Gravel void ratio matching optimization method and device, equipment and storage medium

Abstract

The invention belongs to the technical field of concrete optimization, and discloses a gravel void ratio matching optimization method, device and equipment and a storage medium, and the method comprises the steps: building a training data set according to the obtained grading parameters of single-grade gravel, the obtained needle-sheet-shaped particle content index and the obtained historical void ratio data of the corresponding gravel; constructing an initial void ratio optimization model based on a convolutional neural network; inputting the grading parameters of the single-fraction broken stones and the content index of the needle-sheet-shaped particles into the void ratio optimization model to obtain a predicted void ratio; training the initial void ratio optimization model through the training data set until a target void ratio optimization model is obtained; the grading parameters and the needle-sheet-shaped indexes of the to-be-optimized broken stone are input into the target void ratio optimization model, the target void ratio predicted value is obtained, the use amount of the cementing material is optimized through the target void ratio predicted value, the association relation between the grading parameters, the needle-sheet-shaped particle content indexes and the void ratio is fused in the mode, and the matching optimization precision is improved.

Description

Technical Field

[0001] This invention relates to the field of concrete optimization technology, and in particular to a method, apparatus, equipment and storage medium for optimizing the porosity of crushed stone. Background Technology

[0002] In concrete mix design, the porosity of crushed stone is a key parameter that determines the amount of cementitious material used: too high porosity will lead to an increase in the amount of cementitious material used (more voids need to be filled), which not only increases production costs, but may also cause the concrete to shrink and crack due to excessive cementitious material; too low porosity may affect the workability of concrete (such as reduced fluidity).

[0003] Traditional porosity prediction methods rely on empirical formulas (such as Fuller gradation curves) or single-parameter regression models, which cannot comprehensively consider the complex influence of crushed stone gradation distribution and particle morphology characteristics (such as the content of needle-like and flaky particles), resulting in low matching optimization accuracy.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, and storage medium for optimizing the porosity of crushed stone, aiming to solve the technical problem of low matching optimization accuracy caused by the inability to comprehensively consider the complex influence of crushed stone gradation distribution and particle morphology characteristics.

[0006] To achieve the above objectives, the present invention provides a method for optimizing the porosity of crushed stone, the method comprising the following steps: Acquire the gradation parameters, flaky and needle-like particle content index, and historical porosity data of single-size crushed stone, and construct a training dataset based on the gradation parameters, flaky and needle-like particle content index, and historical porosity data of the corresponding crushed stone. An initial porosity optimization model is constructed based on a convolutional neural network; The gradation parameters of the single-size crushed stone and the content index of needle-like and flaky particles are input into the initial porosity optimization model to obtain the predicted porosity. The training dataset is input into the initial porosity optimization model and the model parameters are iteratively optimized through the backpropagation algorithm until the predicted porosity meets the conditions, so that the model learns the matching mapping relationship between the gradation parameters, the content index of needle-like and flaky particles and the porosity, and obtains the target porosity optimization model. The target porosity optimization model is the porosity optimization model after optimization. Obtain the gradation parameters and flaky / needle-like index of the crushed stone to be optimized, and input the gradation parameters and flaky / needle-like index of the crushed stone to be optimized into the target porosity optimization model to obtain the target porosity prediction value, and use the target porosity prediction value to optimize the amount of cementitious material.

[0007] In some embodiments, obtaining the gradation parameters, flaky and needle-like particle content index, and historical porosity data of a single-size crushed stone, and constructing a training dataset based on the gradation parameters, flaky and needle-like particle content index, and historical porosity data of the corresponding crushed stone, includes: The target particle size range of single-grade crushed stone is determined based on the concrete design strength grade, and the target particle size range is divided into several continuous particle size intervals. The mass percentage of particles in each size range was determined by standard sieve analysis to obtain the gradation parameters of single-size crushed stone. Randomly select no less than a preset number of crushed stone particles and test them one by one using a needle-and-sheet gauge. The percentage of the total mass of needle-shaped and flaky particles in the total mass of crushed stone particles is determined based on the test results to obtain the needle-shaped and flaky particle content index. Historical porosity data of crushed stone is read from the historical database, and the historical porosity data is bound with the gradation parameters and needle-like particle content of single-size crushed stone to form structured data tuples. The structured data tuples are preprocessed, and the preprocessed data is used to construct a training dataset. The data preprocessing includes outlier removal and normalization.

[0008] In some embodiments, the method constructs an initial porosity optimization model based on a convolutional neural network, including: An initial convolutional neural network architecture is established, comprising an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer has two parallel input channels, receiving a tiered parameter vector and a needle-like index vector, respectively. The tiered parameter vector and the needle-like index vector have different dimensions. The convolutional layer is configured with three convolutional blocks, each containing a 3×3 convolutional kernel, a batch normalization layer, and a ReLU activation function. The pooling layer connects a 2×2 max-pooling layer after each convolutional block with a stride of 2. The fully connected layer has two hidden layers and employs a Dropout layer. The output layer uses a single neuron output with a linear activation function. Define the loss function and optimizer configuration. The loss function uses weighted mean squared error, the optimizer is Adam optimizer, and the initial learning rate is set to 0.001. An initial porosity optimization model is constructed based on the network architecture, the loss function, and the optimizer configuration.

