Separation principle intelligent recommendation method based on random forest

Abstract

The invention discloses a separation principle intelligent recommendation method based on a random forest, and the method comprises the steps: training a relation between a physical quantity parameter describing a physical conflict in a previous case and a separation principle through the random forest, and building the relation between the physical quantity parameter and the separation principle; when a new system is designed, physical quantity parameters describing physical conflicts are input into a trained mapping model to intelligently recommend a separation principle, and the system is improved in a concept design stage according to the separation principle obtained through recommendation to counteract the influence brought by the physical conflicts, so that it is ensured that the concept design can be normally and stably operated subsequently; according to the intelligent recommendation separation principle of the method, the design knowledge retrieval efficiency and accuracy can be effectively improved, designers can quickly and accurately obtain the corresponding separation principle conveniently, the physical conflict solving efficiency is improved, and the method has important significance in improving the design efficiency of a system and reducing system problems caused by physical conflicts.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent recommendation of mechanical product design technical knowledge, and in particular to an intelligent recommendation method based on the separation principle of random forests. Background Art

[0002] In the process of conceptual design, most current systems are designed based on user needs. However, technical systems have complex characteristics such as openness and time-varying. As the complexity of technical systems increases, the coupling and interaction between subsystems within the system become more complex, and the conflicts generated in the product design process also increase. Among them, physical conflicts are more acute than technical conflicts. If they cannot be discovered and eliminated in time, they often lead to the failure of the system to achieve the expected results or operation failures in the actual application process, resulting in multiple subsequent iterative designs. Therefore, these factors need to be comprehensively considered in the conceptual design, and improved designs should be made to ensure the actual application effect of the system.

[0003] Physical conflict problems in technical systems can be expressed using terms such as physical quantity parameters in physics. The important research method for solving physical conflict problems in TRIZ theory is that designers select and use corresponding separation principles to solve problems by determining engineering parameters. However, the conceptual design stage has limited information, uncertainty, and multidisciplinary nature, which makes the determination of separation principles difficult, subjective, and inefficient. Due to the constraints of design knowledge and experience, it is difficult for designers to directly and objectively analyze the relationship between parameter effects and separation principles to select separation principles. In the process of physical conflict resolution, separation principles are often selected based on experience, which affects the resolution of conflicts, and then has an adverse impact on the system, affecting the reliability and safety of the system, making the conflict resolution process less efficient and the resulting solutions less effective.

[0004] Therefore, it becomes an important issue how to effectively assist designers in analyzing the conflict problems existing in the system at the parameter level during the conceptual design stage, and then objectively and accurately select the separation principle to improve the design of the system according to the physical quantity parameters to reduce the adverse effects of the physical conflict problems on the system.

[0005] To solve the above problems, the relationship between physical quantity parameters and separation principles is considered, and separation principles are intelligently recommended according to physical quantity parameters, so that designers can objectively and accurately select and improve the system according to the recommended separation principles to offset the impact of physical conflicts. Therefore, designing a separation principle intelligent recommendation method based on random forest is of great significance to improve the efficiency and accuracy of designers' separation principle selection and reduce system problems caused by physical conflicts. Summary of the invention

[0006] In view of this, the present invention provides a separation principle intelligent recommendation method based on random forest. The relationship between physical quantity parameters and separation principles in previous cases is trained through random forest. When a new system is designed, the separation principle is intelligently recommended through the trained mapping model. In the conceptual design stage, the influence of physical conflicts is offset by the heuristic design of separation principle selection and physical conflict resolution through physical quantity parameters, thereby ensuring that the conceptual design can run normally and stably afterwards, which is of great significance to improving the design efficiency of the system and reducing system problems caused by physical conflicts.

[0007] The technical solution for implementing the present invention is as follows:

[0008] Step 1: Analysis of the Engineering Case Collection

[0009] Obtain engineering cases from patent data sets and innovative design case collections, analyze them before and after optimization, extract physical quantity parameters of system interactions, and determine the separation principles used for system improvement or problem solving through judgment;

[0010] Step 2: Generation of input and output codes.

[0011] Standardize and assign values ​​to the analysis results. Encode the input physical quantity parameters through LT-KC, and encode the separation principle classification corresponding to the output of problem solving and system improvement in turn;

[0012] Step 3: Construction of parameter effect and separation principle mapping model.

