Foundation soil slurry quality analysis method, system and equipment and storage medium

Abstract

The invention provides a foundation soil slurry quality analysis method, system and equipment and a storage medium, and relates to the technical field of soil slurry detection. The method comprises the following steps: acquiring multi-modal data of liquid at a drilling fluid inlet and a flow-back liquid outlet by utilizing multi-modal sensors arranged at the drilling fluid inlet and the flow-back liquid outlet; preprocessing the multi-modal data to obtain preprocessed multi-modal data, and performing feature extraction on the preprocessed multi-modal data to obtain multi-modal feature data; and according to the multi-modal characteristic number, extracting the quality parameters of the foundation soil slurry to obtain the quality parameters of the foundation soil slurry, and according to the quality parameters of the foundation soil slurry, analyzing the quality of the foundation soil slurry to obtain a quality analysis result of the foundation soil slurry. Through the alternative scheme, the effect of the pension service is improved.

Description

Technical Field

[0001] The present application relates to the technical field of soil slurry detection, and in particular to a method, system, equipment and storage medium for analyzing the quality of foundation soil slurry. Background Art

[0002] In the field of engineering construction, the quality of the foundation is directly related to the safety and stability of the building. In the process of foundation construction, the quality of the slurry, as an important pouring material, has a decisive influence on the strength and uniformity of the foundation. Therefore, quality analysis and monitoring of the foundation slurry is the key to ensuring the quality of foundation construction.

[0003] At present, the existing technology mainly uses manual sampling and laboratory testing to analyze the quality of soil slurry. Specifically, construction workers sample the soil slurry at the construction site, and then send the samples to the laboratory for testing and analysis of indicators such as chemical composition, suspended particles, and rheology, so as to determine whether the soil slurry meets the design and construction requirements. However, this traditional soil slurry quality analysis method has certain limitations and shortcomings. First, the process of manual sampling and laboratory testing is time-consuming, and it is impossible to achieve real-time monitoring and feedback of soil slurry quality, and there is a lag. Secondly, since the composition and properties of soil slurry may change with the progress of construction and changes in environmental conditions, it is difficult to fully reflect the dynamic changes in soil slurry quality by simply relying on intermittent sampling tests. Furthermore, laboratory testing mainly analyzes a single physical and chemical indicator of soil slurry, and lacks accurate evaluation of the quality of foundation soil slurry. Summary of the invention

[0004] The present application provides a foundation soil slurry quality analysis method, system, equipment and storage medium, which can improve the accuracy of foundation soil slurry quality assessment.

[0005] In a first aspect, the present application provides a method for analyzing the quality of foundation slurry, the method comprising: Using multimodal sensors disposed at the drilling fluid inlet and the flowback fluid outlet, multimodal data of the fluid at the drilling fluid inlet and the flowback fluid outlet are obtained; Preprocessing the multimodal data to obtain preprocessed multimodal data, and extracting features from the preprocessed multimodal data to obtain multimodal feature data; The foundation soil slurry quality parameters are extracted according to the multi-modal characteristic numbers to obtain the foundation soil slurry quality parameters, and the foundation soil slurry quality is analyzed according to the foundation soil slurry quality parameters to obtain the foundation soil slurry quality analysis results.

[0006] By adopting the above technical solution and setting up multimodal sensors at the drilling fluid inlet and the return fluid outlet, multi-dimensional information of the slurry liquid can be obtained in real time, including multi-spectral data, image data, and particle size distribution data. This multimodal data collection method can comprehensively reflect the physical and chemical properties of the slurry and provide a rich data basis for subsequent quality analysis.

[0007] After acquiring the multimodal data, the present invention adopts a complete set of data processing and analysis processes. First, the multimodal data is preprocessed. Through operations such as outlier cleaning, timestamp extraction and alignment, and standardization processing, the noise and interference in the data can be effectively eliminated, and the quality and consistency of the data can be improved. On this basis, feature extraction is performed on the preprocessed multimodal data, and feature information related to soil slurry quality is extracted from multispectral data, image data, and particle size distribution data, and multimodal feature data is obtained through feature fusion. This step can convert the original high-dimensional data into an information-rich feature representation, reducing the complexity of data analysis.

[0008] By using the extracted multimodal feature data, the present invention further realizes the extraction of foundation soil slurry quality parameters. By extracting chemical feature values, image features, and particle size features from multispectral feature fusion data, image feature fusion data, and particle size distribution feature fusion data, respectively, and combining the corresponding regression equations and models, the chemical composition parameters, suspended particle parameters, and rheological parameters of the soil slurry can be accurately determined, thereby obtaining comprehensive foundation soil slurry quality parameters. This quality parameter extraction method based on multimodal features comprehensively considers the various properties of the soil slurry and can more accurately and reliably reflect the true quality status of the soil slurry.

[0009] Finally, according to the extracted foundation soil slurry quality parameters, the present invention uses the principal component analysis method to comprehensively evaluate the soil slurry quality. By calculating the parameter qualification rate and performing principal component analysis, a series of indicators such as the principal component load matrix, the principal component score matrix, and the principal component cumulative variance contribution rate are obtained. Combined with the preset thresholds and judgment rules, the influencing factors and key parameters of the soil slurry quality can be comprehensively analyzed, quality anomalies and potential risks can be identified, and an important basis for quality control during the construction process can be provided.

[0010] In a second aspect of the present application, a foundation slurry quality analysis method system is provided, comprising: A data acquisition module, used to acquire multimodal data of the liquid at the drilling fluid inlet and the flowback fluid outlet by using multimodal sensors arranged at the drilling fluid inlet and the flowback fluid outlet; A data processing module is used to preprocess the multimodal data to obtain preprocessed multimodal data, and to extract features from the preprocessed multimodal data to obtain multimodal feature data; The data analysis module is used to extract foundation soil slurry quality parameters according to the multimodal feature data to obtain the foundation soil slurry quality parameters, and analyze the foundation soil slurry quality according to the foundation soil slurry quality parameters to obtain the foundation soil slurry quality analysis results.

[0011] In a third aspect of the present application, a computer storage medium is provided, wherein the computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the above method steps.

[0012] In the fourth aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the above method.

[0013] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. This application can obtain multi-dimensional information of soil slurry liquid in real time by setting multi-modal sensors at the drilling fluid inlet and the return fluid outlet, including multi-spectral data, image data, and particle size distribution data. This multi-modal data collection method can comprehensively reflect the physical and chemical properties of soil slurry and provide a rich data basis for subsequent quality analysis.

[0014] 2. After acquiring multimodal data, the present invention adopts a complete set of data processing and analysis processes. First, the multimodal data is preprocessed. Through operations such as outlier cleaning, timestamp extraction and alignment, and standardization processing, the noise and interference in the data can be effectively eliminated, and the quality and consistency of the data can be improved. On this basis, feature extraction is performed on the preprocessed multimodal data, and feature information related to soil slurry quality is extracted from multispectral data, image data, and particle size distribution data, and multimodal feature data is obtained through feature fusion. This step can convert the original high-dimensional data into an information-rich feature representation, reducing the complexity of data analysis.

[0015] 3. This application utilizes the extracted multimodal feature data, and the present invention further realizes the extraction of foundation soil slurry quality parameters. By extracting chemical feature values, image features, and particle size features from multispectral feature fusion data, image feature fusion data, and particle size distribution feature fusion data, respectively, and combining the corresponding regression equations and models, the chemical composition parameters, suspended particle parameters, and rheological parameters of the soil slurry can be accurately determined, thereby obtaining comprehensive foundation soil slurry quality parameters. This quality parameter extraction method based on multimodal features comprehensively considers the various properties of the soil slurry and can more accurately and reliably reflect the true quality status of the soil slurry. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a flow chart of a method for analyzing the quality of foundation soil slurry provided in an embodiment of the present application; Figure 2 An architecture diagram of a foundation soil slurry quality analysis system provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in this application. DETAILED DESCRIPTION

[0017] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0018] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0019] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0020] In order to facilitate understanding of the method and system provided by the embodiments of the present application, before introducing the embodiments of the present application, the background of the embodiments of the present application is first introduced.

[0021] Looking at the current engineering practice, the traditional soil slurry quality analysis method mainly relies on manual sampling and laboratory testing. Specifically, construction workers need to take regular or irregular samples of soil slurry at the construction site, and then send the collected samples to a special laboratory for a series of index tests and analyses, including chemical composition, suspended particles, rheology and other aspects, to determine whether the soil slurry meets the design and construction standard requirements. However, although this traditional approach has been in use for many years, its limitations and shortcomings are becoming increasingly prominent.

