Cement slurry proportioning self-adaptive control method and system

Abstract

The application provides a cement slurry proportioning self-adaptive regulation method and system, relates to the technical field of building engineering, and comprises the following steps: collecting geological and environmental data and analyzing to obtain a scene constraint vector, obtaining slurry raw material data to establish a proportioning boundary space, mapping and solving an initial proportioning scheme, and performing rheological simulation and deviation correction to obtain a proportioning adjustment amount, iteratively correcting to obtain a target proportioning scheme, evaluating a fault tolerance margin to generate a segmented regulation strategy, determining an execution parameter, and issuing a driving. The cement slurry proportioning accurate regulation is realized, and the construction adaptability is improved.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, and in particular to a method and system for adaptive control of cement slurry mix proportion. Background Technology

[0002] In the field of cement slurry mix proportion control, conventional practices typically rely on engineers' on-site experience and pre-set fixed formulas. Before construction, a basic mix proportion is determined based on geological survey reports and engineering experience. During construction, key indicators such as slurry fluidity and setting time are manually sampled and tested. The test results are then compared with experienced thresholds or standard ranges. If deviations are found, the proportions of each component are manually adjusted. This static control model, centered on manual experience, cannot respond in real time to the spatial heterogeneity of the geological structure and the temporal fluctuations of the environmental conditions at the construction site, resulting in insufficient adaptability of the mix proportion scheme.

[0003] Furthermore, existing technologies rely on local feedback signals for mix proportion adjustments, lacking a systematic utilization of the physical properties and historical rheological data of the slurry raw materials. Construction sites often accumulate a large number of performance records from different batches of raw materials, but current methods only provide a rough classification and fail to construct quantifiable material characteristic benchmarks based on this data, thus making it impossible to accurately define the feasible boundaries for mix proportion adjustments. When encountering changes in raw material batches or sudden geological conditions, manual adjustments often require multiple trial and error processes, which not only prolongs construction preparation time but also easily leads to deviations in slurry performance from expectations due to inappropriate adjustment magnitudes, increasing material waste and construction quality risks. Summary of the Invention

[0004] This invention provides a method and system for adaptive control of cement slurry mix proportions, which can at least solve some of the problems existing in the prior art.

[0005] A first aspect of the present invention provides a method for adaptive control of cement slurry mix proportions, comprising:

[0006] Geological structure data and environmental status data of the construction site are collected. Spatial layer analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental status data to obtain fluctuation trend characteristics. The scene constraint vector is constructed by combining the intensity distribution characteristics.

[0007] The physical properties and rheological history data of the slurry raw material are obtained and similarity matching is performed to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition calculation is performed to obtain the proportioning boundary space.

[0008] The scenario constraint vector is mapped to the mix proportion boundary space and the constraint is solved to obtain the initial mix proportion scheme. The initial mix proportion scheme is subjected to rheological simulation calculation to obtain the predicted performance curve. Based on the predicted performance curve and the preset construction performance requirements, the deviation quantification analysis is performed to obtain the correction vector. The correction vector is subjected to gradient projection operation to obtain the mix proportion adjustment amount.

[0009] The initial mix proportion scheme is iteratively corrected based on the mix proportion adjustment amount to obtain the target mix proportion scheme. The stability of the target mix proportion scheme is evaluated and the fault tolerance margin is calculated. A segmented control strategy is generated based on the fault tolerance margin and associated with the corresponding construction stage identifier. The mix proportion execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive the mix proportion execution.

[0010] In one alternative implementation,

[0011] Geological structure data and environmental condition data of the construction site are collected. Spatial layering analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental condition data to obtain fluctuation trend characteristics. Combined with the intensity distribution characteristics, a scene constraint vector is constructed, including:

[0012] Geological structure data and environmental status data of the construction site are collected by a distributed sensing unit. Stratigraphic depth information and rock mass type identification are extracted from the geological structure data. Based on the stratigraphic depth information, intervals are divided to obtain a layered depth sequence. The rock mass type identification in each depth interval is mapped to the strength level to obtain a layered strength vector. The gradient of the layered strength vector is calculated to identify stress concentration areas and mark the area boundaries to obtain the strength distribution characteristics.

[0013] Temperature and humidity time series data are extracted from the environmental state data, and sliding window sampling is performed to calculate the statistical features within the window to obtain the environmental fluctuation sequence. The cross-correlation coefficient between the temperature and humidity time series data is calculated and eigenvalue decomposition is performed to obtain a set of feature vectors. Based on the environmental fluctuation sequence, the set of feature vectors is filtered to obtain the dominant mode vector and used as the fluctuation trend feature.

[0014] The intensity distribution features are extended according to the layered depth sequence to obtain a spatial feature tensor, and tensor cross product operation is performed with the fluctuation trend features to obtain a spatiotemporal correlation tensor. Based on the region boundary of the stress concentration region, the spatiotemporal correlation tensor is divided into multiple sub-tensors, and singular value decomposition is performed to extract principal components. The principal components are spliced ​​along the depth dimension to reconstruct the scene constraint vector.

[0015] In one alternative implementation,

[0016] The material reference features are obtained by acquiring the physical properties and rheological history data of the slurry raw materials and performing similarity matching. Based on the material reference features, a component correlation matrix is ​​established and decomposed to obtain the proportioning boundary space, including:

[0017] The physical property parameters of each raw material component are extracted from the slurry raw material library. The mutual information coefficient between different physical property parameters is calculated and cluster analysis is performed to obtain a subset of key physical property parameters. Historical mixing schemes are extracted from pre-stored rheological history data. Based on the subset of key physical property parameters, historical mixing schemes similar to the current physical property parameters in the rheological history data are screened and the corresponding component ratio sequences are extracted and feature-encoded to obtain a mixing feature vector. The distance metric between the physical property parameters of the current construction scenario and the physical property parameters of the scenario corresponding to the historical mixing scheme is calculated and the mixing feature vector is weighted and fused to obtain the material benchmark features.

[0018] The proportions of each group in the material reference features are arranged in a matrix according to the component type to obtain the proportion matrix. The strength index and viscosity index corresponding to the material reference features are extracted to construct the performance matrix and cross-correlation calculation is performed with the proportion matrix to obtain the component correlation matrix.

[0019] The component correlation matrix is ​​subjected to singular value decomposition to extract left and right singular vectors. The component synergy is calculated based on the left singular vectors, and the right singular vectors are filtered to retain the dominant combination pattern. The singular values ​​corresponding to the dominant combination pattern are extracted, and the boundary constraints are constructed based on the singular values ​​and the dominant combination pattern to obtain the matching boundary space.

[0020] In one alternative implementation,

[0021] The initial mix design is obtained by mapping the scenario constraint vector to the mix design boundary space and solving for the constraints. The predicted performance curve is then obtained by performing rheological simulation on the initial mix design, including:

[0022] Projection operation is performed on the scene constraint vector and the dominant combination mode in the ratio boundary space to obtain a projection coefficient set. Based on the projection coefficient set and the pre-acquired boundary constraint conditions, the constraint satisfaction interval is determined. Discrete sampling is performed on the ratio dimension of each group within the constraint satisfaction interval to obtain a candidate ratio set. The Euclidean distance between each candidate ratio point in the candidate ratio set and the scene constraint vector is calculated, and the candidate ratio point with the smallest Euclidean distance is selected. The selected candidate ratio point is consistent with the pre-acquired component synergy degree and adjusted to obtain the initial ratio scheme.

[0023] The correlation coefficients corresponding to the initial proportioning scheme are extracted from the component correlation matrix. Element-wise multiplication of each component proportion in the initial proportioning scheme with its corresponding correlation coefficient is performed to obtain the component interaction matrix. The component interaction matrix is ​​iteratively calculated according to the time step to obtain the state evolution matrix. Shear stress components and flow rate components are extracted from the state evolution matrix and the viscosity evolution sequence is calculated. The hydration reaction progress and the corresponding strength development sequence are calculated based on the cement component proportions in the initial proportioning scheme. The viscosity evolution sequence and the strength development sequence are respectively fitted with time axis curves to obtain viscosity curves and strength curves. The data are then stitched together according to the time dimension to obtain the predicted performance curve.

[0024] In one alternative implementation,

[0025] Based on the predicted performance curve and the preset construction performance requirements, a deviation quantification analysis is performed to obtain a correction vector. Gradient projection calculation is then performed on the correction vector to obtain the mix proportion adjustment amount, including:

[0026] Viscosity and strength curves are extracted from the predicted performance curves and discretely sampled according to time nodes to obtain viscosity sampling sequences and strength sampling sequences. Target viscosity and target strength sequences are extracted from preset construction performance requirements, and point-by-point deviations between these sequences and the viscosity and strength sampling sequences are calculated to obtain viscosity deviation sequences and strength deviation sequences. Deviation tensors are constructed according to construction stages and time nodes. Tucker decomposition is performed on the deviation tensors to extract core tensors and factor matrices. Based on the factor matrices, stage coupling coefficients for construction stages are calculated, and the core tensors are reconstructed to obtain comprehensive deviation sequences. Dimensional mapping is performed on the comprehensive deviation sequences according to the component types in the initial mix design to obtain correction vectors.

