Grouting construction parameter prediction method and device based on BP neural network

Abstract

The invention provides a grouting construction parameter prediction method and device based on a BP neural network, and relates to the field of data processing. In the method, original engineering data for a target engineering is obtained, and the original engineering data is any one of soil density, permeability coefficient, equipment initial pressure, grouting amount and environment temperature; performing data processing on the original engineering data to obtain a to-be-input vector; inputting a vector to be input into the BP neural network model to obtain a prediction result; and carrying out reverse normalization processing on the prediction result to obtain grouting construction parameter prediction quantity. By implementing the technical scheme provided by the invention, the accuracy of grouting construction parameter prediction can be improved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly relates to a method and device for predicting grouting construction parameters based on a BP neural network. Background Art

[0002] In the field of grouting construction, the selection of construction parameters has a crucial impact on project quality and construction efficiency.

[0003] Currently, grouting construction parameters (such as grouting pressure, grouting volume, grouting speed, etc.) often rely on experience and manual judgment to determine. However, due to the fact that the construction site is often affected by various factors such as soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, traditional manual experience is difficult to capture the complex interactions between these factors, resulting in relatively low accuracy of prediction results.

[0004] Therefore, there is an urgent need for a method and device for predicting grouting construction parameters based on a BP neural network. Summary of the Invention

[0005] This application provides a method and device for predicting grouting construction parameters based on a BP neural network, which is convenient for improving the accuracy of predicting grouting construction parameters.

[0006] In the first aspect of this application, a method for predicting grouting construction parameters based on a BP neural network is provided. The method includes: obtaining the original project data for the target project, where the original project data is any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature; performing data processing on the original project data to obtain an input vector to be input; inputting the input vector to be input into the BP neural network model to obtain a prediction result; and performing inverse normalization processing on the prediction result to obtain the predicted value of the grouting construction parameter.

[0007] By adopting the above technical solution, by obtaining engineering data such as soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, a data-driven BP neural network model can be constructed, which can more accurately reflect the non-linear relationship between engineering parameters, improve the prediction accuracy, and is more scientific and stable compared to the method relying on manual experience. The traditional manual judgment method is easily affected by the experience level and subjective judgment of construction personnel, while this method uses a BP neural network for automatic calculation, which can reduce human errors and improve the consistency and reliability of construction parameter prediction. Since the BP neural network has a strong non-linear mapping ability, this method can effectively predict parameters under different soil conditions and different environmental factors, making the construction method have stronger adaptability and generalization ability. By performing normalization preprocessing on the original data, the numerical magnitude differences between different engineering parameters can be eliminated, the stability and convergence speed of the model can be enhanced, and the training and prediction processes of the neural network can be ensured to be more efficient. Therefore, it is convenient to improve the accuracy of grouting construction parameter prediction.

[0008] Optionally, the data processing of the original engineering data to obtain the input vector to be input specifically includes: performing missing value processing and outlier detection on the original engineering data to obtain original engineering features; performing normalization processing on the original engineering features by using a preset normalization formula to obtain the input vector to be input.

[0009] By adopting the above technical solution, through missing value processing and outlier detection, incorrect data can be effectively identified and corrected, avoiding deviation in model training caused by data missing or anomalies, thereby improving the stability and prediction accuracy of the model. Processing outliers can remove the interference of extreme data points on model training, make the input data more conform to the actual construction law, improve the generalization ability of the neural network, and enable it to be applicable to more engineering scenarios. By using a preset normalization formula, feature data with different dimensions can be converted to the same numerical range, reducing the numerical differences between data, preventing certain features from dominating model training due to large numerical values, helping to accelerate the convergence of the BP neural network, and improving the training efficiency.

[0010] Optionally, the preset normalization formula is calculated using the following formula: ; where, x i represents the value of the i-th original engineering feature, including any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, x min and x max are respectively the minimum and maximum values of the i-th original engineering feature in the dataset, used to determine the value range of this feature, and x i norm is the input vector to be input, used to represent the value after normalization.

[0011] By adopting the above technical solution, the soil density, permeability coefficient, and initial equipment pressure are converted to the same numerical range using a normalization formula, avoiding the influence of dimensional differences between features on model training and making the optimization process of the BP neural network more stable. If some feature values are much larger than others, the unnormalized input may cause some features to dominate the training of the network, affecting the learning ability of the model. The normalization process ensures that each feature has an equal impact on the model, improves the generalization ability of the network, and makes it applicable to different construction environments. The normalized input data is more concentrated in the numerical range, avoiding large gradient fluctuations, enabling the gradient descent algorithm of the BP neural network to converge faster, reducing the training time, and improving the calculation efficiency. By dynamically calculating the minimum and maximum values of the features, the normalization method can automatically adapt to different data sets without the need for manual adjustment of the data range, improving the applicability of this method in different projects.

[0012] Optionally, inputting the to-be-input vector into the BP neural network model to obtain a prediction result specifically includes: calculating the to-be-input vector using a preset forward propagation formula to obtain the prediction result; The preset forward propagation formula is specifically as follows: ; where x i norm is the to-be-input vector, used to represent the normalized value, n is the number of neurons in the input layer, that is, the number of input features, is the weight of the first hidden layer, representing the connection weight from the i-th feature in the input layer to the j-th neuron in the first hidden layer, determining the contribution of this feature to this neuron, is the bias of the first hidden layer, used to adjust the activation value of the j-th neuron in the first hidden layer, making the output of the neuron more robust, is the activation function, m 1 is the number of neurons in the first hidden layer, is the weight of the second hidden layer, representing the connection weight from the j-th neuron in the first hidden layer to the k-th neuron in the second hidden layer, is the bias of the second hidden layer, used to adjust the activation value of the k-th neuron in the second hidden layer, m 2 is the number of neurons in the second hidden layer, is the weight of the output layer, representing the weight from the k-th neuron in the second hidden layer to the output layer, is the bias of the output layer, used to adjust the final prediction value, is the prediction result.

