Neural network intelligent control method and system based on piezoelectric fiber drive vibration reduction

Abstract

The present invention provides a neural network intelligent control method and system based on piezoelectric fiber driven vibration reduction, the method comprising the following steps: using LQR algorithm to obtain MFC response signal and control force, and using CNN-LSTM neural network training as training data to be applied to neural network intelligent control module, directly mapping the control current signal required by the current controller according to the acceleration signal, velocity signal and displacement signal output by the signal acquisition module; transmitting the control current signal to the current controller, the current controller generates a corresponding current, and drives the MFC actuator after amplification by a high-voltage amplifier; the MFC actuator generates a control force to suppress the vibration of the controlled structure, and reduces the vibration response of the controlled structure in real time. The present invention improves the operating speed and stability of the control system, realizes rapid data collection, analysis, and vibration control; does not need to establish a mathematical model and parameters of the controlled object, and can adapt to complex vibration conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of structural engineering and structural vibration control, and in particular to a neural network intelligent control method and system based on piezoelectric fiber driven vibration reduction. Background Art

[0002] With the popularization and development of mechanization in production and life, machines are widely used as the driving force in industry and transportation. The harm of mechanical vibration is becoming more and more serious, and the demand for vibration control is becoming more and more urgent.

[0003] Power machinery such as steam turbines, water turbines and motors, transportation vehicles such as cars, trains, ships and airplanes, as well as machine tools, mining machinery and engineering machinery, are all developing in the direction of high speed and heavy load, so the vibration is becoming increasingly strong. Strong vibration can also cause serious damage to machinery, equipment, structures and even buildings.

[0004] At present, with the rapid development of the aerospace industry, people have higher expectations for vibration control in order to meet the needs of aircraft, missiles and other aircraft for faster speed, wider adaptability, stronger concealment and longer service life. Excellent vibration control can effectively solve the structural fatigue problem caused by buffeting, extend the life of the aircraft and improve the passenger experience. Therefore, the study of vibration control has important theoretical significance and engineering practical value.

[0005] In the past half century, the theories and technologies of seismic isolation, energy dissipation, and active and semi-active control have made great progress and have been continuously applied in practical engineering. However, the control effect is still limited for complex structures, and the progress of the above control technologies is hindered by the fact that there are many structural parameters and they are difficult to obtain. Summary of the invention

[0006] In view of this, the purpose of the present invention is to combine CNN with LSTM to construct a centralized control strategy that can simultaneously reduce the time and space dimensions and visualize the acquired information, so as to achieve the purpose of improving the operation speed and stability of the centralized control system while ensuring the calculation accuracy.

[0007] MFC piezoelectric fiber composite material is formed by combining piezoelectric ceramics as the functional phase, epoxy resin as the matrix, and interdigitated electrodes in a certain spatial structure. It not only has excellent piezoelectric properties, but also has good flexibility of polymers, ensuring that MFC will not be damaged due to fiber breakage during vibration bending. Because the electrical response generated by piezoelectric ceramics has the characteristics of high voltage, low current, and low actual power, it has broad application prospects in the field of vibration control.

[0008] Intelligent control combines the advantages of traditional control algorithms and artificial intelligence, and does not require the establishment of precise structural models, so it has become a research hotspot in recent years.

[0009] The present invention adopts an LQR algorithm to obtain the displacement, velocity, acceleration response and control force training data of the MFC position of the controlled structure, and adopts the training data to train a CNN-LSTM neural network to maturity; the trained CNN-LSTM neural network is used in a neural network intelligent control module, and a control current signal required by a current controller is directly mapped according to an acceleration signal, a velocity signal and a displacement signal output by a signal acquisition module, and the control current signal is transmitted to the current controller, and the current controller generates a corresponding current, which is amplified by a high-voltage amplifier and then drives an MFC actuator, and the MFC generates a corresponding control force to suppress the vibration of the controlled structure to be controlled, thereby achieving the effect of reducing the structural vibration response in real time.

[0010] The present invention provides a neural network intelligent control method based on piezoelectric fiber driven vibration reduction, comprising the following steps:

[0011] S1: The LQR algorithm is used to calculate the MFC response signal and control force of the MFC actuator arranged on the controlled structure, and the CNN-LSTM neural network is trained as training data;

[0012] S2: Apply the well-trained CNN-LSTM neural network to the neural network intelligent control module (intelligent controller), and directly map the control current signal required by the current controller according to the acceleration signal, velocity signal and displacement signal output by the signal acquisition module;

[0013] S3: transmitting the control current signal to a current controller, the current controller generates a corresponding current, and drives the MFC actuator after being amplified by a high-voltage amplifier;

[0014] S4: The MFC actuator generates a control force to suppress the vibration of the controlled structure, thereby reducing the vibration response of the controlled structure in real time.