[0027] Extract the set of normal vectors corresponding to the boundary constraints from the ratio boundary space and orthogonalize them to obtain the constraint space basis vectors. Project the correction vector onto the constraint space basis vectors to obtain the tangential component. Calculate the inner product between the tangential component and the pre-acquired component synergy degree and construct synergy consistency constraints to correct the tangential component to obtain the feasible correction component. Calculate the step size factor based on the boundary distance between the feasible correction component and the constraint satisfaction interval and scale the feasible correction component to obtain the ratio adjustment amount.

[0028] In one alternative implementation,

[0029] The initial proportioning scheme is iteratively corrected based on the aforementioned proportioning adjustment to obtain the target proportioning scheme. The stability assessment of the target proportioning scheme yields the fault tolerance margin, including:

[0030] The modified proportioning scheme is obtained by performing vector addition on the adjustment amount and the initial proportioning scheme. The modified proportioning scheme is then subjected to rheological simulation to obtain the modified predicted performance curve. The deviation between the modified and the predicted performance curve is calculated to obtain the performance improvement degree. The ratio of the performance improvement degree to the modulus of the adjustment amount is calculated to obtain the convergence index. When the convergence index is greater than the preset convergence threshold, the modified proportioning scheme is taken as the target proportioning scheme. When the convergence index is less than or equal to the preset convergence threshold, the deviation quantization analysis and gradient projection operation are re-executed on the modified proportioning scheme and iteratively corrected until the convergence index is greater than the preset convergence threshold to obtain the target proportioning scheme.

[0031] Extract each allocation ratio from the target allocation scheme to construct an allocation vector. Calculate the minimum distance between the allocation vector and the boundary based on the allocation boundary space to obtain a boundary margin vector. Perform perturbation sampling on the allocation vector to generate a perturbation allocation set and perform rheological simulation to obtain a perturbation performance curve set. Calculate the variance of the deviation distribution between the perturbation performance curve set and the corrected predicted performance curve as a performance fluctuation index. Calculate the tolerance margin based on the boundary margin vector and the performance fluctuation index.

[0032] In one alternative implementation,

[0033] Based on the fault tolerance margin, a segmented control strategy is generated and associated with the corresponding construction stage identifier. The proportioning execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive proportioning execution, including:

[0034] Based on the fault tolerance margin, the target mix design is decomposed into intervals to obtain the mix ratio fluctuation range. The mix ratio fluctuation range and the layer depth sequence are correlated and mapped to obtain the mix ratio constraint interval for each construction stage. The mix ratio constraint interval between adjacent construction stages is continuously constrained to generate a segmented control strategy. Each mix ratio segment in the segmented control strategy is marked with a construction stage identifier.

[0035] The mixing scheme corresponding to each construction stage is extracted from the segmented control strategy, and the component feeding amount is calculated to obtain the benchmark feeding parameters. The dominant mode vector is extracted from the fluctuation trend characteristics, and the projection coefficient between the current environmental state data and the dominant mode vector is calculated to obtain the environmental offset. The benchmark feeding parameters are dynamically compensated based on the environmental offset to obtain the real-time feeding parameters. The fault tolerance margin is converted into the feeding amount tolerance and superimposed with the real-time feeding parameters to obtain the mixing execution parameters. The mixing execution parameters are sent to the material supply control unit to drive the mixing execution.

[0036] A second aspect of the present invention provides a cement slurry mix proportion adaptive control system, comprising:

[0037] The scene constraint unit is used to collect geological structure data and environmental status data at the construction site, perform spatial layering analysis on the geological structure data to obtain intensity distribution characteristics, perform temporal correlation analysis on the environmental status data to obtain fluctuation trend characteristics, and combine the intensity distribution characteristics to construct a scene constraint vector.

[0038] The proportioning boundary unit is used to obtain the physical property parameters and rheological history data of the slurry raw material and perform similarity matching to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition operation is performed to obtain the proportioning boundary space.

[0039] The mix proportion correction unit is used to map the scenario constraint vector to the mix proportion boundary space and solve the constraints to obtain an initial mix proportion scheme, perform rheological simulation calculations on the initial mix proportion scheme to obtain a predicted performance curve, perform deviation quantification analysis based on the predicted performance curve and preset construction performance requirements to obtain a correction vector, and perform gradient projection calculations on the correction vector to obtain the mix proportion adjustment amount.

[0040] The strategy execution unit is used to iteratively correct the initial proportion scheme based on the proportion adjustment amount to obtain the target proportion scheme, perform stability assessment calculation on the target proportion scheme to obtain the fault tolerance margin, generate a segmented control strategy based on the fault tolerance margin and associate it with the corresponding construction stage identifier, determine the proportion execution parameters of each construction stage according to the segmented control strategy and send them to the material supply control unit to drive the proportion execution.

[0041] A third aspect of the present invention provides an electronic device, comprising:

[0042] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0043] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0044] In this invention, by performing spatial layering analysis on geological data and temporal correlation analysis on environmental data, a scenario constraint vector is constructed to ensure that the mix design closely matches the actual working conditions, avoiding subjective deviations from empirical mix proportions. Multiple iterative corrections and gradient projection calculations ensure that the rheological properties of the final target mix proportion accurately meet construction requirements, significantly reducing construction risks such as pipe blockage and segregation caused by improper mix proportions. Based on the decomposition calculation of material baseline characteristics and component correlation matrices, a scientifically reasonable mix boundary space is constructed, effectively limiting the search range of invalid mix proportion schemes and reducing the number of trial mixes and raw material waste. Combining rheological simulation and deviation quantification analysis, the initial mix proportion scheme is quickly corrected and the mix proportion adjustment is output, significantly shortening the on-site debugging time. The adaptive control mechanism throughout the process minimizes manual intervention, improving construction efficiency and control precision. A stability assessment of the target mix proportion scheme is performed, the fault tolerance margin is calculated, and a segmented control strategy is generated accordingly, enabling flexible responses to environmental disturbances such as geological changes and temperature fluctuations, ensuring consistent grout quality throughout the grouting process. Attached Figure Description

[0045] Figure 1 This is a schematic flowchart of the adaptive control method for cement slurry mix proportion according to an embodiment of the present invention;

[0046] Figure 2 This is a flowchart illustrating the generation and stability evaluation of the target mix ratio scheme for the adaptive control method of cement slurry mix ratio in an embodiment of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0049] Figure 1 This is a schematic flowchart of the adaptive control method for cement slurry mix proportions according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0050] Geological structure data and environmental status data of the construction site are collected. Spatial layer analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental status data to obtain fluctuation trend characteristics. The scene constraint vector is constructed by combining the intensity distribution characteristics.

[0051] The physical properties and rheological history data of the slurry raw material are obtained and similarity matching is performed to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition calculation is performed to obtain the proportioning boundary space.

[0052] The scenario constraint vector is mapped to the mix proportion boundary space and the constraint is solved to obtain the initial mix proportion scheme. The initial mix proportion scheme is subjected to rheological simulation calculation to obtain the predicted performance curve. Based on the predicted performance curve and the preset construction performance requirements, the deviation quantification analysis is performed to obtain the correction vector. The correction vector is subjected to gradient projection operation to obtain the mix proportion adjustment amount.

[0053] The initial mix proportion scheme is iteratively corrected based on the mix proportion adjustment amount to obtain the target mix proportion scheme. The stability of the target mix proportion scheme is evaluated and the fault tolerance margin is calculated. A segmented control strategy is generated based on the fault tolerance margin and associated with the corresponding construction stage identifier. The mix proportion execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive the mix proportion execution.

[0054] In one alternative implementation,

[0055] Geological structure data and environmental condition data of the construction site are collected. Spatial layering analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental condition data to obtain fluctuation trend characteristics. Combined with the intensity distribution characteristics, a scene constraint vector is constructed, including:

[0056] Geological structure data and environmental status data of the construction site are collected by a distributed sensing unit. Stratigraphic depth information and rock mass type identification are extracted from the geological structure data. Based on the stratigraphic depth information, intervals are divided to obtain a layered depth sequence. The rock mass type identification in each depth interval is mapped to the strength level to obtain a layered strength vector. The gradient of the layered strength vector is calculated to identify stress concentration areas and mark the area boundaries to obtain the strength distribution characteristics.

[0057] Temperature and humidity time series data are extracted from the environmental state data, and sliding window sampling is performed to calculate the statistical features within the window to obtain the environmental fluctuation sequence. The cross-correlation coefficient between the temperature and humidity time series data is calculated and eigenvalue decomposition is performed to obtain a set of feature vectors. Based on the environmental fluctuation sequence, the set of feature vectors is filtered to obtain the dominant mode vector and used as the fluctuation trend feature.

[0058] The intensity distribution features are extended according to the layered depth sequence to obtain a spatial feature tensor, and tensor cross product operation is performed with the fluctuation trend features to obtain a spatiotemporal correlation tensor. Based on the region boundary of the stress concentration region, the spatiotemporal correlation tensor is divided into multiple sub-tensors, and singular value decomposition is performed to extract principal components. The principal components are spliced ​​along the depth dimension to reconstruct the scene constraint vector.

[0059] Geological structure data and environmental status data of the construction site are collected by distributed sensing units. These units are arranged according to a pre-defined spatial grid within the construction area, covering different depth layers and surface environmental monitoring points. They can simultaneously collect continuous time-series records of stratigraphic profile information, rock mass mechanical property identifiers, and environmental parameters such as temperature and humidity. The geological structure data includes core analysis results collected along the borehole depth direction, stratigraphic interface delineation information, and rock mass type identifiers for each stratum. The environmental status data records time-series temperature and humidity data of the construction area at a fixed sampling period. Both types of data are aligned at the timestamp level, providing a foundation for subsequent joint analysis.

