing process. SUMMARY

[0005] Therefore, the present application provides a flow state solidified soil mixing material ratio fluctuation early warning adjustment method, which can solve the technical problem of real-time prediction and accurate control of material ratio fluctuation in the flow state solidified soil mixing process in the prior art.

[0006] The present application is implemented as follows: The present application provides a flow state solidified soil mixing material ratio fluctuation early warning adjustment method, which includes: establishing a material ratio fluctuation curve model, real-time collecting the input amount of solidifying agent, the input amount of additive and the input amount of water through sensors, and drawing a material ratio-time curve; setting a material ratio stable interval to form an upper limit early warning value and a lower limit early warning value; recording material ratio abnormal points, the system automatically marking and recording relevant parameters; calculating the material ratio change trend, applying a material ratio dynamic prediction equation set to analyze the historical change law of each component material ratio and predict the future change trend, and evaluating the fluctuation risk coefficient; analyzing the correlation of material ratio abnormal points; constructing a material ratio abnormal point and equipment start-stop relationship model; calculating the material ratio fluctuation influence area; executing a material ratio fluctuation compensation program; and implementing material ratio recovery adjustment.

[0007] The material ratio fluctuation curve model is established by real-time collection of the curing agent input amount, the additive input amount and the water input amount through sensors to draw a material ratio-time curve.

[0008] The material ratio stable interval is set based on historical production data to determine the fluctuation range of the curing agent input amount, the additive input amount and the water input amount, and to form high and low limit early warning values.

[0009] The material ratio abnormal point is recorded when the real-time monitoring value exceeds the preset material ratio stable interval, and the material ratio abnormal point occurrence time and related parameters are recorded.

[0010] The material ratio abnormal point correlation is analyzed by time series clustering of the marked material ratio abnormal points to identify the time interval, amplitude similarity and triggering conditions between the material ratio abnormal points.

[0011] The material ratio abnormal point and equipment start-stop relationship model is constructed by establishing the correlation mapping between the material ratio abnormal point occurrence and the equipment running state, the disc start-stop and the material conveying delay.

[0012] The material ratio fluctuation influence area is calculated according to the material ratio abnormal point occurrence time and the material flow rate to determine the affected material location and range.

[0013] The material ratio fluctuation compensation program is executed when the material ratio abnormal point is detected, and the system automatically adjusts the curing agent input amount, the additive input amount and the water input amount to direct the fluctuating material to the preset area for temporary storage.

[0014] The material ratio recovery adjustment is implemented after the material ratio fluctuation compensation program processing to gradually restore the curing agent input amount, the additive input amount and the water input amount to the standard value to ensure production continuity and mixed material performance.

[0015] The material ratio dynamic prediction equation group includes a material ratio trend equation, a fluctuation amplitude equation and a prediction correction equation.

[0016] The present application forms a complete material ratio fluctuation early warning adjustment technical system through nine steps of establishing a material ratio fluctuation curve model, setting a material ratio stable interval, recording material ratio abnormal points, calculating material ratio variation trends, analyzing material ratio abnormal point correlations, constructing a material ratio abnormal point and device start-stop relationship model, calculating material ratio fluctuation influence areas, executing material ratio fluctuation compensation procedures and implementing material ratio recovery adjustments.

[0017] The method overcomes the defects of material ratio fluctuation monitoring lag and inaccurate adjustment in traditional technologies.

[0018] The present application changes material ratio fluctuation in a fluidized solidification soil mixing process from passive response to active prediction and precise control through mathematical modeling and intelligent analysis, effectively solves the technical problem of difficult real-time prediction and precise control of material ratio fluctuation in the fluidized solidification soil mixing process, and significantly improves the automation level and product quality stability of the fluidized solidification soil production. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The flowchart of the method of the present application. DETAILED DESCRIPTION

[0020] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0021] As Figure 1 shown is a flow chart of a flow state solidified soil mixing material ratio fluctuation early warning adjustment method provided by the present application, and the method comprises the following steps:

[0022] S01, a material ratio fluctuation curve model is established, three component input amounts of solidified agent input amount, additive input amount and water input amount are collected in real time through a sensor, and a material ratio-time change curve is drawn;

[0023] S02, a material ratio stable interval is set, based on historical production data, a fluctuation range of the material ratio of the three components of the solidified agent input amount, the additive input amount and the water input amount is determined, a high limit early warning value and a low limit early warning value are formed;

[0024] S03, a material ratio abnormal point is recorded, when the real-time monitoring value exceeds the preset material ratio stable interval, the system automatically marks the material ratio abnormal point, and records the time when the material ratio abnormal point occurs and related parameters;

[0025] S04, a material ratio change trend is calculated, based on continuous monitoring data, a material ratio dynamic prediction equation set is applied to analyze the historical change law of each component material ratio and predict the future change trend, and a fluctuation risk coefficient is evaluated;

[0026] S05, the correlation of the material ratio abnormal point is analyzed, the marked material ratio abnormal points are time series clustered, and the time interval between the material ratio abnormal points, the amplitude similarity and the triggering condition are identified;

[0027] S06, a material ratio abnormal point and equipment start-stop relationship model is constructed, a correlation mapping between the occurrence of the material ratio abnormal point and the equipment running state, the disc start-stop and the material conveying delay is established;

[0028] S07, a material ratio fluctuation influence area is calculated, according to the material ratio abnormal point occurrence time and the material flow rate, the affected material position and range are determined;

[0029] S08, a material ratio fluctuation compensation program is executed, when the material ratio abnormal point is detected, the system automatically adjusts the solidified agent input amount, the additive input amount and the water input amount, and the fluctuated material is guided into a preset area for temporary storage processing;

[0030] S09, a material ratio recovery adjustment is implemented, after the material ratio fluctuation compensation program processing, the solidified agent input amount, the additive input amount and the water input amount are gradually restored to the standard value, ensuring the production continuity and the mixed material performance meeting the standard;

[0031] The material ratio dynamic prediction equation set includes a material ratio trend equation, a fluctuation amplitude equation, and a prediction correction equation.

[0032] The material ratio trend equation is used to calculate the component feeding ratio change rate and acceleration in a selected time window, and the input includes the current time component feeding ratio value, the previous N time point component feeding ratio historical value, the time window length, the component type identifier, and the sampling frequency, and the output is the component feeding ratio change rate and the predicted change direction in the near future;

[0033] The fluctuation amplitude equation is used to evaluate the material ratio fluctuation amplitude and frequency characteristics, and the input includes the continuous M time point component feeding ratio value, the standard mixing ratio reference value, the historical fluctuation maximum peak value, the equipment working condition parameter, and the material property parameter, and the output is the fluctuation amplitude weighted mean value and the fluctuation period characteristic value.

[0034] The prediction correction equation is used to correct the prediction results according to the device start-stop state, and the input includes the initial prediction trend value, the device start-stop state parameter, the historical start-stop induced fluctuation statistical data, the material ratio abnormal point and device start-stop time correlation coefficient, and the material delivery delay time, and the output is the prediction trend value and the fluctuation risk coefficient after the device start-stop factor correction.

[0035] The material ratio fluctuation curve model is a curve drawn with time as the horizontal coordinate and each component feeding ratio as the vertical coordinate, which is used to visually display the change trend of the curing agent feeding amount, the additive feeding amount, and the water feeding amount with time in the mixing process.

[0036] The material ratio stable interval is the allowable fluctuation range of each component feeding ratio in the fluidified soil mixing process, which is usually set to be 0.5% of the standard mixing ratio, and exceeding this range is considered abnormal and triggers an early warning.

[0037] The material ratio abnormal point is the actual feeding ratio of each component at a certain time point in the mixing process, which exceeds the preset material ratio stable interval limit point, indicating that the mixing ratio deviates at that time and needs to be adjusted.

[0038] The material ratio change trend is the change rate and direction of each component feeding ratio calculated by analyzing the continuous monitoring data, which is used to predict whether the mixing ratio has a continuous deviation risk.

[0039] The material ratio abnormal point relevance is the time series analysis of multiple material ratio abnormal points to find the abnormal occurrence rules, including time interval mode, amplitude change characteristics, and common causes.

[0040] Among them, the material ratio anomaly point and equipment start-up and shutdown relationship model specifically refers to establishing the mapping relationship between the occurrence of material ratio anomaly points and changes in equipment operating status, revealing the mechanism of material ratio fluctuation caused by equipment start-up and shutdown and switching of operating conditions;

[0041] Among them, the material ratio fluctuation impact area specifically refers to the spatial location range of the material affected by the fluctuation, calculated based on the material flow rate and the duration of the abnormal point in the material ratio, which is used to determine the volume of the mixture that needs special treatment.

[0042] The material ratio fluctuation compensation program specifically refers to automatically adjusting the dosage of each component to offset the impact of fluctuations when an abnormal point in the material ratio is detected, while simultaneously importing the affected materials into a temporary storage area for secondary processing.