[0015] Furthermore, the method of obtaining the displacement, velocity, and acceleration response of the position of the MFC actuator in step S1 includes:

[0016] The acceleration signal of the MFC actuator 6 is acquired through the acceleration sensor 1, the acceleration signal is filtered and conditioned through the signal acquisition module, and the velocity signal and the displacement signal are obtained through integration.

[0017] Furthermore, the training method of the CNN-LSTM neural network includes:

[0018] The MFC response signal and external excitation are used as input values ​​of the CNN-LSTM neural network, and the control force of the MFC actuator 6 at the next moment is used as the output value of the CNN-LSTM neural network. The deep learning network framework of the CNN-LSTM neural network is trained and tested to obtain a mature CNN-LSTM neural network.

[0019] Furthermore, the verification method of the deep learning network framework of the CNN-LSTM neural network includes:

[0020] By inputting external excitation into the maturely trained CNN-LSTM neural network, the output value of the CNN-LSTM neural network is obtained and compared with the theoretically calculated control force to verify the effectiveness of the deep learning network framework of the CNN-LSTM neural network. The prediction process loss function is expressed as the mean square error MSE:

[0021]

[0022] Where: MSE refers to mean square error, N is the size of the database; y i , are the measured value and predicted value at time i respectively.

[0023] Furthermore, the deep learning network framework of the CNN-LSTM neural network includes: sequence input layer, sequence folding layer, convolution layer (Conv1), batch normalization layer, dropout layer (Dropout1), activation layer (Relu1), maximum pooling layer (Pool1), convolution layer (Conv2), dropout layer (Dropout2), activation layer (Relu2), maximum pooling layer (Pool2), sequence expansion layer, flattening layer, Lstm layer (Lstm1), dropout layer (Dropout3), Lstm layer (Lstm2), dropout layer (Dropout4), fully connected layer and regression layer;

[0024] The present invention adopts a two-layer convolutional neural network and two-layer maximum pooling layer alternately connected to reduce the dimension of the two-dimensional time-frequency domain features, and adds a discard layer to prevent overfitting; then it is input into two layers of continuous bidirectional long short-term memory network for processing, and the time domain intermediate features are output; the connection layer is a sequence folding layer-sequence expansion layer, the former is a sequence folding layer, which is placed before CNN and can fold the sequence and perform independent convolution operations at each time step; the latter is a sequence expansion layer, which is placed before LSTM and can restore the sequence structure of the input data after the sequence is folded.

[0025] Furthermore, the MFC response signal includes:

[0026] A combination of one or more of acceleration signals, velocity signals and displacement signals.

[0027] The present invention also provides a neural network intelligent control system based on piezoelectric fiber driven vibration reduction, which executes the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described above, including: an acceleration sensor 1, a signal acquisition module 2, a neural network intelligent control module 3, a current controller 4, a high-voltage amplifier 5, an MFC actuator 6, and a controlled structure 7;

[0028] The acceleration sensor 1 and the MFC actuator 6 are both arranged on the surface of the controlled structure 7. The acceleration sensor 1 is used to obtain the acceleration signal of the MFC actuator 6. The acceleration sensor 1 is electrically connected to the signal acquisition module 2 to transmit the acceleration signal to the signal acquisition module 2.

[0029] Preferably, the adhesive used to adhere the MFC actuator 6 to the surface of the controlled structure 7 is epoxy resin;

[0030] Furthermore, the controlled structure 7 is provided with a plurality of acceleration sensors 1 at the same time.

[0031] The signal acquisition module 2 filters and conditions the acceleration signal, and obtains the velocity signal and the displacement signal by integration; the signal acquisition module 2 is electrically connected to the neural network module 3, and transmits the acceleration signal, the velocity signal and the displacement signal to the neural network intelligent control module 3;

[0032] The neural network module 3 is electrically connected to the current controller 4. The neural network intelligent control module 3 directly maps the output control current signal according to the signal transmitted by the signal acquisition module 2 as the current input MFC response signal, and transmits the control current signal to the current controller 4.