[0060] After extracting stratigraphic depth information and rock mass type identifiers from geological structural data, the depth axis is divided into intervals based on natural dividing points of stratigraphic interfaces, forming a layered depth sequence. Each depth interval corresponds to an independent geological unit, and the interval boundary is determined by the lithological abrupt change location of adjacent rock layers, ensuring the homogeneity of rock mass types within the same interval. Strength level mapping is performed on the rock mass type identifiers within each depth interval. The mapping rules convert rock types into quantified strength level values ​​according to rock mass classification standards, thus transforming discrete lithological identifiers into calculable layered strength vectors. The layered strength vectors are arranged along the depth direction, reflecting the strength variation pattern from shallow to deep layers in the construction area. Gradient calculation is performed on the layered strength vectors, i.e., the strength level difference between adjacent depth intervals is calculated layer by layer. Locations where the absolute value of the gradient exceeds a preset threshold are identified as stress concentration areas. The upper and lower boundaries of these locations are marked to form a set of regional boundaries, which, together with the layered strength vectors, constitute the strength distribution characteristics. The identification of stress concentration areas is of great significance for subsequent mix design control, because the requirements for grouting pressure and grout permeability in these areas differ significantly from those in conventional areas.

[0061] After extracting temperature and humidity time-series data from the environmental condition data, a sliding window is used to sample the two sets of time-series data separately. The window length is set according to the construction shift cycle, and an overlap area of ​​a certain step size is set between adjacent windows to ensure the continuity of the time series. Within each window, statistical characteristics such as mean, standard deviation, and range are calculated. The statistical characteristics of each window are arranged in chronological order to obtain an environmental fluctuation sequence reflecting the dynamic changes of environmental parameters. The environmental fluctuation sequence includes temperature fluctuation subsequences and humidity fluctuation subsequences, which respectively describe the fluctuation amplitude and rhythm of temperature and humidity along the construction time axis.

[0062] When calculating the cross-correlation coefficients between temperature and humidity time-series data, Pearson correlation coefficients are calculated for each time lag, yielding cross-correlation function curves. The cross-correlation coefficients are arranged by time lag to form a cross-correlation matrix. Eigenvalue decomposition is then performed on this matrix to obtain a set of eigenvalues ​​and corresponding eigenvectors. Each vector in the eigenvector set represents a potential mode of joint temperature and humidity variation, and the magnitude of the eigenvalue reflects the contribution of that mode to the overall variance. Based on the environmental fluctuation sequence, the eigenvector set is filtered, and the projection coefficient between each eigenvector and the environmental fluctuation sequence is calculated. Several eigenvectors with larger absolute projection coefficient values ​​are retained, as the modes corresponding to these vectors can explain the main changing components in the environmental fluctuation sequence. These are then used as dominant mode vectors, constituting the fluctuation trend features. The selection process for dominant mode vectors effectively reduces the redundant dimensionality of the environmental data, allowing the fluctuation trend features to focus on the temperature and humidity coupling variation patterns that have the most significant impact on slurry performance.

[0063] The intensity distribution characteristics are extended according to the layered depth sequence. Specifically, the intensity level value corresponding to each depth interval in the layered intensity vector is copied and expanded along the spatial dimension to form a spatial feature tensor corresponding to the spatial coordinate system of the construction area. The resolution of this tensor in the depth dimension is consistent with the number of intervals in the layered depth sequence. A tensor outer product operation is then performed between the spatial feature tensor and the fluctuation trend characteristics. This outer product operation multiplies each element of the spatial feature tensor with each component of the fluctuation trend feature vector, generating a higher-dimensional spatiotemporal correlation tensor. This spatiotemporal correlation tensor simultaneously encodes the spatial intensity distribution information of the construction area and the temporal modal information of environmental fluctuations, enabling it to express the comprehensive constraint characteristics of different depth intervals under different environmental conditions.

[0064] Based on the regional boundaries of stress concentration areas, the spatiotemporal correlation tensor is segmented into multiple sub-tensors along the depth dimension. Each sub-tensor corresponds to a geological unit defined by the regional boundary. Singular value decomposition (SVD) is performed on each sub-tensor to extract principal components. The number of principal components is determined by the cumulative contribution rate of the singular values, retaining the minimum number of singular vectors required for the cumulative contribution rate to reach a preset proportion. SVD decomposes the sub-tensor into the product of a left singular vector, a diagonal matrix of singular values, and a right singular vector. The column vectors corresponding to the largest singular values ​​in the left singular vector are taken as the principal components of the sub-tensor. These principal components compactly represent the spatiotemporal coupling information contained in the sub-tensor in a low-dimensional space.

[0065] The principal components extracted from each sub-tensor are concatenated and reconstructed along the depth dimension. The principal component vectors of each sub-tensor are arranged sequentially according to the layered depth sequence, forming a continuous high-dimensional vector, namely the scene constraint vector. The scene constraint vector structurally preserves the spatial hierarchy of the construction area from shallow to deep layers, while embedding corresponding environmental fluctuation mode information at each depth layer, enabling it to comprehensively characterize the spatiotemporal constraints of the construction site. The dimension of the scene constraint vector is jointly determined by the number of intervals in the layered depth sequence, the number of principal components in each sub-tensor, and the dimension of the dominant mode vector. This achieves effective compression and structured representation of the original multi-source data while ensuring information integrity, providing a unified input format for subsequently mapping scene constraints to the ratio boundary space and solving for constraints.

[0066] In one alternative implementation,

[0067] The material reference features are obtained by acquiring the physical properties and rheological history data of the slurry raw materials and performing similarity matching. Based on the material reference features, a component correlation matrix is ​​established and decomposed to obtain the proportioning boundary space, including:

[0068] The physical property parameters of each raw material component are extracted from the slurry raw material library. The mutual information coefficient between different physical property parameters is calculated and cluster analysis is performed to obtain a subset of key physical property parameters. Historical mixing schemes are extracted from pre-stored rheological history data. Based on the subset of key physical property parameters, historical mixing schemes similar to the current physical property parameters in the rheological history data are screened and the corresponding component ratio sequences are extracted and feature-encoded to obtain a mixing feature vector. The distance metric between the physical property parameters of the current construction scenario and the physical property parameters of the scenario corresponding to the historical mixing scheme is calculated and the mixing feature vector is weighted and fused to obtain the material benchmark features.

[0069] The proportions of each group in the material reference features are arranged in a matrix according to the component type to obtain the proportion matrix. The strength index and viscosity index corresponding to the material reference features are extracted to construct the performance matrix and cross-correlation calculation is performed with the proportion matrix to obtain the component correlation matrix.

[0070] The component correlation matrix is ​​subjected to singular value decomposition to extract left and right singular vectors. The component synergy is calculated based on the left singular vectors, and the right singular vectors are filtered to retain the dominant combination pattern. The singular values ​​corresponding to the dominant combination pattern are extracted, and the boundary constraints are constructed based on the singular values ​​and the dominant combination pattern to obtain the matching boundary space.

[0071] Physical property parameters of each raw material component are extracted from the slurry raw material library, including basic physicochemical indicators such as cement fineness, specific surface area, initial heat of hydration, particle size distribution, density, and water absorption. For the extracted multidimensional physical property parameters, the mutual information coefficient between any two physical property parameters is calculated to quantify the nonlinear dependence between the parameters. Let the... The first physical property parameter and the first The mutual information coefficient between the individual physical property parameters is By examining all parameter pairs A mutual information matrix is ​​constructed, and then hierarchical cluster analysis is performed on the matrix to group parameters with high mutual information coefficients into the same cluster. From each cluster, the most representative parameters are retained, and finally, a subset of key physical property parameters is selected. The core significance of this step is to eliminate the interference of redundant parameters on subsequent matching calculations, while retaining the physical property dimensions that have the most significant impact on slurry performance.

[0072] Historical formulation schemes are extracted from pre-stored rheological history data. Using a subset of key physical property parameters as the selection criterion, the distance metric between the current raw material's key physical property parameter vector and the corresponding raw material's key physical property parameter vectors in the historical formulation database is calculated. The distance metric uses a weighted Euclidean distance. Let the current scenario be... The key physical property parameters are as follows: The corresponding parameter value in the historical scheme is The weight is Then the distance metric for Distance metric The smaller the value, the closer the historical mixing ratio is to the raw material condition in the current construction scenario. (Based on distance metric) Sort the historical pairing schemes from smallest to largest, select the ones with the highest similarity, extract the pairing scheme sequences corresponding to these schemes, and encode the features of each pairing scheme sequence to map the discrete pairing values ​​into continuous pairing feature vectors for subsequent vector space fusion operations.

[0073] For each historical allocation scheme obtained through screening, based on the distance metric... The corresponding fusion weights are calculated, with higher weights assigned to historical schemes that are closer in distance. The proportion feature vectors of each historical scheme are then weighted and fused to obtain material benchmark features that comprehensively reflect the characteristics of raw materials in the current construction scenario. The material benchmark features not only include the typical proportion ranges of each component, but also the performance response patterns of this type of raw material under different proportion conditions in historical construction, providing a data foundation for the subsequent construction of the component correlation matrix.