[0013] By adopting the above technical solution, through the preset forward propagation formula, the normalized input data is calculated layer by layer, ensuring that the model can effectively extract the non-linear relationship between features, making the prediction result more accurate and overcoming the limitations of the traditional empirical method. The weights and biases of the first hidden layer and the second hidden layer enable the model to perform complex feature transformations, thereby better learning and fitting the non-linear mapping of grouting construction parameters and improving the prediction ability. The multi-hidden layer structure provides a deeper feature extraction ability, enabling the BP neural network to adapt to different working conditions and improving the adaptability and generalization ability of the model. The introduction of the activation function enables the neural network to learn complex non-linear relationships, thereby overcoming the limitations of traditional linear models and improving the accuracy of grouting parameter prediction. The weights reflect the importance of features in the neural network, ensuring that each input feature can reasonably contribute to the final prediction result. The bias is used to adjust the activation value of the neuron, making the model more robust, capable of adapting to different data distributions and improving the calculation stability. Through layer-by-layer calculation, feature information is gradually extracted and combined, making the network calculation more hierarchical, improving the calculation efficiency, reducing error propagation and improving the prediction accuracy.

[0014] Optionally, calculating the prediction result by using the preset forward propagation formula for the to-be-input vector specifically includes: inputting the to-be-input vector into the first hidden layer of the BP neural network model to calculate a first result; using the first result as an input to be passed to the second hidden layer of the BP neural network model to calculate a second result; and inputting the second result into the output layer of the BP neural network to calculate the prediction result.

[0015] By adopting the above technical solution, through layer-by-layer calculation, the neural network can gradually extract and combine the information of the input features, enabling the model to learn the pattern of the input data more deeply and improving the prediction accuracy. The first hidden layer mainly completes the preliminary feature extraction and transformation, mapping the input data to a high-dimensional space. The second hidden layer further performs complex non-linear transformations on the data, enhancing the model's understanding of the input features, improving the prediction effect and making it applicable to the prediction of grouting construction parameters under different working conditions. The calculation results are transmitted layer by layer, avoiding information loss, ensuring that each layer can make full use of the output of the previous layer, improving the calculation efficiency and reducing the accumulation of calculation errors at the same time. Through the design of two hidden layers, compared with the single hidden layer structure, it can better adapt to complex engineering environments, improve the adaptability of the model under different construction conditions, and make the prediction result more stable and accurate. The clear hierarchical structure makes the model design more transparent, and the number of hidden layers, the number of neurons and the activation function can be adjusted according to different construction scenarios, so as to optimize the model performance and improve the engineering applicability.

[0016] Optionally, the inverse normalization process is performed on the prediction result to obtain the predicted grouting construction parameter, which specifically includes: obtaining the parameters of the preset normalization formula; constructing an inverse normalization formula according to the parameters; and performing the inverse normalization process on the prediction result by using the inverse normalization formula to obtain the predicted grouting construction parameter.

[0017] By adopting the above technical solution, the inverse normalization process can convert the normalized prediction result back to the actual physical quantity, enabling construction personnel to directly use the prediction result for engineering operations and improving the usability and guiding value of the prediction result. Since the input data is normalized before training, if the prediction result is not inverse-normalized, numerical distortion will occur. Therefore, inverse normalization is used to ensure that the numerical range of the predicted data is consistent with the original engineering data, prevent information loss, and make the prediction result more meaningful in engineering. The inverse normalization process constructs a traceable conversion formula based on the known normalization parameters, making the entire data processing process clear and transparent, facilitating subsequent optimization and adjustment, and improving the interpretability of the model. By dynamically obtaining the parameters of the normalization formula, it can adapt to the data ranges of different construction projects, making this method applicable not only to specific projects but also to different grouting construction scenarios, improving the adaptability and generality of the model. Inverse normalization can prevent the direct output of abnormal values caused by normalization calculations, thereby improving the stability and reliability of the prediction result.

[0018] Optionally, the method further includes: generating notification data according to the predicted grouting construction parameter; and sending the notification data to the user equipment carried by the user corresponding to the target project, so as to display the notification data to the user through the user equipment.

[0019] By adopting the above technical solution, by automatically generating notification data and sending it to the user equipment, the predicted grouting construction parameters can be provided to construction personnel in real time, avoiding manual query and calculation, and improving the timeliness of construction decision-making. Construction personnel can directly receive prediction information on user equipment such as mobile phones and tablets without manually checking the prediction result, facilitating the acquisition of key data anytime and anywhere, and improving the convenience and efficiency of construction management. By automatically calculating and pushing the prediction parameters by the system, errors that may occur during manual filling and transmission are avoided, improving the accuracy of the data, ensuring that the construction is carried out according to the optimal parameters, and enhancing the grouting quality and construction safety. This method can push personalized prediction data for different projects or users in different positions, enabling management personnel and construction personnel to obtain the information they need and improving the collaborative work efficiency. The construction environment is complex. If the parameter transmission is not timely, it may lead to operation delays or construction quality problems. This method ensures that the prediction result can be pushed immediately, reduces information lag, and improves the accuracy of construction progress management.