[0013] By using the random forest algorithm, the processed sample data set is trained and a mapping model between the interaction of physical quantity parameters and the separation principle is constructed to recommend the separation principle.

[0014] Step 4: Intelligent recommendation based on the separation principle of the mapping model

[0015] By inputting the physical quantity parameters obtained by analyzing the system to be designed into the constructed mapping model based on random forest, the recommended separation principle is obtained, and the improved design is carried out accordingly.

[0016] Further, step 2-1 includes the following steps:

[0017] Step 11: Determination of the project case set

[0018] A sample dataset is constructed by combining patent datasets and innovative design engineering case collections.

[0019] Step 12: Obtaining sample data

[0020] The learning samples need to include the correspondence between input and output, and the engineering cases in the case set need to be analyzed to determine the parameters that cause the problem and the separation principles used to solve the problem.

[0021] Furthermore, step 2 includes the following steps:

[0022] Step 21: Determine the encoding rules for input physical quantity parameters

[0023] The LT-KC encoding of possible physical conflict parameters in the system working process of the relevant case set is used as input. The input of random forest is a set of physical quantity parameters, and the number of input physical quantity parameters is set to 3. When the system physical quantity parameters need both positive and negative values, when there is a physical conflict in the system, the physical quantity parameters describing the physical conflict and the deteriorated physical quantity parameters and improved physical quantity parameters affected by the conflict parameters are encoded according to the encoding rules respectively;

[0024] Step 22: Determination of output encoding rules

[0025] The separation principle coding of the relevant case set is used as output. First, the physical conflicts existing in the original system of the case are determined, and then the solution or improvement method of the case for the existing problem is analyzed, and then the separation principle used is determined and determined as the output result of the case data.

[0026] Step 3: Construction of parameter effect and separation principle mapping model.

[0027] By using the random forest algorithm, the processed sample data set is trained and a mapping model between the interaction of physical quantity parameters and the separation principle is constructed to recommend the separation principle.

[0028] Furthermore, step three includes the following steps:

[0029] Step 31: Data Preprocessing

[0030] Clean the data, including processing missing values ​​and outliers.

[0031] Step 32: Data Segmentation

[0032] Split the dataset into training and testing sets

[0033] Step 33: Initialization

[0034] Configuration parameters: Set the model's hyperparameters, such as the number of trees, the maximum depth of the trees, etc.

[0035] Step 34: Train the model

[0036] Train the model using the training set data.

[0037] Step 35: Model Evaluation

[0038] Use the test set to evaluate the model performance using appropriate evaluation metrics (such as precision, recall, F1 score, MSE, RMSE, etc.).

[0039] Step 36: Parameter Tuning

[0040] Adjust the model parameters according to the evaluation results until the stopping conditions are met;

[0041] Step 37: Model Validation

[0042] Cross-validation and other techniques are used to further verify the stability and performance of the model.

[0043] Step 38: Model Optimization:

[0044] Further optimize the model according to the verification results until the stopping condition is reached (such as reaching the maximum number of iterations or the fitness meets the requirements);

[0045] Step 39: Output

[0046] Output the optimized model for prediction tasks.

[0047] Furthermore, step 4 includes the following steps:

[0048] Step 41: Determine the system to be designed;

[0049] Step 42 Step 41: Determine the system to be designed;

[0050] Step 42: Analyze and determine physical quantity parameters that may cause physical conflict problems during the operation of the system;

[0051] Step 43: converting the physical quantity parameters determined by the analysis into the physical quantity coding form of LT-KC;

[0052] Step 44: Input the physical quantity code into the physical quantity parameter and separation principle mapping model built based on random forest;

[0053] Step 45: Obtain corresponding separation principle recommendation through the mapping model;

[0054] Step 46: Determine the system optimization design scheme based on the recommended separation principle.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] (1) The method of the present invention obtains a mapping model between physical quantity parameters and separation principles through random forest training, and improves the conceptual design scheme according to the separation principle recommended by the model, effectively improving the efficiency and accuracy of design knowledge retrieval, facilitating designers to quickly and accurately obtain the corresponding separation principle, and reducing the subjectivity and blindness of the improved design under the influence of physical conflict problems;