[0043] Among them, the material ratio recovery adjustment specifically refers to the process by which the system gradually restores the amount of curing agent, admixture, and water to the standard ratio value after the material ratio fluctuation compensation procedure is executed, including gradient adjustment of the feeding rate and monitoring of the performance indicators of the mixture.

[0044] Among them, the rate of change of component feeding ratio specifically refers to the change in the proportion of curing agent, additive, and water input per unit time, which is used to characterize the speed of change of material ratio.

[0045] Among them, the acceleration of the change in the component feeding ratio specifically refers to the change in the rate of change of the component feeding ratio per unit time, which is used to characterize whether the trend of the material ratio change is accelerating or slowing down.

[0046] Among them, the weighted average of fluctuation range specifically refers to the average value calculated by assigning different weights to the fluctuation range of the material ratio in different time windows, which mainly reflects the recent fluctuation situation.

[0047] Specifically, the fluctuation cycle characteristic value refers to the intensity value of the main periodic component in the material ratio fluctuation extracted by Fourier analysis, which is used to identify the periodic fluctuation pattern.

[0048] Among them, the volatility risk coefficient specifically refers to the quantitative score of the risk of an abnormal point in the material ratio in the near future after comprehensively considering the current trend of material ratio change, historical fluctuation patterns and equipment status. The value ranges from 0 to 1, and the larger the value, the higher the risk of an abnormal point in the material ratio.

[0049] The specific implementation methods of the above steps are described in detail below.

[0050] The specific implementation of step S01 involves establishing a material ratio fluctuation curve model by constructing a data acquisition and visualization system. First, high-precision electromagnetic flowmeters or mass flowmeters are deployed as sensors, installed on the conveying pipelines for the three components: curing agent, admixture, and water. The sampling frequency is set to 10Hz to ensure data acquisition accuracy. The acquired data is filtered by a signal conditioning circuit, using a Butterworth low-pass filter algorithm to remove high-frequency noise, with the cutoff frequency set to 1 / 5 of the sampling frequency. Then, the processed component input data is aligned by timestamps, and the percentage of each component input at each time point is calculated to plot the material ratio time curve. This system employs a sliding time window technique, updating and displaying the material ratio changes over the past 30 minutes in real time, with a window sliding step of 1 second. The curve uses a spline interpolation algorithm to achieve a smooth transition, reducing the impact of data jitter on the curve display. The purpose of this step is to establish a real-time visualization monitoring system, providing fundamental data support for material ratio fluctuation analysis.

[0051] The specific implementation of step S02 is based on determining the stable range of the material ratio using statistical analysis methods. First, historical production data is collected, including at least 30 days of continuous production records. The component input data from qualified batches are selected as the basic dataset. The probability density distribution of each component's input ratio is calculated using kernel density estimation, and the main peak position is identified as the standard ratio baseline value. Then, the standard deviation of each component in qualified products is calculated, and the range of ±2 times the standard ratio baseline value is defined as the stable range of the material ratio. This range is typically approximately ±0.5% of the standard ratio. For curing agents, the upper limit warning value is the standard ratio baseline value plus 0.5%, and the lower limit warning value is the standard ratio baseline value minus 0.5%. For admixtures, the upper limit warning value is the standard ratio baseline value plus 0.4%, and the lower limit warning value is the standard ratio baseline value minus 0.4%. For water input, the upper limit warning value is the standard ratio baseline value plus 0.6%, and the lower limit warning value is the standard ratio baseline value minus 0.6%. The purpose of this step is to establish a scientifically reasonable material ratio fluctuation warning threshold, providing a basis for subsequent anomaly detection.

[0052] The specific implementation of step S03 involves realizing the detection and recording function of material ratio anomalies. The system uses a real-time monitoring algorithm to continuously compare the current feeding ratio of each component with the preset stable feeding ratio range. When the feeding ratio of any component exceeds its corresponding high-limit warning value or low-limit warning value for more than 3 seconds, the anomaly point marking logic is triggered. The anomaly point marking uses a mutation point detection algorithm combined with a cumulative sum control chart method to improve the ability to identify true anomalies and reduce the false alarm rate. When a material ratio anomaly is detected, the system automatically creates an anomaly event record, including the following parameters: anomaly occurrence timestamp, anomaly duration, anomaly component identifier, feeding ratio value before the anomaly, feeding ratio value at the time of the anomaly, anomaly magnitude value (percentage deviation between actual value and benchmark value), and relevant equipment operating status parameters. The anomaly record is stored in a time-series database for easy subsequent analysis and querying. The purpose of this step is to achieve real-time detection and detailed recording of material ratio anomalies, providing a data foundation for subsequent analysis and processing.

[0053] The specific implementation of step S04 involves applying a mathematical model to analyze the trend of material ratio changes. First, the material ratio trend equation is defined in the dynamic prediction equation set. This equation calculates the rate of change of the component feeding ratio based on the sliding window linear regression method. The input parameters include: the current component feeding ratio value P. t Historical values ​​of component feeding ratios P ​​at the previous N time points t-1 P t-2 , ..., P t-N (N is usually 60, corresponding to 60 seconds of data), time window length W (usually 30 seconds), component type identifier C i (i=1 represents the curing agent, i=2 represents the admixture, and i=3 represents water), sampling frequency f (usually 10Hz). For the data within each time window, a first-order function P(t) = αt + β is fitted using the least squares method, where the slope α is the rate of change of the component feeding ratio. Then, the acceleration of the component feeding ratio change is obtained by differentiating the rate of change of adjacent time windows. Based on the calculated rate of change and acceleration, a simple linear extrapolation method is used to predict the direction of the feeding ratio change in the near future (10-20 seconds). At the same time, the system calculates a fluctuation risk coefficient, which comprehensively considers the current rate of change, acceleration, historical abnormal frequency, and the closeness of the current material ratio to the warning threshold, with a value ranging from 0 to 1. The purpose of this step is to quantitatively analyze the trend of material ratio change through a mathematical model, providing a basis for predictive adjustments.

[0054] The specific implementation of step S05 involves using time-series data mining technology to analyze the correlation of material ratio anomalies. First, historical material ratio anomaly data undergoes time-series clustering preprocessing. A dynamic time warping algorithm is used to calculate the similarity between anomaly sequence sequences, with a similarity threshold set to 0.85. Then, hierarchical clustering is used to classify similar anomaly sequences, resulting in an anomaly pattern library. For each anomaly pattern, the system extracts time features, including: the distribution of time intervals between anomalies (minimum, maximum, average, standard deviation), the distribution of anomaly occurrence periods, and the relationship between anomalies and shift changes. Simultaneously, amplitude similarity is analyzed, including: anomaly amplitude distribution characteristics, anomaly duration relationship, and anomaly recovery pattern characteristics. Finally, anomaly triggering conditions are identified, and an association rule mining algorithm is applied to analyze the correlation between anomalies and production parameters. The association rule support threshold is set to 0.2, and the confidence threshold is set to 0.7. High-confidence rules are extracted as anomaly triggering conditions. The purpose of this step is to discover hidden patterns among material ratio anomalies through data mining, providing knowledge support for optimizing the early warning mechanism.

[0055] The specific implementation of step S06 involves establishing a correlation model between material ratio anomalies and equipment operating status. First, equipment operating log data is collected, including the start-up and shutdown times of key equipment, operating condition switching times, and material conveying system status change times. Then, the equipment operating data is time-aligned with the material ratio anomaly time series. A sliding time window cross-correlation analysis method is used to calculate the time delay distribution between equipment status changes and material ratio anomalies, with the time window size set to 10 minutes. For highly correlated equipment status change events, their time delay statistical characteristics with material ratio anomalies are calculated, including minimum delay, maximum delay, average delay, and standard deviation. Based on the statistical analysis results, a Bayesian network model is constructed to describe the causal relationship network between equipment status changes and material ratio anomalies. The model nodes include key equipment operating status variables, disc start-up and shutdown status variables, material conveying delay time variables, and material ratio anomaly indicator variables. Historical data is used to train the Bayesian network parameters, and the network learning uses the maximum likelihood estimation method with a learning rate set to 0.05. The purpose of this step is to establish a mathematical correlation model between equipment operating status and material ratio fluctuations, providing equipment-level explanation capabilities for the early warning mechanism.

