Digital twinning-based process flow and equipment energy efficiency operation optimization method and system

Abstract

The invention discloses a process flow and equipment energy efficiency operation optimization method and system based on digital twinning, and belongs to the technical field of data processing, and the method comprises the steps: collecting first data of a production process, carrying out the preprocessing of the first data, and obtaining second data; analyzing the second data to construct a mathematical model of the process flow; setting an optimization target and a constraint target of equipment energy efficiency, and generating an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model; solving the optimization function and the constraint function, obtaining a plurality of optimization schemes, constructing an equipment aging model, extracting equipment parameters from the optimization schemes, inputting the equipment parameters into the equipment aging model for simulation operation, and calculating the maintenance cost; and selecting an optimal scheme from the optimization schemes based on the maintenance cost, and optimizing the technological process based on the optimal scheme. According to the method, the technological process can be optimized under the condition that the energy efficiency of the equipment is controlled, and the optimization cost can be calculated.

Description

Technical Field

[0001] The present invention belongs to the field of digital twin technology, and specifically relates to a method and system for optimizing process flow and equipment energy efficiency operation based on digital twin. Background Art

[0002] The emergence of digital twin technology has provided a new solution for factory production management and simulation, and optimizing production processes based on digital twin technology has become the current mainstream research direction.

[0003] For example, the Chinese patent document with publication number CN116663725A discloses a process flow intelligent prediction and optimization method and system for industrial Internet. The method first establishes an association network of interactions between various nodes, then uses neural network technology to realize feature extraction and intelligent classification of process flow nodes, and observes and screens the operating status of each process flow node in real time. According to the feature classification, the process flow nodes that need to be maintained and optimized are judged to realize integrated process flow intelligent optimization. For another example, the Chinese patent document with publication number CN111428923A discloses a process flow optimization method for intelligent factories based on big data. The method uses a decision tree algorithm to train and learn process data to generate process rules, and then stores them in a historical process rule model library; then, by monitoring the actual production process, real-time process flow data is generated, and training is performed through a neural network algorithm and process flow data to establish a real-time process rule model library; the real-time process rule model library is compared and analyzed with the historical process rule model library to generate process decisions, and then the production process is optimized.

[0004] The optimization goal of the above scheme is relatively simple, and only considers the optimization of process product quality. Generally, changing the process of production will inevitably involve the adjustment of equipment parameters, and the adjustment of equipment parameters will lead to accelerated aging of equipment and increased production energy efficiency. Summary of the invention

[0005] To solve the above problems, the present invention provides a method and system for optimizing process flow and equipment energy efficiency operation based on digital twins, so as to provide an optimization method that comprehensively considers process flow and equipment energy efficiency.

[0006] In order to achieve the above-mentioned purpose of the invention, the present invention proposes a method for optimizing process flow and equipment energy efficiency operation based on digital twin, comprising:

[0007] Collecting first data of a production process, the first data including production quality data, equipment parameter data and production link data, and preprocessing the first data to obtain second data;

[0008] Analyzing the second data to construct a mathematical model of the process flow, the mathematical model comprising dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, the independent variables comprising equipment parameters;

[0009] Selecting and setting an optimization target from the dependent variables, setting a constraint target for equipment energy efficiency, and generating an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model;

[0010] Solving the optimization function and the constraint function to obtain a plurality of optimization schemes, wherein the optimization schemes include a numerical matrix including adjustment values ​​of a plurality of the independent variables;

[0011] Constructing an equipment aging model, the equipment aging model is used to simulate the aging process of production equipment under different equipment parameters, extracting equipment parameters from the independent variables of the optimization scheme, inputting the equipment parameters into the equipment aging model for simulation operation, and calculating maintenance costs;

[0012] An optimal solution is selected from the optimization solutions based on the maintenance cost, and the process flow is optimized based on the optimal solution.

