A collaborative optimization system for energy consumption and efficiency in masterbatch processing based on production data.

Abstract

This invention relates to the field of production management technology, specifically to a collaborative optimization system for energy consumption and efficiency in masterbatch processing based on production data. The system includes: a data acquisition and analysis module that integrates production, energy consumption, and efficiency data to form a structured dataset; an objective function construction module that deeply mines the production, energy consumption, and efficiency features in the data and constructs a set of multi-objective optimization functions accordingly; and a multi-objective collaborative optimization module that employs an improved NSGA-II algorithm, introducing a simulated annealing mechanism to enhance the local search capability of elite solutions, and combining it with crossover and mutation probabilities dynamically adjusted based on multimodal features and targeted search strategies, while simultaneously applying dynamic process, equipment, and safety constraints to obtain globally optimal multi-dimensional scheduling parameters. This invention effectively improves the overall production management efficiency of masterbatch production through the combination of deep data analysis and advanced optimization algorithms.

Description

Technical Field

[0001] This invention relates to the field of production management technology, specifically to a collaborative optimization system for energy consumption and efficiency in color masterbatch processing based on production data. Background Technology

[0002] In existing masterbatch production processes, the setting of processing parameters mainly relies on manual experience. This results in a lack of necessary flexibility and adaptability when the system responds to changes in different batches, formulations, and market demands. Furthermore, traditional management models fail to fully consider the synergistic optimization between processing energy consumption and production efficiency. This management approach often leads to excessive energy consumption or low efficiency, ultimately restricting the overall management efficiency and economic benefits of masterbatch production.

[0003] In traditional systems, the processing parameters for color masterbatch are often set manually, which lacks flexibility; and the synergistic optimization of processing energy consumption and production efficiency is not fully considered, resulting in limited efficiency in color masterbatch production management.

[0004] To address this, a collaborative optimization system for energy consumption and efficiency in masterbatch processing based on production data is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a collaborative optimization system for energy consumption and efficiency in masterbatch processing based on production data. The data acquisition and analysis module integrates production, energy consumption, and efficiency data to form a structured dataset. The objective function construction module deeply mines the production, energy consumption, and efficiency features in the data and constructs a set of multi-objective optimization functions accordingly. The multi-objective collaborative optimization module employs an improved NSGA-II algorithm, introducing a simulated annealing mechanism to enhance the local search capability of elite solutions. It combines this with dynamically adjusted crossover and mutation probabilities based on multimodal features and targeted search strategies, while simultaneously applying dynamic process, equipment, and safety constraints to obtain globally optimal multi-dimensional scheduling parameters. This invention effectively improves the overall production management efficiency of masterbatch production through the combination of deep data analysis and advanced optimization algorithms.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A collaborative optimization system for energy consumption and efficiency in masterbatch processing based on production data includes:

[0008] The data acquisition and analysis module is used to acquire first production data, second energy consumption data, and third efficiency data; based on the analysis of these three types of data, a first structured dataset is obtained.

[0009] The objective function construction module is used to construct a multimodal analysis model to analyze the first structured dataset and obtain the first multimodal features. The first multimodal features include production features, energy consumption features, and efficiency features. Based on these three features, the first set of multi-objective functions is obtained.

[0010] A multi-objective collaborative optimization module is used to construct an improved NSGA-II algorithm to optimize the first multi-objective function set. The improved NSGA-II algorithm introduces the simulated annealing algorithm into the NSGA-II algorithm and applies dynamic constraints to the first multi-objective function set to finally determine the first multi-dimensional scheduling parameters. The first multi-dimensional scheduling parameters include production process scheduling parameters, energy consumption scheduling parameters, and efficiency scheduling parameters.

[0011] The production management module is used to dynamically manage the production process based on the first multi-dimensional scheduling parameters.

[0012] Preferably, the first structured dataset is obtained by weighted fusion of the first production data, the second energy consumption data, and the third efficiency data.

