A TMD vibration reduction design method and system for a furnace front robot based on a TNN neural network

Abstract

The application provides a TMD vibration reduction design method and system for a furnace robot based on a TNN neural network, and belongs to the technical field of vibration control of industrial robots. The application determines a target main vibration frequency through field testing; constructs a high-precision sample data set in combination with theoretical calculation, finite element simulation and a scale model experiment; and on this basis, constructs a cascaded neural network model composed of a preposed reverse structure parameter prediction network and a postposed forward frequency prediction network. Through training of the forward network first and freezing of the weight of the forward network, the training of the reverse network is guided by the feedback of the forward network, so that reverse mapping from the target vibration absorption frequency to the specific structure parameters of the TMD is realized. The method of the application solves the problems of multiple solutions and convergence difficulty in traditional reverse design, can quickly output optimal structure parameters meeting the space constraints of engineering installation, effectively suppresses the vibration of the robot boom, and improves the operation stability of the equipment.

Description

Technical Field

[0001] This invention relates to the field of industrial robot vibration control technology, and in particular to a TMD vibration reduction design method and system for furnace-front robots based on TNN neural networks. Background Technology

[0002] With the development of automation in the smelting industry, furnace-front operation robots have gradually replaced manual operation. However, in actual working conditions, when the robot's pitch mechanism performs operations such as opening holes and carrying probes, the rapid start and stop of the hydraulic cylinders causes severe impact vibrations in the vertical direction of the boom structure. This long-term high-amplitude vibration not only reduces the fatigue life of the pitch and travel mechanisms but also seriously affects the robot's operational stability and safety.

[0003] In existing technologies, the TMD (Transient Damping Device) is a classic passive vibration control device, consisting of a mass block, spring, and damper. It absorbs energy by tuning its natural frequency to the vibration frequency of the main structure. The vibration reduction effect of the TMD device is highly dependent on the tuning of its mass, stiffness, and damping. However, different models or operating conditions of furnace-front robots have different dominant vibration frequencies. Traditional TMD device design usually requires repeated theoretical calculations and trial and error for specific frequencies, resulting in low design efficiency. Although artificial intelligence and neural network technology have been applied in fields such as material structure prediction, existing research on intelligent design of TMD devices mainly focuses on estimating abstract performance parameters such as damping ratio and period, lacking an effective method to directly derive specific engineering structural geometric parameters from the target vibration suppression frequency. In addition, reverse design often suffers from the "multiple solutions" problem, meaning that different parameter combinations may correspond to the same frequency, and directly training the reverse neural network often fails to converge or generates invalid structural parameters. Summary of the Invention

[0004] To address the problems in the existing technology, this invention provides a TMD (Turn-On-Demand) vibration reduction design method and system for furnace-front robots based on a TNN (Transcription Neural Network). Firstly, this invention solves the problems of difficulty in training a single inverse network and the existence of multi-value mappings in the design process of TMD vibration reduction devices by using a cascaded neural network. It can quickly generate structural parameters based on the target frequency, and the predicted frequency highly matches the target frequency. Secondly, the TMD vibration reduction device structure has strong engineering practicality, with adjustable frequency and consideration of engineering constraints, enabling it to adapt to the vibration reduction needs of different models of furnace-out robots and possessing broad industrial application value. To achieve the above objectives, the technical solution is as follows: On the one hand, this invention provides a vibration reduction design method for a furnace-front robot (TMD) based on a TNN neural network, the method comprising: S1. Obtain the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. S2. Based on the parametric model of the TMD vibration reduction device, a sample dataset containing structural parameters and corresponding vibration absorption frequencies is obtained through analysis. S3. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, construct and train a positive frequency prediction neural network to obtain a positive prediction model. S4. Based on the positive prediction model, construct a serial inverse neural network model to obtain the TNN neural network model; S5. Based on the target main vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained by iterating through the loss function. S6. Based on the target structural parameters of the TMD vibration damping device, the target TMD vibration damping device is manufactured and obtained.

