Ultrathin ceramic tile batching optimization method and system based on real-time data acquisition

Abstract

The invention discloses an ultra-thin ceramic tile batching optimization method and system based on real-time data acquisition, and relates to the technical field of ultra-thin ceramic tiles. Comprising the following steps: S1, obtaining a final adjustment parameter according to the difference and external environment data; s2, comparing a simulation result corresponding to the adjustment parameter with a simulation result corresponding to a standard formula, if the comparison results are the same, directly executing the next step, otherwise, returning to the previous step; and S3, determining the dosage of each raw material according to the raw material formula corresponding to the compared simulation result, and adjusting the production formula dosage of the current batch of raw materials according to the dosage. According to the present invention, the adjustment parameter is obtained, and the current formula is compared with the standard formula according to the adjustment parameter and the simulation model, such that the use amount of each raw material in the corresponding formula can be determined on the premise of ensuring the production result of the current batch of raw materials.

Description

Technical Field

[0001] The invention relates to the technical field of ultra-thin ceramic tiles, and in particular to an ultra-thin ceramic tile batching optimization method and system based on real-time data collection. Background Art

[0002] With the development of the construction industry and the increasing demand of consumers for the quality of building materials, ultra-thin ceramic tiles have become widely popular due to their advantages such as light weight, high strength and beautiful appearance. However, during the production process, ensuring the consistency of each batch of products, especially color consistency, has always been a major challenge for manufacturers. This is mainly because factors such as fluctuations in raw material composition and changes in environmental conditions will affect the quality of the final product.

[0003] Traditionally, the production process of ceramic tiles relies on manual experience and fixed formulas to control the proportion of ingredients. This method has several significant problems: Human error: Reliance on operator experience may result in inaccurate dosing, which can affect product quality.

[0004] Slow response: When raw materials change, traditional adjustment methods often lag behind and cannot respond in time, resulting in low production efficiency.

[0005] High cost: In order to ensure product quality, sometimes more expensive or more stable raw materials have to be used, which increases production costs.

[0006] Waste of resources: Due to the lack of precise control, some materials may be overused, resulting in unnecessary waste.

[0007] In recent years, with the advancement of sensor technology and data analysis capabilities, the degree of industrial automation has continued to improve. Especially in the manufacturing industry, the use of sensors for real-time monitoring combined with advanced data analysis technology has become one of the important means to improve production efficiency and product quality. By collecting key parameters in the production process in real time and analyzing them using intelligent algorithms such as machine learning, more refined control of the production process can be achieved.

[0008] The Chinese invention patent with publication number CN113413973A discloses a high-strength ultra-thin ceramic tile intelligent batching device, including a storage mechanism, a weighing mechanism, a front sealing mechanism, a ball milling mechanism, a rear sealing mechanism, a fine grinding mechanism and a control mechanism arranged in sequence according to the production process; the storage mechanism is used to store various raw materials after homogenization and add the raw materials separately; the weighing mechanism is used to weigh the raw materials added by the storage mechanism; the front sealing mechanism is used to add a certain amount of raw materials to the ball milling mechanism and seal the front end opening of the ball milling mechanism; the ball milling mechanism is used to mix and ball mill the certain amount of raw materials and prepare them into coarse powder; the rear sealing mechanism is used to seal the rear end opening of the ball milling mechanism and spray the coarse powder after ball milling into the fine grinding mechanism; the fine grinding mechanism is used to finely grind the coarse powder and prepare it into fine powder. The intelligent batching device for high-strength ultra-thin ceramic tiles of the present invention ensures that the raw materials meet the production requirements and improves production efficiency.

[0009] The batching method of the above-mentioned and similar intelligent batching devices mainly depends on the fixed formula ratio and the operator's work experience for actual adjustment. However, since the iron oxide content in different batches of raw materials cannot be kept exactly the same, there will be slight differences between the batches of raw materials. Therefore, if the formula is not adjusted, the color of the final product will have obvious color difference, which will affect the market competitiveness of the product. If the formula is adjusted, it needs to be readjusted, which will reduce production efficiency. Summary of the invention

[0010] The purpose of the present invention is to provide a method and system for optimizing the ingredients of ultra-thin ceramic tiles based on real-time data collection to solve the problems raised in the above-mentioned background technology.

