A method and system for collaborative control of grouting parameters based on intelligent monitoring

Abstract

The application discloses a kind of based on intelligent monitoring's grouting parameter's collaborative control method and system, it is related to intelligent control technical field, including the time series data of the acquisition grouting control parameter set in preset time window, and corresponding trend feature encoding vector is constructed based on the time series data;Trend form is identified to each parameter based on trend feature encoding vector, and control intent matched with trend form is determined;The control adjustment value of control parameter is calculated based on the control intent and current trend direction;According to the control adjustment value, each parameter in grouting control parameter set is updated, current control period is completed, and next control period is entered.The grouting diffusion effect, density control precision and overall system regulation efficiency are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to a collaborative control method and system for grouting parameters based on intelligent monitoring. Background Technology

[0002] With the continuous expansion of underground engineering, tunnel construction, and coal mining operations, grouting reinforcement, as an important means of waterproofing, sealing, and surrounding rock stabilization, has been widely used in various geotechnical engineering projects. Control parameters during the grouting process include grouting pressure, grouting flow rate, diffusion radius, and grouting density. These parameters exhibit complex dynamic coupling relationships and are influenced by multiple factors such as geological conditions, material properties, and construction pace, resulting in a highly nonlinear and time-varying control process.

[0003] Existing grouting parameter control methods mainly employ fixed threshold adjustment, PID control, or manual adjustment based on historical experience. These methods struggle to respond in real-time to the interconnected behaviors of different parameters and have weak capabilities in identifying abnormal trends, failing to achieve refined control of the grouting process under complex conditions. Furthermore, current control strategies often focus on the parameters themselves, neglecting their different roles in the overall system trend evolution. This lack of a structured control mechanism based on trend behavior easily leads to problems such as adjustment delays and unstable grouting quality. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing grouting control methods have problems such as insufficient trend perception, fixed adjustment methods, and weak coordination between parameters, as well as how to construct trend maps and realize the path propagation of control intentions to drive multi-parameter coordinated adjustment.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a collaborative control method for grouting parameters based on intelligent monitoring, comprising:

[0007] Collect time series data of grouting control parameter set within a preset time window, and construct corresponding trend feature encoding vectors based on the time series data;

[0008] Based on the trend feature encoding vector, the trend pattern of each parameter is identified, and the control intention matching the trend pattern is determined.

[0009] Based on the control intent, path integral propagation is performed in the control parameter trend graph to generate a trend intent propagation value, and the control adjustment value of the control parameter is calculated in combination with the current trend direction;

[0010] Update each parameter in the grouting control parameter set according to the control adjustment value, complete the current control cycle, and enter the next control cycle.

[0011] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, wherein: the response speed characteristic parameter of the grouting area;

[0012] The construction of the corresponding trend feature encoding vector includes: preprocessing the time series data of each control parameter; calculating the velocity and acceleration values ​​of each control parameter based on the preprocessed time series data; and generating the corresponding trend feature encoding vector based on the current value, velocity value, and acceleration value of each control parameter.

[0013] The control period is a continuous time period formed by extracting the time series of control parameters according to a fixed sampling frequency based on a sliding time window of a preset length.

[0014] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, the step of identifying the trend pattern of each parameter includes: constructing a control parameter trend map based on a trend feature encoding vector; for each node in the control parameter trend map, calculating the sum of the trend coupling weights between the node and all other nodes, as an indicator of the trend influence of the node.

[0015] When the trend influence index of a node is higher than the average trend influence, the control parameter corresponding to the node is marked as a trend-driven parameter; otherwise, the control parameter corresponding to the node is marked as a response parameter.

[0016] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, the construction of the control parameter trend map includes: mapping each control parameter in the control parameter set to a node in a graph structure to form a node set; establishing a directed edge between nodes corresponding to any two different control parameters, and calculating a trend coupling weight based on the change of the trend feature encoding vector of the two nodes in two adjacent control cycles, wherein the trend coupling weight characterizes the degree of consistency of the trend change direction between the corresponding control parameters; and generating a control parameter trend map representing the trend coordination relationship between control parameters based on the node set, the directed edge set, and the corresponding trend coupling weight.

[0017] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, the determination of the control intent matching the trend pattern includes determining the control intent of trend-driven parameters and determining the control intent of response parameters.

[0018] The control intent for determining the trend-driven parameters includes extracting the trend feature encoding vector of the trend-driven control parameters over k consecutive control cycles, and constructing trend trajectory input data:

[0019] S i =[s i (tk),s i (t-k+1),…,s i (t)]

[0020] The trend trajectory input data is input into the trend state modeling function. Generate the trend latent variable state for the next control cycle:

[0021]

[0022] The state of the trend latent variable is input into the control intent mapping function to generate the original control intent:

[0023]

[0024] Constructing a structural adjustment vector based on the feedback effect of response parameters:

[0025]

[0026] The original control intent of the trend-driven parameters is fused with the structural adjustment vector to generate the control intent of the trend-driven parameters:

[0027]

[0028] Among them, S i This represents the trend trajectory input data for the i-th trend-driven control parameter; s i (t) represents the trend feature encoding vector of the i-th control parameter during control period t, where k represents the time window length and t represents the time index; z i (t+1) represents the trend latent variable state of the trend-driven control parameter in the next control cycle; This represents a function for modeling trend states. W represents the original control intent vector for trend-driven parameters. d R represents the mapping matrix from the latent trend variables to the control intention space; R represents the set of response parameters. This represents the structural feedback weight of the responsive parameter node j to the trend-driven parameter node i; Represents the predictive control intent vector of the response parameter j; π ji In the trend graph, from node v i to node v j The propagation impact weight; δ jThis represents the trend influence index of the response parameter j in the current control period t; u i The vector represents the control intent of trend-driven parameters; tanh(·) represents the hyperbolic tangent function; r i Indicates the structural feedback correction term;