[0009] In some embodiments, the step of inputting the training dataset into the initial porosity optimization model and iteratively optimizing the model parameters through a backpropagation algorithm until the predicted porosity meets the conditions, so that the model learns the matching mapping relationship between gradation parameters, needle-like particle content index and porosity, and obtains the target porosity optimization model, includes: The training dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio to ensure that the distribution of the training set, validation set, and test set is consistent. Set the batch size to 32, the training rounds to 200, and calculate the validation set loss after each round. If the validation set loss does not decrease for 10 consecutive rounds, the early stopping mechanism is triggered to save the current optimal model; The weights are updated by backpropagation algorithm, and the feature extraction weights of the convolutional layer and the mapping coefficients of the fully connected layer are optimized so that the model learns the matching mapping relationship between the gradation parameters, the content index of needle-like particles and the porosity. The mean absolute error and coefficient of determination of the predicted porosity are calculated on the test set. When the mean absolute error and the coefficient of determination simultaneously meet the conditions, the model is deemed to have completed training, and the target porosity optimization model is obtained.

[0010] In some embodiments, the step of inputting the gradation parameters of the crushed stone to be optimized and the needle-like / flaky index into the target porosity optimization model to obtain a predicted target porosity value, and using the predicted target porosity value to optimize the amount of cementitious material, includes: The gradation parameters of the crushed stone to be optimized and the needle-like index are input into the target porosity optimization model to obtain several candidate porosity values ​​output by the target porosity optimization model. Select the target porosity prediction value that meets the workability requirements from the candidate porosity values; The target amount of cementitious material is calculated based on the predicted target porosity value to optimize the amount of cementitious material used. The calculation formula is as follows: Vs = V1 × (1-p) × 1.05; Where Vs represents the target amount of cementitious material, V1 represents the target amount of cementitious material, and p represents the predicted value of the target porosity.

[0011] In some embodiments, the method further includes: The real-time usage of cementitious materials is automatically collected according to a set cycle. When the fluctuation of real-time cementitious material dosage exceeds the preset ratio, the current porosity of crushed stone is obtained; The current porosity of the crushed stone is merged with the historical porosity data of the crushed stone to obtain a new training dataset; The target porosity optimization model is incrementally trained using the new training dataset, and the new porosity optimization model is deployed. The amount of cementitious material is then optimized based on the new porosity optimization model.

[0012] Furthermore, to achieve the above objectives, the present invention also proposes a crushed stone porosity matching optimization device, the crushed stone porosity matching optimization device comprising: The acquisition module is used to acquire the gradation parameters, needle-like and flaky particle content index, and historical porosity data of single-size crushed stone, and to construct a training dataset based on the gradation parameters, needle-like and flaky particle content index, and historical porosity data of the corresponding crushed stone. A building block is used to construct an initial porosity optimization model based on a convolutional neural network. The input module is used to input the gradation parameters of the single-size crushed stone and the content index of needle-like and flaky particles into the initial porosity optimization model to obtain the predicted porosity. The training module is used to input the training dataset into the initial porosity optimization model and iteratively optimize the model parameters through the backpropagation algorithm until the predicted porosity meets the conditions, so that the model learns the matching mapping relationship between the gradation parameters, the content index of needle-like particles and porosity, and obtains the target porosity optimization model. The target porosity optimization model is the porosity optimization model after optimization. The optimization module is used to obtain the gradation parameters and flaky index of the crushed stone to be optimized, and input the gradation parameters and flaky index of the crushed stone to be optimized into the target porosity optimization model to obtain the target porosity prediction value, and use the target porosity prediction value to optimize the amount of cementitious material.

[0013] In some embodiments, the acquisition module is used to determine the target particle size range of single-size crushed stone based on the concrete design strength grade, and to divide the target particle size range into several continuous particle size intervals. The mass percentage of particles in each size range was determined by standard sieve analysis to obtain the gradation parameters of single-size crushed stone. Randomly select no less than a preset number of crushed stone particles and test them one by one using a needle-and-sheet gauge. The percentage of the total mass of needle-shaped and flaky particles in the total mass of crushed stone particles is determined based on the test results to obtain the needle-shaped and flaky particle content index. Historical porosity data of crushed stone is read from the historical database, and the historical porosity data is bound with the gradation parameters and needle-like particle content of single-size crushed stone to form structured data tuples. The structured data tuples are preprocessed, and the preprocessed data is used to construct a training dataset. The data preprocessing includes outlier removal and normalization.