[0011] To achieve the above object, the present invention provides the following technical solution: a method for optimizing the ingredients of ultra-thin ceramic tiles based on real-time data collection, comprising: S1: Obtain adjustment parameters: Obtain the difference in composition between the current batch of raw materials and the standard formula through the key parameters of the current batch of raw materials, and obtain the final adjustment parameters based on the difference and external environmental data; S2: Obtain simulation results: According to the adjustment parameters and the simulation model, obtain the simulation results corresponding to the adjustment parameters and the standard formula, and compare the simulation results corresponding to the adjustment parameters with the simulation results corresponding to the standard formula. When the comparison results are the same, directly execute the next step, otherwise return to the previous step; S3: batching operation: according to the raw material formula corresponding to the compared simulation result, determine the amount of each raw material, and adjust the production formula amount of the current batch of raw materials according to the amount; Get the final adjustment parameters, specifically: S1.1: Obtain key parameters and external environmental data of the current batch of raw materials through deployed sensors, wherein the key parameters include chemical composition, and the external environmental data include temperature and relative humidity in the current production environment, outdoor temperature and humidity, and wind speed; S1.2: Establish multiple prediction models, and obtain the initial difference prediction value corresponding to each prediction model through the key parameters, and determine the final difference prediction value according to each of the initial difference prediction values; S1.3: Establish an environmental monitoring model, and obtain the impact parameters of environmental parameters on the current batch of raw materials through the external environmental data; S1.4: According to the difference prediction value and the influencing parameter, the final adjustment parameter is obtained through the adjustment model, and the adjustment model is specifically: ,in: is the final adjustment parameter, is the weighting coefficient, is the final difference prediction value, It is the impact parameter of environmental parameters on the current batch of raw materials.

[0012] Furthermore, the final difference prediction value is determined as follows: M1: Establish a linear regression model to obtain the first initial difference prediction value, specifically: ,in: is the first initial difference prediction value, is the weight matrix, is the key parameter after preprocessing, is the bias vector;

[0013] M2: Establish a random forest model to obtain the second initial difference prediction value, specifically: ,in: is the second initial difference prediction value, is the total number of decision trees, is the difference prediction value corresponding to the i-th decision tree, is the key parameter after preprocessing, is the index of the decision tree; M3: Establish a support vector machine model to obtain the third initial difference prediction value, specifically: ,in: is the third initial difference prediction value, is the Lagrange multiplier, is the index of the sample, is the total number of samples, is the label of the jth sample in the key parameters after preprocessing, is the bias term, is the kernel function, is the feature vector of the jth sample in the key parameters after preprocessing, is the feature vector of the jth sample in the key parameters after preprocessing currently in operation; M4: Obtain a final difference prediction value according to the first initial difference prediction value, the second initial difference prediction value and the third initial difference prediction value, specifically: ,in: is the final difference prediction value, is the weight corresponding to the first initial difference prediction value, is the first initial difference prediction value, is the weight corresponding to the second initial difference prediction value, is the second initial difference prediction value, is the weight corresponding to the third initial difference prediction value, is the third initial difference prediction value.

[0014] Furthermore, the weight corresponding to the initial difference prediction value is obtained as follows: N1: Obtain the root mean square error corresponding to each prediction model, and normalize each of the root mean square errors; N2: According to the normalized root mean square error, determine the allocation weight corresponding to each of the initial difference prediction values, and obtain the final weight corresponding to each of the initial difference prediction values ​​through the allocation weight, specifically: ,in: is the weight corresponding to the k-th prediction model, is the distribution weight corresponding to the root mean square error after normalization, is the initial weight corresponding to the m-th prediction model, is the index corresponding to the prediction model.