[0029] The control intent for determining the response parameters includes taking the control intent for all trend-driven parameters as input and inputting it into the response mapping function G. j (·), generating the predictive control intent vector:

[0030]

[0031] Based on the path structure in the control parameter trend graph, calculate the propagation weight of structural influence:

[0032]

[0033] Construct the structural influence vector of the response parameters:

[0034]

[0035] By fusing the structural influence vector with the responsive predictive control intent, a final responsive control intent is generated:

[0036]

[0037] Where D represents the trend-driven parameter set; α represents the damping coefficient in the graph structure path propagation; l represents the step index of the propagation path; P represents the transition probability matrix of the control parameter trend graph, which is generated by normalizing the trend coupling weights between the control parameters; u j Indicates the response parameter p j The ultimate intention of control.

[0038] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, the current trend direction is calculated by subtracting the trend feature encoding vector of each control parameter in the control parameter set under the current control cycle from the trend feature encoding vector under the previous control cycle to form a vector of trend change direction.

[0039] The calculation of the control adjustment value of the control parameter includes matching the control intention vector of the control parameter in the current control cycle with the trend change vector, determining the control adjustment coefficient based on the degree of directional consistency between the control intention vector and the trend change vector, and generating the control adjustment value in combination with the preset adjustment gain parameter.

[0040] As a preferred embodiment of the collaborative control method for grouting parameters based on intelligent monitoring described in this invention, the step of updating each parameter in the grouting control parameter set includes combining the original value of each control parameter in the control parameter set in the current control cycle with its corresponding control adjustment value to generate the initial control parameter value for the next control cycle.

[0041] As a preferred embodiment of the collaborative control system for grouting parameters based on intelligent monitoring described in this invention, it includes a trend encoding module, a trend analysis module, a control intent generation module, a parameter adjustment module, and a parameter update module.

[0042] The trend encoding module is used to collect time series data of grouting control parameter set within a preset time window, construct a trend feature encoding vector for each parameter, and output a trend feature encoding matrix.

[0043] The trend analysis module is used to construct a trend map of control parameters based on the trend feature encoding matrix, identify the trend pattern of each control parameter in the current control cycle, and classify trend-driven parameters and response parameters according to the degree of trend influence.

[0044] The control intent generation module is used to generate control intent vectors for trend-driven parameters and control intent vectors for response parameters based on the coupling relationship between trend-driven parameters and response parameters in the trend graph, and then fuse and output them.

[0045] The parameter adjustment module is used to calculate the control adjustment coefficient based on the directional consistency between the trend direction vector of each parameter and the control intention vector under the current control cycle, and generate the control adjustment value in combination with the preset adjustment gain parameter.

[0046] The parameter update module is used to update the grouting control parameter set based on the control adjustment value, and input the updated parameters into the next control cycle to form a closed-loop control process.

[0047] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement a collaborative control method for grouting parameters based on intelligent monitoring.

[0048] A computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of a collaborative control method for grouting parameters based on intelligent monitoring.

[0049] The beneficial effects of this invention are as follows: The collaborative control method for grouting parameters based on intelligent monitoring provided by this invention constructs trend feature encoding vectors for parameters such as grouting pressure, grouting flow rate, and diffusion radius. This allows for real-time plotting of the changing trends of each parameter within a continuous control cycle, and identifies the coupling relationship between key influencing parameters and controlled response parameters based on trend maps. Based on the identification of trend-driven parameters, the system constructs control intentions by combining the intensity of their trend influence on response parameters. Furthermore, it optimizes the original intentions using response behavior feedback through a trend correction mechanism, improving the adaptability and accuracy of the control strategy. Finally, the control intention vector is matched with the direction of trend change, dynamically adjusting the control amplitude to ensure the stability and target orientation of the collaborative changes of each parameter, significantly improving the grouting diffusion effect, density control accuracy, and overall system regulation efficiency. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 The first embodiment of the present invention provides an overall flowchart of a collaborative control method for grouting parameters based on intelligent monitoring. Detailed Implementation

[0052] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0053] Example 1, referring to Figure 1 As one embodiment of the present invention, a collaborative control method for grouting parameters based on intelligent monitoring is provided, comprising:

[0054] S1: Collect time series data of grouting control parameter set within a preset time window, and construct a corresponding trend feature encoding vector based on the time series data.

[0055] The control cycle is the time period corresponding to one grouting status assessment and control strategy update by the system. At a sampling frequency of one second, the system continuously acquires the raw data of the grouting control parameters and extracts a fixed-length time series using a sliding time window of length k, which serves as the complete input for the current control cycle. The starting point of the control cycle is the first sampling moment after the system completes the previous round of control parameter updates, and the ending point is the last sampling point required to complete the current sliding window length. The cycle length is the same as the sliding time window.

[0056] Within each control cycle, the system collects time-series data of multiple control parameters from the grouting site to construct an input vector for trend analysis. The set of control parameters includes: grouting pressure, grouting flow rate, diffusion radius, grouting density index, and response velocity characteristic parameters of the grouting area.

