A PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration

Abstract

The invention relates to the technical field of PVB production, and specifically discloses a PVB production process energy efficiency optimization system based on an Apriori algorithm and edge cloud collaboration, which is used to solve the problem of insufficient control accuracy of process parameters of a heating kettle and a condensation kettle. The system comprises a perception layer and an edge computing layer, the perception layer and the edge computing layer are connected, the edge computing layer is connected to a data middle platform layer, the data middle platform layer is connected to an application display layer and a rule execution and data processing device, the edge computing layer mines preliminary frequent item sets based on an improved dynamic hash-Apriori algorithm, generates preliminary association rules, which are used for processing and analyzing local data of a condensation kettle and a heating kettle in a short time, and transmits the processed data to a cloud analysis and storage layer, the data middle platform layer executes an asymptotic identification method and a multi-objective heuristic evolutionary algorithm; the invention realizes intelligent optimization of the process parameters of a heating kettle and a condensation kettle in the PVB production process through a dynamic hash-Apriori algorithm.

Description

Technical Field

[0001] The present invention relates to the field of PVB production technology, and more specifically, to a PVB production process energy efficiency optimization system based on an Apriori algorithm and edge-cloud collaboration. Background Art

[0002] Polyvinyl butyral (PVB) is an important polymer material widely used in safety glass, laminated glass, and other fields. During the PVB production process, precise control of key process parameters (such as temperature, heating rate, and time) has a decisive impact on product quality and energy efficiency. Due to the limited technological maturity of PVB production processes in my country, there is still considerable room for improvement, especially in the accuracy, timeliness, and consistency of temperature control. Traditional temperature control systems often struggle to adapt to the complex changes in the production process, resulting in high energy consumption and unstable product quality. During the PVB production process, temperature control typically relies on fixed rule control or simple PID control. These methods cannot adjust parameters in a timely manner when dealing with dynamic production environments, which can easily lead to process deviations. At the same time, existing systems lack in-depth data mining and correlation analysis, and are unable to effectively identify the potential relationships between process parameters, resulting in the system's energy efficiency optimization potential not being fully utilized. Specifically, in the condensation kettle and heating kettle of PVB production, the temperature control accuracy is not enough, and overheating or insufficient heating is prone to occur, resulting in a decline in product quality. The existing control system is unable to analyze and adjust energy consumption parameters in real time, failing to effectively reduce energy consumption. The relationship between process parameters (such as temperature, time, heating rate, etc.) is complex, and there is a lack of analysis and optimization methods based on big data, making it difficult to find the best control strategy. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, the present invention provides a PVB production process energy efficiency optimization system based on the Apriori algorithm and edge cloud collaboration. Through the dynamic hash-Apriori algorithm, the intelligent optimization of the process parameters of the heating kettle and condensation kettle in the PVB production process is realized, thereby reducing energy consumption and improving the product quality and production efficiency of the heating kettle and condensation kettle.

[0004] In PVB production, the dynamic production environment proposed in this solution specifically refers to the dynamic changes in the status of PVB production equipment, fluctuations in external environmental conditions, dynamic adjustments to production batches and production loads, dynamic fluctuations in energy consumption and energy supply, nonlinear changes in chemical reaction conditions, complex correlations between process parameters, and fluctuations in real-time data feedback.

[0005] Limited by process and operational limitations, my country's current PVB (polyvinyl butyral) production technology remains immature, leaving significant room for technological advancement and product quality improvement, particularly in the timeliness, accuracy, and consistency of temperature control in condensation and heating kettles. The Apriori algorithm is applied to mine frequent item sets between data points in the sample set and generate association rules for these frequent item sets, calculating the support and confidence of each association rule. Based on preset minimum support and minimum confidence thresholds, a specified algorithm is used to cluster and optimize the association rules of these frequent item sets. Based on production process monitoring, the rules serve as a guide to adjust process parameters to ensure controllable quality throughout the entire production process. An asymptotic identification method is used to estimate the parameters of the identification model, and a multi-objective heuristic evolutionary algorithm is designed to further optimize the model parameters. Build a platform architecture of perception layer, edge computing layer, data middle platform layer, and application display layer; develop an edge computing collection gateway to support multiple interfaces and communication protocols, and develop a high-performance analysis and application data middle platform layer to support the association and fusion application of multi-structured data; based on the platform system architecture of edge-cloud collaboration, analyze the functions implemented by the system, use Python to build the system model algorithm, provide performance evaluation analysis of the control loop in multiple dimensions such as days and months, and use data mining technology to analyze the changes in PVB operation control quality and the key factors that cause the changes.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The PVB production process energy efficiency optimization system based on the Apriori algorithm and edge-cloud collaboration includes a perception layer and an edge computing layer. The perception layer and the edge computing layer are connected. The edge computing layer is connected to a data middle platform layer. The data middle platform layer is connected to an application display layer and a rule execution and data processing device. The rule execution and data processing device is connected to a cloud analysis and storage layer. The edge computing layer preprocesses the real-time data of the condensation kettle and the heating kettle in the PVB production process transmitted by the perception layer, mines preliminary frequent item sets based on the improved dynamic hash-Apriori algorithm, generates preliminary association rules for processing and analyzing local data in a short time, and transmits the processed data to the cloud analysis and storage layer. The data middle platform layer summarizes the data transmitted by the edge computing layer and historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, executes the asymptotic identification method and the multi-objective heuristic evolutionary algorithm, finds the optimal process parameter configuration that can reduce energy consumption while ensuring PVB production quality, and transmits the generated optimization results to the application display layer. The multi-objective heuristic evolutionary algorithm aims to minimize temperature deviation and energy consumption. Its objective function formula is:

[0008] ;

[0009] Where: is the objective function value in the multi-objective heuristic evolutionary algorithm, is the set of process parameters to be optimized, 、 are the weight factors of temperature deviation and energy consumption, which are set according to production requirements. is the time point index during the data collection process, is the number of data points collected, determined according to the production cycle, For the The temperature deviation collected at each time point, For the Energy consumption collected at each time point;

[0010] The application display layer displays the optimized results to the production operators and adjusts the process parameters of the condensation kettle and the heating kettle based on the optimization results.