[0015] Furthermore, the influence parameters of the environmental parameters on the current batch of raw materials are obtained, as follows: W1: Establish an initial neural network model to obtain the initial environmental impact coefficient, specifically: ,in: is the initial environmental impact coefficient, is the weight matrix corresponding to the neural network model, is the activation function, is the environmental parameter, is the bias vector; W2: Optimize the initial neural network model through a sliding window algorithm to obtain an optimized neural network model, specifically: ,in: is the optimized environmental impact coefficient, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters; W3: Based on the optimized neural network model, an environmental monitoring model is established to obtain the impact parameters of environmental parameters on the current batch of raw materials. The environmental monitoring model is specifically: ,in: is the impact parameter of environmental parameters on the current batch of raw materials, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters, For the The weights corresponding to the interaction terms between the environmental parameters are For the The interaction terms between the environmental parameters, is the total number of interaction terms between environmental parameters, is the index of the interaction term between environmental parameters.

[0016] Furthermore, the simulation results corresponding to the adjustment parameters are compared with the simulation results corresponding to the standard recipe, as follows: S2.1: taking the adjustment parameters and the standard formula as inputs of a simulation model, and outputting simulation results corresponding to the adjustment parameters and the standard formula; S2.2: Compare the simulation result corresponding to the adjustment parameter with the simulation result corresponding to the standard recipe. If the comparison results are the same, execute step S3; otherwise, execute step S2.3; S2.3: Determine the influencing factors from the environmental parameters and the weighting coefficients, repeat steps S1.4 to S2.2, and re-acquire the adjustment parameters based on the influencing factors.

[0017] Furthermore, the influencing factors are determined from the environmental parameters and weighting coefficients as follows: E1: reacquire the current environmental parameters and compare the current environmental parameters with the test environmental parameters. When the current environmental parameters are the same as the test environmental parameters, execute step E2. Otherwise, reacquire the parameters affecting the current batch of raw materials according to the current environmental parameters. E2: adjusting the size of the weighting coefficient, and reacquiring the adjustment parameter according to the adjusted weighting coefficient and the adjustment model.

[0018] A system for optimizing ultra-thin ceramic tile batching based on real-time data collection, wherein the system for optimizing ultra-thin ceramic tile batching uses any of the above-mentioned methods for optimizing ultra-thin ceramic tile batching based on real-time data collection.

[0019] Compared with the prior art, the present invention has the following beneficial effects: First, the present invention obtains the difference in composition and environmental impact between the current batch of raw materials and the standard formula, determines the adjustment parameters, and compares the current formula with the standard formula based on the adjustment parameters and the simulation model, so that the current batch of raw materials can determine the amount of each raw material in the corresponding formula under the premise of ensuring the production results, thereby improving the quality consistency of the final product; Second, the present invention can quickly respond to changes in the production process by collecting and processing data in real time, adjust the batching parameters in time, reduce production interruptions caused by inaccurate recipes, and support flexible recipe switching; Thirdly, the present invention further ensures the accuracy of raw material ratio by influencing the production process by combining environmental factors, thereby reducing the waste of raw materials and lowering the production cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic flow chart of the method for optimizing the ingredients of ultra-thin ceramic tiles of the present invention; Figure 2 A schematic diagram of a process for obtaining adjustment parameters of the present invention; Figure 3 A schematic diagram of a process for obtaining influencing parameters of the present invention; Figure 4 It is a schematic diagram of a flow chart of comparing the simulation results corresponding to the adjustment parameters with the simulation results corresponding to the standard formula according to the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In the process of specific implementation of the existing intelligent batching device, its batching method mainly relies on the fixed formula ratio and the operator's work experience for actual adjustment. However, since the content of iron oxide in different batches of raw materials cannot be kept exactly the same, there will be slight differences between the batches of raw materials. The technical solution of this application can obtain the difference in composition and environmental impact between the current batch of raw materials and the standard formula by real-time collection and processing of key parameters of the current batch of raw materials, and obtain adjustment parameters based on the difference and environmental impact, and compare the current formula with the standard formula based on the adjustment parameters and the simulation model, so as to determine the amount of each raw material in the corresponding formula of the current batch of raw materials under the premise of ensuring the production results.