[0057] (2) The method of the present invention introduces a random forest algorithm to train the relationship between physical quantity parameters and separation principles, thereby improving the accuracy and stability of the model;

[0058] (3) The method of the present invention can effectively improve the efficiency and accuracy of design knowledge retrieval, allowing designers to devote more energy to product design rather than knowledge retrieval;

[0059] (4) The method of the present invention selects the separation principle according to the solution to the conflict problem. According to the recommended separation principle, an improved conceptual design scheme can be obtained to reduce the occurrence of physical conflict problems in advance, thereby eliminating system problems caused by physical conflict problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a process model diagram of an intelligent recommendation method based on the separation principle of random forests;

[0061] Figure 2 Training input groups for physical quantity parameters;

[0062] Figure 3 Encoding rules for physical quantity parameters;

[0063] Figure 4 Output encoding rules;

[0064] Figure 5 Flowchart of random forest model training;

[0065] Figure 6 To predict interfaces for separation principles;

[0066] Figure 7 Output of separation principle of the target system of the embodiment of the present invention;

[0067] Figure 8 This is a schematic diagram of a blade rotating power generation state in a first embodiment of the present invention;

[0068] Fig. 9 This is a schematic diagram of the vibration power generation state of the first embodiment of the present invention;

[0069] Fig.10 This is a diagram of the start-up mode of a resistance-type fan according to scheme 2 of the embodiment of the present invention;

[0070] Fig.11This is a diagram of a lift-type fan mode according to scheme 2 of the embodiment of the present invention;

[0071] Fig.12 This is a diagram of the resistance-type fan mode when the wind speed is relatively high in the second embodiment of the present invention. DETAILED DESCRIPTION

[0072] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0073] Specific implementation method one: This implementation method is an intelligent recommendation method for separation principles based on random forests. This method uses random forests to train the relationship between physical quantity parameters and separation principles in previous cases that have physical conflicts, and establishes the relationship between physical quantity parameters and separation principles. When designing a new system, the separation principle is intelligently recommended by inputting the physical quantity parameters that have physical conflicts into the trained mapping model. According to the recommended separation principle, the system is improved in the conceptual design stage to offset the impact of the physical conflict, thereby ensuring that the conceptual design can run normally and stably in the future. Intelligent recommendation of separation principles according to this method can effectively improve the efficiency and accuracy of design knowledge retrieval, facilitate designers to quickly and accurately obtain the corresponding separation principles, improve the efficiency of physical conflict resolution, and is of great significance to improving the design efficiency of the system and reducing system problems caused by physical conflicts.

[0074] Taking the vertical axis fan as an example, the separation principle intelligent recommendation method based on random forest proposed in the present invention is explained. The vertical axis fan is a device that uses wind energy to generate electricity. Compared with the traditional horizontal axis fan, the vertical axis fan has the advantages of simple structure, strong wind direction adaptability, low noise, etc., and has gradually attracted attention and application. Before applying the mapping model intelligent recommendation separation principle to design the vertical axis fan, it is necessary to first construct a parameter action and separation principle mapping model, including the following steps:

[0075] Step 1: Analysis of the Engineering Case Collection

[0076] Obtain engineering cases from patent data sets and innovative design case collections, analyze them before and after optimization, extract physical quantity parameters of system interactions, and determine the separation principles used for system improvement or problem solving through judgment;

[0077] Step 1 specifically includes the following steps:

[0078] Step 11: Determination of the project case set

[0079] The sample data set was constructed by combining the patent data set and the innovative design engineering case set. The patent data set was retrieved using the PatSnap patent search platform and the IHS Goldfire knowledge base; the innovative design engineering case set was extracted from engineering cases in textbooks such as "Application Cases of Innovative Methods for Frontline Engineers", "Innovative Design Cases for Manufacturing", and "Mechanical Innovative Design Cases and Comments".

[0080] Step 12: Obtaining sample data

[0081] The learning samples need to include the correspondence between input and output, and the engineering cases in the case set need to be analyzed to determine the parameters that cause the problem and the separation principles used to solve the problem.

[0082] Step 2: Generation of input and output codes.