[0011] At the data middle platform level, the specific implementation process of the asymptotic identification method is as follows:

[0012] Step 1: System modeling and parameter definition: Based on the key process parameters of the condensation kettle and the heating kettle, the mathematical models of the condensation kettle and the heating kettle are established using a linear state space model;

[0013] Step 2: Parameter definition and collection: Collect input data from the PVB production process, establish a historical database of process parameters for the heating reactor and condensation reactor, and define parameters to be identified by the asymptotic identification method;

[0014] Step 3: Parameter estimation using the asymptotic identification method: Based on the core formula of the asymptotic identification method, an adaptive algorithm with asymptotic convergence characteristics is used to gradually approximate the true parameters of the condensation reactor and the heating reactor from the system input and output data. The parameters to be identified using the asymptotic identification method are iteratively updated until the error between the model's predicted output and the actual output is less than a set threshold, or the maximum number of iterations is reached. Preliminary parameter estimates are obtained and used as the initial values ​​for the next optimization step.

[0015] Step 4: Multi-objective inspired evolutionary algorithm parameter optimization: Based on the preset multi-objective optimization function, energy consumption minimization and temperature deviation minimization optimization are performed, a population is initialized, each individual represents a parameter set, and the initial estimate value obtained based on the asymptotic identification method is generated. The objective function value corresponding to each individual is calculated, and its fitness function value is obtained. Based on the fitness function value, individuals with fitness exceeding the preset threshold are selected to enter the next generation, and a new parameter set is generated through a crossover operation. Some individual parameters are randomly mutated, and the fitness calculation, selection, crossover and mutation operations are repeated to update the parameter set in the population. The process is iterative until the preset convergence condition is reached.

[0016] As a further solution of the present invention, the perception layer collects the set temperature, actual temperature, temperature deviation, heating rate, heating time, reaction time, energy consumption, and power of the heating kettle during the PVB production process, and collects the set temperature, actual temperature, temperature deviation, reaction time, pressure, energy consumption, and reaction rate of the condensation kettle during the PVB production process, and transmits the collected raw data to the edge computing layer.

[0017] As a further solution of the present invention, the specific operation process of the edge computing layer generating preliminary association rules includes:

[0018] Step 11, data preprocessing: remove noise and outliers in the original data, and use linear interpolation to interpolate missing values, converting the continuous original data into discrete data;

[0019] Step 12, mining preliminary frequent itemsets: define the minimum support threshold and the minimum confidence threshold, convert the preprocessed discrete data into itemsets, use the initial hash function to hash each first-order itemset, generate a hash value and perform frequency counting, use a high-order hash function to hash and count each second-order itemset and third-order itemset, prune the candidate itemsets, remove the itemsets that do not meet the support threshold, and filter out the itemsets that meet the minimum support according to the counting results after hash mapping as preliminary frequent itemsets;

[0020] Step 13, generating preliminary association rules: generating preliminary association rules based on the preliminary frequent item sets;

[0021] Step 14, data storage: transfer the preliminary frequent item sets and preliminary association rules to the cloud analysis and storage layer for data storage.

[0022] As a further solution of the present invention, in the edge computing layer, the initial hash function involved in step 12 is mapped to a specific position in the hash table by linearly combining the values ​​of the first-order item set and then performing a modulo operation. The formula of the initial hash function is:

[0023] ;

[0024] Where: is the representative value of the discretized temperature interval, is the representative value of the discretized heating rate, is the hash value of the first-order itemset, 、 is the prime number weight coefficient, set according to actual needs, is a prime number larger than the initial hash table size, It is a modular operation.

[0025] As a further solution of the present invention, in the edge computing layer, the high-order hash function involved in step 12 includes a heating kettle high-order hash function and a condensation kettle high-order hash function. The heating kettle high-order hash function maps the second-order item set and the third-order item set to a specific position of the hash table through the first heating kettle hash mapping function and the second heating kettle hash mapping function. The formula of the heating kettle high-order hash function is:

[0026] ;

[0027] Where: 、 、 are the discretized temperature, discretized heating rate and discretized heating time of the heating kettle, The hash function mapping value for the heating kettle, 、 They are the mapping values ​​of the hash mapping function for the first heating kettle and the mapping values ​​of the hash mapping function for the second heating kettle respectively.

[0028] As a further solution of the present invention, in the edge computing layer, the first heating kettle hash mapping function in the high-order hash function of the heating kettle processes the discretized temperature, discretized heating rate, and discretized heating time of the heating kettle through a sine function, a natural logarithm function, and an exponential function with base e, and then performs a modulo operation on the sum of the values. The second heating kettle hash mapping function processes the temperature, heating rate, and heating time of the heating kettle through a cosine function, an arithmetic square root function, and a logarithm function with base 2, and then performs a modulo operation on the sum of the values. The formulas of the first heating kettle hash mapping function and the second heating kettle hash mapping function are as follows:

[0029] ;

[0030] ;

[0031] Where: is the modulo operation of the first heating kettle function, This is the modulo operation of the second heating kettle.

[0032] As a further solution of the present invention, in the edge computing layer, the condensation kettle high-order hash function maps the second-order item set and the third-order item set to specific positions in the hash table through the first condensation kettle hash mapping function and the second condensation kettle hash mapping function. The formula of the condensation kettle high-order hash function is:

[0033] ;

[0034] Where: 、 、 、 are the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle, is the mapping value of the high-order hash function of the condensation kettle, 、 They are the hash function mapping value of the first condensation kettle and the hash function mapping value of the second condensation kettle respectively.

