Low-carbon solid waste cementing material production control method and system based on artificial intelligence

Abstract

The invention discloses a low-carbon solid waste cementing material production control method and system based on artificial intelligence, and relates to the technical field of production control. According to the method, historical production data are acquired, a machine learning algorithm is utilized to train a quality prediction model and a production process control variable data relation prediction model respectively, target variable data influencing the production process are collected in real time and input into the two models, corresponding prediction results are acquired respectively, and the production process control variable data relation prediction model is obtained by combining model output. And correcting the variable data, obtaining an optimal correction parameter according to the correction strategy database, and converting the optimal correction parameter into a production equipment control parameter to realize intelligent adjustment of the production process. According to the method, the problem of insufficient control precision in the production process of the low-carbon solid waste cementing material is solved, accurate prediction and dynamic optimization of the production process are realized, and the quality stability and the production efficiency of the low-carbon solid waste cementing material are effectively improved.

Description

Technical Field

[0001] This application relates to the field of production control technology, and particularly to a low-carbon solid waste cementitious material production control method and system based on artificial intelligence. Background Art

[0002] Low-carbon solid waste cementitious materials are widely used in fields such as construction and environmental protection, and play an important role especially in the process of solid waste treatment. Similar prior arts include a Chinese patent with the publication number CN118192456A, which proposes an automatic glue adding method for film production. By monitoring and analyzing the production demand status parameters in the process of forage film production, the glue adding value is obtained, and the glue adding parameters are dynamically adjusted based on this. At the same time, based on this, the glue quality status parameters, glue adding equipment status parameters, and environmental temperature in the current monitoring time period during the forage film production process are analyzed and processed to obtain the glue adding influence coefficient, so as to analyze and calibrate the glue adding value. On the one hand, it ensures that the glue adding effect in the production process is consistent with the actual demand, reducing the situation of over-glue adding or under-glue adding. On the other hand, it helps to timely detect and correct the glue adding value deviation caused by factors such as glue quality, glue adding equipment, and environmental temperature, thereby realizing the automatic adjustment of glue adding parameters, improving production efficiency and product quality, and realizing intelligent production. In addition, a similar prior art is a Chinese patent with the publication number CN107291039A, which proposes an automatic batching control method for rubber powder. This method is jointly realized by hardware and software. The software is installed in a PLC controller system, and the hardware includes several powder feeding devices, several colloid feeding devices, and an air compressor. The discharge ports of each device are concentrated together, and a container cup is provided at the lower end of the discharge port. The container cup is placed on a precision weighing instrument; the powder feeding device is connected to the PLC controller, and the PLC controller controls the feeding speed, vibration frequency, motor speed, and operating voltage magnitude of the powder feeding device through the weight feedback information measured by the precision weighing instrument, and achieves precise proportioning, improving the batching accuracy and repeatability, and can be mechanically operated in a closed environment, which can effectively ensure that the raw materials will not cause product quality problems due to environmental humidity. Although the technical solutions in the above two patent documents have both solved the problems of adding materials or batching in the production process, they do not consider that when liquid admixtures are directly added to cementitious materials, problems such as uneven dispersion and agglomeration are likely to occur, resulting in unstable material properties in the production process, and thus prone to quality problems. Summary of the Invention

[0003] The present application provides a production control method and system for low-carbon solid waste cementitious materials based on artificial intelligence, which is used to precisely adjust various control variables during the production process of low-carbon solid waste cementitious materials, effectively improving production efficiency, reducing errors and unqualified products during the production process, and ensuring the quality stability of the cementitious materials and the sustainability of production through intelligent control and correction schemes. The method includes: Obtain historical production data, train a first model based on the historical target variable data and corresponding quality data in the historical production data through a machine learning algorithm, and train a second model based on the historical target variable data. Collect target variable data affecting production quality in real time, input the target variable data into the first model and the second model, and obtain a first prediction result and a second prediction result respectively. Obtain the number of differential variables from the second prediction result and the target variable data, and determine whether the second model needs to be retrained based on the number. When necessary, train the second model. Correct the target variable data according to the second prediction result and the second model, obtain and input the corrected data into the first model to obtain a third prediction result, and obtain the optimal correction parameter according to the third prediction result, the corrected data and the correction strategy database. Adjust the output information of the production equipment based on the optimal correction parameter.