[0022] First, the process of manual sampling and laboratory testing is often time-consuming, and multiple links and steps are required from sample collection to result feedback, which leads to the lag of soil slurry quality monitoring. The actual situation at the construction site changes rapidly, and the quality parameters of the soil slurry may also change at any time. However, traditional methods cannot respond to quality problems in real time and make timely adjustments, and there is a certain time difference. Secondly, as a complex multiphase system, the composition and properties of soil slurry may change dynamically with the progress of construction and changes in environmental conditions. For example, factors such as fluctuations in the quality of raw materials, slight adjustments in the proportions, and optimization and improvement of construction processes may affect the performance of soil slurry. Simply relying on intermittent sampling and testing is difficult to fully and continuously reflect the dynamic changes in the quality of soil slurry, and it is easy to produce one-sided or lagging quality assessment results. Furthermore, laboratory testing usually analyzes a single physical and chemical indicator of soil slurry, such as chemical composition, particle size, density, viscosity, etc. Although this targeted test can provide quality information in some aspects, it lacks a comprehensive evaluation of the overall performance of soil slurry. As an organic whole, the macroscopic properties of soil slurry are the result of the interaction and mutual influence of various microscopic parameters. The test of a single indicator is difficult to accurately reflect the comprehensive quality level and applicability of soil slurry.

[0023] After the background introduction of the above content, those skilled in the art can understand the problems existing in the prior art. The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0024] Based on the above background technology, please refer to Figure 1 , Figure 1 A flow chart of a method for analyzing the quality of soil slurry for foundation provided in an embodiment of the present application. The system can be implemented by a computer program or can be run as an independent tool application. Specifically, in an embodiment of the present application, the method can be applied on a server, but can also be applied on electronic devices such as servers. A method for analyzing the quality of soil slurry for foundation includes the following steps: S101, using a multimodal sensor disposed at a drilling fluid inlet and a flowback fluid outlet to obtain multimodal data of liquid at the drilling fluid inlet and the flowback fluid outlet; Specifically, multimodal sensors such as multispectral sensors, image sensors, and particle size distribution sensors are installed at the inlet of the drilling fluid and the outlet of the return fluid. Multispectral sensors can collect reflectance or absorbance data of the slurry liquid in different spectral bands, which contain rich information about the chemical composition of the slurry. Image sensors can capture the flow state of the slurry liquid and the morphological characteristics of the suspended particles, providing an intuitive basis for analyzing the uniformity and stability of the slurry. Particle size distribution sensors can measure the particle size distribution and content of solid particles in the slurry liquid, reflecting the fineness and proportion characteristics of the slurry.

[0025] By deploying multimodal sensors at both the drilling fluid inlet and the flowback fluid outlet, the present invention realizes the real-time collection of multi-dimensional information of the slurry liquid. The sensor data at the drilling fluid inlet reflects the initial state of the fresh slurry, while the sensor data at the flowback fluid outlet reflects the quality change of the slurry after the drilling cycle. By comparing and analyzing the multimodal data at these two locations, the quality dynamics of the slurry during the drilling process can be comprehensively evaluated, problems can be discovered in a timely manner, and corresponding measures can be taken.

[0026] Based on the above embodiment, as an optional embodiment, the multimodal sensor includes a spectral detector, a laser particle size analyzer and a high-speed camera; the multimodal sensor arranged at the drilling fluid inlet and the flowback fluid outlet is used to obtain multimodal data of the liquid at the drilling fluid inlet and the flowback fluid outlet, including: S201, using spectral detectors arranged at the drilling fluid inlet and the flowback fluid outlet to obtain multi-spectral data of the liquid at the drilling fluid inlet and the flowback fluid outlet; Specifically, spectral detectors, such as visible-near infrared spectrometers, mid-infrared spectrometers, etc., are installed at the inlet of the drilling fluid and the outlet of the return fluid. These spectral detectors can perform fast and non-destructive optical detection of the slurry liquid in different spectral bands to obtain rich spectral information.

[0027] In actual operation, when the slurry flows through the spectrum detector, the incident light hits the liquid surface at a certain angle, and the chemical components and suspended particles in the liquid selectively absorb, reflect or scatter light of different wavelengths. The spectrum detector measures the changes in light intensity of the liquid at different wavelengths to obtain a series of spectral data, including reflection spectrum, absorption spectrum, scattering spectrum, etc.

[0028] These multispectral data contain rich information about the chemical composition of the slurry liquid. Different chemical components, such as cement, clay, additives, etc., have specific spectral response patterns in different spectral bands. By analyzing the characteristic peak position, peak intensity, peak shape, etc. of multispectral data, qualitative identification and quantitative analysis of the chemical composition of the slurry can be achieved. For example, the type and content of cement in the slurry can be identified through visible light-near infrared spectral data; the type and concentration of organic additives in the slurry can be detected through mid-infrared spectral data.

[0029] At the same time, multispectral data can also reflect the suspended particle characteristics of soil slurry liquid. The particle size, shape, concentration and other parameters of suspended particles will affect the scattering and absorption behavior of light, and thus affect the morphological characteristics of spectral data. By establishing a quantitative relationship model between spectral characteristics and particle characteristics, rapid characterization and monitoring of soil slurry suspended particles can be achieved.

[0030] S202, using high-speed cameras disposed at the drilling fluid inlet and the flowback fluid outlet to obtain image data of the liquid at the drilling fluid inlet and the flowback fluid outlet; Specifically, high-speed cameras, such as industrial digital cameras, high-speed CCD cameras, etc., are installed at the drilling fluid inlet and the return fluid outlet. These high-speed cameras have high frame rates and resolutions, and can continuously capture the flow process of the slurry liquid and obtain clear and coherent image sequences.

[0031] In actual operation, when the slurry flows through the field of view of the high-speed camera, the camera continuously images the liquid at a certain frame rate and exposure time. By adjusting the appropriate light source and shooting parameters, the slurry and suspended particles in the image are clearly presented. The image data captured by the high-speed camera contains rich visual information such as the flow state of the slurry, the morphology of suspended particles, and the size of aggregates.

[0032] S203, using a laser particle size analyzer disposed at the drilling fluid inlet and the flowback fluid outlet to obtain particle size distribution data of the liquid at the drilling fluid inlet and the flowback fluid outlet; Specifically, laser particle size analyzers, such as laser diffraction particle size analyzer, laser scattering particle size analyzer, etc., are installed at the inlet of drilling fluid and the outlet of flowback fluid. These particle size analyzers use laser light sources to measure the scattering and diffraction effects of the suspended particles in the slurry on the laser beam to quickly and accurately obtain the particle size distribution information.

[0033] In actual operation, when the soil slurry flows through the measuring cell of the laser particle size analyzer, the instrument emits a parallel monochromatic laser beam to illuminate the suspended particles in the liquid. According to the Mie scattering theory and Fraunhofer diffraction principle, particles of different sizes will produce scattered light and diffracted light of different intensities at different angles. The particle size analyzer receives scattered and diffracted signals at different angles through multiple annular detector arrays, and converts the optical signals into electrical signals. After amplification, filtering, A / D conversion and other processing, a set of digital particle size distribution data is obtained, including the volume percentage, cumulative distribution, characteristic particle size and other parameters of the particles.

[0034] S204, taking the multispectral data, the image data and the particle size distribution data as multimodal data.

[0035] Specifically, after obtaining the multispectral data, image data, and particle size distribution data of the slurry liquid at the drilling fluid inlet and the return fluid outlet, it is necessary to integrate and organize these data of different types and formats to form a unified multimodal data set. This process is usually achieved through data fusion and data alignment.

[0036] First, for each data collection moment, the synchronously acquired multispectral data, image data, and particle size distribution data are matched according to the timestamp to form a time-aligned data group. In this way, the data of different modes establish a temporal correspondence, which is convenient for subsequent correlation analysis and fusion processing. Secondly, feature extraction and data normalization are performed on the data of different modes. For multispectral data, the characteristic absorption peaks and peak intensities reflecting the chemical composition of the slurry are extracted; for image data, the image texture and color histogram reflecting the morphology of the slurry particles are extracted; for particle size distribution data, the characteristic particle size and distribution width reflecting the particle size distribution of the slurry particles are extracted. Through feature extraction, the original high-dimensional data is converted into compact and effective feature vectors, reducing data redundancy and computational complexity. At the same time, the extracted features are normalized so that the feature data of different modes are comparable on a numerical scale. Finally, the feature vectors extracted from multiple modes are spliced ​​or fused in a certain way to generate a multimodal feature vector.

[0037] S102, preprocessing the multimodal data to obtain preprocessed multimodal data, and extracting features from the preprocessed multimodal data to obtain multimodal feature data; Specifically, the multimodal data is preprocessed. Since the raw data collected by the sensor may contain some noise, outliers or inconsistencies, directly using these data for analysis may affect the accuracy of the results. Therefore, a series of preprocessing operations are required for the multimodal data, including data cleaning, timestamp alignment, data standardization, etc. Data cleaning mainly removes outliers and invalid data points in the data, such as erroneous readings caused by sensor failures, extreme values ​​beyond the normal range, etc. By setting reasonable thresholds and rules, these "dirty data" are filtered out to improve the quality and reliability of the data. Timestamp alignment is to synchronize the data collected by different sensors according to the time axis to ensure that the data of different modes are consistent in the time dimension, which is convenient for subsequent fusion analysis. Data standardization is to normalize the data of different modes, eliminate the differences in dimensions and scales, and make the features of different sources comparable.

