Energy-saving scheduling method for prefabricated metal panel composite wall production line based on artificial intelligence

Abstract

The present invention belongs to the field of intelligent manufacturing energy-saving technology, and discloses an energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence. The method extracts energy consumption influencing parameters from multi-source heterogeneous data, constructs an energy consumption influencing factor matrix F, and performs normalization processing to obtain a normalized matrix of energy consumption influencing factors. Then, the influencing factor interaction matrix A and the time decay factor λ are introduced to convert the time decay energy influencing factor interaction representation matrix X′, which is input into the energy consumption prediction deep learning model. Through the production scheduling optimization model of the genetic algorithm, combined with the production line equipment information and operation constraints, energy consumption optimization is achieved. The obtained optimal scheduling scheme can be directly converted into an equipment operation sequence, realizing real-time monitoring and continuous optimization, and enhancing production efficiency and energy-saving effects. The method effectively integrates the dynamic interaction and time decay characteristics of influencing factors, and provides a solid energy-saving strategy foundation for production scheduling.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent manufacturing energy-saving technology, and specifically relates to an energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence. Background Art

[0002] In recent years, the concepts of building industrialization and green building have been deeply rooted in people's hearts, and prefabricated and assembled building components have been widely promoted and applied. As an important part of the building envelope structure, prefabricated metal panel composite walls are favored for their excellent thermal insulation performance and convenient construction and installation. However, the production process of prefabricated metal panel composite walls is complicated, involving multiple processes such as raw material preparation, steel plate forming, thermal insulation layer laying, and internal and external decorative panel installation. The production equipment consumes a lot of energy, and the production efficiency and energy utilization rate need to be further improved.

[0003] At present, the transformation of manufacturing industry towards digitalization, intelligence and greening has become an inevitable trend. Scholars from both inside and outside the industry have carried out a lot of fruitful research work on improving the energy efficiency of manufacturing processes. Advanced data-driven modeling methods such as deep learning are used to model and predict equipment energy consumption, and multi-objective optimization algorithms such as genetic algorithms are used to solve production scheduling optimization problems. However, existing research has mainly focused on traditional fields such as mechanical manufacturing and process industries, and rarely considers the production characteristics of prefabricated and assembled building components. The production of prefabricated components, such as prefabricated metal panel composite walls, has complex factors affecting energy consumption, involving multi-source heterogeneous time-varying data such as equipment state parameters, workpiece processing characteristics, production task plans, and ambient temperature and humidity. The data presents complex characteristics such as high dimensionality, strong correlation, and nonlinear time-varying. How to extract key influencing factors from massive multimodal data, characterize the laws of dynamic time series evolution, and then accurately predict the energy consumption per unit product is a key problem that needs to be solved urgently to realize intelligent green manufacturing of prefabricated components. Traditional multivariate statistical learning methods are difficult to accurately model the mechanism of energy consumption, and the use of deep learning methods alone also faces technical bottlenecks such as insufficient model generalization ability. It is hoped that the interactive modeling method of energy consumption influencing factors that integrates multi-source heterogeneous data can be explored to fully tap the value of spatiotemporal data, improve the accuracy of modeling prediction, and provide precise guidance for energy-saving control of workshop production.

[0004] In addition, on the basis of obtaining the energy consumption prediction results, it is necessary to further optimize the scheduling of production equipment to minimize energy consumption while meeting process constraints and production tasks. This is essentially a multi-objective, multi-constrained combinatorial optimization scheduling problem. The traditional mathematical programming method has low solution efficiency and is difficult to adapt to the needs of real-time scheduling optimization. General heuristic intelligent optimization algorithms lack pertinence in dealing with complex constraints and coding design, resulting in algorithm searches falling into local optimality and unsatisfactory optimization effects. It is urgent to design an intelligent energy-saving scheduling optimization algorithm with compact coding and flexible and controllable constraints based on the characteristics of prefabricated component production, accelerate the solution of high-quality feasible scheduling solutions, and provide precise strategies for workshop intelligent production and energy management.

[0005] In summary, the present invention provides an energy-saving scheduling system and method for a prefabricated metal panel composite wall production line based on artificial intelligence. Summary of the invention

[0006] The purpose of the present invention is to provide an energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence, which is used to solve the technical problems raised in the existing background technology.

