CN114779650A - Electromagnetic vibration table harmonic control system and method based on neural network inverse model - Google Patents
Electromagnetic vibration table harmonic control system and method based on neural network inverse model Download PDFInfo
- Publication number
- CN114779650A CN114779650A CN202210684067.6A CN202210684067A CN114779650A CN 114779650 A CN114779650 A CN 114779650A CN 202210684067 A CN202210684067 A CN 202210684067A CN 114779650 A CN114779650 A CN 114779650A
- Authority
- CN
- China
- Prior art keywords
- neural network
- vibration table
- vibration
- electromagnetic vibration
- inverse model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
- Y02E40/40—Arrangements for reducing harmonics
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to the technical field of vibration harmonic control, in particular to an electromagnetic vibration table harmonic control system and method based on a neural network inverse model. The system is based on a neural network inverse model, an electromagnetic vibration table is excited by setting input quantity, an input voltage signal and an output vibration signal are processed to obtain a neural network inverse system training sample, the output of the trained neural network inverse model is connected with the input of the electromagnetic vibration table in series, and accurate tracking of the output vibration signal of the electromagnetic vibration table on an expected signal of the input neural network inverse model is achieved. The control system has the advantages of simple composition structure, simple and convenient operation process, wide applicable frequency range and higher control precision on harmonic waves of vibration signals output by the electromagnetic vibration table.
Description
Technical Field
The invention relates to the technical field of vibration harmonic control, in particular to a harmonic control system and method for an electromagnetic vibration table based on a neural network inverse model.
Background
Under the normal condition, the harmonic control of the vibration acceleration signals output by the electromagnetic vibration table for vibration measurement sensor calibration or product vibration environment simulation test can be realized by constructing a closed-loop feedback control system or an amplitude precision adjustment system based on successive approximation.
The implementation process of the closed-loop feedback control system usually needs to be provided with a high-precision vibration signal detection sensor with high price, and the closed-loop feedback control system also has the technical problems of high stability requirement, high parameter adjustment difficulty and the like. The vibration signal amplitude adjusting system based on the successive approximation method needs to drive the electromagnetic vibration table to generate a vibration signal with a certain random amplitude at first, and then corrects the input amplitude of the electromagnetic vibration table successively based on the deviation between the random amplitude and the target amplitude until the vibration signal amplitude is stabilized in the target deviation range. Obviously, the number of operation cycles and the time consumption required by the vibration signal amplitude adjustment process based on the successive approximation system have great randomness. In particular, considering that the detection time of the vibration signal gradually increases with the decrease of the frequency, the adjustment of the amplitude of the low-frequency vibration signal by the successive approximation process will be very slow, and the working efficiency of the electromagnetic vibration table will be seriously affected.
In order to overcome the problems of the methods, in recent years, researchers with relevant researches adopt some intelligent methods to construct a high-precision model of a system and apply precise control to the model, but the current intelligent methods cannot adapt to the control target requirements under a load state aiming at an electromagnetic vibration table with a complex input-output structure and load parameter variation.
Disclosure of Invention
The invention provides an electromagnetic vibration table harmonic control system and method based on a neural network inverse model, aiming at the problems of complicated operation process, low efficiency, limited application range and the like of a traditional control and adjustment system for outputting vibration acceleration signal harmonic waves by an electromagnetic vibration table and aiming at realizing open-loop intelligent control on a load parameter change electromagnetic vibration table.
The invention adopts the following technical scheme:
a harmonic control system of an electromagnetic vibration table based on a neural network inverse model comprises a program control signal source, an electromagnetic vibration table system, a vibration detection unit, a data acquisition unit and the neural network inverse model; the program control signal source is used for generating an input vibration signal; the electromagnetic vibration table system comprises a power amplifier and an electromagnetic vibration table; the vibration detection unit comprises a vibration acceleration sensor and an amplifier or an adapter; the data acquisition unit comprises a data acquisition card and a computer; the neural network inverse model is obtained by processing an output vibration acceleration signal and input vibration signal data of the electromagnetic vibration table by data processing software installed in a computer to form a neural network training sample set and identifying the neural network training sample set.