[0035] As a further solution of the present invention, in the edge computing layer, the first condensation kettle hash mapping function of the condensation kettle high-order hash function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the inverse tangent function, logarithmic function, arithmetic square root function, and logarithmic function with base 2, and then performs a modulo operation on the sum of the values. The second condensation kettle hash mapping function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the exponential function with base e, the fractional function, the sine function, and the natural logarithm function. The formulas of the first condensation kettle hash mapping function and the second condensation kettle hash mapping function are as follows:

[0036] ;

[0037] ;

[0038] Where: is the modulo operation of the first condensation kettle function, This is the modulo operation of the second condensation kettle.

[0039] As a further solution of the present invention, the specific implementation process of step 13 in the implementation process of generating preliminary association rules at the edge computing layer includes:

[0040] Step 131, input preliminary frequent item sets: use the preliminary frequent item sets from step 12 as the basic input for association rule generation;

[0041] Step 132, generating the association rule objective: deriving the association relationship objective between the process parameters of the heating kettle and the condensation kettle from the preliminary frequent item set;

[0042] Step 133, calculating support and confidence: inputting the process parameter combination of the condensation kettle and the heating kettle and the associated parameters derived therefrom into the support and confidence formula to obtain support and confidence;

[0043] Step 134, generating candidate association rules from the preliminary frequent item sets: for each preliminary frequent item set, generating all candidate association rule combinations;

[0044] Step 135, pruning: pruning the candidate association rules generated in step 134, removing the rules that do not meet the minimum confidence threshold;

[0045] Step 136, rule evaluation and sorting: sort the association rules pruned in step 135 according to support and confidence. Based on the sorting results, give priority to rules whose support and confidence are higher than preset values.

[0046] As a further solution of the present invention, the data middle platform layer aggregates the data transmitted by the edge computing layer and the historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, and executes the asymptotic identification method and the multi-objective heuristic evolutionary algorithm. The specific implementation process includes:

[0047] Step 21, data aggregation and integration: The data middle platform aggregates the real-time and historical production data of the heating kettle and condensation kettle transmitted by the edge computing layer, matches and integrates the real-time data and historical data to ensure time series consistency, and handles data redundancy and missing values;

[0048] Step 22, mining global frequent itemsets: Based on the entire PVB heating kettle and condensation kettle production cycle and historical data, define the global minimum support threshold and confidence threshold. Use the improved dynamic hash-Apriori algorithm to perform hash mapping on the first-order itemsets, generate hash values ​​and perform frequency counting. Perform multiple hash mappings on the second-order or third-order itemsets and perform pruning operations to remove itemsets that do not meet the global support threshold. After hash counting and pruning, generate frequent itemsets that meet the global support threshold, which serve as the basis for subsequent global association rule generation.

[0049] Step 23, generating global association rules: Based on the global frequent itemsets in step 22, generate global candidate association rules, use global data to calculate the support and confidence of each candidate association rule, and compare them with the global support threshold and confidence threshold, remove association rules that do not meet the global minimum confidence threshold, and sort the remaining rules by support and confidence;

[0050] Step 24: Execute a multi-objective heuristic evolutionary algorithm: Based on the global association rules and using a preset objective function formula, the multi-objective heuristic evolutionary algorithm is used to optimize the process parameters of the condensation kettle and the heating kettle. The multi-objective optimization function is solved using a gradient descent method to obtain the optimal process parameter configuration of the condensation kettle and the heating kettle that can minimize temperature deviation and energy consumption while ensuring PVB production quality.

[0051] Step 25, output and adaptive adjustment: The optimized process parameter configuration results of the condensation kettle and the heating kettle are output to the application display layer and fed back to the rule execution and data processing device. The rule execution and data processing device automatically adjusts the process parameters of the condensation kettle and the heating kettle according to the optimized process parameter configuration and the output process parameter configuration obtained by the asymptotic identification method.

[0052] As a further solution of the present invention, in step 25, the process parameter configuration obtained by the multi-objective heuristic evolutionary algorithm and the output process parameter configuration obtained by the asymptotic identification method each account for 50% of the weight.

[0053] Compared with the existing technology, the technical effect of the present invention is as follows: the present invention significantly improves the accuracy, timeliness and consistency of temperature control in the PVB production process through the collaborative work of data perception, edge computing and cloud storage. Unlike traditional fixed rules or PID control systems, the system uses dynamic hashing-Apriori algorithm to mine the potential correlation between process parameters, generate frequent item sets and association rules, and then optimizes the process parameters of the heating kettle and condensation kettle through asymptotic identification method and multi-objective heuristic evolutionary algorithm, and finds the optimal configuration scheme that minimizes energy consumption while ensuring product quality. Compared with the existing technology, the system can analyze the relationship between complex variables such as temperature, time, and heating rate in real time, dynamically adjust process parameters, avoid overheating or insufficient heating, and solve the problems of slow response, inaccurate adjustment, and low energy efficiency of existing temperature control systems, thereby improving production efficiency and product quality stability. Through multi-dimensional data mining and intelligent algorithms, the system greatly reduces energy consumption and realizes automation and intelligent optimization of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0055] The following is a clear and complete description of the technical solutions of the present invention, in conjunction with the accompanying drawings. Obviously, the technical solutions described are only part of the present invention, not all of it. Based on the contents of the present invention, all other technical solutions obtained by persons of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0056] like Figure 1As shown, the PVB production process energy efficiency optimization system based on the Apriori algorithm and edge-cloud collaboration proposed by the present invention includes a perception layer and an edge computing layer, the perception layer and the edge computing layer are connected, the edge computing layer is connected to a data middle platform layer, the data middle platform layer is connected to an application display layer and a rule execution and data processing device, the rule execution and data processing device is connected to a cloud analysis and storage layer, the edge computing layer pre-processes the real-time data of the condensation kettle and the heating kettle in the PVB production process transmitted by the perception layer, mines preliminary frequent item sets based on the improved dynamic hash-Apriori algorithm, generates preliminary association rules for processing and analyzing local data in a short time, and transmits the processed data to the cloud analysis and storage layer, the data middle platform layer summarizes the data transmitted by the edge computing layer and the historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, executes the asymptotic identification method and the multi-objective heuristic evolutionary algorithm, finds the optimal process parameter configuration that can reduce energy consumption while ensuring the quality of PVB production, and transmits the generated optimization results to the application display layer, the multi-objective heuristic evolutionary algorithm aims to minimize temperature deviation and energy consumption, and its objective function formula is:

[0057] ;

[0058] Where: is the objective function value in the multi-objective heuristic evolutionary algorithm, is the set of process parameters to be optimized, 、 are the weight factors of temperature deviation and energy consumption, which are set according to production requirements. is the time point index during the data collection process, is the number of data points collected, determined according to the production cycle, For the The temperature deviation collected at each time point, For the Energy consumption collected at each time point;

[0059] The application display layer displays the optimized results to the production operators and adjusts the process parameters of the condensation kettle and the heating kettle based on the optimization results.

[0060] It should be noted that at the data middle platform level, the specific implementation process of the asymptotic identification method is as follows:

[0061] Step 1: System modeling and parameter definition: Based on the key process parameters of the condensation kettle and the heating kettle, the mathematical models of the condensation kettle and the heating kettle are established using a linear state space model;

[0062] Step 2: Parameter definition and collection: Collect input data from the PVB production process, establish a historical database of process parameters for the heating reactor and condensation reactor, and define parameters to be identified by the asymptotic identification method;

[0063] Step 3: Parameter estimation using the asymptotic identification method: Based on the core formula of the asymptotic identification method, an adaptive algorithm with asymptotic convergence characteristics is used to gradually approximate the true parameters of the condensation reactor and the heating reactor from the system input and output data. The parameters to be identified using the asymptotic identification method are iteratively updated until the error between the model's predicted output and the actual output is less than a set threshold, or the maximum number of iterations is reached. Preliminary parameter estimates are obtained and used as the initial values ​​for the next optimization step.

[0064] Step 4: Multi-objective inspired evolutionary algorithm parameter optimization: Based on the preset multi-objective optimization function, energy consumption minimization and temperature deviation minimization optimization are performed, a population is initialized, each individual represents a parameter set, and the initial estimate value obtained based on the asymptotic identification method is generated. The objective function value corresponding to each individual is calculated, and its fitness function value is obtained. Based on the fitness function value, individuals with fitness exceeding the preset threshold are selected to enter the next generation, and a new parameter set is generated through a crossover operation. Some individual parameters are randomly mutated, and the fitness calculation, selection, crossover and mutation operations are repeated to update the parameter set in the population. The process is iterative until the preset convergence condition is reached.

[0065] By combining asymptotic identification with a multi-objective evolutionary algorithm, the system accurately identifies key process parameters for the condensation and heating reactors, and gradually optimizes temperature control and energy consumption. The asymptotic identification method uses an adaptive algorithm to gradually approximate the true system parameters, ensuring accurate parameter estimation. The multi-objective evolutionary algorithm further optimizes these parameters, achieving the dual goals of minimizing energy consumption and temperature deviation. Through iterative calculation, selection, crossover, and mutation operations, the system dynamically updates the parameter set, ensuring the continuity and adaptability of parameter optimization in complex production environments. Ultimately, this improves the efficiency and product quality of the PVB production process while reducing energy consumption.

[0066] It should be noted that the perception layer collects the set temperature, actual temperature, temperature deviation, heating rate, heating time, reaction time, energy consumption, and power of the heating kettle during the PVB production process, and collects the set temperature, actual temperature, temperature deviation, reaction time, pressure, energy consumption, and reaction rate of the condensation kettle during the PVB production process, and transmits the collected raw data to the edge computing layer.

[0067] It should be noted that the specific operation process of the edge computing layer to generate preliminary association rules includes:

[0068] Step 11, data preprocessing: remove noise and outliers in the original data, and use linear interpolation to interpolate missing values, converting the continuous original data into discrete data;

[0069] Step 12, mining preliminary frequent itemsets: define the minimum support threshold and the minimum confidence threshold, convert the preprocessed discrete data into itemsets, use the initial hash function to hash each first-order itemset, generate a hash value and perform frequency counting, use a high-order hash function to hash and count each second-order itemset and third-order itemset, prune the candidate itemsets, remove the itemsets that do not meet the support threshold, and filter out the itemsets that meet the minimum support according to the counting results after hash mapping as preliminary frequent itemsets;

[0070] Step 13, generating preliminary association rules: generating preliminary association rules based on the preliminary frequent item sets;

[0071] Step 14, data storage: transfer the preliminary frequent item sets and preliminary association rules to the cloud analysis and storage layer for data storage.

[0072] By generating preliminary association rules at the edge computing layer, the system can effectively solve the problem of real-time data processing and analysis in the PVB production process. Data preprocessing ensures the integrity and accuracy of input data by removing noise and filling missing values. Frequent item set mining uses an improved hash algorithm to improve the efficiency of item set mapping and calculation, especially in large-scale data sets. Item sets that do not meet the support requirements are removed through pruning to reduce computational redundancy. Based on these frequent item sets, preliminary association rules are generated, which helps to reveal the potential relationship between the process parameters of the condensation kettle and the heating kettle, providing a basis for subsequent optimization. After the preliminary rules are stored in the cloud, further global analysis and parameter adjustment are facilitated. The entire process improves the level of intelligent process control in PVB production, solves the defect that traditional systems cannot be dynamically adjusted and optimized, and significantly improves production efficiency and product quality.

[0073] It should be noted that in the edge computing layer, the initial hash function involved in step 12 maps the values ​​of the first-order item set to a specific position in the hash table by performing a linear combination and then performing a modulo operation. The formula of the initial hash function is:

[0074] ;

[0075] Where: is the representative value of the discretized temperature interval, is the representative value of the discretized heating rate, is the hash value of the first-order itemset, 、 is the prime number weight coefficient, set according to actual needs, is a prime number larger than the initial hash table size, It is a modular operation.