[0004] As a preferred technical solution of the present invention, the training of the second model includes: Use the historical target variable data as the second sample data, and train the second model through the second sample data. The second model learns the correlation between the historical target variable data through a machine learning algorithm, extracts key information from the historical target variable data, encodes the key information, and also restores the historical target variable data based on the encoded key information and the learned correlation between the historical target variable data.

[0005] As a preferred technical solution of the present invention, the correction of the target variable data includes: Compare the second prediction result with the input target variable data one by one to obtain the variable difference corresponding to each variable data, regard the variable with the variable difference greater than the set threshold as a differential variable, and count the number of differential variables. When the difference variable does not exist in the target variable data, the target variable data does not need to be corrected. When the number of difference variables is less than the first set number, the remaining variable data of the target variable data excluding the difference variables is input into the second model to obtain a set of predicted variable data. The values corresponding to the difference variables in the set of predicted variable data are used to replace the values of the difference variables in the target variable data, and the replaced target variable data is used as the corrected data.

[0006] As a preferred technical solution of the present invention, determining whether the second model needs to be retrained includes: When the number of difference variables is greater than the second set number and the first prediction result is normal, the target variable data is used as the data to be learned. Multiple sets of the data to be learned are continuously collected and added to the second sample data corresponding to the second model, and the second model is retrained based on the second sample data.

[0007] As a preferred technical solution of the present invention, obtaining the optimal correction parameter includes: When there is no difference variable in the target variable data or after correction, the target variable data is input into the first model to obtain the first prediction result. The first prediction result is compared with the quality standard to obtain a comparison result; When the comparison result is less than or equal to the set value, the quality is qualified; when the comparison result is greater than the set value, the quality is unqualified. Based on the input data of the first prediction result, a variable data set with a similarity greater than the preset similarity to the target variable data and the first prediction result is searched in the correction strategy database, and the control parameters corresponding to the target variable data are corrected according to the correction strategy corresponding to each variable combination in the variable data set to obtain a correction parameter. The correction parameter is input into the first model to obtain a third prediction result, and the correction parameter corresponding to the third prediction result closest to the quality standard is used as the optimal correction parameter.

[0008] As a preferred technical solution of the present invention, the target variable data at least includes the spraying amount and spraying speed of the liquid admixture, as well as the stirring speed, stirring time, mixing uniformity and material temperature of the stirring equipment.

[0009] As a preferred technical solution of the present invention, the quality data at least includes the compressive strength and stability of the low-carbon solid waste cementitious material.

[0010] The present invention also provides a production control system for low-carbon solid waste cementitious materials based on artificial intelligence for implementing the above method. The system includes: A training unit, configured to obtain historical production data, and train a first model based on historical target variable data and corresponding quality data in the historical production data through a machine learning algorithm, and train a second model based on the historical target variable data; A prediction unit, configured to collect in real time target variable data affecting production quality, input the target variable data into the first model and the second model, and respectively obtain a first prediction result and a second prediction result; A judgment unit, configured to obtain the number of differential variables from the second prediction result and the target variable data, judge whether the second model needs to be retrained according to the number, and retrain the second model when necessary; A correction unit, configured to correct the target variable data according to the second prediction result and the second model, obtain and input the corrected data into the first model to obtain a third prediction result, and obtain the optimal correction parameter according to the third prediction result, the corrected data and a correction strategy database; A control unit, configured to control the output information of production equipment based on the optimal correction parameter.

[0011] The present invention further provides a computer-readable storage medium, on which instructions are stored, and when the instructions are executed by a processor, the above method is implemented.

