A steam curing process management and optimization method and system based on a smart beam factory

Abstract

The application provides a kind of based on wisdom beam factory's steaming process management and optimization method and system, it is related to steaming process management technical field, the method includes: step 1, constructs and trains an intelligent analysis model, obtains the intelligent analysis model after training;Real-time acquisition's steaming environment parameters and the temperature and humidity data of the internal multiple position points of precast beam body are input into the intelligent analysis model after training, obtain the temperature and humidity field distribution in space and time on the internal precast beam body;Step 2, extract the change characteristics of temperature and humidity field distribution in space and time dimension, determine a feature operating point according to change characteristics, and generate corresponding temperature and humidity gradient change trend line based on feature operating point.The application realizes the management and continuous improvement of steaming process, improves the steaming quality of precast beam body and production management efficiency by temperature and humidity field accurate prediction, structure state evaluation and individualized steaming regulation, combined with micro-crack risk control and model adaptive optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of steam curing process management, in particular to a steam curing process management and optimization method and system based on a smart beam factory. BACKGROUND

[0002] In the steam curing process of a precast beam, effective monitoring of the temperature and humidity field inside the beam body is one of the key links that affect the final performance of the concrete. Currently, in actual production, the steam curing process control usually relies on preset process curves or limited number of sensor feedbacks. Such methods may face some challenges in actual application. Since the monitoring points are relatively limited, the data obtained is mostly difficult to accurately reflect the complete temperature and humidity distribution details of the beam body in the spatial and temporal dimensions. For example, in a certain beam factory project, although multiple temperature sensors are arranged, the analysis mainly relies on the independent values of each monitoring point, and the data is not fully integrated to build a complete temperature field distribution model. This processing method may cause a certain lag in the perception of the internal state changes of the structure, so that the adjustment of the steam curing process parameters still relies more on the experience judgment of the operators. In this case, there may be local differences in the curing conditions of different areas of the beam body, and if not properly controlled, it may increase the risk of micro-cracks in the concrete, which may have a potential impact on the long-term durability of the precast beam. SUMMARY

[0003] The technical problem to be solved by the present application is to provide a steam curing process management and optimization method and system based on a smart beam factory, to reduce the micro-cracks caused by the superposition of thermal stress and dry shrinkage stress inside the beam body, and to improve the structural integrity and durability of the precast beam body.

[0004] To solve the above technical problems, the technical solutions of the present application are as follows:

[0005] In a first aspect, a steam curing process management and optimization method based on a smart beam factory is provided, which comprises the following steps:

[0006] Step 1: constructing and training an intelligent analysis model to obtain a trained intelligent analysis model; inputting the real-time collected steam curing environment parameters and the temperature and humidity data of multiple position points inside the precast beam body into the trained intelligent analysis model to obtain the temperature and humidity field distribution of the precast beam body in space and time;

[0007] Step 2: extracting the change characteristics of the temperature and humidity field distribution in the spatial and temporal dimensions, determining a characteristic operation point according to the change characteristics, and generating a corresponding temperature and humidity gradient change trend line based on the characteristic operation point;

[0008] Step 3: Calculate the coordination angle between the temperature and humidity gradient trend lines, and define an optimization operation domain based on the coordination angle; combine the temperature and humidity field distribution with the optimization operation domain to determine the structural state assessment results of the precast beam; and generate parameter correction amounts for the steam curing process parameters based on the structural state assessment results.

[0009] Step 4: Based on the structural condition assessment results, determine the risk level of microcrack formation in the precast beam; integrate the microcrack formation risk level with the parameter correction amount to generate a personalized steam curing control strategy for the current precast beam.

[0010] Step 5: Implement personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

[0011] Secondly, a steam curing process management and optimization system based on a smart beam factory includes:

[0012] The training module is used to build and train an intelligent analysis model to obtain the trained intelligent analysis model. The real-time collected steam curing environment parameters and temperature and humidity data of multiple locations inside the precast beam are input into the trained intelligent analysis model to obtain the spatial and temporal temperature and humidity field distribution inside the precast beam.

[0013] The extraction module is used to extract the variation characteristics of the temperature and humidity field distribution in the spatial and temporal dimensions, determine a feature operation point based on the variation characteristics, and generate the corresponding temperature and humidity gradient change trend line based on the feature operation point;

[0014] The evaluation module is used to calculate the coordination angle between the temperature and humidity gradient trend lines, and define an optimization operation domain based on the coordination angle; it combines the temperature and humidity field distribution with the optimization operation domain to determine the structural state evaluation result of the precast beam; and it generates parameter correction amounts for the steam curing process parameters based on the structural state evaluation result.

[0015] The judgment module is used to determine the risk level of microcrack formation in the precast beam based on the structural condition assessment results; and to generate a personalized steam curing control strategy for the current precast beam by integrating the microcrack formation risk level with the parameter correction amount.

[0016] The optimization module is used to execute personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

[0017] Thirdly, a computing device includes:

[0018] One or more processors;

[0019] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0020] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0021] The above-described solution of the present invention has at least the following beneficial effects:

[0022] By constructing an intelligent analysis model, integrating real-time environmental parameters with multi-point temperature and humidity data inside the beam, a two-dimensional temperature and humidity field distribution is generated in both space (various areas of the beam) and time (current and future). This overcomes the monitoring blind spots of discrete sensors, making the details of temperature and humidity changes inside the beam clearly visible and avoiding one-sided judgments about the structural state. By extracting the spatiotemporal variation characteristics of temperature and humidity, calculating the coordination angle, and assessing the structural stress state, fuzzy experience is transformed into quantitative structural state assessment results and parameter corrections. Furthermore, based on the structural state assessment results, the risk level of microcrack formation is determined. The risk level and parameter corrections are integrated to generate personalized steam curing control strategies, which can specifically adjust key steam curing parameters, such as heating rate and isothermal parameters, to reduce microcracks caused by the superposition of thermal stress and drying shrinkage stress inside the beam, thereby improving the structural integrity and durability of the precast beam. Through a closed loop of executing control strategies, feeding back actual quality differences, and optimizing model parameters, the intelligent analysis model can adapt to actual working conditions such as beam plant equipment characteristics, concrete batch differences, and environmental fluctuations, achieving continuous improvement in control accuracy and steam curing quality. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for managing and optimizing the steam curing process based on a smart beam factory, as provided in an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of a steam curing process management and optimization system based on a smart beam factory, provided by an embodiment of the present invention. Detailed Implementation

[0025] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0026] like Figure 1 As shown, an embodiment of the present invention proposes a method for managing and optimizing the steam curing process based on a smart beam factory. The method includes the following steps:

[0027] Step 1: Build and train an intelligent analysis model to obtain the trained intelligent analysis model; input the real-time collected steam curing environment parameters and temperature and humidity data of multiple locations inside the precast beam into the trained intelligent analysis model to obtain the spatial and temporal temperature and humidity field distribution inside the precast beam.

[0028] Step 2: Extract the variation characteristics of the temperature and humidity field distribution in the spatial and temporal dimensions, determine a feature operation point based on the variation characteristics, and generate the corresponding temperature and humidity gradient change trend line based on the feature operation point;

[0029] Step 3: Calculate the coordination angle between the temperature and humidity gradient trend lines, and define an optimization operation domain based on the coordination angle; combine the temperature and humidity field distribution with the optimization operation domain to determine the structural state assessment results of the precast beam; and generate parameter correction amounts for the steam curing process parameters based on the structural state assessment results.

[0030] Step 4: Based on the structural condition assessment results, determine the risk level of microcrack formation in the precast beam; integrate the microcrack formation risk level with the parameter correction amount to generate a personalized steam curing control strategy for the current precast beam.