[0074] The proportions of each group in the material's baseline characteristics are arranged in a matrix according to component type to obtain a proportion matrix. Rows in the proportion matrix correspond to different historical proportion samples, and columns correspond to different raw material components (such as cement, fly ash, water-reducing agent, water, etc.). Each element in the matrix represents the mass percentage or dosage ratio of the corresponding component in the corresponding sample. Simultaneously, strength indices (such as 28-day compressive strength and early strength development rate) and viscosity indices (such as initial viscosity and plastic viscosity) corresponding to each historical proportion sample are extracted from the material's baseline characteristics to construct a performance matrix. The rows of the performance matrix also correspond to historical proportion samples, and the columns correspond to different performance indices. Cross-correlation calculations are performed on the proportion matrix and performance matrix, i.e., the correlation coefficients between each column (component) of the proportion matrix and each column (performance index) of the performance matrix are calculated to obtain a component correlation matrix. The elements in the component correlation matrix reflect the degree and direction of the influence of changes in the dosage of specific components on specific performance indices, providing a structured mathematical expression for subsequent decomposition operations.

[0075] Singular value decomposition (SVD) is performed on the component correlation matrix, decomposing it into a product of a left singular matrix, a diagonal matrix of singular values, and a right singular matrix. The left singular vector corresponds to the principal direction in the component space, and the right singular vector corresponds to the principal direction in the performance space. The singular values ​​reflect the strength of the correlation between components and performance along the corresponding direction. Component synergy is calculated based on the left singular vector, i.e., the projection coefficients of each raw material component along the principal direction are analyzed. Components with high synergy indicate that their dosage changes have similar directional effects in the proportioning space and can be considered as objects for synergistic control. When screening the right singular vector, the dominant combination patterns with larger corresponding singular values ​​are retained, while the secondary patterns with smaller singular values ​​are discarded. This approach reduces dimensionality while preserving the component combination patterns that play a decisive role in slurry performance.

[0076] The set of singular values ​​corresponding to the dominant combination patterns is extracted, and boundary constraints for proportioning are established based on the coefficient ranges of each component within the dominant combination pattern. Specifically, for each dominant combination pattern, the allowable variation range of the pattern in the proportioning space is determined according to the magnitude of the singular values. The larger the singular value, the wider the constraint range of the corresponding pattern, allowing the combination pattern to vary within a larger range; dominant patterns with smaller singular values ​​correspond to stricter boundary constraints to prevent proportioning deviations in that direction from causing significant performance degradation. The boundary constraints of all dominant combination patterns are summarized to form a multi-dimensional proportioning boundary space. This proportioning boundary space uses the dominant combination patterns as coordinate axes and the constraint ranges derived from the singular values ​​as boundaries, ensuring that the proportioning schemes are within the physically feasible domain and that proportioning adjustments are always made along the direction with the most significant impact on performance, avoiding wasting adjustment resources on ineffective dimensions.

[0077] The construction of the mix design boundary space provides a structured feasible region definition for subsequent mapping and constraint solving of scenario constraint vectors. The component synergistic relationships revealed by singular value decomposition enable efficient mix design optimization within the dimensionality-reduced dominant mode space. Boundary constraints ensure the engineering feasibility and performance reliability of the final mix design. In actual construction, the physical properties of different batches of raw materials may fluctuate. By re-executing the similarity matching and weighted fusion process, the material baseline features can be dynamically updated, thereby reconstructing the component correlation matrix and updating the mix design boundary space. This achieves an adaptive response to batch differences in raw materials, ensuring the stability of slurry performance throughout the entire construction cycle.

[0078] In one alternative implementation,

[0079] The initial mix design is obtained by mapping the scenario constraint vector to the mix design boundary space and solving for the constraints. The predicted performance curve is then obtained by performing rheological simulation on the initial mix design, including:

[0080] Projection operation is performed on the scene constraint vector and the dominant combination mode in the ratio boundary space to obtain a projection coefficient set. Based on the projection coefficient set and the pre-acquired boundary constraint conditions, the constraint satisfaction interval is determined. Discrete sampling is performed on the ratio dimension of each group within the constraint satisfaction interval to obtain a candidate ratio set. The Euclidean distance between each candidate ratio point in the candidate ratio set and the scene constraint vector is calculated, and the candidate ratio point with the smallest Euclidean distance is selected. The selected candidate ratio point is consistent with the pre-acquired component synergy degree and adjusted to obtain the initial ratio scheme.

[0081] The correlation coefficients corresponding to the initial proportioning scheme are extracted from the component correlation matrix. Element-wise multiplication of each component proportion in the initial proportioning scheme with its corresponding correlation coefficient is performed to obtain the component interaction matrix. The component interaction matrix is ​​iteratively calculated according to the time step to obtain the state evolution matrix. Shear stress components and flow rate components are extracted from the state evolution matrix and the viscosity evolution sequence is calculated. The hydration reaction progress and the corresponding strength development sequence are calculated based on the cement component proportions in the initial proportioning scheme. The viscosity evolution sequence and the strength development sequence are respectively fitted with time axis curves to obtain viscosity curves and strength curves. The data are then stitched together according to the time dimension to obtain the predicted performance curve.

[0082] The scene constraint vector carries information on the geological intensity distribution and environmental fluctuations at the construction site, while the mix design boundary space describes the feasible domain structure of each allocation ratio in the form of dominant combination patterns. When mapping the scene constraint vector to the mix design boundary space, an inner product operation is performed on the scene constraint vector and each dominant combination pattern vector extracted from the mix design boundary space to obtain the projection coefficient of the scene constraint vector in each dominant combination pattern direction. All projection coefficients together constitute the projection coefficient set. This set reflects the activation intensity of each dominant combination pattern by the current construction scene; the larger the projection coefficient, the higher the degree of matching between the corresponding combination pattern and the current scene.

[0083] Based on the projection coefficient set and combined with pre-acquired boundary constraints (including physical upper and lower limits of each group's allocation ratio, total normalization constraints, and proportional constraints between specific components), the constraint satisfaction interval is determined in the projection coefficient space. The process for determining the constraint satisfaction interval is as follows: the component range corresponding to each coefficient in the projection coefficient set is compared one by one with the boundary constraints, and the coefficient interval segment that simultaneously satisfies all constraints is retained, forming a feasible interval after multi-dimensional cross-constraints. If the projection coefficient of a dominant combination mode exceeds the boundary constraint range, it is truncated to the nearest constraint boundary value to ensure that subsequent sampling processes are always conducted within the legal allocation domain.

[0084] Within the constraint-satisfied interval, discrete sampling is performed on each component's proportion dimension. The sampling step size is determined based on the minimum adjustable precision of each component in actual construction. For example, the water-cement ratio dimension can be uniformly discretized with a step size of 0.02, and the admixture dosage dimension can be discretized with a step size of 0.1%, thereby generating a candidate proportion set covering each component's proportion dimension within the constraint-satisfied interval. Each element in the candidate proportion set is called a candidate proportion point, representing a specific combination of component dosages.

[0085] Calculate the Euclidean distance between each candidate matching point in the candidate matching set and the scene constraint vector. ,in Indicates the first The Euclidean distance from each candidate matching point to the scene constraint vector. This refers to the index number of the candidate matching points. (The process involves) filtering out... The candidate mix designation point with the smallest Euclidean distance is selected as the initial mix designation result that best matches the current construction scenario. This selection logic is based on the fact that the candidate mix designation point with the smallest Euclidean distance is the closest to the geometric distance of the scenario constraint vector in the mix designation parameter space, which means that its mix designation characteristics best meet the requirements of the current scenario.

[0086] The selected candidate formulations need to be validated against the pre-obtained component synergy. Component synergy describes the mutually promoting or inhibiting relationships between different components in hydration reactions, rheological behavior, etc., and is stored in the form of a synergy matrix. During the validation process, the dosage of each component at the candidate formulation is substituted into the synergy matrix to calculate the synergistic response value between each component pair. If the synergistic response value of a component pair is lower than the preset synergistic threshold, the component pair that deviates significantly from the synergistic range is fine-tuned. The adjustment direction is to move closer to the optimal synergy point, and the adjustment magnitude does not exceed twice the discrete sampling step size of that component, to ensure that the adjusted formulation remains within the constraint-satisfied range. After consistency validation and adjustment, the initial formulation scheme is finally determined.

[0087] The correlation coefficients corresponding to the initial formulation scheme are extracted from the component correlation matrix. The component correlation matrix records the interaction strength between components at the physicochemical level. The correlation coefficients are extracted by locating and reading the coefficient values ​​at the corresponding row and column positions in the component correlation matrix based on the type index of each component in the initial formulation scheme, forming a set of correlation coefficients matching the dimensions of the initial formulation scheme. The element-wise product operation is then performed between the formulation value vector of each component in the initial formulation scheme and the corresponding correlation coefficient vector to obtain the component interaction matrix. Each element of this matrix... Indicates the first The component and the first The strength of the interaction between components is determined by the ratio of the two components and their correlation coefficient, and can comprehensively reflect the degree of mutual influence of different components in the slurry system.

[0088] The component interaction matrix is ​​iteratively calculated according to time steps to simulate the state changes of the slurry during stirring, pumping, and grouting. Within each time step, the component interaction matrix at the current moment is used as input, and environmental parameters such as temperature and pressure are combined to update the matrix elements, obtaining the state matrix for the next moment. This process is repeated until the preset total simulation time is reached, ultimately forming the state evolution matrix. The row dimension of the state evolution matrix corresponds to the time step, and the column dimension corresponds to the state variables of each component.