[0013] Further, analyzing the second data comprises the following steps:

[0014] The second data is divided into a first variable and a second variable, one of the first variables is extracted as an analysis variable, and one of the second variables is extracted as a target variable, a demarcation threshold is determined based on the target variable, and the values ​​of the target variable that are less than the demarcation threshold are defined as first values, and the rest are defined as second values, the analysis variable corresponding to the first value is a third value, and the analysis variable corresponding to the second value is a fourth value, and the influence factor L is calculated based on a first formula, and the first formula is: Wherein, P1 is the average value of the first values, P2 is the average value of the second values, P3 is the average value of the third values, and P4 is the average value of the fourth values;

[0015] Repeat this step to traverse all the second variables, sort all the influencing factors obtained, screen the target variables corresponding to the first M influencing factors, filter the screened target variables, and use the filtered target variables as the independent variables of the analysis variables, and the corresponding analysis variables as the dependent variables. If the independent variable does not exist for the analysis variable, use the analysis variable as the independent variable.

[0016] Further, determining the demarcation threshold based on the target variable comprises the following steps:

[0017] A progressive value is set, and an advanced value is obtained by increasing the target variable with the smallest value by the progressive value, and the target variable is divided into a first part and a second part using the advanced value, and an evaluation value of the advanced value is calculated, and the evaluation value is used to evaluate the division effect of the advanced value, and the progressive value is increased to obtain a new advanced value, and the evaluation value is recalculated based on the new advanced value, and this step is repeated until the advanced value exceeds the maximum value of the second variable, and the advanced value corresponding to the maximum evaluation value is used as the demarcation threshold.

[0018] Furthermore, calculating the evaluation value includes the following steps:

[0019] A second formula is set, and the evaluation value R is calculated based on the second formula, where the second formula is: Among them, V1 is the average value of the first part after division, V2 is the average value of the second part after division, U1 is the sum of square deviations of the first part, U2 is the sum of square deviations of the second part, M1 is the number of target variables included in the first part, and M2 is the number of target variables included in the second part.

[0020] Furthermore, filtering the target variable includes the following steps:

[0021] The target variables corresponding to the first M influencing factors are combined into an analysis array, two types of target variables are extracted from the analysis array as a third variable and a fourth variable, and the parallelism between the third variable and the fourth variable is calculated based on a third formula, and the third formula is: Wherein, a is the sum of the number of the third variable and the fourth variable whose values ​​are less than the demarcation threshold, b is the sum of the number of the third variable and the fourth variable whose values ​​are greater than or equal to the demarcation threshold, G is the total number of the first variable and the second variable, and ε is the correction value;

[0022] Keep the third variable unchanged, re-extract the target variable from the analysis array as the fourth variable, continue to calculate the parallelism between the third variable and the newly extracted fourth variable, repeat this step until the analysis array is traversed to obtain a numerical set of the third variable, the numerical set including all the parallelisms between the third variable and the remaining target variables in the analysis array, if the parallelisms in the numerical set that are less than the second threshold value exceed a preset number, eliminate the third variable, otherwise retain the third variable.

[0023] Furthermore, constructing the equipment aging model includes the following steps:

[0024] Collect multiple historical data as input data sets, the historical data include operation records in the past time period, set a first time point, a second time point and a third time point as output data sets for each historical data, the period between the first time point and the second time point is a stable period, the period between the second time point and the third time point is a decay period, and the period after the third time point is defined as a failure period, build a neural network model, train the neural network model based on the input data set and the output data set, and use the trained neural network model as the equipment aging model.

[0025] Further, calculating the maintenance cost includes the following steps:

[0026] After inputting the current equipment parameters and the equipment parameters in the optimization scheme into the equipment aging model, the first duration, second duration and third duration of the equipment in the stable period, the decay period and the failure period are obtained, a unit maintenance cost is set for each duration, and the maintenance cost of the optimization scheme is calculated based on the unit maintenance cost and the corresponding duration.

[0027] Furthermore, the preprocessing includes missing value supplementation, duplicate value cleaning and outlier removal.