[0056] The specific implementation of step S07 involves using a material flow model to calculate the area affected by material ratio fluctuations. First, the material flow rate is determined based on process parameters, including the flow velocity within the pipeline (typically 0.8-1.2 m / s), agitator speed, and operating parameters of the material conveying equipment. Then, combining the occurrence time and duration of the material ratio anomaly, the affected material volume V = Q × T is calculated, where Q is the material flow rate and T is the duration of the anomaly. Next, a fluid dynamics network model is used to simulate the material flow trajectory within the system. This model considers the network topology, pressure distribution at each node, and flow distribution patterns, and the Hardy-Cross iterative method is used for model solving. Based on the simulation results, the spatial distribution range of the abnormal material is determined, including the starting and ending coordinates and the outline of the covered area. Finally, a three-dimensional visualization model is generated to visually display the affected area; lighter colors indicate the degree of influence, with darker colors indicating a higher deviation from the standard ratio. The purpose of this step is to accurately locate the spatial range of materials affected by material ratio fluctuations, providing guidance for targeted treatment.

[0057] The specific implementation of step S08 is to implement a material ratio fluctuation compensation control strategy. When the system detects an abnormal point in the material ratio, it triggers the compensation control logic, which is based on fuzzy control theory to achieve multi-variable coordinated control. First, the current feeding ratio deviation of each component is calculated. The deviation is used as the input variable of the fuzzy controller, and the deviation is quantified into a fuzzy set ("significantly low", "slightly low", "normal", "slightly high", "significantly high") using a triangular membership function. Then, the adjustment direction and magnitude of each component are determined according to a preset fuzzy rule library, which contains about 25 IF-THEN rules, such as "IF curing agent ratio significantly high AND water ratio normal THEN curing agent input significantly reduced AND water input slightly increased". Fuzzy inference uses the Mamdani algorithm, and defuzzification uses the centroid method. Based on the defuzzification results, the system automatically adjusts the feeding equipment of each component, including speed adjustment and valve opening adjustment. At the same time, the material guiding mechanism is activated to guide the abnormal material into a preset temporary storage area, the design capacity of which is usually 5% of the normal production volume. The purpose of this step is to achieve real-time compensation and adjustment of material ratio fluctuations through intelligent control technology, reducing the impact of fluctuations on product quality.

[0058] The specific implementation of step S09 involves controlling the material ratio recovery and adjustment process. After the material ratio fluctuation compensation procedure is completed, the system enters the recovery and adjustment phase. This phase employs a piecewise linear control strategy to achieve a smooth recovery of the feeding ratio of each component. First, the difference between the current feeding ratio of each component and the standard mix ratio is calculated, and a recovery path is designed, typically divided into three phases: "rapid approach zone," "deceleration transition zone," and "fine adjustment zone." In the rapid approach zone, the adjustment rate is set to 80% of the maximum allowable adjustment rate, with the goal of quickly bringing the feeding ratio of each component close to within ±1% of the standard mix ratio. In the deceleration transition zone, the adjustment rate is linearly reduced to 30% of the maximum allowable adjustment rate to avoid overshoot. In the fine adjustment zone, small-step precise adjustments are used until the feeding ratio of each component reaches within ±0.2% of the standard mix ratio. Throughout the recovery process, the system continuously monitors key mixture performance indicators, including fluidity and strength development, to ensure that the mixture performance meets design requirements. When the feeding ratio of each component stabilizes within ±0.2% of the standard mix ratio within 10 minutes of continuous monitoring, and the mixture performance indicators are normal, the system confirms that the recovery and adjustment is complete. The purpose of this step is to achieve a smooth recovery of the production process and maintain production continuity while ensuring product quality.

[0059] The mathematical model or calculation process involved in this invention will be described in detail below.

[0060] The calculation process for establishing the material ratio fluctuation curve model in step S01 is specifically represented as follows:

[0061]

[0062] In the formula, R i (t) represents the proportion of component i fed at time point t; Q i (t) represents the amount of component i added at time point t; i is the component type index (i=1 represents curing agent, i=2 represents admixture, i=3 represents water); t is the time variable, in seconds.

[0063] The transfer function of the Butterworth low-pass filter algorithm is expressed as follows:

[0064]

[0065] In the formula, H(s) is the filter transfer function; ω c ω is the cutoff angular frequency, which is 1 / 5 of the sampling frequency; n is the filter order, which is usually 2; s is a complex frequency domain variable.

[0066] The formula for calculating the filtered component input data is as follows:

[0067]

[0068] In the formula, The input amount of the filtered component i at time t; h(k) is the filter impulse response coefficient; N is the filter length, usually taken as 20; Q i (tk) represents the initial input amount of component i at time point tk.

[0069] The calculation process for the material ratio stability range in step S02 is specifically shown below:

[0070] The kernel density estimation method calculates the probability density distribution of the component feed ratio using the following formula:

[0071]

[0072] In the formula, f i (r) is the probability density function of the proportion r of component i; n is the number of historical data samples; h is the bandwidth parameter, usually taken as 0.05; K is the kernel function, using a Gaussian kernel function. r ij This represents the proportion of component i in the j-th sample from the historical data.

[0073] Formula for calculating the standard deviation of component feeding ratio:

[0074]

[0075] In the formula, σ i μ is the standard deviation of the feed ratio of component i; i is the mean of the proportions of component i, i.e., the standard proportion baseline; n is the sample size; r ij This represents the proportion of component i in the j-th sample from the historical data.

[0076] Formulas for calculating the upper and lower limits of the material ratio stability range:

[0077] UL i =μ i +2σ i ;

[0078] LL i =μ i -2σ i ;

[0079] In the formula, UL i The upper limit warning value for the feeding ratio of component i; LL i μ is the lower limit warning value for the proportion of component i in the feed; i σ is the standard proportion benchmark value for the feeding ratio of component i; i Let be the standard deviation of the proportion of component i in the feed.

[0080] The calculation process for detecting abnormal material ratios in step S03 is specifically shown below:

[0081] The formula for calculating the statistics in the cumulative sum control chart method is as follows:

[0082] S i (t)=max(0,S i (t-1)+(R i (t)-μ i )-k);

[0083] In the formula, S i (t) represents the cumulative sum statistic of component i at time point t; R i (t) represents the actual feed ratio of component i at time point t; μ i The standard proportion of component i is the reference value; k is the allowable deviation parameter, which is usually taken as σ. i / 2;S i (0) = 0 is the initial value.

[0084] Anomaly detection criteria:

[0085] When S i (t)>h or R i (t)>UL i Or R i (t) <LL i If the duration exceeds 3 seconds, it is marked as an abnormal point in the material ratio;

[0086] In the formula, h is the threshold for the cumulative sum statistic, which is usually taken as 4σ. i UL i The upper limit warning value for component i; LL i This is the lower limit warning value for component i.

[0087] Formula for calculating abnormal amplitude values:

[0088]

[0089] In the formula, D i (t) represents the anomalous amplitude value of component i at time point t; R i (t) represents the actual feed ratio of component i at time point t; μ i This is the standard proportion benchmark value for component i.

[0090] The specific representation of the dynamic prediction equations for material ratio in step S04 is as follows:

[0091] Material ratio trend equation:

[0092]

[0093] In the formula, V i (t) represents the rate of change of the feed ratio of component i at time point t; P t-jThe feeding ratio of component i at time point tj; W represents the average proportion of component i fed within the time window; W is the length of the time window, which is usually 30 seconds.

[0094] Formula for calculating the acceleration of changes in component feeding ratio:

[0095]

[0096] In the formula, A i (t) represents the acceleration of the change in the feed ratio of component i at time point t; V i (t) represents the rate of change of the feed ratio of component i at time point t; V i (t-Δt) represents the rate of change of the feeding ratio of component i at time point t-Δt; Δt is the time step, which is usually taken as 1 second.

[0097] Formula for predicting future feed ratio:

[0098]

[0099] In the formula, P i (t+τ) represents the predicted feed ratio of component i at a future time point t+τ; P i (t) represents the feed ratio of component i at the current time point t; V i (t) represents the rate of change of the current feed ratio; A i (t) represents the acceleration of the current feed ratio change; τ is the prediction time length, which is usually taken as 10-20 seconds.

[0100] Fluctuation amplitude equation:

[0101]

[0102] In the formula, M i (t) is the weighted average of the fluctuation amplitude of component i at time t; P i (tj·Δt) represents the proportion of component i fed at time point tj·Δt; μ i w is the standard proportion reference value for component i; j Let be the weighting coefficient, satisfying And w1>w2>...>w m Exponentially decaying weights are typically used. Where λ is the decay rate parameter, usually taken as 0.2; m is the number of historical time points considered, usually taken as 10; Δt is the time step, usually taken as 1 second.

[0103] Formula for calculating the characteristic value of fluctuation cycle:

[0104]

[0105] In the formula, F i (f) is the frequency spectrum of the fluctuation in the feed ratio of component i; P i (t) represents the proportion of component i fed at time point t; T represents the total sampling time; f represents the frequency variable; and j represents the imaginary unit.

[0106] The cyclic characteristic value is defined as the amplitude corresponding to the main peak frequency:

[0107] C i =max f F i (f);

[0108] In the formula, C i F represents the oscillation periodic characteristic value of component i; i (f) is the frequency spectrum of the fluctuation of the feeding ratio of component i.