[0012] Effect Through the quality prediction model and the production process control variable data relationship prediction model trained by the machine learning algorithm, the present invention provides accurate prediction capabilities for the production process. These models can analyze and predict the relationships between multiple variable data affecting production quality, enabling real-time adjustment of the production process, so as to ensure that the quality of the final product meets the predetermined standards. This technical solution avoids quality fluctuations caused by variable fluctuations through multiple predictions and optimized control, and improves production stability; secondly, the system collects in real time target variable data affecting production quality, and inputs these data into the trained prediction models to obtain prediction results respectively. By comparing the prediction results with the actual data, the system can timely detect deviations and perform corrections. The corrected data will be re-input into the model to further improve the prediction accuracy, thereby realizing dynamic optimization and precise control in the production process. This closed-loop control mechanism effectively avoids quality fluctuations caused by equipment performance aging or environmental changes, and improves production consistency and stability; in addition, the present invention dynamically adjusts variable data through a correction strategy database. When the prediction result does not meet the quality standard, the system can automatically select the optimal correction parameter according to similar situations in historical data and perform real-time adjustment on production equipment. Through this automated adjustment process, the complexity of manual intervention is reduced, and the efficiency of the production process and the response speed of quality control are improved. Description of the Drawings

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0014] Figure 1 It is a flowchart of the production control method of low-carbon solid waste cementitious materials based on artificial intelligence in the embodiment; Figure 2 It is a flowchart of the correction method of target variable data in the embodiment; Figure 3 It is a flowchart of the method for obtaining the optimal correction parameters in the embodiment; Figure 4 It is a structural diagram of the production control system of low-carbon solid waste cementitious materials based on artificial intelligence in the embodiment. Specific Embodiments

[0015] The embodiments of the present application provide a production control method and system for low-carbon solid waste cementitious materials based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and above-mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0016] For ease of understanding, the following describes the specific process of the embodiments of the present application. An embodiment of the production control method of low-carbon solid waste cementitious materials based on artificial intelligence in the embodiments of the present application is as Figure 1 shown and includes: Step S1: Obtain historical production data, train a first model based on the historical target variable data and corresponding quality data in the historical production data through a machine learning algorithm, and train a second model based on the historical target variable data; Specifically, in the production process of low-carbon solid waste cementitious materials, when liquid admixtures are directly added to the cementitious materials, problems such as uneven dispersion and agglomeration are likely to occur, resulting in unstable material properties during the production process, and thus prone to quality problems. Therefore, in order to ensure its uniform distribution and then improve the quality and production stability of the cementitious materials, atomizing devices are widely used for the dispersion of liquid admixtures, and the spray volume, spray speed of the atomizing device corresponding to the liquid admixture, the stirring speed, stirring time, mixing uniformity of the stirring equipment, and the material temperature are taken as target variables, and the compressive strength and stability of the cementitious materials are taken as quality data. ， By obtaining historical target variable data and corresponding quality data, machine learning algorithms are used to train the first model and the second model. The first model is a quality prediction model that learns the relationship between quality data and historical target variables during the production process based on historical data. The second model: a production process historical target variable data relationship prediction model that establishes the correlation between target variables by learning the historical target variable data in historical production data. By training these two models, accurate first and second models can be established based on historical data, providing reference and basis for subsequent production control, and ensuring the accurate modeling of the relationship between target variables and the quality of the final product during the production process.

[0017] Step S2: Collect target variable data affecting production quality in real time, input the target variable data into the first model and the second model, and obtain the first prediction result and the second prediction result respectively. Specifically, by real-time monitoring multiple target variable data in the production process, these real-time data are automatically input into the first and second models respectively to obtain the first prediction result, that is, the production quality predicted based on the quality prediction model, and the second prediction result, that is, the historical target variable relationship predicted based on the historical control variable data relationship model. The above technical solution can reflect the changes of target variables in the production process in real time, and the real-time prediction can provide effective information for subsequent data correction and adjustment, thus improving the intelligent level of the production process.

