Concrete vibration quality prediction method based on AWMI-Light GBM algorithm

Abstract

The invention relates to the technical field of data processing, in particular to a concrete vibration quality prediction method based on an AWMI-Light GBM algorithm. According to the method, the multi-granularity fusion AWMI algorithm and the LightGBM algorithm are combined, so that the concrete vibration quality is accurately predicted. Firstly, original vibration data including vibration characteristics and vibration quality are obtained, and a basis is provided for subsequent processing. And then, weighting processing is performed on the original data by using a multi-granularity fusion AWMI algorithm, so that the influence of key features is effectively improved, and the data is enabled to better meet actual prediction requirements. And then, the weighted data is used for training a LightGBM algorithm, and a high-precision vibration quality prediction model is obtained. And finally, by inputting real-time vibration data, the model can rapidly predict the corresponding vibration quality, and powerful support is provided for quality control in the construction process. The method not only improves the prediction accuracy, but also improves the construction efficiency, and has important practical value.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method for predicting concrete vibration quality based on an AWMI-LightGBM algorithm. Background Art

[0002] Concrete is a crucial material in the construction industry. Currently, concrete quality testing primarily focuses on properties such as strength and compactness. Traditional strength and compactness testing typically relies on construction worker experience, rebound testing, and core sampling. However, these methods have significant limitations. Complex construction environments and individual differences in construction worker experience can easily lead to errors in vibration quality assessment. Therefore, it is necessary to establish a scientific and precise concrete vibration quality testing system to accurately monitor vibration quality and dynamically adjust vibration parameters during the vibration process.

[0003] The aforementioned concrete quality tests are all conducted after concrete pouring and cannot provide a reference for quality control during concrete construction. Currently, domestic quality testing during concrete vibration relies primarily on real-time monitoring of vibration parameters and visual recognition technology. Real-time monitoring of vibration parameters typically uses sensors such as ultrasonic waves, inclination sensor lines, cameras, and satellite positioning to collect parameters such as vibration trajectory, vibration depth, and vibration inclination, and then inputs them into established vibration quality evaluation mechanisms for analysis. However, the mechanisms influencing concrete vibration quality in engineering practice are complex, and relying solely on vibration parameters cannot accurately assess concrete quality. Furthermore, in actual engineering applications, harsh operating environments can easily affect sensor accuracy, leading to detection errors. Visual recognition technology can analyze concrete surface image features and evaluate quality by comparing features such as bubble area and bubble distribution on the exposed concrete surface. While this technology has some feasibility, it requires high precision and stability from the image sensor, resulting in poor adaptability. Accurate image capture is difficult in harsh concrete vibration environments. Furthermore, visual recognition technology can only detect the concrete surface condition, limiting its ability to assess overall concrete quality.

[0004] In view of the problems of low accuracy and poor adaptability of the current concrete vibration quality assessment scheme, it is necessary to find a highly feasible concrete vibration quality prediction method to achieve accurate detection of concrete vibration quality and real-time adjustment of vibration parameters during the operation process to ensure the construction quality of the project. Summary of the Invention

[0005] In order to avoid and overcome the technical problems existing in the prior art, the present invention provides a concrete vibration quality prediction method based on the AWMI-LightGBM algorithm. The present invention can more accurately predict the concrete vibration quality.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A concrete vibration quality prediction method based on the AWMI-LightGBM algorithm includes the following prediction steps:

[0008] S1. Obtaining original vibration data during the concrete vibration process, where the original vibration data includes vibration characteristics and vibration quality;

[0009] S2, using the multi-granularity fusion AWMI algorithm to perform weighted processing on the original vibration data to obtain corresponding weighted vibration data;

[0010] S3. Use the weighted vibration data to train the LightGBM algorithm to obtain a vibration quality prediction model;

[0011] S4. Acquire real-time vibration data during the concrete vibration process, and input the real-time vibration data into a vibration quality prediction model to predict the vibration quality corresponding to the real-time vibration data.