[0057] Among them, grouting pressure reflects the real-time advancement intensity of the grouting process, grouting flow rate characterizes the injection rate of grout per unit time; the diffusion radius estimate is used to describe the propagation range of grout in pores or fractures, and can be obtained indirectly through ground-penetrating radar, sonic inversion and other means; grouting density index is used to measure the structural filling effect of the injection area, and is usually calculated by density sensing equipment or image analysis system; in addition, it also includes response speed parameters of the grouting area, such as borehole echo vibration, borehole structure amplitude response and other characteristic quantities, which are used to reflect the dynamic response after the grout enters.

[0058] Let the above control parameters correspond to p1, p2, ..., p n We set the sampling time to once per second and use a sliding time window of length k (5 minutes) to construct the time series of each parameter:

[0059] p i (t)=[p i (t-k+1),p i (t-k+2),…,p i (t)]

[0060] Where, p i The i-th grouting control parameter is represented by n; n represents the total number of control parameters; t represents the time corresponding to the current control cycle; k represents the length of the sliding time window; p i (t) represents the time series vector of the i-th parameter within the current time window, consisting of continuous sampled values ​​from time t-k+1 to t, p i (t-k+1),p i (t-k+2),…,p i (t) represent the parameters p i The specific sampled values ​​at each historical moment are arranged in chronological order, reflecting the evolution of the parameter within the current time window. 'i' represents the index.

[0061] Outlier removal is performed on the time series data of each parameter using the sliding median filtering method, and the processed values ​​are mapped to the interval [0, 1] using the linear normalization method to improve the comparability between parameters of different dimensions and enhance the stability and consistency of subsequent trend modeling.

[0062] Based on the processed time series data, numerical difference calculations are used to calculate the velocity and acceleration characteristics of each control parameter in the current cycle:

[0063]

[0064] Among them, v i (t) represents the velocity characteristic of the i-th control parameter at time t, that is, the rate of change of this parameter within the current period; p i (t) represents the value of the i-th control parameter at the current time t; p i (t-1) represents the value of the i-th control parameter at the previous control time; Δt represents the time interval between two adjacent control cycles, usually set to 1 second; a i (t) represents the acceleration characteristic of the i-th control parameter at time t, i.e., the change in its rate of change; v i (t-1) represents the velocity characteristic value of this parameter in the previous control cycle.

[0065] A moving average process is introduced to suppress sampling noise, expressed by the formula:

[0066]

[0067] in, This represents the smoothed velocity value of the i-th control parameter at the current time t; v i (tj) represents the velocity value of this parameter before the first j cycles; This represents the smoothed acceleration value of the i-th control parameter at the current time t; a i (tj) represents the acceleration value of the parameter before the previous j periods; m represents the number of historical periods used for the moving average, i.e. the length of the smoothing window; j represents the number of time offset steps being processed in the current calculation, from 0 to m-1.

[0068] Furthermore, for each control parameter p i Construct trend feature encoding vectors:

[0069]

[0070] Among them, s i (t) represents the trend feature encoding vector of the i-th control parameter at the current time t; pi (t) represents the original value of this parameter in the current control cycle; v i (t) represents the velocity characteristic value of the parameter at the current moment, reflecting the rate of change of the parameter; a i (t) represents the acceleration characteristic value of the parameter at the current moment, reflecting the change in the rate of change of the parameter.

[0071] By collecting time-series data of grouting control parameters within a preset time window, including grouting pressure, grouting flow rate, diffusion radius, density index, and response velocity characteristic parameters, and preprocessing the time series of each parameter to calculate the current value, velocity value, and acceleration value, a trend feature encoding vector is constructed based on this. The encoding vectors of each control parameter are combined in index order to form a trend feature encoding matrix, providing a quantitative representation of dynamic behavior for subsequent trend map construction and control intent generation.

[0072] Existing technologies often employ fixed-period sampling followed by direct input into the control model or utilize static weights for parameter adjustment, failing to consider the dynamic trend characteristics during parameter evolution. This makes it difficult to capture the directionality and speed of parameter changes, leading to adjustment lag and misadjustment problems. In contrast, this method, by introducing a trend encoding mechanism for velocity and acceleration, overcomes the traditional limitation of "adjusting solely based on current values." It establishes a multi-dimensional representation of multi-parameter trend behavior, enabling the control logic to perceive the directional trend of parameter changes over time, achieving a more agile and forward-looking control response.

[0073] The trend feature encoding mechanism provides a high-fidelity input foundation for subsequent trend map construction, driving parameter identification, and control intent propagation. While improving the accuracy of trend identification, it also significantly enhances the system's ability to model nonlinearity, coupling, and dynamic evolution. By introducing dynamic trend vectors to replace static parameter values, the adaptability and robustness of the control system to complex grouting environments can be effectively improved. It exhibits superior adjustment response speed and control stability compared to traditional control schemes, providing more predictive and safer control capabilities for grouting operations in high-risk scenarios.

[0074] S2: Identify the trend pattern of each parameter based on the trend feature encoding vector, and determine the control intent matching the trend pattern.

[0075] The system constructs a trend graph structure G(t) based on the trend feature matrix to characterize the dynamic trend synergy among various control parameters.