[0076] By performing linear combinations and modular operations on first-order itemsets using an initial hash function, the system can efficiently map discretized temperature and heating rate data into a hash table, enabling rapid frequency counting and data processing. The introduction of prime number weight coefficients ensures that the effects of temperature and heating rate on the hash value are relatively independent, reducing hash conflicts and ensuring that different itemsets are discretely mapped to different locations in the hash table. This approach effectively improves the efficiency of large-scale data processing, lays a stable data foundation for subsequent frequent itemset mining and association rule generation, and resolves the frequent mapping conflicts encountered in traditional methods.

[0077] Example 1

[0078] Regarding the application of the initial hash function, the following embodiment parameter calculation is performed to illustrate in detail the technical effect of the initial hash function in solving the technical problem.

[0079] is the representative value of the discretized temperature interval, is the representative value of the heating rate after discretization. Both are integer intervals. In the early stage of heating of the heating kettle, the heating rate needs to be gradually increased, but it needs to be controlled to avoid a sudden temperature rise that causes a too fast reaction or instability. 、 As the weight of the heating kettle temperature and heating rate, at the end of the condensation kettle reaction, the temperature of the condensation kettle gradually decreases, the pressure change is small, and the smooth transition of temperature is more important. It is necessary to prevent the sudden drop in temperature from affecting the product quality. 、 As the weight of the condensation kettle temperature and heating rate, set the prime number of the modular operation , which is larger than the size of the hash table to ensure that the hash values ​​are evenly distributed, according to the formula of the initial hash function:

[0080] ;

[0081] for The remainder when divided by 31 is equal to Therefore, the first-order item set The hash value of is 12 and is mapped to the 12th position in the hash table.

[0082] It should be noted that in the edge computing layer, the high-order hash functions involved in step 12 include the heating kettle high-order hash function and the condensation kettle high-order hash function. The heating kettle high-order hash function maps the second-order item set and the third-order item set to a specific position in the hash table through the first heating kettle hash mapping function and the second heating kettle hash mapping function. The formula of the heating kettle high-order hash function is:

[0083] ;

[0084] Where: 、 、 are the discretized temperature, discretized heating rate and discretized heating time of the heating kettle, The hash function mapping value for the heating kettle, 、 They are the mapping values ​​of the hash mapping function for the first heating kettle and the mapping values ​​of the hash mapping function for the second heating kettle respectively.

[0085] It should be noted that in the edge computing layer, the first heating kettle hash mapping function in the heating kettle high-order hash function processes the discretized temperature, discretized heating rate, and discretized heating time of the heating kettle through the sine function, natural logarithm function, and exponential function with base e, and then performs a modulo operation on the sum of the values. The second heating kettle hash mapping function processes the temperature, heating rate, and heating time of the heating kettle through the cosine function, arithmetic square root function, and logarithm function with base 2, and then performs a modulo operation on the sum of the values. The formulas of the first heating kettle hash mapping function and the second heating kettle hash mapping function are as follows:

[0086] ;

[0087] ;

[0088] Where: is the modulo operation of the first heating kettle function, This is the modulo operation of the second heating kettle.

[0089] It should be noted that in the edge computing layer, the condensation kettle high-order hash function maps the second-order item set and the third-order item set to specific positions in the hash table through the first condensation kettle hash mapping function and the second condensation kettle hash mapping function. The formula of the condensation kettle high-order hash function is:

[0090] ;

[0091] Where: 、 、 、 are the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle, is the mapping value of the high-order hash function of the condensation kettle, 、 They are the hash function mapping value of the first condensation kettle and the hash function mapping value of the second condensation kettle respectively.

[0092] It should be noted that in the edge computing layer, the first condensation kettle hash mapping function of the condensation kettle high-order hash function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the inverse tangent function, logarithmic function, arithmetic square root function, and logarithmic function with base 2, and then performs a modulo operation on the sum of the values. The second condensation kettle hash mapping function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the exponential function with base e, the fractional function, the sine function, and the natural logarithm function. The formulas of the first condensation kettle hash mapping function and the second condensation kettle hash mapping function are as follows:

[0093] ;

[0094] ;

[0095] Where: is the modulo operation of the first condensation kettle function, This is the modulo operation of the second condensation kettle.

[0096] In the heating kettle, high-order hash functions are used to map temperature, heating rate, and heating time, effectively solving the technical challenges of complex data processing and frequent itemset mining. Nonlinear processing of these parameters using sine, natural logarithm, and exponential functions ensures that differences between parameters are fully amplified, reducing conflicts that arise during the hash mapping process. Furthermore, the use of modular arithmetic ensures a uniform distribution of generated hash values, preventing data from being concentrated in specific locations and improving the storage and retrieval efficiency of the hash table. For second- and third-order itemsets, the combination of two different hash mappings effectively reduces conflicts and duplicate calculations that may arise from different parameter combinations in high-order itemsets, thereby improving the efficiency of frequent itemset mining. This technology enables more precise control of the temperature rise rate of the heating kettle during production, ensuring a stable heating process and avoiding process deviations caused by excessively fast or slow heating. Ultimately, this improves the reliability of the PVB production process and product quality.

[0097] In the condensation reactor, high-order hash functions are applied to map temperature, pressure, reaction time, and energy consumption. Different process parameters are processed using inverse tangent functions, logarithmic functions, square root functions, and exponential functions with base e. This nonlinear mapping approach rationally reflects the correlations between process parameters while effectively avoiding conflicts and duplications in the hash mapping. The use of two hash mappings ensures that different combinations of second-order and third-order item sets are evenly distributed in the hash table, reducing data redundancy and improving the accuracy and speed of item set processing. This allows the condensation reactor to more precisely control key parameters such as reaction time, temperature, and pressure during the production process, ensuring stability during the reaction, reducing energy consumption and unnecessary resource waste, optimizing production energy efficiency, and guaranteeing final product quality. This technology can effectively improve the operating efficiency of the condensation reactor under complex process conditions and reduce technical deviations and failure rates during production.

