Intelligent concrete formula adjusting method
Through the intelligent concrete formula adjustment method, the neural network model is used to predict the compressive strength and slump of concrete, which solves the problem of difficult to predict compressive strength in traditional methods, and achieves resource optimization and production cost reduction.
Patent Information
- Application Number
- CN202411417674.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-11
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
During the traditional concrete production process, due to fluctuations in the aggregate moisture content, the actual water consumption does not match the design formula, and the compressive strength is difficult to accurately predict, resulting in waste of concrete resources and increased production costs.
Using intelligent concrete formula adjustment method, by obtaining historical production information and current curve data, using neural network model training and optimization algorithms, predicting the compressive strength and slump of concrete, and adjusting the raw material mix ratio according to the predicted value to ensure that the production requirements are met. .
Accurate prediction of the compressive strength and slump of concrete is achieved, reducing resource waste, improving production efficiency and reducing production costs.
Smart Images

Figure CN118916652A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of concrete formula adjustment, and in particular to an intelligent concrete formula adjustment method. Background Art
[0002] Concrete mix design is an important link in the concrete production process; it refers to determining the performance indicators of concrete such as compressive strength and slump according to the purpose and use environment of concrete, and designing the proportion of cement, aggregate, admixture and water required for concrete through experiments and calculations on the premise of meeting the performance requirements and material properties of concrete.
[0003] In the actual production process, due to production volume and supply and demand reasons, the sand and gravel raw materials used by the mixing station in different time periods come from different manufacturers. There are differences in the production specifications and production processes of different raw materials, which affects the working performance of concrete. Among them, the actual water consumption is inconsistent with the design formula mainly due to the fluctuation of aggregate moisture content. Usually, the production control personnel visually inspect the concrete mixing state and make corrections to the mix ratio based on experience to meet the requirements of engineering construction. For example, when the water content in the concrete is too much, the slump of the concrete will increase, while the compressive strength will be relatively reduced. At this time, the amount of cementitious materials used in the concrete can be increased, or the water consumption can be reduced in the next batch to comprehensively adjust the slump of the concrete in the mixing tank truck after loading.
[0004] Compared with slump, the compressive strength of concrete is affected by more factors, and it is difficult to predict the compressive strength after adjusting the mix ratio. Generally, production control personnel will be more conservative when making adjustments to ensure that there is a certain surplus in the strength of the concrete to meet the quality requirements of the project; but surplus concrete strength means a waste of resources and increases the production cost of concrete. Therefore, a more accurate intelligent concrete formula adjustment method is needed. Summary of the invention
[0005] The invention discloses an intelligent concrete formula adjustment method, which mainly solves the problem that in the production process of traditional concrete, the formula of concrete is usually adjusted by manual experience to meet the slump requirements of engineering construction, but since the compressive strength of concrete is difficult to predict, it will be over-controlled, resulting in waste of concrete resources and increased production costs.
[0006] To achieve the above object, the technical solution of the present invention is implemented as follows: The present invention provides a method for adjusting a smart concrete formula, comprising the following steps: S1. Feature selection and data analysis: Obtain historical concrete production information, current curves, and data on actual raw material usage in the formula, and analyze the current curves with the same water-cement ratio in the acquired data. Select the mean of the time period with obvious current differences as the characteristic value of the current curve based on the statistical parameters of standard deviation, maximum value, and minimum value; S2. Correlation between characteristic values of current curve and compressive strength and slump: According to the characteristic values of historical current curve, the current curve characteristics of each pot of concrete are compared with the median of the characteristic values of historical current curve, and the corresponding compressive strength and slump labels are assigned; S3. Model training and algorithm optimization: The actual usage data of concrete raw materials is obtained, and the characteristics of the mix ratio are extracted. The characteristics of the mix ratio and the current curve are used as input vectors, and the mass of each cubic meter of concrete raw materials is used as input features. Multiple neural network models are designed and trained for different concrete configuration strengths. The neural network model is trained, and regularization and dropout techniques are used to reduce overfitting and enhance the generalization ability of the neural network model. A concrete compressive strength and slump detection model with a mapping relationship between the current curve and the corresponding compressive strength and slump labels is obtained; S4. Adjustment of concrete formula: obtain the concrete mixing current curve and raw material measurement data, extract features and input them into the concrete compressive strength and slump detection model, obtain the predicted values of concrete compressive strength and slump, and compare them with production requirements. When there is a deviation, obtain the historical data set, use pattern recognition and optimization algorithms to obtain the adjustment plan of the raw material mix ratio.