[0013] Preferably, the multimodal analysis model includes a multidimensional data input layer, a multidimensional data preprocessing layer, a multimodal feature extraction layer, and a multimodal feature output layer;

[0014] The multidimensional data input layer is used to input the first structured dataset into the multimodal analysis model;

[0015] The multidimensional data preprocessing layer is used to preprocess the first structured dataset to obtain the second structured dataset;

[0016] The multimodal feature extraction layer includes a production feature extraction sublayer, an energy consumption feature extraction sublayer, and an efficiency feature extraction sublayer; each sublayer performs feature analysis by fusing LSTM and CNN to obtain production features, energy consumption features, and efficiency features, respectively.

[0017] The multimodal feature output layer is used to output production features, energy consumption features, and efficiency features.

[0018] Preferably, the process of obtaining the first multi-objective function set includes: obtaining an energy consumption feature sub-function based on energy consumption characteristics; obtaining an efficiency feature sub-function based on efficiency characteristics; obtaining an energy consumption target sub-function based on the energy consumption feature sub-function and a first influence coefficient; obtaining an efficiency target sub-function based on the efficiency feature sub-function and a second influence coefficient; and thus obtaining the first multi-objective function set.

[0019] Preferably, the improved NSGA-II algorithm dynamically adjusts the crossover probability, mutation probability, selection pressure, and distribution estimation parameters of the NSGA-II algorithm based on the first multimodal features; for elite individuals in the Pareto front solution set output by NSGA-II, a simulated annealing algorithm is introduced to perform local depth search and perturbation.

[0020] Preferably, the improved NSGA-II algorithm constructs a targeted search strategy based on the importance analysis of each first multimodal feature; the strategy analyzes the correlation strength between the first multimodal features and each sub-objective of the first multi-objective function set; and dynamically adjusts the search focus of the improved NSGA-II algorithm.

[0021] Preferably, dynamic constraints include process constraints, equipment constraints, and safety constraints; process constraints are used to constrain the extrusion temperature range, screw speed limit, material ratio boundary, and product quality requirements; equipment constraints are used to constrain the motor power limit, melt pressure range, cooling system capacity, and production line capacity limit; and safety constraints are used to constrain the temperature safety boundary, pressure safety limit, energy consumption peak limit, and environmental emission standards.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] 1. This invention constructs a multimodal analysis model that integrates LSTM and CNN to perform deep feature extraction and fusion analysis on production, energy consumption and efficiency data. Compared with traditional methods that rely on experience or single parameters, it can more accurately and comprehensively identify key influencing factors under complex working conditions and effectively analyze production, energy consumption and efficiency characteristics; thus effectively improving the overall production management efficiency of masterbatch production.

[0024] 2. This invention employs an improved NSGA-II algorithm, innovatively integrating simulated annealing to enhance the local depth search and perturbation of Pareto frontier elite individuals. It also dynamically adjusts algorithm parameters and constructs a targeted search strategy based on multimodal characteristics. This hybrid optimization strategy can more effectively perform global searches in a complex multi-objective space of energy consumption and efficiency, overcoming the shortcomings of traditional optimization methods that easily get trapped in local optima or struggle to balance multiple conflicting objectives. This achieves a deeper level of synergistic optimization between energy consumption and efficiency, effectively improving the overall production management efficiency of masterbatch production.

[0025] 3. This invention not only optimizes the NSGA-II algorithm through improvements, but also enables continuous dynamic management of the production process based on the optimized multi-dimensional scheduling parameters. Combined with dynamic constraints, it forms a closed-loop intelligent system encompassing data acquisition, analysis, optimization decision-making, and execution feedback. This mechanism allows production management to respond in real-time to changes in operating conditions and continuously optimize resource allocation, thereby significantly improving overall operational efficiency, material utilization, and energy efficiency, effectively enhancing the overall production management efficiency of masterbatch production. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of a system for collaborative optimization of energy consumption and efficiency in masterbatch processing based on production data, provided in an embodiment of the present invention.

[0027] Figure 2 This is a schematic diagram of the structure of a multimodal analysis model provided in an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Example 1

[0030] To improve the efficiency of color masterbatch production management, a collaborative optimization system for color masterbatch processing energy consumption and efficiency based on production data was applied, such as... Figure 1 A schematic diagram of a system for collaborative optimization of energy consumption and efficiency in masterbatch processing based on production data, provided in an embodiment of the present invention, specifically includes:

[0031] The data acquisition and analysis module is used to acquire first production data, second energy consumption data, and third efficiency data; based on the analysis of these three types of data, a first structured dataset is obtained.