[0020] In a second aspect of the present application, a device for predicting grouting construction parameters based on a BP neural network is provided. The device is characterized in that it includes an acquisition module (31) and a processing module (32). Among them, the acquisition module (31) is used to acquire the original project data for the target project, and the original project data is any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature; the processing module (32) is used to perform data processing on the original project data to obtain an input vector to be input; the processing module (32) is further used to input the input vector to be input into the BP neural network model to obtain a prediction result; the processing module (32) is further used to perform anti-normalization processing on the prediction result to obtain a predicted value of the grouting construction parameter.

[0021] In a third aspect of the present application, an electronic device is provided. The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. Both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory so that the electronic device executes the method described above.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions, and when the instructions are executed, the method described above is executed.

[0023] In summary, one or more technical solutions provided in the present application have at least the following technical effects or advantages: By acquiring engineering data such as soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, and constructing a data-driven BP neural network model, the nonlinear relationship between engineering parameters can be more accurately reflected, improving the prediction accuracy, and being more scientific and stable compared to the method relying on manual experience. The traditional manual judgment method is easily affected by the experience level and subjective judgment of construction personnel, while this method uses a BP neural network for automatic calculation, which can reduce human errors and improve the consistency and reliability of the prediction of construction parameters. Since the BP neural network has a strong nonlinear mapping ability, this method can effectively predict parameters under different soil conditions and different environmental factors, making the construction method have stronger adaptability and generalization ability. By performing normalization preprocessing on the original data, the numerical magnitude difference between different engineering parameters can be eliminated, enhancing the stability and convergence speed of the model, and ensuring that the training and prediction processes of the neural network are more efficient. Therefore, it is convenient to improve the accuracy of predicting grouting construction parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic flowchart of a method for predicting grouting construction parameters based on a BP neural network provided by an embodiment of the present application; Figure 2 Another schematic flow diagram of a grouting construction parameter prediction method based on a BP neural network provided by an embodiment of the present application; Figure 3 A schematic module diagram of a grouting construction parameter prediction device based on a BP neural network provided by an embodiment of the present application; Figure 4 A schematic structural diagram of an electronic device provided by an embodiment of the present application.

[0025] Explanation of reference numerals: 31, acquisition module; 32, processing module; 41, processor; 42, communication bus; 43, user interface; 44, network interface; 45, memory. Detailed implementation manners

[0026] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0027] In the description of the embodiments of the present application, words such as "for example" or "for instance" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "for example" or "for instance" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Specifically, the use of words such as "for example" or "for instance" is intended to present relevant concepts in a specific manner.

[0028] In the description of the embodiments of the present application, the meaning of the term "a plurality" refers to two or more. For example, a plurality of systems refers to two or more systems, and a plurality of screen terminals refers to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the technical features indicated. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0029] In the field of grouting construction, the reasonable selection of construction parameters plays a crucial role in project quality and construction efficiency.

[0030] Currently, the grouting construction parameters (such as grouting pressure, grouting volume, grouting speed, etc.) mainly rely on the experience of construction workers and manual judgment. However, due to the complex construction site environment, the interaction of various factors such as soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, it is difficult for the traditional empirical method to accurately describe the complex relationship between these factors, often resulting in insufficient accuracy in parameter selection and affecting the construction effect and project quality.

[0031] To solve the above technical problems, this application provides a grouting construction parameter prediction method based on the BP neural network. Refer to Figure 1 , Figure 1 which is a schematic flow chart of a grouting construction parameter prediction method based on the BP neural network provided by an embodiment of this application. This method is applied to a server and includes steps S110 to S140. The above steps are as follows: S110. Obtain the original project data for the target project, where the original project data is any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature.

[0032] Specifically, the server will obtain the original project data for a specific project, and these data are the basic parameters used to guide the grouting operation during the construction process. The original project data can be any one or more of the following data: Soil density: It represents the ratio of the mass to the volume of the soil and is an important index for judging the properties of the soil layer. Permeability coefficient: It reflects the ability of water to flow in soil or rock and affects the diffusion and consolidation effect of the grouting slurry. Initial equipment pressure: It refers to the preset pressure value of the grouting equipment before the start of construction and determines the state of the slurry before entering the soil body. Grouting volume: It refers to the volume of the slurry injected into the soil body during the construction process and is an important parameter to ensure the reinforcement effect of the project. Environmental temperature: It reflects the temperature condition at the construction site, and the temperature may affect the fluidity and curing time of the slurry. After the server obtains these data, it can provide basic information for subsequent data processing, model training, and prediction analysis, so as to achieve automatic and refined grouting parameter prediction.

[0033] For example, assume that in a diaphragm wall project, grouting reinforcement is required. Multiple sensors and monitoring devices are arranged on-site, including: a sensor measures the soil density, and the recorded value is 2.1 g / cm³. Another sensor monitors the permeability coefficient, and the recorded value is 0.15 mm / s. The control system of the construction equipment shows that the initial pressure of the equipment is 0.95 MPa. The on-site operation system records the grouting volume as 48 L. The environmental temperature sensor shows that the current temperature is 25.5 °C. The server automatically collects these data through the network to form the original engineering dataset. Whether the actually collected data is a single parameter (such as only obtaining the soil density) or a combination of multiple parameters, it will serve as an important basis for the subsequent processing and input of the prediction model. In this way, the subsequent BP neural network model can use these data for normalization processing, feature extraction, and parameter prediction, providing accurate grouting construction parameter suggestions for the construction personnel.