[0007] On the one hand, the present invention provides an energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence, comprising: obtaining energy consumption influencing parameters of the production of prefabricated metal panel composite walls, determining an energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of prefabricated metal panel composite walls; normalizing the energy consumption influencing factor matrix F; and obtaining an energy consumption influencing factor normalized matrix F;

[0008] The interaction matrix A of influencing factors is introduced, and the normalized matrix F of energy consumption influencing factors is transformed to obtain the interaction representation matrix X of energy influencing factors; the time decay factor λ is introduced to transform the interaction representation matrix X of energy influencing factors to determine the interaction representation matrix X′ of time decay energy influencing factors; the interaction representation matrix X′ of time decay energy influencing factors is input into the energy consumption prediction deep learning model, and the output of the energy consumption prediction deep learning model is the energy consumption prediction value of the time segment; the energy consumption prediction value of the time segment is input into the production scheduling optimization model based on genetic algorithm, and the production line equipment information, optimization objectives and constraints and genetic algorithm parameters are input at the same time, and the scheduling plan with optimal energy consumption is obtained through iterative evolution of genetic algorithm; the optimal scheduling plan is converted into an operation sequence executable by the equipment, and control instructions are issued to monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model.

[0009] On the one hand, the present invention provides an energy-saving scheduling system for a prefabricated metal panel composite wall production line based on artificial intelligence, the system comprising:

[0010] The processing module is used for S1 to obtain the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall, determine the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall; normalize the energy consumption influencing factor matrix F; obtain the energy consumption influencing factor normalized matrix; introduce the influencing factor interaction matrix A, transform the energy consumption influencing factor normalized matrix to obtain the energy influencing factor interaction representation matrix X; introduce the time attenuation factor λ to transform the energy influencing factor interaction representation matrix X to determine the time attenuation energy influencing factor interaction representation matrix X′;

[0011] An energy consumption prediction deep learning model determination module is used to input the time-attenuated energy impact factor interaction representation matrix X′ into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment;

[0012] The optimal solution determination module is used to input the energy consumption forecast value of the time segment into the production scheduling optimization model based on the genetic algorithm, and at the same time input the production line equipment information, optimization objectives and constraints and genetic algorithm parameters, and obtain the scheduling solution with the best energy consumption through iterative evolution of the genetic algorithm;

[0013] The optimization module is used to convert the optimal scheduling plan into an operation sequence that can be executed by the equipment, issue control instructions, monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model.

[0014] The present invention also provides an electronic device, comprising: a processor and a memory, wherein: the memory stores a computer program that can be called by the processor; the processor executes the energy-saving scheduling method for the prefabricated metal panel composite wall production line based on artificial intelligence in the background by calling the computer program stored in the memory.

[0015] The present invention also provides a computer-readable storage medium on which an erasable computer program is stored; when the computer program is run on a computer device, the computer device executes the energy-saving scheduling method for the prefabricated metal panel composite wall production line based on artificial intelligence in the background.

[0016] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0017] 1. The present invention proposes an energy-saving scheduling method for the production of prefabricated metal panel composite walls. The method first obtains multi-source heterogeneous energy consumption influencing parameters of the production process, such as production tasks, environmental conditions, equipment operating status, ambient temperature and humidity, raw material quality, etc., and obtains the normalized matrix of energy consumption influencing factors through factor matrix construction and data normalization. The interaction matrix of influencing factors and time attenuation factors are further introduced, and the normalized matrix is ​​transformed to obtain the interactive representation matrix of energy influencing factors, which comprehensively considers the self-effects and interactive effects of each influencing factor, as well as the dynamic attenuation law of the intensity of influence over time. The obtained representation matrix is ​​input into the energy consumption prediction deep learning model, which can significantly improve the accuracy of the energy consumption prediction of the production line. The energy consumption impact modeling method of the present invention overcomes the shortcomings of traditional multivariate statistical learning and single deep learning methods, and lays a solid foundation for subsequent production scheduling optimization.

[0018] 2. On the basis of obtaining the predicted value of production energy consumption in each time segment, the present invention inputs the energy consumption prediction results, production line equipment information, optimization objectives, constraints and genetic algorithm parameters into the optimization model, and quickly searches for the optimal scheduling solution under complex constraints by designing chromosome encoding methods, fitness functions, genetic operators such as selection, crossover and mutation. With the minimum energy consumption as the optimization goal, the fitness function is designed to be the inverse of the energy consumption value. The lower the energy consumption, the higher the fitness. In the process of genetic evolution, high energy consumption solutions are continuously eliminated, and the population is promoted to converge to a more energy-saving scheduling strategy, and finally the optimal scheduling solution that takes into account production efficiency and energy utilization is obtained. Compared with traditional mathematical programming and heuristic algorithms, the scheduling optimization method described in the present invention has higher solving efficiency, more flexible constraint processing, and more significant optimization effect.