[0007] In one embodiment, the neural network model includes two layers of fully connected neural networks connected in sequence, with hidden layers of 32 neurons and 16 neurons, a ReLU activation function, and an output layer with a linear activation function.
[0008] In one embodiment, the optimizer in the neural network model is an Adam optimizer, and the loss function is the mean absolute error.
[0009] In one embodiment, the initial learning rate of the neural network model is 0.001, the decay exponent is 0.98, the decay rate is 1, and a fixed step decay strategy is used, and an early stopping mechanism is set.
[0010] In one embodiment, the early stopping mechanism stops training when the validation set loss does not improve for 20 consecutive epochs.
[0011] In one embodiment, the mass of each cubic meter of concrete raw materials includes the mass of gravel per unit cubic meter 5-1, the mass of gravel per unit cubic meter 1-2, the mass of gravel per unit cubic meter 1-3, the mass of sand per unit cubic meter, the mass of water reducing agent per unit cubic meter, the mass of cement per unit cubic meter, the mass of water per unit cubic meter, and the current characteristics per unit cubic meter.
[0012] In one embodiment, when the neural network model is trained, samples in which the mix ratio is not adjusted during the concrete production mixing process are used as training data sets and compared with the original formula to obtain an unadjusted production order, and training is performed based on compressive strength and slump labels, mix ratio characteristics and current curve characteristics.
[0013] In one embodiment, the historical data set is a prediction database of historical concrete compressive strength and slump.
[0014] In one embodiment, the method of the pattern recognition and optimization algorithm is to match the mix ratio corresponding to the compressive strength and slump corresponding to the predicted values of concrete compressive strength and slump from a historical concrete compressive strength and slump prediction database, and compare the mix ratio with the current batch production requirements to obtain the required water or cementitious material adjustment amount.
[0015] The advantages or beneficial effects of the above technical solution include at least: by adopting feature selection and data analysis, associating the characteristic value of the current curve with the compressive strength and slump, model training and algorithm optimization and concrete formula adjustment steps, it is possible to obtain and process various data in the historical concrete mixing process to obtain the current curve and the corresponding compressive strength and slump labels, and it is possible to train a neural network model based on the current curve and the corresponding compressive strength and slump labels to obtain a concrete compressive strength and slump detection model with a mapping relationship between the current curve and the corresponding compressive strength and slump labels, so as to predict the concrete compressive strength and slump in the actual production process through the model, and it is possible to match the mix ratio under the same circumstances based on the historical production data, so as to obtain the required water or gel material adjustment amount, thereby ensuring the production quality of concrete, reducing production costs and improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The accompanying drawings illustrate exemplary embodiments of the present invention and together with the description serve to explain the principles of the present invention. These drawings are included to provide a further understanding of the present invention and are incorporated in and constitute a part of this specification.
[0017] Figure 1 A schematic flow chart of a method for adjusting a formula of intelligent concrete according to an exemplary embodiment of the present invention is shown; Figure 2A schematic diagram showing the distribution of current characteristics under the same recipe according to an exemplary embodiment of the present invention is shown. DETAILED DESCRIPTION
[0018] Embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.
[0019] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0020] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.
[0021] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.
[0022] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise clearly indicated in the context, it should be understood as "one or more".
[0023] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes, and are not used to limit the scope of these messages or information.
[0024] See also Figure 1 The embodiment of the present invention provides a method for adjusting a smart concrete formula, comprising the following steps: S1. Feature selection and data analysis: Obtain historical concrete production information, current curves, and data on actual raw material usage in the formula, and analyze the current curves with the same water-cement ratio in the acquired data. Select the mean of the time period with obvious current differences as the characteristic value of the current curve based on the statistical parameters of standard deviation, maximum value, and minimum value; Among them, during the mixing process of concrete production, the current curve of the mixer reflects the mixing resistance of the mixture. According to experimental observation and analysis, under a specific production formula, the current curve forming point in certain time periods changes significantly. These changing time periods reflect the mixing and reaction process of concrete and can have a decisive influence on the quality of the final concrete. Therefore, it is necessary to obtain the above data for the subsequent training of the neural network model.