[0023] refer to Figure 1-Figure 4 This embodiment provides an ultra-thin ceramic tile ingredient optimization method based on real-time data collection, comprising the following steps: Step S1: Obtain adjustment parameters. That is, collect the key parameters of the current batch of raw materials through the deployed sensors, and obtain the difference in composition and environmental impact between the current batch of raw materials and the standard formula based on the key parameters obtained, and obtain the final adjustment parameters, as follows: Step S1.1: Obtain key parameters of the current batch of raw materials through deployed sensors. Specifically, obtain the chemical composition of the current batch of raw materials, such as iron oxide, silicon dioxide, etc. through a spectrometer. Monitor the temperature and relative humidity in the current production environment through temperature and humidity sensors, and obtain outdoor temperature, humidity and wind speed through an external meteorological service API.

[0024] Further, the key parameters of the current batch of raw materials obtained above are subjected to corresponding preprocessing. The preprocessing operations in the present embodiment include data cleaning, feature extraction and data standardization. Among them, data cleaning and removing outliers from the key parameters obtained, and performing corresponding smoothing operations. Feature extraction is to extract the absorbance value at a specific wavelength from the spectral data, because the size of the absorbance value is related to the chemical composition in the raw material. Data cleaning is to perform corresponding standardization operations on the key data through Z-score standardization operations. Specifically, the data cleaning, feature extraction and data standardization are all existing mature conventional data processing methods, so in the present embodiment, no specific elaboration will be given.

[0025] Step S1.2: Establish multiple prediction models, and use the key parameters preprocessed in step S1.1 as the input of each prediction model, output the initial difference prediction value corresponding to each prediction model, and extract the final difference prediction value from the multiple initial difference prediction values. The details are as follows: Step M1: Establish a linear regression model, and use the key parameters preprocessed in step S1.1 as the input of the linear regression model, and output the first initial difference prediction value. In this embodiment, the linear regression model is specifically: ,in: is the first initial difference prediction value, is the weight matrix, is the key parameter after preprocessing, is the bias vector.

[0026] In the specific implementation process, the key parameters of the pre-treated raw materials in the current batch are: absorbance value A is 0.75, temperature T is 25°C, relative humidity RH is 60%, outdoor temperature is 20°C, outdoor relative humidity RH0 is 70%, wind speed V w 5m / s.

[0027] That is to say, based on the above data, the following key parameters after preprocessing can be obtained: , weight matrix and the bias vector , specifically: , , Therefore, in this embodiment, the first initial difference prediction values ​​obtained by the linear regression model are output as 7.8 and 6.1875, where 7.8 is the difference between the current batch and the standard formula in the first component (iron oxide), and 6.1875 is the difference between the current batch and the standard formula in the second component (silicon dioxide).

[0028] Step M2: Establish a random forest model, and use the key parameters preprocessed in step S1.1 as the input of the random forest, and output the second initial difference prediction value. In this embodiment, the random forest model is specifically: ,in: is the second initial difference prediction value, is the total number of decision trees, is the difference prediction value corresponding to the i-th decision tree, is the key parameter after preprocessing, is the index of the decision tree.

[0029] In the specific implementation process, the key parameters of the pre-treated raw materials in the current batch are: absorbance value A is 0.75, temperature T is 25°C, relative humidity RH is 60%, outdoor temperature is 20°C, outdoor relative humidity RH0 is 70%, wind speed V w At the same time, the number of decision trees in the random forest model in this embodiment is set to 100.

[0030] That is to say, according to the above data, the sum of the difference prediction values ​​of all decision trees for the first component (iron oxide) is 760, and the sum of the difference prediction values ​​of all decision trees for the second component (silicon dioxide) is 660. That is, the second initial difference prediction values ​​obtained by output are 7.6 and 6.6. Among them, 7.6 is the difference between the current batch and the standard formula in the first component (iron oxide), and 6.6 is the difference between the current batch and the standard formula in the second component (silicon dioxide).

[0031] Step M3: Establish a support vector machine model, and use the key parameters preprocessed in step S1.1 as the input of the support vector machine model, and output the third initial difference prediction value. In this embodiment, the support vector machine model is specifically: ,in: is the third initial difference prediction value, is the Lagrange multiplier, is the index of the sample, is the total number of samples, is the label of the jth sample in the key parameters after preprocessing, is the bias term, is the kernel function, is the feature vector of the jth sample in the key parameters after preprocessing, It is the feature vector of the jth sample in the key parameters after preprocessing currently in operation.