[0019] 3. The invention has constructed an energy-saving scheduling system for a prefabricated metal panel composite wall production line based on the industrial Internet architecture. By deploying intelligent sensors, data acquisition and communication terminals on production equipment, multi-source heterogeneous data of the production process is collected in real time and uploaded to the industrial cloud platform. The cloud platform receives and pre-processes multimodal data, uses the energy consumption impact modeling and production scheduling optimization method described in the present invention to form an energy-saving scheduling strategy, and issues executable equipment control instructions. At the same time, the production execution status and energy consumption data are monitored in real time, the modeling parameters and scheduling strategies are dynamically updated, and continuous iterative optimization is achieved through industrial big data analysis and machine learning algorithms to achieve closed-loop control of the entire production process. This system has an advanced architecture and seamless integration of software and hardware, which can significantly improve the production efficiency of prefabricated components, reduce energy consumption per unit product, promote the upgrading of green and intelligent manufacturing in the industry, and have significant economic and social benefits. . BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 A schematic diagram of a process of the present invention is shown;

[0022] Figure 2 A timing diagram of the present invention is shown;

[0023] Figure 3 A system diagram of the present invention is shown;

[0024] Figure 4 An electronic device diagram of the present invention is shown;

[0025] Figure 5 A storage medium diagram of the present invention is shown. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] Example 1

[0028] S1. Obtain energy consumption influencing parameters of the production of prefabricated metal panel composite walls, and determine the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of prefabricated metal panel composite walls;

[0029] Among them, the consumption impact factor matrix F is as follows:

[0030]

[0031] Among them, in the production of prefabricated metal panel composite walls, the parameters affecting energy consumption are numerous and complex. In addition to the equipment operating status, they are also related to production tasks, environmental conditions, etc. nmIt indicates the influence value of the mth factor in the nth time segment. For example, the energy consumption of the equipment in different operating states, such as on, off, and standby, varies greatly. The energy required for processing metal panels of different models and quantities is also different. Ambient temperature: Too high or too low temperature will increase the energy consumption burden of the equipment. In addition, factors such as raw material quality and worker operating habits will also affect energy consumption to a certain extent.

[0032] For example, we divide the entire production process into several time segments, such as 30 minutes per segment. For the mth influencing factor in the nth time segment, we give it an impact value f nm This value can be calculated using different methods based on the characteristics of different factors.

[0033] We use corresponding quantitative mapping methods for different types of energy consumption influencing factors and convert them into unified numerical representations to facilitate subsequent mathematical modeling and calculations. Specifically:

[0034] Device operating state: Based on the three typical working modes of the device, we define the impact value of the "on" state as 1, the "off" state as 0, and the "standby" state as 0.3. This metric value intuitively reflects the relative high and low energy consumption of the device in different operating states. The larger the value, the greater the impact of the state on energy consumption.

[0035] Production task volume: In each time segment, we use the number of processed metal panels as the impact value of this factor. The number of processed panels is positively correlated with the energy consumption of the equipment. The more the number, the higher the equipment load, and the greater the corresponding energy consumption impact. Therefore, this metric can quantitatively reflect the impact of production task intensity on energy consumption.

[0036] Energy consumption: For some devices, we can directly obtain their actual energy consumption value in a time segment, such as the 50 kWh of electricity consumed by the high-speed running device in the i-th segment. This value can be directly used as an impact value to accurately characterize the energy consumption performance of the device in a specific state.

[0037] Ambient temperature: As a continuous variable, we can directly use the value of Celsius as its impact value. Temperature and equipment energy consumption often show a nonlinear relationship. Too high or too low will significantly increase cooling / heating energy consumption. Therefore, we can also consider using methods such as mean or piecewise linear function to map temperature to a measurement domain that better matches the energy efficiency characteristics of the equipment, etc.

[0038] Through the above methods, the originally heterogeneous energy consumption influencing factors are uniformly mapped to the numerical space to form a standardized impact value representation, which lays a solid foundation for the subsequent construction of the energy consumption influencing factor matrix and the quantitative analysis of the interaction of factors. At the same time, the mapping and quantification methods of the impact values ​​can be flexibly adjusted according to actual needs and data characteristics, reflecting the adaptability and scalability of this method. Through the above methods, we get an energy consumption influencing factor matrix F, where the number of rows represents the time segment and the number of columns represents the factors considered. However, the dimensions of different factors are different, and the numerical values ​​vary greatly, so they cannot be used directly.

[0039] S2 normalizes the energy consumption impact factor matrix F to obtain the energy consumption impact factor normalized matrix

[0040] The purpose of normalization is to scale the value range of each factor to a similar interval, such as [0, 1], so that their impact weights on energy consumption are relatively balanced. The specific calculation of normalization is to subtract the mean of each impact value from the impact value, and then divide it by the standard deviation of the impact value to obtain the normalized matrix of energy consumption impact factors. To solve the above problems, we can design a factor interaction representation (FIR) method to transform and enhance F and obtain the energy factor interaction representation matrix X for subsequent modeling.

[0041] Normalize F to eliminate the dimension effect:

[0042]

[0043] where μ j and σ j are the mean and standard deviation of the jth factor, respectively.