[0012] As a further solution of the present invention, the specific content of step S2 is as follows:

[0013] S21. Select n groups of original samples (vibration characteristics, vibration quality), and construct the original data set after noise reduction processing;

[0014] S22. Calculating multi-granularity fusion mutual information and weighting functions of all categories of vibration features based on feature values ​​and vibration quality values;

[0015] S23. Based on the multi-granularity fusion mutual information and the weighting function, weighted processing is performed on the original data set to obtain weighted data, and the data is defined as weighted data.

[0016] As a further solution of the present invention, the original data set D is expressed as:

[0017] D={(x ji ,y i )|j=1,2,…,J; i=1,2,…,n};

[0018] Where x ji represents the characteristic value of the jth vibration feature in the i-th group of original samples; y i represents the value of the vibration quality in the i-th group of original samples; J represents the total number of categories of vibration features.

[0019] As a further solution of the present invention: the calculation formula of mutual information is expressed as follows:

[0020]

[0021] In the formula, I(x j ,y) represents the mutual information between the j-th vibration feature and the vibration quality y; p(x ji ,y i ) represents x ji and y i The joint probability distribution between ji ) represents x ji The marginal probability distribution of p(y i ) represents y i The marginal probability distribution of .

[0022] As a further solution of the present invention: the calculation formula of the weighting function is expressed as follows:

[0023]

[0024] Where, w(x j ) represents the weighting function of the j-th vibration characteristic; Var(x j ) represents the variance of the j-th vibration feature in the original data set; Var avg Represents the average value of the vibration characteristic variance of all categories.

[0025] As a further solution of the present invention: the calculation formula of multi-granularity fusion mutual information is expressed as follows:

[0026] When η1(I1)>max(η2(I2),η3(I3)):

[0027]

[0028] When η1(I1)≤max(η2(I2),η3(I3)):

[0029]

[0030] Where, m = 1, 2, 3, is the normalized mutual information; It is the polynomial weighted sum of the coarse-grained mutual information; α, β, γ, ρ are weight coefficients, reflecting the influence of each particle size level on the vibration quality.

[0031] As a further solution of the present invention: the calculation formula of weighted data is expressed as follows:

[0032]

[0033] Where x' ji Represents x ji The weighted data is formed after weighted processing.

[0034] As a further solution of the present invention: the vibration characteristics include concrete water-cement ratio, concrete sand ratio, concrete temperature, vibration time, vibration depth and vibration frequency; and the compressive strength is used to characterize the vibration quality of the concrete.

[0035] As a further solution of the present invention: weighted data replaces the feature values ​​of the vibration features in the original samples to form weighted samples, and each group of weighted samples constitutes a weighted data set; in the process of training the LightGBM algorithm using the weighted data set, the mean square error is used as the loss function; the calculation formula of the mean square error is expressed as follows:

[0036]

[0037] Where y i represents y i Predicted value; represents y i and y i The mean square error between .

[0038] As a further solution of the present invention: the first-order derivative of the loss function is expressed as follows:

[0039]

[0040] Where g i represents the first-order derivative value of the i-th group of weighted samples; Represents the symbol for multivariate derivatives.

[0041] As a further solution of the present invention: in the LightGBM algorithm, the calculation formula of the vibration quality prediction value is as follows:

[0042]

[0043] Where, Represents the predicted value of the t+1th round of iterative training of the LightGBM algorithm; represents the predicted value of the tth round of iterative training of the LightGBM algorithm; η represents the learning rate; Indicates the output value of the decision tree after t+1 rounds of iterative training in the LightGBM algorithm.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention realizes the accurate prediction of concrete vibration quality by combining the AWMI algorithm and the LightGBM algorithm. First, by obtaining the original vibration data, including vibration characteristics and vibration quality, a basis is provided for subsequent processing. Then, the original data is weighted by using the AWMI algorithm to effectively enhance the influence of key features, making the data more in line with actual prediction needs. Next, the LightGBM algorithm is trained using the weighted data to obtain a high-precision vibration quality prediction model. Finally, by inputting real-time vibration data, the model can quickly predict the corresponding vibration quality, providing strong support for quality control during the construction process. This method not only improves the accuracy of the prediction, but also improves the construction efficiency, and has important practical value.