[0025] S2. Correlation between characteristic values of current curve and compressive strength and slump: According to the characteristic values of historical current curve, the current curve characteristics of each pot of concrete are compared with the median of the characteristic values of historical current curve, and the corresponding compressive strength and slump labels are assigned; Among them, since the current curve of concrete with the same proportion in the actual production process will also show differences, the characteristic distribution of the overall current curve is normally distributed. The overall high current curve is related to the increase in stirring resistance. Therefore, it shows that the concrete is relatively dry and hard, with a smaller slump and a larger compressive strength. Therefore, the above data need to be labeled with compressive strength and slump.
[0026] S3. Model training and algorithm optimization: The actual usage data of concrete raw materials is obtained, and the characteristics of the mix ratio are extracted. The characteristics of the mix ratio and the current curve are used as input vectors, and the mass of each cubic meter of concrete raw materials is used as input features. Multiple neural network models are designed and trained for different concrete configuration strengths. The neural network model is trained, and regularization and dropout techniques are used to reduce overfitting and enhance the generalization ability of the neural network model. A concrete compressive strength and slump detection model with a mapping relationship between the current curve and the corresponding compressive strength and slump labels is obtained; Among them, the neural network model includes two layers of fully connected neural networks connected in sequence, with hidden layers of 32 neurons and 16 neurons, ReLU activation function and output layer with linear activation function; the optimizer in the neural network model is Adam optimizer, and the loss function is mean absolute error; the initial learning rate of the neural network model is 0.001, the decay exponent is 0.98, the decay rate is 1 fixed step decay strategy, and an early stopping mechanism is set; the early stopping mechanism is to stop training when the validation set loss has not improved for 20 consecutive epochs, that is, to obtain the optimal concrete compressive strength and slump detection model. The mass of each cubic meter of concrete raw materials includes the mass of 5-1 gravel per unit cubic meter, the mass of 1-2 gravel per unit cubic meter, the mass of 1-3 gravel per unit cubic meter, the mass of sand per unit cubic meter, the mass of water reducer per unit cubic meter, the mass of cement per unit cubic meter, the mass of water per unit cubic meter and the current characteristics per unit cubic meter. When the neural network model is trained, samples in which the mix ratio is not adjusted during the concrete production and mixing process are used as training data sets and compared with the original formula to obtain the production order without adjustment, and training is performed based on the compressive strength and slump labels, the characteristics of the mix ratio and the characteristics of the current curve.
[0027] S4. Adjustment of concrete formula: obtain the concrete mixing current curve and raw material measurement data, extract features and input them into the concrete compressive strength and slump detection model, obtain the predicted values of concrete compressive strength and slump, and compare them with production requirements. When there is a deviation, obtain the historical data set, use pattern recognition and optimization algorithms to obtain the adjustment plan of the raw material mix ratio.
[0028] The historical data set is a prediction database of historical concrete compressive strength and slump. The method of pattern recognition and optimization algorithm is to match the compressive strength and slump corresponding to the predicted values of concrete compressive strength and slump from the prediction database of historical concrete compressive strength and slump, and compare the mix ratio with the current production requirements to obtain the required water or cementitious material adjustment amount.
[0029] By adopting the above method, feature selection and data analysis are adopted, the characteristic value of the current curve is associated with the compressive strength and slump, model training and algorithm optimization and concrete formula adjustment steps are performed, and the current curve and the corresponding compressive strength and slump labels can be obtained and processed based on various data in the historical concrete mixing process, and the current curve and the corresponding compressive strength and slump labels can be obtained. A neural network model can be trained according to the current curve and the corresponding compressive strength and slump labels to obtain a concrete compressive strength and slump detection model with a mapping relationship between the current curve and the corresponding compressive strength and slump labels, so as to predict the concrete compressive strength and slump in the actual production process through the model, and the mix ratio under the same situation can be matched based on the historical production data, so as to obtain the required water or gel material adjustment amount, thereby ensuring the production quality of concrete, reducing production costs and improving production efficiency. Example
[0030] See also Figure 1 and Figure 2 , obtain all production order information of a mixing station from 2021 to 2023, the current curve of the mixing host and the actual raw material usage data of the formula.