[0028] The present invention also provides a process flow and equipment energy efficiency operation optimization system based on digital twins, which is used to implement the above-mentioned process flow and equipment energy efficiency operation optimization method based on digital twins, and the system includes:

[0029] A collection module, used for collecting first data of a production process, wherein the first data includes production quality data, equipment parameter data and production link data, and preprocessing the first data to obtain second data;

[0030] An analysis module, configured to analyze the second data to construct a mathematical model of a process flow, wherein the mathematical model includes dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, and the independent variables include equipment parameters;

[0031] A generation module, used to select and set an optimization target from the dependent variables, set a constraint target for equipment energy efficiency, generate an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model, solve the optimization function and the constraint function, and obtain multiple optimization schemes, wherein the optimization schemes include a numerical matrix, and the numerical matrix includes adjustment values ​​of multiple independent variables;

[0032] The accounting module is used to construct an equipment aging model, which is used to simulate the aging process of production equipment under different equipment parameters, extract equipment parameters from the independent variables of the optimization plan, input the equipment parameters into the equipment aging model for simulation operation, and calculate the maintenance cost, select the best plan from the optimization plan based on the maintenance cost, and optimize the process flow based on the best plan.

[0033] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0034] The present invention constructs a mathematical model, namely a digital twin model, based on the production process data, so as to obtain the influence relationship between various production quality data and equipment parameter data and production link data. On this basis, an optimization goal is set, and constraints on equipment energy efficiency are introduced. By jointly solving the optimization function and the constraints, the obtained optimization schemes are all within the preset energy efficiency limits, avoiding the problem of excessive equipment energy efficiency after optimization, which causes a significant increase in production costs. In addition, the present invention also establishes an equipment aging model, and obtains the equipment aging process under each optimization scheme through the equipment aging model, and further calculates the maintenance cost of the optimization scheme based on this, which can help production personnel better select optimization schemes and improve the optimization effect of the process flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flowchart of the steps of the process flow and equipment energy efficiency operation optimization method based on digital twin of the present invention;

[0036] Figure 2 It is a structural schematic diagram of the process flow and equipment energy efficiency operation optimization system based on digital twin of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0038] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0039] like Figure 1 As shown, a method for optimizing process flow and equipment energy efficiency operation based on digital twins includes:

[0040] S1: Collect first data of the production process, the first data including production quality data, equipment parameter data and production link data, pre-process the first data to obtain second data.

[0041] Depending on the application scenario, the specific types of production quality data will be different. For example, in the chemical production scenario, production quality data includes the content of various chemical elements in the product, or various physical properties of the product, such as hardness and stiffness. In the semiconductor production scenario, production quality data includes yield, throughput and thickness uniformity, etc. Equipment parameter data is various setting parameters of production equipment, such as temperature, humidity, vibration frequency, transmission speed, etc. Production link data is the specific process of product production, such as calcination time, grinding fineness, cooling time, etc., as well as the ratio of various chemical components in each production link. After obtaining the first data, data preprocessing is required to remove abnormal data.

[0042] The preprocessing in this embodiment includes missing value supplementation, duplicate value cleaning and outlier removal, which are specific existing technologies and will not be described in detail here.

[0043] S2: Analyze the second data to construct a mathematical model of the process flow, the mathematical model including dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, and the independent variables include equipment parameters.

[0044] Independent variables are variables that can be actively changed by humans, such as equipment parameters, ingredient ratios, etc. Dependent variables are variables that change with the independent variables, such as hardness, thickness uniformity, etc. When constructing a mathematical model, the independent variables are first specified manually, and then the second data is analyzed according to the specified independent variables to find out the dependent variables affected by the independent variables, as well as the influence factors of the independent variables and the dependent variables. The larger the value of the influence factor in this embodiment, the greater the influence of the independent variable on the dependent variable; the independent variable is a variable that is not affected by other parameters.

[0045] S3: Select and set an optimization target from the dependent variables, set a constraint target for the energy efficiency of the equipment, and generate an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model.

[0046] Specifically, firstly, manually select the dependent variable and set the specific value you want the dependent variable to reach. The system uses the dependent variable and the corresponding specific value as the optimization target, such as selecting the chip yield of 30% as the optimization target. Since there are independent variables that affect the dependent variable in the mathematical model, a linear regression function for the dependent variable can be generated according to the independent variable, such as E≤A*W1+B*W2+C*W3, where E is the optimization target, A, B, C are the corresponding independent variables, and W1, W2 and W3 are the corresponding influencing factors; similarly, the upper limit of the equipment energy efficiency, that is, the constraint condition, can be set for a single or multiple production equipment. The equipment energy efficiency calculation function is, for example, E≥[(A*W4+B*W5)*P*T] / R, where P is the rated power, W4 and W5 are the weights of the independent variables A and B on the rated power, which can be obtained through testing or product manuals, T is the operating time, and R is the number of products produced within the operating time. The above function is just an example. In practical applications, more complex energy efficiency evaluation methods proposed in the prior art can be used, which will not be introduced in detail here.