[0046] 2. A detailed weighted processing process ensures data accuracy and validity. First, an original dataset is constructed, providing a foundation for subsequent calculations. Next, multi-granularity fusion mutual information and weighting functions are calculated, fully accounting for the relationship between feature values ​​and vibration quality, as well as the variability between features. Finally, based on this information, the original dataset is weighted to obtain weighted data that better meets prediction requirements. This optimization step makes the subsequent training prediction model more accurate and improves overall prediction performance.

[0047] 3. The raw dataset is clearly represented, using a matrix format to intuitively display the vibration characteristics and vibration quality values ​​for each raw sample. This representation not only facilitates data processing and analysis but also facilitates subsequent weighting and model training. Furthermore, by clarifying the total number of vibration feature categories, it provides a foundation for calculating mutual information and weighting functions, further ensuring data accuracy and validity.

[0048] 4. The mutual information calculation formula fully considers the correlation between vibration characteristics and vibration quality. By calculating the joint probability distribution and marginal probability distribution, the mutual information value between the two is obtained. This formula not only reflects the degree of influence of the characteristics on vibration quality, but also provides an important basis for subsequent weighting processing. By optimizing the calculation of mutual information, the importance of each feature can be more accurately assessed, thereby improving the accuracy of the prediction model.

[0049] 5. The weighting function calculation formula implements weighted processing of features by considering the variance and average variance of each vibration feature. This formula not only accounts for the differences between features but also fully considers the range of feature values, making the weighted data more consistent with actual prediction requirements. By optimizing the weighting function, the performance of the prediction model can be further improved, enhancing the accuracy and stability of the prediction.

[0050] 6. The weighted data calculation formula combines mutual information with a weighting function to achieve weighted processing of the raw data. This formula not only considers the correlation between features but also fully considers the importance of feature values, making the weighted data more consistent with prediction requirements. By optimizing the calculation of weighted data, the accuracy of the prediction model can be further improved, providing stronger support for quality control during the construction process.

[0051] 7. By selecting vibration characteristics such as concrete water-cement ratio, concrete sand content, concrete temperature, vibration time, vibration depth, and vibration frequency, and characterizing concrete vibration quality through compressive strength, this method can comprehensively reflect the key factors and results of the concrete vibration process. This characterization method not only meets actual construction needs but also provides a foundation for subsequent prediction model training. By optimizing the characterization of vibration characteristics and vibration quality, the accuracy and practicality of the prediction model can be further improved.

[0052] 8. By forming a weighted dataset and using the mean square error as the loss function, this method can more accurately evaluate the performance of the prediction model. The weighted dataset fully considers the importance of each feature, making model training more consistent with actual prediction requirements. Furthermore, the mean square error as a loss function can intuitively reflect the difference between predicted and actual values, providing an important basis for model optimization. By optimizing the weighted dataset and loss function, the accuracy and stability of the prediction model can be further improved.

[0053] 9. The representation of the first-order derivative of the loss function facilitates the implementation of the gradient descent algorithm during subsequent model training. By calculating the first-order derivative, we can determine the direction and step size for model parameter updates, thereby optimizing model performance. This representation not only simplifies the model training process but also improves training efficiency. By optimizing the calculation of the first-order derivative of the loss function, we can further improve the accuracy and convergence speed of the prediction model.

[0054] 10. The calculation formula for vibration quality prediction in the LightGBM algorithm gradually optimizes model performance through iterative training. Each round of iterative training adjusts the prediction results based on the previous round, bringing the predicted value closer to the actual value. At the same time, by introducing parameters such as the learning rate and the output value of the decision tree, the complexity and generalization ability of the model can be further controlled. This calculation formula not only improves the accuracy of the prediction, but also enhances the stability and robustness of the model. By optimizing the calculation method for vibration quality prediction in the LightGBM algorithm, the overall prediction performance can be further improved, providing more reliable support for quality control during the construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is the overall prediction flow chart of the present invention.

[0056] Figure 2 Detailed prediction flow chart of the present invention.