[0005] Optionally, in S2, based on the parameterized model of the TMD vibration damping device, a sample dataset containing structural parameters and corresponding absorption frequencies is obtained through analysis, including: S21. Based on the parametric model of the TMD vibration reduction device, multiple sets of structural parameters are randomly generated by establishing the axial stiffness formula and frequency formula of the TMD device. S22. Based on the multiple sets of structural parameter datasets, modal analysis is performed using finite element analysis software to obtain the modal analysis dataset; S23. Based on the modal analysis dataset, a physical model is created for verification, resulting in a sample dataset containing structural parameters and corresponding vibration absorption frequencies.

[0006] Optionally, the structural parameters include: the outer diameter, inner diameter, and height of the rubber block, and the mass of the iron block.

[0007] Optionally, the TMD vibration damping device includes: an iron housing, rubber blocks, and fastening units; the iron housing consists of a fixed mass housing and an adjustable mass housing.

[0008] Optionally, in step S3, a positive frequency prediction neural network is constructed and trained based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies to obtain a positive prediction model, including: S31. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, the Adam optimization algorithm is used, and an initial learning rate is set to obtain the initial positive prediction model. S32. Based on the initial positive prediction model, with structural parameters as input and vibration absorption frequency as output, and using mean square error (MSE) as loss function, iterative training is performed to obtain the optimized parameters of the positive prediction model. S33. Save and update the optimized parameters of the positive prediction model to obtain the positive prediction model.

[0009] Optionally, in S4, a cascaded inverse neural network model is constructed based on the forward prediction model to obtain a TNN neural network model, including: S41. Based on the forward prediction model, construct a cascaded inverse neural network model, keep the optimized parameters of the forward prediction model, optimize the parameters of the inverse neural network model, and obtain the inverse neural network model. S42. Based on the inverse neural network model, the prediction output of the forward prediction model is used to supervise the results of the inverse neural network model to obtain the TNN neural network model.

[0010] Optionally, in step S5, based on the target dominant vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained through iteration using a loss function, including: S51. Based on the TNN neural network model, set a constraint layer to obtain a TNN neural network model with constraints. S52. Based on the constrained TNN neural network model and the target main vibration frequency to be suppressed, the target structural parameters of the TMD vibration reduction device are obtained by iterating using the mean square error (MSE) as the loss function.

[0011] Optionally, the constraint layer is based on the actual installation space of the furnace-front operating robot and the process dimensions of the rubber material.

[0012] On the other hand, the present invention provides a TMD vibration reduction design system for a furnace-front robot based on a TNN neural network. This system is applied to a TMD vibration reduction design method for a furnace-front robot based on a TNN neural network. The system includes: The frequency acquisition module is used to acquire the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. The data acquisition module is used to obtain a sample dataset containing structural parameters and corresponding vibration absorption frequencies by analyzing the parameterized model of the TMD vibration reduction device. The first model building module is used to build and train a positive frequency prediction neural network based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, so as to obtain a positive prediction model. The second model building module is used to build a cascaded inverse neural network model based on the positive prediction model, thereby obtaining the TNN neural network model; The structural parameter acquisition module is used to obtain the target structural parameters of the TMD vibration reduction device by iterating through a loss function based on the target main vibration frequency to be suppressed and the TNN neural network model. The device fabrication module is used to fabricate and obtain the target TMD vibration damping device based on the target structural parameters of the TMD vibration damping device.

[0013] Compared with the prior art, the technical solution of the present invention has at least the following beneficial effects: The above-mentioned solution solves the problems of difficulty in training a single inverse network and the existence of multi-value mapping in the design of TMD vibration reduction device by using a serial neural network. It can quickly generate structural parameters according to the target frequency and the predicted frequency is highly consistent with the target frequency. On the other hand, the TMD vibration reduction device structure has strong engineering practicality, frequency adjustability and takes into account engineering constraints, and can adapt to the vibration reduction needs of different models of furnace robots, and has broad industrial application value. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of an embodiment of the TMD vibration reduction design method for furnace front robot based on TNN neural network of the present invention; Figure 2 This is a flowchart of a sample dataset containing structural parameters and corresponding vibration absorption frequencies obtained in an embodiment of the TMD vibration reduction design method for furnace front robot based on TNN neural network of the present invention. Figure 3 This is a flowchart of the positive prediction model obtained in an embodiment of the TMD vibration reduction design method for furnace front robot based on TNN neural network of the present invention; Figure 4 This is a flowchart of the TNN neural network model obtained in an embodiment of the TMD vibration reduction design method for furnace front robot based on TNN neural network of the present invention; Figure 5 This is a flowchart illustrating the target structural parameters of the TMD vibration reduction device obtained in an embodiment of the TNN-based furnace front robot TMD vibration reduction design method of the present invention. Figure 6 This is a schematic diagram of the TMD vibration reduction device in an embodiment of the TNN neural network-based furnace front robot TMD vibration reduction design method of the present invention; Figure 7 This is a system block diagram of an embodiment of the TMD vibration reduction design system for furnace front robot based on TNN neural network of the present invention.