[0038] Based on the above embodiment, as an optional embodiment, preprocessing the multimodal data to obtain preprocessed multimodal data includes: S301, using a statistical method to identify outlier points in the multimodal data, and cleaning the outlier points to obtain first multimodal data; Specifically, after obtaining the fused multimodal data, statistical methods are first used to detect outliers in the data. Outliers refer to extreme values ​​or outliers that are significantly deviated from the rest of the data points and do not conform to the overall distribution law. These outliers may be caused by factors such as sensor failure, measurement error, and data transmission error, which will seriously interfere with data analysis and modeling and must be identified and processed.

[0039] Commonly used outlier detection methods include distance-based methods, distribution-based methods, and clustering-based methods. Distance-based methods such as the Euclidean distance method and the Mahalanobis distance method calculate the distance between the data point and the sample mean or center point, and mark the points exceeding a certain threshold as outliers; distribution-based methods such as the 3σ principle and the boxplot method assume that the data obeys a certain probability distribution and mark the extreme points at the tail of the distribution as outliers; clustering-based methods such as DBSCAN and LOF divide the data into several clusters through clustering algorithms, and mark the points far away from the main cluster center and with significantly lower density as outliers.

[0040] S302, extracting a timestamp from the first multimodal data to obtain a timestamp of the first multimodal data, and performing time alignment on the first multimodal data according to the timestamp of the first multimodal data to obtain second multimodal data; Specifically, after obtaining the first multimodal data that has been cleaned of outliers, it is necessary to further extract the timestamp information in the data. The timestamp is the time point or time interval recorded when the data is collected, reflecting the time and sequence of data generation. In the actual soil slurry quality monitoring system, multispectral data, image data, and particle size distribution data are usually collected independently by different sensors. Although preliminary matching has been performed according to timestamps in the data fusion stage, due to the influence of factors such as the sampling frequency of the sensor and data transmission delay, the timestamps of different modal data may have certain deviations and inconsistencies, which require further calibration and alignment.

[0041] There are two main methods for extracting timestamps: one is to read the timestamp directly from the metadata or additional information of the data file, such as the file name, time tag, etc.; the other is to identify the time information contained in the data through data content analysis, such as the time watermark in image data, the acquisition time in spectral data, etc. When extracting timestamps, the data format and time representation of different sensors should be fully considered, and a unified time base and format specification should be adopted, such as UTC time, Unix timestamp, etc.

[0042] After extracting the timestamp of the first multimodal data, the data of different modes need to be time aligned. The purpose of time alignment is to unify the timestamps of data of different modes to a common time base, eliminating the time offset and inconsistency caused by asynchronous acquisition of sensors. Common time alignment methods include nearest neighbor interpolation, linear interpolation, and spline interpolation. Nearest neighbor interpolation matches the data of each mode to the common time point closest to the timestamp; linear interpolation performs linear fitting between two adjacent timestamps to estimate the data value at the common time point; spline interpolation uses piecewise polynomial functions for higher-order time matching and smoothing.

[0043] S303: performing standardization processing on the second multimodal data to obtain preprocessed multimodal data.

[0044] Specifically, after obtaining the time-aligned second multimodal data, the data of different modes need to be standardized. Standardization is a commonly used data normalization method, which transforms the original data into a standard normal distribution with a mean of 0 and a standard deviation of 1 by linear transformation. Since the soil slurry quality data comes from multiple sensors, data of different modes often have different physical meanings and measurement units, and the numerical ranges vary greatly. If these raw data are directly input into the quality prediction model, it is easy to cause some features to be magnified or reduced, introduce instability in numerical calculations, and affect the accuracy and generalization performance of modeling.

[0045] Standardization can effectively solve this problem. Commonly used standardization methods include Z-score standardization and Min-Max standardization. Z-score standardization is to standardize each feature dimension separately, subtract the feature mean from the original data, and then divide it by the feature standard deviation to obtain the standardized data. Suppose a feature in the second multimodal data is x, its mean is μ, and its standard deviation is σ, then the Z-score standardization formula is: x'= (x - μ) / σ Where x' is the normalized data. Min-Max normalization maps the data to the interval [0, 1], and the formula is: x'= (x - xmin) / (xmax - xmin) Where xmin and xmax are the minimum and maximum values ​​of the feature respectively. In addition to the above two methods, other normalization methods such as logarithmic normalization and quantile normalization can also be used according to the characteristics of data distribution.

[0046] Based on the above embodiment, as an optional embodiment, feature extraction is performed on the preprocessed multimodal data to obtain multimodal feature data, including: S401, extracting features from multispectral data to obtain multispectral feature data; Specifically, after obtaining the preprocessed multispectral data, it is necessary to extract features to obtain multispectral feature data. Multispectral data is usually composed of spectral reflectance or absorbance of multiple bands, and each band corresponds to a feature dimension. Directly using the original high-dimensional spectral data for quality modeling not only has high computational complexity, but also easily introduces noise and redundant information, affecting the generalization performance of the model. Therefore, it is necessary to use feature extraction methods to select the most representative and discriminative low-dimensional features from high-dimensional spectral data.

[0047] Commonly used multispectral feature extraction methods include principal component analysis (PCA), independent component analysis (ICA), partial least squares regression (PLSR), etc. PCA extracts the principal components that maximize the data variance by performing an orthogonal transformation on the spectral data, and projects the high-dimensional data into a low-dimensional space; ICA extracts independent source signal components by maximizing the statistical independence between components; PLSR combines principal component analysis and multivariate linear regression, and considers the correlation of the target variable while extracting the principal components, thereby improving the predictive power of the features. In addition to the above traditional statistical methods, some feature extraction methods based on deep learning can also be used, such as convolutional neural networks (CNN), autoencoders (AE), etc., to automatically learn the deep-level feature representation of spectral data.

[0048] When selecting a feature extraction method, the characteristics of the soil slurry quality prediction task and the distribution of the data should be fully considered, and factors such as feature interpretability, computational efficiency, and prediction performance should be weighed. Generally, an exploratory analysis of multispectral data is performed first to understand the distribution law and correlation structure of the data, and then a suitable feature extraction method is selected. The extracted multispectral feature data usually includes principal component scores, independent component coefficients, latent variable scores, etc. The dimensions are much lower than the original spectral data, but they can retain quality-related information to the maximum extent.

[0049] S402, extracting features from the image data to obtain image feature data; Specifically, after obtaining the preprocessed image data, it is necessary to extract features to obtain image feature data. Image data is usually a two-dimensional matrix composed of a large number of pixels, and each pixel contains the intensity values ​​of the three color channels of RGB. Directly using the original high-resolution image for quality modeling not only requires huge computational effort, but is also easily affected by factors such as lighting and angle, affecting the robustness and generalization ability of the features. Therefore, it is necessary to use image feature extraction methods to extract the key features that best characterize the quality of the soil slurry from the original image.

[0050] Commonly used image feature extraction methods include traditional image processing methods and deep learning-based methods. Traditional methods mainly design some feature descriptors manually based on prior knowledge of soil slurry images, such as color histogram, local binary pattern (LBP), scale-invariant feature transform (SIFT), etc., to describe the color distribution, texture structure, key points and other visual features of the image; deep learning methods use deep models such as convolutional neural networks (CNN) to automatically learn hierarchical feature representations of images, such as features at different abstract levels such as edges, textures, and semantics.

[0051] When selecting an image feature extraction method, the characteristics of soil slurry quality images and the requirements of quality prediction tasks should be fully considered. Since the imaging environment and shooting conditions of soil slurry images are relatively controllable, low-level features such as color and texture have good discriminative power for quality status, so traditional image feature extraction methods such as color moment and gray-level co-occurrence matrix (GLCM) can be given priority. For some complex quality discrimination tasks, such as water extraction and segregation, deep learning-based methods can be used to extract more abstract and high-level semantic features. The extracted image feature data usually includes color statistics, texture parameters, CNN feature maps, etc. The dimensions are much lower than the original image, but they can effectively characterize the visual characteristics of soil slurry quality.

[0052] S403, extracting features from the particle size distribution data to obtain particle size distribution feature data; Specifically, after obtaining the preprocessed particle size distribution data, it is necessary to extract its features to obtain particle size distribution feature data. Particle size distribution data is usually composed of mass percentages of several particle size intervals, showing a certain distribution curve shape. Directly using the original particle size distribution curve for quality modeling not only has a high dimension, but also makes it difficult to characterize the overall characteristics and morphological differences of the curve, affecting the discriminability and stability of the features. Therefore, it is necessary to adopt a particle size distribution feature extraction method to extract the key parameters and characteristic quantities that best reflect the quality of the slurry from the particle size distribution curve.