[0057] Figure 3 This is a curve chart of concrete vibration quality prediction results based on the AWMI-LightGBM model. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] See also Figure 1 and Figure 2 In an embodiment of the present invention, a method for predicting the quality of concrete vibration based on the AWMI-LightGBM algorithm includes the following steps:

[0060] 1. Obtaining the original dataset

[0061] The original vibration data during the concrete vibration process is obtained, and the original vibration data includes vibration characteristics and vibration quality.

[0062] When selecting concrete quality detection parameters, the present invention combines engineering practice and detection accuracy requirements, selects existing or easily available vibration parameters and concrete characteristic parameters to form the original data set, and uses concrete compressive strength as the detection indicator of concrete vibration quality, where the vibration parameters are vibration time and vibration depth, and the concrete characteristic parameters are concrete water-cement ratio, sand ratio and temperature.

[0063] In actual engineering, due to various reasons such as fluctuations in raw material quality, environmental factors, and inconsistent construction operations, the concrete characteristic parameters will have strong uncertainty, which in turn affects the accuracy of the training model. At the same time, due to the complex nonlinear relationship between features and target variables, especially when there are fewer input features, it will seriously affect the accuracy of the prediction model. Therefore, the present invention proposes an adaptive weighted mutual information algorithm (AWMI) based on multi-granularity fusion to preprocess the data. Compared with the traditional weighted mutual information algorithm (WMI), the AWMI based on multi-granularity fusion can consider the complex influence of global statistics and local relations of features on the vibration quality, and effectively capture the complex relationship between features and target vectors by adaptively adjusting feature weights. It has certain advantages in solving the various uncertainties and complex dependency problems in the concrete vibration process.

[0064] Concrete's water-cement ratio and sand content can be determined by sampling on-site and using specialized equipment. For example, a concrete moisture meter can be used to measure the water-cement ratio of concrete. Sand content is typically determined by laboratory analysis of the aggregate ratio in concrete samples.

[0065] Concrete temperature can be measured using a concrete thermometer or thermistor. These devices can be inserted directly into the concrete to read the temperature in real time.

[0066] The vibration time can be recorded by a timer to ensure that each vibration reaches the predetermined time.

[0067] The vibration depth can be determined by measuring the insertion depth of the vibrating rod, which usually needs to be measured manually or automatically during the vibration process.

[0068] The vibration frequency can be read through the frequency display of the vibration equipment, or calculated by measuring the number of vibrations of the vibrating rod per unit time.

[0069] 2. Calculate multi-granularity weighted vibration data

[0070] 1. Build the original dataset

[0071] The AWMI algorithm is used to weight the noise-reduced vibration data to obtain the corresponding weighted vibration data. The AWMI algorithm can obtain the complex dependency relationship between features and target variables by assigning different weights to different features.

[0072] The concrete water-cement ratio, concrete sand ratio, concrete temperature, vibration time, vibration depth, and vibration frequency are used as vibration features (input features), and the concrete compressive strength is used as vibration quality (output feature). n groups of original samples (vibration features, vibration quality) are selected to construct the original dataset; the original dataset D is expressed as:

[0073] D={(x ji ,y i )|j=1,2,…,J; i=1,2,…,n};

[0074] Where x ji represents the characteristic value of the jth vibration feature in the i-th group of original samples; y i represents the value of the vibration quality in the i-th group of original samples; J represents the total number of categories of vibration features, which is 6 in this embodiment, that is, D = {(x ji ,y i )|j=1,2,…,6;i=1,2,…,n}.

[0075] 2. Calculate multi-granularity weighted mutual information

[0076] Let the set of characteristic values ​​of the j-th vibration characteristic be x j ={x j1 ,x j2 ,…,x jn}; The mutual information calculation formula can be expressed as:

[0077]

[0078] The calculation formula of the weighting function is as follows:

[0079]

[0080] Conventional feature weighting methods only consider global features, limiting their ability to capture multi-dimensional factors influencing concrete vibration quality. For example, temperature and vibration frequency may have a combined effect on the ultimate compressive strength of concrete. Multi-granularity weighted mutual information analyzes the relationship between features and vibration quality at different granularities, assessing their importance accordingly. Here, weighted mutual information is calculated at three granularity levels: coarse, medium, and fine.