[0034] S120. Perform data processing on the original engineering data to obtain the vector to be input.

[0035] Specifically, the server will perform a series of data preprocessing operations on the original engineering data (such as soil density, permeability coefficient, etc.) obtained from the target project, converting the original data into a standardized data format suitable for input into the subsequent BP neural network, that is, forming a vector to be input. This process includes the following steps: Missing value processing: Check whether there is missing data in the original data and fill it using appropriate methods (such as mean, median filling, etc.). Outlier detection: Identify and process abnormal or incorrect data in the data to ensure the accuracy and consistency of the data. Uniformly convert the original data with different dimensions (for example, the unit of soil density may be g / cm³, while the unit of grouting volume may be L) to the same numerical range (usually 0 to 1) for efficient training of the subsequent model. Combine the processed multiple engineering data (such as soil density, permeability coefficient, initial pressure of the equipment, etc.) into a vector, and this vector is the so-called "vector to be input". This vector can reflect various key feature information of the engineering site as the input of the model.

[0036] In a possible implementation manner, performing data processing on the original engineering data to obtain the vector to be input specifically includes: performing missing value processing and outlier detection on the original engineering data to obtain the original engineering features; using a preset normalization formula to perform normalization processing on the original engineering features to obtain the vector to be input.

[0037] Specifically, in actual engineering data, data missing may occur due to various reasons (such as sensor failures or data transmission errors). By processing the missing values in the original engineering data, the missing data can be filled with the mean, median, or other methods to ensure that each engineering feature has a complete data record. Sometimes, there may be obviously unreasonable or out-of-normal-range values in the data (for example, an extremely low or high value suddenly appears in the soil density). If these data are not processed, they may have a negative impact on the model. The purpose of outlier detection is to identify and remove or correct these abnormal data, so as to obtain accurate and reliable original engineering features. The data dimensions of different engineering features may vary greatly. For example, the soil density (g / cm³) and the grouting volume (L) are at different orders of magnitude. To make each feature have a similar influence during model training, it is necessary to scale them to the same numerical range. The advantage of doing this is to ensure high-quality and consistent input data, eliminate the adverse effects brought by different dimensions, and thus improve the model training efficiency and prediction accuracy.

[0038] In a possible implementation manner, the preset normalization formula is calculated using the following formula: ; where, x i represents the value of the i-th original engineering feature, including any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and ambient temperature, x min and x max are respectively the minimum and maximum values of the i-th original engineering feature in the dataset, used to determine the value range of this feature, and x i norm is the vector to be input, used to represent the value after normalization.

[0039] Specifically, x i represents the numerical value of the i-th original engineering feature. For example, this feature can be any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, or ambient temperature. x min and x max are respectively the minimum and maximum values of the i-th engineering feature in the entire dataset. They determine the value range of this feature and are used as reference values for standardizing the data during the normalization process. The function of this formula is to convert the original numerical values to a unified numerical interval, so that the data scales of all features are consistent. x i norm represents the value after normalization, which constitutes the corresponding component in the vector to be input and is used by the subsequent neural network model.

[0040] For example, suppose there is an engineering dataset that contains the feature of "soil density" with the unit of g / cm 3。Assume that in the dataset, the minimum value of soil density is 1.8 g / cm 3 , and the maximum value is 2.5 g / cm 3 。If the soil density at a certain measurement point is 2.1 g / cm 3 , then using the normalization formula, the result is 0.429. In this way, the original soil density of 2.1 g / cm 3 is normalized to approximately 0.429 and becomes an element corresponding in the input vector. If other features (such as permeability coefficient, initial pressure of the equipment, etc.) are also normalized in the same way according to their respective minimum and maximum values, finally, these normalized values can be combined into a unified input vector to be input, for the neural network model to perform calculations and predictions. The advantage of this normalization process is that the numerical ranges of different features are unified, so that the model will not cause training deviation due to the value of a certain feature being too large or too small during training, thus improving the convergence speed and prediction accuracy of the model.

[0041] S130. Input the input vector to be input into the BP neural network model to obtain a prediction result.

[0042] Specifically, the server passes the preprocessed (such as normalized) input data (input vector to be input) into the BP neural network model, and then through the forward propagation calculation inside the model, generates a result for describing the prediction of engineering parameters. In other words, the input vector to be input contains various normalized feature data of the engineering site, and the BP neural network uses its internal weights, biases, and activation functions to convert these data into the final predicted values. In the previous steps, the original engineering data (such as soil density, permeability coefficient, initial pressure of the equipment, grouting volume, ambient temperature) after data cleaning, outlier detection, and normalization processing constitute a standardized input vector. For example, the vector may be [0.43, 0.40, 0.375, 0.4, 0.55], where each number corresponds to an engineering feature. The server passes this input vector to be input as input data to the BP neural network. The BP neural network usually includes an input layer, a hidden layer, and an output layer. In the input layer, each component of the vector directly corresponds to a neuron. In the first hidden layer, each neuron will calculate the product of the input vector and its corresponding weight and add the bias, and then through the activation function conversion, generate the output of the first layer. Next, the output of the first hidden layer is passed as input to the second hidden layer, and the same processing of weighted summation, adding bias, and activation function is performed. Finally, the output of the second hidden layer enters the output layer, and the output layer performs the final calculation to obtain one or more predicted values (according to the needs of the actual problem), which is the prediction result. The prediction result may be a normalized value, representing the model's prediction of grouting construction parameters (such as grouting pressure). If necessary, it can also be converted back to the physical quantity in the actual project through the inverse normalization step.