[0032] Furthermore, the first production data includes extruder screw speed, barrel temperature distribution, melt pressure, and material ratio; the second energy consumption data includes extruder motor power consumption, heating system energy consumption, cooling system energy consumption, and auxiliary equipment energy consumption; and the third efficiency data includes product output, product qualification status, and material usage.

[0033] The barrel temperature distribution represents the temperature data of each heating zone and each cooling zone; the material ratio represents the proportion of each raw material.

[0034] Furthermore, the first structured dataset includes:

[0035] The first structured dataset is obtained by cleaning and normalizing the first production data, the second energy consumption data, and the third efficiency data, and then weighted and fused together.

[0036] The objective function construction module is used to construct a multimodal analysis model to analyze the first structured dataset and obtain the first multimodal features. The first multimodal features include production features, energy consumption features, and efficiency features. Based on these three features, the first set of multi-objective functions is obtained.

[0037] Furthermore, the multimodal analysis model includes a multidimensional data input layer, a multidimensional data preprocessing layer, a multimodal feature extraction layer, and a multimodal feature output layer; Figure 2 This is a schematic diagram of the structure of a multimodal analysis model provided in an embodiment of the present invention;

[0038] The multidimensional data input layer is used to input the first structured dataset into the multimodal analysis model;

[0039] The multidimensional data preprocessing layer is used to preprocess the first structured dataset to obtain a second structured dataset; the preprocessing includes normalization.

[0040] The multimodal feature extraction layer includes a production feature extraction sublayer, an energy consumption feature extraction sublayer, and an efficiency feature extraction sublayer; each sublayer performs feature analysis by fusing LSTM and CNN to obtain production features, energy consumption features, and efficiency features, respectively.

[0041] The multimodal feature output layer is used to output production features, energy consumption features, and efficiency features.

[0042] Specifically, each of the multimodal feature extraction layers deploys parallel LSTM and CNN branches, respectively used for deep extraction of temporal dynamic features and local features of the input data. After obtaining the feature vectors extracted by the LSTM and CNN branches, they are weighted based on preset weights of the two branches to obtain a fused "final feature vector." The preset weights are used to evaluate the confidence of the feature vectors extracted by the LSTM and CNN branches in real time based on a built-in quantization method (e.g., Monte Carlo Dropout). Features from branches with higher confidence are assigned higher weights during fusion.

[0043] The production characteristics include the average temperature of each region of the barrel, the maximum temperature difference of each region of the barrel, the average melt pressure, the ratio of screw speed to melt pressure, and the proportion of each material; the energy consumption characteristics include the total average power consumption, the average energy consumption of the heating system, and the average energy consumption of the cooling system; the efficiency characteristics include the output per unit time, the product qualification rate, and the material utilization rate.

[0044] This embodiment constructs a multimodal analysis model that integrates LSTM and CNN to perform deep feature extraction and fusion analysis on production, energy consumption, and efficiency data. Compared with traditional methods that rely on experience or single parameters, it can more accurately and comprehensively identify key influencing factors under complex working conditions, effectively analyze production, energy consumption, and efficiency characteristics, and effectively improve the overall production management efficiency of masterbatch production.

[0045] Furthermore, the process of obtaining the first multi-objective function set includes: obtaining energy consumption feature sub-functions based on energy consumption characteristics; obtaining efficiency feature sub-functions based on efficiency characteristics; obtaining energy consumption target sub-functions based on energy consumption feature sub-functions and a first influence coefficient; obtaining efficiency target sub-functions based on efficiency feature sub-functions and a second influence coefficient; and thus obtaining the first multi-objective function set.