[0081] In the coarse-grained analysis, the weighted mutual information between all vibration features and vibration quality is calculated to evaluate the impact of each feature on vibration quality at the global level. The calculation formula is as follows:

[0082]

[0083] Medium-grained analysis can introduce direct or indirect relationships between unidentified vibration characteristics in concrete vibration into the model training process. Based on the physical properties of concrete vibration characteristics, the features are divided into two feature subsets: S1 = {water-cement ratio, sand content, temperature}, and S2 = {vibration depth, vibration time, vibration frequency}. For each subset, the mutual information between it and the vibration quality is calculated, and the joint contribution of the feature subsets to the vibration quality is analyzed. The calculation formula is as follows:

[0084]

[0085] Fine-grained analysis mainly examines the local relationship between features and calculates the mutual information between the vibration feature and its neighboring features, thereby discovering possible local dependencies between features. The calculation formula is as follows:

[0086]

[0087] Where x j ,x j+1 Represents the vibration characteristics x j and its neighboring features x j+1 The mutual information between them.

[0088] Using the weighted fusion method, the weighted mutual information of coarse-grained, medium-grained, and fine-grained data is weighted and summed to obtain the final multi-grained comprehensive weighted mutual information:

[0089] When η1(I1)>max(η2(I2),η3(I3)):

[0090]

[0091] When η1(I1)≤max(η2(I2),η3(I3)):

[0092]

[0093] Where, m = 1, 2, 3, is the normalized mutual information; It is the polynomial weighted sum of the coarse-grained mutual information; α, β, γ, ρ are weight coefficients, reflecting the influence of each particle size level on the vibration quality.

[0094] Based on the mutual information and the weighting function, the original data set is weighted to obtain weighted data, and the data is defined as weighted data.

[0095] The calculation formula of weighted data is as follows:

[0096]

[0097] Where x' ji Represents x ji The weighted data is formed after weighted processing.

[0098] 3. Obtaining the Vibration Quality Prediction Model

[0099] The weighted data replaces the feature values ​​of the vibration features in the original samples to form weighted samples, and each group of weighted samples constitutes a weighted data set; the LightGBM algorithm is trained using the weighted data set to obtain a vibration quality prediction model.

[0100] The LightGBM algorithm is a gradient boosting ensemble method based on decision trees. When processing regression tasks, it can gradually reduce the prediction error by training multiple decision trees. The final output of the model is the weighted prediction result of all decision trees. LightGBM uses a histogram algorithm, which occupies less memory and greatly reduces the computational cost. Compared with other algorithms, it can effectively improve computational efficiency in actual engineering applications.

[0101] The modeling idea of ​​the concrete vibration quality prediction model based on the LightGBM algorithm is as follows: input the preprocessed concrete characteristic parameters and vibration parameter data set, use the histogram algorithm to find the optimal segmentation point of the features; use the Leaf-wise leaf growth strategy with depth limitation to generate a decision tree; calculate the residual of the first round of decision trees, use the residual as the training sample of the next decision tree, and continuously fit the residual iterative training; the weighted sum of the decision trees generated in each round is obtained to obtain the final prediction model.

[0102] The LightGBM concrete vibration quality prediction model is a regression model based on a decision tree. The model hyperparameters need to be set. The relevant parameters are as follows.

[0103] 1. Generate a decision tree

[0104] Concrete vibration quality prediction is a regression problem. By using the mean square error as the loss function, the loss function of the mean square error is:

[0105]

[0106] Where y i represents y i Predicted value; represents y i and y i The mean square error between .

[0107] The gradient is the first-order derivative of the loss function with respect to the model output, namely:

[0108]

[0109] Where g i represents the first-order derivative value of the i-th group of weighted samples; Represents the symbol for multivariate derivatives.

[0110] LightGBM will traverse each feature data and calculate the gain of each possible split point to select the optimal split point. The gain is the information gain G after splitting:

[0111]

[0112] Among them, n k is the left subset of the split point; n s is the right subset of the splitting points.