[0033] The current controller 4 is electrically connected to the high-voltage amplifier 5, and the high-voltage amplifier 5 is electrically connected to the MFC actuator 6. The current controller 4 amplifies the control current signal through the high-voltage amplifier 5 to drive the MFC actuator 6 to generate a control force acting on the controlled structure 7, thereby suppressing the vibration of the controlled structure 7.

[0034] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the above-mentioned neural network intelligent control method based on piezoelectric fiber driven vibration reduction are implemented.

[0035] The present invention also provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described above are implemented.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention combines CNN with LSTM to construct a control strategy that can simultaneously reduce the two dimensions of time and space and can visualize the acquired information, thereby improving the operating speed and stability of the control system and also ensuring the calculation accuracy; the CNN-LSTM neural network performs dimensionality reduction processing on acceleration signals, velocity signals and displacement signals through convolution kernels of different sizes, and then transmits the signals to the LSTM intelligent control system, generates an MFC actuator control force and applies it to the controlled structure, thereby realizing rapid data collection, analysis and vibration control; in addition, the neural network intelligent control system of the present invention does not need to establish a mathematical model of the controlled structure, nor does it need parameters of the controlled structure itself, such as the mass, inertia, geometric dimensions, stiffness, etc. of the controlled structure, has strong self-adaptation ability, and can adapt to complex vibration working condition changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiment.The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention.

[0039] In the attached picture:

[0040] Figure 1 It is a flow chart of the neural network intelligent control method based on piezoelectric fiber driven vibration reduction of the present invention;

[0041] Figure 2 A schematic diagram of the structure of a computer device according to an embodiment of the present invention;

[0042] Figure 3 It is a structural schematic diagram of a neural network intelligent control system based on piezoelectric fiber driven vibration reduction according to an embodiment of the present invention;

[0043] Figure 4 This is a CNN-LSTM neural network learning flow chart of an embodiment of the present invention;

[0044] Figure 5 This is a diagram of the internal structure of the CNN-LSTM neural network in an embodiment of the present invention;

[0045] Figure 6 It is a comparison diagram of the acceleration response of the CNN-LSTM neural network in an embodiment of the present invention and the uncontrolled state.

[0046] The symbols in the accompanying drawings are:

[0047] 1. Acceleration sensor; 2. Signal acquisition module; 3. Neural network intelligent control module; 4. Current controller; 5. High-voltage amplifier; 6. MFC actuator; 7. Controlled structure. DETAILED DESCRIPTION

[0048] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and products consistent with some aspects of the present disclosure as detailed in the appended claims.

[0049] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. The singular forms of "a", "said" and "the" used in this disclosure and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0050] It should be understood that although the terms first, second, third, etc. may be used in the present disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present disclosure, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0051] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0052] The embodiment of the present invention provides a neural network intelligent control method based on piezoelectric fiber driven vibration reduction, see Figure 1 As shown, the following steps are included:

[0053] S1: The LQR algorithm is used to calculate the MFC response signal and control force of the MFC actuator arranged on the controlled structure, and the CNN-LSTM neural network is trained as training data;

[0054] The MFC response signal includes:

[0055] A combination of one or more of an acceleration signal, a velocity signal and a displacement signal;

[0056] The method for obtaining the displacement, velocity and acceleration response of the position of the MFC actuator includes:

[0057] The acceleration signal of the MFC actuator 6 is obtained by the acceleration sensor 1, the acceleration signal is filtered and conditioned by the signal acquisition module, and the velocity signal and the displacement signal are obtained by integration;

[0058] In this embodiment, the training method of the CNN-LSTM neural network includes:

[0059] The MFC response signal and the external excitation are used as the input values ​​of the CNN-LSTM neural network, the control force of the MFC actuator 6 at the next moment is used as the output value of the CNN-LSTM neural network, the deep learning network framework of the CNN-LSTM neural network is trained and a mature CNN-LSTM neural network is obtained through testing;

[0060] In this embodiment, the verification method of the deep learning network framework of the CNN-LSTM neural network includes:

[0061] By inputting external excitation into the maturely trained CNN-LSTM neural network, the output value of the CNN-LSTM neural network is obtained and compared with the theoretically calculated control force to verify the effectiveness of the deep learning network framework of the CNN-LSTM neural network. The prediction process loss function is expressed as the mean square error MSE:

[0062]

[0063] Where: MSE refers to mean square error, N is the size of the database; y i , are the measured value and predicted value at time i respectively;

[0064] The deep learning network framework of the CNN-LSTM neural network in this embodiment is shown in Figure 5 The following are shown: sequence input layer, sequence folding layer, convolution layer (Conv1), batch normalization layer, drop layer (Dropout1), activation layer (Relu1), maximum pooling layer (Pool1), convolution layer (Conv2), drop layer (Dropout2), activation layer (Relu2), maximum pooling layer (Pool2), sequence expansion layer, flattening layer, Lstm layer (Lstm1), drop layer (Dropout3), Lstm layer (Lstm2), drop layer (Dropout4), fully connected layer and regression layer;

[0065] The present invention adopts a two-layer convolutional neural network and two-layer maximum pooling layer alternately connected to reduce the dimension of the two-dimensional time-frequency domain features, and adds a discard layer to prevent overfitting; then inputs it into two layers of continuous bidirectional long short-term memory networks for processing, and outputs the intermediate features in the time domain; the connection layer is a sequence folding layer-sequence expansion layer, the former is a sequence folding layer, which is placed before CNN and can fold the sequence and perform independent convolution operations at each time step; the latter is a sequence expansion layer, which is placed before LSTM and can restore the sequence structure of the input data after the sequence is folded;

[0066] S2: Apply the well-trained CNN-LSTM neural network to the neural network intelligent control module (intelligent controller), and directly map the control current signal required by the current controller according to the acceleration signal, velocity signal and displacement signal output by the signal acquisition module;

[0067] S3: transmitting the control current signal to a current controller, the current controller generates a corresponding current, and drives the MFC actuator after being amplified by a high-voltage amplifier;

[0068] S4: The MFC actuator generates a control force to suppress the vibration of the controlled structure, thereby reducing the vibration response of the controlled structure in real time.

[0069] The present invention also provides a neural network intelligent control system based on piezoelectric fiber driven vibration reduction, which executes the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described above, see Figure 3 As shown, it includes: an acceleration sensor 1, a signal acquisition module 2, a neural network intelligent control module 3, a current controller 4, a high-voltage amplifier 5, an MFC actuator 6, and a controlled structure 7;

[0070] The acceleration sensor 1 and the MFC actuator 6 are both arranged on the surface of the controlled structure 7. The acceleration sensor 1 is used to obtain the acceleration signal of the MFC actuator 6. The acceleration sensor 1 is electrically connected to the signal acquisition module 2 to transmit the acceleration signal to the signal acquisition module 2.

[0071] Preferably, the adhesive used to adhere the MFC actuator 6 to the surface of the controlled structure 7 is epoxy resin;

[0072] Preferably, the controlled structure 7 is provided with a plurality of acceleration sensors 1 at the same time.

[0073] The signal acquisition module 2 filters and conditions the acceleration signal, and obtains the velocity signal and the displacement signal by integration; the signal acquisition module 2 is electrically connected to the neural network module 3, and transmits the acceleration signal, the velocity signal and the displacement signal to the neural network intelligent control module 3;

[0074] The neural network module 3 is electrically connected to the current controller 4. The neural network intelligent control module 3 directly maps the output control current signal according to the signal transmitted by the signal acquisition module 2 as the current input MFC response signal, and transmits the control current signal to the current controller 4.

[0075] The current controller 4 is electrically connected to the high-voltage amplifier 5, and the high-voltage amplifier 5 is electrically connected to the MFC actuator 6. The current controller 4 amplifies the control current signal through the high-voltage amplifier 5 to drive the MFC actuator 6 to generate a control force acting on the controlled structure 7, thereby suppressing the vibration of the controlled structure 7.

[0076] In this embodiment, MFC actuators 6 are pasted in the middle and four corners of the plate-shell structure filling plate with a size of 1.0m×0.5m×0.02m. The filling plate density is 1510kg / m3, the four internal fillings are 0.344m×0.25m×0.02m in size, and the Poisson's ratio is 0.3. El-centro wave is selected as the input excitation to act on the center of the plate. The LQR control algorithm is used to obtain the displacement, velocity, acceleration and control force of the MFC actuator position under the excitation of the structure, and the CNN-LSTM neural network is trained to maturity. Then according to the following Figure 4 The steps shown use the CNN-LSTM neural network as an intelligent control algorithm to control the current output and then control the piezoelectric fiber to drive the MFC to generate a control force acting on the plate shell, and obtain the acceleration response of the center of the plate after control, and compare it with the uncontrolled state, see Figure 6 As shown, it can be clearly seen that there is a good control effect.