[0089] Shear stress and flow rate components are extracted from the state evolution matrix. The shear stress component reflects the grout's ability to resist deformation during flow, while the flow rate component reflects the actual flow rate of the grout under grouting pressure. Based on these two components, the viscosity evolution sequence is calculated according to fundamental rheological relationships: at each time step, the shear stress value at that moment is divided by the flow rate value to obtain the apparent viscosity at that moment. The apparent viscosity values ​​at all moments are arranged in chronological order to form the viscosity evolution sequence. This sequence fully records the dynamic change of viscosity over time in the grout from initial mixing to the end of grouting.

[0090] Based on the cement composition ratio in the initial mix design, the hydration reaction progress is calculated. The calculation of the hydration reaction progress is based on a cement hydration kinetic model, using cement dosage, water-cement ratio, and ambient temperature as input parameters. The degree of hydration is calculated step-by-step over time, and the strength value at each time point is obtained by mapping the empirical relationship between the degree of hydration and strength. The strength values ​​at all times are arranged in chronological order to form a strength development sequence. This strength development sequence describes the complete process of the grout from initial setting to final setting and then to the later strength increase, reflecting the influence of different cement dosages and water-cement ratios on the mechanical properties of the grout solidification.

[0091] Time-axis curve fitting was performed on both the viscosity evolution sequence and the intensity development sequence. For the viscosity evolution sequence, polynomial or piecewise linear fitting methods were used to fit discrete viscosity time points into a continuous viscosity curve. For the intensity development sequence, a logarithmic growth function or power function was used for fitting to obtain a continuous intensity curve. During the fitting process, the optimal fitting parameters were automatically selected with the goal of minimizing the sum of squared residuals, ensuring that the accuracy of the fitted curve in restoring the original sequence met a preset error threshold.

[0092] Both the viscosity and strength curves are plotted on the horizontal axis using time, but their vertical axes use different physical quantities, corresponding to viscosity units and strength units, respectively. The two curves are stitched together along the time dimension, merging them into a single multi-channel predictive performance curve on a unified time axis. Each time point on the predictive performance curve contains information from both the predicted viscosity and strength values, comprehensively characterizing the evolution of the slurry's rheological and mechanical properties throughout the entire construction period under the initial mix design. This provides a unified data carrier for subsequent quantitative analysis of deviations from construction performance requirements.

[0093] In one alternative implementation,

[0094] Based on the predicted performance curve and the preset construction performance requirements, a deviation quantification analysis is performed to obtain a correction vector. Gradient projection calculation is then performed on the correction vector to obtain the mix proportion adjustment amount, including:

[0095] Viscosity and strength curves are extracted from the predicted performance curves and discretely sampled according to time nodes to obtain viscosity sampling sequences and strength sampling sequences. Target viscosity and target strength sequences are extracted from preset construction performance requirements, and point-by-point deviations between these sequences and the viscosity and strength sampling sequences are calculated to obtain viscosity deviation sequences and strength deviation sequences. Deviation tensors are constructed according to construction stages and time nodes. Tucker decomposition is performed on the deviation tensors to extract core tensors and factor matrices. Based on the factor matrices, stage coupling coefficients for construction stages are calculated, and the core tensors are reconstructed to obtain comprehensive deviation sequences. Dimensional mapping is performed on the comprehensive deviation sequences according to the component types in the initial mix design to obtain correction vectors.

[0096] Extract the set of normal vectors corresponding to the boundary constraints from the ratio boundary space and orthogonalize them to obtain the constraint space basis vectors. Project the correction vector onto the constraint space basis vectors to obtain the tangential component. Calculate the inner product between the tangential component and the pre-acquired component synergy degree and construct synergy consistency constraints to correct the tangential component to obtain the feasible correction component. Calculate the step size factor based on the boundary distance between the feasible correction component and the constraint satisfaction interval and scale the feasible correction component to obtain the ratio adjustment amount.

[0097] After extracting the viscosity and strength curves from the predicted performance curves, discrete sampling is performed according to the time nodes in the construction process. The continuous curves are transformed into numerical sequences at a finite number of time points, resulting in viscosity and strength sampling sequences. The time nodes are selected based on the construction stage division, with uniform or non-uniform distribution. Sampling is intensified at key construction nodes (such as the initial setting stage and the final setting stage) to improve the accuracy of deviation capture. Simultaneously, target viscosity and target strength sequences are extracted from the preset construction performance requirements at the corresponding time nodes, ensuring strict alignment of the two sets of sequences on the time axis. The viscosity deviation sequence is obtained by subtracting the target viscosity sequence from the viscosity sampling sequence at each point; similarly, the strength deviation sequence is obtained by subtracting the target strength sequence from the strength sampling sequence at each point. The point-by-point deviation is calculated by subtracting the target value from the current predicted value, retaining the sign information to reflect the direction of performance deviation (higher or lower).

[0098] After obtaining the viscosity and strength deviation sequences, the two types of deviation sequences are organized into third-order deviation tensors according to the construction stage and time node dimensions. The three dimensions correspond to the performance type (viscosity and strength), construction stage number, and time node number, respectively. The construction of the deviation tensor unifies the deviation information of multiple stages and multiple performance indicators into a single data structure, facilitating subsequent cross-stage and cross-performance indicator coupled analysis. Tucker decomposition is performed on the constructed deviation tensor, decomposing it into a product of a low-order core tensor and three factor matrices, i.e. ,in For the core tensor, , , These are the factor matrices corresponding to the three dimensions. Tucker decomposition is solved iteratively using alternating least squares or higher-order singular value decomposition. The rank parameter of the core tensor is pre-set based on the number of major changing components of the bias tensor, and is usually chosen as the minimum rank combination that can cover more than 95% of the cumulative variance.

[0099] Based on the factor matrix obtained from Tucker decomposition, the stage coupling coefficients between construction stages are calculated. These coefficients reflect the degree of interrelationship among different construction stages in terms of deviation structure, specifically determined by analyzing the factor matrices corresponding to each construction stage. The cosine similarity between the column vectors is normalized to obtain the coupling coefficient. A higher coupling coefficient indicates more similar deviation patterns between adjacent construction stages, requiring coordinated adjustment rather than independent processing in subsequent corrections. The stage coupling coefficient is used to adjust the core tensor. A weighted reconstruction is performed, embedding the coupling coefficient as a weighting factor for the construction stage dimension into the reconstruction process, resulting in a comprehensive deviation sequence that reflects the cross-stage coupling effect. The comprehensive deviation sequence is expanded around time nodes, and the comprehensive deviation value at each time node integrates the performance deviations of viscosity and strength, as well as the coupling effects between stages, thus exhibiting higher information completeness compared to processing viscosity deviation or strength deviation separately.

[0100] The overall deviation sequence is dimensionally mapped according to the component types in the initial formulation scheme, converting the deviation information in the time domain into a correction direction vector in the component adjustment space, i.e., the correction vector. The dimensional mapping is implemented by establishing a mapping matrix based on the sensitivity relationship between each component and its viscosity and strength properties. , where matrix elements Indicates the first The component is related to the first The sensitivity coefficients of each performance index are obtained through regression analysis of historical experimental data or physicochemical mechanism analysis. Correction vector. From the integrated deviation sequence and mapping matrix Multiplying by the transpose yields, i.e. ,in This is the vectorized representation of the overall deviation sequence. Correction vector. Each component corresponds to the adjustment direction and adjustment intensity of different components.

[0101] After obtaining the correction vector Next, the gradient projection calculation stage begins. The set of normal vectors corresponding to the boundary constraints is extracted from the proportioning boundary space. Each boundary constraint (such as upper and lower limits of component dosage, water-cement ratio range, etc.) corresponds to a normal vector, and the direction of the normal vector points to the infeasible side of the constraint. All normal vectors are then subjected to Gram-Schmidt orthogonalization to obtain the set of basis vectors in the constraint space. ,in To determine the number of effective constraints, orthogonalization is used to eliminate linear dependence between normal vectors, ensuring the numerical stability of subsequent projection decomposition. The correction vector is then... It is decomposed into two parts: a normal component and a tangential component within the constraint space. The normal component runs along the normal direction of the constraint boundary, while the tangential component lies within the tangential plane of the constraint boundary. Tangential component The calculation method is to subtract the sum of the projections onto all basis vectors in the constraint space from the correction vector, that is... ,in This is the index of the constraint space basis vectors. The tangential component preserves the valid adjustment directions in the correction vector that do not violate the constraint boundaries.

[0102] Calculate the tangential component With the pre-acquired component synergy vector The inner product between them yields the collaborative consistency metric. ,in Each component reflects the degree of synergistic cooperation between different components in chemical reactions and physical properties, and is pre-calibrated by material compatibility experiments or mechanism models. If If the adjustment direction of the tangential component is below the preset coordination consistency threshold, it indicates a conflict between the current adjustment direction of the tangential component and the coordination law of the components, requiring correction of the tangential component. The correction method is to adjust the tangential component along the coordination degree vector. The direction is enhanced by projection, and components that are opposite to the synergy degree vector are suppressed, thereby obtaining feasible modified components that satisfy the synergy consistency constraint. The feasible correction measures, while maintaining the effectiveness of the adjustment, ensure that the dosage adjustments of each component conform to the material compatibility rules, and avoid the deterioration of slurry performance due to excessive adjustment of a single component.