[0109] Prediction correction equation:

[0110]

[0111] In the formula, This is the predicted value after correction for equipment start-up and shutdown factors; P i (t+τ) is the initial predicted value; E k The start / stop status parameters (E) of device k k =1 indicates startup, E k =-1 indicates stop, E k =0 indicates stable operation); δ k G represents the typical fluctuation range caused by the start-up and shutdown of device k; ik The influence coefficient of equipment k's start-up and shutdown on component i; t k t is the time of the most recent state change of device k; α is the current time; α is the time decay coefficient, which is usually taken as 0.1; K is the number of critical devices to be considered.

[0112] Formula for calculating volatility risk coefficient:

[0113]

[0114] In the formula, R risk (t) represents the volatility risk coefficient; P i (t) represents the feed ratio of the current component i; μ i The standard proportion reference value for component i; UL i V is the upper limit warning value for component i; i (t) represents the rate of change of the current feed ratio; V max The highest rate of change in history; A i (t) represents the acceleration of the current feed ratio change; A maxThis represents the greatest acceleration of change in history; H f The recent anomaly frequency factor has a value range of 0-1; w1, w2, w3, and w4 are weighting coefficients that satisfy... The values ​​are usually taken as w1 = 0.4, w2 = 0.3, w3 = 0.2, and w4 = 0.1.

[0115] The calculation process of temporal clustering in step S05 is specifically represented as follows:

[0116] Dynamic time warping algorithm for calculating the similarity of outlier sequences:

[0117]

[0118] In the formula, DTW(X, Y) is the dynamic time warp distance between sequences X and Y; X 1：i Y represents the first i elements of sequence X; 1：j d(x) represents the first j elements of sequence Y; i y j () is a single-point distance function, usually using Euclidean distance. d is the feature dimension, which is usually taken as 3 in this method (corresponding to the three components).

[0119] Similarity calculation formula:

[0120]

[0121] In the formula, Sim(X,Y) is the similarity between sequences X and Y, with a value range of 0-1; DTW(X,Y) is the dynamic time warping distance.

[0122] The formula for calculating support in association rule mining:

[0123]

[0124] In the formula, For rules The support level; count(X∪Y) is the number of records that contain both X and Y; N is the total number of records.

[0125] The formula for calculating confidence in association rule mining is as follows:

[0126]

[0127] In the formula, For rules The confidence level; count(X∪Y) is the number of records that contain both X and Y; count(X) is the number of records that contain X.

[0128] The calculation process for establishing the model of the relationship between material ratio anomalies and equipment start-up / shutdown in step S06 is specifically represented as follows:

[0129] Sliding time window cross-correlation analysis method:

[0130]

[0131] In the formula, R xy (τ) is the cross-correlation coefficient between sequences x and y with a delay of τ; x t The value of the sequence of device state changes at time t; y t The value of the material ratio anomaly indicator sequence at time t; and τ represents the mean of sequences x and y, respectively; T represents the total time length; and τ represents the time delay variable.

[0132] Formula for calculating conditional probability in a Bayesian network model:

[0133]

[0134] In the formula, P(A) i |Pa(A i )) is node A i In its parent node Pa(A) i The conditional probability given a certain value; N ijk For node A in historical data i The value is k and its parent node Pa(A) i The number of samples for a given combination of values ​​j.

[0135] Maximum likelihood estimation for Bayesian network parameter learning:

[0136]

[0137] In the formula, L is the likelihood function; θ ijk For node A i When its parent node has a value of j, A i The conditional probability parameter with a value of k; N ijk q represents the sample size for the corresponding case; n represents the number of network nodes; q i For node A i The number of possible combinations of values ​​for the parent node; r i For node A i The number of possible values.

[0138] The calculation process for the influence area of ​​material ratio fluctuation in step S07 is specifically shown below:

[0139] Formula for calculating the volume of affected materials:

[0140]

[0141] In the formula, V is the volume of the affected material; Q is the total material flow rate; Q i Let be the flow rate of component i; T be the duration of the anomaly.

[0142] Hardy-Cross iterative method for solving fluid dynamics pipe network models:

[0143]

[0144] In the formula, ΔQ j Here is the flow correction value for loop j; r i Q is the resistance coefficient of pipe segment i; i Let be the flow rate of pipe segment i; m is the exponent of the relationship between flow rate and pressure loss, typically taken as 1.85-2; n j Let j be the number of pipe segments contained in loop j.

[0145] Iterative update flow formula:

[0146]

[0147] In the formula, Let ΔQ be the flow rate of pipe segment i in the kth iteration; j δ is the flow correction value for loop j; ij It is a direction factor, which is 1 when pipe segment i is clockwise in loop j and -1 when it is counterclockwise.

[0148] The calculation process for material ratio fluctuation compensation control in step S08 is specifically represented as follows:

[0149] Formula for calculating the deviation of the current feed ratio of each component in fuzzy control:

[0150] e i (t)=R i (t)-μ i ;

[0151] In the formula, e i (t) represents the deviation in the feed ratio of component i at time point t; R i (t) represents the actual feed ratio of component i at time point t; μ i This is the standard proportion benchmark value for component i.

[0152] Definition of fuzzy membership function:

[0153]

[0154]

[0155] In the formula, μ NB (e), μNS (e), μ ZO (e), μ PS (e), μ PB (e) represents the membership functions for “significantly low”, “slightly low”, “normal”, “slightly high”, and “significantly high”, respectively; e represents the feed ratio deviation; and σ represents the standard deviation of the component feed ratio.

[0156] Mamdani fuzzy inference calculation formula:

[0157] μ C (z)=max x，y [min(μ A (x), μ B (y), μ R (x, y, z))];

[0158] In the formula, μ C (z) is the membership function of the reasoning result; μ A (x) and μ B (y) is the membership function of the input variable; μ R (x, y, z) is the membership function of the fuzzy rule.

[0159] Formula for calculating the centroid of the defuzzified model:

[0160]

[0161] In the formula, z * The definite value after defuzzification; z j μ represents the discrete points of the output universe of discourse. C (z j ) represents the membership degree value of the corresponding point; m represents the number of discrete points.

[0162] The calculation process for material ratio recovery adjustment in step S09 is specifically represented as follows:

[0163] The formula for calculating the adjustment rate of a piecewise linear control strategy is as follows:

[0164]

[0165] In the formula, v i (t) represents the adjustment rate of component i at time point t; v max e is the maximum allowable adjustment rate; i (t) represents the deviation in the feed ratio of component i; e H The threshold for the fast approach region is typically set to 1%; e M This is the threshold for the deceleration transition zone, typically set to 0.5%; e L The threshold for the fine-tuning area is typically set to 0.2%.

[0166] Formula for calculating the adjustment amount of component feeding ratio:

[0167] ΔR i (t)=v i (t)·Δt·sign(e i (t));

[0168] In the formula, ΔR i (t) represents the adjustment amount of the feed ratio of component i at time point t; v i (t) represents the adjustment rate; Δt represents the control period, typically taken as 1 second; sign(e i (t) is the deviation sign function, when e i When (t)>0, the value is -1, and when e i When (t) < 0, the value is 1.

[0169] The construction principles and significance of the above equations are as follows: The material ratio trend equation uses the least squares method for linear regression, which can effectively estimate the rate of change of the component feeding ratio, consider all data points within the time window, and avoid the influence of single-point fluctuations; The fluctuation amplitude equation uses the weighted average principle, giving higher weight to recent data and more accurately reflecting the current fluctuation state; The prediction correction equation combines physical models and statistical learning methods, and simulates the decay characteristics of equipment start-up and shutdown over time through an exponential decay function, thereby improving prediction accuracy; The dynamic time warping algorithm can handle the problems of inconsistent time series lengths and time axis distortion, and improve the accuracy of abnormal pattern recognition; The Bayesian network model uses probabilistic graph theory to construct causal relationships between variables, which can effectively handle uncertainty; The Hardy-Cross iterative method is a classic method for solving nonlinear pipeline network equations and is suitable for complex pipeline network systems; The fuzzy control strategy can handle system uncertainty and nonlinear characteristics, realize complex control logic through language rules, and improve the robustness of the control system; The piecewise linear control strategy achieves smooth transition, avoids oscillations during the adjustment process, and ensures the stability of the production process.

[0170] Specifically, the principle of this invention is as follows: The technical solution of this invention is based on the organic combination of system monitoring, mathematical modeling, predictive analysis and intelligent control. It achieves comprehensive monitoring and prediction of material ratio fluctuations through the material ratio fluctuation curve model and the material ratio dynamic prediction equation set, and achieves precise control through the material ratio fluctuation compensation program.