[0031] Step 5: Implement personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

[0032] In this embodiment of the invention, an intelligent analysis model is constructed to integrate real-time environmental parameters and multi-point temperature and humidity data inside the beam, generating a two-dimensional temperature and humidity field distribution in space (various areas of the beam) and time (current and future). This overcomes the monitoring blind spots of discrete sensors, making the details of temperature and humidity changes inside the beam clearly visible and avoiding one-sided judgments on the structural state. By extracting the spatiotemporal variation characteristics of temperature and humidity, calculating the coordination angle, and evaluating the structural stress state, fuzzy experience is transformed into quantitative structural state assessment results and parameter corrections. Furthermore, based on the structural state assessment results, the risk level of microcrack formation is determined. The risk level and parameter corrections are integrated to generate a personalized steam curing control strategy, which can specifically adjust key steam curing parameters, such as heating rate and isothermal parameters, to reduce microcracks caused by the superposition of thermal stress and drying shrinkage stress inside the beam, thereby improving the structural integrity and durability of the precast beam. Through the closed loop of executing the control strategy, feeding back actual quality differences, and optimizing model parameters, the intelligent analysis model can adapt to actual working conditions such as beam plant equipment characteristics, concrete batch differences, and environmental fluctuations, achieving continuous improvement in control accuracy and steam curing quality.

[0033] In a preferred embodiment of the present invention, step 1 includes:

[0034] Step 100: Using historical data on ambient temperature, humidity, and steam pressure during the steam curing process, along with corresponding data on the dimensions, reinforcement information, and concrete material properties of the precast beam structure, the intelligent analysis model is trained. This training process includes: using a Convolutional Neural Network (CNN) to extract spatial features, and combining this with a Long Short-Term Memory (LSTM) network to process the time series, constructing a CNN-LSTM hybrid intelligent analysis model. The CNN layer captures the spatial distribution characteristics of temperature and humidity inside the precast beam, while the LSTM layer learns the dynamic changes in temperature and humidity over time. The model output is the three-dimensional temperature and humidity distribution inside the beam. Secondly, multiple sets of historical data for a single steam curing cycle, from the start of heating to the end of cooling, are collected. Each set of data must include... Two types of related information are required. The first is dynamic environmental parameters, which need to be recorded every 10 minutes for environmental temperature, humidity, and steam pressure data during each steam curing process to ensure coverage of parameter changes at each stage of steam curing (heating, constant temperature, and cooling). The second is static beam characteristic parameters, which need to be recorded corresponding to the structural dimensions of the precast beams being steam-cured, including beam length, beam height, cross-sectional width, and cross-sectional shape, such as T-type or box-type. The reinforcement information includes steel bar type, steel bar diameter, reinforcement spacing, and number of reinforcement layers. The concrete material characteristics include concrete strength grade, water-cement ratio, aggregate type, and admixture type. The dynamic environmental parameters (with timestamps) in each set of data are matched one-to-one with the static beam characteristic parameters, and the data is organized into a structured training dataset according to beam ID, steam curing cycle, environmental parameters, and characteristic parameters.

[0035] The iterative training of the intelligent analysis model (CNN-LSTM hybrid structure) is initiated as follows: First, the environmental parameters (expanded according to time series) and beam characteristic parameters in the dataset are used as input features of the intelligent analysis model. The measured temperature and humidity data of the three-dimensional grid nodes inside the beam during this steam curing process, i.e., the measured temperature and humidity distribution, are used as output labels. This ensures that the input feature dimension matches the model input layer dimension and the output label dimension matches the model output layer dimension. Second, the initial weight matrices and bias terms of each layer of the model are initialized. The initial value range of the convolution kernel weight matrix of the CNN layer (used to extract spatial features) is [-0.05, 0.05]. The input gate, forget gate, and output gate of the LSTM layer (used to process time series) are initialized. The initial value range of the weight matrix for cell state updates is [-0.02, 0.02]; the initial value range of the bias terms for all layers, including the biases of the CNN layer and the biases of each category of the LSTM layer, is [-0.1, 0.1]; the mean square error function is used to calculate the deviation between the temperature and humidity distribution predicted by the intelligent analysis model (CNN-LSTM hybrid structure) and the measured temperature and humidity distribution. The specific calculation method is as follows: traverse all three-dimensional mesh nodes inside the beam, calculate the difference between the predicted temperature value and the measured temperature value, and the difference between the predicted humidity value and the measured humidity value for each node, and sum the squares of the temperature difference and humidity difference for all nodes to obtain the total sum of squares of the temperature and humidity difference; divide the total sum of squares by the total number of three-dimensional mesh nodes to obtain the deviation value of this iteration;

[0036] The backpropagation algorithm adjusts the weight matrices and bias terms of each layer based on the deviation. For the adjustment of the convolutional kernel weights in the CNN layer, during gradient calculation, the partial derivative of the output layer error with respect to the CNN layer's output feature map is first calculated. This is done element-wise to quantify the impact of each element's change on the overall model error. Then, this partial derivative is multiplied element-wise by the derivative of the CNN layer's activation function (using ReLU as an example). The ReLU derivative rule is that it takes 1 when the non-activated output element is positive and 0 when it is negative. This yields the partial derivative of the error with respect to the non-activated output of the CNN layer. Finally, this result is cross-correlated with the original CNN layer's input feature map (the operation rule is the same as for convolution, only the input feature map does not need to be flipped). This yields the partial derivative of the error with respect to each element in the convolutional kernel weight matrix, i.e., the gradient of the convolutional kernel weights. During parameter updates, the preset learning rate is 0.001. The new value of each convolutional kernel weight element is equal to its original value minus the product of the learning rate and the element's gradient.

[0037] When adjusting the parameters of the LSTM layer, the adjustment logic for the input gate, forget gate, and output gate is consistent. Taking the input gate as an example, during gradient calculation, the partial derivative of the error with respect to the input gate output is calculated first. By taking the derivative element by element, the influence of changes in the input gate output elements on the overall error is reflected. Then, this partial derivative is multiplied element by element by the derivative of the sigmoid activation function. The sigmoid derivative is calculated based on the input gate output, that is, the result after sigmoid activation is multiplied by (1 minus the input gate output) to obtain the partial derivative of the error with respect to the unactivated output of the input gate. Finally, this result is multiplied by the LSTM layer input sequence to obtain the partial derivative of the error with respect to the input gate weight matrix, that is, the gating weight gradient. Parameter updates are performed according to the principle that the new value is equal to the original value minus 0.001 multiplied by the gradient. The adjustment logic for cell state update parameters is consistent with the above gating parameters, with only two differences. First, the inactive output of cell state update is obtained by multiplying the LSTM layer input sequence with the cell state update weight matrix and superimposing the cell state bias term. Second, the activation function is replaced with the tanh function. Therefore, the partial derivative of the error with respect to the cell state update activation output needs to be multiplied element-wise with the tanh derivative. The tanh derivative is calculated by subtracting the square of the cell state update activation output from 1. The cell state update activation output is the result of the inactive output of cell state update after tanh processing. After calculating the partial derivative of the error with respect to the cell state update parameters through the same steps, the update is completed according to the formula: new value of cell state update parameter = original value of the parameter - (0.001 × gradient of the parameter). Third, the bias calculation, gradient calculation, and parameter optimization process is repeated, and the bias value is recorded in each iteration. The iteration continues until the bias value of 50 consecutive iterations is lower than the original value. When the value reaches 0.00001, the model has established a stable mapping relationship between environmental parameters, beam characteristic parameters, and the temperature and humidity distribution inside the beam. Training is then stopped, and the trained intelligent analysis model is obtained.