[0032] In the specific implementation process, the key parameters of the pre-treated raw materials in the current batch are: absorbance value A is 0.75, temperature T is 25°C, relative humidity RH is 60%, outdoor temperature is 20°C, outdoor relative humidity RH0 is 70%, wind speed V w is 5m / s. At the same time, the bias term in this embodiment Set to 0.1, kernel function Set to 0.1.

[0033] That is to say, through the support vector machine model, the third initial difference prediction values ​​obtained by output are 7.6 and 6.385, where 7.6 is the difference between the current batch and the standard formula in the first component (iron oxide), and 6.385 is the difference between the current batch and the standard formula in the second component (silicon dioxide).

[0034] Step M4: Combine the first initial difference prediction value obtained in step M1, the second initial difference prediction value obtained in step M2, and the third initial difference prediction value obtained in step M3 to obtain a final difference prediction value, specifically: ,in: is the final difference prediction value, is the weight corresponding to the first initial difference prediction value, is the first initial difference prediction value, is the weight corresponding to the second initial difference prediction value, is the second initial difference prediction value, is the weight corresponding to the third initial difference prediction value, is the third initial difference prediction value.

[0035] In this embodiment, the weight corresponding to the first initial difference prediction value is , the weight corresponding to the second initial difference prediction value The weight corresponding to the third initial difference prediction value The sum 3 is 1. At the same time, the specific size of each weight is obtained as follows: Step N1: Obtain the root mean square error corresponding to the linear regression model in step M1, the root mean square error corresponding to the linear regression model in step M2, and the root mean square error corresponding to the support vector machine model in step M3, and normalize the root mean square error corresponding to each model.

[0036] Step N2: According to the normalized root mean square error in step N1, a weight is assigned to each initial difference prediction value, and the assigned weight is normalized, wherein the normalized weight value corresponding to each initial difference prediction value is the final corresponding weight value. In this embodiment, the formula for obtaining the weight value is specifically: ,in: is the weight corresponding to the k-th prediction model, is the distribution weight corresponding to the root mean square error after normalization, is the initial weight corresponding to the m-th prediction model, is the index corresponding to the prediction model.

[0037] In the specific implementation process, the root mean square error corresponding to the linear regression model in step M1 is 0.5, the root mean square error corresponding to the linear regression model in step M2 is 0.4, and the root mean square error corresponding to the support vector machine model in step M3 is 0.6. In other words, the normalized root mean square error corresponding to the linear regression model in step M1 is 0.333, the normalized root mean square error corresponding to the linear regression model in step M2 is 0.267, and the normalized root mean square error corresponding to the support vector machine model in step M3 is 0.4.

[0038] Furthermore, according to the normalized root mean square error corresponding to each of the above-mentioned prediction models, the distribution weight corresponding to the linear regression model in step M1 is 3, the distribution weight corresponding to the linear regression model in step M2 is 3.75, and the distribution weight corresponding to the support vector machine model in step M3 is 2.5. Therefore, after further normalizing the distribution weight, the final weight corresponding to each prediction model can be obtained. Specifically: the final weight corresponding to the linear regression model in step M1 is 0.324, the final weight corresponding to the linear regression model in step M2 is 0.405, and the final weight corresponding to the support vector machine model in step M3 is 0.271. In other words, the final difference prediction values ​​obtained in this embodiment are 7.7 and 6.4.

[0039] Step S1.3: Establish an environmental monitoring model, and obtain the parameters affecting the current batch of raw materials based on the temperature and relative humidity, outdoor temperature, humidity and wind speed in the current production environment obtained in step S1.1. The details are as follows: Step W1: Establish an initial neural network model, process the acquired environmental parameters, and obtain the initial environmental impact coefficient. The initial neural network model in this embodiment is specifically: ,in: is the initial environmental impact coefficient, is the weight matrix corresponding to the neural network model, is the activation function, is the environmental parameter, is the bias vector.

[0040] Step W2: Optimize the initial neural network model in step W1 through the sliding window algorithm, specifically: ,in: is the optimized environmental impact coefficient, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters.