[0076] Let each control parameter correspond to a graph node v i Introduce a directed edge (v) between any two nodes. i →v j The trend coupling weights of the edges are calculated using the following formula:

[0077]

[0078] Based on the above weights, construct a trend graph of the control parameters:

[0079] G(t) = (V, ε(t), W(t))

[0080] Among them, s i (t) represents the trend feature encoding vector of the i-th control parameter in the current control period t; s i (t-1) represents the trend feature encoding vector of the i-th parameter in the previous control period t-1; s j (t), s j (t-1) are the trend feature encoding vectors of the j-th control parameter in the current period and the previous period, respectively; i and j represent the index numbers of any two different control parameters, satisfying i≠j, i,j∈[1,n]; (s i (t)-s i (t-1)) represents the trend change vector of the i-th parameter between adjacent periods; (s j (t)-s j (t-1)) represents the trend change vector of the j-th parameter; w ij (t) represents the directional similarity between the trend change of the i-th parameter and the trend change of the j-th parameter, i.e., the trend coupling weight, with a value range of [-1, 1]. V = {v1, v2, ..., v n} represents the set of control parameter nodes; ε(t) represents the set of directed edges formed by the trend coupling relationship between parameters at period t; W(t) is the set of nodes formed by all w ij The trend coupling weight matrix consists of G(t); G(t) represents the trend map of control parameters constructed during the control period t.

[0081] To facilitate the propagation and iteration of structural influences, the system normalizes the weight matrix W(t) to obtain the transition matrix P of the trend graph:

[0082]

[0083] Among them, P ij This indicates that in the control parameter trend graph, from node v i Transfer to node v j Normalized transition probability; max(w) ij (t),0): indicates that only the trend coupling weight w is taken. ij The non-negative part of (t) is considered to have no positive effect if it is negative, and its value is 0; k represents the index variable used to traverse the nodes, and its range is 1≤k≤n.

[0084] The system performs centrality analysis on the trend graph structure to identify the structural role of each control parameter in the trend graph within the current period. Each node v is defined. i The trend influence index is:

[0085]

[0086] when At that time, the marker parameter p i If it is a "trend-driven parameter", it means that its trend changes have a significant impact on the overall trend of the system; otherwise, it is considered a "response parameter", meaning that its trend changes mainly due to the influence of other parameters.

[0087] Where, ζ i (t) represents the trend influence index of the i-th parameter node under control period t; n represents the total number of control parameters; j represents the target parameter index, with a value range from 1 to n; w ij (t) represents the parameter p i Pointer to parameter p j Trend coupling weights; |w ij (t)| represents the absolute value of the trend coupling weight, used to avoid the cancellation of the overall influence of positive and negative directions; This indicates the degree of influence of the average trend.

[0088] After completing the identification and classification of parameter trend patterns, the control parameters are divided into a trend-driven parameter set D and a response parameter set R.

[0089] The determination of control intent matching the trend pattern includes determining control intent for trend-driven parameters and determining control intent for response parameters.

[0090] Determining the control intent for trend-driven parameters includes: for each trend-driven parameter p i ∈D, the system constructs the original control intent based on its trend change vector:

[0091]

[0092] For each of the response parameters affected by it Calculate the absorption error vector:

[0093]

[0094] in, Indicates the control parameter p i The trend change vector under the control period t; s i (t) represents the control parameter p i The trend feature encoding vector under the current control period t; s i (t-1) represents the control parameter pi The trend feature encoding vector under the previous control period t-1; ε ij (t) represents the response parameter p under control period t. j For trend-driven parameter p i The trend absorption error vector; s j (t) represents the response parameter p j The trend feature encoding vector in the current period; s j (t-1) represents the response parameter p j The trend feature encoding vector from the previous period; α ij (t) represents the trend-driven parameter p under control period t. i For the response parameter p j Trend absorption coefficient As before, this indicates the trend-driven parameter p. i The trend change vector.

[0095] We construct a trend intention correction term by weighted aggregation of all absorbed errors in the responses:

[0096]

[0097] Calculate the adaptive correction coefficient for trend intent:

[0098]

[0099] The final control intent vector of the trend-driven parameters is obtained as follows:

[0100]

[0101] Where, Δ i (t) represents the trend-driven parameter p. i The trend intention correction term under the control period t; β ij Indicates parameter p j For parameter p i The absorption error weighting coefficient; R i Indicates the trend-driven parameter p i The set of response parameters that affect; g i (t) represents the trend-driven parameter p. i The trend intention correction gain coefficient is used to dynamically adjust the trend correction magnitude; ∥Δ i (t)∥ represents the Euclidean norm (modulus) of the trend intention correction term, reflecting the current cumulative deviation strength; κ represents the slope adjustment factor of the gain function; δ represents the deviation threshold term, used to shift the midpoint of the sigmoidal curve; exp(·) represents the exponential function; The trend-driven parameter p represents the control period t. iThe control intent vector, integrating original trends and feedback corrections; γ i (t) represents the trend enhancement gain factor of the trend-driven parameter, which controls the strength of the original trend; g i (t)·Δ i (t) represents the trend feedback term based on the absorption error correction, which is used to dynamically suppress or compensate for the original trend.

[0102] Determining the control intent for response parameters includes:

[0103] Each response parameter p j The control intent ∈R partly stems from the combined influence of multiple trend-driven parameters. Let the set of its trend-driven influences be . By each p i ∈D j The influence vector obtained by projection is:

[0104]

[0105] Constructing a weighted fusion vector for trend-driven control projection:

[0106]

[0107] in, Indicates the response parameter The trend-driven subset of parameters that have an impact, i.e., all parameters related to p j The set of driving parameters with trend projection, p i ∈D j This indicates a trend-driven parameter that belongs to this set. Indicates the response parameter p j The external control intent generated by trend-driven parameters under control period t; ω ij (t) represents the trend-driven parameter p. i For the response parameter p j The fusion weighting coefficient; s j (t) represents the current trend feature encoding of the response parameter.