[0038] Step 101: Integrate the real-time collected environmental temperature, humidity, and steam pressure data from the steam curing environment, along with the temperature and humidity data directly measured by sensors placed at multiple locations inside the precast beam, into a set of real-time input data. Input this real-time input data into the trained intelligent analysis model for data fusion processing to obtain the continuous temperature and humidity distribution within the precast beam's internal space. Specifically, based on the input format of the trained intelligent analysis model in Step 100, process the real-time data in three steps. First, deploy monitoring equipment and sensors, placing one environmental monitoring device on each of the four sides (east, west, south, north) and the center of the steam curing area, monitoring every 5 minutes. The system collects temperature, humidity, and steam pressure data of the current steam curing environment at a high frequency. Sensors are deployed inside the precast beam according to a full-coverage rule for key areas: along the beam length, one monitoring point is set at each beam end (1m from the end), mid-span (midpoint of beam length), and 1 / 4 of the beam length. Within each section, one sensor is placed every 20cm from the surface (5cm from the beam surface) to the core area (center of the section). Along the beam width, one sensor is placed at the midpoint of the width and 10cm from each side. All sensors simultaneously collect temperature and humidity data to ensure coverage of areas prone to temperature and humidity differences, such as the core area of ​​the section and the transition zone between the surface and middle layers. The second step involves setting up environmental monitoring... The collected timestamps, ambient temperature, ambient humidity, and steam pressure data are categorized into the environmental parameter unit. The collected timestamps, sensor 3D coordinates (x for beam length, y for beam height, z for beam width), temperature, and humidity data are categorized into the beam internal point data unit. The data from the two modules are aligned using the same timestamp, and duplicate timestamp entries are removed to form non-redundant structured real-time input data, ensuring that the data format is completely consistent with the input feature format of the intelligent analysis model in step 100. In the third step, the structured real-time input data is input into the trained intelligent analysis model, and outliers are removed using the 3σ criterion. That is, if a data point deviates from the average value of the sensor's last 5 collections, outliers are removed. Data exceeding three times the standard deviation was identified as faulty sensor data and discarded. Cubic spline interpolation was used to supplement the gaps in sensor data, specifically by using measured data from three adjacent sensors to generate temperature and humidity values ​​at any point between them, ensuring data continuity. Based on the established mapping relationship in the model, the discrete sensor data was extrapolated to temperature and humidity data covering all three-dimensional grid nodes (grid spacing 10cm×10cm×10cm) within the beam. The final output is a continuous temperature and humidity distribution presented in a three-dimensional grid format, with each grid node corresponding to a unique (x, y, z) coordinate and the corresponding temperature and humidity values, fully reflecting the temperature and humidity conditions throughout the beam.

[0039] Step 102: Based on the continuous temperature and humidity distribution, extrapolate the temperature and humidity changes at future time points to obtain the temperature and humidity field distribution including spatial three-dimensional coordinates and future time points. Specifically, this includes: Based on the input format of the intelligent analysis model trained in Step 100, process the real-time data in three steps. First, deploy one environmental monitoring device on each of the four sides (east, west, south, north) and the center of the steam curing area to collect the temperature, humidity, and steam pressure data of the current steam curing environment every 5 minutes. Second, deploy sensors inside the precast beam according to the rule of full coverage of key areas. That is, along the beam length, set one cross-sectional monitoring point at the beam end (1m from the end), the mid-span (midpoint of beam length), and 1 / 4 of the beam length. Within each cross-section, from the surface (5cm from the beam surface) to the core area (center point of the cross-section), deploy one sensor at a spacing of 20cm. Along the beam width, one sensor is installed at the midpoint of the width and 10cm from each side. All sensors simultaneously collect temperature and humidity data to ensure coverage of areas of the beam prone to temperature and humidity differences, such as the core area of ​​the cross-section and the transition zone between the surface and middle layers. In the second step, real-time input data is integrated. The collection timestamps, ambient temperature, ambient humidity, and steam pressure data collected by the environmental monitoring equipment are included in the environmental parameter unit, while the collection timestamps, sensor three-dimensional coordinates (x for beam length, y for beam height, and z for beam width), temperature, and humidity data collected by the sensors are included in the beam internal point data unit. The data of the two units are aligned according to the same timestamp, and duplicate timestamp entries are removed to form non-redundant structured real-time input data, ensuring that the data format is completely consistent with the input feature format of the intelligent analysis model in step 100.

[0040] The third step involves inputting the structured real-time input data into the trained intelligent analysis model. First, outliers are removed using the 3σ criterion: if a data point deviates from the average of the sensor's last five acquisitions by more than three times the standard deviation, it is considered faulty data and removed. Then, cubic spline interpolation is used to fill in the gaps between sensor points. Specifically, for any two adjacent sensors (denoted as Sensor 1 and Sensor 2, with three-dimensional coordinates of...),... , The measured temperature and humidity values ​​were respectively In the blank area between the adjacent sensor groups, select the third sensor immediately adjacent to the adjacent sensor group, denoted as sensor 3, and assign it three-dimensional coordinates. Measured temperature and humidity values Using the coordinates of these three sensors and the temperature and humidity values ​​as constraints, a piecewise cubic polynomial function is constructed, with the function formula as follows: = + + + + + + + + + + + + + + + + ,in Represents any position within the blank area Temperature and humidity values, The coefficients of the polynomial to be determined need to satisfy three constraints: first, the calculated values ​​of the function at the three sensor locations must be consistent with the measured values; second, the calculated values ​​must be consistent with the measured values. Secondly, the piecewise functions of adjacent blank regions have continuous first derivatives at the boundary, i.e., at the coordinates of sensor 1 or sensor 2; thirdly, the piecewise functions of adjacent piecewise functions have continuous second derivatives at the boundary. By solving the linear equation system formed by the above constraints, the specific values ​​of all coefficients can be obtained. When calculating the temperature and humidity values ​​at any location within the blank region, the location is first determined. To determine which two adjacent sensors' blank region it belongs to, retrieve the piecewise cubic polynomial function with already determined coefficients for that region; then... Substitute into the above formula and calculate sequentially. Equal values ​​and sum them up to obtain This represents the temperature and humidity value at that location, ensuring data continuity. Finally, based on the established mapping relationship in the model, the discrete sensor point data is extrapolated to temperature and humidity data covering all three-dimensional grid nodes (grid spacing 10cm×10cm×10cm) inside the beam. The final output is a continuous temperature and humidity distribution presented in the form of a three-dimensional grid, with each grid node corresponding to a unique value. The coordinates and the temperature and humidity values ​​at those coordinates can fully reflect the temperature and humidity conditions throughout the entire interior of the beam.

[0041] This embodiment trains an intelligent analysis model using multi-dimensional historical data, enabling the model to establish a correlation between environmental parameters, beam characteristics, and temperature and humidity distribution, avoiding the limitations of relying on only a single data source. By integrating and fusing multi-source real-time data, it solves the problems of limited sensor locations and discrete data in steam curing monitoring. The resulting continuous temperature and humidity distribution can fully reflect the temperature and humidity conditions throughout the beam, eliminating monitoring blind spots. Furthermore, it predicts future temperature and humidity changes and forms a spatiotemporally combined temperature and humidity field distribution, breaking the limitation of only being able to obtain current temperature and humidity data. This allows for early prediction of the temperature and humidity state inside the beam, avoiding a lag in the perception of structural state changes.

[0042] In a preferred embodiment of the present invention, step 2 includes:

[0043] Step 200: Process the temperature data and humidity data in the temperature and humidity field distribution, calculate the temperature gradient modulus and humidity gradient modulus in the spatial dimension; analyze the change of the temperature and humidity field distribution over time, and calculate the temperature change rate and humidity change rate in the time dimension. Specifically, use the global temperature and humidity data at the starting time node of the constant temperature stage in the temperature and humidity field distribution in Step 102 as the calculation basis. The temperature and humidity changes at this node are stable and the data is highly representative. Calculate according to the following process: First, select 30 groups of spatial position combinations inside the beam body. Each group contains 2 different positions, ensuring that the surface layer (y ≤ 0.1m), middle layer (0.1m < y < 0.8m), and core area (y ≥ 0.8m) are covered. And the connection direction of each group of positions includes the three dimensions of the x-axis, y-axis, and z-axis. For each group of positions and , calculate the temperature gradient modulus, that is, read point temperature , point temperature , calculate the absolute temperature difference , and calculate the spatial distance between the two points according to the formula ; divide by to obtain the temperature gradient in this direction ; then calculate the modulus according to the formula , where ,

[0044] are the temperature gradient components in the x, y, and z axis directions respectively, to obtain the temperature gradient modulus of this group of positions; adopt the same process as the temperature gradient modulus, read point humidity , point humidity , calculate the absolute humidity difference between the two , and divide the absolute humidity difference obtained above by the spatial distance between position and to obtain the humidity gradient ; then calculate the modulus according to the formula ; where , are the humidity gradient components in the x, y, and z axis directions respectively, to obtain the humidity gradient modulus. Repeat the above calculation to obtain the temperature gradient modulus and humidity gradient modulus of 30 groups of position combinations, and summarize them to form a global gradient modulus dataset.