[0014] Furthermore, to achieve the above objectives, the present invention also proposes a crushed stone porosity matching optimization device, which includes: a memory, a processor, and a crushed stone porosity matching optimization program stored in the memory and executable on the processor. The crushed stone porosity matching optimization program is configured to implement the steps of the crushed stone porosity matching optimization method described above.

[0015] Furthermore, to achieve the above objectives, the present invention also proposes a storage medium storing a gravel porosity matching optimization program, wherein when the gravel porosity matching optimization program is executed by a processor, it implements the steps of the gravel porosity matching optimization method described above.

[0016] In this invention, the gradation parameters, flaky and needle-like particle content, and historical porosity data of single-size crushed stone are obtained, and a training dataset is constructed based on these parameters. An initial porosity optimization model is built using a convolutional neural network. The gradation parameters and flaky and needle-like particle content of the single-size crushed stone are input into the initial porosity optimization model to obtain the predicted porosity. The training dataset is then input into the initial porosity optimization model, and the model parameters are iteratively optimized using a backpropagation algorithm until the predicted porosity is obtained. The model learns the matching mapping relationship between gradation parameters, needle-like particle content index, and porosity to obtain a target porosity optimization model, which is the optimized porosity optimization model. The gradation parameters and needle-like particle index of the crushed stone to be optimized are obtained and input into the target porosity optimization model to obtain a predicted target porosity value. The predicted target porosity value is then used to optimize the amount of cementitious material. This method integrates the correlation between gradation parameters, needle-like particle content index, and porosity, improving the matching optimization accuracy. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the first embodiment of the crushed stone porosity matching optimization method of the present invention; Figure 2 This is a structural block diagram of the first embodiment of the crushed stone porosity matching and optimization device of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0020] This invention provides a method for optimizing the porosity of crushed stone, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the crushed stone porosity matching optimization method of the present invention.

[0021] In this embodiment, the crushed stone porosity matching optimization method includes the following steps: Step S10: Obtain the gradation parameters, needle-like and flaky particle content index, and historical porosity data of single-size crushed stone, and construct a training dataset based on the gradation parameters, needle-like and flaky particle content index, and historical porosity data of the corresponding crushed stone.

[0022] In this embodiment, the executing entity is a crushed stone porosity matching and optimization device, which has functions such as data processing, data communication and program execution. The crushed stone porosity matching and optimization device can be a computer terminal device or other network device, or other devices with similar functions. This embodiment does not limit the scope of the device.

[0023] It is important to note that in concrete mix design, the porosity of crushed stone is a key parameter determining the amount of cementitious material used. An excessively high porosity leads to an increase in cementitious material usage (requiring more voids to be filled), increasing production costs and potentially causing shrinkage cracking in the concrete due to excessive cementitious material. Conversely, an excessively low porosity may affect the workability of the concrete (e.g., reduced fluidity). Traditional porosity prediction methods rely on empirical formulas (such as Fuller gradation curves) or single-parameter regression models, failing to comprehensively consider the complex influences of crushed stone gradation distribution and particle morphology characteristics (e.g., the content of needle-like and flaky particles), resulting in low accuracy in matching optimization.

[0024] To address the aforementioned technical issues, this embodiment acquires the gradation parameters, flaky and needle-like particle content index, and historical porosity data of single-size crushed stone, and constructs a training dataset based on these parameters. An initial porosity optimization model is then built using a convolutional neural network. The gradation parameters and flaky and needle-like particle content index of the single-size crushed stone are input into the initial porosity optimization model to obtain a predicted porosity. The training dataset is then input into the initial porosity optimization model, and the model parameters are iteratively optimized using a backpropagation algorithm until the predicted porosity is achieved. The following conditions are met to enable the model to learn the matching mapping relationship between gradation parameters, needle-like particle content index, and porosity, thereby obtaining a target porosity optimization model. The target porosity optimization model is the optimized porosity optimization model after completion. The gradation parameters and needle-like particle index of the crushed stone to be optimized are obtained, and the gradation parameters and needle-like particle index of the crushed stone to be optimized are input into the target porosity optimization model to obtain the target porosity prediction value. The amount of cementitious material is then optimized using the target porosity prediction value. The above method integrates the correlation between gradation parameters, needle-like particle content index, and porosity, improving the matching optimization accuracy. Specifically, it can be implemented as follows.

[0025] In this specific implementation, it is necessary to first obtain the gradation parameters of single-size crushed stone, the content index of needle-like and flaky particles, and the historical porosity data of the corresponding crushed stone, and then construct a training dataset based on these data.