[0018] Step S3: Obtain the number of differential variables from the second prediction result and the target variable data, and judge whether the second model needs to be retrained according to the number. When necessary, train the second model. Specifically, based on the second prediction result and the actually collected target variable data, the system will compare the data differences of each variable to determine whether there are significant differences: Differential variable: If the data difference of a certain variable exceeds the set threshold, it is regarded as a differential variable. Number of differential variables: Count the number of differential variables and determine whether the second model needs to be retrained based on the number. Through the above technical solution, the accuracy of the target variable data is ensured. When there are significant deviations in the target variable data, they can be detected and corrected in a timely manner. When the number of differential variables is large, it is judged whether to update the model, that is, to perform retraining, to ensure that the model always adapts to the changes in production and further improve the prediction and control accuracy.

[0019] Step S4: Correct the target variable data according to the second prediction result and the second model, obtain and input the corrected data into the first model to obtain a third prediction result, and obtain the optimal correction parameter according to the third prediction result, the corrected data, and the correction strategy database. Specifically, through the corrected target variable data, the system will re-input it into the first model to obtain a third prediction result. At the same time, the system retrieves historical correction data similar to the third prediction result from the correction strategy database and extracts the optimal correction parameter from it. The correction strategy database contains historical data of different production variables and correction measures, which can find a suitable adjustment strategy for the current production process based on similarity. It also selects the optimal correction parameter based on the quality prediction result and the correction strategy database and inputs it into the production equipment to adjust the actual production process. Through the above technical solution, the output information of the production equipment can be accurately adjusted to ensure that each variable in the production process can be as close as possible to the predetermined target. The automatic adjustment optimizes the production process, reduces human intervention, and improves production efficiency and product quality stability.

[0020] Step S5: Adjust the output information of the production equipment based on the optimal correction parameter.

[0021] Specifically, based on the optimal correction parameter obtained in Step S4, the system transmits the corrected control parameter to the production equipment to adjust the operations in the production process, such as the dispersion of liquid admixtures, and the spraying amount, spraying speed of liquid admixtures, and stirring speed, stirring time, mixing uniformity, and material temperature of the stirring equipment, to ensure that the production quality meets the target requirements. Through the above technical solution, by real-time adjusting the control parameters of the production equipment, it is ensured that the quality control in the production process is always in the best state. By automatically obtaining the optimal correction parameter and controlling the equipment, it reduces human intervention and improves production efficiency and quality consistency.

[0022] Furthermore, the training of the first model and the second model includes: Use the historical target variable data as the second sample data, and train the second model with the second sample data. The second model learns the correlation relationships between the historical target variable data through machine learning algorithms, extracts key information from the historical target variable data, encodes the key information, and also decodes and reconstructs the historical target variable data based on the encoded key information and the learned correlation relationships between the historical target variable data.

[0023] Specifically, in the production process of low-carbon solid waste cementitious materials, when liquid admixtures are directly added to the cementitious materials, problems such as uneven dispersion and agglomeration are likely to occur, resulting in unstable material properties during the production process and thus prone to quality problems. Therefore, in order to timely adjust the above-mentioned relevant variables that affect quality problems, that is, the above-mentioned historical target variable data, based on machine learning algorithms, learn the correlation relationships between the quality data and the relevant variable data in the historical production data, and establish a first model for predicting quality. Among them, the above-mentioned quality data at least includes the compressive strength and stability of low-carbon solid waste cementitious materials, and the above-mentioned relevant variable data at least includes the spraying amount, spraying speed of liquid admixtures, and the stirring speed, stirring time, mixing uniformity, and material temperature of the stirring equipment. It effectively realizes the accurate prediction and dynamic optimization of quality during the production process, ensuring the quality stability of cementitious material production. Since there are also certain correlation relationships between the above-mentioned historical target variable data on the premise of the stable production quality of the above-mentioned low-carbon solid waste cementitious materials, the production quality is ensured through the mutual adaptation between the above-mentioned historical target variable data. Therefore, through the second model, analyze the mutual relationships between the historical target variable data in the above-mentioned historical production data, learn the association rules between the target variables. During the training process of the second model, first extract key information from the above-mentioned historical target variable data. The above-mentioned key information is the N data with the top ranking in terms of the influence degree on the above-mentioned production quality, encode it, and decode it based on the encoded key information and the learned correlation relationships between the historical target variable data, and use the decoded data as the output data of the second model. By comparing the input data and output data of the second model, it can be judged whether the input data is accurate. And when the input data of the second model does not completely include the above-mentioned target variable data, predict the missing variables in the input data through the learned correlation relationships between the variables, and output the complete predicted target variable data. Through the above technical solutions, it is possible to judge whether the target variable data input into the first model by the second model is accurate, avoiding the deterioration of the quality prediction accuracy caused by inaccurate acquisition of the above-mentioned target variable data during the production process, improving the accurate prediction and dynamic optimization of production quality during the production process, ensuring the quality stability of cementitious material production, and at the same time laying a foundation for optimizing the above-mentioned target variable data.