[0045] Select 20 fixed spatial positions inside the beam body, covering the surface layer, middle layer, and core area. Select 6 to 7 points in each area. Use the temperature and humidity data at two adjacent predicted time nodes with an interval of 1 hour in Step 102 as the calculation basis. For each fixed position ​Read the previous time node temperature The next time node temperature Calculate the absolute temperature difference between the two. Time interval Hours; use Divide by The temperature change at that location was obtained. Read Humidity at all times , Humidity at all times Calculate the absolute humidity difference between the two. ;use Divide by The humidity change rate was obtained. Repeat the above calculations to obtain the temperature and humidity change rates at 20 fixed locations at all adjacent time points, and summarize them to form a global change rate dataset.

[0046] Step 201: Combine the temperature gradient modulus, humidity gradient modulus, temperature change rate, and humidity change rate to form a change feature. Specifically, this includes: based on the two datasets obtained in step 200—the global gradient modulus dataset, containing the temperature gradient modulus and humidity gradient modulus of various spatial locations; and the global change rate dataset, containing the temperature change rate and humidity change rate of each fixed location at adjacent time points—the two datasets are first merged according to spatial location and time point dimensions to form a comprehensive dataset, where the spatial location corresponds to the three-dimensional coordinates inside the beam. The time nodes correspond to the predicted time nodes in step 102, such as 1 hour or 2 hours after the start of steam curing. The four types of parameters are clearly defined as temperature gradient modulus, humidity gradient modulus, temperature change rate, and humidity change rate, ensuring that each spatial location and time node combination corresponds to the specific values ​​of the four types of parameters. The comprehensive dataset is input into the trained intelligent analysis model. The model performs correlation analysis through two types of algorithms, specifically as follows: For any two of the four types of parameters, such as temperature gradient modulus and temperature change rate, humidity gradient modulus and humidity change rate, the Pearson correlation coefficient between the two is calculated, with a value range of [-1, 1]. A coefficient close to 1 indicates a strong positive correlation between the two types of parameters (when one type increases, the other type increases synchronously), close to -1 indicates a strong negative correlation (when one type increases, the other type decreases synchronously), and close to 0 indicates no significant correlation. This is used to determine the relationship between the parameters. The correlation direction and strength were analyzed. A linear regression model was constructed using the overall fluctuation of the temperature and humidity field inside the beam as the dependent variable (obtained by calculating the sum of squared deviations between the temperature and humidity values ​​of all three-dimensional mesh nodes of the beam and the global average, reflecting the overall instability of the temperature and humidity field), and four types of parameters as independent variables. The variance contribution rate of each independent variable was calculated, i.e., the proportion of the independent variable's explanation of the dependent variable's change. Parameters with a variance contribution rate exceeding 10% were selected as key parameters, indicating a significant impact on the fluctuation of the temperature and humidity field, such as the temperature gradient modulus in the core area and the temperature change rate of the surface. Finally, the correlations between key parameters and parameters were integrated, such as the positive correlation between the temperature gradient modulus and the temperature change rate, to generate a structured parameter matrix. Each row of the matrix corresponds to a combination of spatial location and time node, such as coordinates. With time nodes The combination of =2h, each column corresponds to a key parameter, such as the core temperature gradient modulus, the surface humidity change rate, etc., and the matrix elements are the specific values ​​of the key parameters corresponding to this combination; through this matrix, the comprehensive characteristics of the spatial differences (reflected by the gradient modulus) and temporal changes (reflected by the change rate) of the temperature and humidity field are clearly presented, that is, the changing characteristics of the temperature and humidity field are formed.

[0047] Step 202: Identify the spatial location corresponding to the maximum value of the temperature gradient modulus or humidity gradient modulus in the changing features, and use the corresponding spatial location as the feature operation point. Specifically, this includes: filtering the structured parameter matrix generated in step 201. First, extract all values ​​of the temperature gradient modulus from the values ​​corresponding to all spatial locations and time nodes in the matrix. By comparing row by row, that is, comparing the temperature gradient modulus value of each spatial location and time node combination, determine the maximum value of the temperature gradient modulus, denoted as . and record Spatial coordinates of the row ( Simultaneously, extract all values ​​of the humidity gradient modulus, and similarly determine the maximum value of the humidity gradient modulus, denoted as . Record the spatial coordinates of its row ( The second step is to compare. and The magnitude of the value, if This indicates that the spatial differences in temperature gradients are more significant, so (selecting...) ) represents the target coordinates; if This indicates that the spatial differences in humidity gradients are more significant. (Selection) The first step is to use the target coordinates as the coordinates; if the two values ​​are equal, choose any set of coordinates as the target coordinates; the third step is to combine the actual structure of the precast beam and set the target coordinates as the target coordinates. Converted to physical location, where The distance along the beam's length, e.g., p=10m means 10m from one end of the beam, i.e., the mid-span position. The distance corresponding to the beam height direction, such as =1.5m indicates that the distance from the bottom of the beam is 1.5m, which is the core area of ​​the cross-section. The distance corresponding to the beam width direction, such as r=0.5m, means 0.5m from one side of the beam, i.e., the midpoint of the width; this physical location is the characteristic operating point, where the temperature and humidity changes most drastically, and it is a high-risk area for stress concentration in the beam structure.

[0048] Step 203: Based on the temperature gradient direction and humidity gradient direction at the feature operation point, construct a temperature and humidity gradient trend line. Specifically, this includes: First, using the feature operation point, coordinates (… Centered on the x-axis (beam length direction), extend 1m forward and backward, i.e. -1m to +1m, extending 1m both vertically and horizontally along the y-axis (beam height direction), i.e. -1m to +1m, extending 0.5m to the left and right along the z-axis (beam width direction) (i.e.) -0.5m to +0.5m); take 5 data points in each direction at 0.2m intervals, such as Shaft selection -1m, -0.8m , +0.8m +1m, a total of 15 spatial points (3 directions × 5 points) temperature gradient values ​​were obtained, i.e., the temperature gradient modulus of each point, forming a spatial temperature gradient dataset; all prediction time nodes from step 102 were extracted, a total of 5, such as =1h、 =2h、 =3h、 =4h、 =5h corresponding to the characteristic operation point temperature gradient modulus, forming 5 sets of time nodes and temperature gradient modulus data; the second step, for each direction, such as The x-axis is defined by the distance extending outward from the feature operation point. , such as characteristic operation points Coordinates are a certain point Coordinates are +0.4m, then x=0.4m, with the temperature gradient at that point as the ordinate. That is, the temperature gradient modulus at that point, which is fitted with a quadratic curve using the least squares method, is given by the following formula: ,in The temperature gradient value corresponding to a spatial point, that is, the specific value of the temperature gradient modulus at that point; It is the distance extending outward from the feature operation point, reflecting the spatial distance between that point and the feature operation point; It is the coefficient of the quadratic term, which determines the convexity / concavity direction and curvature of the curve; It is the coefficient of the linear term, which determines the slope of the curve; It is a constant term, corresponding to =0, which is the temperature gradient value at the characteristic operating point. This curve is the temperature gradient trend line in the spatial dimension, reflecting the spatial distribution of the temperature gradient around the characteristic operating point; it is calculated using the least squares method. (Coefficient of quadratic term) (coefficient of the first term) (Constant term), the core is to minimize the actual value of the data points. Value and Fitted Curve Calculation The residual sum of squares of the values ​​is calculated by first calculating the sum of squares of all values ​​in the dataset. The sum The sum The sum The sum, and × The sum × The sum, all The sum; then use these statistics to construct three linear equations, such as × The sum = × The sum + × The sum + × The sum × The sum = × The sum + × The sum + × The sum The sum = × The sum + The sum of ×x + 5× Finally, by using elimination or matrix methods to solve this set of linear equations, we obtain... , , The specific value is represented by the curve, which is the temperature gradient trend line in the spatial dimension, reflecting the spatial distribution law of the temperature gradient around the characteristic operation point.