[0041] Step W3: Through the mutual influence of environmental parameters, the optimized neural network model is further optimized to obtain the final environmental parameters affecting the current batch of raw materials, specifically: ,in: is the impact parameter of environmental parameters on the current batch of raw materials, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters, For the The weights corresponding to the interaction terms between the environmental parameters are For the The interaction terms between the environmental parameters, is the total number of interaction terms between environmental parameters, is the index of the interaction term between environmental parameters.

[0042] In the process of specific implementation, the data shown in Table 1 below are obtained, specifically: Table 1 time Temperature / ℃ humidity / % Outdoor temperature / ℃ Outdoor humidity / % Wind speed / m / s Production results T1 25 60 20 70 5 7.6 T2 26 62 21 72 6 7.8 T3 24 59 19 69 4 7.5 T4 27 63 22 73 7 8.0 T5 23 57 18 67 3 7.3 T6 28 65 23 75 8 8.2 T7 23 58 19 68 4 7.5 According to the data in the above table, the current environmental parameters include indoor temperature, indoor humidity, outdoor temperature, outdoor humidity and wind speed, and the weights of the environmental parameters are 0.2, 0.15, 0.1, 0.1, and 0.05, respectively. In this embodiment, only the interaction term between indoor temperature and indoor humidity is referenced, and the weight corresponding to the interaction term is 0.01, so the influence parameter of the environmental parameters on the current batch of raw materials in this embodiment is 38.25.

[0043] Step S1.4: Combine the final difference prediction value (7.7 and 6.4) obtained in step M4 with the parameter (38.25) of the environmental parameter on the current batch of raw materials obtained in step W3 to obtain the final adjustment parameter. In this embodiment, the final adjustment parameter is obtained by adjusting the model, and the adjustment model is specifically: ,in: is the final adjustment parameter, is the weighting coefficient, is the final difference prediction value, It is the impact parameter of environmental parameters on the current batch of raw materials.

[0044] Furthermore, the weighting coefficient in this embodiment is There is no specific requirement for the size setting, which can be adjusted according to the production results of ultra-thin tiles in the actual production process. It is worth noting that the weighting coefficient The initial value of the size is set to 0.5, and its specific size is adjusted according to specific production.

[0045] In the specific implementation process, the weighting coefficient in this embodiment The size of is set to 0.8, so the final adjustment parameters obtained in this embodiment are 13.81 and 12.77.

[0046] Step S2: Get simulation results. That is, according to the final adjustment parameters (13.81 and 12.77) obtained in step S1.4, the final adjustment parameters are used as the input of the simulation model, and the simulation results corresponding to the adjustment parameters are obtained as output, and the simulation results are compared with the simulation results corresponding to the standard formula. When the comparison results of the two are consistent, the next step S3 is executed, otherwise it returns to the previous step S1.4, re-acquires the adjustment parameters, and repeats the current step S2 until the comparison results of the two are consistent. The details are as follows: Step S2.1: The final adjustment parameters (13.81 and 12.77) obtained in step S1.4 are used as inputs of the simulation model, and the simulation results of the ultra-thin ceramic tiles corresponding to the adjustment parameters are prepared through the existing process flow model (such as water-based tape casting preparation). Further, the simulation results of the ultra-thin ceramic tiles corresponding to the standard formula are prepared with the same process flow model (such as water-based tape casting preparation).

[0047] Step S2.2: Compare the simulation results of the ultra-thin ceramic tiles corresponding to the adjustment parameters obtained in step S2.1 with the simulation results of the ultra-thin ceramic tiles corresponding to the standard formula. If the comparison results are consistent, execute the next step S3, otherwise execute step S2.3 to adjust the weight coefficients. The size is adjusted or the environmental parameters are modified to modify the parameters affecting the current batch of raw materials, and the process returns to step S1.4 to reacquire the adjustment parameters and repeat steps S2.1-S2.2.

[0048] In the specific implementation process, the simulation results of the ultra-thin ceramic tiles corresponding to the adjustment parameters are: density (2.7g / cm³), bending strength (150MPa), and water absorption (0.5%). The simulation results of the ultra-thin ceramic tiles corresponding to the standard formula are: density (2.8g / cm³), bending strength (160MPa), and water absorption (0.4%). In other words, according to the data, it can be seen that there are obvious differences in bending strength and water absorption, so the adjustment parameters need to be re-obtained.