[0053] Commonly used methods for extracting particle size distribution features include statistical moment method, fractal dimension method, function fitting method, etc. The statistical moment method is to characterize the distribution characteristics of the curve such as position, dispersion, skewness, etc. by calculating the moments of each order of the particle size distribution curve (such as mean, variance, skewness, kurtosis, etc.); the fractal dimension method is to use the self-similarity of the particle size distribution to measure the complexity and filling characteristics of the curve through fractal parameters such as box dimension and mass dimension; the function fitting method is to use certain specific functions (such as Rosin-Rammler distribution, lognormal distribution, etc.) to fit the particle size distribution curve and extract the optimal function parameters as the particle size distribution characteristics.

[0054] When selecting a method for extracting particle size distribution features, the needs of soil slurry quality prediction and the characteristics of the particle size distribution curve should be fully considered. Since the soil slurry particle size distribution curve has strong structure and regularity, the statistical moment method and function fitting method can usually better characterize its distribution characteristics and morphological differences, so they can be used as the preferred feature extraction methods. For some occasions with higher quality requirements, such as high-strength and low-bleeding soil slurry, the fractal dimension method can be combined to extract more detailed and comprehensive particle size distribution information. The extracted particle size distribution feature data usually includes average particle size, uniformity coefficient, Rosin-Rammler distribution index, etc. The dimension is much lower than the original particle size distribution curve, but it can effectively characterize the grading characteristics of soil slurry particles.

[0055] S404, performing feature fusion on the multi-spectral feature data, the image feature data and the particle size distribution feature data to obtain multi-modal feature data.

[0056] Specifically, after obtaining the multispectral feature data, image feature data and particle size distribution feature data, they need to be fused to obtain multimodal feature data. Since the three modal features have large differences in physical meaning, numerical range, dimensional scale, etc., they cannot be directly spliced ​​and used. It is necessary to adopt an appropriate feature fusion strategy to achieve optimal combination and unified representation of features.

[0057] Commonly used feature fusion methods include concatenation fusion, weighted fusion, serial fusion and deep fusion. Concatenation fusion is to simply concatenate different modal features into a high-dimensional vector as the input of the model; weighted fusion is to assign different weights to different modal features to generate a weighted combination feature; serial fusion is to input different modal features into multiple sub-models in sequence, and then fuse the output results of the sub-models; deep fusion is to use the deep learning model to automatically learn the interaction and combination of features at different levels to generate deep fusion features.

[0058] When selecting feature fusion methods, the characteristics and correlations of different modal features, as well as the complexity of the quality prediction task, should be fully considered. For multispectral features and particle size distribution features, since their physical meanings and data structures are relatively close, they can be combined by connection fusion or weighted fusion; for image features, since their dimensions and data types are quite different from the other two features, the serial fusion method can be used first to extract the high-level semantic information of the image, and then fuse it with the multispectral features and particle size distribution features; for some complex quality prediction tasks, such as strength prediction, water seepage prediction, etc., the deep fusion method can be further used to adaptively learn the optimal combination of the three modal features.

[0059] S103, extracting foundation soil slurry quality parameters according to the multi-modal characteristic numbers to obtain foundation soil slurry quality parameters, and analyzing the foundation soil slurry quality according to the foundation soil slurry quality parameters to obtain foundation soil slurry quality analysis results.

[0060] Specifically, for different types of quality parameters, the corresponding machine learning model is selected for training and prediction. For example, for the extraction of chemical component content, the support vector regression (SVR) model is used, with spectral features as input, and the nonlinear relationship between features and component content is learned through training sample data, and then the component content is predicted for the new feature data. For the extraction of suspended particle characteristics, the convolutional neural network (CNN) model is used, with image features as input, and the morphology and distribution information of particles are extracted through convolution and pooling operations, and the fully connected layer is used to output the particle size, shape, content and other parameters. For the extraction of rheological parameters, the long short-term memory network (LSTM) model is used, with the time series of particle size features as input, and the rheological parameters such as viscosity and yield value of soil slurry are predicted by learning the time series pattern of particle size changes.

[0061] By applying machine learning algorithms, the present invention realizes automatic mapping and extraction from multimodal feature data to soil slurry quality parameters, overcomes the limitations of traditional empirical formulas and experimental methods, and improves the intelligence level and accuracy of quality parameter extraction.

[0062] After extracting the foundation soil slurry quality parameters, the last step is to conduct a comprehensive analysis of the quality parameters to obtain an overall evaluation result of the soil slurry quality. The present invention adopts the principal component analysis method to reduce the dimension and integrate multiple quality parameters, extract the main factors affecting the soil slurry quality, and calculate the comprehensive quality score.

[0063] Specifically, the extracted quality parameter data is first standardized to eliminate dimension and scale differences. Then, the correlation coefficient matrix between the quality parameters is calculated, and the eigenvalue decomposition is performed to obtain the principal component load matrix and the principal component score matrix. According to the cumulative variance contribution rate of the principal component, the first few principal components are selected as the comprehensive evaluation index of the soil slurry quality, and the scores of each sample on these principal components are calculated to obtain the comprehensive score vector of the soil slurry quality. Finally, the comprehensive score is compared with the preset quality grade threshold to judge the quality grade of the soil slurry and form the final quality analysis result.

[0064] Through principal component analysis, the present invention compresses and integrates the information of multiple quality parameters into a few principal components, which not only reduces the dimension of the evaluation index, but also retains the main variation information of the quality parameters, and realizes the comprehensive evaluation of the soil slurry quality. At the same time, the quality grade is associated with the comprehensive score, making the quality judgment more intuitive and quantitative, which is convenient for construction management personnel to monitor and make decisions on quality.

[0065] Based on the above embodiment, as an optional embodiment, the multimodal feature data includes multispectral feature fusion data, image feature fusion data and particle size distribution feature fusion data; the foundation soil slurry quality parameters are extracted according to the multimodal feature data to obtain the foundation soil slurry quality parameters, including: S501, extracting chemical characteristic values ​​related to chemical composition from the multi-spectral characteristic fusion data, and determining the soil slurry chemical composition parameters according to the chemical characteristic values ​​and a preset soil slurry chemical composition regression equation; Specifically, after obtaining the multimodal feature fusion data, it is first necessary to extract the chemical feature values ​​related to the chemical composition from the multispectral feature fusion data. Multispectral data is very sensitive to the chemical composition of materials, and different chemical components will show characteristic spectral absorption or reflection peaks in different bands. By analyzing the spectral feature values ​​such as characteristic bands and characteristic absorption valleys in the multispectral feature fusion data, the qualitative and quantitative information of key chemical components such as cement, fly ash, and mineral admixtures in the soil slurry can be inverted.

[0066] After extracting the chemical characteristic values, it is necessary to determine the chemical composition parameters of the soil slurry based on the chemical characteristic values ​​and the preset soil slurry chemical composition regression equation. There is usually a relatively stable quantitative relationship between the chemical composition of the soil slurry and its spectral characteristic values. A regression equation between the chemical characteristic values ​​and the chemical composition content can be established through a large amount of experimental data. Commonly used regression models include partial least squares regression, support vector regression, artificial neural network, etc. Using the established regression equation, substituting the extracted chemical characteristic values ​​into the equation, it is possible to quantitatively predict the various chemical composition parameters of the soil slurry, such as cement content, type and amount of mineral admixtures, water-cement ratio, etc.

[0067] S502, extracting image features related to suspended particles from the image feature fusion data, and determining soil slurry suspended particle parameters according to the image features and a preset image feature regression equation; Specifically, after obtaining the image feature fusion data, it is necessary to extract image features related to suspended particles. The image data intuitively reflects the visual morphological characteristics of the slurry and contains rich suspended particle information. By analyzing the image feature fusion data, key image features reflecting the characteristics of suspended particles can be obtained, such as particle size distribution, aspect ratio, roundness, grayscale histogram and other morphological parameters. These image features are closely related to the physical properties of suspended particles such as content, particle size, shape, etc., and can be used as important indicators to characterize suspended particles.

[0068] Commonly used image feature extraction methods include edge detection, region segmentation, morphological filtering, texture analysis, etc. In view of the characteristics of soil slurry suspended particles, a particle extraction algorithm based on threshold segmentation can be used. By setting an appropriate grayscale threshold, the image is segmented into a particle area and a background area, and then the particle area is morphologically processed to extract the particle size, shape and other characteristic parameters. At the same time, the texture analysis method can also be combined to calculate the image's grayscale co-occurrence matrix, local binary pattern and other texture features to characterize the surface roughness and spatial distribution characteristics of the particles.

[0069] After extracting the image features related to suspended particles, it is necessary to determine the soil slurry suspended particle parameters based on the image features and the preset image feature regression equation. There is also a quantitative relationship between image features and suspended particle parameters. Through a large amount of test data, a regression model between parameters such as particle content and particle size distribution and image features can be established. Commonly used regression models include multivariate linear regression, partial least squares regression, support vector regression, etc. Using the established regression equation, substituting the extracted image features into the equation, various parameters of soil slurry suspended particles, such as particle volume fraction, median particle size, uniformity coefficient, etc., can be quantitatively predicted.