[0108] Calculate the degree of consistency between the internal trend and the external trend:

[0109]

[0110] Generate fusion weights based on consistency:

[0111]

[0112] The ultimate control intent of response parameters The vector is formed by a weighted fusion of external trends and its own trends:

[0113]

[0114] Where, ρ j (t) represents the response parameter p j Trend consistency indicators; The dot product of autonomous tendencies and external control intentions; Let λ and λ represent the Euclidean norms (i.e., vector lengths) of the two vectors, respectively; j (t) represents the response parameter p j External trend fusion coefficient; η represents the fusion sensitivity adjustment factor; exp(·) represents the exponential function; 1-λ j (t) represents the proportion of autonomous trends, inversely proportional to the consistency of external trends; γ j (t) represents the enhancement gain coefficient of the autonomous trend of the response parameter.

[0115] The control intent of trend-driven parameters and the final control intent of response parameters are combined to construct a unified control intent set matrix U(t) as follows:

[0116]

[0117] in, This represents the control intent vector of the k-th trend-driven parameter under period t; p represents the control intent vector of the k-th response parameter at period t; k This represents the k-th control parameter in the system; Represents a trend-driven parameter set; p represents the set of response parameters. k ∈D indicates that the parameter is a trend-driven parameter; This indicates that the parameter is a response parameter.

[0118] After constructing the trend feature encoding vector, the dynamic collaborative relationship between various control parameters in the system can be accurately characterized by calculating the similarity of trend change directions among control parameters and constructing a trend graph structure. Unlike traditional methods that adjust parameters based on fixed coupling relationships or static control rules, this scheme introduces a directed graph structure driven by trend features and further extracts the dominant trend parameters and passive response parameters through node centrality indicators, achieving dynamic identification of the master-slave relationship of system control behavior at the structural level.

[0119] Building upon this foundation, a feedback adjustment mechanism for control intent is achieved by constructing a trend absorption error vector for trend-driven parameters on response parameters. This absorption error vector does not rely on static residual thresholds or regular mappings; instead, it accurately models the deviation of trend intent through the difference between the trend change direction of the response parameters and the expected projection. The error terms are weighted and aggregated to form a trend correction vector, and an adjustment coefficient is generated by an adaptive sigmoid gain function. This enables dynamic callback of trend control intent, effectively avoiding system oscillations and instability problems caused by excessive or insufficient trend control in traditional methods.

[0120] The control intent generation mechanism for responsive parameters further breaks through the existing passive linear superposition of response behaviors. In this scheme, the responsive parameter not only integrates the control projections of multiple trend-driven parameters, but also adaptively adjusts the fusion ratio between trend control and its own behavior by constructing a directional consistency index between its own trend changes and external trend intent. This mechanism constructs a trend consistency fusion factor through angle similarity and a sigmoid fusion function, enabling the responsive parameter to have self-driven adjustment capabilities when its behavior cannot be explained by external trends. This avoids the problem of overfitting of the response parameter to trend behavior in traditional methods, and significantly improves the stability and controllability of the system under abrupt disturbances.

[0121] Ultimately, the generated control intent matrix achieves coordinated adjustment of trend-driven and response parameters within each control cycle. It can also construct parameter adjustment paths based on the directional relationship between control intent and trend changes, providing a structurally complete and behaviorally defined input foundation for subsequent trend direction matching and adjustment amplitude generation.

[0122] Overall, this solution breaks through the technical limitations of traditional grouting control algorithms, such as residual adjustment, rule-based decision-making, or static feedback structures, in multiple aspects, including trend control graph construction, parameter master-slave classification, feedback callback mechanism, and adaptive fusion logic. It has stronger time-varying adaptability, structural expression ability, and intelligent control performance, and is suitable for highly complex and highly coupled industrial dynamic parameter control scenarios.

[0123] S3: Calculate the control adjustment value of the control parameters based on the control intention and the current trend direction.

[0124] During the current cycle, the system no longer propagates the control intent graph structure. Instead, it takes the control intent as input and combines the trend and direction characteristics of each parameter to generate control adjustment values ​​with directional constraints.

[0125] Specifically, for each control parameter p k Its current trend direction vector d k (t) is defined as:

[0126] d k (t)=s k (t)-s k (t-1)

[0127] Among them, s k (t) represents the trend feature encoding vector of the control parameters under the control period t, including the current value, velocity, and acceleration features, d k (t) reflects the direction of its trend change within adjacent periods. The system uses a direction-weighted fusion method to calculate the control adjustment value Δp of the control parameters. k (t):

[0128]

[0129] Where, d k (t) represents the control parameter p k The trend direction vector under control period t; s k (t) represents the control parameter p k Trend feature encoding vector under control period t; s k (t-1) represents the trend characteristic encoding vector of this control parameter in the previous control cycle; t represents the index of the current control cycle; k represents the index number of the control parameter, i.e., the kth control parameter. Δp k (t) represents the control parameter p k The control adjustment value under control period t; λ is the adjustment coefficient, representing the overall amplitude of the control adjustment value; Indicates the control parameter p k The control intent vector under control period t is derived from the structure fusion control output of S2; it represents a very small positive constant to avoid numerical instability caused by division by zero.

[0130] In the current control cycle, a trend direction vector is introduced as the basis for generating control adjustment values. By performing differential calculations on the trend feature encoding of control parameters in previous and subsequent cycles, a trend direction vector containing the current value, rate of change, and acceleration is constructed. This direction vector not only reflects the dynamic trend of the control parameters but also their evolution trend in the time series. Based on this, the system adopts a direction-weighted mechanism to combine the structural fusion control intent with the trend direction, forming control adjustment values ​​with directional constraints. This effectively avoids the problem of adjustment direction deviating from the trend change path, enhancing the predictability and target consistency of control actions.