[0026] The specific process for obtaining the gradation parameters of the aforementioned single-size crushed stone is as follows: The target particle size range of the single-size crushed stone is determined based on the concrete design strength grade; this target particle size range is then divided into several continuous particle size intervals; the mass percentage of particles in each particle size interval is determined using standard sieve analysis to obtain the gradation parameters of the single-size crushed stone. For example, in this embodiment, the target particle size range is determined to be 5-20 mm, which is then divided into five continuous particle size intervals: 5-10 mm, 10-12 mm, 12-15 mm, 15-18 mm, and 18-20 mm. The mass percentage of particles in each interval is then determined using standard sieve analysis. For example, a crushed stone sample dried to constant weight is passed sequentially through square-hole sieves with apertures of 20mm, 18mm, 15mm, 12mm, 10mm, and 5mm. The mass of the residue on each sieve is weighed, and the particle mass percentage of each particle size interval is calculated: Pi = mi / mtotal, where Pi is the particle mass percentage of the i-th interval, mi is the mass of the residue on the sieve in that interval, and mtotal is the total mass of the sample. Finally, the gradation parameters P1, P2, P3, P4, and P5 for the above five intervals can be obtained. The number of intervals can be set according to actual needs; this embodiment does not impose any restrictions on this.

[0027] Furthermore, the process of obtaining the needle-like and flaky particle content index specifically involves randomly selecting no less than a preset number of crushed stone particles and testing them one by one using a needle-like and flaky particle gauge. Based on the test results, the percentage of the total mass of needle-like and flaky particles in the total mass of crushed stone particles is determined to obtain the needle-like and flaky particle content index. It should be noted that, in this embodiment, 100 crushed stone particles are used as an example. No less than 100 crushed stone particles are randomly selected and tested one by one using a needle-like and flaky particle gauge. If the particle length is greater than the corresponding aperture of the needle-like gauge, it is determined to be needle-like; if the thickness is less than the corresponding groove width of the flaky gauge, it is determined to be flaky. The percentage of the total mass of needle-like and flaky particles in the total mass of crushed stone particles is then calculated, A=m. 针片 / m 试样 Where A is the percentage of the total mass of needle-like and flaky particles to the total mass of crushed stone particles, i.e., the content index of needle-like and flaky particles, m 针片 and m 试样 These represent the total mass of the needle-like and flaky particles and the total mass of the 100 extracted gravel particles, respectively.

[0028] Finally, historical porosity data of crushed stone is read from the historical database, and the historical porosity data is bound with the gradation parameters and needle-like particle content of single-size crushed stone to form structured data tuples. The structured data tuples are preprocessed, and the preprocessed data is used to construct a training dataset. The data preprocessing includes outlier removal and normalization.

[0029] Step S20: Construct an initial porosity optimization model based on a convolutional neural network.

[0030] In the specific implementation, the initial gap ratio optimization model is constructed by initializing the network architecture of a convolutional neural network, which includes an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer has two parallel input channels, receiving a gradation parameter vector and a needle-like index vector, respectively. The gradation parameter vector and the needle-like index vector have different dimensions. The convolutional layer is configured with three convolutional blocks, each containing a 3×3 convolutional kernel, a batch normalization layer, and a ReLU activation function. The pooling layer connects each convolutional block to a 2×2 max-pooling layer with a stride of 2. The fully connected layer has two hidden layers and uses a Dropout layer. The output layer uses a single neuron output with a linear activation function. A loss function and optimizer configuration are defined. The loss function is a weighted mean squared error, the optimizer is the Adam optimizer, and the initial learning rate is set to 0.001. Based on the network architecture, the loss function, and the optimizer configuration, the initial gap ratio optimization model is constructed.

[0031] Step S30: Input the gradation parameters of the single-size crushed stone and the content index of needle-like and flaky particles into the initial porosity optimization model to obtain the predicted porosity.

[0032] It should be noted that the initial porosity optimization model is an untrained model, and the predicted porosity obtained has a large error with the actual porosity. In this embodiment, the gradation parameters of single-size crushed stone and the content index of needle-shaped and flaky particles are input into the initial porosity optimization model to obtain the predicted porosity. The model is then trained based on this predicted porosity.

[0033] Step S40: Input the training dataset into the initial porosity optimization model and iteratively optimize the model parameters through the backpropagation algorithm until the predicted porosity meets the conditions, so that the model learns the matching mapping relationship between the gradation parameters, the content index of needle-like particles and the porosity, and obtains the target porosity optimization model. The target porosity optimization model is the optimized porosity optimization model.

[0034] In the specific implementation, the training process of the model is as follows: the training dataset is divided into a training set, a validation set, and a test set in a ratio of 7:2:1 to ensure that the distribution of the training set, validation set, and test set is consistent; the batch size is set to 32, the training rounds are 200, and the validation set loss is calculated after each round. When the validation set loss does not decrease for 10 consecutive rounds, an early stopping mechanism is triggered to save the current optimal model. The weights are updated through the backpropagation algorithm, and the feature extraction weights of the convolutional layer and the mapping coefficients of the fully connected layer are optimized so that the model learns the matching mapping relationship between the gradation parameters, the needle-like particle content index and the porosity. The mean absolute error and the coefficient of determination of the predicted porosity are calculated on the test set. When the mean absolute error and the coefficient of determination simultaneously meet the conditions, the model is determined to have completed training, and the target porosity optimization model is obtained.