[0103] Based on feasible correction components Calculate the distance between it and the boundary of the constraint-satisfied interval to obtain the step size factor. The step size factor is calculated using the maximum feasible step size principle: starting from the current allocation point along the feasible correction component direction, the maximum allowable movement distance to reach the nearest constraint boundary is determined. Step size factor Take as With preset safety margin coefficient The product of, i.e. ,in This is used to ensure that the adjusted proportioning points lie strictly inside the constraint-satisfying interval, rather than just on the boundary. (For feasible correction components) Multiply by step factor Scaling is performed to obtain the final ratio adjustment. Adjustment amount of proportion Each component directly corresponds to the adjustment range of the dosage of each component in the initial proportioning scheme, and can be directly used for subsequent iterative correction calculations. Through the aforementioned deviation quantification and gradient projection calculation process, the complex multi-stage, multi-performance index deviation information is transformed into executable adjustment instructions that meet the proportioning constraints, ensuring the engineering feasibility of the proportioning optimization process and the physical rationality of the adjustment direction.

[0104] Figure 2 This is a flowchart illustrating the generation and stability evaluation of the target mix ratio scheme for the adaptive control method of cement slurry mix ratio in an embodiment of the present invention.

[0105] In one alternative implementation,

[0106] The initial proportioning scheme is iteratively corrected based on the aforementioned proportioning adjustment to obtain the target proportioning scheme. The stability assessment of the target proportioning scheme yields the fault tolerance margin, including:

[0107] The modified proportioning scheme is obtained by performing vector addition on the adjustment amount and the initial proportioning scheme. The modified proportioning scheme is then subjected to rheological simulation to obtain the modified predicted performance curve. The deviation between the modified and the predicted performance curve is calculated to obtain the performance improvement degree. The ratio of the performance improvement degree to the modulus of the adjustment amount is calculated to obtain the convergence index. When the convergence index is greater than the preset convergence threshold, the modified proportioning scheme is taken as the target proportioning scheme. When the convergence index is less than or equal to the preset convergence threshold, the deviation quantization analysis and gradient projection operation are re-executed on the modified proportioning scheme and iteratively corrected until the convergence index is greater than the preset convergence threshold to obtain the target proportioning scheme.

[0108] Extract each allocation ratio from the target allocation scheme to construct an allocation vector. Calculate the minimum distance between the allocation vector and the boundary based on the allocation boundary space to obtain a boundary margin vector. Perform perturbation sampling on the allocation vector to generate a perturbation allocation set and perform rheological simulation to obtain a perturbation performance curve set. Calculate the variance of the deviation distribution between the perturbation performance curve set and the corrected predicted performance curve as a performance fluctuation index. Calculate the tolerance margin based on the boundary margin vector and the performance fluctuation index.

[0109] The adjusted proportions are vector-added with the initial proportions, meaning they are added element-wise along each component dimension to obtain the revised proportions. The amount of each component in the revised proportions is equal to the sum of the amount of the corresponding component in the initial proportions and the proportion adjustment in that dimension. This addition ensures that both the direction and magnitude of the adjustment are fully superimposed onto the initial proportions, forming a new proportioning point with physical interpretability.

[0110] Rheological simulations were performed on the modified mix design. The corrected proportions of each component were input, and the viscosity evolution, yield stress changes, and setting characteristics of the slurry under different shear rates, temperatures, and time conditions were simulated using a slurry rheological constitutive model. The modified predicted performance curves were then output. The differences between the modified and initial predicted performance curves were calculated at each time point, and the weighted norm of the deviations of each performance index was taken as the degree of performance improvement. , The dimensions are consistent with the performance indicators, and the larger the value, the more significant the correction effect.

[0111] Convergence index Defined as the ratio of performance improvement to the modulus of proportion adjustment, i.e. ,in The Euclidean mold length is used to adjust the proportions. This reflects the efficiency of performance improvement brought about by the unit adjustment range, when A larger value indicates that the current adjustment direction is effective and the step size is appropriate; when A smaller value indicates that the adjustment has reached saturation or deviated from its intended direction, requiring a recalculation of the correction vector. (Preset convergence threshold) The performance requirements are predetermined based on the stringency of the construction requirements, and are usually fixed constants within the range of empirical values.

[0112] when When the current modified mixing scheme is deemed to have met the convergence condition, it is directly output as the target mixing scheme and enters the subsequent stability evaluation process. At this point, using the revised mix design as a new starting point, the deviation quantification analysis is re-executed. This involves comparing the revised predicted performance curve with the construction performance requirements, recalculating the comprehensive deviation sequence, and performing gradient projection calculations again to obtain a new round of mix design adjustments. This new adjustment is then added to the current revised mix design to form a new revised mix design, and the process begins the next iteration. Each iteration involves recalculation. until The iteration terminates and outputs the target proportion scheme. A maximum iteration count is set during the iteration process to prevent infinite loops due to numerical oscillations. When the maximum iteration count is reached, the most recent iteration is used. The largest corrected ratio is taken as the target ratio.

[0113] Extract the proportion values ​​of each component from the target formulation scheme, and construct a proportion vector by arranging the components in order. The dimension equals the total number of components involved in the proportioning. The proportioning boundary space is composed of upper and lower bound constraints determined by the component incidence matrix decomposition operation, with each component corresponding to a feasible interval. Calculation The boundary margin vector is obtained by taking the smaller of the distances between each component value and its corresponding upper and lower bounds. The boundary margin components of all components are then arranged in order. . Each component reflects the margin of the target proportion scheme from the constraint boundary in the corresponding component dimension. The smaller the component value, the closer the component is to the constraint boundary and the weaker its tolerance to disturbances.

[0114] For the ratio vector Perturbation sampling is performed using a uniform or Gaussian distribution. Within the neighborhood of the sample, several perturbation points are generated. The perturbation amplitude is set with reference to the typical range of material measurement errors in actual construction. Each perturbation point must satisfy the constraints of the mix proportion boundary space, and sampling points exceeding the boundary are discarded. All legal perturbation points are used to form a perturbation mix proportion set, where each element is a complete component mix proportion vector. Rheological simulation calculations are performed on each perturbation mix proportion scheme in the perturbation mix proportion set to obtain the corresponding performance curves. The performance curves of all perturbation mix proportion schemes together constitute a perturbation performance curve set.

[0115] The performance deviation sequence for each perturbation sample is obtained by calculating the time-node difference between each curve in the set of perturbation performance curves and the corrected predicted performance curve. The variance of the distribution of all sample performance deviation sequences is then calculated and used as a performance fluctuation index. . The larger the value, the worse the performance stability of the target formulation under raw material disturbance; The smaller the value, the stronger the robustness of the ratio scheme to disturbances.

[0116] fault tolerance margin Synthetic boundary margin vector With performance fluctuation indicators Calculations show that, specifically, taking Weighted average of each component As a representative value of space margin, the weights are determined by the sensitivity of each component to performance indicators; components with higher sensitivity correspond to larger weights. Fault tolerance margin Defined as The numerator reflects the average margin of the target proportion scheme from the constraint boundary in the proportion space, while the denominator reflects the degree of performance fluctuation of the scheme under disturbance. The larger the value, the more the target mix design is far from the boundary constraints and insensitive to disturbances, the wider the allowable error range during construction, and the stronger the control robustness. When the value is small, stricter material supply accuracy requirements need to be set for this construction stage in the segmented control strategy, and real-time monitoring trigger conditions need to be added when the mix ratio execution parameters are issued to ensure construction quality.

[0117] In real-world engineering scenarios, geological conditions and environmental states vary across different construction stages, resulting in varying tolerance margins for the target mix proportions at each stage. By comparing the tolerance margins at each stage, the weakest stage with the smallest margin can be identified. A higher-precision material supply control unit can be prioritized for this stage, and a more frequent feedback acquisition frequency can be set for it in the segmented control strategy. This, in turn, improves the overall reliability and engineering adaptability of the adaptive control of cement slurry mix proportions.

[0118] In one alternative implementation,

[0119] Based on the fault tolerance margin, a segmented control strategy is generated and associated with the corresponding construction stage identifier. The proportioning execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive proportioning execution, including:

[0120] Based on the fault tolerance margin, the target mix design is decomposed into intervals to obtain the mix ratio fluctuation range. The mix ratio fluctuation range and the layer depth sequence are correlated and mapped to obtain the mix ratio constraint interval for each construction stage. The mix ratio constraint interval between adjacent construction stages is continuously constrained to generate a segmented control strategy. Each mix ratio segment in the segmented control strategy is marked with a construction stage identifier.

[0121] The mixing scheme corresponding to each construction stage is extracted from the segmented control strategy, and the component feeding amount is calculated to obtain the benchmark feeding parameters. The dominant mode vector is extracted from the fluctuation trend characteristics, and the projection coefficient between the current environmental state data and the dominant mode vector is calculated to obtain the environmental offset. The benchmark feeding parameters are dynamically compensated based on the environmental offset to obtain the real-time feeding parameters. The fault tolerance margin is converted into the feeding amount tolerance and superimposed with the real-time feeding parameters to obtain the mixing execution parameters. The mixing execution parameters are sent to the material supply control unit to drive the mixing execution.

[0122] Generating a segmented control strategy based on fault tolerance margin first requires decomposing the target allocation scheme into intervals. Specifically, based on fault tolerance margin... As a baseline reference for the fluctuation range, symmetrical expansions are applied upwards and downwards to the proportion values ​​of each component in the target formulation, respectively, to obtain the allowable fluctuation range of each component during execution. The width of this fluctuation range is related to... Positive correlation: When the fault tolerance margin is large, it indicates that the target mix design is far from the constraint boundary and the performance fluctuation is small, so the corresponding fluctuation range can be appropriately widened; when the fault tolerance margin is small, it indicates that the design is near the constraint boundary or the performance is more sensitive to disturbances, so the mix design fluctuation range should be narrowed accordingly to ensure construction safety.