[0047] S4: Solve the optimization function and the constraint function to obtain multiple optimization solutions, where the optimization solutions include a numerical matrix, and the numerical matrix includes adjustment values ​​of multiple independent variables.

[0048] After obtaining the optimization function and constraint function, the particle swarm algorithm or Monte Carlo method can be used to solve various A, B, and C parameter combinations that satisfy the two functions. The parameter combinations are presented in the form of numerical matrices. The A, B, and C values ​​in each parameter combination are the specific values ​​of the independent variables that need to be adjusted. Since there are multiple parameter combinations, there will be multiple optimization schemes, and the obtained optimization schemes will be within the pre-set energy efficiency requirements, thereby avoiding the problem of excessive equipment energy efficiency after optimization.

[0049] S5: Construct an equipment aging model. The equipment aging model is used to simulate the aging process of production equipment under different equipment parameters. The equipment parameters are extracted from the independent variables of the optimization scheme, and the equipment parameters are input into the equipment aging model for simulation operation and maintenance cost calculation.

[0050] The equipment aging model is built based on a neural network. The specific construction method will be introduced later. Changing equipment parameters will generally speed up or slow down the aging rate of the equipment. For example, increasing the motor speed will speed up the wear of the motor drive shaft, which will shorten the motor life and accelerate the aging of the motor. Therefore, after changing the equipment parameters, the changed equipment parameters are input into the equipment aging model for simulation to obtain the aging process of the equipment after adjusting the parameters. Finally, the maintenance cost of the equipment is calculated based on the aging process.

[0051] S6: Select the best solution from the optimization solutions based on the maintenance cost, and optimize the process flow based on the best solution.

[0052] After calculating the maintenance cost of each optimization plan, it can be presented to production personnel, who can select the appropriate optimization plan as the best plan based on the size of the maintenance cost. Selection rules can also be set, such as the minimum maintenance cost rule, by selecting the optimization plan corresponding to the minimum maintenance cost as the best plan.

[0053] The present invention first constructs a mathematical model, namely a digital twin model, based on the collected production process data, so as to obtain the influence relationship between various production quality data and equipment parameter data and production link data. On this basis, an optimization goal is set, and constraints on equipment energy efficiency are introduced. By jointly solving the optimization function and the constraints, the obtained optimization schemes are all within the preset energy efficiency limits, avoiding the problem of excessive equipment energy efficiency after optimization, which causes a significant increase in production costs. In addition, the present invention also establishes an equipment aging model, and obtains the equipment aging process under each optimization scheme through the equipment aging model, and further calculates the maintenance cost of the optimization scheme based on this, which can help production personnel better select optimization schemes and improve the optimization effect of the process flow.

[0054] It is particularly noteworthy that, through the present invention, the process flow can be optimized while controlling the energy efficiency of the equipment, and the optimization cost can also be calculated to achieve comprehensive optimization of the process flow and the energy efficiency of the equipment.

[0055] In this embodiment, analyzing the second data includes the following steps:

[0056] The second data is divided into the first variable and the second variable, one first variable is extracted as the analysis variable, one second variable is extracted as the target variable, the demarcation threshold is determined based on the target variable, the values ​​of the target variable that are less than the demarcation threshold are defined as the first values, and the rest are defined as the second values, the analysis variable corresponding to the first value is the third value, and the analysis variable corresponding to the second value is the fourth value, and the influence factor L is calculated based on the first formula, which is: Among them, P1 is the average value of the first value, P2 is the average value of the second value, P3 is the average value of the third value, and P4 is the average value of the fourth value.