[0035] It should be noted that the formula for calculating the mean absolute error is: N is the number of samples. and The coefficient of determination between the predicted and actual values. , where p is the average true porosity.

[0036] Step S50: Obtain the gradation parameters and flaky / needle-like index of the crushed stone to be optimized, and input the gradation parameters and flaky / needle-like index of the crushed stone to be optimized into the target porosity optimization model to obtain the target porosity prediction value, and use the target porosity prediction value to optimize the amount of cementitious material.

[0037] In a specific implementation, the gradation parameters of the crushed stone to be optimized and the needle-like index are input into the target porosity optimization model to obtain several candidate porosity values ​​output by the target porosity optimization model. Select the target porosity prediction value that meets the workability requirements from the candidate porosity values; The target amount of cementitious material is calculated based on the predicted target porosity value to optimize the amount of cementitious material used. The calculation formula is as follows: Vs = V1 × (1-p) × 1.05; Where Vs represents the target amount of cementitious material, V1 represents the target amount of cementitious material, and p represents the predicted value of the target porosity.

[0038] In practical applications, for example, the amount of crushed stone used in C40 concrete is 1050 kg / m³. 3 Apparent density 2650 kg / m³ 3 →V1 = 1050 / 2650 ≈ 0.396m 3 Target porosity p = 42.0% → Vs = 0.396 × (1 - 0.42) × 1.05 ≈ 0.396 × 0.58 × 1.05 ≈ 0.243 m 3 The density of the cementitious material is 2400 kg / m³. 3 → Dosage = 0.243 × 2400 ≈ 583 kg / m³ 3 .

[0039] Furthermore, to adapt to fluctuations in raw materials (such as changes in the gradation / needle-like content due to changes in the origin of crushed stone), the model can be updated periodically. Specifically, the real-time cementitious material dosage is automatically collected according to a set cycle. When the fluctuation in the real-time cementitious material dosage exceeds a preset ratio, the current crushed stone porosity is obtained. The current crushed stone porosity is merged with the historical porosity data of the crushed stone to obtain a new training dataset. The target porosity optimization model is incrementally trained using the new training dataset, and the new porosity optimization model is deployed. The cementitious material dosage is then optimized based on the new porosity optimization model.

[0040] It should be noted that the preset ratio can be set to 5%, that is, when the real-time cementitious material dosage fluctuates by more than 5%, the current porosity of the crushed stone is obtained. The above preset ratio can also be adjusted according to actual needs, and this embodiment does not impose any restrictions on it.

[0041] In this embodiment, the gradation parameters, flaky and needle-like particle content, and historical porosity data of single-size crushed stone are obtained, and a training dataset is constructed based on these parameters. An initial porosity optimization model is built based on a convolutional neural network. The gradation parameters and flaky and needle-like particle content of the single-size crushed stone are input into the initial porosity optimization model to obtain the predicted porosity. The training dataset is input into the initial porosity optimization model, and the model parameters are iteratively optimized using a backpropagation algorithm until the predicted porosity is obtained. The model learns the matching mapping relationship between gradation parameters, needle-like particle content index, and porosity to obtain a target porosity optimization model, which is the optimized porosity optimization model. The gradation parameters and needle-like particle index of the crushed stone to be optimized are obtained and input into the target porosity optimization model to obtain a predicted target porosity value. The predicted target porosity value is then used to optimize the amount of cementitious material. This method integrates the correlation between gradation parameters, needle-like particle content index, and porosity, improving the matching optimization accuracy.

[0042] Furthermore, this embodiment of the invention also proposes a storage medium storing a gravel porosity matching optimization program, wherein when the gravel porosity matching optimization program is executed by a processor, it implements the steps of the gravel porosity matching optimization method described above.

[0043] Reference Figure 2 , Figure 2 This is a structural block diagram of the first embodiment of the crushed stone porosity matching and optimization device of the present invention.