[0046] Furthermore, based on various energy consumption characteristics and preset target values, an energy consumption characteristic sub-function is obtained; the specific formula is as follows:

[0047] ;

[0048] in, Represents the energy consumption characteristic sub-function; Indicates the number of energy consumption characteristics; Indicates the first One energy consumption characteristic; Indicates the first Target values ​​for each energy consumption characteristic;

[0049] The energy consumption target subfunction is obtained by multiplying the energy consumption characteristic subfunction and the first influence coefficient; the efficiency characteristic function is obtained based on the efficiency characteristic; the specific formula is as follows:

[0050] ;

[0051] in, Represents the efficiency characteristic sub-function; Indicates the number of efficiency features; Indicates the first One efficiency characteristic; Indicates the first The target value of each efficiency characteristic;

[0052] The efficiency objective function is obtained by multiplying the efficiency characteristic function and the second influence coefficient; and the first multi-objective function set is obtained based on the energy consumption objective function and the efficiency objective function.

[0053] The first set of multi-objective functions includes energy consumption objective sub-functions and efficiency objective sub-functions;

[0054] The first set of multi-objective functions is shown below:

[0055] ;

[0056] in, Represents the first set of multi-objective functions; Represents the energy consumption objective sub-function; Represent the efficiency objective subfunction;

[0057] The energy consumption objective function is:

[0058] ;

[0059] in, Indicates the first influence coefficient; Represents the energy consumption characteristic sub-function;

[0060] The efficiency objective function is:

[0061] ;

[0062] in, This represents the second influence coefficient; Represents the efficiency characteristic sub-function;

[0063] Furthermore, the specific acquisition process for the first influence coefficient and the second influence coefficient is as follows: Based on energy consumption characteristics and production characteristics, a correlation analysis is performed using the Pearson coefficient to obtain the energy consumption-production correlation coefficient; the energy consumption-production correlation coefficient is used as the first influence coefficient; based on efficiency characteristics and production characteristics, a correlation analysis is performed using the Pearson coefficient to obtain the efficiency-production correlation coefficient; the efficiency-production correlation coefficient is used as the second influence coefficient.

[0064] The multi-objective collaborative optimization module is used to construct an improved NSGA-II algorithm to optimize the first multi-objective function set, and to dynamically constrain the first multi-objective function set to finally determine the first multi-dimensional scheduling parameters; the first multi-dimensional scheduling parameters include production process scheduling parameters, energy consumption scheduling parameters, and efficiency scheduling parameters;

[0065] Furthermore, the improved NSGA-II algorithm dynamically adjusts the crossover probability, mutation probability, selection pressure, and distribution estimation parameters of the NSGA-II algorithm based on the first multimodal feature; specifically including:

[0066] By analyzing the energy consumption characteristics, the characteristic variance of each batch is obtained; if the variance fluctuates greatly, the global exploration capability needs to be enhanced, thus increasing the crossover probability.

[0067] By analyzing the variance of efficiency characteristics of different batches of masterbatch products, if the fluctuation is small, it indicates that the efficiency is relatively stable. In this case, it is necessary to reduce the probability of variation to achieve local optimization and focus.

[0068] When the correlation between the production characteristics of the masterbatch and the objective function is strong, it indicates that the production characteristics have a significant impact on the optimization objective. In this case, selection pressure should be increased to prioritize the retention of high-quality individuals.

[0069] The distribution estimation parameter controls the search range of crossover and mutation operations and affects the uniformity of solution distribution. When the multimodal features (production features, efficiency features, and energy consumption features) are unevenly distributed, the distribution estimation parameter is reduced to focus on local search; conversely, the search range is expanded.

[0070] For elite individuals in the Pareto front solution set output by NSGA-II, a simulated annealing algorithm is introduced to perform local depth search and perturbation.

[0071] The specific process includes:

[0072] Elite Individual Selection: Elite individuals are extracted from the Pareto Frontier of NSGA-II;

[0073] Using crowding distance filtering, individuals with sparser distributions are selected to maintain the diversity of the solution set; for example, if the front contains 100 solutions, the top 20% of individuals by crowding distance are selected for simulated annealing.

[0074] Simulated annealing initialization: with selected elite individuals As a starting point; Includes multi-dimensional scheduling parameters (such as extruder screw speed, barrel temperature distribution, melt pressure, and material ratio); determines the initial temperature, cooling coefficient (0.95), and termination temperature;

[0075] Disturbance mechanisms: including those related to The parameter vector is perturbed to generate a neighborhood solution. The perturbation method is based on a Gaussian distribution: the perturbation step size is adjusted according to (the average variance of each multimodal feature); the variance of the multimodal features;

[0076] Acceptance criteria: Calculation The objective function values; including energy consumption objective function values ​​and efficiency objective function values; and determine... Is it not inferior to If so, then accept. Otherwise, accept it according to the Metropolis criterion with probability.