[0031] Among them, the compressive strength in the production order information includes eight strength grades: C15, C20, C25, C30, C35, C40, C45, and C50. Filter the formula with the largest sample size (because there is no measurement record of the exact value of the aggregate moisture content, the total water consumption in the historical mix ratio data is incorrect, but the deviation from the actual moisture content satisfies the normal distribution. Therefore, the mix ratio with the largest sample size is closer to the true value), analyze the current curves of all samples under this formula, and select the current mean of the time period with obvious differences in the current curve as the current characteristic value. Reference Figure 2 Taking a formula with C30 compressive strength and 150±30mm slump requirements as an example, the compressive strength label of the sample with current characteristic value distribution at the 5% quantile is set to the lower deviation boundary of 25MPa, and the slump requirement label is set to the lower deviation boundary of 120mm; the compressive strength label of the sample with current characteristic value distribution at the median is set to 30MPa, and the slump requirement label is set to 150mm; the compressive strength label of the sample with current characteristic value distribution at the 90% quantile is set to the upper deviation boundary of 35Mpa, and the slump requirement label is set to the upper deviation boundary of 180mm. Different compressive strength and slump labels are assigned according to the deviation of the current characteristics of each pot and the median of the historical current characteristic value distribution. Similarly, compressive strength and slump labels are assigned to samples of each formula.
[0032] Next, the mix ratio characteristics are extracted from the actual raw material usage data of historical production orders, and the mass of each cubic meter of concrete raw materials is used as the model input feature (unit Kg / m3), including the mass of gravel per unit cubic meter (particle size 5-10mm), the mass of gravel per unit cubic meter (particle size 10-20mm), the mass of gravel per unit cubic meter (particle size 16-31.5mm), the mass of sand per unit cubic meter, the mass of water reducer per unit cubic meter, the mass of cement per unit cubic meter, the mass of water per unit cubic meter, and the current characteristics per unit cubic meter.
[0033] Samples whose mix proportions were not adjusted during the concrete production and mixing process were selected as training data sets. By comparing them with the original formula, the production orders without adjustments can be obtained, which means that the raw material fluctuations are small and the concrete quality is relatively stable, which is conducive to the model learning rules; then, according to the slump and compressive strength labels, the neural network model is used to learn the corresponding mix proportion characteristics and current characteristics; the structure of the neural network model is a two-layer fully connected neural network, the number of neurons in the hidden layer is 32 and 16 respectively, the activation function is set to ReLU, and finally an output layer with a linear activation function is connected to regress the weighted results of the neurons; at the same time, regularization and dropout techniques are used to reduce overfitting and enhance the generalization ability of the model; among them, the optimizer selects the Adam optimizer, the loss function uses the mean absolute error evaluation, the initial learning rate is 0.001, and the decay exponent is set to 0.98, and the decay rate is set to 1. The fixed step decay strategy, the early stopping mechanism is added during training, and the training is stopped when the loss of the validation set has not improved for 20 consecutive epochs, and the optimal model is saved, so that the concrete compressive strength and slump detection model with the optimal current curve and the corresponding compressive strength and slump label mapping relationship can be obtained.
[0034] During concrete production and mixing, the raw material metering data and the current curve of the mixing host are read, and the features are extracted and input into the concrete compressive strength and slump detection model to obtain the predicted values of the compressive strength and slump of the concrete batch; while saving the predicted value results, the corresponding slump and compressive strength corresponding mix ratio is matched from the historical predicted compressive strength and slump database, and the mix ratio is selected for comparison with the current batch to obtain the required water or cementitious material adjustment amount.
[0035] In actual use, the present invention requires the mixing station to have the ability to accurately measure the actual amount of raw materials used, and to record the historical order data of the mixing station and the actual amount of raw materials used. At the same time, it is able to obtain the working current of the mixing host, and requires the mixing station to have good production quality control in historical production tasks to ensure the accuracy of the assignment of slump and compressive strength labels by the present invention. In the description of the present invention, it should be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they cannot be understood as limitations on the present invention.
[0036] It should be understood by those skilled in the art that the above embodiments are only for the purpose of clearly illustrating the present invention, and are not intended to limit the scope of the present invention. For those skilled in the art, other changes or modifications may be made based on the above invention, and these changes or modifications are still within the scope of the present invention.