[0049] Using the predicted time point as the x-axis That is, the steam curing time point in step 102, such as =1h, 2h, with the temperature gradient modulus of the characteristic operation point as the ordinate y, i.e., the temperature gradient modulus at that time node, an exponential curve is fitted using the least squares method, the formula is: ,in It is the temperature gradient modulus value of the feature operation point at a certain time node; It is the predicted time point, that is, the time coordinate in the steam curing process; This is the initial gradient coefficient, ranging from 0.5 to 30℃ / m. The specific value depends on the intensity of the temperature gradient in the beam during the initial steam curing stage. A larger value indicates a higher baseline for the initial temperature gradient, reflecting the initial time point. The gradient change base is close to 0; It is the decay coefficient, with a value ranging from 0.01 to 21 / h. The larger the value, the faster the temperature gradient modulus decays over time, and vice versa. It is the steady-state gradient value, reflecting a sufficiently long time ( The steady-state value of the temperature gradient modulus up to infinity; It is a natural constant and the base of the exponential function. This curve is the temperature gradient trend line in the time dimension, which can reflect the change law of temperature gradient at the characteristic operating point with steam curing time; the calculation process is as follows: first estimate Take the last two data points from the time dimension dataset. corresponding The average value of the y-value is calculated because the temperature and humidity tend to stabilize in the later stages of steam curing, at which point the y-value is close to the steady-state value. Then, each data point in the dataset is processed and calculated. ( That is, the temperature gradient modulus value at that point. and The difference is used to transform a non-linear exponential relationship into a linear one, and then... Take the natural logarithm (denoted as ) ), the original exponential formula Transform into linear form Where A is the initial gradient coefficient. The corresponding intermediate parameters (A= ),because The value ranges from 0.5 to 30℃ / m, therefore the range of values ​​for A is... 0.5 to 30, B is related to the attenuation coefficient The corresponding intermediate parameters, ,because The value ranges from 0.01 to 21 / h, therefore the value of B ranges from -2 to -0.01. The time nodes are then calculated. The sum The sum × The sum The sum of these statistics is used to construct two linear equations, which are respectively... × The sum = A × Sum + B × The sum The sum = the number of samples in the time dimension dataset (5 × A + B ×) The sum of these two equations is used to solve for the values ​​of A and B. Then, the values ​​of A and B are derived from A and B. and ,Right now , Finally , , Substitute into the original exponent formula Verification, if actual If the sum of squared residuals between the calculated and uncalculated values ​​is too large, readjustment is required. The estimated value is obtained by repeating the above data processing, equation construction and solution steps until the residual meets the requirements. The curve obtained through this process is the temperature gradient change trend line in the time dimension, which can reflect the change law of temperature gradient at the characteristic operating point with the steam curing time.

[0050] The construction of the humidity gradient trend line follows the same process as the temperature gradient trend line, specifically as follows: First, taking the feature operation point as the center, 15 spatial points are selected along the x, y, and z axes with the same range (x-axis ±1m, y-axis ±1m, z-axis ±0.5m) and spacing (0.2m). The humidity gradient values ​​of each point are obtained to form a spatial dimension dataset. Simultaneously, the humidity gradient modulus of the feature operation points corresponding to the 5 prediction time nodes in step 102 is extracted to form a time node and humidity gradient modulus dataset. Second, the trend line is fitted and the coefficients are calculated. The formula for the spatial dimension (quadratic curve fitting) fitting curve is... ,in This represents the humidity gradient value at a spatial point. The distance extending outward from the feature operation point. For the coefficient of the quadratic term, For the coefficient of the linear term, The constant term, all three are obtained through fitting, and their meanings are the same as those of the quadratic temperature gradient curve. , , Consistency (determining the convexity and concavity, and the tendency of inclination of the curve, respectively) =0 (Value), the calculation logic is also similar to , , Completely consistent, meaning that by calculating the humidity dataset , , , and × , × , The sum of these equations is used to construct a system of linear equations and solve it to obtain the following results. , , Specific numerical values; time dimension (exponential curve fitting), the fitting curve formula is: , where y is the humidity gradient modulus value of the feature operation point at a certain time node. To predict time points, For the initial gradient correlation coefficient, For attenuation coefficient, These are the steady-state gradient values; all three are obtained through fitting and their meaning is similar to that of the temperature gradient exponential curve. , , Consistent, reflecting the initial base, decay rate, and steady-state value respectively. It is a natural constant; the calculation logic is also the same as... , , Completely consistent, that is, estimate first (Take the mean of the later y-values), calculate and take Transform into linear form (in ), use the least squares method to find And reverse the process Finally, the residuals were verified and adjusted. Through the above fitting and coefficient calculation, the humidity gradient change trend line at the feature operation point was finally obtained.

[0051] This embodiment quantifies the spatial differences (through gradient modulus) and temporal variations (through rate of change) of the temperature and humidity field distribution, solving the problem of focusing only on independent values ​​and failing to reflect the drastic changes in temperature and humidity. By integrating four types of parameters to form change characteristics, it achieves a global integrated analysis of temperature and humidity field changes, avoiding the one-sidedness of analysis caused by interpreting scattered data individually, and more comprehensively reflecting the overall change pattern of the temperature and humidity field. It identifies characteristic operation points, focusing on key locations where temperature and humidity changes are most drastic, avoiding the blindness of global analysis and improving its targeting. It constructs a temperature and humidity gradient change trend line, intuitively presenting the change pattern of temperature and humidity gradients at key locations, avoiding the problem of lacking clear trend references in regulation.

[0052] In a preferred embodiment of the present invention, step 3 includes:

[0053] Step 300: Measure the angle between the temperature and humidity gradient trend lines at different feature operation points to obtain the coordination angle, and set the boundary of the optimized operation domain according to the range of the coordination angle. Specifically, this includes: selecting all feature operation points determined in step 202, such as the core area and surface of different sections of the beam, and extracting the spatial dimension temperature and humidity gradient trend lines generated in step 203 at each feature operation point; for any two feature operation points with the same type of trend lines, such as the temperature gradient trend line at point A and the temperature gradient trend line at point B, first calculate the slope of the tangent line of each trend line at the feature operation point, that is, take the derivative with respect to the quadratic curve, and obtain the slope of the temperature gradient trend line. Substitute 0 ; Slope of humidity gradient trend line Substitute =0 Use the arctangent function to convert the slope into an angle. ,in, It is the gradient trend line of the same type at the first feature operation point, such as the angle corresponding to the slope of the tangent line of the temperature gradient trend line at point A at the feature operation point. It is the gradient trend line of the same type at the second feature operation point, such as the angle corresponding to the slope of the tangent line of the temperature gradient trend line at point B at the feature operation point. The slope of the temperature gradient trend line at another feature point is given, and the coordination angle is the absolute value of the difference between the two angles. Based on the structural safety requirements of the beam, the range of values ​​for the coordination angle is set, such as... ≤30° is the coordination interval; the temperature and humidity gradient range corresponding to the characteristic operation point within the coordination angle interval, such as temperature gradient 0.5 to 15℃ / m and humidity gradient 1 to 10% / m, is set as the boundary of the optimization operation domain, and the gradient range outside the interval is excluded from the optimization domain.

[0054] Step 301 involves performing a matching analysis between the temperature and humidity field distribution and the boundary of the optimized operating domain to obtain a matched temperature and humidity field distribution that meets the boundary conditions of the optimized operating domain. Specifically, this includes: extracting the global temperature and humidity field distribution data from the comprehensive dataset in Step 201, which includes temperature and humidity values ​​corresponding to each spatial location and time node of the beam; comparing the temperature gradient value and humidity gradient value at each spatial location of the beam with the boundary of the optimized operating domain set in Step 300; if both the temperature gradient value and humidity gradient value at that location are within the boundary of the optimized operating domain, then the temperature and humidity data at that location are retained; if any gradient value exceeds the boundary of the optimized operating domain, then the temperature and humidity data at that location are removed; and integrating all the retained temperature and humidity data to form the matched temperature and humidity field distribution.