[0171] First, this invention establishes a material ratio fluctuation curve model. By using sensors to collect real-time data on the input amounts of the three components—curing agent, additives, and water—the system plots a material ratio change curve over time, enabling full monitoring of the mixing process. Simultaneously, based on historical production data, a material ratio stability range is set, generating high-limit and low-limit warning values ​​to provide a benchmark for anomaly detection. When real-time monitored values ​​exceed the preset material ratio stability range, the system automatically marks it as an anomaly point and records relevant parameters, providing a data foundation for subsequent analysis.

[0172] Secondly, this invention constructs a set of dynamic prediction equations for the material ratio, including a material ratio trend equation, a fluctuation amplitude equation, and a prediction correction equation. The material ratio trend equation predicts the direction of change in the near future by calculating the rate and acceleration of change in the component feeding ratio; the fluctuation amplitude equation assesses the fluctuation amplitude and frequency characteristics of the material ratio, and outputs the weighted average of the fluctuation amplitude and the characteristic value of the fluctuation period; the prediction correction equation corrects the prediction results based on the equipment start-up and shutdown status, and outputs the predicted trend value and fluctuation risk coefficient after correction for equipment start-up and shutdown factors. These three equations constitute a scientific and complete prediction system that conforms to the principles of material feeding dynamics.

[0173] Furthermore, this invention analyzes the correlation between material ratio anomalies and constructs a model of the relationship between material ratio anomalies and equipment start-up and shutdown, revealing the mechanism by which equipment start-up, shutdown, and switching of operating conditions cause material ratio fluctuations, providing a theoretical basis for preventing material ratio fluctuations. Simultaneously, by calculating the area affected by material ratio fluctuations, the location and range of materials affected by the fluctuations are accurately located, providing spatial positioning for subsequent processing.

[0174] Finally, the present invention executes a material ratio fluctuation compensation procedure. When an abnormal point in the material ratio is detected, the system automatically adjusts the input amount of each component to offset the impact of the fluctuation, and the affected materials are introduced into a preset area for secondary processing. After the material ratio fluctuation compensation, the system gradually restores the input amount of each component to the standard value, ensuring production continuity and that the performance of the mixture meets the standards.

[0175] Based on the above principles, this invention achieves real-time prediction and precise control of material ratio fluctuations during the mixing process of fluidized solidified soil, which is in line with the development trend of modern industrial automation and intelligence, and reflects the deep integration of mathematical modeling and process control.

[0176] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0177] The specific implementation of step S01 involves establishing a material ratio fluctuation curve model by constructing a data acquisition and visualization system. First, high-precision electromagnetic flowmeters or mass flowmeters are deployed as sensors. These sensors are installed on the conveying pipelines for the three components: curing agent, admixture, and water. The sampling frequency is set to 10Hz to ensure data acquisition accuracy. The acquired data is filtered by a signal conditioning circuit, and a Butterworth low-pass filter algorithm is used to remove high-frequency noise. The filter transfer function is expressed as follows: Where H(s) is the filter transfer function, ω c ω is the cutoff angular frequency, which is taken as 1 / 5 of the sampling frequency; n is the filter order, usually taken as 2; and s is a complex frequency domain variable. The formula for calculating the filtered component input data is as follows: in Let h(k) be the input amount of the filtered component i at time t, h(k) be the filter impulse response coefficient, N be the filter length (usually 20), and Q be the input amount of the filtered component i at time t. i (tk) represents the initial input amount of component i at time point tk. Then, the processed component input data is aligned by timestamps, and the percentage of each of the three components' input amounts to the total input amount is calculated. A material ratio-time curve is then plotted using the following formula: Where R i (t) represents the proportion of component i fed at time point t, Q i (t) represents the input amount of component i at time point t, where i is the component type index (i=1 for curing agent, i=2 for admixture, i=3 for water), and t is the time variable in seconds. This system employs a sliding time window technique to update and display the material ratio changes over the past 30 minutes in real time, with a window sliding step of 1 second. The curve uses spline interpolation to achieve a smooth transition, reducing the impact of data jitter on the curve display. The purpose of this step is to establish a real-time visual monitoring system to provide basic data support for material ratio fluctuation analysis.

[0178] The specific implementation of step S02 is based on determining the stable range of the material ratio using statistical analysis methods. First, historical production data is collected, including at least 30 days of continuous production records. The component input data from batches that meet quality standards are selected as the basic dataset. The probability density distribution of the input ratio of each component is calculated using the kernel density estimation method, with the following formula: Where f i (r) is the probability density function of the proportion r of component i, n is the number of historical data samples, h is the bandwidth parameter, usually taken as 0.05, and K is the kernel function, using a Gaussian kernel function. r ijLet i be the proportion of component i in the j-th sample from historical data. Identify the main peak position as the standard ratio baseline. Then calculate the standard deviation of each component in qualified products using the following formula: Where σ i μ is the standard deviation of the proportion of component i in the feed. i The average proportion of component i in the feed ratio, i.e., the standard mix ratio baseline value, where n is the sample size and r is the mean. ij Let be the feed ratio of component i in the j-th sample of historical data. The range of ±2 standard deviations from the standard ratio baseline is defined as the stable feed ratio interval, calculated using the formula UL. i =μ i +2σ i and LL i =μ i -2σ i UL i LL is the upper limit warning value for the feeding ratio of component i. i μ is the lower limit warning value for the proportion of component i in the feed. i σ is the standard proportion benchmark value for component i. i This represents the standard deviation of the proportion of component i. This range is typically ±0.5% of the standard mix ratio. For curing agents, the upper limit warning value is 0.5% above the standard mix ratio baseline, and the lower limit warning value is 0.5% below the standard mix ratio baseline; for admixtures, the upper limit warning value is 0.4% above the standard mix ratio baseline, and the lower limit warning value is 0.4% below the standard mix ratio baseline; for water dosage, the upper limit warning value is 0.6% above the standard mix ratio baseline, and the lower limit warning value is 0.6% below the standard mix ratio baseline. The purpose of this step is to establish a scientifically reasonable warning threshold for component ratio fluctuations, providing a basis for subsequent anomaly detection.

[0179] The specific implementation of step S03 involves realizing the detection and recording function of abnormal material ratios. The system uses a real-time monitoring algorithm to continuously compare the current feeding ratio of each component with the preset stable material ratio range. When the feeding ratio of any component exceeds its corresponding high-limit warning value or low-limit warning value for more than 3 seconds, the abnormal point marking logic is triggered. The abnormal point marking uses a mutation point detection algorithm combined with a cumulative sum control chart method to improve the ability to identify true abnormalities and reduce the false alarm rate. The cumulative sum statistic calculation formula is S. i (t)=max(0,S i (t-1)+(R i (t)-μ i )-k), where S i R(t) is the cumulative sum statistic of component i at time point t. i (t) represents the actual feed ratio of component i at time point t, μ i Let k be the standard proportion reference value for component i, and k be the allowable deviation parameter, typically taken as σ.i / 2,S i (0) = 0 is the initial value. The outlier determination condition is when S i (t)>h or R i (t)>UL i Or R i (t) <LL i When the duration exceeds 3 seconds, it is marked as an outlier in the material ratio, where h is the threshold of the cumulative sum statistic, which is usually taken as 4σ. i UL i LL is the upper limit warning value for component i. i This is the lower limit warning value for component i. When an abnormal material ratio is detected, the system automatically creates an abnormal event record, including parameters such as the timestamp of the abnormality, the duration of the abnormality, the identifier of the abnormal component, the feed ratio before the abnormality, the feed ratio at the time of the abnormality, and the abnormality amplitude value. The formula for calculating the abnormality amplitude value is as follows: Where D i (t) represents the anomalous amplitude value of component i at time point t, R i (t) represents the actual feed ratio of component i at time point t, μ i This represents the standard proportion baseline value for component i. Anomaly records are stored in a time-series database for easy retrieval and analysis later. The purpose of this step is to achieve real-time detection and detailed recording of anomalies in the material ratio, providing a data foundation for subsequent analysis and processing.