[0131] Unlike traditional control strategies that often rely on graph structure propagation or directly output control intentions, the current solution uses trend direction constraint logic to achieve a structure-embedded mechanism that generates effective adjustment quantities without relying on path diffusion in the graph. This design reduces the risk of structural propagation errors being transmitted during the control phase and can quickly converge to stable control values ​​under conditions of abrupt trend changes or complex interactions, effectively improving the system's steady-state response and adjustment accuracy in nonlinear dynamic scenarios.

[0132] Furthermore, this control adjustment mechanism breaks through the bottleneck of existing control based on static adjustment or structural diffusion by combining trend direction perception with structural intent. It achieves sensitive response to multi-dimensional trend information and directional control of adjustment path, significantly improving the timeliness, stability and consistency of parameter adjustment and execution constraints.

[0133] S4: Update each parameter in the grouting control parameter set according to the control adjustment value, complete the current control cycle, and enter the next control cycle.

[0134] Within the current control period t, the system updates each control parameter in the grouting control parameter set based on the control adjustment value generated by the trend intention propagation process. Specifically, this includes combining the original value of each control parameter in the current control period with the corresponding control adjustment value to generate the initial parameter input value for the next control period, expressed as:

[0135] p k (t+1)=p k (t)+Δp k (t)

[0136] Where, p k (t) represents the original value of the k-th control parameter in the control parameter set during the current control period t, Δp k (t) represents the control adjustment value calculated according to the control intent generation process, p k (t+1) represents the updated control parameter value. After updating the values ​​of all control parameters in the control parameter set, the system uses the updated control parameter set as input to re-enter the next control cycle and continue to execute steps such as trend feature encoding, trend pattern recognition, control intent generation, and trend intent propagation, forming a closed-loop control process driven by trend behavior.

[0137] Within the current control cycle, the system updates each parameter in the grouting control parameter set based on the control adjustment value generated by the trend intent propagation mechanism, forming the initial input parameters for the next control cycle. By combining the original parameter values ​​with the propagated adjustment values, a stable and more responsive parameter input set is constructed, enabling the control logic to have adaptive adjustment capabilities and continuous response capabilities within the cycle, providing a more stable initial input for subsequent trend feature analysis and intent generation.

[0138] Compared to traditional grouting control methods that generally employ fixed threshold adjustment or linear gain correction, which only update the parameters directly based on error feedback, this method cannot dynamically perceive the coordinated change path between control parameters. Furthermore, the update strategy is often independent of the overall trend evolution structure of the system, resulting in lagging control strategy, slow system response, and limited adjustment accuracy, making it difficult to achieve efficient closed-loop control under multi-parameter coupling.

[0139] By directly using the control intent values ​​generated from propagation in the trend graph for parameter update calculations, this method no longer relies on preset rules or static control factors. Instead, it performs cross-parameter and cross-period information fusion and behavior transfer based on the propagation results of the graph structure, breaking the traditional model of separating control parameter updates from trend structures. This path integral propagation mechanism assigns structural trend inputs to each control parameter, giving its update behavior globally collaborative attributes and constructing a control closed-loop system with structural self-organizing characteristics. This breakthrough overcomes the locality bottlenecks of existing technologies at both the algorithm mechanism and system level.

[0140] Example 2 is an embodiment of the present invention, which provides a collaborative control method for grouting parameters based on intelligent monitoring. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiment.

[0141] A -630m deep coal mine cut-out section was selected as the test section. The surrounding rock in this section is mainly medium- to fine-grained feldspathic sandstone, with well-developed fissures and significant stress concentration. Conventional grouting parameter control is insufficient to simultaneously control the diffusion range and the density of the surrounding rock. To verify the effectiveness of the "collaborative control method for grouting parameters based on intelligent monitoring," four sets of grouting holes were arranged in the test roadway, with three holes in each set, a depth of 6m, and a diameter of 42mm, arranged in a staggered "quincunx" pattern. 24 hours before the test, an integrated pressure-flow electronic grout pump, an online density meter, a distributed acoustic inversion device (for calculating the diffusion radius), and a high-frequency excitation response sensor (for extracting the response velocity characteristics of the surrounding rock) were installed. The sampling frequency was uniformly set to 1 time / minute; the control cycle was set to 5 minutes; and the sliding time window was set to 15 minutes (i.e., k=15). The initial system parameters were set as follows: grouting pressure 1.20MPa, grouting flow rate 28.50L / min, target diffusion radius 2.10m, and density index 0.70.

[0142] After continuously collecting raw data for 15 minutes, the five parameters were filtered using a moving median filter to remove outliers, and then linearly normalized to the [0, 1] interval. Subsequently, the first and second differences were calculated to obtain velocity and acceleration features, which were then concatenated with the current value to generate a trend feature encoding vector of length 3. The vectors of the five parameters were stacked vertically to form a 5×3 trend feature encoding matrix, which was then written into the database.

[0143] Using the encoding matrices from the latest two cycles as input, the trend coupling weights between nodes are calculated using cosine similarity, and a 5-node directed weighted graph is dynamically constructed. The sum of the absolute values ​​of the incident weights of each node is calculated to obtain the trend influence degree; nodes with an influence degree greater than the average are identified as trend-driven parameters. The first cycle's determination results are: grouting pressure and grouting flow rate are trend-driven, while diffusion radius, density, and response speed are response-driven.