[0024] Further, the correction of the target variable data, such as Figure 2 shown, includes: Comparing the second prediction result with the input target variable data one by one to obtain the variable difference corresponding to each variable data, taking the variable with the variable difference greater than the set threshold as the differential variable, and counting the number of the differential variables; When there is no differential variable in the target variable data, the target variable data does not need to be corrected. When the number of the differential variables is less than the first set number, the remaining variable data of the target variable data excluding the differential variables is input into the second model to obtain a predicted variable data set, and the values corresponding to the differential variables in the predicted variable data set are used to replace the values of the differential variables in the target variable data, and the replaced target variable data is used as the corrected data.

[0025] Specifically, during the production process of low-carbon solid waste cementitious materials, due to performance changes caused by equipment performance loss and aging or changes in the production environment, the output of the control parameters, i.e., the above-mentioned target variable data corresponding to the equipment, may deviate from the preset standard, resulting in unstable production quality or non-compliance with expectations. Traditional control methods are difficult to accurately and real-time adjust multiple variables. Especially when the production quality is about to show abnormalities and no timely and effective correction is made, it will lead to fluctuations in production quality. For example, when the spraying amount and spraying speed of the atomizing device for liquid admixtures do not match the stirring speed, stirring time, mixing uniformity, and material temperature of the stirring equipment, the production quality is unqualified. Therefore, by respectively inputting the above-mentioned target variable data collected in real time into the above-mentioned first model and the second model, and respectively obtaining the above-mentioned first prediction result and the above-mentioned second prediction result, wherein the data types of the above-mentioned target variable data and the above-mentioned historical target variable data are the same, comparing each variable data in the above-mentioned second prediction result with each variable data corresponding to the above-mentioned multiple variables, and obtaining the variable difference corresponding to each variable. When the variable difference corresponding to each variable is less than or equal to the above-mentioned set threshold, that is, the above does not exist, it indicates that the above-mentioned target variable data satisfy the correlation relationship and no correction is required; taking the variable corresponding to the variable difference greater than the above-mentioned set threshold in the above-mentioned target variables as the above-mentioned differential variable, that is, the variable with deviation. When the number of the above-mentioned differential variables is less than the first set number, the first set number is 2, that is, when there is a deviation in one of the above-mentioned target variable data, the above-mentioned second model can output the prediction variable data group with the highest similarity based on the remaining variable data after excluding the differential variable from the target variable data, and replacing the value corresponding to the above-mentioned abnormal variable in the above-mentioned prediction variable data group with the value of the differential variable in the above-mentioned target variable data, and using the replaced above-mentioned target variable data as the above-mentioned correction data. The above technical solution, through the mutual cooperation of the first model and the second model, can not only verify the accuracy of the above-mentioned target variable data, but also correct the above-mentioned target variable data, further improving the accuracy of the prediction result of the above-mentioned second model.