[0098] In order to illustrate the specific calculation process of the high-order hash function of the heating kettle and the high-order hash function of the condensation kettle in detail, they are respectively described through Example 2 and Example 3.

[0099] Example 2

[0100] In the heating kettle, the three process parameters of temperature, heating rate and heating time need to be hash mapped to generate corresponding high-order hash values ​​to facilitate the rapid screening of frequent item sets and the mining of association rules. The data input is set to 45, 3 and 8 for the discretized temperature, heating rate and heating time respectively. , the hash mapping function of the first heating kettle is , The value of is rounded to 11;

[0101] The hash mapping function of the second heating kettle is ;

[0102] So the high-order hash function of the heating kettle is , the rounded result is 16.

[0103] Example 3

[0104] In the condensation reactor, the four process parameters of temperature, pressure, reaction time and energy consumption need to be hash mapped to generate high-order hash values. The discrete input data of temperature, pressure, reaction time and energy consumption are 60, 5, 10 and 100 respectively. When the first condensation kettle hash mapping function ;

[0105] Second condensation kettle hash mapping function ;

[0106] So the condensation kettle high-order hash function is , the rounded result is 17.

[0107] The above technical effects specifically demonstrate the application of high-order hash functions of the heating kettle and the condensation kettle through Examples 2 and 3, and illustrate the significant advantages of the scheme in data processing, hash conflict reduction, frequent item set mining, process control accuracy and large-scale data processing. In Example 2, the system performs hash mapping on the three process parameters of the heating kettle, namely, temperature, heating rate and heating time, and uses sine function, natural logarithm function and exponential function for nonlinear processing, and generates hash values ​​through two different hash mapping functions. When the data input discretization temperature is 45, the heating rate is 3, and the heating time is 8, the hash result is 16. Through this hash mapping, it is ensured that the multidimensional parameter data can be evenly distributed in the hash table, effectively reducing data conflicts, improving the speed of frequent item set mining, and ensuring the process control accuracy of the heating kettle. In Example 3, the four process parameters of the condensation kettle, namely, temperature, pressure, reaction time and energy consumption, are nonlinearly processed by inverse tangent function, logarithm function, square root function, etc., and the hash value is finally calculated to be 17 after the first and second mappings. The data inputs are respectively discretized temperature 60, pressure 5, reaction time 10, and energy consumption 100. This processing method ensures the independence of the condensation kettle process parameters, avoids hash conflicts, makes the data evenly distributed in the hash table, and effectively improves the efficiency of the system under complex process conditions. Through the calculation process of Example 2 and Example 3, the advantages of this scheme in large-scale data processing, real-time process parameter control and frequent item set mining are verified. In particular, through two hash mappings, the system can accurately and quickly process the complex process parameters in the heating kettle and the condensation kettle, greatly improving the accuracy and response speed of process control, further optimizing production energy efficiency, and ensuring the stability of product quality and the improvement of production efficiency.

[0108] It should be noted that the specific implementation process of step 13 in the implementation process of generating preliminary association rules at the edge computing layer includes:

[0109] Step 131, input preliminary frequent item sets: use the preliminary frequent item sets from step 12 as the basic input for association rule generation;

[0110] Step 132, generating the association rule objective: deriving the association relationship objective between the process parameters of the heating kettle and the condensation kettle from the preliminary frequent item set;

[0111] Step 133, calculating support and confidence: inputting the process parameter combination of the condensation kettle and the heating kettle and the associated parameters derived therefrom into the support and confidence formula to obtain support and confidence;

[0112] Step 134, generating candidate association rules from the preliminary frequent item sets: for each preliminary frequent item set, generating all candidate association rule combinations;

[0113] Step 135, pruning: pruning the candidate association rules generated in step 134, removing the rules that do not meet the minimum confidence threshold;

[0114] Step 136, rule evaluation and sorting: sort the association rules pruned in step 135 according to support and confidence. Based on the sorting results, give priority to rules whose support and confidence are higher than preset values.

[0115] It should be noted that the data middle platform layer aggregates the data transmitted by the edge computing layer and historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, and executes the asymptotic identification method and multi-objective heuristic evolutionary algorithm. The specific implementation process includes:

[0116] Step 21, data aggregation and integration: The data middle platform aggregates the real-time and historical production data of the heating kettle and condensation kettle transmitted by the edge computing layer, matches and integrates the real-time data and historical data to ensure time series consistency, and handles data redundancy and missing values;

[0117] Step 22, mining global frequent itemsets: Based on the entire PVB heating kettle and condensation kettle production cycle and historical data, define the global minimum support threshold and confidence threshold. Use the improved dynamic hash-Apriori algorithm to perform hash mapping on the first-order itemsets, generate hash values ​​and perform frequency counting. Perform multiple hash mappings on the second-order or third-order itemsets and perform pruning operations to remove itemsets that do not meet the global support threshold. After hash counting and pruning, generate frequent itemsets that meet the global support threshold, which serve as the basis for subsequent global association rule generation.

[0118] Step 23, generating global association rules: Based on the global frequent itemsets in step 22, generate global candidate association rules, use global data to calculate the support and confidence of each candidate association rule, and compare them with the global support threshold and confidence threshold, remove association rules that do not meet the global minimum confidence threshold, and sort the remaining rules by support and confidence;

[0119] Step 24: Execute a multi-objective heuristic evolutionary algorithm: Based on the global association rules and using a preset objective function formula, the multi-objective heuristic evolutionary algorithm is used to optimize the process parameters of the condensation kettle and the heating kettle. The multi-objective optimization function is solved using a gradient descent method to obtain the optimal process parameter configuration of the condensation kettle and the heating kettle that can minimize temperature deviation and energy consumption while ensuring PVB production quality.