[0057] The first variable and the second variable are potential dependent variables and independent variables. Specifically, the first variable and the second variable are first specified manually, and then the specified first variable and the second variable are screened by the method of this embodiment to obtain the first variable and the second variable that do affect each other as the independent variable and the dependent variable. Specifically, after dividing the first variable and the second variable, a type of first variable is selected as the analysis variable 1, and a type of second variable is randomly selected as the target variable 1. The first variable and the second variable are a data type. For example, the first variable is current, which has different values ​​at multiple time points; a demarcation threshold is set according to the target variable 1, and a value less than the demarcation threshold is set as the first value, and a value greater than or equal to the demarcation threshold is set as the second value.

[0058] Then verify whether there is any mutual influence between analysis variable 1 and target variable 1. Specifically, obtain the third value that appears at the same time as the first value and the fourth value that appears at the same time as the second value. The third value and the fourth value are both values ​​of analysis variable 1. Then calculate the average value of each of the first to fourth values ​​and substitute them into the first formula for calculation, so as to obtain the influence factor of analysis variable 1 and target variable 1. Through a reasonable demarcation threshold, target variable 1 can be divided into a part with a larger value and a part with a smaller value. If there is a mutual influence between analysis variable 1 and target variable 1, the corresponding target variable 1 will also be divided into a part with a larger value and a part with a smaller value. Finally, the influence of target variable 1 on analysis variable 1 can be obtained through the first formula. For example, if the denominator in the first formula is 10 and the numerator is 5, the influence factor between analysis variable 1 and target variable 1 is calculated to be 0.5.

[0059] Repeat this step to traverse all second variables, sort all the influencing factors obtained, filter the target variables corresponding to the first M influencing factors, filter the filtered target variables, and use the filtered target variables as the independent variables of the analysis variables, and the corresponding analysis variables as the dependent variables. If there is no independent variable for the analysis variable, use the analysis variable as the independent variable.

[0060] For the same analysis variable 1, use the above steps to calculate the influencing factors of all other second variables 2 to second variables 15 and analysis variable 1, and then sort the influencing factors from large to small. For example, M is 6, and the first 6 influencing factors after sorting are selected. Assuming that there will be target variables 1, 5, 6, 8, 11, and 23 corresponding to them, after further screening and removing these 6 target variables, the remaining target variables 1, 5, and 6 are used as independent variables of analysis variable 1, and analysis variable 1 is the dependent variable. The analysis process of the remaining first variables is similar, and will not be repeated here.

[0061] This embodiment determines the demarcation threshold based on the target variable, including the following steps:

[0062] Set a progressive value, increase the progressive value of the target variable with the smallest value to obtain an advanced value, use the advanced value to divide the target variable into the first part and the second part, calculate the evaluation value of the advanced value, the evaluation value is used to evaluate the division effect of the advanced value, increase the progressive value to obtain a new advanced value, recalculate the evaluation value based on the new advanced value, repeat this step until the advanced value exceeds the maximum value of the second variable, and use the advanced value corresponding to the maximum evaluation value as the demarcation threshold.

[0063] Specifically, calculating the evaluation value includes the following steps:

[0064] Set a second formula, and calculate the evaluation value R based on the second formula. The second formula is: Among them, V1 is the average value of the first part after division, V2 is the average value of the second part after division, U1 is the sum of square deviations of the first part, U2 is the sum of square deviations of the second part, M1 is the number of target variables included in the first part, and M2 is the number of target variables included in the second part.

[0065] The progressive value is a fixed value set in advance. When determining the demarcation threshold, first locate the target variable with the smallest value, which is defined as the basic value here. Add the basic value and the progressive value to obtain the advanced value. Set the target variable that is less than the advanced value as the first part, and the target variable that is greater than the advanced value as the second part. Here, the values ​​of a type of target variable appearing at multiple time points are divided; then use the second formula to calculate the evaluation value of the corresponding advanced value. The following example introduces the second formula. Assuming that the numerical distribution of the target variable is bipolar, that is, some values ​​are relatively small and some values ​​are relatively large, if a suitable advanced value is selected, that is, the advanced value can be better divided into the first part and the second part with smaller and larger values, then it can be known that the numerator value in the second formula is relatively large, and in the denominator part, the sum of the squares of the deviations of the first part and the second part is relatively small, which will eventually lead to a large calculation result of the second formula, so that a higher evaluation value will be obtained; on the contrary, if the advanced value is unreasonable, a smaller evaluation value will be obtained. In this way, by continuously increasing the advanced value and calculating the evaluation value of the advanced value after each increase, the most appropriate advanced value can be finally obtained as the demarcation threshold.