[0043] In a possible implementation, the input vector to be input is input into the BP neural network model to obtain a prediction result, which specifically includes: calculating the input vector to be input using a preset forward propagation formula to obtain the prediction result; The preset forward propagation formula is specifically as follows: ; where x i norm is the input vector to be input, used to represent the normalized value, n is the number of neurons in the input layer, that is, the number of input features, is the weight of the first hidden layer, representing the connection weight from the i-th feature in the input layer to the j-th neuron in the first hidden layer, determining the contribution of this feature to this neuron, is the bias of the first hidden layer, used to adjust the activation value of the j-th neuron in the first hidden layer, making the output of the neuron more robust, is the activation function, m 1 is the number of neurons in the first hidden layer, is the weight of the second hidden layer, representing the connection weight from the j-th neuron in the first hidden layer to the k-th neuron in the second hidden layer, is the bias of the second hidden layer, used to adjust the activation value of the k-th neuron in the second hidden layer, m 2 is the number of neurons in the second hidden layer, is the weight of the output layer, representing the weight from the k-th neuron in the second hidden layer to the output layer, is the bias of the output layer, used to adjust the final prediction value, is the prediction result.

[0044] Specifically, the input vector to be input is an engineering data vector after normalization processing, and each element represents a standardized engineering feature (such as soil density, permeability coefficient, etc.). After normalization, each feature value falls within a unified numerical range (usually 0 to 1). The input layer and the parameter represent the number of neurons in the input layer, that is, the number of input features. For example, if there are 5 original features, then n = 5. The weight is the connection weight from the i-th neuron in the input layer to the j-th neuron in the first hidden layer, used to represent the influence degree of the input feature on the output of this hidden neuron. The bias is the bias term of each neuron in the first hidden layer, used to adjust the calculated weighted sum, making the input of the neuron in the activation function more appropriate, thereby improving the robustness of the output. The activation function performs a non-linear transformation on the weighted sum of each neuron in the first hidden layer. Commonly used activation functions include ReLU, tanh, or Sigmoid. The number of neurons in the first hidden layer represents the total number of neurons in the first hidden layer.

[0045] Among them, the above content elaborates in detail the forward propagation process of the BP neural network. Starting from the normalized input vector, it successively passes through the input layer, the first hidden layer, the second hidden layer to the output layer. Each layer performs calculations through preset weights, biases, and activation functions, and finally obtains the prediction result. This hierarchical calculation method helps to gradually extract and combine the information of the input features, thereby more accurately capturing complex non-linear relationships and improving the accuracy of grouting construction parameter prediction.

[0046] In a possible implementation, a preset forward propagation formula is used to calculate the input vector to be processed to obtain the prediction result, specifically including: inputting the input vector to be processed into the first hidden layer of the BP neural network model to calculate the first result; using the first result as the input and passing it to the second hidden layer of the BP neural network model to calculate the second result; inputting the second result into the output layer of the BP neural network to calculate the prediction result.

[0047] Specifically, the input vector to be processed, such as the engineering data vector after normalization processing. This vector is fed into the first hidden layer of the BP neural network. In this layer, each neuron will perform a weighted sum of each element of the input vector and its corresponding weight, add the bias, and finally perform a non-linear transformation through an activation function (such as ReLU, Sigmoid, or tanh). The values output after each neuron's calculation constitute the output vector of the first hidden layer, that is, the first result. The second hidden layer inputs the output vector of the first hidden layer, that is, the first result. Using the first result as the input and passing it to the second hidden layer. Similar to the first hidden layer, each neuron in the second hidden layer will also perform a weighted sum of the input value and its own weight, add the bias, and undergo a non-linear transformation through the activation function. The output of each neuron constitutes the output vector of the second hidden layer, that is, the "second result". The output layer inputs the output vector of the second hidden layer, that is, the second result. The output layer usually uses linear activation (or does not use an activation function), performs a final weighted sum and adds the bias to the second result to obtain the final prediction result of the model. The prediction result reflects the prediction of the BP neural network on the target parameters (such as grouting construction parameters) based on the input data.

[0048] In a possible implementation, assume n = 5 (5 input features), and assume the first hidden layer has m 1 = 10 neurons, and the second hidden layer has m 2 = 8 neurons. is the activation function (such as ReLU or tanh).

[0049] Calculation in the first hidden layer, for the j-th neuron: ; Among them, is the calculation result of the first hidden layer, is the output vector of the first hidden layer, is the weight of the first hidden layer, representing the connection weight from the i-th feature of the input layer to the j-th neuron of the first hidden layer, which determines the contribution of this feature to this neuron. is the bias of the first hidden layer, used to adjust the activation value of the j-th neuron of the first hidden layer, making the output of the neuron more robust. is the activation function, usually selected or , introducing non-linearity. m 1 is the number of neurons in the first hidden layer.

[0050] Calculation for the second hidden layer, for the k-th neuron: ; where, is the calculation result of the second hidden layer, is the output vector of the second hidden layer, is the weight of the second hidden layer, representing the connection weight from the j-th neuron of the first hidden layer to the k-th neuron of the second hidden layer. is the bias of the second hidden layer, used to adjust the activation value of the k-th neuron of the second hidden layer. m 2 is the number of neurons in the second hidden layer.

[0051] Calculation for the output layer, the output layer uses a linear activation function: ; Finally, the prediction result is obtained: ; where, is the output vector of the output layer, is the weight of the output layer, representing the weight from the k-th neuron of the second hidden layer to the output layer. is the bias of the output layer, used to adjust the final prediction value. is the final prediction result, representing the grouting construction parameters calculated based on the input features, such as the optimal grouting pressure, slurry diffusion range, etc.