[0077] The embodiment of the present invention combines CNN with LSTM to construct a control strategy that can simultaneously reduce the two dimensions of time and space and can visualize the acquired information, thereby improving the operating speed and stability of the control system and also ensuring the calculation accuracy; the CNN-LSTM neural network performs dimensionality reduction processing on acceleration signals, velocity signals and displacement signals through convolution kernels of different sizes, and then transmits them to the LSTM intelligent control system to generate MFC actuator control force and apply it to the controlled structure, thereby realizing rapid data collection, analysis and vibration control; in addition, the neural network intelligent control system of the present invention does not need to establish a mathematical model of the controlled structure, nor does it need the parameters of the controlled structure itself, such as the mass, inertia, geometric dimensions, stiffness, etc. of the controlled structure, and has strong self-adaptation ability and can adapt to complex vibration working condition changes.

[0078] The embodiment of the present invention further provides a computer device, Figure 2 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention; see the accompanying drawings Figure 2As shown, the computer device includes: an input device 23, an output device 24, a memory 22 and a processor 21; the memory 22 is used to store one or more programs; when the one or more programs are executed by the one or more processors 21, the one or more processors 21 implement the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as provided in the above embodiment; wherein the input device 23, the output device 24, the memory 22 and the processor 21 can be connected through a bus or other means, Figure 2 The example of connecting through bus is taken in the following.

[0079] The memory 22 is a readable and writable storage medium of a computing device, which can be used to store software programs and computer executable programs, such as program instructions corresponding to the neural network intelligent control method based on piezoelectric fiber driven vibration reduction described in the embodiment of the present invention; the memory 22 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required for a function; the data storage area can store data created according to the use of the device, etc.; in addition, the memory 22 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device; in some instances, the memory 22 can further include a memory remotely arranged relative to the processor 21, and these remote memories can be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0080] The input device 23 may be used to receive input digital or character information, and to generate key signal input related to user settings and function control of the device; the output device 24 may include a display device such as a display screen.

[0081] The processor 21 executes various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 22, that is, realizes the above-mentioned neural network intelligent control method based on piezoelectric fiber drive vibration reduction.

[0082] The computer device provided above can be used to execute the active control method of controlling MFC drive by the CNN-LSTM algorithm provided in the above embodiment, and has corresponding functions and beneficial effects.

[0083] The embodiment of the present invention also provides a storage medium containing computer executable instructions, which are used to execute the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as provided in the above embodiment when executed by a computer processor. The storage medium is any of various types of memory devices or storage devices, and the storage medium includes: installation media, such as CD-ROM, floppy disk or tape device; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (such as hard disk or optical storage); registers or other similar types of memory elements, etc.; the storage medium may also include other types of memory or a combination thereof; in addition, the storage medium may be located in the first computer system in which the program is executed, or may be located in a different second computer system, the second computer system is connected to the first computer system via a network (such as the Internet); the second computer system may provide program instructions to the first computer for execution. The storage medium includes two or more storage media that can reside in different locations (for example, in different computer systems connected via a network). The storage medium can store program instructions (for example, specifically implemented as a computer program) that can be executed by one or more processors.

[0084] Of course, the computer executable instructions of a storage medium containing computer executable instructions provided in an embodiment of the present invention are not limited to the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described in the above embodiment, and can also execute related operations in the neural network intelligent control method based on piezoelectric fiber driven vibration reduction provided in any embodiment of the present invention.

[0085] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A neural network intelligent control method based on piezoelectric fiber driven vibration reduction, characterized in that: The following steps are involved: S1: The LQR algorithm is used to calculate the MFC response signal and control force of the MFC actuator arranged on the controlled structure, and the CNN-LSTM neural network is trained as training data; S2: Apply the well-trained CNN-LSTM neural network to the neural network intelligent control module, and directly map the control current signal required by the current controller according to the acceleration signal, velocity signal and displacement signal output by the signal acquisition module; S3: transmitting the control current signal to a current controller, the current controller generates a corresponding current, and drives the MFC actuator after being amplified by a high-voltage amplifier; S4: The MFC actuator generates a control force to suppress the vibration of the controlled structure, thereby reducing the vibration response of the controlled structure in real time.