[0070] S503, extracting particle size characteristics related to rheological properties from the particle size distribution characteristic fusion data, and determining the rheological parameters of the slurry according to the particle size characteristics and a preset particle size model; Specifically, after obtaining the fusion data of particle size distribution characteristics, it is necessary to extract the particle size characteristics related to rheological properties. The particle size distribution data reflects the gradation composition characteristics of the slurry particles. The content and distribution of particles in different particle size segments have an important influence on the rheological properties of the slurry. By analyzing the fusion data of particle size distribution characteristics, key particle size characteristics reflecting the rheological properties of the slurry can be obtained, such as characteristic particle size, uniformity coefficient, fractal dimension, etc. These particle size characteristics comprehensively reflect factors such as the size distribution, gradation continuity, and particle morphology complexity of the slurry particles. They have a good correspondence with the rheological parameters of the slurry and can be used as an important indicator for predicting and regulating rheological properties.

[0071] Commonly used particle size feature extraction methods include characteristic particle size extraction, gradation curve fitting, fractal dimension calculation, etc. According to the characteristics of soil slurry particle size distribution, theoretical models such as lognormal distribution function or Rosin-Rammler distribution function can be used to fit the particle size distribution data and extract characteristic particle size parameters such as median particle size and mean particle size. At the same time, the slope, curvature and other geometric characteristics of the particle size distribution curve can also be calculated to evaluate the uniformity and continuity of the gradation. In addition, using fractal theory, the fractal dimension of the particle size distribution can be calculated to quantitatively describe the complexity and self-similarity of the particle morphology and reflect the influence of the particle stacking state on the rheological properties.

[0072] After extracting the particle size characteristics related to rheological properties, it is necessary to determine the rheological parameters of the slurry based on the particle size characteristics and the preset particle size model. There is a certain quantitative relationship between the particle size characteristics and the rheological parameters. Through rheological tests and particle size analysis tests, a mathematical model between the rheological parameters and the particle size characteristics can be established. Commonly used particle size models include the Krieger-Dougherty equation, the Mooney equation, the Farris equation, etc. These models describe the influence of factors such as particle volume fraction, particle size distribution, and particle shape on rheological parameters from different angles. Using the established particle size model, the extracted particle size characteristics are substituted into the model to quantitatively estimate the rheological parameters of the slurry, such as viscosity and yield stress.

[0073] S504, taking the rheological parameters of the soil slurry, the chemical composition parameters of the soil slurry and the parameters of the suspended particles of the soil slurry as the quality parameters of the foundation soil slurry.

[0074] Specifically, the rheological parameters of the slurry reflect the fluidity and deformation characteristics of the slurry, such as viscosity, yield stress, plasticity index, etc., which determine the construction process performance of the slurry, such as pumpability, filling, and self-compacting properties. The chemical composition parameters of the slurry reflect the material composition and proportion characteristics of the slurry, such as cement content, amount of mineral admixtures, water-cement ratio, etc., which determine the hydration reaction characteristics and mechanical property development of the slurry. The suspended particle parameters of the slurry reflect the dispersion state and morphological characteristics of the slurry particles, such as particle content, particle size distribution, shape characteristics, etc., which affect the particle stacking structure and mechanical properties of the slurry. These three types of parameters are interrelated and influence each other, and jointly determine the comprehensive quality characteristics of the slurry.

[0075] Therefore, taking the rheological parameters, chemical composition parameters and suspended particle parameters of the slurry as the quality parameters of the foundation slurry can comprehensively characterize and evaluate the slurry quality from multiple dimensions. On the one hand, a single parameter can only reflect a certain aspect of the slurry quality. For example, a single rheological parameter can only characterize the process performance of the slurry, a single chemical composition parameter can only reflect the proportion of the slurry, and a single suspended particle parameter can only describe the particle characteristics of the slurry. The slurry quality is the result of the combined effect of multiple factors. It is difficult to comprehensively judge the quality of the slurry by considering only a single parameter. On the other hand, there are inherent connections and mutual constraints between different parameters. For example, chemical composition affects rheological properties, and particle characteristics affect mechanical properties. Isolating a parameter and ignoring the interaction between parameters makes it difficult to accurately grasp the essential laws of slurry quality.

[0076] Based on the above embodiment, as an optional embodiment, the foundation soil slurry quality is analyzed according to the foundation soil slurry quality parameter to obtain the foundation soil slurry quality analysis result, including: S601, comparing the foundation soil slurry quality parameters with preset standard parameter data to determine the pass rate of the foundation soil slurry quality parameters; Specifically, after obtaining the soil slurry quality parameters, it is necessary to establish a corresponding standard parameter database as a reference for quality evaluation. The standard parameter database contains standard data such as design values, specification values, and experience values ​​of various quality parameters, reflecting the qualified requirements and optimal goals of soil slurry quality. These standard data can be derived from design documents, construction specifications, industry standards, etc., or they can be determined based on engineering experience and statistical analysis. When establishing a standard parameter database, it is necessary to comprehensively consider factors such as project type, construction technology, and environmental conditions, select applicable quality standards, and ensure the scientificity and feasibility of the evaluation criteria.

[0077] During the quality analysis, the measured soil slurry quality parameters are compared with the standard parameter data one by one, and the qualified rate of each parameter is calculated. The qualified rate can be expressed by indicators such as the ratio of the measured value to the standard value and the frequency of the measured value falling within the standard range, which reflects the degree to which the measured parameters meet the quality requirements. By setting the qualified rate threshold, it is possible to determine whether each quality parameter is qualified, and further analyze the degree of quality parameters and the extent of exceeding the standard. For parameters with a higher qualified rate, it indicates that the soil slurry quality performs well in this indicator and meets the design and specification requirements; for parameters with a lower qualified rate, it indicates that the soil slurry quality is insufficient in this indicator and needs to be focused on and optimized.

[0078] Parameter qualification rate analysis can help comprehensively evaluate the quality of soil slurry and quantitatively judge the quality level. Through statistical analysis of the qualification rates of various quality parameters, quality evaluation indicators such as comprehensive qualification rate and key parameter qualification rate can be calculated, and soil slurry quality can be graded and ranked. For example, soil slurry with a comprehensive qualification rate higher than 90% is rated as excellent quality, and soil slurry with a comprehensive qualification rate lower than 70% is rated as unqualified quality. The division of quality grades provides a quantitative basis for project acceptance and quality control, which helps to timely discover and deal with quality problems and ensure project quality and safety.

[0079] S602, performing principal component analysis on the foundation slurry quality parameters to obtain a principal component load matrix, a principal component score matrix, a principal component cumulative variance contribution rate, a principal component score scatter plot, and a principal component load two-dimensional plot; Specifically, after obtaining the soil slurry quality parameter data, it is necessary to perform data preprocessing, such as data standardization, missing value filling, etc., to eliminate the impact of dimension and order of magnitude differences and improve data quality and analysis reliability. Then, use statistical software or programming to implement the principal component analysis algorithm, calculate the covariance matrix or correlation matrix of the quality parameters, perform eigenvalue decomposition on it, obtain eigenvalues ​​and eigenvectors, and then calculate the principal component loading matrix, principal component score matrix, principal component cumulative variance contribution rate and other results.

[0080] The principal component load matrix reflects the correlation between the original quality parameters and the principal components. The absolute value of the load value indicates the contribution of the original parameters to the principal components, and the positive and negative signs of the load value indicate the positive and negative correlation between the original parameters and the principal components. By analyzing the load matrix, the key parameters that play a decisive role in the quality of the soil slurry can be identified, the inherent connection between different parameters can be revealed, and the basis for parameter selection can be provided for quality optimization control.

[0081] The principal component score matrix reflects the score of each soil slurry sample on each principal component, and the size of the score value indicates the relative position and importance of the sample on the principal component. By analyzing the principal component scores, cluster analysis and outlier detection can be performed on soil slurry samples, sample groups with similar quality characteristics can be identified, and individual samples with abnormal quality can be found, providing a basis for quality grading and problem diagnosis.

[0082] The cumulative variance contribution rate of the principal components reflects the ability of the first k principal components to explain the total variance of the original data. The higher the contribution rate, the more original information the first k principal components contain, and the better the dimensionality reduction effect. Usually, the first k principal components with a cumulative variance contribution rate of more than 85% are selected as the final analysis object, which can not only retain the original information to the maximum extent, but also effectively reduce the data dimension and simplify the quality analysis and evaluation process.

[0083] The principal component score scatter plot is a two-dimensional or three-dimensional scatter plot of the scores of the soil slurry samples on the first two or three principal components, which intuitively displays the distribution characteristics of the samples in the principal component space. By observing the degree of aggregation and dispersion of the sample points in the scatter plot, the similarities and differences between the samples can be found, and the sample subgroups with significantly different quality characteristics can be identified, providing graphical support for quality classification and abnormal diagnosis.