[0044] like Figure 2 As shown, the gravel porosity matching optimization device proposed in this embodiment of the invention includes: The acquisition module 10 is used to acquire the gradation parameters, needle-like and flaky particle content index, and historical porosity data of single-size crushed stone, and to construct a training dataset based on the gradation parameters, needle-like and flaky particle content index, and historical porosity data of the corresponding crushed stone. Module 20 is used to build an initial porosity optimization model based on a convolutional neural network; Input module 30 is used to input the gradation parameters of the single-size crushed stone and the content index of needle-like and flaky particles into the initial porosity optimization model to obtain the predicted porosity; Training module 40 is used to input the training dataset into the initial porosity optimization model and iteratively optimize the model parameters through backpropagation algorithm until the predicted porosity meets the conditions, so that the model learns the matching mapping relationship between gradation parameters, needle-like particle content index and porosity, and obtains the target porosity optimization model, which is the porosity optimization model after optimization. The optimization module 50 is used to obtain the gradation parameters and flaky index of the crushed stone to be optimized, and input the gradation parameters and flaky index of the crushed stone to be optimized into the target porosity optimization model to obtain the target porosity prediction value, and use the target porosity prediction value to optimize the amount of cementitious material.

[0045] In this embodiment, the gradation parameters, flaky and needle-like particle content, and historical porosity data of single-size crushed stone are obtained, and a training dataset is constructed based on these parameters. An initial porosity optimization model is built based on a convolutional neural network. The gradation parameters and flaky and needle-like particle content of the single-size crushed stone are input into the initial porosity optimization model to obtain the predicted porosity. The training dataset is input into the initial porosity optimization model, and the model parameters are iteratively optimized using a backpropagation algorithm until the predicted porosity is obtained. The model learns the matching mapping relationship between gradation parameters, needle-like particle content index, and porosity to obtain a target porosity optimization model, which is the optimized porosity optimization model. The gradation parameters and needle-like particle index of the crushed stone to be optimized are obtained and input into the target porosity optimization model to obtain a predicted target porosity value. The predicted target porosity value is then used to optimize the amount of cementitious material. This method integrates the correlation between gradation parameters, needle-like particle content index, and porosity, improving the matching optimization accuracy.

[0046] In some embodiments, the acquisition module 10 is used to determine the target particle size range of single-size crushed stone based on the concrete design strength grade, and to divide the target particle size range into several continuous particle size intervals. The mass percentage of particles in each size range was determined by standard sieve analysis to obtain the gradation parameters of single-size crushed stone. Randomly select no less than a preset number of crushed stone particles and test them one by one using a needle-and-sheet gauge. The percentage of the total mass of needle-shaped and flaky particles in the total mass of crushed stone particles is determined based on the test results to obtain the needle-shaped and flaky particle content index. Historical porosity data of crushed stone is read from the historical database, and the historical porosity data is bound with the gradation parameters and needle-like particle content of single-size crushed stone to form structured data tuples. The structured data tuples are preprocessed, and the preprocessed data is used to construct a training dataset. The data preprocessing includes outlier removal and normalization.

[0047] In some embodiments, the construction module 20 is used to initialize the network architecture of a convolutional neural network. The network architecture includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The input layer has two parallel input channels, which respectively receive a gradation parameter vector and a needle-like index vector. The gradation parameter vector and the needle-like index vector have different dimensions. The convolutional layer is configured with three convolutional blocks in sequence. Each convolutional block contains a 3×3 convolutional kernel, a batch normalization layer, and a ReLU activation function. The pooling layer connects a 2×2 max pooling layer after each convolutional block with a stride of 2. The fully connected layer has two hidden layers and uses a Dropout layer. The output layer uses a single neuron output and the activation function is a linear function. Define the loss function and optimizer configuration. The loss function uses weighted mean squared error, the optimizer is Adam optimizer, and the initial learning rate is set to 0.001. An initial porosity optimization model is constructed based on the network architecture, the loss function, and the optimizer configuration.

[0048] In some embodiments, the training module 40 is used to divide the training dataset into a training set, a validation set, and a test set in a 7:2:1 ratio, so that the distribution of the training set, validation set, and test set is consistent. Set the batch size to 32, the training rounds to 200, and calculate the validation set loss after each round. If the validation set loss does not decrease for 10 consecutive rounds, the early stopping mechanism is triggered to save the current optimal model; The weights are updated by backpropagation algorithm, and the feature extraction weights of the convolutional layer and the mapping coefficients of the fully connected layer are optimized so that the model learns the matching mapping relationship between the gradation parameters, the content index of needle-like particles and the porosity. The mean absolute error and coefficient of determination of the predicted porosity are calculated on the test set. When the mean absolute error and the coefficient of determination simultaneously meet the conditions, the model is deemed to have completed training, and the target porosity optimization model is obtained.

[0049] In some embodiments, the optimization module 50 is used to input the gradation parameters of the crushed stone to be optimized and the needle-like index into the target porosity optimization model to obtain a number of candidate porosity values ​​output by the target porosity optimization model. Select the target porosity prediction value that meets the workability requirements from the candidate porosity values; The target amount of cementitious material is calculated based on the predicted target porosity value to optimize the amount of cementitious material used. The calculation formula is as follows: Vs = V1 × (1-p) × 1.05; Where Vs represents the target amount of cementitious material, V1 represents the target amount of cementitious material, and p represents the predicted value of the target porosity.