[0113] By calculating the splitting gain, we can determine the improvement in model performance before and after a feature is split. We then compare the splitting gains of each feature split point to select the optimal split point. Then, among all the candidate split points, we select the feature with the highest gain for splitting. This cycle continues until the preset stopping conditions, such as the maximum tree depth or maximum number of leaves, are met, completing the decision tree construction.

[0114] 2. Model update

[0115] Repeat the decision tree generation process based on the prediction value of the generated decision tree, calculate the sample gradient after the prediction value is updated, reselect the split point to generate a new decision tree, add the newly generated decision tree to the previous round of prediction, and update the model, that is:

[0116]

[0117] Where, Represents the predicted value of the t+1th round of iterative training of the LightGBM algorithm; represents the predicted value of the tth round of iterative training of the LightGBM algorithm; η represents the learning rate; Indicates the output value of the decision tree after t+1 rounds of iterative training in the LightGBM algorithm.

[0118] This process is repeated continuously, with each new decision tree constructed based on the predictions of the previous model. This continuous updating of the model reduces the error and improves the model's predictive power. If the model's error does not improve significantly within 10 iterations, training is terminated early.

[0119] 3. Hyperparameter Optimization

[0120] Hyperparameters are parameters that need to be manually set before model training and cannot be learned by the model itself. In the LightGBM model, the choice of hyperparameters directly affects the model's prediction performance, so they need to be set through experience or experimental parameter adjustment.

[0121] Random search can effectively solve the problem of low computational efficiency of the LightGBM model due to the large parameter space by randomly sampling in a predefined parameter space and then using the loss function to find the optimal parameter combination. It is more practical than other hyperparameter optimization algorithms.

[0122] Assume that the set of hyperparameters to be optimized is:

[0123] θ=[θ1,θ2,θ3,θ4,θ5];

[0124] Among them, the hyperparameters θ1, θ2, θ3, θ4, and θ5 are the maximum number of leaves in the tree, the learning rate, the tree depth, the number of trees, and the minimum number of samples in the leaf nodes, respectively.

[0125] A 5-fold cross-validation was used to evaluate the performance of each hyperparameter combination selected by random search, and the optimal parameter combination was finally selected.

[0126]

[0127] Among them, θ best is the best hyperparameter combination; is the mean square error; θ m is the model hyperparameter combination; Θ is the hyperparameter space.

[0128] The best LightGBM model was obtained through hyperparameter tuning, and the root mean square error (RMSE) of the trained model was evaluated on the test set.

[0129]

[0130] The present invention realizes the accurate prediction of concrete vibration quality by combining the AWMI algorithm and the LightGBM algorithm. First, by obtaining the original vibration data, including vibration characteristics and vibration quality, a basis is provided for subsequent processing. Then, the original data is weighted by using the AWMI algorithm to effectively enhance the influence of key features, making the data more in line with actual prediction needs. Next, the LightGBM algorithm is trained using the weighted data to obtain a high-precision vibration quality prediction model. Finally, by inputting real-time vibration data, the model can quickly predict the corresponding vibration quality, providing strong support for quality control during the construction process. This method not only improves the accuracy of the prediction, but also improves the construction efficiency, and has important practical value.

[0131] The concrete vibration quality prediction results based on the AWMI-LightGBM model are as follows: Figure 3 As shown in the figure, it can be found that using multi-granularity fusion mutual information to preprocess concrete vibration data can effectively reduce the impact of abnormal data on the results, and the prediction results are improved. As shown in Table 1, the AWMI-LightGBM model used in this invention has higher prediction accuracy for concrete vibration quality than the prediction results of the SVM model and LSBoost model.

[0132] Table 1 Prediction accuracy of concrete vibration quality for 3 models

[0133]

[0134] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A concrete vibration quality prediction method based on AWMI-LightGBM algorithm, characterized in that: The prediction steps include: S1. Obtaining original vibration data during the concrete vibration process, where the original vibration data includes vibration characteristics and vibration quality; S2, using the multi-granularity fusion AWMI algorithm to perform weighted processing on the original vibration data to obtain corresponding weighted vibration data; S3. Use the weighted vibration data to train the LightGBM algorithm to obtain a vibration quality prediction model; S4. Acquire real-time vibration data during the concrete vibration process, and input the real-time vibration data into a vibration quality prediction model to predict the vibration quality corresponding to the real-time vibration data.