[0077] Iterate until the termination temperature is reached to obtain the optimized elite individuals.

[0078] Furthermore, the improved NSGA-II algorithm constructs a targeted search strategy based on the importance analysis of each first multimodal feature; the strategy analyzes the correlation strength between the first multimodal features and each sub-objective of the first multi-objective function set; and dynamically adjusts the search focus of the improved NSGA-II algorithm.

[0079] Furthermore, the targeted search strategy includes obtaining a comprehensive importance weight based on the feature values ​​(such as the average temperature of each region of the barrel and the maximum temperature difference of each region of the barrel) and the objective function values ​​(such as the energy consumption objective function value and the efficiency objective function value) in the multimodal features;

[0080] This embodiment employs an improved NSGA-II algorithm, innovatively integrating simulated annealing to enhance the local depth search and perturbation of Pareto frontier elite individuals. It also dynamically adjusts algorithm parameters and constructs targeted search strategies based on multimodal characteristics. This hybrid optimization strategy can more effectively perform global searches in a complex multi-objective space of energy consumption and efficiency, overcoming the shortcomings of traditional optimization methods that easily get trapped in local optima or struggle to balance multiple conflicting objectives. This achieves a deeper level of synergistic optimization between energy consumption and efficiency, effectively improving the overall production management efficiency of masterbatch production.

[0081] Furthermore, the specific process for obtaining the comprehensive importance weight is as follows: Correlation coefficients are calculated using the Pearson coefficient formula based on the correlation values ​​of each feature value (e.g., average temperature of each region of the barrel, maximum temperature difference of each region of the barrel) and the objective function values ​​(e.g., energy consumption objective function value and efficiency objective function value) in the multimodal features; then, a correlation matrix is ​​obtained based on each correlation coefficient; the ratio is obtained by taking the quotient of the correlation coefficients of each feature and each objective function with the sum of the correlation coefficients of all features, and this ratio represents the importance weight of each feature to each objective function; finally, the comprehensive importance weight is obtained by weighting the feature and the importance weight of each objective function. Features with larger comprehensive importance weights: narrow the search range, focus on the neighborhood of the current high-quality solution, and promote local precision search. Features with smaller comprehensive importance weights: expand the search range.

[0082] The first multi-dimensional scheduling parameters include production process scheduling parameters (such as screw speed, barrel temperature, and material ratio), energy consumption scheduling parameters (such as motor power, heating energy consumption, and cooling energy consumption), and efficiency scheduling parameters (such as output, pass rate, and material utilization rate).

[0083] Furthermore, dynamic constraints include process constraints, equipment constraints, and safety constraints; process constraints are used to constrain the extrusion temperature range, screw speed limits, material ratio boundaries, and product quality; equipment constraints are used to constrain motor power limits, melt pressure range, cooling system capacity, and production line capacity limits; and safety constraints are used to constrain temperature safety boundaries, pressure safety limits, peak energy consumption limits, and environmental emission standards.

[0084] The production management module is used to dynamically manage the production process based on the first multi-dimensional scheduling parameters.

[0085] This embodiment not only optimizes the NSGA-II algorithm but also enables continuous dynamic management of the production process based on the optimized multi-dimensional scheduling parameters. Combined with dynamic constraints, it forms a closed-loop intelligent system encompassing data acquisition, analysis, optimization decision-making, and execution feedback. This mechanism allows production management to respond in real-time to changes in operating conditions and continuously optimize resource allocation, thereby significantly improving overall operational efficiency, material utilization, and energy efficiency, effectively enhancing the overall production management efficiency of masterbatch production.

[0086] This invention integrates production, energy consumption, and efficiency data into a structured dataset through a data acquisition and analysis module. The objective function construction module deeply mines the production, energy consumption, and efficiency features within the data and constructs a set of multi-objective optimization functions accordingly. The multi-objective collaborative optimization module employs an improved NSGA-II algorithm, introducing a simulated annealing mechanism to enhance the local search capability of elite solutions. It combines this with dynamically adjusted crossover and mutation probabilities based on multimodal features and targeted search strategies, while simultaneously applying dynamic process, equipment, and safety constraints to obtain globally optimal multi-dimensional scheduling parameters. This invention effectively improves the overall production management efficiency of masterbatch production through the combination of deep data analysis and advanced optimization algorithms.