[0084] The principal component load two-dimensional diagram is a two-dimensional vector diagram that plots the loads of the original quality parameters on the first two principal components to show the geometric relationship between the parameters and the principal components. The length of the load vector indicates the importance of the parameter, the angle indicates the correlation between the parameters, and the quadrant position indicates the positive and negative correlation between the parameter and the principal component. By analyzing the load two-dimensional diagram, the parameter combination that plays a key role in the quality of the soil slurry can be intuitively identified, the promotion or inhibition effect between the parameters can be revealed, and the optimization direction can be provided for quality control.

[0085] S603, judging according to the principal component load matrix, the principal component score matrix, the principal component cumulative variance contribution rate pass rate, the preset quality parameter threshold and the judgment rule, and obtaining the foundation soil slurry quality analysis result.

[0086] Specifically, the last key step of quality assessment is to make a comprehensive judgment based on the principal component load matrix, principal component score matrix, principal component cumulative variance contribution rate, quality parameter pass rate, preset quality parameter threshold and judgment rules to obtain the foundation soil slurry quality analysis results. Through principal component analysis and parameter pass rate analysis, the principal component index reflecting the comprehensive characteristics of soil slurry quality and the pass rate index reflecting the compliance of soil slurry quality have been obtained. However, it is difficult to make an accurate judgment and grade classification of soil slurry quality based on a single indicator. It is necessary to combine the pre-set threshold standards and judgment rules, conduct a comprehensive analysis of various indicators and make multi-criteria decisions to obtain objective and reliable quality analysis results.

[0087] Based on the above embodiment, as an optional embodiment, principal component analysis is performed on the foundation soil slurry quality parameters to obtain the principal component load matrix, the principal component score matrix and the principal component cumulative variance contribution rate, including: S701, calculating the Pearson correlation coefficient between the foundation soil slurry quality parameters, and constructing a correlation coefficient matrix according to the Pearson correlation coefficient; Specifically, foundation soil slurry quality parameters usually include multiple indicators, such as fluidity, density, moisture content, segregation rate, water seepage rate, etc. There may be a certain correlation between these indicators, that is, some indicators show a positive or negative correlation trend. Understanding the correlation between quality parameters can help identify the parameter combination that plays a key role in soil slurry quality and optimize parameter selection and quality control strategies. The Pearson correlation coefficient is a statistical indicator that measures the strength of the linear correlation between two continuous variables. The value range is between -1 and 1. The larger the absolute value, the stronger the correlation. A positive value indicates a positive correlation, a negative value indicates a negative correlation, and 0 indicates no correlation.

[0088] When calculating the Pearson correlation coefficient, the soil slurry quality parameter data must first be preprocessed, such as removing outliers and missing values, to ensure the validity and representativeness of the data. Then, the Pearson correlation coefficient between any two quality parameters is calculated using the correlation function in the statistical software or programming language, such as the pandas.DataFrame.corr() function in Python, to obtain a symmetrical correlation coefficient matrix. The rows and columns of the matrix represent different quality parameters, and the matrix elements represent the correlation coefficients between the corresponding row and column parameters. The elements on the main diagonal are all 1, indicating that the parameters are completely correlated with themselves.

[0089] By analyzing the correlation coefficient matrix, we can intuitively understand the strength of the correlation and the positive and negative relationship between the quality parameters. If the correlation coefficient between two parameters is close to 1 or -1, it indicates that there is a significant linear correlation between them. In the subsequent analysis, we can consider retaining only one of the parameters to reduce data redundancy and computational complexity; if the correlation coefficient between two parameters is close to 0, it indicates that they are relatively independent and the influence of the two parameters on the soil slurry quality needs to be considered at the same time; if the correlation coefficient between multiple parameters is large, forming a correlation coefficient cluster, it indicates that these parameters may be affected by the same potential factor. We can consider extracting this potential factor as the principal component to simplify the quality evaluation model.

[0090] S702, performing eigendecomposition on the correlation coefficient matrix to obtain matrix eigenvalues ​​and matrix eigenvectors; Specifically, as a real symmetric matrix, the correlation coefficient matrix can be decomposed into the product of eigenvalues ​​and eigenvectors according to the matrix eigendecomposition theorem. The eigenvalue represents the expansion multiple of the matrix in the direction of the eigenvector, and the eigenvector represents the direction of the main change of the matrix. Performing eigendecomposition on the correlation coefficient matrix is ​​to find a set of mutually orthogonal eigenvectors so that the projection of the matrix in these directions has the largest variance, thereby revealing the main variation mode of the data.

[0091] When performing eigendecomposition, we first need to select a suitable eigendecomposition algorithm, such as the Jacobi method, power method, QR decomposition method, etc., according to the dimension and numerical characteristics of the correlation coefficient matrix. Then, we use the eigendecomposition function in linear algebra software or programming language, such as the numpy.linalg.eig() function in Python, to calculate the eigenvalues ​​and eigenvectors of the correlation coefficient matrix. The obtained eigenvalues ​​are usually arranged in order from large to small, and the corresponding eigenvectors form the column vectors of the matrix in the order of the eigenvalue size.

[0092] By analyzing the eigenvalues ​​and eigenvectors, we can understand the main modes and contributions of the coordinated changes in quality parameters. The size of the eigenvalue reflects the importance of data variation in the direction of the corresponding eigenvector. The larger the eigenvalue, the greater the difference in the data in this direction and the more information it contains. The sum of the eigenvalues ​​is equal to the trace of the correlation coefficient matrix, which reflects the total variation of the quality parameters. The components of the eigenvector represent the projection coefficients of the original quality parameters in the direction of the principal component. The larger the absolute value of the component, the greater the contribution of the parameter to the principal component. The positive and negative signs indicate the positive and negative correlation between the parameter and the principal component.

[0093] According to the analysis results of eigenvalues ​​and eigenvectors, the number and interpretation significance of principal components can be preliminarily determined. Usually, the first few eigenvectors with eigenvalues ​​greater than 1 are selected as principal components, which can not only retain the information of the original data to the maximum extent, but also effectively reduce the dimension of the data and reduce the computational complexity. For each principal component, according to the size and sign of its eigenvector component and the physical meaning of the quality parameter, it can be inferred that the principal component reflects which typical combination patterns of quality parameters, such as cohesion, fluidity, uniformity, etc., which provides a reference for the naming and interpretation of the principal component.

[0094] S703, according to the size of the matrix eigenvalue, select the matrix eigenvectors corresponding to the first k largest matrix eigenvalues ​​as the load vector of the principal component; Specifically, after eigendecomposing the correlation coefficient matrix, a set of eigenvectors sorted by eigenvalue size are obtained. According to the basic principle of principal component analysis, the larger the eigenvalue, the greater the data difference in the direction of the corresponding eigenvector, the more information it contains, and the greater the contribution to the overall variation of the quality parameters. Therefore, selecting the eigenvectors corresponding to the first k largest eigenvalues ​​as the load vectors of the principal component is to select k orthogonal directions that can best represent the difference and information of the original data as new coordinate axes, mapping the quality parameters from the original high-dimensional space to the low-dimensional space of the principal component, and realizing data dimensionality reduction and feature extraction.

[0095] When selecting the principal component load vector, we first need to determine the number of principal components k to be extracted based on the size and trend of the eigenvalues, as well as the interpretation significance and application requirements of the principal components. The following common criteria are usually used: parallel analysis method, that is, determining the optimal number of principal components based on the eigenvalue distribution of random data; scree plot method, that is, determining the cutoff position of the number of principal components based on the mutation point of the eigenvalue size; cumulative variance contribution rate method, that is, determining the minimum number of principal components based on the sum of the variance contribution rates of the first k principal components reaching a certain threshold (such as 80% or 90%).

[0096] After determining the number of principal components, select the eigenvectors corresponding to the first k largest eigenvalues ​​to form the principal component loading matrix. Each column of the loading matrix corresponds to a principal component, and the components of the column vector represent the projection coefficients of the original quality parameters on the principal component. The larger the absolute value of the component, the greater the contribution of the parameter to the principal component. According to the loading matrix, the load values ​​of each quality parameter on each principal component can be calculated, that is, the correlation coefficient between the original parameter and the principal component, which reflects the strength of the correlation between the parameter and the principal component.

[0097] The physical meaning of the principal components can be explained and named through the principal component load matrix and load values. Generally, based on the size and sign of the load value and combined with the professional knowledge of quality parameters, it is inferred that each principal component reflects which typical combination mode of quality parameters, such as fluidity, cohesion, anti-segregation, etc. The naming of the principal components should be concise, accurate and comprehensive, reflecting the mathematical characteristics of the principal components and conforming to the engineering practice of soil slurry quality evaluation.