[0180] The specific implementation of step S04 involves applying a mathematical model to analyze the trend of material ratio changes. First, the material ratio trend equation is defined in the dynamic prediction equation set. This equation calculates the rate of change of the component feeding ratio based on the sliding window linear regression method. The calculation formula is as follows: Where V i (i) represents the rate of change of the feed ratio of component i at time point t, P t-j The feeding ratio of component i at time point tj. W represents the average proportion of component i fed within the time window, where W is the length of the time window, typically 30 seconds. Input parameters include the current component feeding proportion P. t Historical values ​​of component feeding ratios P ​​at the previous N time points t-1 P t-2 , ..., P t-N (N is usually 60, corresponding to 60 seconds of data), time window length W (usually 30 seconds), component type identifier C i (i=1 represents the curing agent, i=2 represents the admixture, i=3 represents water), sampling frequency f (usually 10Hz). Then, the acceleration of the component feeding ratio change is obtained by performing differential calculations on the change rates of adjacent time windows. The calculation formula is: Where A i(t) represents the acceleration of the change in the feed ratio of component i at time point t, V i (t) represents the rate of change of the feed ratio of component i at time point t, V i (t-Δt) represents the rate of change of the feed ratio of component i at time point t-Δt, where Δt is the time step, typically taken as 1 second. Based on the calculated rate of change and acceleration, a simple linear extrapolation method is used to predict the direction of feed ratio change in the short term (10-20 seconds). The prediction formula is as follows: Where P i (t+τ) represents the predicted feed ratio of component i at the future time point t+τ, P i (t) represents the feed ratio of component i at the current time point t, V i (t) represents the rate of change of the current feed ratio, A i (t) represents the acceleration due to the change in the current feed ratio, and τ is the prediction time length, typically taken as 10-20 seconds. The fluctuation amplitude equation is used to assess the fluctuation amplitude, and the calculation formula is as follows: Where M i (t) is the weighted average of the fluctuation amplitude of component i at time t, P i (tj·Δt) represents the feed ratio of component i at time point tj·Δt, μ i w is the standard proportion reference value for component i. j Here, the weighting coefficients are typically exponentially decaying. The formula for calculating the variability period characteristic value is: Where F i (f) is the frequency spectrum of the fluctuation in the feed ratio of component i, P i (t) represents the proportion of component i fed at time point t, T is the total sampling time, f is the frequency variable, and j is the imaginary unit. The characteristic value of the fluctuation period is defined as C. i =max f F i (f), where C i F is the variability characteristic value of component i. i (f) represents the frequency spectrum of the fluctuation in the feed ratio of component i. The prediction correction equation is used to correct the prediction results based on the equipment start-up and shutdown status, and the calculation formula is as follows: in P is the predicted value after correction for equipment start-up and shutdown factors. i (t+τ) is the initial predicted value, E k Let δ be the start / stop status parameter of device k. k G represents the typical fluctuation range caused by the start-up and shutdown of device k. ikLet be the influence coefficient of device k's start-up and shutdown on component i, t be the time of the most recent state change of device k, t be the current time, α be the time decay coefficient, typically taken as 0.1, and K be the number of critical devices considered. Simultaneously, the system calculates the fluctuation risk coefficient, using the following formula: Where R risk (t) is the volatility risk coefficient, P i (t) represents the feed ratio of the current component i, μ i For the standard proportioning reference value of component i, UL i V is the upper limit warning value for component i. i (t) represents the rate of change of the current feed ratio, V max For the largest rate of change in history, A i (t) represents the acceleration of the current feed ratio change, A max H represents the greatest acceleration of change in history. f The recent anomaly frequency factor has a value range of 0-1, and w1, w2, w3, and w4 are weighting coefficients that satisfy the following conditions: Typically, the values ​​are w1 = 0.4, w2 = 0.3, w3 = 0.2, and w4 = 0.1. The purpose of this step is to quantitatively analyze the trend of material ratio changes through mathematical models, providing a basis for predictive adjustments.

[0181] The specific implementation of step S05 involves using time-series data mining techniques to analyze the correlation of material ratio anomalies. First, time-series clustering preprocessing is performed on the historical material ratio anomaly data. Then, the dynamic time warping algorithm is used to calculate the similarity between anomaly sequences. The calculation formula is as follows:

[0182] Where DTW(X, Y) is the dynamic time-warped distance between sequences X and Y, X 1：i Y represents the first i elements of sequence X. 1：j Let d(x) represent the first j elements of sequence Y. i y j The similarity function is a single-point distance function, typically using Euclidean distance. The formula for calculating similarity is... Where Sim(X, Y) represents the similarity between sequences X and Y, ranging from 0 to 1, and DTW(X, Y) represents the dynamic time warping distance. The similarity threshold is set to 0.85. Then, hierarchical clustering is used to classify similar anomaly sequences, resulting in an anomaly pattern library. For each anomaly pattern, the system extracts time features, including the distribution of time intervals between anomalies, the distribution of anomaly occurrence periods, and the relationship between anomalies and shift changes. Simultaneously, amplitude similarity is analyzed, including anomaly amplitude distribution characteristics, anomaly duration relationships, and anomaly recovery pattern characteristics. Finally, anomaly triggering conditions are identified, and an association rule mining algorithm is applied to analyze the correlation between anomalies and production parameters. The formula for calculating the support of association rules is... in For rules The support is given by count(X∪Y), where count(X∪Y) is the number of records that contain both X and Y, and N is the total number of records. The formula for calculating the confidence of an association rule is: in For rules The confidence level is calculated as follows: count(X∪Y) represents the number of records containing both X and Y, and count(X) represents the number of records containing X. The support threshold for association rules is set to 0.2, and the confidence threshold is set to 0.7. High-confidence rules are extracted as anomaly triggering conditions. The purpose of this step is to discover implicit patterns between material ratio anomalies through data mining, providing knowledge support for optimizing the early warning mechanism.

[0183] The specific implementation of step S06 involves establishing a correlation model between material ratio anomalies and equipment operating status. First, equipment operation log data is collected, including start-up and shutdown times of key equipment, operating condition switching times, and material conveying system status change times. Then, the equipment operation data is time-aligned with the material ratio anomaly time series. A sliding time window cross-correlation analysis method is used to calculate the time delay distribution between equipment status changes and material ratio anomalies. The cross-correlation analysis calculation formula is as follows: Where R xy (τ) is the cross-correlation coefficient between sequences x and y with a delay of τ, x t The value of y is the sequence of device state changes at time t. t The value of the material ratio anomaly indicator sequence at time t, and Let x and y be the mean values ​​of sequences x and y, respectively, T be the total time length, and τ be the time delay variable. The time window size is set to 10 minutes. For highly correlated equipment state change events, the statistical characteristics of their time delay relative to material ratio anomalies are calculated, including minimum delay, maximum delay, average delay, and standard deviation. Based on the statistical analysis results, a Bayesian network model is constructed to describe the causal relationship between equipment state changes and material ratio anomalies. The model nodes include key equipment operating state variables, disc start / stop state variables, material conveying delay time variables, and material ratio anomaly indicator variables. The conditional probability calculation formula in the Bayesian network model is as follows: Where P(A) i |Pa(A i )) is node A i In its parent node Pa(A) i The conditional probability N given a certain value. ijk For node A in historical data i The value is k and its parent node Pa(A) i The number of samples for a given combination of values ​​j. The maximum likelihood estimate formula for Bayesian network parameter learning is: Where L is the likelihood function, θ ijk For node A i When its parent node has a value of j, A i The conditional probability parameter takes the value k, N ijk Here, n represents the number of samples for the corresponding scenario, and q represents the number of network nodes. i For node A i The number of possible combinations of values ​​for the parent node, r i For node A i The number of possible values ​​is determined. The Bayesian network parameters are trained using historical data, employing maximum likelihood estimation with a learning rate of 0.05. The purpose of this step is to establish a mathematical correlation model between equipment operating status and material ratio fluctuations, providing equipment-level explanation capabilities for the early warning mechanism.

[0184] The specific implementation of step S07 involves applying a material flow model to calculate the area affected by material ratio fluctuations. First, the material flow rate is determined based on process parameters, including the flow velocity within the pipeline (typically 0.8-1.2 m / s), agitator speed, and operating parameters of the material conveying equipment. Then, considering the occurrence time and duration of material ratio anomalies, the volume of affected material is calculated using the following formula: Where V is the volume of the affected material, Q is the total material flow rate, and Q i Let be the flow rate of component i, and T be the duration of the anomaly. Next, a fluid dynamics network model is used to simulate the material flow trajectory within the system. This model considers the network topology, pressure distribution at each node, and flow distribution patterns. The Hardy-Cross iterative method is used to solve the model, and the calculation formula is as follows: Where ΔQ j r is the flow correction value for loop j. i Let Q be the resistance coefficient of pipe segment i. i Let n be the flow rate of pipe segment i, m be the exponent of the relationship between flow rate and pressure loss, which is usually taken as 1.85-2, and n be the flow rate of pipe segment i. j Let j be the number of pipe segments contained in loop j. The iterative flow rate update formula is: in Let ΔQ be the flow rate of pipe segment i in the kth iteration. j δ is the flow correction value for loop j. ij The direction factor is 1 when pipe segment i is clockwise in loop j and -1 when it is counterclockwise. Based on the simulation results, the spatial distribution range of abnormal materials is determined, including the starting position coordinates, ending position coordinates, and the outline of the covered area. Finally, a 3D visualization model is generated to intuitively display the affected area, with color depth indicating the degree of impact, and darker colors indicating a higher degree of deviation from the standard ratio. The purpose of this step is to accurately locate the spatial range of materials affected by material ratio fluctuations, providing guidance for targeted treatment.