[0026] Further, determining whether the second model needs to be retrained includes: When the number of the differential variables is greater than or equal to the second set number and the first prediction result is normal, the above-mentioned target variable data is used as the data to be learned, multiple groups of the above-mentioned data to be learned are continuously collected and added to the second sample data corresponding to the second model, and the second model is retrained again based on the second sample data.

[0027] Specifically, when the above-mentioned difference variable data is greater than or equal to the above-mentioned first set quantity and less than the above-mentioned second set quantity, it is possible that the data collected by the above-mentioned sensor acquisition unit is incorrect. At this time, the first prediction result output by inputting the above-mentioned target variable data into the first model cannot guarantee accuracy. Moreover, due to the large number of variables, it cannot be accurately corrected by the above-mentioned second model. Therefore, a fault code or warning information can be sent to the user terminal for timely maintenance or production stoppage to prevent quality problems. However, when the above-mentioned number of difference variables is greater than or equal to the above-mentioned second set quantity, where the above-mentioned second set quantity is 50% of the total number of types of the above-mentioned target variable data and rounded up, and when the above-mentioned first prediction result is normal, it is possible that due to environmental changes and equipment performance fluctuations during the production process, there may be significant differences in the above-mentioned target variable data. During the production process, the updates of the above-mentioned first model and the above-mentioned second model are not synchronized or the above-mentioned second model is not updated in a timely manner, resulting in some production modes or production parameter combinations in the above-mentioned production that the above-mentioned second model has not learned. This makes the reconstruction result of the above-mentioned second model for the input above-mentioned target variable data quite different from the input data. Therefore, the above-mentioned target variable data is used as the above-mentioned data to be learned, multiple groups of the above-mentioned data to be learned are collected and added to the second sample data corresponding to the above-mentioned second model, and the above-mentioned second model is retrained. The retrained model will be able to better learn the relationships between new control variables and better cooperate with the above-mentioned first model, thereby improving the prediction ability of production quality. Through more accurate prediction, more precise control parameters can be provided for the production process to ensure the achievement of quality standards.

[0028] Furthermore, the acquisition of the optimal correction parameter is as Figure 3 shown and includes: When there are no difference variables in the above-mentioned target variable data or after correction, input it into the above-mentioned first model to obtain the above-mentioned first prediction result, compare the above-mentioned first prediction result with the quality standard to obtain a comparison result; When the comparison result is less than or equal to the set value, the quality is qualified; when the comparison result is greater than the set value, the quality is unqualified. Based on the above-mentioned first prediction result and the corresponding input target variable data, search the correction strategy database for a variable data set with a similarity greater than the preset similarity, and correct the above-mentioned target variable data according to the correction strategy corresponding to each variable combination in the variable data set to obtain a correction parameter. Input the correction parameter into the above-mentioned first model to obtain a third prediction result, and use the correction parameter corresponding to the third prediction result closest to the above-mentioned quality standard as the optimal correction parameter.

[0029] Specifically, since there is no correlation between the above-mentioned target variable data or calibration data of the differential variables that conforms to the correlation between variables, when inputting the above first model, an accurate above-mentioned first prediction result can be obtained. By comparing the first prediction result with the quality standard, it is judged whether the current production quality is qualified. If there is a large difference between the first prediction result and the quality standard, that is, the prediction result does not meet the set quality standard. Since multiple variables are interrelated, the unqualified production quality is the result of the interaction of multiple variables. It is difficult to determine which specific control parameter of the equipment should be adjusted to improve the production quality. Therefore, historical data with a high similarity to the current prediction result is searched in the database, and the corresponding correction strategy is extracted. The correction strategy database contains the target variable data set, corresponding quality data, and correction strategies in the historical production process. These data sets are associated with various quality deviations, control strategies, and correction parameters in the production process. According to the correction strategy corresponding to the variable group with a high similarity to the current target variable data and the first prediction result, that is, the predicted quality data, the control parameters of the current equipment are corrected, and the corrected parameters, that is, the above-mentioned target variable data output by the corrected production equipment, are simulated and input into the above first model again to obtain the above third prediction result. The above correction strategy at least includes adjusting the spraying amount, spraying speed, stirring speed, stirring time, mixing uniformity, and material temperature, etc., and the above correction parameter corresponding to the third prediction result closest to the above quality standard is used as the above optimal correction parameter. The above technical solution can automatically select the most suitable correction parameter through the correction strategy database, reduce the cumbersome steps of manual adjustment, improve the production efficiency and the response speed and accuracy of quality control, ensure that the quality of the cementitious material meets the quality standard, reduce quality fluctuations, production downtime, and the appearance of unqualified products, and improve the overall production efficiency.