[0098] Selecting the eigenvectors corresponding to the first k largest eigenvalues ​​as the principal component load vector can retain the information of the original quality parameters to the maximum extent, significantly reduce the dimension of the data, and reduce data redundancy and noise interference. The principal component load matrix not only reveals the inherent connection and change pattern between the quality parameters, but also provides a new indicator system and calculation basis for quality evaluation and prediction. Through the principal component score, each soil slurry sample can be quantitatively represented and comprehensively evaluated in the principal component space, intuitively reflecting the quality status and abnormality of the sample. The principal component load matrix can also guide the optimization and simplification of quality parameters, find out the key parameter combination that plays a decisive role in soil slurry quality, and optimize the quality detection and control scheme.

[0099] S704, organizing the load vectors of the selected k principal components into a principal component load matrix by columns, where each column represents a principal component; Specifically, in the previous step, according to the size of the matrix eigenvalue, the eigenvectors corresponding to the first k largest eigenvalues ​​are selected as the load vectors of the principal component. These k load vectors reflect the projection relationship of the original quality parameters in the direction of the principal component and contain the main information and patterns of data variation. In order to facilitate the mathematical representation and calculation application of the principal component, these k load vectors need to be organized into a matrix in a standardized form.

[0100] The specific steps to form the principal component loading matrix are as follows: First, arrange the k load vectors in order of eigenvalue size. The first principal component corresponds to the eigenvector of the largest eigenvalue, the second principal component corresponds to the eigenvector of the second largest eigenvalue, and so on. Then, arrange the k column vectors in order from left to right to form a matrix with n rows and k columns, where n is the number of original quality parameters. This matrix is ​​the principal component loading matrix, denoted by L. Each column of the matrix represents the load vector of a principal component, reflecting the linear combination relationship between the principal component and the original quality parameter; each row of the matrix corresponds to an original quality parameter, and the components of the row vector represent the load values ​​of the parameter on the k principal components. The principal component loading matrix is ​​an orthogonal matrix, the column vectors are mutually orthogonal, and the modulus length is 1, which meets the requirements of matrix orthogonality and unitization.

[0101] Through the principal component loading matrix, the original n-dimensional quality parameter space can be transformed into a k-dimensional principal component space to achieve data dimension reduction and feature compression. Each column of the loading matrix is ​​a coordinate axis of the principal component space, and there is a linear transformation relationship between it and the coordinate axis of the original parameter space. This transformation can map each sample point in the original data set from the parameter space to the principal component space, and obtain the score of the sample point on the principal component, that is, the principal component score. The principal component score reflects the projection length of each sample point in the principal component direction, and can be used as the coordinate of the sample point in the new space for comprehensive evaluation and similarity measurement of samples.

[0102] The principal component loading matrix is ​​not only the coordinate basis of the principal component space, but also the key matrix for calculating the principal component score. According to the mathematical principle of principal component analysis, the principal component score of a sample point can be calculated by the matrix product of the original data matrix and the loading matrix. Let the original data matrix be X, where each row represents a sample and each column represents a quality parameter, then the principal component score matrix F can be expressed as: F=XL. This formula reveals the linear transformation relationship between the original data, the loading matrix, and the principal component score, and provides a matrix operation tool for the interpretation and application of the principal component.

[0103] S705, multiplying the foundation soil slurry quality parameter by the principal component load matrix to obtain the scores of each principal component, and composing the scores of each principal component into a principal component score matrix by row; Specifically, the principal component score is the coordinate of the sample point in the principal component space, reflecting the projection length of the sample point in the principal component direction. Since the principal component is a linear combination of the original quality parameters, the principal component score can be calculated by the matrix product of the original data and the principal component load matrix. Let the original data matrix be X, where each row represents a soil slurry sample and each column represents a quality parameter; let the principal component load matrix be L, where each column represents a load vector of a principal component. Then the principal component score matrix F can be expressed as: F=XL. This matrix product operation can be implemented in mathematical software or programming languages ​​such as MATLAB and Python.

[0104] The specific steps for calculating the principal component score are as follows: First, construct a data matrix X based on the collection and recording method of the original data. Generally, the quality parameter values ​​of each soil slurry sample are arranged in the form of row vectors, and the row vectors of different samples are stacked according to the sample number or time order to form a matrix of m rows and n columns, where m is the number of samples and n is the number of parameters. Then, according to the principal component loading matrix L obtained in the previous step, the product of X and L is calculated by matrix multiplication to obtain the principal component score matrix F. The dimension of the matrix F is m rows and k columns, where k is the number of principal components extracted. Each row of F represents the principal component score vector of a sample, which contains the score value of the sample on k principal components; each column of F represents a principal component, and the column vector contains the scores of all samples on the principal component. The principal component score vector reflects the relative position and difference of the sample in the principal component space. The larger the score value, the more outstanding the performance of the sample on the principal component.

[0105] The principal component score matrix can be used to achieve comprehensive evaluation and similarity analysis of soil slurry quality. In the principal component space, each sample corresponds to a principal component score vector, which can be used for quantitative comparison and ranking of samples. The components of the score vector on each principal component reflect the performance of the sample in different quality modes. The larger the component value, the more typical the sample is in this mode. The comprehensive quality index of the sample can be obtained by the modulus length or component weighted average of the score vector, which is used for the overall evaluation and quality judgment of the sample. The distance or correlation measurement between the score vectors, such as Euclidean distance, cosine similarity, etc., can be used to analyze the similarities and differences between different samples, and discover the typical combination pattern and change trend of quality parameters.

[0106] The principal component score matrix can be used not only for the quality evaluation of a single sample, but also for the overall analysis and modeling of the sample set. By performing statistical analysis on the principal component scores, such as mean, variance, correlation, etc., the distribution characteristics and overall laws of the sample set in the principal component space can be revealed. By performing cluster analysis and discriminant analysis on the principal component scores, the internal structure and classification laws of the sample set can be discovered, providing data support for the grading and quality standard formulation of soil slurry quality. By performing regression analysis and predictive modeling on the principal component scores, the quantitative relationship between quality parameters and construction performance, strength performance, etc. can be established, providing a decision-making tool for soil slurry mix optimization and quality control.

[0107] S706, calculating the cumulative variance contribution rates of the first k principal components according to the principal component score matrix to obtain the principal component cumulative variance contribution rates.

[0108] Specifically, the variance contribution rate of the principal component reflects the proportion of each principal component in the overall variance of the original data, and is an indicator of the importance and information content of the principal component. According to the mathematical principle of principal component analysis, the variance of each principal component is equal to the corresponding eigenvalue, so the variance contribution rate of the principal component can be calculated by the proportion of the eigenvalue to the sum of the total eigenvalues. Let the covariance matrix or correlation coefficient matrix of the original data matrix be S, its eigenvalues ​​are λ1,λ2,…,λn, and the corresponding eigenvectors are e1,e2,…,en. Then the variance contribution rate vi of the i-th principal component can be expressed as: vi=λi / ∑λj, where ∑λj is the sum of all eigenvalues. The variance contribution rate of the principal component reflects the importance of the principal component in the overall variance. The larger the contribution rate, the more information the principal component contains and the stronger its representativeness of the data.

[0109] In principal component analysis, the first few principal components with larger variance contribution rates are usually selected as the main features of the data, while the subsequent principal components with smaller variance contribution rates are ignored. This is because the first few principal components usually contain most of the data information and variation characteristics, while the variance contribution rates of the subsequent principal components are very small and have limited explanatory effect on the data. In order to evaluate the cumulative explanatory power of the first k principal components selected for the original data, it is necessary to calculate their cumulative variance contribution rates. The cumulative variance contribution rate Vk of the first k principal components can be expressed as: Vk=∑vi(i=1,2,…,k), which is the sum of the variance contribution rates of the first k principal components. The cumulative variance contribution rate reflects the cumulative explanation ratio of the first k principal components to the overall variance of the original data, and is an important indicator for measuring the principal component extraction effect and information loss.

[0110] The specific steps for calculating the cumulative variance contribution of the principal component are as follows: First, calculate the variance contribution of each principal component based on the principal component loading matrix and the principal component score matrix. Each column of the principal component loading matrix corresponds to a principal component, and the square of the modulus of the column vector is equal to the eigenvalue of the principal component. Therefore, the eigenvalue of each principal component can be calculated by the modulus of the column vector of the loading matrix, and then divided by the sum of the eigenvalues ​​to obtain the variance contribution of each principal component. Alternatively, the variance contribution of each principal component can be obtained by dividing the variance of each column of the principal component score matrix by the sum of the variances of all columns. The two calculation methods are equivalent. Then, the variance contribution rates of the first k principal components are accumulated to obtain the cumulative variance contribution rate Vk. Generally, according to the research purpose and experience, the minimum number of principal components with a cumulative variance contribution rate of 80%~95% is selected as the number of principal components finally extracted.