[0055] Step 302: Calculate the thermal stress distribution and drying shrinkage stress distribution corresponding to the matched temperature and humidity field distribution using the trained intelligent analysis model. Specifically, step 302a involves inputting the temperature distribution data from the matched temperature and humidity field distribution into the trained intelligent analysis model. Based on the temperature distribution data and the thermal expansion coefficient of the precast beam material, the thermal stress distribution at each spatial location within the precast beam is calculated. This includes: inputting the temperature distribution data from the matched temperature and humidity field distribution into the trained intelligent analysis model; the model retrieving the inherent parameters of concrete from the precast beam material parameter library, including the elastic modulus (range 20 to 40 GPa) and the thermal expansion coefficient (range 10 × 10⁻⁶ GPa). Up to 13× / ℃, Poisson's ratio, typically ranging from 0.15 to 0.25; after training, the intelligent analysis model calculates the temperature at each spatial location in the matched temperature and humidity field distribution, determining the difference between the temperature at that location and the global average temperature, which is the temperature change at that location; the model calculates the thermal stress at each spatial location according to the following logic: first, calculate the product of the elastic modulus and the coefficient of thermal expansion, then multiply the product by the temperature change at that location, divide the result by (1 minus Poisson's ratio), and the final result is the thermal stress at that location; subsequently, the model summarizes the thermal stress at all spatial locations of the beam to form the internal thermal stress distribution of the precast beam.

[0056] Step 302b involves inputting the humidity distribution data from the matched temperature and humidity field distribution into the trained intelligent analysis model. Based on the humidity distribution data and the drying shrinkage characteristics of the precast beam material, the drying shrinkage stress distribution at each spatial location within the precast beam is calculated. Specifically, this includes inputting the humidity distribution data from the matched temperature and humidity field distribution into the trained intelligent analysis model. The model retrieves the inherent parameters of the concrete from the precast beam material parameter library, including the drying shrinkage coefficient, with a value range of 200× Up to 400× The elastic modulus, with a value range of 20 to 40 GPa (same as step 302a), and the Poisson's ratio, with a value range of 0.15 to 0.25 (same as step 302a), are used. After training, the intelligent analysis model calculates the difference between the saturated humidity of the concrete (95% to 100%) and the humidity value at that location. This difference represents the humidity change at that location. The model calculates the drying shrinkage stress at each spatial location according to the following logic: first, it calculates the product of the drying shrinkage coefficient and the humidity change at that location to obtain the drying shrinkage deformation at that location; then, it multiplies the elastic modulus by the drying shrinkage deformation, and divides the result by (1 minus Poisson's ratio). The final result is the drying shrinkage stress at that location. The drying shrinkage stresses at all spatial locations of the beam are summarized to form the internal drying shrinkage stress distribution of the precast beam.

[0057] Step 303: Integrate the thermal stress distribution and drying shrinkage stress distribution to form a structural condition assessment result. Compare the maximum stress value in the structural condition assessment result with the concrete tensile strength threshold to obtain the parameter correction amount for the steam curing process. Specifically, this includes: for each spatial location of the beam, superimposing the thermal stress and drying shrinkage stress at that location, where tensile stress is taken as positive and compressive stress as negative, to obtain the total stress at that location; integrating the total stress at all locations to form a structural condition assessment result, which includes the total stress at each spatial location and the maximum tensile stress value among all total stresses; retrieving the concrete tensile strength threshold, with a value ranging from 2.0 to 3.5 MPa, specifically varying with the concrete strength grade, such as C30 concrete. For C60 concrete, the maximum tensile stress is approximately 2.0 MPa, and for C60 concrete, it is approximately 3.5 MPa. The maximum tensile stress value in the structural condition assessment results is compared with the concrete tensile strength threshold. If the maximum tensile stress value is less than or equal to the concrete tensile strength threshold, it indicates that the current stress of the beam is within the safe range, and the correction amount for the steam curing process parameters is 0. If the maximum tensile stress value is greater than the concrete tensile strength threshold, the correction amount is calculated by dividing (maximum tensile stress value minus concrete tensile strength threshold) by the concrete tensile strength threshold and then multiplying by a preset correction coefficient. The preset correction coefficient ranges from 0.8 to 1.2 and is used to convert the stress difference into an executable adjustment range for the process parameters. The final result is the parameter correction amount for the steam curing process.

[0058] This embodiment filters out areas with coordinated temperature and humidity gradients by using a coordination angle to eliminate interference from extreme gradients. This ensures that the matched temperature and humidity field distribution better matches the actual safe operating range of the beam, avoiding stress calculation deviations caused by extreme gradients. It calculates and superimposes thermal stress and drying shrinkage stress separately, covering the two main types of stress generated by temperature and humidity changes in the beam. Compared to single stress calculation, this provides a more comprehensive reflection of the internal structural stress state of the beam, reducing the risk of missing stress assessments. By comparing the maximum stress with the tensile strength threshold, the structural safety state is transformed into a quantifiable process parameter correction, replacing empirical adjustments. This provides a clear and executable quantitative standard for optimizing the steam curing process, improving the accuracy of process adjustments.

[0059] In a preferred embodiment of the present invention, step 4 includes:

[0060] Step 400: Compare the maximum tensile stress value in the structural condition assessment results with the concrete tensile strength threshold to obtain a comparison result; determine the microcrack formation risk level based on the comparison result, specifically including: comparing the maximum tensile stress value in the structural condition assessment results obtained in step 303 with the concrete tensile strength threshold to obtain three comparison results: Result 1, the maximum tensile stress value is less than or equal to 0.8 times the concrete tensile strength threshold; Result 2, the maximum tensile stress value is greater than 0.8 times the concrete tensile strength threshold, but less than or equal to the concrete tensile strength threshold; Result 3, the maximum tensile stress value is greater than the concrete tensile strength threshold. Based on the above comparison results, determine the microcrack formation risk level accordingly: Result 1 corresponds to low risk, Result 2 corresponds to medium risk, and Result 3 corresponds to high risk.

[0061] Step 401 involves performing a correlation analysis between the microcrack generation risk level and the parameter correction amount to determine the adjustment values ​​for the heating rate, isothermal temperature, and isothermal duration. Specifically, this includes: correlating the microcrack generation risk level determined in step 400 with the steam curing process parameter correction amount obtained in step 303; calculating the three types of steam curing parameter adjustment values ​​based on the correlation rules stored in the trained intelligent analysis model; the low-risk heating rate adjustment value is equal to the parameter correction amount multiplied by the correlation coefficient 0.5, and the isothermal temperature adjustment value is equal to the parameter correction amount multiplied by the correlation coefficient 0. 0.3, the adjustment value for constant temperature duration is equal to the parameter correction amount multiplied by the correlation coefficient 0.2; the adjustment value for medium-risk heating rate is equal to the parameter correction amount multiplied by the correlation coefficient 0.8, the adjustment value for constant temperature is equal to the parameter correction amount multiplied by the correlation coefficient 0.6, and the adjustment value for constant temperature duration is equal to the parameter correction amount multiplied by the correlation coefficient 0.4; the adjustment value for high-risk heating rate is equal to the parameter correction amount multiplied by the correlation coefficient 1.2, the adjustment value for constant temperature is equal to the parameter correction amount multiplied by the correlation coefficient 0.9, and the adjustment value for constant temperature duration is equal to the parameter correction amount multiplied by the correlation coefficient 0.7.