[0083] Standardize and assign values ​​to the analysis results. Encode the input physical quantity parameters through LT-KC, and encode the separation principle classification corresponding to the output of problem solving and system improvement in turn;

[0084] Step 2 specifically includes the following steps:

[0085] Step 21: Determine the encoding rules for input physical quantity parameters

[0086] The LT-KC encoding of possible physical conflict parameters in the system working process of the relevant case set is used as input. The input of random forest is a set of physical quantity parameters, and the number of input physical quantity parameters is set to 3. When the system physical quantity parameters need both positive and negative values, when there is a physical conflict in the system, the physical quantity parameters describing the physical conflict and the deteriorated physical quantity parameters and improved physical quantity parameters affected by the conflict parameters are encoded according to the encoding rules respectively;

[0087] Step 22: Determination of output encoding rules

[0088] The separation principle coding of the relevant case set is used as output. First, the physical conflicts existing in the original system of the case are determined, and then the solution or improvement method of the case for the existing problem is analyzed, and then the separation principle used is determined and determined as the output result of the case data.

[0089] Step 3: Construction of parameter effect and separation principle mapping model.

[0090] By using the random forest algorithm, the processed sample data set is trained and a mapping model between the interaction of physical quantity parameters and the separation principle is constructed to recommend the separation principle.

[0091] Step three specifically includes the following steps: Figure 5 As shown:

[0092] Step 31: Data Preprocessing

[0093] Clean the data, including processing missing values ​​and outliers.

[0094] Step 32: Data Segmentation

[0095] Split the dataset into training and testing sets

[0096] Step 33: Initialization

[0097] Configuration parameters: Set the model's hyperparameters, such as the number of trees, the maximum depth of the trees, etc.

[0098] Step 34: Train the model

[0099] Train the model using the training set data.

[0100] Step 35: Model Evaluation

[0101] Use the test set to evaluate the model performance using appropriate evaluation metrics (such as precision, recall, F1 score, MSE, RMSE, etc.).

[0102] Step 36: Parameter Tuning

[0103] Adjust the model parameters according to the evaluation results until the stopping conditions are met;

[0104] Step 37: Model Validation

[0105] Cross-validation and other techniques are used to further verify the stability and performance of the model.

[0106] Step 38: Model Optimization:

[0107] Further optimize the model according to the verification results until the stopping condition is reached (such as reaching the maximum number of iterations or the fitness meets the requirements);

[0108] Step 39: Output

[0109] Output the optimized model for prediction tasks.

[0110] The following takes the design of a vertical axis fan assisted by the separation principle based on random forest intelligent recommendation as an example, that is, the relevant content of step 4, to provide a specific embodiment of the present invention;

[0111] Vertical axis wind turbine is a kind of equipment that uses wind energy to generate electricity. Compared with traditional horizontal axis wind turbine, vertical axis wind turbine has the advantages of simple structure, strong adaptability to wind direction and low noise, and is gradually attracting attention and application.

[0112] Step 4: Intelligent recommendation based on the separation principle of the mapping model

[0113] By inputting the physical quantity parameters obtained by analyzing the system to be designed into the constructed mapping model based on random forest, the recommended separation principle is obtained, and the improved design is carried out accordingly.

[0114] The design steps of a vertical axis fan are as follows:

[0115] Step 41: Determine the system to be designed;

[0116] Based on the above background, the research on vertical axis fans is promoted to discover and solve possible conflict problems, promote the use of clean energy, improve energy efficiency, and promote sustainable development. The system to be designed is determined to be a vertical axis fan system.

[0117] Step 42: Analyze and determine physical quantity parameters that may cause physical conflict problems during the operation of the system;

[0118] When a vertical axis wind turbine is working, in a breeze, it is hoped that the blade area is large enough to meet the output power of the wind turbine to improve the utilization rate of wind energy. When the wind speed exceeds the rated wind speed range of the wind turbine, the wind turbine may be at risk of overload. In order to protect the wind turbine from damage, it is hoped that the blade area is small enough, otherwise the wind turbine needs to automatically shut down to reduce the force on the blades and expand the wind speed threshold for wind turbine operation.