2. The concrete vibration quality prediction method based on the AWMI-LightGBM algorithm according to claim 1 is characterized in that: The specific content of step S2 is as follows: S21. Select n groups of original samples (vibration characteristics, vibration quality), and construct the original data set after noise reduction processing; S22. Calculating multi-granularity fusion mutual information and weighting functions of all categories of vibration features based on feature values ​​and vibration quality values; S23. Based on the multi-granularity fusion mutual information and the weighting function, weighted processing is performed on the original data set to obtain weighted data, and the data is defined as weighted data.

3. The concrete vibration quality prediction method based on the AWMI-LightGBM algorithm according to claim 2 is characterized in that: The original dataset D is represented as: D={(x ji ,y i )|j=1,2,…,J;i=1,2,…,n}; Where x ji represents the characteristic value of the jth vibration feature in the i-th group of original samples; y i represents the value of the vibration quality in the i-th group of original samples; J represents the total number of categories of vibration features.

4. The method for predicting concrete vibration quality based on the AWMI-LightGBM algorithm according to claim 3, characterized in that: The calculation formula of mutual information is as follows: In the formula, I(x j ,y) represents the mutual information between the j-th vibration feature and the vibration quality y; p(x ji ,y i ) represents x ji and y i The joint probability distribution between ji ) represents x ji The marginal probability distribution of p(y i ) represents y i The marginal probability distribution of .

5. The concrete vibration quality prediction method based on the AWMI-LightGBM algorithm according to claim 4 is characterized in that: The calculation formula of the weighting function is as follows: Where, w(x j ) represents the weighting function of the j-th vibration characteristic; Var(x j ) represents the variance of the j-th vibration feature in the original data set; Var avg Represents the average value of the vibration characteristic variance of all categories.

6. The method for predicting concrete vibration quality based on the AWMI-LightGBM algorithm according to claim 5, characterized in that: The calculation formula of multi-granularity fusion mutual information is as follows: When η(I1)>max(η(I2), η(I3)): When η(I1)≤max(η(I2), η(I3)): Where, is the normalized mutual information; It is the polynomial weighted sum of the coarse-grained mutual information; α, β, γ, ρ are weight coefficients, reflecting the influence of each particle size level on the vibration quality.

7. The method for predicting concrete vibration quality based on the AWMI-LightGBM algorithm according to claim 6, characterized in that: The calculation formula of weighted data is as follows: Where x' ji Represents x ji The weighted data is formed after weighted processing.

8. A concrete vibration quality prediction method based on the AWMI-LightGBM algorithm according to any one of claim 7, characterized in that: Vibration characteristics include concrete water-cement ratio, concrete sand ratio, concrete temperature, vibration time, vibration depth and vibration frequency; the vibration quality of concrete is characterized by compressive strength.

9. The method for predicting concrete vibration quality based on the AWMI-LightGBM algorithm according to claim 8, characterized in that: The weighted data replaces the feature values ​​of the vibration features in the original samples to form weighted samples, and each group of weighted samples constitutes a weighted data set. In the process of training the LightGBM algorithm using the weighted data set, the mean square error is used as the loss function. The calculation formula of the mean square error is expressed as follows: Where y i represents y i Predicted value; represents y i and y i The mean square error between .

10. The method for predicting concrete vibration quality based on the AWMI-LightGBM algorithm according to claim 9, characterized in that: The first-order derivative of the loss function is expressed as follows: Where g i represents the first-order derivative value of the i-th group of weighted samples; Represents the symbol for multivariate derivatives; In the LightGBM algorithm, the calculation formula for the vibration quality prediction value is as follows: Where, Represents the predicted value of the t+1th round of iterative training of the LightGBM algorithm; Represents the predicted value of the t-th round of iterative training of the LightGBM algorithm; η represents the learning rate; Indicates the output value of the decision tree after t+1 rounds of iterative training in the LightGBM algorithm.

Images