[0144] For the two trend-driven parameters, the original intention is first constructed in the vector space. Then, based on the actual trends of the three response parameters under their influence, an absorption error vector is constructed. The vectors are aggregated and the correction gain is calculated using the Sigmoid function to complete the direction-amplitude correction of the original intention. For the three response parameters, the two driving parameters are first fused and their trend projections are obtained to obtain the external intention. Then, the cosine consistency is calculated with their respective autonomous trend directions to generate the fusion weight, forming the final control intention.

[0145] The system calculates the angle between the current trend direction vector and the corresponding control intention vector for each of the five parameters, mapping the cosine value to a directional consistency coefficient. Combined with a preset gain parameter of 0.15, it outputs a control adjustment value (positive values ​​represent increase, negative values ​​represent decrease). Adjustment commands are sent from the PLC to the variable frequency slurry pump and proportional valve, updating pressure and flow rate in real time. Simultaneously, the water-cement ratio is adjusted at the liquid preparation station to indirectly regulate density.

[0146] The updated parameters are then entered into the next control cycle, repeating steps ①–④. The entire process runs continuously for 5 control cycles (25 minutes), with the original and adjusted data recorded by the OPC server throughout to ensure traceability.

[0147] Over five consecutive control cycles, the grouting pressure increased from 1.20 MPa, 1.35 MPa, 1.50 MPa, 1.48 MPa to 1.52 MPa; the grouting flow rate was recorded as 28.50 L / min, 30.10 L / min, 31.80 L / min, 32.00 L / min, and 32.50 L / min; the grout diffusion radius calculated by acoustic inversion was 2.10 m, 2.25 m, 2.42 m, 2.39 m, and 2.51 m, respectively; the density index increased from 0.70 to 0.75, 0.82, 0.81, and finally reached 0.85; the arrival time of the surrounding rock vibration signal decreased from 105.00 ms to 98.50 ms, 93.80 ms, and 94.50 ms, and finally stabilized at 91.60 ms.

[0148] Data from five control cycles show that the method of this invention can achieve a better balance of multiple parameters in the grouting system within a short time (25 minutes). The grouting pressure rapidly increases by 12.50% in the second cycle and continues to increase by 11.11% in the third cycle, before being finely adjusted to 1.52 MPa. Simultaneously, the grouting flow rate is adjusted upwards according to the pressure-flow coupling curve, with a total increase of 14.04%. Traditional constant-pressure grouting often results in pressure fluctuations >0.10 MPa due to untimely flow-pressure matching, while the maximum fluctuation in this experiment was only 0.04 MPa, demonstrating the pressure stabilization advantage brought by trend control. The diffusion radius ultimately increased by 0.41 m, a 19.52% increase relative to the first cycle, indicating a significant increase in the penetration coverage of the grout in the fractures. If empirical constant flow rate control is used, the diffusion radius can only be increased by an average of about 10%, significantly lower than the effect achieved by this invention. The density index improved by 21.43% (from 0.70 to 0.85), and after a slight decline in the fourth cycle, it rebounded in the fifth cycle, indicating that the trend absorption error correction logic successfully suppressed the local backflow problem caused by over-dense grouting, achieving adaptive compensation. The continuous decrease in the arrival time of surrounding rock vibration indicates that the dynamic response of the surrounding rock-grout coupling system has accelerated. With the optimization of pressure and flow rate linkage, the dynamic steady-state switching time of the system has been shortened from ≈120ms under traditional control to ≤95ms, with the measured optimal value of 91.60ms, an improvement of 23.24%.

[0149] Compared with traditional grouting control schemes based on PID or fixed rules, this embodiment presents three significant advantages: First, the parameter classification driven by the trend graph avoids the coupling conflict caused by the unified adjustment of five parameters, enabling the main control pressure and flow rate to achieve rapid and stable gains; Second, the trend absorption error correction term allows the system to detect the mismatch between trend-driven and response-following in real time and implement vector-level correction, most clearly demonstrated by the third-cycle diffusion radius, which is reduced from 2.42m to 2.39m and then increased again; Third, the autonomous trend weight adjustment function of the response parameters ensures that the proportion of external projection is automatically reduced when the external trend is inconsistent with its own safety threshold, avoiding pulsating pressure rises caused by excessive following. Comprehensive analysis shows that the method of this invention significantly outperforms existing technologies in four dimensions: stability, response speed, grouting effect coverage, and surrounding rock reinforcement quality, fully demonstrating the creativity and novelty of the trend-behavior collaborative algorithm.

[0150] Example 3, an embodiment of the present invention, provides a collaborative control system for grouting parameters based on intelligent monitoring, including a trend encoding module, a trend analysis module, a control intent generation module, a parameter adjustment module, and a parameter update module.

[0151] The trend encoding module is used to collect time series data of grouting control parameter set within a preset time window, construct a trend feature encoding vector for each parameter, and output a trend feature encoding matrix.

[0152] The trend analysis module is used to construct a trend map of control parameters based on the trend feature encoding matrix, identify the trend pattern of each control parameter in the current control cycle, and classify trend-driven parameters and response parameters according to the degree of trend influence.

[0153] The control intent generation module is used to generate control intent vectors for trend-driven parameters and control intent vectors for response parameters based on the coupling relationship between trend-driven parameters and response parameters in the trend graph, and then fuse and output them.

[0154] The parameter adjustment module is used to calculate the control adjustment coefficient based on the directional consistency between the trend direction vector of each parameter and the control intention vector under the current control cycle, and generate the control adjustment value in combination with the preset adjustment gain parameter.

[0155] The parameter update module is used to update the grouting control parameter set based on the control adjustment value, and input the updated parameters into the next control cycle to form a closed-loop control process.