[0052] Among them, the above content details how to input the normalized input vector into the first hidden layer, the second hidden layer, and the output layer of the BP neural network in sequence through a preset forward propagation formula, and calculate the final prediction result layer by layer. The first hidden layer transforms the original input features into preliminary high-dimensional feature expressions; the second hidden layer further extracts deeper feature information; the output layer combines this information and outputs the final predicted value. This hierarchical calculation process enables the neural network to effectively capture the non-linear relationships in the input data, improve the prediction accuracy of the grouting construction parameters, and provide reliable support for engineering decisions.

[0053] S140. Perform inverse normalization on the prediction result to obtain the predicted value of the grouting construction parameter.

[0054] Specifically, when training and predicting the neural network model, the input data usually needs to be normalized, that is, the original data is mapped to a smaller numerical range, such as [0, 1] or [−1, 1]. The advantages of this include: making the input features with different dimensions have similar numerical ranges, improving the stability of model training; avoiding unstable gradient updates caused by too large feature values; improving the convergence speed and making the training more efficient. However, the prediction result of the neural network is also a normalized value and cannot be directly used in engineering applications. Therefore, inverse normalization must be performed to convert the predicted value back to the actual unit of the original data, such as MPa (pressure), m 3 / h (grouting volume), etc.

[0055] Therefore, by obtaining engineering data such as soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature, and constructing a data-driven BP neural network model, the non-linear relationship between engineering parameters can be more accurately reflected, the prediction accuracy can be improved, and it is more scientific and stable compared to relying on manual experience. The traditional manual judgment method is easily affected by the experience level and subjective judgment of construction personnel, while this method uses a BP neural network for automatic calculation, which can reduce human errors and improve the consistency and reliability of construction parameter prediction. Due to the strong non-linear mapping ability of the BP neural network, this method can effectively predict parameters under different soil conditions and different environmental factors, making the construction method have stronger adaptability and generalization ability. By performing normalization preprocessing on the original data, the numerical magnitude differences between different engineering parameters can be eliminated, the stability and convergence speed of the model can be enhanced, and the training and prediction processes of the neural network can be ensured to be more efficient. Therefore, it is convenient to improve the accuracy of grouting construction parameter prediction.

[0056] In a possible implementation, the predicted result is denormalized to obtain the predicted value of the grouting construction parameter, which specifically includes: obtaining the parameters of the preset normalization formula; constructing a denormalization formula according to the parameters; and denormalizing the predicted result using the denormalization formula to obtain the predicted value of the grouting construction parameter.

[0057] Specifically, in the BP neural network, the input and output data are usually normalized to improve the training stability and prediction accuracy of the model. For example, the numerical ranges of the original construction parameters (such as grouting pressure, grouting volume, permeability coefficient, etc.) may vary greatly, and directly inputting them into the neural network may lead to unstable training. Therefore, it is necessary to scale them to a standard range. However, the predicted result of the BP neural network is a normalized value. If these values are directly used, they have no practical engineering significance. Therefore, it is necessary to perform denormalization processing to restore them to the original numerical range.

[0058] In a possible implementation, the denormalization formula is as follows: ; where y actual is the actual predicted value of the grouting construction parameter, and y min and y max are the minimum and maximum values of the predicted result.

[0059] Specifically, the server uses the denormalization formula to convert the predicted result of the BP neural network into the actual grouting construction parameter, making it available for engineering. The normalized predicted value cannot be directly used for construction and must be denormalized. The server uses the minimum and maximum values for linear transformation to restore the original physical value. The denormalized parameters, such as grouting pressure, grouting volume, and grouting speed, can be directly used for engineering construction, improving the scientificity and accuracy of decision-making. This calculation method ensures that the predicted result of the BP neural network can converge stably during model training and be operable in actual applications.

[0060] In a possible implementation, referring to Figure 2 , Figure 2 is another process schematic diagram of a method for predicting grouting construction parameters based on a BP neural network provided by an embodiment of the present application. It includes steps S210 to S220, and the above steps are as follows: S210. Generate notification data according to the predicted value of the grouting construction parameter; S220. Send the notification data to the user equipment carried by the user corresponding to the target project to display the notification data to the user through the user equipment.

[0061] Specifically, the server needs to ensure that the appropriate user devices receive the notifications. Therefore, it queries the user information related to the target project and then pushes the notification data to the user's devices. The target project refers to the specific project where grouting construction is currently underway, such as a subway tunnel construction site or a dam reinforcement project. The target users are engineers, technicians, or managers at the construction site. The user devices refer to the intelligent devices carried by the users, such as smartphones, tablets, smartwatches, and construction-specific terminals (such as intelligent handheld devices equipped with industrial-grade displays). The server sends the notification data to the target users' devices through wireless communication methods (such as 4G / 5G, Wi-Fi, or local area network).

[0062] When the user device receives the notification data, it will display it to the user in an appropriate manner. The specific forms include: pop-up reminder (applicable to emergencies). For example, a notification pops up on an engineer's mobile phone: "The current recommended grouting pressure is 1.52 MPa. Please adjust the equipment.". Display on the APP or Web interface. In the project management APP, users can view the detailed prediction results of grouting parameters and relevant guidance information. SMS or email notification. If the network at the construction site is unstable, the server can also send the construction parameters via SMS or email. Voice broadcast or intelligent voice assistant. If the construction environment is noisy, intelligent devices (such as headphones or smart speakers) can voice broadcast the notification. For example: "Please note that the current recommended best grouting pressure is 1.52 MPa.".