2. The neural network intelligent control method based on piezoelectric fiber driven vibration reduction according to claim 1 is characterized in that: The method of obtaining the displacement, velocity, and acceleration response of the position of the MFC actuator in step S1 includes: The acceleration signal of the MFC actuator (6) is acquired through an acceleration sensor (1), the acceleration signal is filtered and conditioned through a signal acquisition module, and a velocity signal and a displacement signal are acquired through integration.

3. The neural network intelligent control method based on piezoelectric fiber driven vibration reduction according to claim 1 is characterized in that: The training method of the CNN-LSTM neural network includes: The MFC response signal and the external excitation are used as the input values ​​of the CNN-LSTM neural network, and the control force of the MFC actuator (6) at the next moment is used as the output value of the CNN-LSTM neural network. The deep learning network framework of the CNN-LSTM neural network is trained and tested to obtain a mature CNN-LSTM neural network.

4. The neural network intelligent control method based on piezoelectric fiber driven vibration reduction according to claim 3 is characterized in that: The verification method of the deep learning network framework of the CNN-LSTM neural network includes: By inputting external excitation into the maturely trained CNN-LSTM neural network, the output value of the CNN-LSTM neural network is obtained and compared with the theoretically calculated control force to verify the effectiveness of the deep learning network framework of the CNN-LSTM neural network. The prediction process loss function is expressed as the mean square error MSE: Where: MSE refers to mean square error, N is the database size; , are the measured value and predicted value at time i respectively.

5. The neural network intelligent control method based on piezoelectric fiber driven vibration reduction according to claim 4 is characterized in that: The deep learning network framework of the CNN-LSTM neural network includes: a sequence input layer, a sequence folding layer, a convolution layer Conv1, a batch normalization layer, a discard layer Dropout1, an activation layer Relu1, a maximum pooling layer Pool1, a convolution layer Conv2, a discard layer Dropout2, an activation layer Relu2, a maximum pooling layer Pool2, a sequence expansion layer, a flattening layer, an Lstm layer Lstm1, a discard layer Dropout3, an Lstm layer Lstm2, a discard layer Dropout4, a fully connected layer and a regression layer.

6. The neural network intelligent control method based on piezoelectric fiber drive vibration reduction according to claim 1 is characterized in that: The MFC response signal includes: A combination of one or more of acceleration signals, velocity signals and displacement signals.

7. A neural network intelligent control system based on piezoelectric fiber driven vibration reduction, characterized in that: A neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described in any one of claims 1 to 6, comprising: an acceleration sensor (1), a signal acquisition module (2), a neural network intelligent control module (3), a current controller (4), a high voltage amplifier (5), an MFC actuator (6), and a controlled structure (7); The acceleration sensor (1) and the MFC actuator (6) are both arranged on the surface of the controlled structure (7); the acceleration sensor (1) is used to obtain an acceleration signal of the MFC actuator (6); the acceleration sensor (1) is electrically connected to a signal acquisition module (2) and transmits an acceleration signal to the signal acquisition module (2); The signal acquisition module (2) filters and conditions the acceleration signal, and obtains a velocity signal and a displacement signal by integration; the signal acquisition module (2) is electrically connected to the neural network intelligent control module (3), and transmits the acceleration signal, velocity signal and displacement signal to the neural network intelligent control module (3); The neural network intelligent control module (3) is electrically connected to the current controller (4); the neural network intelligent control module (3) directly maps an output control current signal based on the signal transmitted by the signal acquisition module (2) as the currently input MFC response signal, and transmits the control current signal to the current controller (4); The current controller (4) is electrically connected to the high-voltage amplifier (5), and the high-voltage amplifier (5) is electrically connected to the MFC actuator (6). The current controller (4) amplifies the control current signal through the high-voltage amplifier (5) to drive the MFC actuator (6) to generate a control force acting on the controlled structure (7), thereby suppressing the vibration of the controlled structure (7).

8. The neural network intelligent control system based on piezoelectric fiber drive vibration reduction according to claim 7 is characterized in that: The controlled structure (7) is also provided with a plurality of acceleration sensors (1).

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the neural network intelligent control method based on piezoelectric fiber driven vibration reduction described in any one of claims 1 to 6 are implemented.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the neural network intelligent control method based on piezoelectric fiber driven vibration reduction as described in any one of claims 1 to 6 are implemented.

Images