[0044] In such a complex production system, there is often an interactive coupling effect between the various energy consumption influencing factors, that is, the combined effect of certain factors may produce an energy consumption impact far exceeding the independent linear superposition. In order to fully consider and quantify this interactive effect in the modeling process, this embodiment introduces an influencing factor interaction matrix A. The matrix A is a symmetric matrix, in which the element α ij Characterizes the strength of the interaction between the i-th factor and the j-th factor, α ij The value of can be defined based on engineering domain knowledge, expert experience or data analysis results. When the interaction between factors i and j is significant, the corresponding α ij The value should be set larger; on the contrary, if the interaction effect between the two factors is weak or even negligible, then α ij The value should be set smaller. When i=j, α ij Represents the independent influence strength of factor i itself.

[0045] To put it more simply, the factors affecting energy consumption interact with each other. When certain factors are combined, their combined impact on system energy consumption is far greater than the simple linear sum of the independent effects of each factor. This "1+1>2" effect is common in actual production processes. For example, the superposition of high temperature environment and high-load operation of equipment will cause a sharp drop in equipment energy efficiency, resulting in an increase in energy consumption that is far higher than the sum of the two independent effects. For another example, the impact of different production tasks on the energy consumption of equipment may also have an interactive effect, and certain task combinations may cause a jump in energy consumption. Considering the interaction will help to more comprehensively and accurately characterize the energy consumption law of the production process, and provide a reliable basis for subsequent prediction and optimization.

[0046] Introduce the impact factor interaction matrix A, A is an m×m symmetric matrix, element α (ij) =α (ji) It represents the strength of the interaction between factors i and j, which can be determined based on domain knowledge or data analysis.

[0047] Specifically, the impact factor interaction matrix A is a symmetric matrix, each element α ij =α ji Indicates the strength of the interaction between factor i and factor j. This value can be given based on our understanding and experience of the production process. We give a larger value to a pair of factors with a strong interaction, and a smaller value to a pair of factors with a strong interaction. When i=j, α ii Represents the influence strength of factor i itself.

[0048]

[0049] For example, consider the three variables of equipment status (S), production task volume (T) and ambient temperature (E), which have an interaction in affecting energy consumption. Specifically: the interaction intensity between equipment status and production task volume is high. When the equipment is running at full load, its energy consumption is significantly more responsive to the task volume than when it is in the no-load state; the interaction between equipment status and ambient temperature is moderate. The energy consumption of equipment running at full load under high temperature conditions increases significantly, while the impact is small when it is in the no-load state; the interaction between production task volume and ambient temperature is weak, and their impact on energy consumption can be considered relatively independent.

[0050] Based on these qualitative analyses, the impact factor interaction matrix A can be constructed to quantify these interaction effects:

[0051]

[0052] In the above formula, α SS , α TT , α EE Indicates the influence strength of the factor itself, all set to 1.

[0053] Off-diagonal elements α ST , α SE , α TE It represents the interaction strength between factors, and its values ​​are set to 0.8, 0.5, and 0.2 according to the degree of influence.

[0054] Step 4: Transform the energy consumption influencing factor normalized matrix The energy impact factor interaction representation matrix X is obtained.

[0055] Specifically, use A to normalize Transform and obtain the energy impact factor interaction representation matrix X:

[0056]

[0057] The expanded form is:

[0058]

[0059] x ik It represents the interactive representation of the influencing factors of the i-th time segment and the k-th time segment, taking into account the interaction between various factors. The element x in the energy influencing factor interactive representation matrix X ik The self-influence and interactive influence of each factor in the i-th and k-th time segments are comprehensively considered.

[0060] It integrates the cross-combination effects of the influencing factors between two time segments. ik The larger the value of is, the stronger the correlation between the energy consumption of the i-th and k-th time segments is, and there may be some common influence patterns. When i=k, x ii It reflects the self-influence and interaction of each factor in the i-th time segment. In this way, the energy influencing factor interaction representation matrix X realizes the interactive feature extraction of energy consumption influencing factors in two dimensions, namely, between time segments and within segments, so that the subsequent energy consumption modeling and prediction can fully consider the complex correlation of dynamic factors, thereby significantly improving the performance of the system. This interactive representation method also enables the algorithm to adaptively capture and learn the evolution law of energy consumption in the production process, with greater robustness and practical value.

[0061] Step 5: Introduce the time decay factor λ to transform the energy influence factor interaction representation matrix X to determine the time decay energy influence factor interaction representation matrix X′;

[0062] The time decay energy influence factor interaction representation matrix X′ in which x′ ik The specific formula of the element is as follows:

[0063] x′ ik =λ|i-k| x ik i,k=1,2,…,n

[0064] Among them, λ∈(0,1] is the time decay factor, and |ik| is the absolute value of the time segment index difference. The smaller λ is, the faster the dependency between time segments decays.