[0063] This application also provides a grouting construction parameter prediction device based on the BP neural network. Refer to Figure 3 , Figure 3 which is a schematic diagram of the modules of a grouting construction parameter prediction device provided by an embodiment of this application. The grouting construction parameter prediction device includes an acquisition module 31 and a processing module 32. Among them, the acquisition module 31 acquires the original project data for the target project. The original project data is any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature. The processing module 32 performs data processing on the original project data to obtain an input vector to be input. The processing module 32 inputs the input vector to be input into the BP neural network model to obtain a prediction result. The processing module 32 performs anti-normalization processing on the prediction result to obtain the predicted value of the grouting construction parameter.

[0064] In a possible implementation manner, the processing module 32 performs data processing on the original project data to obtain an input vector to be input, specifically including: the processing module 32 performs missing value processing and outlier detection on the original project data to obtain the original project features. The processing module 32 performs normalization processing on the original project features using a preset normalization formula to obtain the input vector to be input.

[0065] In a possible implementation manner, the preset normalization formula is calculated using the following formula: ; Among them, x i represents the value of the i-th original engineering feature, including any one of soil density, permeability coefficient, initial equipment pressure, grouting volume, and environmental temperature. x min and x max are respectively the minimum and maximum values of the i-th original engineering feature in the dataset, used to determine the value range of this feature. x i norm is the input vector to be input, used to represent the normalized value.

[0066] In a possible implementation manner, the processing module 32 inputs the input vector to be input into the BP neural network model to obtain a prediction result, specifically including: the processing module 32 uses a preset forward propagation formula to calculate the input vector to be input to obtain a prediction result; The preset forward propagation formula is specifically as follows: ; Among them, x i norm is the input vector to be input, used to represent the normalized value, n is the number of neurons in the input layer, that is, the number of input features, is the weight of the first hidden layer, representing the connection weight from the i-th feature in the input layer to the j-th neuron in the first hidden layer, determining the contribution of this feature to this neuron, is the bias of the first hidden layer, used to adjust the activation value of the j-th neuron in the first hidden layer, making the output of the neuron more robust, is the activation function, m 1 is the number of neurons in the first hidden layer, is the weight of the second hidden layer, representing the connection weight from the j-th neuron in the first hidden layer to the k-th neuron in the second hidden layer, is the bias of the second hidden layer, used to adjust the activation value of the k-th neuron in the second hidden layer, m 2 is the number of neurons in the second hidden layer, is the weight of the output layer, representing the weight from the k-th neuron in the second hidden layer to the output layer, is the bias of the output layer, used to adjust the final prediction value, is the prediction result.

[0067] In a possible implementation, the processing module 32 calculates the input vector to be processed using a preset forward propagation formula to obtain a prediction result, specifically including: the processing module 32 inputs the input vector to be processed into the first hidden layer of the BP neural network model, and calculates to obtain a first result; the processing module 32 uses the first result as an input and passes it to the second hidden layer of the BP neural network model, and calculates to obtain a second result; the processing module 32 inputs the second result into the output layer of the BP neural network and calculates to obtain a prediction result.

[0068] In a possible implementation, the processing module 32 performs denormalization processing on the prediction result to obtain a predicted value of the grouting construction parameter, specifically including: the acquisition module 31 acquires the parameters of the preset normalization formula; the processing module 32 constructs a denormalization formula according to the parameters; the processing module 32 performs denormalization processing on the prediction result using the denormalization formula to obtain a predicted value of the grouting construction parameter.

[0069] In a possible implementation, the processing module 32 generates notification data according to the predicted value of the grouting construction parameter; the processing module 32 sends the notification data to the user equipment carried by the user corresponding to the target project, so as to display the notification data to the user through the user equipment.

[0070] It should be noted that: when the device provided in the above embodiments realizes its functions, only the above-mentioned division of each functional module is used for illustration. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be seen in the method embodiments, which will not be elaborated here.

[0071] The present application also provides an electronic device. Refer to Figure 4 , Figure 4 which is a schematic structural diagram of an electronic device provided in an embodiment of the present application. The electronic device may include: at least one processor 41, at least one network interface 44, a user interface 43, a memory 45, and at least one communication bus 42.

[0072] Among them, the communication bus 42 is used to realize the connection and communication between these components.

[0073] Among them, the user interface 43 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 43 may further include a standard wired interface and a wireless interface.

[0074] Among them, the network interface 44 may optionally include a standard wired interface and a wireless interface (such as a Wi-Fi interface).

[0075] Among them, the processor 41 may include one or more processing cores. The processor 41 connects various parts within the entire server through various interfaces and lines, and executes various functions of the server and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 45, and by calling the data stored in the memory 45. Optionally, the processor 41 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor 41 may integrate a combination of one or several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 41 and may be implemented separately by a single chip.

[0076] Among them, the memory 45 may include random access memory (RAM) and may also include read-only memory. Optionally, the memory 45 includes a non-transitory computer-readable storage medium. The memory 45 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 45 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store the data involved in the above-mentioned various method embodiments. Optionally, the memory 45 may also be at least one storage device located far from the aforementioned processor 41. As Figure 4 shown, the memory 45, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program of a grouting construction parameter prediction method based on a BP neural network.

