CN-121997737-A - Method, device, equipment and storage medium for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction

CN121997737ACN 121997737 ACN121997737 ACN 121997737ACN-121997737-A

Abstract

The application provides a method, a device, equipment and a storage medium for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction, and relates to the technical field of concrete strength prediction. The method comprises the steps of obtaining a plurality of groups of concrete test samples containing cement, slag, fly ash, water reducer, stone and sand dosage and corresponding compressive strength actual measurement values, constructing a GA-BP neural network model, enabling the input layer node number M to be consistent with the concrete component parameter types, optimizing by using a quadratic error criterion function as an objective function through a genetic algorithm to obtain a weight and threshold optimal solution, completing model construction, inputting the model after normalization treatment of the test samples for training, normalizing the concrete component parameters to be predicted, inputting the trained model, and outputting the compressive strength predicted value. The application realizes accurate and stable prediction of the strength of the concrete with the complex mixing proportion, and solves the problem of large error of the traditional gray water ratio empirical formula.

Inventors

WANG JIANGLONG
JIANG ZHONGYANG
Xin Jiubin
LIU GUOLONG
ZHANG CHEN
GAO YANG
YANG QIAN

Assignees

甘肃省建筑设计研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260123

Claims (10)

1. A method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction is characterized by comprising the following steps: Obtaining a plurality of groups of medium-high-strength concrete test samples with different mixing ratios, wherein the high-strength concrete test samples comprise concrete composition parameters and corresponding compressive strength actual measurement values, and the concrete composition parameters comprise cement consumption, slag consumption, fly ash consumption, water reducing agent consumption, stone consumption and sand consumption; constructing a GA-BP neural network model to realize the prediction of the compressive strength of high-strength concrete, comprising: Setting the number M of the input layer nodes, the number q of the hidden layer nodes and the number L of the output layer nodes, wherein the number M of the input layer nodes is consistent with the type number of the concrete composition parameters; Taking a quadratic error criterion function as an objective function, solving to obtain an optimal solution of a weight and a threshold value through the steps of initializing, selecting, crossing, mutating and individual evaluating of a genetic algorithm, and constructing a GA-BP neural network model; after carrying out normalization treatment on the high-strength concrete test sample, inputting the high-strength concrete test sample into a constructed GA-BP neural network model to complete model training; and normalizing the composition parameters of the high-strength concrete to be predicted, inputting the composition parameters into a trained GA-BP neural network model, and outputting a concrete compressive strength predicted value.
2. The method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction as recited in claim 1, wherein the input of the i-th node of said hidden layer Calculated by the following formula: (1) in the formula, J is the number of the nodes, M is the number of the nodes; The weight value from the ith node of the hidden layer to the jth node of the input layer is obtained; a threshold value for the i node of the hidden layer; the output of the ith node of the hidden layer Calculated by the following formula: (2) in the formula, Representing the excitation function of the hidden layer.
3. The method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction as recited in claim 2, wherein the output of the kth node of the output layer The normalized value of the concrete compressive strength predicted value is calculated by the following formula: (3) in the formula, Representing an excitation function of the output layer; Representing the weight value from the kth node of the output layer to the ith node of the hidden layer; representing the threshold of the kth node of the output layer.
4. A method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction as set forth in claim 3, wherein the quadratic error criterion function is expressed as: (4) in the formula, Is a quadratic error criterion function; Is the target value of the sample; Is the number of output layer nodes.
5. The method for predicting the compressive strength of high-strength concrete based on GA-BP neural network model prediction according to claim 1, wherein the method for normalizing the high-strength concrete test sample is as follows: (5) in the formula, Is the normalized value of the parameters of the concrete composition, Is the original value of the parameters of the concrete composition, For the minimum value of the concrete composition parameter in all test samples, Is the maximum value of the concrete composition parameter in all test samples.
6. The method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction according to any one of claims 1 to 5, wherein the excitation function of the hidden layer and the excitation function of the output layer are both logarithmic S functions.
7. The method for predicting the compressive strength of high-strength concrete in GA-BP neural network model prediction according to claim 1, wherein the weight of an input layer and an implied layer, the weight of an implied layer and an output layer, the threshold of the implied layer and the threshold of the output layer are all satisfied, the weight of the input layer and the implied layer is 7 rows and 30 columns of matrixes, the 7 rows respectively correspond to input nodes of cement, slag, fly ash, water reducing agent, stone and sand, the weight of the implied layer and the output layer is 1 row and 30 columns of matrixes, and the threshold of the implied layer is 1 row and 30 columns of matrixes.
8. A high-strength concrete compressive strength prediction device based on GA-BP neural network model prediction is characterized in that the device comprises: The data acquisition module is configured to acquire a plurality of groups of medium-high-strength concrete test samples with different mix ratios, wherein the high-strength concrete test samples comprise concrete composition parameters and corresponding compressive strength actual measurement values, and the concrete composition parameters comprise cement consumption, slag consumption, fly ash consumption, water reducer consumption, stone consumption and sand consumption; a model building module configured to build a GA-BP neural network model to enable high-strength concrete compressive strength prediction, comprising: Setting the number M of the input layer nodes, the number q of the hidden layer nodes and the number L of the output layer nodes, wherein the number M of the input layer nodes is consistent with the type number of the concrete composition parameters; Taking a quadratic error criterion function as an objective function, solving to obtain an optimal solution of a weight and a threshold value through the steps of initializing, selecting, crossing, mutating and individual evaluating of a genetic algorithm, and constructing a GA-BP neural network model; after carrying out normalization treatment on the high-strength concrete test sample, inputting the high-strength concrete test sample into a constructed GA-BP neural network model to complete model training; and normalizing the composition parameters of the high-strength concrete to be predicted, inputting the composition parameters into a trained GA-BP neural network model, and outputting a concrete compressive strength predicted value.
9. An electronic device comprising a processor and a memory communicatively coupled to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method for predicting compressive strength of high-strength concrete in GA-BP-based neural network model prediction as set forth in any one of claims 1 to 7.
10. A computer-readable storage medium, wherein computer-executable instructions are stored in the computer-readable storage medium, and the computer-executable instructions, when executed by a processor, are for implementing the method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction according to any one of claims 1 to 7.