[0030] Further, the target variable data at least includes the spraying amount and spraying speed of the liquid admixture, and the stirring speed, stirring time, mixing uniformity, and material temperature of the stirring equipment.

[0031] Further, the quality data at least includes the compressive strength and stability of the low-carbon solid waste cementitious material.

[0032] Specifically, both the above-mentioned compressive strength and stability are measured and quantified data. For example, the compressive strength and stability can be divided into multiple grades respectively to quantify the compressive strength and stability of the low-carbon solid waste cementitious material.

[0033] The present invention also provides a production control system for low-carbon solid waste cementitious materials based on artificial intelligence, which is used to implement the above method, as Figure 4 shown, the system includes: A training unit, configured to obtain historical production data, and train a first model based on historical target variable data and corresponding quality data in the historical production data through a machine learning algorithm, and train a second model based on the historical target variable data; A prediction unit, configured to collect in real time target variable data affecting production quality, input the target variable data into the first model and the second model, and respectively obtain a first prediction result and a second prediction result; A judgment unit, configured to obtain the number of differential variables from the second prediction result and the target variable data, judge whether the second model needs to be retrained according to the number, and retrain the second model when necessary; A correction unit, configured to correct the target variable data according to the second prediction result and the second model, obtain and input the corrected data into the first model to obtain a third prediction result, and obtain the optimal correction parameter according to the third prediction result, the corrected data and a correction strategy database; A control unit, configured to control the output information of production equipment based on the optimal correction parameter.

[0034] The present invention also provides a computer-readable storage medium, on which instructions are stored, and when the instructions are executed by a processor, the above method is implemented.

[0035] In summary, the quality prediction model and the production process control variable data relationship prediction model trained by the machine learning algorithm of the present invention provide accurate prediction capabilities for the production process. These models can analyze and predict the relationships between multiple variable data affecting production quality, enabling the production process to be adjusted in real time, so as to ensure that the quality of the final product meets the predetermined standards. This technical solution avoids quality fluctuations caused by variable fluctuations through multiple predictions and optimized control, and improves production stability; secondly, the system collects in real time target variable data affecting production quality and inputs these data into the trained prediction models to obtain prediction results respectively. By comparing the prediction results with the actual data, the system can timely detect deviations and perform corrections. The corrected data will be re-input into the model to further improve the prediction accuracy, thereby realizing dynamic optimization and precise control in the production process. This closed-loop control mechanism effectively avoids quality fluctuations caused by equipment performance aging or environmental changes, and improves production consistency and stability; in addition, the present invention dynamically adjusts variable data through a correction strategy database. When the prediction result does not meet the quality standard, the system can automatically select the optimal correction parameter according to similar situations in historical data and perform real-time adjustment on production equipment. Through this automated adjustment process, the complexity of manual intervention is reduced, and the efficiency of the production process and the response speed of quality control are improved.

[0036] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the foregoing method embodiments and will not be described herein again.

[0037] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0038] The above is the case. The above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or equivalently replace some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A production control method for low-carbon solid waste cementitious materials based on artificial intelligence, characterized in that The method includes: Obtaining historical production data, training a first model based on historical target variable data and corresponding quality data in the historical production data through a machine learning algorithm, and training a second model based on the historical target variable data; Collecting target variable data affecting production quality in real time, inputting the target variable data into the first model and the second model, and respectively obtaining a first prediction result and a second prediction result; Obtaining the number of difference variables from the second prediction result and the target variable data, judging whether the second model needs to be retrained according to the number, and training the second model when necessary; Correcting the target variable data according to the second prediction result and the second model, obtaining and inputting the corrected data into the first model to obtain a third prediction result, and obtaining the optimal correction parameter according to the third prediction result, the corrected data and the correction strategy database; Adjusting the output information of the production equipment based on the optimal correction parameter.