[0049] Step S2.3: Determine the environmental parameters and weighting coefficients respectively, and identify the influencing factors that need to be adjusted. At the same time, according to the determined influencing factors, return to step S1.4, re-acquire the adjustment parameters, and repeat steps S2.1-S2.2. The details are as follows: Step E1: reacquire the current environmental parameters, and compare the current environmental parameters with the test environmental parameters acquired in step 1.3. Specifically, when the current environmental parameters and the test environmental parameters are the same, directly execute the next step E2, otherwise directly adjust the parameters affecting the current batch of raw materials according to the current environmental parameters, and adjust the final adjustment parameters according to the parameters affecting the current batch of raw materials after the adjustment of the environmental parameters.

[0050] Step E2: Adjust weighting coefficients The size of the weighting factor is increased or decreased by the same multiple. , and according to the weighting coefficient after increase or decrease , adjust the adjustment model and the final adjustment parameters.

[0051] Step S3: batching operation. That is, according to the simulation results compared in step S2.2, determine the formula corresponding to the simulation results, and calculate the new amount of each raw material according to the ratio of each raw material in the formula. And according to the new amount of each raw material, adjust the parameters of the current production equipment and the amount of production ingredients.

[0052] This embodiment also provides an ultra-thin ceramic tile batching optimization system based on real-time data collection. The ultra-thin ceramic tile batching optimization system uses an ultra-thin ceramic tile batching optimization method based on real-time data collection in the above embodiment.

[0053] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is limited by the attached embodiments and their equivalents.

Claims

1. A method for optimizing ultra-thin ceramic tile ingredients based on real-time data collection, characterized in that: Included are: S1: Obtain adjustment parameters: Obtain the difference in composition between the current batch of raw materials and the standard formula through the key parameters of the current batch of raw materials, and obtain the final adjustment parameters based on the difference and external environmental data; S2: Obtain simulation results: According to the adjustment parameters and the simulation model, obtain the simulation results corresponding to the adjustment parameters and the standard formula, and compare the simulation results corresponding to the adjustment parameters with the simulation results corresponding to the standard formula. When the comparison results are the same, directly execute the next step, otherwise return to the previous step; S3: batching operation: according to the raw material formula corresponding to the compared simulation result, determine the amount of each raw material, and adjust the production formula amount of the current batch of raw materials according to the amount; Get the final adjustment parameters, specifically: S1.1: Obtain key parameters and external environmental data of the current batch of raw materials through deployed sensors, wherein the key parameters include chemical composition, and the external environmental data include temperature and relative humidity in the current production environment, outdoor temperature and humidity, and wind speed; S1.2: Establish multiple prediction models, and obtain the initial difference prediction value corresponding to each prediction model through the key parameters, and determine the final difference prediction value according to each of the initial difference prediction values; S1.3: Establish an environmental monitoring model, and obtain the impact parameters of environmental parameters on the current batch of raw materials through the external environmental data; S1.4: According to the difference prediction value and the influencing parameter, the final adjustment parameter is obtained by adjusting the model, and the adjustment model is specifically: ,in: is the final adjustment parameter, is the weighting coefficient, is the final difference prediction value, It is the impact parameter of environmental parameters on the current batch of raw materials.

2. The method for optimizing ultra-thin ceramic tile ingredients based on real-time data acquisition according to claim 1 is characterized in that: The final difference prediction value is determined as follows: M1: Establish a linear regression model to obtain the first initial difference prediction value, specifically: ,in: is the first initial difference prediction value, is the weight matrix, is the key parameter after preprocessing, is the bias vector; M2: Establish a random forest model to obtain the second initial difference prediction value, specifically: ,in: is the second initial difference prediction value, is the total number of decision trees, is the difference prediction value corresponding to the i-th decision tree, is the key parameter after preprocessing, is the index of the decision tree; M3: Establish a support vector machine model to obtain the third initial difference prediction value, specifically: ,in: is the third initial difference prediction value, is the Lagrange multiplier, is the index of the sample, is the total number of samples, is the label of the jth sample in the key parameters after preprocessing, is the bias term, is the kernel function, is the feature vector of the jth sample in the key parameters after preprocessing, is the feature vector of the jth sample in the key parameters after preprocessing currently in operation; M4: Obtain a final difference prediction value according to the first initial difference prediction value, the second initial difference prediction value and the third initial difference prediction value, specifically: ,in: is the final difference prediction value, is the weight corresponding to the first initial difference prediction value, is the first initial difference prediction value, is the weight corresponding to the second initial difference prediction value, is the second initial difference prediction value, is the weight corresponding to the third initial difference prediction value, is the third initial difference prediction value.