[0120] Step 25, output and adaptive adjustment: The optimized process parameter configuration results of the condensation kettle and the heating kettle are output to the application display layer and fed back to the rule execution and data processing device. The rule execution and data processing device automatically adjusts the process parameters of the condensation kettle and the heating kettle according to the optimized process parameter configuration and the output process parameter configuration obtained by the asymptotic identification method.

[0121] It should be noted that in step 25, the process parameter configuration obtained by the multi-objective heuristic evolutionary algorithm and the output process parameter configuration obtained by the asymptotic identification method each account for 50% of the weight.

[0122] In step 25, a multi-objective heuristic evolutionary algorithm and an asymptotic identification method, each weighted 50%, are used to obtain the final process parameter configuration. This aims to combine the strengths of both algorithms to achieve more precise process control for the condensation and heating reactors. The asymptotic identification method provides relatively stable preliminary parameter estimates by identifying historical data and system dynamic behavior, while the multi-objective heuristic evolutionary algorithm optimizes and searches based on production objectives (such as minimizing energy consumption and controlling temperature) to provide a local optimal solution. By combining the two methods, each weighted 50%, the system maintains the stability of the historical process while dynamically responding to real-time changes, ensuring the accuracy and flexibility of process parameters within the established goals. This approach improves the system's adaptability and real-time adjustment capabilities, optimizing production efficiency and product quality.

[0123] The present invention significantly improves the accuracy, timeliness and consistency of temperature control in the PVB production process through the collaborative work of data perception, edge computing and cloud storage. Unlike traditional fixed rules or PID control systems, the system uses a dynamic hash-Apriori algorithm to mine the potential correlations between process parameters, generate frequent item sets and association rules, and then optimizes the process parameters of the heating kettle and condensation kettle through asymptotic identification method and multi-objective heuristic evolutionary algorithm, and finds the optimal configuration scheme that minimizes energy consumption while ensuring product quality. Compared with the existing technology, the system can analyze the relationship between complex variables such as temperature, time, and heating rate in real time, dynamically adjust process parameters, avoid overheating or insufficient heating, and solve the problems of slow response, inaccurate adjustment, and low energy efficiency of existing temperature control systems, thereby improving production efficiency and product quality stability. Through multi-dimensional data mining and intelligent algorithms, the system greatly reduces energy consumption and realizes automation and intelligent optimization of the production process.

[0124] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0125] Finally: The above description is only an embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A PVB production process energy efficiency optimization system based on Apriori algorithm and edge cloud collaboration includes a perception layer and an edge computing layer. The perception layer and the edge computing layer are connected, and are characterized in that: The edge computing layer is connected to the data middle platform layer, which is connected to the application display layer and the rule execution and data processing device. The rule execution and data processing device is connected to the cloud analysis and storage layer. The edge computing layer pre-processes the real-time data of the condensation kettle and the heating kettle in the PVB production process transmitted by the perception layer, mines the preliminary frequent item sets based on the improved dynamic hash-Apriori algorithm, generates preliminary association rules for processing and analyzing local data in a short time, and transmits the processed data to the cloud analysis and storage layer. The data middle platform layer summarizes the data transmitted by the edge computing layer and the historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, executes the asymptotic identification method and the multi-objective heuristic evolutionary algorithm, finds the optimal process parameter configuration that can reduce energy consumption while ensuring the quality of PVB production, and transmits the generated optimization results to the application display layer. The multi-objective heuristic evolutionary algorithm aims to minimize temperature deviation and energy consumption, and its objective function formula is: ； Where: is the objective function value in the multi-objective inspired evolutionary algorithm, is the set of process parameters to be optimized, , are the weight factors of temperature deviation and energy consumption, which are set according to production requirements. is the time point index during data collection. is the number of data points collected, determined according to the production cycle, For the The temperature deviation collected at each time point, For the Energy consumption collected at each time point; The application display layer displays the optimized results to the production operators and adjusts the process parameters of the condensation kettle and the heating kettle based on the optimized results.

2. According to claim 1, the PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration is characterized in that: The perception layer collects the set temperature, actual temperature, temperature deviation, heating rate, heating time, reaction time, energy consumption, and power of the heating kettle during the PVB production process, and collects the set temperature, actual temperature, temperature deviation, reaction time, pressure, energy consumption, and reaction rate of the condensation kettle during the PVB production process, and transmits the collected raw data to the edge computing layer.

3. According to claim 2, the PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration is characterized in that: The specific operation process of the edge computing layer to generate preliminary association rules includes: Step 11, data preprocessing: remove noise and outliers in the original data, and use linear interpolation to interpolate missing values ​​to convert continuous original data into discrete data; Step 12, mining preliminary frequent item sets: define the minimum support threshold and the minimum confidence threshold, convert the preprocessed discrete data into item sets, use the initial hash function to hash map each first-order item set, generate hash values ​​and perform frequency counting, use high-order hash functions to hash map and count each second-order item set and third-order item set, prune the candidate item sets, remove the item sets that do not meet the support threshold, and filter out the item sets that meet the minimum support according to the counting results after hash mapping as preliminary frequent item sets; Step 13, generating preliminary association rules: generating preliminary association rules based on preliminary frequent item sets; Step 14, data storage: The preliminary frequent item sets and preliminary association rules are transmitted to the cloud analysis and storage layer for data storage.

4. According to claim 3, the PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration is characterized in that: In the edge computing layer, the initial hash function involved in step 12 maps the values ​​of the first-order item set to a specific position in the hash table by linearly combining them and performing a modulo operation. The formula of the initial hash function is: ； Where: is the representative value of the discretized temperature range, is the representative value of the heating rate after discretization, is the hash value of the first-order itemset, , is the prime number weight coefficient, which is set according to actual needs. is a prime number larger than the initial hash table size, It is a modular operation.