[0119] Step 43: converting the physical quantity parameters determined by the analysis into the physical quantity coding form of LT-KC;

[0120] The physical quantity parameters related to the system variation caused by the LT matrix logical multiplication are encoded according to the encoding rules. The input physical quantity parameters are area, power and speed. The L-bit and T-bit encodings are encoded according to the exponent of the LT dimension of each physical quantity parameter. According to the physical quantity parameter dimension query system, the power and speed belong to the mechanics field, and the K-bit encoding is 1; the area belongs to the space and time field, and the K-bit encoding is 0. According to the current state value, the C-bit encoding of area, power, and speed is -1.

[0121] Step 44: Input the physical quantity code into the physical quantity parameter and separation principle mapping model built based on random forest;

[0122] The physical quantity parameters blade area, power and speed related to the conflict problem are input into the pre-trained mapping model according to the encoding rules.

[0123] Step 45: Obtain corresponding separation principle recommendation through the mapping model;

[0124] The separation principle recommended by them is time separation.

[0125] Step 46: Determine the system optimization design scheme based on the recommended separation principle.

[0126] According to the output rule interpretation of the mapping model, the vertical axis wind turbine system is improved by using time separation. According to the recommended separation principle, the vertical axis wind turbine system is improved, and the target system of the improvement is the blades of the vertical axis wind turbine. The following two feasible solutions are determined.

[0127] Option 1:

[0128] Design a special blade contraction structure such as Figure 8 As shown in the figure, the blade support rods can be retracted. The support rods A and B are hinged to the fan support column and the inner side of the blade respectively. The support rods C and D are hinged to the fan support column and the inner side of the blade respectively. When the support rods C and D slide to the lowest end, A and B overlap, and C and D overlap, locking the connection structures AB and CD. At this time, the vertical axis fan is as follows Fig. 9 At this time, the vertical axis fan is a cylinder, which can use the Karman vortex street effect to achieve vibration power generation under the impact of wind to prevent excessive wind force from causing damage to the blades.

[0129] The wind speed is detected by the wind speed sensor. When the wind speed is normal, the normal blade rotation power generation state is adopted. When the wind speed exceeds a certain threshold, in order to protect the vertical wind turbine from damage, the vibration power generation state is adopted. The power generation state is switched according to different wind speeds.

[0130] Option 2:

[0131] A passive variable pitch mechanism based on a combination of gears and connecting rods is designed to achieve passive pitch change of wind turbine blades during rotation. The central gear is located on the support shaft, and the circumferential gear is located around the central gear. The rotation center of the circumferential gear and the offset hole on it that is concentric with the center of the circle constitute a crank. The component that pushes the blade to passively change the pitch is the push rod of the four-bar mechanism. The blade itself acts as a rocker, and the components that support the blades on both sides are support rods. The pitch amplitude can be changed by changing the length of the crank and the push rod. In order to ensure that the pitch change of the variable pitch mechanism is also a complete cycle within one rotation of the wind turbine, the radius of the circumferential gear and the central gear must be equal.

[0132] In the breeze state, in order to ensure the output power of the wind turbine, the initial passive adaptive system design is adopted to improve the utilization rate of wind energy. When the wind speed exceeds the rated wind speed range of the wind turbine, the wind turbine may be at risk of overload. Switching the wind turbine to the resistance-type blade mode and reducing the blade area in contact with the wind can reduce the force on the blades and expand the wind speed threshold for the operation of the wind turbine. In order to solve the problem that the wind turbine needs to shut down automatically during strong winds, the shutdown state of the vertical axis wind turbine is improved. In the case of a lift-type blade structure in normal operation, the telescopic push rod is retracted, and the connection between the blade body and the support frame is used as the retracted folding area until the blade body stops retracting when it is parallel to the support frame. The shape of the blade at this time is similar to the initial resistance-type blade structure, but the force area of ​​the blade and the wind is reduced, which can ensure that the wind turbine will not be overloaded in strong winds and reduce the problem of blades needing to shut down, such as Fig.12 As shown in Figure 1, the resistance type small blade structure after the blade transformation.

[0133] According to the ambient wind speed in different situations, the following four working modes are determined.

[0134] (1) Mode 1, start-up mode, using large resistance blades is conducive to the start-up of the fan, such as Fig.10 shown.

[0135] (2) Mode 2: At normal wind speed, the lift-type fixed propeller mode is used to ensure the normal operation of the wind turbine. Fig.11 shown.

[0136] (3) Mode 3: At low wind speeds, the lift-type follow-up variable pitch mode is used to improve the wind energy conversion efficiency of the wind turbine. Fig.11 shown.