3. The method for optimizing ultra-thin ceramic tile ingredients based on real-time data acquisition according to claim 2 is characterized in that: The method for obtaining the weight corresponding to the initial difference prediction value is as follows: N1: Obtain the root mean square error corresponding to each prediction model, and normalize each of the root mean square errors; N2: According to the normalized root mean square error, determine the allocation weight corresponding to each of the initial difference prediction values, and obtain the final weight corresponding to each of the initial difference prediction values ​​through the allocation weight, specifically: ,in: is the weight corresponding to the k-th prediction model, is the distribution weight corresponding to the root mean square error after normalization, is the initial weight corresponding to the m-th prediction model, is the index corresponding to the prediction model.

4. The method for optimizing ultra-thin ceramic tile ingredients based on real-time data acquisition according to claim 2 is characterized in that: Get the impact parameters of environmental parameters on the current batch of raw materials, as follows: W1: Establish an initial neural network model to obtain the initial environmental impact coefficient, specifically: ,in: is the initial environmental impact coefficient, is the weight matrix corresponding to the neural network model, is the activation function, is the environmental parameter, is the bias vector; W2: Optimize the initial neural network model through a sliding window algorithm to obtain an optimized neural network model, specifically: ,in: is the optimized environmental impact coefficient, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters; W3: Based on the optimized neural network model, an environmental monitoring model is established to obtain the impact parameters of environmental parameters on the current batch of raw materials. The environmental monitoring model is specifically: ,in: is the impact parameter of environmental parameters on the current batch of raw materials, For the The weights corresponding to the environmental parameters are For the environmental parameters, is the index of the environment parameter, is the total number of environmental parameters, For the The weights corresponding to the interaction terms between environmental parameters are For the The interaction terms between the environmental parameters, is the total number of interaction terms between environmental parameters, is the index of the interaction term between environmental parameters.

5. The method for optimizing ultra-thin ceramic tile ingredients based on real-time data acquisition according to claim 1 is characterized in that: The simulation results corresponding to the adjustment parameters are compared with the simulation results corresponding to the standard formula, as follows: S2.1: taking the adjustment parameters and the standard formula as inputs of a simulation model, and outputting simulation results corresponding to the adjustment parameters and the standard formula; S2.2: Compare the simulation result corresponding to the adjustment parameter with the simulation result corresponding to the standard recipe. If the comparison results are the same, execute step S3; otherwise, execute step S2.3; S2.3: Determine the influencing factors from the environmental parameters and the weighting coefficients, repeat steps S1.4 to S2.2, and re-acquire the adjustment parameters based on the influencing factors.

6. The method for optimizing ultra-thin ceramic tile ingredients based on real-time data acquisition according to claim 5 is characterized in that: The influencing factors are determined from the environmental parameters and weighting coefficients as follows: E1: reacquire the current environmental parameters and compare the current environmental parameters with the test environmental parameters. When the current environmental parameters are the same as the test environmental parameters, execute step E2. Otherwise, reacquire the parameters affecting the current batch of raw materials according to the current environmental parameters. E2: adjusting the size of the weighting coefficient, and reacquiring the adjustment parameter according to the adjusted weighting coefficient and the adjustment model.

7. An ultra-thin ceramic tile batching optimization system based on real-time data collection, characterized in that: The ultra-thin ceramic tile batching optimization system uses an ultra-thin ceramic tile batching optimization method based on real-time data acquisition as described in any one of claims 1-6.

Images