[0065] Generally speaking, the closer the time segments are, the stronger the correlation of their energy consumption should be, while the correlation of energy consumption of the far-separated time segments may be weak. In order to reflect this time dependence, we represent each element x in the energy impact factor interaction representation matrix X ik Multiply by an attenuation factor λ |i-k| .

[0066] The value range of λ is (0, 1], and its value reflects the decay speed of the time interval on the energy consumption correlation. |ik| is the absolute value of the order difference between the i-th time segment and the k-th time segment. The larger |ik| is, the farther the interval between the two time segments is, and λ is |i-k| The smaller the value of ik The more it decays.

[0067] For example, if we set λ = 0.5, if |ik| = 1, it means that the interaction between two adjacent time segments will be multiplied by 0.5 1 = 0.5; if |ik| = 3, it means that there are 3 time segments apart, and their interaction effect will be multiplied by 0.5 3 =0.125, which decays to a very small value. This is equivalent to applying a time-decayed "filter" to the matrix X, allowing the energy influencing factor interaction representation matrix to pay more attention to recent energy consumption-related patterns.

[0068] Through the above influencing factor interaction characterization method, the original energy consumption influencing factor matrix F is mapped to a new feature space, which reflects the multi-factor interaction and time series dependency, greatly improves the energy consumption prediction performance of the deep learning model, and lays the foundation for subsequent intelligent scheduling optimization.

[0069] S6 inputs the time-attenuated energy impact factor interaction representation matrix X′ into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment. The model is continuously trained and optimized with actual energy consumption data to improve the prediction accuracy. The deep learning model can be a long short-term memory network (LSTM), a convolutional neural network (CNN), a graph neural network (GNN), a Transformer model, or a combination and integration of multiple models, such as CNN+LSTM, GNN+Transformer, etc., to make full use of the advantages of different models to improve the performance of energy consumption prediction. At the same time, attention should also be paid to the complexity and computational efficiency of the model, weighing the prediction accuracy and real-time requirements, and selecting the most suitable model.

[0070] Step 7: Input the energy consumption forecast value of the time segment into the production scheduling optimization model based on genetic algorithm, and input the production line equipment information, optimization objectives and constraints and genetic algorithm parameters at the same time, and obtain the scheduling plan with optimal energy consumption through iterative evolution of genetic algorithm.

[0071] It is understandable that production scheduling optimization is performed based on genetic algorithm (GA). Chromosome encoding uses a two-dimensional matrix of n×m to represent a scheme, where n represents the number of devices and m represents the number of time segments. Each element in the matrix represents the operating state of the corresponding device in the corresponding time segment, which can be on, off or standby. For example, if the element in the i-th row and j-th column of the matrix is ​​"on", it means that the i-th device is in the on state in the j-th time segment. This encoding method intuitively maps the scheduling scheme to chromosomes, which is convenient for subsequent genetic algorithm operations. Randomly generating feasible decompositions as the initial population can be understood as we can randomly generate some n×m two-dimensional matrices and check whether each matrix meets the conditions such as equipment energy consumption constraints and production task requirements. The matrices that meet the conditions are used as individuals in the initial population, and the fitness function is defined as the inverse of the total energy consumption prediction value. Iterate evolution until the termination condition is met to obtain the scheduling scheme with the optimal total energy consumption.

[0072] It should be noted that we only used one set of energy consumption prediction deep learning models, not two independent models. The function of this set of energy consumption prediction deep learning models is to predict the energy consumption per unit time. They accept the time-attenuated energy influence factor interaction representation matrix X′ as input and output the predicted value of the energy consumption per unit time of each device in the time segment.

[0073] Specifically, for x devices and y time segments on a production line, we need to train x*y energy consumption prediction deep learning models, each corresponding to a specific device and time segment combination. These models share the same architecture and hyperparameters, but are trained using different training data (i.e., historical operation data under different devices and time segments).

[0074] During the training phase, we paired the time-attenuated energy influencing factor interaction representation matrix with the corresponding actual unit time energy consumption to form training samples. By minimizing the error between the predicted value and the actual value, the model learned the mapping relationship between the influencing factors and the unit time energy consumption.

[0075] In the application phase, for a new scheduling scheme, we first encode it into a device operation matrix, and then use the trained energy consumption prediction deep learning model to predict the unit time energy consumption of each device in each time segment to form a unit time energy consumption matrix. Finally, the device operation matrix is ​​multiplied and summed with the corresponding elements of the unit time energy consumption matrix to obtain the total energy consumption prediction value under the scheduling scheme.

[0076] The total energy consumption prediction value is directly used as the fitness function of the genetic algorithm to guide the genetic algorithm to search for the scheduling plan with the optimal energy consumption. During the search process, we only need to reuse this set of energy consumption prediction deep learning models to calculate the total energy consumption prediction value for each candidate scheduling plan without the need to train a new model.