[0087] Example 2

[0088] To improve the efficiency of color masterbatch production management, a collaborative optimization system for color masterbatch processing energy consumption and efficiency based on production data was applied, such as... Figure 1 A schematic diagram of a system for collaborative optimization of energy consumption and efficiency in masterbatch processing based on production data, provided in an embodiment of the present invention, specifically includes:

[0089] The data acquisition and analysis module is used to acquire first production data, second energy consumption data, and third efficiency data; based on the analysis of these three types of data, a first structured dataset is obtained.

[0090] The objective function construction module is used to construct a multimodal analysis model to analyze the first structured dataset and obtain the first multimodal features. The first multimodal features include production features, energy consumption features, and efficiency features. Based on these three features, the first set of multi-objective functions is obtained.

[0091] A multi-objective collaborative optimization module is used to construct an improved NSGA-II algorithm to optimize the first multi-objective function set. The improved NSGA-II algorithm introduces the simulated annealing algorithm into the NSGA-II algorithm and applies dynamic constraints to the first multi-objective function set to finally determine the first multi-dimensional scheduling parameters. The first multi-dimensional scheduling parameters include production process scheduling parameters, energy consumption scheduling parameters, and efficiency scheduling parameters.

[0092] The production management module is used to dynamically manage the production process based on the first multi-dimensional scheduling parameters.

[0093] Furthermore, the first structured dataset is obtained by weighted fusion of the first production data, the second energy consumption data, and the third efficiency data.

[0094] Furthermore, the multimodal analysis model includes a multidimensional data input layer, a multidimensional data preprocessing layer, a multimodal feature extraction layer, and a multimodal feature output layer; Figure 2 This is a schematic diagram of the structure of a multimodal analysis model provided in an embodiment of the present invention;

[0095] The multidimensional data input layer is used to input the first structured dataset into the multimodal analysis model;

[0096] The multidimensional data preprocessing layer is used to preprocess the first structured dataset to obtain the second structured dataset;

[0097] The multimodal feature extraction layer includes a production feature extraction sublayer, an energy consumption feature extraction sublayer, and an efficiency feature extraction sublayer; each sublayer performs feature analysis by fusing LSTM and CNN to obtain production features, energy consumption features, and efficiency features, respectively.

[0098] The multimodal feature output layer is used to output production features, energy consumption features, and efficiency features.

[0099] Furthermore, the process of obtaining the first multi-objective function set includes: obtaining energy consumption feature sub-functions based on energy consumption characteristics; obtaining efficiency feature sub-functions based on efficiency characteristics; obtaining energy consumption target sub-functions based on energy consumption feature sub-functions and a first influence coefficient; obtaining efficiency target sub-functions based on efficiency feature sub-functions and a second influence coefficient; and thus obtaining the first multi-objective function set.

[0100] Furthermore, the improved NSGA-II algorithm dynamically adjusts the crossover probability, mutation probability, selection pressure, and distribution estimation parameters of the NSGA-II algorithm based on the first multimodal feature; for elite individuals in the Pareto front solution set output by NSGA-II, a simulated annealing algorithm is introduced to perform local depth search and perturbation.

[0101] Furthermore, the improved NSGA-II algorithm constructs a targeted search strategy based on the importance analysis of each first multimodal feature; the strategy analyzes the correlation strength between the first multimodal features and each sub-objective of the first multi-objective function set; and dynamically adjusts the search focus of the improved NSGA-II algorithm.

[0102] Furthermore, dynamic constraints include process constraints, equipment constraints, and safety constraints; process constraints are used to constrain the extrusion temperature range, screw speed limits, material proportioning boundaries, and product quality requirements; equipment constraints are used to constrain motor power limits, melt pressure range, cooling system capacity, and production line capacity limits; and safety constraints are used to constrain temperature safety boundaries, pressure safety limits, peak energy consumption limits, and environmental emission standards.