5. According to claim 3, the PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration is characterized in that: In the edge computing layer, the high-order hash functions involved in step 12 include a heating kettle high-order hash function and a condensation kettle high-order hash function. The heating kettle high-order hash function maps the second-order item set and the third-order item set to a specific position of the hash table through the first heating kettle hash mapping function and the second heating kettle hash mapping function. The formula of the heating kettle high-order hash function is: ； Where: , , are the discretized temperature, discretized heating rate and discretized heating time of the heating kettle, The hash function mapping value for the heating kettle, , They are respectively the mapping value of the hash mapping function of the first heating kettle and the mapping value of the hash mapping function of the second heating kettle.

6. According to claim 5, the PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration is characterized in that: In the edge computing layer, the first heating kettle hash mapping function in the high-order hash function of the heating kettle processes the discretized temperature, discretized heating rate and discretized heating time of the heating kettle through the sine function, natural logarithm function and exponential function with base e, and then performs modulo operation on the sum value. The second heating kettle hash mapping function processes the temperature, heating rate and heating time of the heating kettle through the cosine function, arithmetic square root function and logarithm function with base 2, and then performs modulo operation on the sum value. The formulas of the first heating kettle hash mapping function and the second heating kettle hash mapping function are as follows: ； ； Where: is the modulus operation of the first heating kettle function, This is the modulo operation for the second heating kettle.

7. The PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration according to claim 5 is characterized in that: In the edge computing layer, the condensation kettle high-order hash function maps the second-order item set and the third-order item set to a specific position of the hash table through the first condensation kettle hash mapping function and the second condensation kettle hash mapping function. The formula of the condensation kettle high-order hash function is: ； Where: , , , are the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation reactor, respectively. is the mapping value of the high-order hash function of the condensation kettle, , They are the hash function mapping value of the first condensation kettle and the hash function mapping value of the second condensation kettle respectively.

8. The PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration according to claim 7 is characterized in that: In the edge computing layer, the first condensation kettle hash mapping function of the condensation kettle high-order hash function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the inverse tangent function, logarithmic function, arithmetic square root function, and logarithmic function with base 2, and then performs a modulus operation on the sum. The second condensation kettle hash mapping function processes the discretized temperature, discretized pressure, discretized reaction time, and discretized energy consumption of the condensation kettle through the exponential function with base e, the score function, the sine function, and the natural logarithm function. The formulas of the first condensation kettle hash mapping function and the second condensation kettle hash mapping function are as follows: ； ； Where: is the modulo operation of the first condensation kettle function, This is the modulo operation of the second condensation kettle.

9. The PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration according to claim 8 is characterized in that: The specific implementation process of step 13 in the implementation process of generating preliminary association rules at the edge computing layer includes: Step 131, inputting preliminary frequent item sets: using the preliminary frequent item sets from step 12 as basic input for generating association rules; Step 132, generating the target of association rules: deriving the target of association relationship between the process parameters of the heating kettle and the condensation kettle from the preliminary frequent item set; Step 133, calculating support and confidence: inputting the process parameter combination of the condensation kettle and the heating kettle and the associated parameters derived therefrom into the support and confidence formula to obtain support and confidence; Step 134, generating candidate association rules from the preliminary frequent item sets: for each preliminary frequent item set, generating all candidate association rule combinations; Step 135, pruning: pruning the candidate association rules generated in step 134, and removing the rules that do not meet the minimum confidence threshold; Step 136, rule evaluation and sorting: sort the association rules pruned in step 135 according to support and confidence. Based on the sorting results, give priority to rules whose support and confidence are both higher than preset values.

10. The PVB production process energy efficiency optimization system based on Apriori algorithm and edge-cloud collaboration according to claim 9 is characterized in that: The data middle platform layer aggregates the data transmitted by the edge computing layer and the historical production data, uses the improved dynamic hash-Apriori algorithm to generate global association rules, and executes the specific implementation process of the multi-objective heuristic evolutionary algorithm, including: Step 21, data aggregation and integration: The data middle platform layer aggregates the real-time production data and historical production data of the heating kettle and condensation kettle transmitted by the edge computing layer, matches and integrates the real-time data and historical data to ensure time series consistency, and handles data redundancy and missing values; Step 22, mining global frequent item sets: according to the production cycle and historical data of the heating kettle and condensation kettle of the entire PVB, define the global minimum support threshold and confidence threshold, use the improved dynamic hash-Apriori algorithm, perform hash mapping on the first-order item sets, generate hash values ​​and perform frequency counting, perform multiple hash mapping on the second-order or third-order item sets, and perform pruning operations to remove item sets that do not meet the global support threshold. After hash counting and pruning, generate frequent items that meet the global support threshold as the basis for subsequent global association rule generation; Step 23, generating global association rules: according to the global frequent itemsets in step 22, generating global candidate association rules, using global data to calculate the support and confidence of each candidate association rule, and comparing them with the global support threshold and confidence threshold, removing association rules that do not meet the global minimum confidence threshold, and sorting the remaining rules by support and confidence; Step 24, executing a multi-objective inspired evolutionary algorithm: Based on the global association rules, a multi-objective inspired evolutionary algorithm is used to optimize the configuration of the process parameters of the condensation kettle and the heating kettle through a preset objective function formula, and a gradient descent method is used to solve the multi-objective optimization function to obtain the optimal process parameter configuration of the condensation kettle and the heating kettle that can minimize the temperature deviation and energy consumption while ensuring the quality of PVB production; Step 25, output and adaptive adjustment: output the optimized process parameter configuration results of the condensation kettle and the heating kettle to the application display layer, and feed back to the rule execution and data processing device. The rule execution and data processing device automatically adjusts the process parameters of the condensation kettle and the heating kettle according to the optimized process parameter configuration and the output process parameters obtained by the asymptotic identification method.

Images