[0077] Exemplarily, the chromosome encoding process is: for x devices, y time segments, each segment can be operated as on / off / standby. The encoding uses a two-dimensional matrix of x×y, and each element is 0, 1, and 2. For example, [102; 011] means that the operations of 2 devices in 3 time segments are on-off-standby and off-on-on. Initial population generation: set the population size N, and randomly generate N feasible solutions as the initial population according to the encoding rules. The feasible solutions must meet the process constraints, such as ensuring that the key process is not interrupted. Fitness function construction: Under the conditions of meeting production requirements and process constraints, the optimization goal is to minimize energy consumption. Let the total energy consumption of the i-th individual (scheduling plan) be E. i , the fitness function is defined as Total energy consumption E i It can be estimated from the device operation matrix and the corresponding energy consumption per unit time, and the energy consumption per unit time can be predicted by the deep learning model.

[0078] The genetic operator design process is as follows: Selection operator: Roulette selection is used, and individuals with high fitness have a greater probability of being selected. Crossover operator: Uniform crossover is used to randomly select certain positions of two individuals for binary bit swapping. Mutation operator: Basic bit mutation is used to reverse certain gene bits of individuals with a certain probability. Termination condition: Set the maximum number of iterations or the expected energy consumption threshold. The process is terminated when the maximum number of iterations is reached or the energy consumption of the optimal individual in the population is lower than the threshold.

[0079] The design of genetic operators is closely integrated with the production characteristics of composite walls. The roulette selection with elite retention not only ensures that high-quality solutions are passed down, but also gives inferior solutions a certain chance of survival, thus maintaining population diversity. The crossover operator uses crossover in two dimensions: process and time period. The process dimension crossover is to exchange the scheduling strategies of different equipment; the time period dimension crossover is to reorganize the operating status of the same equipment in different time periods. The mutation operator introduces adaptive mutation probability, which mutates with a higher probability in the early stage of evolution to jump out of the local optimum; in the later stage, it mutates with a lower probability and conducts local search near the optimal solution.

[0080] Step 8: Convert the optimal scheduling plan into an operational sequence that can be executed by the equipment, issue control instructions, monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model

[0081] Result feedback execution: The optimal scheduling plan obtained by the genetic algorithm is converted into an operation sequence executable by the equipment and control instructions are issued. The execution status and energy consumption data are monitored in real time, and the actual energy consumption data is fed back to the deep learning model to continuously improve the prediction performance.

[0082] The iterative convergence of the genetic algorithm is judged by the criteria that the average fitness variance of the population is sufficiently small or the maximum number of iterations is reached. The optimal individual at the time of convergence is decoded into an on / off sequence that can be executed by the equipment and sent to the control systems of key process equipment such as CNC machine tools, CNC punching machines, bending machines, shearing machines, palletizers, welding robots, etc. The scheduling instructions are sent to the MES and EMS systems simultaneously to coordinate and optimize production planning and energy management.

[0083] Example 2

[0084] This embodiment is based on the embodiment 1, and the same parts as the embodiment 1 are not repeated. Figure 2 As shown, the present invention provides an energy-saving scheduling system 200 for a prefabricated metal panel composite wall production line based on artificial intelligence, and the system includes:

[0085] The data processing module 210 is used for S1 obtaining the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall, determining the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall; normalizing the energy consumption influencing factor matrix F; obtaining the energy consumption influencing factor normalized matrix; introducing the influencing factor interaction matrix A, transforming the energy consumption influencing factor normalized matrix to obtain the energy influencing factor interaction representation matrix X; introducing the time attenuation factor λ to transform the energy influencing factor interaction representation matrix X to determine the time attenuation energy influencing factor interaction representation matrix X′;

[0086] The energy consumption prediction deep learning model determination module 220 is used to input the time-attenuated energy impact factor interaction representation matrix X′ into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment;

[0087] The optimal solution determination module 230 is used to input the energy consumption prediction value of the time segment into the production scheduling optimization model based on the genetic algorithm, and at the same time input the production line equipment information, optimization objectives and constraints and genetic algorithm parameters, and obtain the scheduling solution with the optimal energy consumption through iterative evolution of the genetic algorithm;

[0088] The optimization module 240 is used to convert the optimal scheduling plan into an operation sequence executable by the device, issue control instructions, monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model.

[0089] Example 3

[0090] A schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 4 As shown, according to another aspect of the present application, an electronic device 300 is also provided. The electronic device 300 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, the energy-saving scheduling method for the prefabricated metal panel composite wall production line based on artificial intelligence as described above can be executed.