[0066] In this embodiment, filtering the target variable includes the following steps:

[0067] The target variables corresponding to the first M influencing factors are combined into an analysis array, and two types of target variables are extracted from the analysis array as the third variable and the fourth variable. The parallelism between the third variable and the fourth variable is calculated based on the third formula. The third formula is: Among them, a is the sum of the number of third variables and fourth variables whose values ​​are less than the demarcation threshold, b is the sum of the number of third variables and fourth variables whose values ​​are greater than or equal to the demarcation threshold, G is the total number of first variables and second variables, and ε is the correction value.

[0068] The third variable and the fourth variable are two different types of variables, such as voltage and current. Then, count the number 1 of the third variable whose value is less than the demarcation threshold, count the number 2 of the fourth variable whose value is less than the demarcation threshold, and add the number 1 and the number 2 to obtain a. Count the number 3 of the third variable whose value is greater than or equal to the demarcation threshold, and count the number 4 of the fourth variable whose value is greater than or equal to the demarcation threshold, and add the number 3 and the number 4 to obtain b. In this embodiment, the correction value is set to 1. The closer the change law of the third variable is to that of the fourth variable, the closer the numerical distribution after division by the critical threshold is, and the smaller the calculation result substituted into the third formula. On the contrary, if the third variable has no correlation with the fourth variable, a larger value will be obtained through the third formula.

[0069] Keep the third variable unchanged, re-extract the target variable from the analysis array as the fourth variable, continue to calculate the parallelism between the third variable and the newly extracted fourth variable, repeat this step until the analysis array is traversed to obtain a numerical set of the third variable, the numerical set includes all parallelisms of the third variable and the remaining target variables in the analysis array, if the parallelism in the numerical set that is less than the second threshold exceeds a preset number, the third variable is eliminated, otherwise the third variable is retained.

[0070] Then continue to extract the remaining target variables from the analysis array as the fourth variable, calculate the parallelism with the previously extracted third variable, and after completing the parallelism calculation of the third variable and all other target variables in the analysis array, combine the obtained parallelism into a numerical set, and count how many types of target variables in the analysis array are correlated with the third variable based on the numerical set. If there is a correlation with more target variables (that is, the number of parallelisms less than the second threshold in the numerical set exceeds the preset number), it indicates that the third variable will be affected by multiple target variables while affecting the analysis variable. If this third variable is retained, it will greatly increase the complexity of subsequent analysis. Therefore, from the perspective of simplifying the computational complexity, this third variable is eliminated.

[0071] Building a device aging model involves the following steps:

[0072] Collect multiple historical data as input data sets, the historical data include operation records in the past time period, set the first time point, the second time point and the third time point as output data sets for each historical data, the period between the first time point and the second time point is the stable period, the period between the second time point and the third time point is the decay period, and the period after the third time point is defined as the failure period, build a neural network model, train the neural network model based on the input data set and the output data set, and use the trained neural network model as the equipment aging model.

[0073] Specifically, historical data of the same type of equipment running in the past is collected, which can be obtained from multiple factories or equipment providers. The historical data includes the setting parameters, operating time, fault records, etc. of the equipment, and then the first time point, the second time point and the third time point are marked for each piece of historical data. They can be marked manually or automatically according to the equipment aging assessment rules. The neural network model constructed in this embodiment is a fully connected single hidden layer network, in which the number of input layer neurons is determined according to the number of data types in the historical data, the number of hidden neurons is set to 8, the number of output layer neurons is set to 3, the activation function uses the sigmoid function, and the loss function uses the cross entropy loss function. Then, the input training set and the output training set are used to construct a numerical matrix, and the numerical matrix is ​​input into the neural network model for training. When the mean square error MSE of the neural network model after training meets the requirements, it stops, indicating that the model training is completed. The specific function form and training method are prior art and will not be introduced here.