Claims
1. A method for adjusting the formula of intelligent concrete, It is characterized in that The following steps are included: S1. Feature selection and data analysis: Obtain historical concrete production information, current curves, and data on actual raw material usage in the formula, and analyze the current curves with the same water-cement ratio in the acquired data. Select the mean of the time period with obvious current differences as the characteristic value of the current curve based on the statistical parameters of standard deviation, maximum value, and minimum value; S2. Correlation between characteristic values of current curve and compressive strength and slump: According to the characteristic values of historical current curve, the current curve characteristics of each pot of concrete are compared with the median of the characteristic values of historical current curve, and the corresponding compressive strength and slump labels are assigned; S3. Model training and algorithm optimization: The actual usage data of concrete raw materials is obtained, and the characteristics of the mix ratio are extracted. The characteristics of the mix ratio and the current curve are used as input vectors, and the mass of each cubic meter of concrete raw materials is used as input features. Multiple neural network models are designed and trained for different concrete configuration strengths. The neural network model is trained, and regularization and dropout techniques are used to reduce overfitting and enhance the generalization ability of the neural network model. A concrete compressive strength and slump detection model with a mapping relationship between the current curve and the corresponding compressive strength and slump labels is obtained; S4. Adjustment of concrete formula: obtain the concrete mixing current curve and raw material measurement data, extract features and input them into the concrete compressive strength and slump detection model, obtain the predicted values of concrete compressive strength and slump, and compare them with production requirements. When there is a deviation, obtain the historical data set, use pattern recognition and optimization algorithms to obtain the adjustment plan of the raw material mix ratio.
2. The method for adjusting the formula of intelligent concrete according to claim 1, It is characterized in that The neural network model includes two layers of fully connected neural networks connected in sequence, a hidden layer with 32 neurons and 16 neurons, a ReLU activation function, and an output layer with a linear activation function.
3. The intelligent concrete formula adjustment method according to claim 1, It is characterized in that The optimizer in the neural network model is the Adam optimizer, and the loss function is the mean absolute error.
4. The intelligent concrete formula adjustment method according to claim 1, It is characterized in that The initial learning rate of the neural network model is 0.001, the decay exponent is 0.98, the decay rate is 1, and the fixed step decay strategy is set, and an early stopping mechanism is set.
5. The intelligent concrete formula adjustment method according to claim 4, It is characterized in that The early stopping mechanism stops training when the validation set loss does not improve for 20 consecutive epochs.
6. The intelligent concrete formula adjustment method according to claim 1, It is characterized in that The mass of each cubic meter of concrete raw materials includes the mass of gravel per unit cubic meter 5-1, the mass of gravel per unit cubic meter 1-2, the mass of gravel per unit cubic meter 1-3, the mass of sand per unit cubic meter, the mass of water reducing agent per unit cubic meter, the mass of cement per unit cubic meter, the mass of water per unit cubic meter and the current characteristics per unit cubic meter.
7. The method for adjusting the formula of intelligent concrete according to claim 1, It is characterized in that When the neural network model is trained, samples whose mix ratio is not adjusted during the concrete production and mixing process are used as training data sets, and compared with the original formula to obtain an unadjusted production order, and training is performed based on compressive strength and slump labels, through the characteristics of the mix ratio and the characteristics of the current curve.
8. The method for adjusting the formula of intelligent concrete according to claim 1, It is characterized in that The historical data set is a prediction database of historical concrete compressive strength and slump.
9. The method for adjusting the formula of intelligent concrete according to claim 8, It is characterized in that The method of the pattern recognition and optimization algorithm is to match the mix ratio corresponding to the compressive strength and slump corresponding to the predicted values of the concrete compressive strength and slump from the historical concrete compressive strength and slump prediction database, and compare the mix ratio with the current batch production requirements to obtain the required water or cementitious material adjustment amount.
Citation Information
Patent Citations
Model construction method and device and concrete compressive strength prediction method and device
CN113109145A
Slump prediction method and device based on concrete formula and readable medium
CN117195163A
System and method for predicting hydrocarbon well production
US20210224669A1
Cited By
Asphalt mixture aggregate grading online monitoring method
CN119147529A
Concrete slump control system and method combining AI monitoring and ERP
CN120560023A
Concrete preparation formula optimization design method
CN120895131A