[0091] The method or system according to the embodiment of the present application can also be used by Figure 4 The electronic device architecture shown in FIG. Figure 4As shown, the electronic device 300 may include a bus 301, one or more CPUs 302, a read-only memory (ROM) 303, a random access memory (RAM) 304, a communication port 305 connected to a network, an input / output component 306, a hard disk 307, etc. The storage device in the electronic device 300, such as the ROM 303 or the hard disk 307, may store the energy-saving scheduling method for the prefabricated metal panel composite wall production line based on artificial intelligence provided in the present application. The energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence may, for example, include: obtaining energy consumption influencing parameters for the production of prefabricated metal panel composite walls, and determining an energy consumption influencing factor matrix F through the energy consumption influencing parameters for the production of prefabricated metal panel composite walls; normalizing the energy consumption influencing factor matrix F; obtaining a normalized matrix of energy consumption influencing factors; introducing an influencing factor interaction matrix A, transforming the energy consumption influencing factor normalized matrix to obtain an energy influencing factor interaction representation matrix X; introducing a time decay factor λ to transform the energy influencing factor interaction representation matrix X to determine a time decay energy influencing factor interaction representation matrix X′; inputting the time decay energy influencing factor interaction representation matrix X′ into an energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment; inputting the energy consumption prediction value of the time segment into a production scheduling optimization model based on a genetic algorithm, and simultaneously inputting production line equipment information, optimization objectives and constraints, and genetic algorithm parameters, and obtaining an energy-optimal scheduling plan through iterative evolution of the genetic algorithm; converting the optimal scheduling plan into an operation sequence executable by the equipment, issuing control instructions, monitoring the execution status and energy consumption data in real time, and continuously optimizing the scheduling strategy and prediction model. Furthermore, the electronic device 300 may further include a user interface 308. Figure 4 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 4 One or more components of an electronic device are shown.

[0092] Example 4

[0093] A schematic diagram of a computer-readable storage medium structure provided by an embodiment of the present application. Figure 5 As shown, a computer-readable storage medium 400 according to an embodiment of the present application is shown. Computer-readable instructions are stored on the computer-readable storage medium 400. When the computer-readable instructions are executed by the processor, the energy-saving scheduling method for the prefabricated metal panel composite wall production line based on artificial intelligence according to the embodiment of the present application described with reference to the above figures can be executed. The storage medium 400 includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0094] In addition, according to the implementation mode of the present application, the process described in the above reference flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium stores machine-readable instructions, the machine-readable instructions can be executed by a processor to execute instructions corresponding to the method steps provided in the present application, for example: obtaining energy consumption influencing parameters of the production of prefabricated metal panel composite walls, determining the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of prefabricated metal panel composite walls; normalizing the energy consumption influencing factor matrix F; obtaining the energy consumption influencing factor normalized matrix; introducing the influencing factor interaction matrix A, transforming the energy consumption influencing factor normalized matrix to obtain the energy influencing factor interaction representation matrix X; introducing the time attenuation factor λ to the energy influencing factor The sub-interaction characterization matrix X is transformed to determine the time-attenuation energy impact factor interaction characterization matrix X′; the time-attenuation energy impact factor interaction characterization matrix X′ is input into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment; the energy consumption prediction value of the time segment is input into the production scheduling optimization model based on the genetic algorithm, and the production line equipment information, optimization objectives and constraints and genetic algorithm parameters are input at the same time, and the scheduling plan with the best energy consumption is obtained through iterative evolution of the genetic algorithm; the optimal scheduling plan is converted into an operation sequence executable by the equipment, and control instructions are issued, the execution status and energy consumption data are monitored in real time, and the scheduling strategy and prediction model are continuously optimized. When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0095] The methods, apparatuses, and devices of the present application may be implemented in many ways. For example, the methods, apparatuses, and devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless otherwise specified. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the method according to the present application. Therefore, the present application also covers recording media storing programs for executing the method according to the present application.

[0096] In addition, the parts of the above-mentioned technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive redundancy.