[0016] The numbers in the diagram are explained as follows: 1. Iron outer shell; 2. Rubber block; 3. Fastening unit; 11. Fixed mass outer shell; 12. Adjustable mass outer shell. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0020] like Figure 1 The flowchart shown is an embodiment of the TMD vibration reduction design method for furnace-front robots based on TNN neural networks according to the present invention. The present invention provides a TMD vibration reduction design method for furnace-front robots based on TNN neural networks, which is implemented by a TMD vibration reduction design system for furnace-front robots based on TNN neural networks. The method includes: S1. Obtain the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. Furthermore, field tests were conducted to address the vibration issue of the calcium carbide furnace unloading robot during pitching motion. A three-axis vibration acceleration sensor was installed at the end of the robot's arm mechanism. During the rapid start and stop of the hydraulic cylinder-driven pitching motion of the arm, vertical vibration acceleration signals were collected. By performing fast Fourier transform spectral analysis on the collected time-domain signals, the dominant frequency of the arm mechanism under this operating condition was obtained. In this embodiment, the vertical dominant frequency of the end of the calcium carbide furnace unloading robot's arm was measured to be 30Hz, and this frequency was set as the target dominant frequency for suppression by the TMD vibration damping device.

[0021] S2. Based on the parametric model of the TMD vibration reduction device, a sample dataset containing structural parameters and corresponding vibration absorption frequencies is obtained through analysis. Specifically, such as Figure 2 The flowchart shown in this embodiment of the TNN-based TMD vibration reduction design method for furnace-front robots of the present invention obtains a sample dataset containing structural parameters and corresponding vibration absorption frequencies. In step S2, based on the parameterized model of the TMD vibration reduction device, the sample dataset containing structural parameters and corresponding vibration absorption frequencies is obtained through analysis, including: S21. Based on the parametric model of the TMD vibration reduction device, multiple sets of structural parameters are randomly generated by establishing the axial stiffness formula and frequency formula of the TMD device. S22. Based on the multiple sets of structural parameter datasets, modal analysis is performed using finite element analysis software to obtain the modal analysis dataset; S23. Based on the modal analysis dataset, a physical model is created for verification, resulting in a sample dataset containing structural parameters and corresponding vibration absorption frequencies.

[0022] Specifically, the structural parameters include: the outer diameter, inner diameter, and height of the rubber block, and the mass of the iron block.

[0023] Specifically, such as Figure 6 The diagram shown is a schematic diagram of the TMD vibration reduction device in an embodiment of the TNN-based furnace robot TMD vibration reduction design method of the present invention. The TMD vibration reduction device includes: an iron shell 1, a rubber block 2, and a fastening unit 3; the iron shell 1 is composed of a fixed mass shell 11 and an adjustable mass shell 12.

[0024] Furthermore, based on the robot's installation space constraints, the structural design and parametric modeling of the TMD vibration damping device were designed to create an easy-to-install and frequency-adjustable TMD vibration damping device. Physical structure: The device mainly consists of an iron outer shell 1, rubber blocks 2, and fastening units 3. The iron outer shell 1 comprises a fixed-mass shell 11 and an adjustable-mass shell 12; the mass is fine-tuned by increasing or decreasing the number of adjustable-mass shells 12. m Four core geometric and physical parameters that determine the performance of the TMD vibration damping device are selected as design variables: the outer diameter of the cylindrical rubber block. d 1 Inner diameter of cylindrical rubber block d 2 Height of cylindrical rubber block h and the quality of the iron casing m Based on the principle of vibration absorption, the axial stiffness of the TMD device is established. k and axial natural frequency f Theoretical calculation formula: (1) (2) In the formula, E This is the elastic modulus of the rubber material.