2. The method according to claim 1, wherein The training of the second model includes: Taking the historical target variable data as the second sample data, and training the second model through the second sample data. The second model learns the correlation between the historical target variable data through a machine learning algorithm, extracts key information from the historical target variable data, encodes the key information, and also restores the historical target variable data based on the encoded key information and the learned correlation between the historical target variable data.

3. The method according to claim 1, characterized in that The correction of the target variable data includes: Comparing the second prediction result with the input target variable data one by one to obtain the variable difference corresponding to each variable data, taking the variable with the variable difference greater than the set threshold as the difference variable, and counting the number of the difference variables; When there is no difference variable in the target variable data, the target variable data does not need to be corrected. When the number of difference variables is less than the first set number, the remaining variable data of the target variable data excluding the difference variables is input into the second model to obtain a predicted variable data group, and the value corresponding to the difference variable in the predicted variable data group is used to replace the value of the difference variable in the target variable data, and the replaced target variable data is used as the corrected data.

4. The method according to claim 1, wherein Judging whether the second model needs to be retrained includes: When the number of difference variables is greater than the second set number and the first prediction result is normal, the target variable data is used as the data to be learned, multiple groups of the data to be learned are continuously collected and added to the second sample data corresponding to the second model, and the second model is retrained based on the second sample data.

5. The method according to claim 1, characterized in that, The obtaining of the optimal correction parameter includes: When there is no difference variable in the target variable data or after correction, inputting it into the first model to obtain the first prediction result, comparing the first prediction result with the quality standard to obtain a comparison result; When the comparison result is less than or equal to a set value, the quality is qualified; when the comparison result is greater than the set value, the quality is unqualified. Based on the input data of the first prediction result, find in the correction strategy database a variable data set whose similarity to the target variable data and the first prediction result is greater than a preset similarity, and correct the control parameters corresponding to the target variable data according to the correction strategy corresponding to each variable combination in the variable data set to obtain correction parameters. Input the correction parameters into the first model to obtain a third prediction result, and use the correction parameters corresponding to the third prediction result that is closest to the quality standard as the optimal correction parameters.

6. The method according to claim 1, wherein The target variable data at least includes the spraying amount and spraying speed of the liquid admixture, as well as the stirring speed, stirring time, mixing uniformity, and material temperature of the stirring equipment.

7. The method according to claim 1, wherein The quality data at least includes the compressive strength and stability of the low-carbon solid waste cementitious material.

8. A production control system for a low-carbon solid waste cementitious material based on artificial intelligence, which is used to implement the method described in any one of claims 1-7, characterized in that, The system includes: A training unit, configured to obtain historical production data, and train a first model based on the historical target variable data and the corresponding quality data in the historical production data through a machine learning algorithm, and train a second model based on the historical target variable data; A prediction unit, configured to collect in real time the target variable data affecting production quality, input the target variable data into the first model and the second model, and respectively obtain a first prediction result and a second prediction result; A judgment unit, configured to obtain the number of differential variables from the second prediction result and the target variable data, and judge whether the second model needs to be retrained according to the number. When necessary, train the second model; A correction unit, configured to correct the target variable data according to the second prediction result and the second model, obtain and input the corrected data into the first model to obtain a third prediction result, and obtain the optimal correction parameters according to the third prediction result, the corrected data, and the correction strategy database; A control unit, configured to control the output information of the production equipment based on the optimal correction parameters.

9. A computer-readable storage medium having instructions stored thereon, characterized in that, When the instructions are executed by a processor, the method described in any one of claims 1-7 is implemented.

Images