[0097] The specific implementation modes as described above further describe the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation mode of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence, characterized in that: include: Obtain energy consumption influencing parameters of the production of prefabricated metal panel composite walls, and determine the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of prefabricated metal panel composite walls; Normalize the energy consumption influencing factor matrix F; obtain the energy consumption influencing factor normalized matrix Introduce the interaction matrix A of the impact factors and transform the normalized matrix of the energy consumption impact factors Get the energy impact factor interaction representation matrix X; The time decay factor λ is introduced to transform the energy influence factor interaction representation matrix X to determine the time decay energy influence factor interaction representation matrix X′; The time-attenuated energy impact factor interaction representation matrix X′ is input into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment; Input the energy consumption prediction value of the time segment into the production scheduling optimization model based on genetic algorithm, and input the production line equipment information, optimization objectives and constraints and genetic algorithm parameters at the same time, and obtain the scheduling plan with optimal energy consumption through iterative evolution of genetic algorithm; Convert the optimal scheduling plan into an operational sequence that can be executed by the equipment, issue control instructions, monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model; Normalizing the energy consumption influencing factor matrix F; Get the normalized matrix of energy consumption influencing factors as follows: For each element f in the energy consumption influencing factor matrix F nm Perform normalization processing to obtain the normalized matrix of the corresponding energy consumption influencing factors Elements Among them, F and The expression is as follows: where μ m and σ m are the mean and standard deviation of the mth factor, f nm It represents the influence value of the mth factor in the nth time segment; The energy impact factor interaction representation matrix X determination process includes: Construct the impact factor interaction matrix A, The influencing factor interaction matrix A is a symmetric matrix used to characterize the interaction between factors. ij =α ji It represents the strength of the interaction between factors i and j; Using the impact factor interaction matrix A, the normalized Transform and obtain the energy impact factor interaction representation matrix X: The expanded form is: x ik represents the interactive representation value of the influencing factors of the i-th time segment and the k-th time segment, taking into account the interaction between various factors, and the element x in the energy influencing factor interactive representation matrix X ik The self-influence and interactive influence of each factor in the i-th and k-th time segments are comprehensively considered; The introduction of the time decay factor λ to transform the energy influence factor interaction representation matrix X to determine the time decay energy influence factor interaction representation matrix X′ comprises: The specific formula for introducing the time decay factor λ to transform the energy influence factor interaction representation matrix X is as follows: x′ ik =λ |i-k| x ik i,k=1,2,…,n Among them, λ∈(0,1] is the time decay factor, |ik| is the absolute value of the time segment index difference, and the smaller λ is, the faster the dependency between time segments decays.

2. The energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence according to claim 1, characterized in that: The energy consumption influencing parameters of the production of the prefabricated metal panel composite wall include, but are not limited to: production tasks, environmental conditions, equipment operating status, ambient temperature, ambient humidity and raw material quality.

3. The energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence according to claim 1, characterized in that: Also includes: Taking the minimum energy consumption as the optimization goal, let the energy consumption of the i-th scheduling scheme be E i , the fitness function is 4. The energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence according to claim 1, characterized in that: The genetic operator design process based on the genetic algorithm is as follows: Selection operator: Roulette wheel selection is used, and individuals with high fitness have a greater probability of being selected. Crossover operator: Uniform crossover is used to randomly select certain positions of two individuals for binary bit swapping. Mutation operator: Basic bit mutation is used to reverse certain gene bits of individuals with a certain probability. Termination condition: Set the maximum number of iterations or the expected energy consumption threshold. The process is terminated when the maximum number of iterations is reached or the energy consumption of the optimal individual in the population is lower than the threshold.

5. An energy-saving scheduling system for a prefabricated metal panel composite wall production line based on artificial intelligence, which is used to implement an energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence as described in any one of claims 1 to 4, characterized in that: The system comprises: The data processing module is used for S1 to obtain the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall, and determine the energy consumption influencing factor matrix F through the energy consumption influencing parameters of the production of the prefabricated metal panel composite wall; normalize the energy consumption influencing factor matrix F; obtain the energy consumption influencing factor normalized matrix; introduce the influencing factor interaction matrix A, transform the energy consumption influencing factor normalized matrix to obtain the energy influencing factor interaction representation matrix X; introduce the time decay factor λ to transform the energy influencing factor interaction representation matrix X to determine the time decay energy influencing factor interaction representation matrix X′; An energy consumption prediction deep learning model determination module is used to input the time-attenuated energy impact factor interaction representation matrix X′ into the energy consumption prediction deep learning model, and the energy consumption prediction deep learning model outputs the energy consumption prediction value of the time segment; The optimal solution determination module is used to input the energy consumption forecast value of the time segment into the production scheduling optimization model based on the genetic algorithm, and at the same time input the production line equipment information, optimization objectives and constraints and genetic algorithm parameters, and obtain the scheduling solution with the best energy consumption through iterative evolution of the genetic algorithm; The optimization module is used to convert the optimal scheduling plan into an operation sequence that can be executed by the equipment, issue control instructions, monitor the execution status and energy consumption data in real time, and continuously optimize the scheduling strategy and prediction model.

6. An electronic device, characterized in that: include: A processor and a memory, wherein: The memory stores a computer program that can be called by the processor; The processor executes the energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence as described in any one of claims 1 to 4 in the background by calling the computer program stored in the memory.

7. A computer-readable storage medium, characterized in that: A rewritable computer program is stored thereon; When the computer program runs on a computer device, the computer device executes the energy-saving scheduling method for a prefabricated metal panel composite wall production line based on artificial intelligence as described in any one of claims 1 to 4 in the background.

Images