[0025] To train a high-precision neural network, a dataset was constructed using a method combining theoretical calculation, simulation verification, and experimental correction. To verify the accuracy of the theoretical formula, a small-scale TMD vibration damping device model was first designed and manufactured, with structural parameters shown in Table 1. The natural frequencies of the small-scale TMD device were obtained using three methods: theoretical calculation, finite element simulation modal analysis, and laser vibration meter frequency response testing. The results of the three analysis methods show minimal errors, verifying the reliability of the theoretical formula in generating large-scale datasets. Large-scale data generation: Based on the verified theoretical formula, 12,000 sets of structural parameters were randomly generated within the parameter range that conforms to the engineering installation space. d 1 , d 2 , h and m and its corresponding absorption frequency f and angular frequency After normalization, these sets are used as the training and testing sets for the neural network.

[0026] Table 1 Structural Parameters of Small TMD Vibration Damper Devices

[0027] S3. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, construct and train a positive frequency prediction neural network to obtain a positive prediction model. Specifically, such as Figure 3 The flowchart shown in this embodiment of the TNN-based TMD vibration reduction design method for furnace-front robots of the present invention obtains a positive prediction model. In step S3, based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, a positive frequency prediction neural network is constructed and trained to obtain the positive prediction model, including: S31. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, the Adam optimization algorithm is used, and an initial learning rate is set to obtain the initial positive prediction model. S32. Based on the initial positive prediction model, with structural parameters as input and vibration absorption frequency as output, and using mean square error (MSE) as loss function, iterative training is performed to obtain the optimized parameters of the positive prediction model. S33. Save and update the optimized parameters of the positive prediction model to obtain the positive prediction model.

[0028] Furthermore, the positive frequency prediction network structure is constructed and trained: the input layer contains 4 neurons, with corresponding structural parameters... d 1 , d 2 , h and mAfter passing through several hidden layers (5 layers), with node numbers of 1024, 512, 256, 256, and 128 respectively, the output layer contains 2 neurons, corresponding to the vibration absorption frequency. f and angular frequency The ReLU activation function was used. Training process: Supervised learning was performed using the aforementioned 12,000 datasets. The Adam optimization algorithm was used, with a learning rate of 0.002 and mean squared error as the loss function. Validation was performed every 10 training steps, and training stopped when the prediction accuracy stabilized above 95%. After training, the weight parameters of the forward network were saved and used as the later part of the cascaded network. In subsequent steps, its weights were frozen and not updated.

[0029] S4. Based on the positive prediction model, construct a serial inverse neural network model to obtain the TNN neural network model; Specifically, such as Figure 4 The flowchart shown in the embodiment of the TNN neural network-based furnace front robot TMD vibration reduction design method of the present invention obtains the TNN neural network model. In step S4, based on the forward prediction model, a series inverse neural network model is constructed to obtain the TNN neural network model, including: S41. Based on the forward prediction model, construct a cascaded inverse neural network model, keep the optimized parameters of the forward prediction model, optimize the parameters of the inverse neural network model, and obtain the inverse neural network model. S42. Based on the inverse neural network model, the prediction output of the forward prediction model is used to supervise the results of the inverse neural network model to obtain the TNN neural network model.

[0030] Furthermore, the input layer of the pre-inverse network is the target absorption frequency. f and angular frequency The output layer contains the predicted structural parameters. d 1 , d 2 , h and m The output is directly used as the input to the subsequent forward network. A design space constraint is introduced at the output of the inverse network to force the output structural parameters to be within the engineering-allowed range, avoiding the generation of invalid solutions. The randomly generated target frequency is input to the cascaded network; the pre-network outputs predicted structural parameters, and the subsequent forward network calculates the predicted vibration absorption frequency based on these parameters. f and angular frequency By using the backpropagation algorithm, only the weights of the preceding inverse network are updated, forcing the structural parameters generated by the inverse network to match the parameters of the target frequency generated by the forward network. When the accuracy of the predicted frequency matching the target frequency reaches a preset level, such as an error of <1%, the inverse neural network training is complete.