[0050] In some embodiments, the training module 40 is used to automatically collect the real-time amount of cementitious materials according to a set period; When the fluctuation of real-time cementitious material dosage exceeds the preset ratio, the current porosity of crushed stone is obtained; The current porosity of the crushed stone is merged with the historical porosity data of the crushed stone to obtain a new training dataset; The target porosity optimization model is incrementally trained using the new training dataset, and the new porosity optimization model is deployed. The amount of cementitious material is then optimized based on the new porosity optimization model.

[0051] This application embodiment also provides a crushed stone porosity matching optimization device, including a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other through the communication bus. The memory is used to store crushed stone porosity matching optimization programs. When the processor executes the programs stored in the memory, it implements the above-mentioned crushed stone porosity matching optimization method.

[0052] The communication bus mentioned in the aforementioned gravel porosity matching optimization device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0053] The communication interface is used for communication between the aforementioned crushed stone porosity matching and optimization equipment and other equipment.

[0054] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0055] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0056] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0057] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0058] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0059] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0060] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solution of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0061] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0062] In addition, for technical details not described in detail in this embodiment, please refer to the crushed stone porosity matching optimization method provided in any embodiment of the present invention, which will not be repeated here.

[0063] Furthermore, it should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0064] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0065] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory (ROM) / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0066] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

[0067] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

Claims

1. A method for optimizing macadam void ratio matching, characterized by, The gravel void ratio matching optimization method comprises: Obtaining grading parameters, needle and flake particle content indicators of single particle size gravel, and historical void ratio data of corresponding gravel, and constructing a training data set according to the grading parameters, the needle and flake particle content indicators and the historical void ratio data of the single particle size gravel; An initial void ratio optimization model is constructed based on a convolutional neural network; The grading parameters and the needle and flake particle content indicators of the single particle size gravel are input into the initial void ratio optimization model to obtain a predicted void ratio; The training data set is input into the initial void ratio optimization model, and the model parameters are iteratively optimized through a back propagation algorithm until the predicted void ratio meets a condition, so that the model learns the matching mapping relationship between the grading parameters, the needle and flake particle content indicators and the void ratio, and a target void ratio optimization model is obtained, which is the optimized void ratio optimization model; The grading parameters and the needle and flake indicators of the gravel to be optimized are obtained, and the grading parameters and the needle and flake indicators of the gravel to be optimized are input into the target void ratio optimization model to obtain a target void ratio prediction value, and the target void ratio prediction value is used to optimize the amount of cementitious material.

2. The method of claim 1, wherein, The method comprises: Based on the design strength grade of concrete, the target particle size range of single particle size gravel is determined, and the target particle size range is divided into a plurality of continuous particle size intervals; The particle mass proportion of each particle size interval is determined by using a standard sieve analysis method to obtain the grading parameters of single particle size gravel; Randomly extract no less than a preset number of gravel particles, and detect them one by one by using a needle and flake gauge; Based on the detection results, the total mass percentage of the needle and flake particles in the total mass of the gravel particles is determined to obtain the needle and flake particle content indicators; The historical void ratio data of the gravel is read from a historical database, and the historical void ratio data is bound with the grading parameters and the needle and flake particle content indicators of the single particle size gravel to form a structured data tuple; The structured data tuple is preprocessed, and the preprocessed data is used to construct a training data set, wherein the data preprocessing comprises outlier elimination and normalization processing.

3. The method of claim 1, wherein, The method comprises: A network architecture of an initial convolutional neural network is initialized, the network architecture comprising an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer, wherein the input layer is configured with two parallel input channels to receive a gradation parameter vector and a needle-like index vector respectively, the gradation parameter vector and the needle-like index vector have different dimensions, the convolutional layer is sequentially configured with three convolutional blocks, each convolutional block comprising a 3×3 convolutional kernel, a batch normalization layer, and a ReLU activation function, the pooling layer comprises a 2×2 maximum pooling layer connected after each convolutional block, with a step size of 2, the fully connected layer is configured with two hidden layers and uses a Dropout layer, and the output layer uses a single neuron output with a linear activation function; A loss function is defined, and an optimizer is configured, the loss function uses a weighted mean square error, the optimizer selects an Adam optimizer, and the initial learning rate is set to 0.001; An initial void ratio optimization model is constructed based on the network architecture, the loss function, and the optimizer configuration.