[0111] The cumulative variance contribution rate of the principal component can not only be used to determine the number of principal components to be extracted, but also can be used for weighted calculation of the principal component score and sample evaluation. In some applications, in order to highlight the importance differences of the principal components, the variance contribution rate of each principal component can be used as a weight to perform a weighted average on the principal component scores to obtain the comprehensive score or quality index of the sample. The comprehensive score comprehensively considers the information contribution of each principal component and can more comprehensively reflect the overall characteristics and relative advantages and disadvantages of the sample. When evaluating and grading the quality of samples, the comprehensive scores of the samples can be directly compared, or the critical value or threshold of the principal component combination can be divided in combination with the cumulative variance contribution rate to achieve clustering and grading of the samples.

[0112] See also Figure 2 , Figure 2 The present invention provides a foundation soil slurry quality analysis system architecture diagram, the foundation soil slurry quality analysis system may include: The data acquisition module 1 is used to acquire multimodal data of the liquid at the drilling fluid inlet and the flowback fluid outlet by using multimodal sensors arranged at the drilling fluid inlet and the flowback fluid outlet; The data processing module 2 is used to preprocess the multimodal data to obtain preprocessed multimodal data, and to extract features from the preprocessed multimodal data to obtain multimodal feature data; The data analysis module 3 is used to extract the foundation soil slurry quality parameters according to the multimodal characteristic numbers to obtain the foundation soil slurry quality parameters, and analyze the foundation soil slurry quality according to the foundation soil slurry quality parameters to obtain the foundation soil slurry quality analysis results.

[0113] It should be noted that: when the system provided in the above embodiment realizes its functions, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0114] Please refer to Figure 3 The application also discloses an electronic device. Figure 3 The electronic device 300 may include: at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302 or end-to-end wireless communication.

[0115] The communication bus 302 is used to realize the connection and communication between these components.

[0116] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0117] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0118] Among them, the processor 301 may include one or more processing cores. The processor 301 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Optionally, the processor 301 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field~Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 301 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301, and it can be implemented separately through a chip.

[0119] Among them, the memory 305 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read~Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may optionally be at least one storage system located away from the aforementioned processor 301. Refer to Figure 3 The memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module and an application program for a method for analyzing the quality of foundation soil slurry.

[0120] exist Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain the data input by the user; and the processor 301 can be used to call the application program storing the foundation soil slurry quality analysis method in the memory 305. When executed by one or more processors 301, the electronic device 300 executes one or more methods in the above-mentioned embodiments. It should be noted that for the aforementioned method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited to the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application. In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0121] In the several implementations provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are only schematic, such as the division of modules, which is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, indirect coupling or communication connection of systems or modules, which can be electrical or other forms.

[0122] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0123] The present application also provides a computer storage medium that can store multiple instructions, which are suitable for being loaded and executed by a processor as described above. Figure 1 The foundation slurry quality analysis method of the embodiment shown in the figure can be found in the specific implementation process. Figure 1 The specific description of the illustrated embodiment will not be repeated here.

[0124] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

[0125] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: various media that can store program codes, such as USB flash drives, mobile hard drives, magnetic disks or optical disks.

[0126] The above are only exemplary embodiments of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present disclosure.

[0127] This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art not described in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for analyzing the quality of foundation slurry, characterized in that: The method comprises: Using multimodal sensors disposed at the drilling fluid inlet and the flowback fluid outlet, multimodal data of the fluid at the drilling fluid inlet and the flowback fluid outlet are obtained; Preprocessing the multimodal data to obtain preprocessed multimodal data, and extracting features from the preprocessed multimodal data to obtain multimodal feature data; The foundation soil slurry quality parameters are extracted according to the multimodal characteristic numbers to obtain the foundation soil slurry quality parameters, and the foundation soil slurry quality is analyzed according to the foundation soil slurry quality parameters to obtain the foundation soil slurry quality analysis results.

2. The method according to claim 1, characterized in that The multimodal sensor includes a spectral detector, a laser particle size analyzer and a high-speed camera; the multimodal sensor disposed at the drilling fluid inlet and the flowback fluid outlet is used to obtain multimodal data of the liquid at the drilling fluid inlet and the flowback fluid outlet, including: The spectral detectors arranged at the drilling fluid inlet and the flowback fluid outlet are used to obtain multi-spectral data of the liquid at the drilling fluid inlet and the flowback fluid outlet; The high-speed cameras arranged at the drilling fluid inlet and the flowback fluid outlet are used to obtain image data of the liquid at the drilling fluid inlet and the flowback fluid outlet; Using the laser particle size analyzer arranged at the drilling fluid inlet and the flowback fluid outlet, the particle size distribution data of the liquid at the drilling fluid inlet and the flowback fluid outlet are obtained; The multispectral data, the image data and the particle size distribution data are used as the multimodal data.

3. The method according to claim 1, characterized in that The preprocessing of the multimodal data to obtain preprocessed multimodal data includes: Using a statistical method to identify outlier points in the multimodal data, and cleaning the outlier points to obtain first multimodal data; Extracting a timestamp from the first multimodal data to obtain a timestamp of the first multimodal data, and performing time alignment on the first multimodal data according to the timestamp of the first multimodal data to obtain second multimodal data; The second multimodal data is standardized to obtain the preprocessed multimodal data.

4. The method according to claim 2, characterized in that: The step of extracting features from the preprocessed multimodal data to obtain multimodal feature data includes: Extracting features from the multispectral data to obtain multispectral feature data; Extracting features from the image data to obtain image feature data; Performing feature extraction on the particle size distribution data to obtain particle size distribution feature data; The multi-spectral feature data, the image feature data and the particle size distribution feature data are feature fused to obtain the multi-modal feature data.

5. The method according to claim 1, characterized in that The multimodal feature data includes multispectral feature fusion data, image feature fusion data and particle size distribution feature fusion data; the foundation soil slurry quality parameter extraction is performed according to the multimodal feature data to obtain the foundation soil slurry quality parameter, including: Extracting chemical characteristic values ​​related to chemical composition from the multispectral characteristic fusion data, and determining soil slurry chemical composition parameters according to the chemical characteristic values ​​and a preset soil slurry chemical composition regression equation; Extracting image features related to suspended particles from the image feature fusion data, and determining soil slurry suspended particle parameters based on the image features and a preset image feature regression equation; Extracting particle size characteristics related to rheological properties from the particle size distribution characteristic fusion data, and determining soil slurry rheological parameters according to the particle size characteristics and a preset particle size model; The soil slurry rheological parameters, the soil slurry chemical composition parameters and the soil slurry suspended particle parameters are used as the foundation soil slurry quality parameters.

6. The method according to claim 1, characterized in that The step of analyzing the foundation soil slurry quality according to the foundation soil slurry quality parameter to obtain the foundation soil slurry quality analysis result includes: Comparing the foundation soil slurry quality parameters with preset standard parameter data to determine the pass rate of the foundation soil slurry quality parameters; Performing principal component analysis on the foundation soil slurry quality parameters to obtain a principal component load matrix, a principal component score matrix, a principal component cumulative variance contribution rate, a principal component score scatter plot, and a principal component load two-dimensional plot; The foundation soil slurry quality analysis result is obtained by making a judgment based on the principal component load matrix, the principal component score matrix, the principal component cumulative variance contribution rate pass rate, the preset quality parameter threshold and the judgment rule.

7. The method according to claim 6, characterized in that The principal component analysis is performed on the foundation soil slurry quality parameters to obtain the principal component load matrix, the principal component score matrix and the principal component cumulative variance contribution rate, including: Calculating the Pearson correlation coefficient between the foundation soil slurry quality parameters, and constructing a correlation coefficient matrix based on the Pearson correlation coefficient; Perform eigendecomposition on the correlation coefficient matrix to obtain the matrix eigenvalues ​​and matrix eigenvectors; According to the size of the matrix eigenvalue, the matrix eigenvectors corresponding to the first k largest matrix eigenvalues ​​are selected as the load vectors of the principal components; The load vectors of the selected k principal components are organized into the principal component load matrix by columns, and each column represents a principal component; Multiplying the foundation soil slurry quality parameter by the principal component load matrix to obtain each principal component score, and composing the principal component score matrix by row; The cumulative variance contribution rates of the first k principal components are calculated according to the principal component score matrix to obtain the principal component cumulative variance contribution rate.

8. A foundation soil slurry quality analysis system, characterized in that: The system comprises: A data acquisition module, used to acquire multimodal data of the liquid at the drilling fluid inlet and the flowback fluid outlet by using multimodal sensors arranged at the drilling fluid inlet and the flowback fluid outlet; A data processing module, used for preprocessing the multimodal data to obtain preprocessed multimodal data, and performing feature extraction on the preprocessed multimodal data to obtain multimodal feature data; The data analysis module is used to extract foundation soil slurry quality parameters according to the multimodal feature data to obtain foundation soil slurry quality parameters, and analyze the foundation soil slurry quality according to the foundation soil slurry quality parameters to obtain foundation soil slurry quality analysis results.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes a processor, a memory and a transceiver, the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method according to any one of claims 1 to 7.

Images