[0031] S5. Based on the target main vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained by iterating through the loss function. Specifically, such as Figure 5 The flowchart shown in this embodiment of the TNN neural network-based furnace front robot TMD vibration reduction design method of the present invention obtains the target structural parameters of the TMD vibration reduction device. In step S5, based on the target main vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained through iteration using a loss function, including: S51. Based on the TNN neural network model, set a constraint layer to obtain a TNN neural network model with constraints. Furthermore, this constraint layer is based on the actual installation space of the furnace-front operating robot and the process dimensions of the rubber material.

[0032] S52. Based on the constrained TNN neural network model and the target main vibration frequency to be suppressed, the target structural parameters of the TMD vibration reduction device are obtained by iterating using the mean square error (MSE) as the loss function.

[0033] Furthermore, the target main vibration frequency of 30Hz, measured in S1, is input into the trained cascaded inverse neural network model. The model outputs the structural parameters of the optimal TMD vibration damping device for this calcium carbide furnace unloading robot, including the height of the cylindrical rubber block, the inner diameter of the cylindrical rubber block, the outer diameter of the cylindrical rubber block, and the mass of the iron shell, as shown in Table 2.

[0034] Table 2 Structural Parameters of TMD Device

[0035] S6. Based on the target structural parameters of the TMD vibration damping device, the target TMD vibration damping device is manufactured and obtained.

[0036] Specifically, based on the predicted parameters in Table 2, appropriate steel materials were selected for the processing, manufacturing, and installation of the TMD device prototype.

[0037] Before installation: The peak value of the vertical vibration acceleration at the end of the boom was high, and the oscillation decay time was long. After installation: The peak value of the vertical vibration acceleration at the end of the boom was significantly reduced; the time-domain waveform showed rapid vibration convergence and a significant improvement in stability. Vibration in the impact direction and rotation direction was monitored simultaneously, and the results showed that the vibration amplitude in these two directions did not change significantly after the TMD device was installed, proving that the device did not introduce a negative coupling effect.

[0038] like Figure 7The diagram shown is a system block diagram of an embodiment of the TMD vibration reduction design system for a furnace-front robot based on a TNN neural network according to the present invention. The present invention provides a TMD vibration reduction design system for a furnace-front robot based on a TNN neural network. This system is applied to a TMD vibration reduction design method for a furnace-front robot based on a TNN neural network. The system includes: a frequency acquisition module, a data acquisition module, a first model construction module, a second model construction module, a structural parameter acquisition module, and a device fabrication module. Specifically, The frequency acquisition module is used to acquire the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. The data acquisition module is used to obtain a sample dataset containing structural parameters and corresponding vibration absorption frequencies by analyzing the parameterized model of the TMD vibration reduction device. The first model building module is used to build and train a positive frequency prediction neural network based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, so as to obtain a positive prediction model. The second model building module is used to build a cascaded inverse neural network model based on the positive prediction model, thereby obtaining the TNN neural network model; The structural parameter acquisition module is used to obtain the target structural parameters of the TMD vibration reduction device by iterating through a loss function based on the target main vibration frequency to be suppressed and the TNN neural network model. The device fabrication module is used to fabricate and obtain the target TMD vibration damping device based on the target structural parameters of the TMD vibration damping device.

[0039] This invention provides a TMD (Total Motion Damping) vibration reduction design method and system for furnace-front robots based on a TNN (Telematics Neural Network). Firstly, this invention solves the problems of difficulty in training a single inverse network and the existence of multi-value mappings in the design process of TMD vibration reduction devices by using a cascaded neural network. It can quickly generate structural parameters based on the target frequency, and the predicted frequency highly matches the target frequency. Secondly, the TMD vibration reduction device structure has strong engineering practicality, with adjustable frequency and consideration of engineering constraints, enabling it to adapt to the vibration reduction requirements of different models of furnace-front robots. It successfully achieves rapid inverse design of the structural parameters of the TMD vibration reduction device for calcium carbide furnace-front robots under specific working conditions, effectively solving the vibration control problem in practical engineering and possessing broad industrial application value.