[0185] The specific implementation of step S08 is to implement a material ratio fluctuation compensation control strategy. When the system detects an abnormal point in the material ratio, it triggers the compensation control logic, which is based on fuzzy control theory to achieve multivariate coordinated control. First, the deviation of the current feeding ratio of each component is calculated, and the calculation formula is e. i (t)=R i (t)-μ i , where e i (t) represents the deviation of the feed ratio of component i at time point t, R i (t) represents the actual feed ratio of component i at time point t, μ i The standard proportion benchmark value for component i. The deviation is used as the input variable of the fuzzy controller. A triangular membership function is used to quantify the deviation into a fuzzy set. The fuzzy membership function is defined as a piecewise linear function; for example, the membership function for "significantly low" is... Where μ NB (e) is the membership function for "significantly low", where e is the deviation of the feed ratio and σ is the standard deviation of the component feed ratio. Similarly, membership functions for "slightly low", "normal", "slightly high", and "significantly high" are defined. Then, the adjustment direction and magnitude for each component are determined based on a pre-set fuzzy rule base containing approximately 25 IF-THEN rules. Fuzzy inference uses the Mamdani algorithm, with the calculation formula being μ. C (z)=max x，y [min(μ A (x), μ B (y), μ R [(x, t, z))], where μ C (z) is the membership function of the reasoning result, μ A (x) and μ B (y) is the membership function of the input variable, μ R (x, y, z) represents the membership function of the fuzzy rule. Defuzzification uses the centroid method, and the calculation formula is as follows: Where z * To determine the definite value after defuzzification, z j μ represents the discrete points of the output universe of discourse. C (z j ) represents the membership value of the corresponding point, and m represents the number of discrete points. Based on the defuzzification results, the system automatically adjusts the feeding equipment for each component, including speed adjustment and valve opening regulation. Simultaneously, the material guiding mechanism is activated to guide abnormal materials into a preset temporary storage area, the design capacity of which is typically 5% of the normal production volume. The purpose of this step is to achieve real-time compensation and adjustment of material ratio fluctuations through intelligent control technology, reducing the impact of fluctuations on product quality.

[0186] The specific implementation of step S09 involves executing the material ratio recovery and adjustment process control. After the material ratio fluctuation compensation procedure is completed, the system enters the recovery and adjustment phase. This phase employs a piecewise linear control strategy to achieve a smooth recovery of the feeding ratio of each component. First, the difference between the current feeding ratio of each component and the standard ratio value is calculated, and a recovery path plan is designed, typically divided into three phases: "rapid approach zone," "deceleration transition zone," and "fine adjustment zone." The adjustment rate calculation formula for the piecewise linear control strategy is as follows: Where v i (t) represents the adjustment rate of component i at time point t, v max For the maximum allowable adjustment rate, e i (t) represents the deviation in the feed ratio of component i, e H The threshold for the fast approach region is typically set to 1%, e M This is the threshold for the deceleration transition zone, typically set to 0.5%, e L The threshold for fine-tuning the feed ratio is typically set to 0.2%. The formula for calculating the adjustment amount of the component feed ratio is ΔR. i (t)=v i (t)·Δt·sign(e i (t)), where ΔR i (t) represents the adjustment amount of the feed ratio of component i at time point t, v i (t) represents the adjustment rate, Δt represents the control period, which is typically taken as 1 second, and sign(e i (t) is the deviation sign function, when e i When (t)>0, the value is -1, and when e i When (t) < 0, the value is 1. In the rapid approach zone, the adjustment rate is set to 80% of the maximum allowable adjustment rate, aiming to quickly bring the proportion of each component close to the standard mix ratio within ±1%. In the deceleration transition zone, the adjustment rate is linearly reduced to 30% of the maximum allowable adjustment rate to avoid overshoot. In the fine adjustment zone, small-step precise adjustments are used until the proportion of each component reaches the standard mix ratio within ±0.2%. Throughout the recovery process, the system continuously monitors key mixture performance indicators, including flowability and strength development, to ensure that the mixture performance meets design requirements. When the proportion of each component stabilizes within ±0.2% of the standard mix ratio within 10 minutes of continuous monitoring, and the mixture performance indicators are normal, the system confirms that the recovery adjustment is complete. The purpose of this step is to achieve a smooth recovery of the production process and maintain production continuity while ensuring product quality.

[0187] The steps S01 to S09 described above together constitute a complete method for early warning and adjustment of material ratio fluctuations in fluidized bed solidified soil mixing. This method first collects real-time data using high-precision sensors and establishes a material ratio fluctuation curve model (S01). Then, based on historical data, it sets a stable material ratio range and an early warning threshold (S02). When an abnormal point in the material ratio is detected (S03), the system analyzes the trend of material ratio changes and predicts future changes (S04), while simultaneously mining the correlation between abnormal points (S05) and their relationship with equipment start-up and shutdown (S06), and calculating the range of fluctuation impact (S07). Based on the above analysis results, the system executes a material ratio fluctuation compensation program (S08) and implements material ratio recovery adjustment (S09), ultimately achieving stable control of the mixing process. The entire method fully utilizes advanced technologies such as mathematical models, statistical analysis, time-series data mining, fluid dynamics models, and fuzzy control theory, forming a systematic material ratio fluctuation early warning and adjustment mechanism, effectively improving the stability of the fluidized bed solidified soil mixing process and product quality.

[0188] The core advantage of this method lies in the collaborative application of multiple algorithms, such as the Butterworth low-pass filter algorithm to remove data noise, the kernel density estimation method to determine the standard ratio baseline value, the cumulative sum control chart method to detect outliers, the sliding window linear regression to predict changing trends, the dynamic time warping algorithm to identify similar abnormal patterns, the Bayesian network model to analyze the relationship between equipment status and material ratio fluctuations, the Hardy-Cross iterative method to simulate material flow, and a multivariate coordinated control strategy based on fuzzy control theory. These algorithms work together to form a closed-loop material ratio fluctuation early warning and adjustment system, enabling intelligent control throughout the entire process from monitoring, analysis, prediction to adjustment. This method also fully considers practical engineering needs, such as setting reasonable early warning thresholds (curing agent ±0.5%, admixture ±0.4%, water ±0.6%), requiring outlier detection to last for more than 3 seconds to avoid false alarms due to instantaneous fluctuations, setting the prediction time range to 10-20 seconds to ensure sufficient reaction time, comprehensively considering current status and historical data for the fluctuation risk coefficient, including approximately 25 rules in the fuzzy control rule base covering common abnormal situations, and using a piecewise linear control strategy to avoid oscillations during the recovery process. These parameter settings and algorithm selections are all based on the characteristics of fluidized solidified soil mixing process, ensuring the practicality and effectiveness of the method.

[0189] To better understand and implement this invention, a specific application scenario is provided in Example 2: At an underground engineering construction site, researchers used fluidized solidified soil as backfill material, requiring stable mix proportions to ensure project quality. A fluidized solidified soil mixing system was set up on-site, including a supply system for three components: solidifying agent, admixture, and water, as well as an automatic control system. To address the frequent material ratio fluctuations during mixing, researchers implemented the material ratio fluctuation early warning and adjustment method of this invention.

[0190] First, researchers installed high-precision electromagnetic flowmeters on the delivery pipelines for the three components: curing agent, additives, and water, with a sampling frequency set to 10Hz. The collected raw data was processed using a Butterworth low-pass filter to calculate the feed ratio of each component, thus establishing a material ratio fluctuation curve model. Table 1 shows the standard ratio baseline values ​​of the three components and the actual monitoring data during a continuous production process.

[0191] Table 1 Standard proportions and actual monitoring data of the three components of fluidized solidified soil

[0192]

[0193] Based on historical production data analysis, researchers used kernel density estimation to determine the standard proportion benchmark values ​​for each component and set stable ranges for the material ratios based on the calculated standard deviations. The upper limit warning value for the curing agent was 13.0%, and the lower limit warning value was 12.0%; the upper limit warning value for the admixture was 2.4%, and the lower limit warning value was 1.6%; the upper limit warning value for water was 86.1%, and the lower limit warning value was 84.9%. These thresholds were derived through statistical analysis of historical qualified product data, ensuring the sensitivity and accuracy of the warnings.

[0194] During actual operation, the system detected an abnormal fluctuation in the curing agent dosage ratio. As shown in Table 2, the curing agent dosage ratio started to be too high at 9:27:35 and remained so for about 45 seconds, reaching a maximum of 13.3%, exceeding the high-limit warning value. The system immediately marked this as an abnormal dosage ratio point and recorded the relevant parameters.