[0074] The maintenance cost calculation in this embodiment includes the following steps:

[0075] After the current equipment parameters and the equipment parameters in the optimization scheme are input into the equipment aging model, the first duration, second duration and third duration of the equipment in the stable period, decay period and failure period are obtained, and the unit maintenance cost is set for each duration. The maintenance cost of the optimization scheme is calculated based on the unit maintenance cost and the corresponding duration.

[0076] After converting the parameters currently being used by the equipment and the equipment parameters in the optimization scheme into the correct input dimension, they are input into the equipment aging model to predict the occurrence time of the first time point, the second time point, and the third time point. Then, the duration of the stable period, the decay period, and the failure period can be obtained. According to the previously set unit maintenance cost, the unit maintenance cost is multiplied by the corresponding duration to obtain the maintenance cost. The unit maintenance cost can be set based on experience.

[0077] like Figure 2 As shown, the present invention also provides a process flow and equipment energy efficiency operation optimization system based on digital twins, which is used to implement the above-mentioned process flow and equipment energy efficiency operation optimization method based on digital twins, and the system includes:

[0078] A collection module is used to collect first data of the production process, the first data including production quality data, equipment parameter data and production link data, and pre-process the first data to obtain second data;

[0079] An analysis module, used for analyzing the second data to construct a mathematical model of the process flow, the mathematical model including dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, the independent variables including equipment parameters;

[0080] A generation module, used to select and set an optimization target from dependent variables, set a constraint target for equipment energy efficiency, generate an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model, solve the optimization function and the constraint function, and obtain multiple optimization schemes, wherein the optimization scheme includes a numerical matrix, and the numerical matrix includes adjustment values ​​of multiple independent variables;

[0081] The accounting module is used to build an equipment aging model. The equipment aging model is used to simulate the aging process of production equipment under different equipment parameters, extract equipment parameters from the independent variables of the optimization plan, input the equipment parameters into the equipment aging model for simulation operation, and calculate the maintenance cost. The best plan is selected from the optimization plan based on the maintenance cost, and the process flow is optimized based on the best plan.

[0082] It should be understood that the various technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing process flow and equipment energy efficiency based on digital twins, characterized in that: include: Collecting first data of a production process, the first data including production quality data, equipment parameter data and production link data, and preprocessing the first data to obtain second data; Analyzing the second data to construct a mathematical model of the process flow, the mathematical model comprising dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, the independent variables comprising equipment parameters; Selecting and setting an optimization target from the dependent variables, setting a constraint target for equipment energy efficiency, and generating an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model; Solving the optimization function and the constraint function to obtain a plurality of optimization schemes, wherein the optimization schemes include a numerical matrix including adjustment values ​​of a plurality of the independent variables; Constructing an equipment aging model, the equipment aging model is used to simulate the aging process of production equipment under different equipment parameters, extracting equipment parameters from the independent variables of the optimization scheme, inputting the equipment parameters into the equipment aging model for simulation operation, and calculating maintenance costs; An optimal solution is selected from the optimization solutions based on the maintenance cost, and the process flow is optimized based on the optimal solution.

2. The method according to claim 1, characterized in that Analyzing the second data comprises the following steps: The second data is divided into a first variable and a second variable, one of the first variables is extracted as an analysis variable, and one of the second variables is extracted as a target variable, a demarcation threshold is determined based on the target variable, and the values ​​of the target variable that are less than the demarcation threshold are defined as first values, and the rest are defined as second values, the analysis variable corresponding to the first value is a third value, and the analysis variable corresponding to the second value is a fourth value, and the influence factor L is calculated based on a first formula, and the first formula is: Wherein, P1 is the average value of the first values, P2 is the average value of the second values, P3 is the average value of the third values, and P4 is the average value of the fourth values; Repeat this step to traverse all the second variables, sort all the influencing factors obtained, screen the target variables corresponding to the first M influencing factors, filter the screened target variables, and use the filtered target variables as the independent variables of the analysis variables, and the corresponding analysis variables as the dependent variables. If the independent variable does not exist for the analysis variable, use the analysis variable as the independent variable.