[0137] (4) Mode 4: When the wind speed is high or strong, the blades are actively deformed to change into a resistance type small blade area working mode, so that the fan always works at the rated power. Fig.12 shown.

[0138] In summary, the above are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent recommendation method based on the separation principle of random forest, characterized in that: An intelligent recommendation method based on the separation principle of random forests, comprising the following steps: Step 1: Analysis of the Engineering Case Collection Obtain engineering cases from patent data sets and innovative design case collections, analyze them before and after optimization, extract physical quantity parameters of system interactions, and determine the separation principles used for system improvement or problem solving through judgment; Step 2: Generation of input and output codes. Standardize and assign values ​​to the analysis results. Encode the input physical quantity parameters through LT-KC, and encode the separation principle classification corresponding to the output of problem solving and system improvement in turn; Step 3: Construction of parameter effect and separation principle mapping model. By using the random forest algorithm, the processed sample data set is trained and a mapping model between the interaction of physical quantity parameters and the separation principle is constructed to recommend the separation principle. Step 4: Intelligent recommendation based on the separation principle of the mapping model By inputting the physical quantity parameters obtained by analyzing the system to be designed into the constructed mapping model based on random forest, the recommended separation principle is obtained, and the improved design is carried out accordingly.

2. The method for intelligent recommendation based on the separation principle of random forest according to claim 1, characterized in that: Step one includes the following steps: Step 11: Determination of the project case set A sample dataset is constructed by combining patent datasets and innovative design engineering case collections. Step 12: Obtaining sample data The learning samples need to include the correspondence between input and output, and the engineering cases in the case set need to be analyzed to determine the parameters that cause the problem and the separation principles used to solve the problem.

3. The method for intelligent recommendation based on the separation principle of random forest as claimed in claim 1, characterized in that: Step 2 includes the following steps: Step 21: Determine the encoding rules for input physical quantity parameters The LT-KC encoding of possible physical conflict parameters in the system working process of the relevant case set is used as input. The input of random forest is a set of physical quantity parameters, and the number of input physical quantity parameters is set to 3. When the system physical quantity parameters need both positive and negative values, when there is a physical conflict in the system, the physical quantity parameters describing the physical conflict and the deteriorated physical quantity parameters and improved physical quantity parameters affected by the conflict parameters are encoded according to the encoding rules respectively; Step 22: Determination of output encoding rules The separation principle coding of the relevant case set is used as output. First, the physical conflicts existing in the original system of the case are determined, and then the solution or improvement method of the case for the existing problem is analyzed, and then the separation principle used is determined and determined as the output result of the case data. The method for intelligent recommendation based on the separation principle of random forest according to claim 1, wherein step three comprises the following steps: Step 31: Data Preprocessing Clean the data, including processing missing values ​​and outliers. Step 32: Data Segmentation Split the dataset into training and testing sets Step 33: Initialization Configuration parameters: Set the model's hyperparameters, such as the number of trees, the maximum depth of the trees, etc. Step 34: Train the model Train the model using the training set data. Step 35: Model Evaluation Use the test set to evaluate the model performance using appropriate evaluation metrics (such as precision, recall, F1 score, MSE, RMSE, etc.). Step 36: Parameter Tuning Adjust the model parameters according to the evaluation results until the stopping conditions are met; Step 37: Model Validation Cross-validation and other techniques are used to further verify the stability and performance of the model. Step 38: Model Optimization: Further optimize the model according to the verification results until the stopping condition is reached (such as reaching the maximum number of iterations or the fitness meets the requirements); Step 39: Output Output the optimized model for prediction tasks.

4. The method for intelligent recommendation based on the separation principle of random forest according to claim 1, characterized in that: Step 4 includes the following steps: Step 41: Determine the system to be designed; Step 42: Analyze and determine physical quantity parameters that may cause physical conflict problems during the operation of the system; Step 43: converting the physical quantity parameters determined by the analysis into the physical quantity coding form of LT-KC; Step 44: Input the physical quantity code into the physical quantity parameter and separation principle mapping model built based on random forest; Step 45: Obtain corresponding separation principle recommendation through the mapping model; Step 46: Determine the system optimization design scheme based on the recommended separation principle.

Images