[0077] In Figure 4In the electronic device shown, the user interface 43 is mainly used to provide an interface for the user to input and obtain the data input by the user; while the processor 41 can be used to call an application program stored in the memory 45 for a grouting construction parameter prediction method based on a BP neural network. When executed by one or more processors, the electronic device is caused to execute the method as described in one or more of the above embodiments.

[0078] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0079] This application also provides a computer-readable storage medium storing instructions. When executed by one or more processors, the electronic device is caused to execute the method as described in one or more of the above embodiments.

[0080] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0081] In several embodiments provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some service interfaces. The indirect couplings or communication connections of the devices or units can be in electrical or other forms.

[0082] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0083] In addition, in each embodiment of this application, the functional units can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0084] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes: various media such as USB flash drives, mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0085] The foregoing are only exemplary embodiments of the present disclosure, and the scope of the present disclosure cannot be limited thereby. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. Those skilled in the art will readily think of other implementation manners of the present disclosure after considering the specification and the practice of the disclosed truth. This application aims to cover any variations, uses, or adaptive changes of the present disclosure, and these variations, uses, or adaptive changes follow the general principles of the present disclosure and include well-known common knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and the embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A grouting construction parameter prediction method based on BP neural network, characterized in that: The method comprises: Acquire original engineering data for the target project, wherein the original engineering data is any one of soil density, permeability coefficient, equipment initial pressure, grouting volume, and ambient temperature; Processing the original engineering data to obtain a vector to be input; Inputting the vector to be input into the BP neural network model to obtain a prediction result; The prediction results are subjected to inverse normalization processing to obtain the predicted values ​​of the grouting construction parameters.

2. The grouting construction parameter prediction method based on BP neural network according to claim 1 is characterized in that: The processing of the original engineering data to obtain the vector to be input specifically includes: Performing missing value processing and outlier detection on the original engineering data to obtain original engineering features; The original engineering feature is normalized using a preset normalization formula to obtain the vector to be input.

3. The grouting construction parameter prediction method based on BP neural network according to claim 2 is characterized in that: The preset normalization formula is calculated using the following formula: ； Among them, x i represents the value of the i-th original engineering characteristic, including any one of soil density, permeability coefficient, equipment initial pressure, grouting volume and ambient temperature, x min and x max are the minimum and maximum values ​​of the i-th original engineering feature in the data set, which are used to determine the value range of the feature. i norm is the input vector, used to represent the normalized value.

4. The grouting construction parameter prediction method based on BP neural network according to claim 1 is characterized in that: The step of inputting the vector to be input into the BP neural network model to obtain a prediction result specifically includes: Using a preset forward propagation formula to calculate the input vector to obtain the prediction result; The preset forward propagation formula is as follows: ； Among them, x i norm is the input vector, which is used to represent the normalized value, n is the number of neurons in the input layer, that is, the number of input features, is the weight of the first hidden layer, which represents the connection weight from the i-th feature of the input layer to the j-th neuron of the first hidden layer, and determines the contribution of the feature to the neuron. is the bias of the first hidden layer, which is used to adjust the activation value of the jth neuron in the first hidden layer to make the output of the neuron more robust. is the activation function, m1 is the number of neurons in the first hidden layer, is the weight of the second hidden layer, which represents the connection weight from the jth neuron in the first hidden layer to the kth neuron in the second hidden layer. is the bias of the second hidden layer, which is used to adjust the activation value of the kth neuron in the second hidden layer. m2 is the number of neurons in the second hidden layer. is the weight of the output layer, which represents the weight from the kth neuron in the second hidden layer to the output layer. is the bias of the output layer, which is used to adjust the final prediction value. For the prediction results.

5. The grouting construction parameter prediction method based on BP neural network according to claim 4 is characterized in that: The using of a preset forward propagation formula to calculate the input vector to obtain the prediction result specifically includes: Inputting the vector to be input into the first hidden layer of the BP neural network model, and calculating to obtain a first result; Passing the first result as input to the second hidden layer of the BP neural network model to calculate and obtain a second result; The second result is input into the output layer of the BP neural network to calculate and obtain the prediction result.

6. The grouting construction parameter prediction method based on BP neural network according to claim 2 is characterized in that: The prediction result is subjected to a denormalization process to obtain a prediction value of the grouting construction parameter, specifically comprising: Obtaining parameters of the preset normalization formula; According to the parameters, an anti-normalization formula is constructed; The prediction result is denormalized using the denormalization formula to obtain the predicted value of the grouting construction parameter.

7. The grouting construction parameter prediction method based on BP neural network according to claim 1 is characterized in that: The method further comprises: Generate notification data according to the predicted quantity of grouting construction parameters; The notification data is sent to a user device carried by a user corresponding to the target project, so as to display the notification data to the user through the user device.

8. A grouting construction parameter prediction device based on BP neural network, characterized in that: The device comprises an acquisition module (31) and a processing module (32), wherein: The acquisition module (31) is used to acquire original engineering data for the target project, wherein the original engineering data is any one of soil density, permeability coefficient, equipment initial pressure, grouting volume and ambient temperature; The processing module (32) is used to process the original engineering data to obtain a vector to be input; The processing module (32) is also used to input the vector to be input into the BP neural network model to obtain a prediction result; The processing module (32) is also used to perform a denormalization process on the prediction result to obtain a predicted value of the grouting construction parameter.

9. An electronic device, characterized in that: The electronic device comprises a processor (41), a memory (45), a user interface (43) and a network interface (44), wherein the memory (45) is used to store instructions, the user interface (43) and the network interface (44) are both used to communicate with other devices, and the processor (41) is used to execute the instructions stored in the memory (45) so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.

Images