Description

Method, device, equipment and storage medium for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction Technical Field The application relates to the technical field of concrete strength prediction, in particular to a method, a device, equipment and a storage medium for predicting the compressive strength of high-strength concrete based on GA-BP neural network model prediction. Background The medium-high strength concrete has increasingly wide application in high-rise buildings, large-span structures and complex projects due to the excellent mechanical properties. The compressive strength of concrete is a key parameter for evaluating the mechanical property and engineering quality of the concrete, and directly influences the safety and reliability of the structure. Therefore, the accurate and reliable concrete compressive strength prediction model is established, and has important significance for guiding concrete mix proportion design and optimizing construction process. At present, concrete mix designs are mainly based on the general concrete mix design procedure (JGJ 55-2019) which is designed based on the empirical relationship between gray-to-water ratio and compressive strength. However, the strength of concrete is complicated by cement, fly ash, slag, water reducer, sand, stone and other components and interaction thereof, and when each component changes irregularly, a linear model based on the gray-water ratio often has difficulty in accurately reflecting the strength change rule, and has larger prediction deviation, so that the accuracy and reliability of engineering design are affected. In recent years, artificial neural networks have been introduced into the field of concrete strength prediction due to their strong nonlinear fitting ability. For example, BP neural networks have been used for strength prediction of medium-low strength concrete. However, the BP neural network is optimized by adopting a back propagation algorithm based on gradient, is easy to fall into local optimum in the training process, has low convergence speed, is sensitive to an initial weight and a threshold value, and causes insufficient model stability and larger fluctuation of a prediction result. In order to overcome the problems, researchers try to introduce genetic algorithms to optimize the initial parameters of the neural network, such as the prior art that GA-BP neural network is used to predict the compressive strength of the vegetation porous concrete. However, the model still needs to perform secondary optimization of the weight and the threshold value in the training process, the program structure is complex, and the calculation efficiency still has room for improvement. Therefore, it is necessary to provide a concrete compressive strength prediction method with higher efficiency, stability and prediction precision, so as to better meet the requirements of modern engineering on fine control of the performance of concrete materials. Disclosure of Invention The application provides a method, a device, equipment and a storage medium for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction, which are used for solving the following problems in the prior art: 1) The traditional gray water ratio model has the problems of insufficient prediction precision and larger error under the condition of complex mixing ratio; 2) The BP neural network is easy to fall into the problem of local optimum and poor convergence stability in training; 3) The existing GA-BP neural network model has the problems of complex structure, complicated optimization steps and low calculation efficiency. In a first aspect, the application provides a method for predicting compressive strength of high-strength concrete based on GA-BP neural network model prediction, which comprises the following steps: Obtaining a plurality of groups of medium-high-strength concrete test samples with different mixing ratios, wherein the high-strength concrete test samples comprise concrete composition parameters and corresponding compressive strength actual measurement values, and the concrete composition parameters comprise cement consumption, slag consumption, fly ash consumption, water reducing agent consumption, stone consumption and sand consumption; constructing a GA-BP neural network model to realize the prediction of the compressive strength of high-strength concrete, comprising: Setting the number M of the input layer nodes, the number q of the hidden layer nodes and the number L of the output layer nodes, wherein the number M of the input layer nodes is consistent with the type number of the concrete composition parameters; Taking a quadratic error criterion function as an objective function, solving to obtain an optimal solution of a weight and a threshold value through the steps of initializing, selecting, crossing, mutating and individual evaluating of a