3. The method according to claim 2, characterized in that Determining the demarcation threshold based on the target variable comprises the following steps: A progressive value is set, and an advanced value is obtained by increasing the target variable with the smallest value by the progressive value, and the target variable is divided into a first part and a second part using the advanced value, and an evaluation value of the advanced value is calculated, and the evaluation value is used to evaluate the division effect of the advanced value, and the progressive value is increased to obtain a new advanced value, and the evaluation value is recalculated based on the new advanced value, and this step is repeated until the advanced value exceeds the maximum value of the second variable, and the advanced value corresponding to the maximum evaluation value is used as the demarcation threshold.

4. The method according to claim 3, characterized in that Calculating the evaluation value includes the following steps: A second formula is set, and the evaluation value R is calculated based on the second formula, where the second formula is: Among them, V1 is the average value of the first part after division, V2 is the average value of the second part after division, U1 is the sum of square deviations of the first part, U2 is the sum of square deviations of the second part, M1 is the number of target variables included in the first part, and M2 is the number of target variables included in the second part.

5. The method according to claim 2, characterized in that: Filtering the target variable includes the following steps: The target variables corresponding to the first M influencing factors are combined into an analysis array, two types of target variables are extracted from the analysis array as a third variable and a fourth variable, and the parallelism between the third variable and the fourth variable is calculated based on a third formula, and the third formula is: Wherein, a is the sum of the number of the third variable and the fourth variable whose values ​​are less than the demarcation threshold, b is the sum of the number of the third variable and the fourth variable whose values ​​are greater than or equal to the demarcation threshold, G is the total number of the first variable and the second variable, and ε is the correction value; Keep the third variable unchanged, re-extract the target variable from the analysis array as the fourth variable, continue to calculate the parallelism between the third variable and the newly extracted fourth variable, repeat this step until the analysis array is traversed to obtain a numerical set of the third variable, the numerical set including all the parallelisms between the third variable and the remaining target variables in the analysis array, if the parallelisms in the numerical set that are less than the second threshold value exceed a preset number, eliminate the third variable, otherwise retain the third variable.

6. The method according to claim 4, characterized in that Constructing the equipment aging model includes the following steps: Collect multiple historical data as input data sets, the historical data include operation records in the past time period, set a first time point, a second time point and a third time point as output data sets for each historical data, the period between the first time point and the second time point is a stable period, the period between the second time point and the third time point is a decay period, and the period after the third time point is defined as a failure period, build a neural network model, train the neural network model based on the input data set and the output data set, and use the trained neural network model as the equipment aging model.

7. The method according to claim 6, characterized in that Calculating the maintenance cost includes the following steps: After inputting the current equipment parameters and the equipment parameters in the optimization scheme into the equipment aging model, the first duration, the second duration and the third duration of the equipment in the stable period, the decay period and the failure period are obtained, a unit maintenance cost is set for each duration, and the maintenance cost of the optimization scheme is calculated based on the unit maintenance cost and the corresponding duration.

8. The method according to claim 1, characterized in that The preprocessing includes missing value supplementation, duplicate value cleaning and outlier removal.

9. A process flow and equipment energy efficiency operation optimization system based on digital twin, used to implement the method described in any one of claims 1 to 8 above, the system comprising: A collection module, used for collecting first data of a production process, wherein the first data includes production quality data, equipment parameter data and production link data, and preprocessing the first data to obtain second data; An analysis module, configured to analyze the second data to construct a mathematical model of a process flow, wherein the mathematical model includes dependent variables, independent variables and independent variables, and influencing factors between the dependent variables and the independent variables, and the independent variables include equipment parameters; A generation module, used to select and set an optimization target from the dependent variables, set a constraint target for equipment energy efficiency, generate an optimization function and a constraint function based on the optimization target, the constraint target and the mathematical model, solve the optimization function and the constraint function, and obtain multiple optimization schemes, wherein the optimization schemes include a numerical matrix, and the numerical matrix includes adjustment values ​​of multiple independent variables; The accounting module is used to construct an equipment aging model, which is used to simulate the aging process of production equipment under different equipment parameters, extract equipment parameters from the independent variables of the optimization plan, input the equipment parameters into the equipment aging model for simulation operation, and calculate the maintenance cost, select the best plan from the optimization plan based on the maintenance cost, and optimize the process flow based on the best plan.

Images