[0062] Step 402: Based on the heating rate adjustment value, constant temperature adjustment value, and constant temperature duration adjustment value, a personalized steam curing control strategy is generated. Specifically, this includes: retrieving the basic steam curing process parameters currently being executed on the precast beam, including the initial heating rate (range 5 to 20℃ / h), the initial constant temperature (range 40 to 60℃), and the initial constant temperature duration (range 3 to 8h); superimposing the three types of adjustment values ​​obtained in step 401 with the corresponding basic steam curing process parameters; if the adjustment value is negative, subtracting the absolute value of the adjustment value from the basic process parameters to obtain new steam curing process parameters, i.e., the new heating rate equals the initial heating rate plus the heating rate adjustment value; the new constant temperature equals the initial constant temperature plus the constant temperature adjustment value; and the new constant temperature duration equals the initial constant temperature duration plus the constant temperature duration adjustment value; and integrating the new heating rate, new constant temperature, and new constant temperature duration to form a personalized steam curing control strategy targeting the current temperature and humidity field state and stress risk of the beam.

[0063] This embodiment transforms microcrack risk into an intuitive level, quickly identifying the current safety risk level of the beam and clarifying key control directions. For example, high-risk beams require priority process adjustments to avoid wasting control resources. Based on the risk level, different parameter adjustment values ​​are matched. For low-risk beams, the adjustment range is reduced to control costs, while for high-risk beams, the adjustment range is increased to ensure safety. This ensures that process adjustment measures are precisely matched with the risk level, avoiding over-adjustment or under-adjustment. For different beams with varying temperature and humidity conditions and stress risks, personalized control strategies containing specific parameters are generated to improve the adaptability of the steam curing process to different working conditions, effectively reducing the probability of microcrack formation and ensuring the quality of precast beam structures.

[0064] In a preferred embodiment of the present invention, step 5 includes:

[0065] Step 500: The heating rate adjustment value, constant temperature adjustment value, and constant temperature duration adjustment value in the personalized steam curing control strategy are sent to the steam curing equipment execution unit. The heating process is controlled according to the heating rate adjustment value, and the constant temperature process is controlled according to the constant temperature adjustment value and the constant temperature duration adjustment value to complete the steam curing process. Specifically, this includes: transmitting the heating rate adjustment value, constant temperature adjustment value, and constant temperature duration adjustment value included in the personalized steam curing control strategy generated in step 402 to the steam curing equipment execution unit, such as the temperature control system and humidity control system of the steam curing shed; after receiving the heating rate adjustment value, the steam curing equipment execution unit combines it with the initial heating... The heating rate, ranging from 5 to 20℃ / h, is the new heating rate obtained by adding the initial heating rate to the heating rate adjustment value. This controls the temperature rise of the steam curing environment, ensuring that the temperature rises steadily to the target constant temperature at the new heating rate. Once the temperature reaches the target value (initial constant temperature plus constant temperature adjustment value, with the initial constant temperature ranging from 40 to 60℃), the execution unit maintains the ambient temperature at the new constant temperature. It also controls the duration of the constant temperature phase based on the new constant temperature duration obtained by adding the initial constant temperature duration adjustment value (with the initial constant temperature duration ranging from 3 to 8h), until the constant temperature period ends, thus completing the entire steam curing process.

[0066] Step 501: After the steam curing process is completed, obtain the actual strength data and microcrack data of the precast beam. Compare the actual strength data with the expected strength target to obtain strength difference data, and compare the actual microcrack data with the expected microcrack control target to obtain microcrack difference data. Specifically, this includes: after the steam curing process is completed, using strength testing equipment, such as a rebound hammer and a compression testing machine, to test the actual strength data of the precast beam, such as cubic compressive strength and axial tensile strength; using microcrack observation equipment, such as a crack width meter, to test the actual microcrack data on the surface and inside of the beam, such as microcrack length and microcrack width; retrieving the expected strength target from the precast beam production standard, with a value range of 20 to 60 MPa, for example, the target value of the 28-day cubic compressive strength of C30 concrete is 30 MPa. The target compressive strength of C50 concrete cubes at 28 days is 50 MPa, and the target compressive strength of C60 concrete cubes at 28 days is 60 MPa. The strength difference data is obtained by subtracting the expected strength target from the actual strength data. A positive result indicates that the actual strength is higher than the target, and a negative result indicates that the actual strength is lower than the target. The expected microcrack control targets (microcrack width range 0.1 to 0.2 mm, not greater than 0.2 mm; microcrack length range 30 to 50 mm, not greater than 50 mm) are retrieved from the precast beam quality standards. The expected microcrack width target and the expected microcrack length target are subtracted from the actual microcrack width and the expected microcrack length target, respectively, to obtain microcrack difference data. A positive result indicates that the actual microcrack exceeds the target, and a negative result indicates that the actual microcrack meets or exceeds the target.

[0067] Step 502: Based on the strength difference data and microcrack difference data, adjust the parameter weights for stress prediction and temperature / humidity field prediction in the trained intelligent analysis model to complete the adaptive optimization of the trained intelligent analysis model. Specifically, this includes: using the strength difference data and microcrack difference data obtained in step 501 as feedback signals for model optimization; if the strength difference data is negative (actual strength is insufficient), it indicates that the thermal stress prediction parameter weights in the model may be biased, and the prediction parameter weights related to the temperature field need to be adjusted; if the microcrack difference data is positive (actual microcracks exceed the target), it indicates that the drying shrinkage stress or total stress prediction parameter weights in the model may be biased, and the prediction parameter weights related to the humidity field or stress superposition need to be adjusted; retrieve the current parameter weights of the trained intelligent analysis model, where the stress prediction parameter weights, such as the weights of the elastic modulus and thermal expansion coefficient in thermal stress calculation, range from 0.2 to 0.5, and the temperature / humidity field prediction parameter weights, such as the weights of the temperature gradient fitting coefficient and humidity gradient... The weight of the fit coefficient ranges from 0.3 to 0.6. The weight is adjusted logically by adding the current parameter weight to the target value (absolute value of the difference data ÷ target value) multiplied by a preset adjustment coefficient. The preset adjustment coefficient ranges from 0.05 to 0.2 and is used to control the magnitude of the weight adjustment to avoid excessive fluctuations. For example, if the absolute value of the strength difference data is 5 MPa and the expected strength target is 50 MPa, the preset adjustment coefficient is 0.1, and the current elastic modulus weight is 0.3, then the corresponding parameter weight adjustment is (5 ÷ 50) × 0.1 = 0.01, and the adjusted elastic modulus weight is 0.3 + 0.01 = 0.31. The adjusted parameter weights are then updated in the trained intelligent analysis model to re-verify the model's prediction accuracy for temperature, humidity, and stress fields (by comparing the temperature, humidity, and stress values ​​predicted by the updated model with the actual detected values). If the prediction error is within the allowable range (e.g., less than 5%), the model adaptive optimization is complete. If the error exceeds the range, the above weight adjustment steps are repeated until the prediction error meets the requirements.

[0068] This embodiment achieves automated and precise control of the heating and temperature control processes by directly transmitting personalized control parameters to the equipment execution unit, avoiding errors from manual operation and ensuring that the steam curing process is strictly executed according to the optimized strategy, thereby improving the stability of the beam steam curing quality. By collecting actual strength and crack data after steam curing and calculating the difference value, a closed loop of process execution, quality inspection, and difference feedback is established to promptly detect deviations between the steam curing process and quality targets, reducing potential quality risks. Based on the quality difference data, the model parameter weights are adjusted so that the intelligent analysis model can be continuously optimized according to actual production feedback, gradually reducing errors in temperature and humidity fields and stress prediction, and improving the accuracy of the model's prediction of the beam state under different working conditions.

[0069] like Figure 2As shown, embodiments of the present invention also provide a steam curing process management and optimization system based on a smart beam factory, comprising:

[0070] The training module is used to build and train an intelligent analysis model to obtain the trained intelligent analysis model. The real-time collected steam curing environment parameters and temperature and humidity data of multiple locations inside the precast beam are input into the trained intelligent analysis model to obtain the spatial and temporal temperature and humidity field distribution inside the precast beam.