[0103] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A system for collaborative optimization of energy consumption and efficiency in masterbatch processing based on production data, characterized in that, include: The data acquisition and analysis module is used to acquire the first production data, the second energy consumption data, and the third efficiency data. Based on the analysis of these three types of data, the first structured dataset is obtained; The objective function construction module is used to construct a multimodal analysis model to analyze the first structured dataset and obtain the first multimodal features. The first multimodal features include production features, energy consumption features, and efficiency features. Based on these three features, the first set of multi-objective functions is obtained. The process of obtaining the first multi-objective function set includes: obtaining energy consumption feature sub-functions based on energy consumption characteristics; obtaining efficiency feature sub-functions based on efficiency characteristics; obtaining energy consumption target sub-functions based on energy consumption feature sub-functions and a first influence coefficient; obtaining efficiency target sub-functions based on efficiency feature sub-functions and a second influence coefficient; and thus obtaining the first multi-objective function set. The specific acquisition process of the first influence coefficient and the second influence coefficient is as follows: performing correlation analysis based on energy consumption characteristics and production characteristics using the Pearson coefficient to obtain the energy consumption-production correlation coefficient; using the energy consumption-production correlation coefficient as the first influence coefficient; performing correlation analysis based on efficiency characteristics and production characteristics using the Pearson coefficient to obtain the efficiency-production correlation coefficient; and using the efficiency-production correlation coefficient as the second influence coefficient. A multi-objective collaborative optimization module is used to construct an improved NSGA-II algorithm to optimize the first multi-objective function set. The improved NSGA-II algorithm introduces the simulated annealing algorithm into the NSGA-II algorithm and applies dynamic constraints to the first multi-objective function set to finally determine the first multi-dimensional scheduling parameters. The first multi-dimensional scheduling parameters include production process scheduling parameters, energy consumption scheduling parameters, and efficiency scheduling parameters. The improved NSGA-II algorithm dynamically adjusts the crossover probability, mutation probability, selection pressure, and distribution estimation parameters of the NSGA-II algorithm based on the first multimodal feature; for elite individuals in the Pareto front solution set output by NSGA-II, a simulated annealing algorithm is introduced to perform local depth search and perturbation. The improved NSGA-II algorithm constructs a targeted search strategy based on the importance analysis of each first multimodal feature; the targeted search strategy is constructed by analyzing the correlation strength between the first multimodal features and each sub-objective of the first multi-objective function set; and dynamically adjusting the search focus of the improved NSGA-II algorithm. The production management module is used to dynamically manage the production process based on the first multi-dimensional scheduling parameters.

2. The color masterbatch processing energy consumption and efficiency collaborative optimization system based on production data according to claim 1, characterized in that: The first structured dataset is obtained by weighted fusion of the first production data, the second energy consumption data, and the third efficiency data.

3. The color masterbatch processing energy consumption and efficiency collaborative optimization system based on production data according to claim 1, characterized in that: The multimodal analysis model includes a multidimensional data input layer, a multidimensional data preprocessing layer, a multimodal feature extraction layer, and a multimodal feature output layer; The multidimensional data input layer is used to input the first structured dataset into the multimodal analysis model; The multidimensional data preprocessing layer is used to preprocess the first structured dataset to obtain the second structured dataset; The multimodal feature extraction layer includes a production feature extraction sublayer, an energy consumption feature extraction sublayer, and an efficiency feature extraction sublayer; Each sub-layer performs feature analysis by fusing LSTM and CNN to obtain production features, energy consumption features, and efficiency features respectively. The multimodal feature output layer is used to output production features, energy consumption features, and efficiency features.

4. The color masterbatch processing energy consumption and efficiency collaborative optimization system based on production data according to claim 1, characterized in that: The dynamic constraints include process constraints, equipment constraints, and safety constraints. Process constraints are used to constrain the extrusion temperature range, screw speed limit, material ratio boundary, and product quality requirements. Equipment constraints are used to constrain the motor power limit, melt pressure range, cooling system capacity, and production line capacity limit. Safety constraints are used to constrain the temperature safety boundary, pressure safety limit, energy consumption peak limit, and environmental emission standards.

Images