[0123] The mix proportion fluctuation range is mapped to a stratified depth sequence. The stratified depth sequence originates from the intensity distribution characteristics obtained through spatial stratification analysis of geological structure data, with each depth stratum corresponding to a construction stage. The strength characteristics, permeability, and pore structure of the strata differ across construction stages; therefore, the mapping results for the same mix proportion fluctuation range vary across different depth intervals. By incorporating geological characteristic parameters of each depth interval into the mapping relationship, the mix proportion constraint intervals corresponding to shallow geological conditions and those corresponding to deep geological conditions exhibit differentiated numerical distributions, thus forming a set of multiple mix proportion constraint intervals that match the construction stratification structure.

[0124] Continuity constraints are applied to the mix proportion constraint intervals between adjacent construction stages to prevent abrupt changes in mix proportions during stage transitions and to avoid uncontrollable jumps in slurry rheological properties due to drastic changes in component ratios. The implementation of continuity constraints requires that the mix proportion constraint intervals of two adjacent construction stages overlap or connect at their boundaries; that is, the upper boundary of the mix proportion constraint interval of the previous stage is not lower than the lower boundary of the mix proportion constraint interval of the subsequent stage, ensuring that there is a mix proportion transition value within the stage transition area that simultaneously satisfies the constraints of both stages. After continuity constraint processing, the mix proportion constraint intervals of all construction stages together constitute a complete segmented control strategy, and each mix proportion segment is marked with a corresponding construction stage identifier for accurate location during subsequent extraction and distribution.

[0125] After extracting the mix proportions corresponding to each construction stage from the segmented control strategy, it is necessary to calculate the actual feed amount of each component to obtain the baseline feed parameters. Let the target mix proportion for a certain construction stage be... The proportions of the components are as follows: The total amount of grout injected in a single grouting session during this stage is Then the baseline feed amount of this component satisfy In actual engineering, The baseline feeding parameters are determined by the rated flow rate of the grouting pump and the duration of this construction phase; these parameters are for each component. The set of instructions serves as the initial reference commands for the material feeding control unit.

[0126] Dynamic changes in environmental condition data have a real-time impact on the actual performance of the slurry. Therefore, it is necessary to extract the dominant mode vector from the fluctuation trend characteristics and dynamically compensate the baseline feed parameters. The dominant mode vector is a vector representation formed by extracting the modal components with the largest energy proportion after modal decomposition of the time-series correlation analysis results of the environmental condition data. It reflects the most important direction of change in environmental fluctuations. Let the environmental condition data vector at the current moment be... The dominant mode vector is Then the projection coefficient of the current environmental state relative to the dominant mode direction for , This is the environmental offset, whose sign reflects the direction of the offset of the current environmental state relative to the dominant mode, and whose absolute value reflects the magnitude of the offset. When When this occurs, it indicates that the current environmental conditions are shifting positively along the dominant mode direction, such as higher temperature or higher humidity. In this case, the slurry fluidity may increase, requiring an appropriate reduction in the water-cement ratio or an increase in the amount of coagulant components added; when When this occurs, it indicates that the environmental state has shifted in the opposite direction, requiring compensation adjustments in the opposite direction.

[0127] Based on environmental offset The baseline feeding parameters are dynamically compensated to obtain the real-time feeding parameters. The compensation process utilizes a pre-calibrated environment-feeding response coefficient matrix. To achieve the first step of this matrix... Line 1 Column elements Indicates the first When the dominant mode shift in the environment is a unit quantity, the first The adjustment range of the feed amount for each component. The first component in the real-time feed parameters. Feed amount of each component satisfy ,in For the first Projection coefficients in the direction of each dominant mode. This serves as the index number for the dominant mode. Through the aforementioned compensation mechanism, the real-time feeding parameters can adaptively adjust to the dynamic changes in the construction site environment, reducing deviations in slurry performance caused by environmental fluctuations.

[0128] fault tolerance margin Converting to feed rate tolerance maps the stability assessment results to an allowable deviation range that the feed control unit can directly execute. The conversion method is as follows: based on the mapping relationship between tolerance margin and proportion fluctuation range, combined with the density and unit volume mass parameters of each component, [the following is done]... Converted to allowable upper and lower deviations in the amount of each component fed. .Will With real-time feeding parameters By superimposing the ranges, we obtain the first... The range of parameters for the proportioning of each component is as follows: The feeding control unit performs actual feeding control within this range, which not only ensures the accuracy of the proportioning, but also reserves a reasonable tolerance space for equipment execution error and raw material weighing error.

[0129] After the mix proportion execution parameters are sent to the material supply control unit, the control unit drives the feeding mechanisms of each raw material silo to accurately mix materials according to the set feeding amount and tolerance range, based on the mix proportion execution parameter range corresponding to each construction stage identifier. During stage switching, the material supply control unit automatically switches to the mix proportion execution parameters of the next stage based on the construction stage identifier, and uses a transition range ensured by continuity constraints to ensure a smooth transition, avoiding sudden changes in mix proportion that could impact the grouting equipment and the geological structure. Throughout the entire execution process, real-time environmental status data is continuously collected to update the environmental offset. This drives the dynamic updating of real-time material feeding parameters and mix proportion execution parameters, enabling closed-loop adaptive control throughout the entire construction cycle.

[0130] A second aspect of the present invention provides a cement slurry mix proportion adaptive control system, comprising:

[0131] The scene constraint unit is used to collect geological structure data and environmental status data at the construction site, perform spatial layering analysis on the geological structure data to obtain intensity distribution characteristics, perform temporal correlation analysis on the environmental status data to obtain fluctuation trend characteristics, and combine the intensity distribution characteristics to construct a scene constraint vector.

[0132] The proportioning boundary unit is used to obtain the physical property parameters and rheological history data of the slurry raw material and perform similarity matching to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition operation is performed to obtain the proportioning boundary space.

[0133] The mix proportion correction unit is used to map the scenario constraint vector to the mix proportion boundary space and solve the constraints to obtain an initial mix proportion scheme, perform rheological simulation calculations on the initial mix proportion scheme to obtain a predicted performance curve, perform deviation quantification analysis based on the predicted performance curve and preset construction performance requirements to obtain a correction vector, and perform gradient projection calculations on the correction vector to obtain the mix proportion adjustment amount.

[0134] The strategy execution unit is used to iteratively correct the initial proportion scheme based on the proportion adjustment amount to obtain the target proportion scheme, perform stability assessment calculation on the target proportion scheme to obtain the fault tolerance margin, generate a segmented control strategy based on the fault tolerance margin and associate it with the corresponding construction stage identifier, determine the proportion execution parameters of each construction stage according to the segmented control strategy and send them to the material supply control unit to drive the proportion execution.

[0135] A third aspect of the present invention provides an electronic device, comprising:

[0136] A processor and a memory for storing processor-executable instructions, wherein the processor is configured to invoke instructions stored in the memory to perform the aforementioned method.

[0137] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0138] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for adaptive control of cement slurry mix proportions, characterized in that, include: Geological structure data and environmental status data of the construction site are collected. Spatial layer analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental status data to obtain fluctuation trend characteristics. The scene constraint vector is constructed by combining the intensity distribution characteristics. The physical properties and rheological history data of the slurry raw material are obtained and similarity matching is performed to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition calculation is performed to obtain the proportioning boundary space. The scenario constraint vector is mapped to the mix proportion boundary space and the constraint is solved to obtain the initial mix proportion scheme. The initial mix proportion scheme is subjected to rheological simulation calculation to obtain the predicted performance curve. Based on the predicted performance curve and the preset construction performance requirements, the deviation quantification analysis is performed to obtain the correction vector. The correction vector is subjected to gradient projection operation to obtain the mix proportion adjustment amount. The initial mix proportion scheme is iteratively corrected based on the mix proportion adjustment amount to obtain the target mix proportion scheme. The stability of the target mix proportion scheme is evaluated and the fault tolerance margin is calculated. A segmented control strategy is generated based on the fault tolerance margin and associated with the corresponding construction stage identifier. The mix proportion execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive the mix proportion execution.

2. The method according to claim 1, characterized in that, Geological structure data and environmental condition data of the construction site are collected. Spatial layering analysis is performed on the geological structure data to obtain intensity distribution characteristics. Temporal correlation analysis is performed on the environmental condition data to obtain fluctuation trend characteristics. Combined with the intensity distribution characteristics, a scene constraint vector is constructed, including: Geological structure data and environmental status data of the construction site are collected by a distributed sensing unit. Stratigraphic depth information and rock mass type identification are extracted from the geological structure data. Based on the stratigraphic depth information, intervals are divided to obtain a layered depth sequence. The rock mass type identification in each depth interval is mapped to the strength level to obtain a layered strength vector. The gradient of the layered strength vector is calculated to identify stress concentration areas and mark the area boundaries to obtain the strength distribution characteristics. Temperature and humidity time series data are extracted from the environmental state data, and sliding window sampling is performed to calculate the statistical features within the window to obtain the environmental fluctuation sequence. The cross-correlation coefficient between the temperature and humidity time series data is calculated and eigenvalue decomposition is performed to obtain a set of feature vectors. Based on the environmental fluctuation sequence, the set of feature vectors is filtered to obtain the dominant mode vector and used as the fluctuation trend feature. The intensity distribution features are extended according to the layered depth sequence to obtain a spatial feature tensor, and tensor cross product operation is performed with the fluctuation trend features to obtain a spatiotemporal correlation tensor. Based on the region boundary of the stress concentration region, the spatiotemporal correlation tensor is divided into multiple sub-tensors, and singular value decomposition is performed to extract principal components. The principal components are spliced ​​along the depth dimension to reconstruct the scene constraint vector.