[0156] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0157] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0158] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0159] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0160] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A collaborative control method for grouting parameters based on intelligent monitoring, characterized in that, include: Collect time series data of grouting control parameter set within a preset time window, and construct corresponding trend feature encoding vectors based on the time series data; Based on the trend feature encoding vector, the trend pattern of each parameter is identified, and the control intention matching the trend pattern is determined. The identification of the trend pattern of each parameter includes constructing a trend map of control parameters based on the trend feature encoding vector; For each node in the control parameter trend graph, the sum of the trend coupling weights between the node and all other nodes is calculated as the trend influence index of the node. When the trend influence index of a node is higher than the average trend influence, the control parameter corresponding to the node is marked as a trend-driven parameter; otherwise, the control parameter corresponding to the node is marked as a response parameter. The determination of control intent matching the trend pattern includes determining control intent for trend-driven parameters and determining control intent for response parameters; The control intent for determining the trend-driven parameters includes, within the current control cycle, calculating the trend change vector based on the trend feature encoding vector of the trend-driven parameters and the trend feature encoding vector of the previous control cycle, and constructing the original control intent. For the response parameters under the influence of the trend-driven parameters, calculate the difference between the actual trend change and the trend projection to form an absorption error vector; We perform weighted aggregation on all absorption error vectors to construct a trend correction vector; The adaptive correction coefficient is calculated based on the deviation between the norm of the trend correction vector and a set threshold. Based on the original control intent, the trend correction vector, and the adaptive correction coefficient, a control intent vector for trend-driven parameters is constructed. The determination of the control intent for the responsive parameter includes: determining the set of trend-driven parameters that influence each responsive parameter; obtaining the trend projection vector of each trend-driven parameter onto the responsive parameter; and performing weighted fusion of all trend projection vectors of the responsive parameter to construct the external control intent for the responsive parameter. Calculate the current periodic trend change vector of the response parameter and calculate the consistency index with the direction of the external control intention. The fusion coefficient is calculated based on the consistency index, and the control intent vector of the response parameter is constructed by combining the external control intent, the autonomous trend change vector, and the trend enhancement gain. Calculate the control adjustment value of the control parameters based on the control intent and the current trend direction; Update each parameter in the grouting control parameter set according to the control adjustment value, complete the current control cycle, and enter the next control cycle.

2. The collaborative control method for grouting parameters based on intelligent monitoring as described in claim 1, characterized in that: The set of control parameters includes grouting pressure, grouting flow rate, diffusion radius, grouting density index, and response speed characteristic parameters of the grouting area; The construction of the corresponding trend feature encoding vector includes preprocessing the time series data of each control parameter; Based on the preprocessed time series data, the velocity and acceleration values ​​of each control parameter are calculated respectively. And based on the current value, velocity value and acceleration value of each control parameter, a corresponding trend feature encoding vector is generated; The control period is a continuous time period formed by extracting the time series of control parameters according to a fixed sampling frequency based on a sliding time window of a preset length.

3. The collaborative control method for grouting parameters based on intelligent monitoring as described in claim 2, characterized in that: The construction of the control parameter trend map includes: mapping each control parameter in the control parameter set to a node in the graph structure to form a node set; establishing a directed edge between nodes corresponding to any two different control parameters; and calculating the trend coupling weight based on the change of the trend feature encoding vector of the two nodes in two adjacent control periods. The trend coupling weight represents the degree of consistency of the trend change direction between the corresponding control parameters. Based on the set of nodes, the set of directed edges, and the corresponding trend coupling weights, a control parameter trend map representing the trend synergy relationship between control parameters is generated.

4. The collaborative control method for grouting parameters based on intelligent monitoring as described in claim 3, characterized in that: The current trend direction is calculated by taking the difference between the trend feature encoding vector of each control parameter in the control parameter set under the current control cycle and the trend feature encoding vector under the previous control cycle, and forming a vector of trend change direction. The calculation of the control adjustment value of the control parameters includes: obtaining the control intention vector of the control parameters in the current control cycle; calculating the degree of directional consistency between the control intention vector and the corresponding trend direction vector; constructing a control adjustment coefficient based on the degree of directional consistency; and generating the control parameter adjustment value in the current control cycle by combining it with the set adjustment gain parameter.

5. The collaborative control method for grouting parameters based on intelligent monitoring as described in claim 4, characterized in that: The process of updating each parameter in the grouting control parameter set includes combining the original value of each control parameter in the current control cycle with its corresponding control adjustment value to generate the initial control parameter value for the next control cycle.

6. A system employing the collaborative control method for grouting parameters based on intelligent monitoring as described in any one of claims 1 to 5, characterized in that: It includes a trend encoding module, a trend analysis module, a control intent generation module, a parameter adjustment module, and a parameter update module; The trend encoding module is used to collect time series data of grouting control parameter set within a preset time window, construct a trend feature encoding vector for each parameter, and output a trend feature encoding matrix. The trend analysis module is used to construct a trend map of control parameters based on the trend feature encoding matrix, identify the trend pattern of each control parameter in the current control cycle, and classify trend-driven parameters and response parameters according to the degree of trend influence. The control intent generation module is used to generate control intent vectors for trend-driven parameters and control intent vectors for response parameters based on the coupling relationship between trend-driven parameters and response parameters in the trend graph, and then fuse and output them. The parameter adjustment module is used to calculate the control adjustment coefficient based on the directional consistency between the trend direction vector of each parameter and the control intention vector under the current control cycle, and generate the control adjustment value in combination with the preset adjustment gain parameter. The parameter update module is used to update the grouting control parameter set based on the control adjustment value, and input the updated parameters into the next control cycle to form a closed-loop control process.

Images