[0071] The extraction module is used to extract the variation characteristics of the temperature and humidity field distribution in the spatial and temporal dimensions, determine a feature operation point based on the variation characteristics, and generate the corresponding temperature and humidity gradient change trend line based on the feature operation point;

[0072] The evaluation module is used to calculate the coordination angle between the temperature and humidity gradient trend lines, and define an optimization operation domain based on the coordination angle; it combines the temperature and humidity field distribution with the optimization operation domain to determine the structural state evaluation result of the precast beam; and it generates parameter correction amounts for the steam curing process parameters based on the structural state evaluation result.

[0073] The judgment module is used to determine the risk level of microcrack formation in the precast beam based on the structural condition assessment results; and to generate a personalized steam curing control strategy for the current precast beam by integrating the microcrack formation risk level with the parameter correction amount.

[0074] The optimization module is used to execute personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

[0075] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0076] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0077] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0078] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for managing and optimizing the steam curing process based on a smart beam factory, characterized in that, The method includes: Step 1: Build and train an intelligent analysis model to obtain the trained intelligent analysis model; input the real-time collected steam curing environment parameters and temperature and humidity data of multiple locations inside the precast beam into the trained intelligent analysis model to obtain the spatial and temporal temperature and humidity field distribution inside the precast beam. Step 2: Process the temperature and humidity data in the temperature and humidity field distribution, and calculate the temperature gradient modulus and humidity gradient modulus in the spatial dimension; analyze the changes in the temperature and humidity field distribution over time, and calculate the rate of change of temperature and humidity in the time dimension; combine the temperature gradient modulus, humidity gradient modulus, rate of change of temperature, and rate of change of humidity to form a change feature; identify the spatial location corresponding to the maximum value of the temperature gradient modulus or humidity gradient modulus in the change feature, and use the corresponding spatial location as the feature operation point; construct the temperature and humidity gradient change trend line based on the temperature gradient direction and humidity gradient direction at the feature operation point. Step 3: Measure the angle between the temperature and humidity gradient trend lines at different characteristic operation points to obtain the coordination angle, and set the boundary of the optimized operation domain according to the range of the coordination angle; perform matching analysis between the temperature and humidity field distribution and the boundary of the optimized operation domain to obtain the matched temperature and humidity field distribution that meets the boundary conditions of the optimized operation domain; calculate the thermal stress distribution and drying shrinkage stress distribution corresponding to the matched temperature and humidity field distribution through the trained intelligent analysis model; combine the thermal stress distribution and drying shrinkage stress distribution to form the structural state assessment result, and compare the maximum stress value in the structural state assessment result with the concrete tensile strength threshold to obtain the parameter correction amount of the steam curing process parameters; Step 4: Based on the structural condition assessment results, determine the risk level of microcrack formation in the precast beam; integrate the microcrack formation risk level with the parameter correction amount to generate a personalized steam curing control strategy for the current precast beam. Step 5: Implement personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

2. The method for steam curing process management and optimization based on a smart beam factory according to claim 1, characterized in that, Step 1 includes: The intelligent analysis model is trained using data on ambient temperature, humidity and steam pressure during historical steam curing processes, as well as corresponding data on precast beam structure dimensions, reinforcement information and concrete material properties. The ambient temperature, ambient humidity, and steam pressure data collected in real time from the steam curing environment parameters, as well as the temperature and humidity data directly measured by sensors placed at multiple locations inside the precast beam, are integrated into a set of real-time input data. The real-time input data is then input into the trained intelligent analysis model to perform data fusion processing on the real-time input data, thereby obtaining the continuous temperature and humidity distribution inside the precast beam. Based on the continuous temperature and humidity distribution, the temperature and humidity changes at future time points are inferred, resulting in a temperature and humidity field distribution that includes spatial three-dimensional coordinates and future time points.

3. The method for steam curing process management and optimization based on a smart beam factory according to claim 2, characterized in that, The trained intelligent analysis model calculates the thermal stress distribution and drying shrinkage stress distribution corresponding to the matched temperature and humidity field distribution, including: The temperature distribution data in the matched temperature and humidity field distribution is input into the trained intelligent analysis model. Based on the temperature distribution data and the thermal expansion coefficient of the precast beam material, the thermal stress distribution at each spatial location inside the precast beam is calculated. The humidity distribution data in the matched temperature and humidity field distribution is input into the trained intelligent analysis model. Based on the humidity distribution data and the drying shrinkage characteristics of the precast beam material, the drying shrinkage stress distribution at each spatial location inside the precast beam is calculated.

4. The method for steam curing process management and optimization based on a smart beam factory according to claim 3, characterized in that, Step 4 includes: The maximum tensile stress value in the structural condition assessment results is compared with the concrete tensile strength threshold to obtain the comparison results; the risk level of microcrack formation is determined based on the comparison results. By performing a correlation analysis between the risk level of microcrack formation and the parameter correction amount, the adjustment values ​​for heating rate, isothermal temperature, and isothermal duration were determined. A personalized steam curing control strategy is generated based on the adjustment values ​​for heating rate, constant temperature, and constant temperature duration.

5. The method for steam curing process management and optimization based on a smart beam factory according to claim 4, characterized in that, Step 5 includes: The heating rate adjustment value, constant temperature adjustment value, and constant temperature duration adjustment value in the personalized steam curing control strategy are sent to the execution unit of the steam curing equipment. The heating process is controlled according to the heating rate adjustment value, and the constant temperature process is controlled according to the constant temperature adjustment value and constant temperature duration, thus completing the steam curing process. After the steam curing process is completed, the actual strength data and microcrack data of the precast beam are obtained. The actual strength data is compared with the expected strength target to obtain the strength difference data. The actual microcrack data is compared with the expected microcrack control target to obtain the microcrack difference data. Based on the strength difference data and microcrack difference data, the parameter weights of stress prediction and temperature and humidity field prediction in the trained intelligent analysis model are adjusted to complete the adaptive optimization of the trained intelligent analysis model.

6. A steam curing process management and optimization system based on a smart beam factory, the system implementing the method as described in any one of claims 1 to 5, characterized in that, include: The training module is used to build and train an intelligent analysis model to obtain the trained intelligent analysis model. The real-time collected steam curing environment parameters and temperature and humidity data from multiple locations inside the precast beam are input into the trained intelligent analysis model to obtain the spatial and temporal temperature and humidity field distribution inside the precast beam. The extraction module processes temperature and humidity data in the temperature and humidity field distribution, calculates the temperature gradient modulus and humidity gradient modulus in the spatial dimension, analyzes the changes in the temperature and humidity field distribution over time, and calculates the rate of change of temperature and humidity in the time dimension. It integrates the temperature gradient modulus, humidity gradient modulus, rate of change of temperature, and rate of change of humidity to form a change feature. It identifies the spatial location corresponding to the maximum value of the temperature gradient modulus or humidity gradient modulus in the change feature and uses the corresponding spatial location as the feature operation point. Based on the temperature gradient direction and humidity gradient direction at the feature operation point, construct the temperature and humidity gradient change trend line; The evaluation module measures the angle between the temperature and humidity gradient trend lines at different characteristic operation points to obtain the coordination angle, and sets the boundary of the optimized operation domain according to the range of the coordination angle; it performs matching analysis between the temperature and humidity field distribution and the boundary of the optimized operation domain to obtain the matched temperature and humidity field distribution that meets the boundary conditions of the optimized operation domain. The trained intelligent analysis model calculates the thermal stress distribution and drying shrinkage stress distribution corresponding to the matched temperature and humidity field distribution; the thermal stress distribution and drying shrinkage stress distribution are combined to form the structural state assessment result, and the maximum stress value in the structural state assessment result is compared with the concrete tensile strength threshold to obtain the parameter correction amount of the steam curing process parameters. The judgment module is used to determine the risk level of microcrack formation in the precast beam based on the structural condition assessment results. By integrating the risk level of microcrack generation with parameter correction, a personalized steam curing control strategy for the current precast beam is generated. The optimization module is used to execute personalized steam curing control strategies and adaptively optimize the trained intelligent analysis model based on the difference between the actual steam curing quality of the precast beam and the expected quality target.

7. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 5.

Images