3. The method according to claim 1, characterized in that, The material reference features are obtained by acquiring the physical properties and rheological history data of the slurry raw materials and performing similarity matching. Based on the material reference features, a component correlation matrix is ​​established and decomposed to obtain the proportioning boundary space, including: The physical property parameters of each raw material component are extracted from the slurry raw material library. The mutual information coefficient between different physical property parameters is calculated and cluster analysis is performed to obtain a subset of key physical property parameters. Historical mixing schemes are extracted from pre-stored rheological history data. Based on the subset of key physical property parameters, historical mixing schemes similar to the current physical property parameters in the rheological history data are screened and the corresponding component ratio sequences are extracted and feature-encoded to obtain a mixing feature vector. The distance metric between the physical property parameters of the current construction scenario and the physical property parameters of the scenario corresponding to the historical mixing scheme is calculated and the mixing feature vector is weighted and fused to obtain the material benchmark features. The proportions of each group in the material reference features are arranged in a matrix according to the component type to obtain the proportion matrix. The strength index and viscosity index corresponding to the material reference features are extracted to construct the performance matrix and cross-correlation calculation is performed with the proportion matrix to obtain the component correlation matrix. The component correlation matrix is ​​subjected to singular value decomposition to extract left and right singular vectors. The component synergy is calculated based on the left singular vectors, and the right singular vectors are filtered to retain the dominant combination pattern. The singular values ​​corresponding to the dominant combination pattern are extracted, and the boundary constraints are constructed based on the singular values ​​and the dominant combination pattern to obtain the matching boundary space.

4. The method according to claim 1, characterized in that, The initial mix design is obtained by mapping the scenario constraint vector to the mix design boundary space and solving for the constraints. The predicted performance curve is then obtained by performing rheological simulation on the initial mix design, including: Projection operation is performed on the scene constraint vector and the dominant combination mode in the ratio boundary space to obtain a projection coefficient set. Based on the projection coefficient set and the pre-acquired boundary constraint conditions, the constraint satisfaction interval is determined. Discrete sampling is performed on the ratio dimension of each group within the constraint satisfaction interval to obtain a candidate ratio set. The Euclidean distance between each candidate ratio point in the candidate ratio set and the scene constraint vector is calculated, and the candidate ratio point with the smallest Euclidean distance is selected. The selected candidate ratio point is consistent with the pre-acquired component synergy degree and adjusted to obtain the initial ratio scheme. The correlation coefficients corresponding to the initial proportioning scheme are extracted from the component correlation matrix. Element-wise multiplication of each component proportion in the initial proportioning scheme with its corresponding correlation coefficient is performed to obtain the component interaction matrix. The component interaction matrix is ​​iteratively calculated according to the time step to obtain the state evolution matrix. Shear stress components and flow rate components are extracted from the state evolution matrix and the viscosity evolution sequence is calculated. The hydration reaction progress and the corresponding strength development sequence are calculated based on the cement component proportions in the initial proportioning scheme. The viscosity evolution sequence and the strength development sequence are respectively fitted with time axis curves to obtain viscosity curves and strength curves. The data are then stitched together according to the time dimension to obtain the predicted performance curve.

5. The method according to claim 1, characterized in that, Based on the predicted performance curve and the preset construction performance requirements, a deviation quantification analysis is performed to obtain a correction vector. Gradient projection calculation is then performed on the correction vector to obtain the mix proportion adjustment amount, including: Viscosity and strength curves are extracted from the predicted performance curves and discretely sampled according to time nodes to obtain viscosity sampling sequences and strength sampling sequences. Target viscosity and target strength sequences are extracted from preset construction performance requirements, and point-by-point deviations between these sequences and the viscosity and strength sampling sequences are calculated to obtain viscosity deviation sequences and strength deviation sequences. Deviation tensors are constructed according to construction stages and time nodes. Tucker decomposition is performed on the deviation tensors to extract core tensors and factor matrices. Based on the factor matrices, stage coupling coefficients for construction stages are calculated, and the core tensors are reconstructed to obtain comprehensive deviation sequences. Dimensional mapping is performed on the comprehensive deviation sequences according to the component types in the initial mix design to obtain correction vectors. Extract the set of normal vectors corresponding to the boundary constraints from the ratio boundary space and orthogonalize them to obtain the constraint space basis vectors. Project the correction vector onto the constraint space basis vectors to obtain the tangential component. Calculate the inner product between the tangential component and the pre-acquired component synergy degree and construct synergy consistency constraints to correct the tangential component to obtain the feasible correction component. Calculate the step size factor based on the boundary distance between the feasible correction component and the constraint satisfaction interval and scale the feasible correction component to obtain the ratio adjustment amount.

6. The method according to claim 1, characterized in that, The initial proportioning scheme is iteratively corrected based on the aforementioned proportioning adjustment to obtain the target proportioning scheme. The stability assessment of the target proportioning scheme yields the fault tolerance margin, including: The modified proportioning scheme is obtained by performing vector addition on the adjustment amount and the initial proportioning scheme. The modified proportioning scheme is then subjected to rheological simulation to obtain the modified predicted performance curve. The deviation between the modified and the predicted performance curve is calculated to obtain the performance improvement degree. The ratio of the performance improvement degree to the modulus of the adjustment amount is calculated to obtain the convergence index. When the convergence index is greater than the preset convergence threshold, the modified proportioning scheme is taken as the target proportioning scheme. When the convergence index is less than or equal to the preset convergence threshold, the deviation quantization analysis and gradient projection operation are re-executed on the modified proportioning scheme and iteratively corrected until the convergence index is greater than the preset convergence threshold to obtain the target proportioning scheme. Extract each allocation ratio from the target allocation scheme to construct an allocation vector. Calculate the minimum distance between the allocation vector and the boundary based on the allocation boundary space to obtain a boundary margin vector. Perform perturbation sampling on the allocation vector to generate a perturbation allocation set and perform rheological simulation to obtain a perturbation performance curve set. Calculate the variance of the deviation distribution between the perturbation performance curve set and the corrected predicted performance curve as a performance fluctuation index. Calculate the tolerance margin based on the boundary margin vector and the performance fluctuation index.

7. The method according to claim 1, characterized in that, Based on the fault tolerance margin, a segmented control strategy is generated and associated with the corresponding construction stage identifier. The proportioning execution parameters for each construction stage are determined according to the segmented control strategy and sent to the material supply control unit to drive proportioning execution, including: Based on the fault tolerance margin, the target mix design is decomposed into intervals to obtain the mix ratio fluctuation range. The mix ratio fluctuation range and the layer depth sequence are correlated and mapped to obtain the mix ratio constraint interval for each construction stage. The mix ratio constraint interval between adjacent construction stages is continuously constrained to generate a segmented control strategy. Each mix ratio segment in the segmented control strategy is marked with a construction stage identifier. The mixing scheme corresponding to each construction stage is extracted from the segmented control strategy, and the component feeding amount is calculated to obtain the benchmark feeding parameters. The dominant mode vector is extracted from the fluctuation trend characteristics, and the projection coefficient between the current environmental state data and the dominant mode vector is calculated to obtain the environmental offset. The benchmark feeding parameters are dynamically compensated based on the environmental offset to obtain the real-time feeding parameters. The fault tolerance margin is converted into the feeding amount tolerance and superimposed with the real-time feeding parameters to obtain the mixing execution parameters. The mixing execution parameters are sent to the material supply control unit to drive the mixing execution.

8. A cement slurry mix proportion adaptive control system, used to implement the method of any one of claims 1-7, characterized in that, include: The scene constraint unit is used to collect geological structure data and environmental status data at the construction site, perform spatial layering analysis on the geological structure data to obtain intensity distribution characteristics, perform temporal correlation analysis on the environmental status data to obtain fluctuation trend characteristics, and combine the intensity distribution characteristics to construct a scene constraint vector. The proportioning boundary unit is used to obtain the physical property parameters and rheological history data of the slurry raw material and perform similarity matching to obtain the material reference features. Based on the material reference features, a component correlation matrix is ​​established and decomposition operation is performed to obtain the proportioning boundary space. The mix proportion correction unit is used to map the scenario constraint vector to the mix proportion boundary space and solve the constraints to obtain an initial mix proportion scheme, perform rheological simulation calculations on the initial mix proportion scheme to obtain a predicted performance curve, perform deviation quantification analysis based on the predicted performance curve and preset construction performance requirements to obtain a correction vector, and perform gradient projection calculations on the correction vector to obtain the mix proportion adjustment amount. The strategy execution unit is used to iteratively correct the initial proportion scheme based on the proportion adjustment amount to obtain the target proportion scheme, perform stability assessment calculation on the target proportion scheme to obtain the fault tolerance margin, generate a segmented control strategy based on the fault tolerance margin and associate it with the corresponding construction stage identifier, determine the proportion execution parameters of each construction stage according to the segmented control strategy and send them to the material supply control unit to drive the proportion execution.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images