[0195] Table 2 Data Recording of Material Ratio Anomalies

[0196]

[0197]

[0198] The system applied a set of dynamic prediction equations for the material ratio to analyze the changing trend of the curing agent dosage ratio. Using a 30-second sliding time window, the calculated rate of change of the curing agent dosage ratio was 0.015% / second, and the acceleration was 0.002% / s². 2 Based on this data, the system predicts that without intervention, the proportion of hardener added will continue to rise over the next 20 seconds, potentially reaching a maximum of 13.5%.

[0199] Meanwhile, the system calculated a volatility risk coefficient of 0.78, indicating a high risk of severe material ratio anomalies. The weighted average of the volatility amplitude was 0.52%, and the volatility cycle characteristic value analysis showed that such volatility typically lasts 45-60 seconds.

[0200] Through temporal clustering analysis of historical material ratio anomalies, the system found that these anomalies are highly correlated with fluctuations in the No. 1 curing agent supply pump. The similarity calculated by the dynamic time warping algorithm is 0.92, exceeding the set threshold of 0.85, indicating that the current anomaly belongs to a typical pattern of "increased curing agent ratio caused by pump fluctuations". Association rule mining results show that the confidence level of this anomaly and pump start-up and shutdown operations is 0.82, and the support level is 0.25.

[0201] Cross-correlation analysis determined the time delay relationship between changes in equipment status and abnormal material ratios, as shown in Table 3.

[0202] Table 3. Statistics on Time Delay of Equipment Status Changes and Material Ratio Anomalies

[0203] Device state change Minimum delay (s) Maximum delay (s) Average delay (s) Standard deviation (s) Correlation coefficient Pump 1 start 5.2 12.4 8.7 1.8 0.42 Pump 1 stop 3.8 10.5 7.2 1.5 0.38 Pump 1 fluctuation 2.5 5.8 3.8 0.9 0.85 Pump 2 start 6.3 14.1 9.5 2.1 0.25 Admixture valve open 4.2 9.8 6.5 1.4 0.31 Admixture valve close 3.9 8.7 5.8 1.3 0.29

[0204] Based on a material flow rate of 1.0 m / s and an anomaly duration of 45 seconds, the system calculated the affected material volume to be approximately 2.7 m³. 3 The Hardy-Cross iterative method was used to simulate the flow trajectory of materials within the system, and the spatial distribution range of abnormal materials was determined.

[0205] Upon detecting an anomaly in the material ratio, the system immediately executes a material ratio fluctuation compensation procedure. Based on the fuzzy control strategy, the calculated deviation in the curing agent dosage ratio is 0.8%, falling into the "slightly high" category with a membership degree of 0.78. Applying Mamdani fuzzy inference and the center-of-gravity method for defuzzification, the system decides to reduce the curing agent dosage by 8% while increasing the water dosage by 1.5%. Simultaneously, the material diversion mechanism is activated to divert the 2.7m of material already affected by the fluctuation. 3 Materials are imported into a temporary storage area for secondary processing.

[0206] After the compensation measures were implemented, the system began to perform material ratio recovery adjustments. The adjustment process was divided into three stages: in the rapid approach zone (9:28:00-9:28:20), the amount of curing agent added was reduced at 80% of the maximum adjustment rate (0.12% / second); in the deceleration transition zone (9:28:20-9:28:40), the adjustment rate was linearly reduced to 30% of the maximum rate (0.045% / second); in the fine adjustment zone (9:28:40-9:29:00), the system was precisely adjusted to the standard value with the minimum adjustment step size (0.02% / second). As shown in Table 4, through the piecewise linear control strategy, the proportions of each component were smoothly restored to the normal range.

[0207] Table 4 Data on the Material Ratio Recovery and Adjustment Process

[0208]

[0209]

[0210] Traditional fluidized bed mixing processes rely primarily on manual, periodic sampling and adjustments based on experience, making it difficult to promptly detect and address material ratio fluctuations. Operators often only discover problems after the fluctuations have been occurring for a considerable time, and the analysis and handling of the causes of these fluctuations depend mainly on experience-based judgment, lacking scientific basis. Furthermore, traditional methods struggle to quantitatively analyze the relationship between different equipment conditions and material ratio fluctuations, and cannot accurately calculate the scope of the fluctuations' impact or implement targeted solutions.

[0211] This invention represents a significant advancement over traditional methods: First, it enables real-time monitoring and early warning of material ratio fluctuations, reducing the time to detect anomalies from tens of minutes to just a few seconds. Second, it accurately predicts material ratio trends through mathematical models, shifting from passive response to proactive prevention. Third, it establishes a correlation model between material ratio anomalies and equipment status, making fault diagnosis more precise. Fourth, it accurately calculates the range of influence of fluctuations, avoiding the waste caused by misjudging a large amount of qualified materials as unqualified in traditional methods. Finally, it employs intelligent control strategies to achieve automatic compensation and recovery adjustments, reducing human intervention and improving production stability and product quality consistency.

[0212] It should be noted that the variables involved in this invention are explained in detail in Tables 5 and 6 below.

[0213] Table 5. Variable Explanation Table (Part 1)

[0214]

[0215]

[0216] Table 6. Variable Explanation Table (Part Two)

[0217]

[0218]

[0219] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil, characterized in that, include: A material ratio fluctuation curve model is established. The input amounts of three components—curing agent, admixture, and water—are collected in real time by sensors, and the material ratio is plotted as a curve of change over time. A stable range for the material ratio is set, and high-limit and low-limit warning values ​​are generated. Abnormal points in the material ratio are recorded, and the system automatically marks and records the relevant parameters. Calculate the trend of material ratio variation, apply the material ratio dynamic prediction equation set to analyze the historical variation law of material ratio of each component and predict the future trend, and at the same time assess the fluctuation risk coefficient. Analyze the correlation between material ratio anomalies; Construct a model of the relationship between material ratio anomalies and equipment start-up and shutdown; Calculate the area affected by material ratio fluctuations; execute the material ratio fluctuation compensation procedure; implement material ratio recovery adjustment.

2. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 1, characterized in that, The material ratio fluctuation curve model is established by using sensors to collect the input amounts of three components—curing agent, additive, and water—in real time, and plotting the material ratio change curve over time.

3. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 2, characterized in that, The set material ratio stability range is based on historical production data to determine the allowable fluctuation range of the feeding ratio of the three components: curing agent, additives, and water, and to form high-limit warning values ​​and low-limit warning values.

4. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 3, characterized in that, Recording material ratio anomalies means that when the real-time monitored value exceeds the preset material ratio stability range, the system automatically marks it as a material ratio anomaly point and records the time of occurrence and related parameters.

5. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 4, characterized in that, Analyzing the correlation of material ratio anomalies involves performing temporal clustering on the marked material ratio anomalies to identify the time intervals, magnitude similarities, and triggering conditions between them.

6. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 5, characterized in that, Constructing a model of the relationship between material ratio anomalies and equipment start-up and shutdown involves establishing a correlation mapping between the occurrence of material ratio anomalies and equipment operating status, disc start-up and shutdown, and material conveying delay.

7. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 6, characterized in that, The calculation of the area affected by material ratio fluctuations is based on the time of occurrence of material ratio anomalies and the material flow rate to determine the location and range of the affected materials.

8. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 7, characterized in that, The material ratio fluctuation compensation program automatically adjusts the amount of curing agent, additives, and water added when an abnormal point in the material ratio is detected, and imports the fluctuating material into a preset area for temporary storage and processing.

9. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 8, characterized in that, Implementing material ratio restoration adjustment involves gradually restoring the input of curing agent, admixture, and water to standard values ​​after processing the material ratio fluctuation compensation procedure, ensuring production continuity and meeting the performance standards of the mixture.

10. The method for early warning and adjustment of fluctuations in the mixing ratio of fluidized solidified soil according to claim 9, characterized in that, The dynamic prediction equation set for material ratio includes a material ratio trend equation, a fluctuation amplitude equation, and a prediction correction equation. The material ratio trend equation calculates the rate of change and acceleration of the component feeding ratio within a selected time window. Inputs include the current component feeding ratio value, historical component feeding ratio values ​​from the previous N time points, time window length, component type identifier, and sampling frequency. Outputs the rate of change of the component feeding ratio and the predicted direction of change in the near future. The fluctuation amplitude equation evaluates the fluctuation amplitude and frequency characteristics of the material ratio. Inputs include the component feeding ratio values ​​from M consecutive time points, the standard proportion baseline value, the historical maximum peak value of fluctuations, equipment operating parameters, and material property parameters. Outputs the weighted average of the fluctuation amplitude and the characteristic value of the fluctuation period. The prediction correction equation corrects the prediction results based on the equipment start-up and shutdown status. Inputs include the initial prediction trend value, equipment start-up and shutdown status parameters, historical start-up and shutdown-induced fluctuation statistics, the correlation coefficient between material ratio anomalies and equipment start-up and shutdown times, and material conveying delay time. Outputs the predicted trend value and fluctuation risk coefficient after correction for equipment start-up and shutdown factors.

Images