[0040] It is understood that the present invention has been described through the above embodiments and should not be construed as limiting the implementation and scope of the present invention. Those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the present invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A vibration reduction design method for a furnace-front robot (TMD) based on a TNN neural network, characterized in that, The method includes: S1. Obtain the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. S2. Based on the parametric model of the TMD vibration reduction device, a sample dataset containing structural parameters and corresponding vibration absorption frequencies is obtained through analysis. S3. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, construct and train a positive frequency prediction neural network to obtain a positive prediction model; S4. Based on the positive prediction model, construct a serial inverse neural network model to obtain the TNN neural network model; S5. Based on the target main vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained by iterating through the loss function. S6. Based on the target structural parameters of the TMD vibration damping device, the target TMD vibration damping device is manufactured and obtained.

2. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 1, characterized in that, In step S2, based on the parameterized model of the TMD vibration reduction device, a sample dataset containing structural parameters and corresponding absorption frequencies is obtained through analysis, including: S21. Based on the parametric model of the TMD vibration reduction device, multiple sets of structural parameters are randomly generated by establishing the axial stiffness formula and frequency formula of the TMD device. S22. Based on the multiple sets of structural parameter datasets, modal analysis is performed using finite element analysis software to obtain a modal analysis dataset; S23. Based on the modal analysis dataset, a physical model is created for verification to obtain a sample dataset containing structural parameters and corresponding vibration absorption frequencies.

3. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 2, characterized in that, The structural parameters include: the outer diameter, inner diameter, and height of the rubber block, and the mass of the iron block.

4. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 1, characterized in that, The TMD vibration damping device includes: an iron outer shell, a rubber block, and a fastening unit; the iron outer shell consists of a fixed mass shell and an adjustable mass shell.

5. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 1, characterized in that, In step S3, a positive frequency prediction neural network is constructed and trained based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies to obtain a positive prediction model, including: S31. Based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, the Adam optimization algorithm is used, and an initial learning rate is set to obtain an initial positive prediction model. S32. Based on the initial positive prediction model, with structural parameters as input and vibration absorption frequency as output, and using mean square error (MSE) as loss function, iterative training is performed to obtain the optimized parameters of the positive prediction model. S33. Save and update the optimized parameters of the positive prediction model to obtain the positive prediction model.

6. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 5, characterized in that, In step S4, a cascaded inverse neural network model is constructed based on the forward prediction model to obtain a TNN neural network model, including: S41. Based on the forward prediction model, construct a cascaded inverse neural network model, maintain the optimized parameters of the forward prediction model, optimize the parameters of the inverse neural network model, and obtain the inverse neural network model. S42. Based on the inverse neural network model, the prediction output of the forward prediction model is used to supervise the results of the inverse neural network model to obtain the TNN neural network model.

7. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 1, characterized in that, In step S5, based on the target dominant vibration frequency to be suppressed and the TNN neural network model, the target structural parameters of the TMD vibration reduction device are obtained through iteration using a loss function, including: S51. Based on the TNN neural network model, set a constraint layer to obtain a TNN neural network model with constraints. S52. Based on the TNN neural network model with constraints and the target main vibration frequency to be suppressed, the target structural parameters of the TMD vibration reduction device are obtained by iterating using the mean square error (MSE) as the loss function.

8. The TMD vibration reduction design method for furnace-front robot based on TNN neural network according to claim 7, characterized in that, The constraint layer is based on the actual installation space of the furnace-front operating robot and the process dimensions of the rubber material.

9. A TNN-based TMD vibration reduction design system for furnace-front robots, used to implement the TNN-based TMD vibration reduction design method for furnace-front robots as described in any one of claims 1-8, characterized in that, The system includes: The frequency acquisition module is used to acquire the field vibration data of the boom mechanism of the furnace front operation robot under pitching motion, and obtain the target main vibration frequency to be suppressed through spectrum analysis. The data acquisition module is used to obtain a sample dataset containing structural parameters and corresponding vibration absorption frequencies by analyzing the parameterized model of the TMD vibration reduction device. The first model building module is used to build and train a positive frequency prediction neural network based on the sample dataset containing structural parameters and corresponding vibration absorption frequencies, so as to obtain a positive prediction model. The second model construction module is used to construct a serial inverse neural network model based on the forward prediction model to obtain a TNN neural network model. The structural parameter acquisition module is used to obtain the target structural parameters of the TMD vibration reduction device by iterating through a loss function based on the target main vibration frequency to be suppressed and the TNN neural network model. The device fabrication module is used to fabricate and obtain the target TMD vibration damping device based on the target structural parameters of the TMD vibration damping device.

Images