4. The method of claim 1, wherein, The training data set is input into the initial void ratio optimization model, and the model parameters are iteratively optimized through a back propagation algorithm until the predicted void ratio meets the condition, so that the model learns the matching mapping relationship between the gradation parameter, the needle-like particle content index, and the void ratio, and a target void ratio optimization model is obtained, comprising: The training data set is divided into a training set, a validation set, and a test set in a ratio of 7:2:1, so that the gradation distribution of the training set, the validation set, and the test set is consistent; The batch size is set to 32, and the training rounds are set to 200 rounds, and the validation set loss is calculated after each round is completed; When the validation set loss does not decrease for 10 consecutive rounds, an early stopping mechanism is triggered, and the current optimal model is saved; The weights are updated through the back propagation algorithm, and the feature extraction weights of the convolutional layer and the mapping coefficients of the fully connected layer are optimized, so that the model learns the matching mapping relationship between the gradation parameter, the needle-like particle content index, and the void ratio; The average absolute error and the determination coefficient of the predicted void ratio are calculated on the test set, and when the average absolute error and the determination coefficient both meet the conditions, it is determined that the model is trained, and a target void ratio optimization model is obtained.

5. The method of optimizing stone voids matchability of claim 1, wherein, The gradation parameter and the needle-like index of the to-be-optimized macadam are input into the target void ratio optimization model to obtain a target void ratio prediction value, and the target void ratio prediction value is used to optimize the amount of cementitious material, comprising: The gradation parameter and the needle-like index of the to-be-optimized macadam are input into the target void ratio optimization model to obtain a target void ratio prediction value, and the target void ratio prediction value is used to optimize the amount of cementitious material, comprising: The target void ratio prediction value is filtered from the plurality of candidate void ratio values to obtain a target void ratio prediction value that meets the working property requirements; The target cementitious material amount is calculated according to the target void ratio prediction value to optimize the amount of cementitious material, and the calculation formula is: Vs=V1×(1-p)×1.05; Wherein, Vs represents the target cementitious material amount, V1 represents the target cementitious material amount, and p represents the target void ratio prediction value.

6. The stone void ratio matching optimization method according to any one of claims 1 to 5, characterized by, The method further comprises: The real-time cementitious material amount is automatically collected at a set period. When it is detected that the real-time cementitious material dosage fluctuation exceeds the preset proportion, the current gap rate of the gravel is obtained; The current gap rate of the gravel is combined with the historical gap rate data of the gravel to obtain a new training data set; The new training data set is used for model incremental training of the target gap rate optimization model, and a new gap rate optimization model is deployed, and the cementitious material dosage is optimized based on the new gap rate optimization model.

7. A device for optimizing matching of stone voids, characterized by The gravel gap rate matching optimization device comprises: The acquisition module is configured to obtain the grading parameters, the needle and flake particle content index of the single particle size gravel, and the historical gap rate data of the corresponding gravel, and to construct a training data set according to the grading parameters, the needle and flake particle content index of the single particle size gravel, and the historical gap rate data of the corresponding gravel; The construction module is configured to construct an initial gap rate optimization model based on a convolutional neural network; The input module is configured to input the grading parameters and the needle and flake particle content index of the single particle size gravel into the initial gap rate optimization model to obtain a predicted gap rate; The training module is configured to input the training data set into the initial gap rate optimization model and iteratively optimize the model parameters through a back propagation algorithm until the predicted gap rate meets a condition, so that the model learns the matching mapping relationship between the grading parameters, the needle and flake particle content index, and the gap rate, and obtains a target gap rate optimization model, which is an optimized gap rate optimization model; The optimization module is configured to obtain the grading parameters and the needle and flake index of the gravel to be optimized, and input the grading parameters and the needle and flake index of the gravel to be optimized into the target gap rate optimization model to obtain a target gap rate prediction value, and optimize the cementitious material dosage by using the target gap rate prediction value.

8. The apparatus for optimization of stone void matching of claim 7, wherein, The acquisition module is configured to determine a target particle size range of the single particle size gravel based on the design strength grade of the concrete, and divide the target particle size range into a plurality of continuous particle size intervals; The standard sieve analysis method is used to determine the particle mass proportion of each particle size interval to obtain the grading parameters of the single particle size gravel; Randomly extract no less than a preset number of gravel particles, and detect each particle using a needle and flake gauge; Based on the detection results, determine the total mass percentage of the needle and flake particles in the total mass of the gravel particles to obtain the needle and flake particle content index; Read the historical gap rate data of the gravel from a historical database, bind the historical gap rate data with the grading parameters and the needle and flake particle content index of the single particle size gravel to form a structured data tuple; Perform data preprocessing on the structured data tuple, and construct a training data set using the preprocessed data, wherein the data preprocessing includes outlier removal and normalization processing.

9. A stone void matching optimization apparatus, characterized by, The gravel gap rate matching optimization device comprises a memory, a processor, and a gravel gap rate matching optimization program stored on the memory and executable on the processor, and the gravel gap rate matching optimization program is configured to implement the steps of the gravel gap rate matching optimization method according to any one of claims 1 to 6.

10. A storage medium, characterized by The storage medium has stored thereon a macadam void ratio matching optimization program which, when executed by the processor, implements the steps of the macadam void ratio matching optimization method according to any one of claims 1 to 6.

Images