A harmonic control method of an electromagnetic vibration table based on a neural network inverse model comprises the following steps:
step 1, obtaining a variable load electromagnetic vibration table neural network inverse model group:
1.1 in order to make the electromagnetic shaking table meet the requirement of accurately controlling and outputting the harmonic wave of vibration acceleration signal under different loads, have proposed the load segmentation equivalent method, divide the range of the electromagnetic shaking table drivable load into several small sections evenly and regard every section of load intermediate value as the representative value of this section of load, mount the vibration acceleration transducer of the selected load value on the electromagnetic shaking table sequentially, the electromagnetic shaking table is equivalent to representing the electromagnetic shaking table model group under different load segmentations under the working condition of different loads;
1.2 taking the model of the electromagnetic vibration table under the first section of load as an example to identify the neural network inverse model, intensively selecting characteristic points in a set test frequency range and a vibration displacement range of the electromagnetic vibration table, selecting corresponding test frequency points and displacement points according to the characteristic points, generating input vibration signals of each frequency point and displacement stroke by a program control signal source, and driving the electromagnetic vibration table to generate vibration acceleration signals of each frequency point and displacement stroke after being amplified by a power amplifier; the lower limit value of the test frequency range is 0.1 time of the lower limit frequency of the electromagnetic vibration table to which the waveform distortion control needs to be applied, and the upper limit value of the test frequency range is 10 times of the upper limit frequency of the electromagnetic vibration table to which the waveform distortion control needs to be applied; the vibration displacement range is a rated vibration displacement range of the electromagnetic vibration table;
1.3, detecting by a vibration acceleration sensor of a vibration detection unit, generating output vibration acceleration signals of an electromagnetic vibration table corresponding to displacement strokes at each frequency point by an amplifier or an adapter, synchronously acquiring the output vibration acceleration signals and input vibration signals generated by a program control signal source by a data acquisition card, and then sending the output vibration acceleration signals and the input vibration signals to a computer;
1.4 further obtaining an output vibration velocity signal and an output vibration displacement signal corresponding to a first-order integral and a second-order integral of the output vibration acceleration signal in a computer, establishing a neural network training sample set for an output layer by taking the output vibration acceleration signal, the output vibration velocity signal and the output vibration displacement signal as input layers and the input vibration signal as the output layer, carrying out normalization processing on the training sample set, carrying out identification training on a neural network through a neural network identification tool, and obtaining a neural network inverse model of the load segmented electromagnetic vibration table;
1.5, setting the precision requirement of the neural network inverse model as lambda, verifying the identification precision of the neural network inverse model by randomly selecting data in a training sample set, adjusting a neural network training algorithm and changing a neural network transmission function, so that the neural network inverse model meets the precision requirement;
1.6 selecting an electromagnetic vibration table model under a second section of load, identifying a corresponding neural network inverse model according to a neural network inverse model identification method corresponding to the electromagnetic vibration table under the first section of load, and so on, identifying the neural network inverse models corresponding to all the load section electromagnetic vibration tables, so that the electromagnetic vibration table model groups correspond to the neural network inverse model groups one by one;
step 2, harmonic control of the variable-load electromagnetic vibration table based on the neural network inverse model:
2.1 determining the load section of the electromagnetic vibration table load in the step 1, and selecting an electromagnetic vibration table model and a neural network inverse model corresponding to the load section;
2.2, the output of the neural network inverse model is connected with the input of the electromagnetic vibration table in series, and the expected output vibration acceleration, speed and displacement signals of the electromagnetic vibration table are input into the neural network inverse model through a program control signal source to obtain the input vibration signals of the electromagnetic vibration table; then, inputting the input vibration signal into an electromagnetic vibration table, and exciting the electromagnetic vibration table to output a vibration acceleration signal;
2.3, calculating the output vibration acceleration signals obtained in the step 2.2 by adopting harmonic analysis software to obtain the frequency, amplitude, phase and corresponding waveform distortion value of fundamental frequency and harmonic components of each order contained in the signals;
2.4 setting the maximum allowable value of the waveform distortion degree of the output acceleration signal of the electromagnetic vibration table as alpha, judging whether the waveform distortion degree value of the output vibration acceleration signal is less than or equal to alpha, and if so, indicating that the neural network inverse model controller meets the control requirement; if not, executing the step 1.2-1.5 to re-identify the neural network inverse model corresponding to the load segment, increasing the data volume of the training sample set by refining the feature points, increasing the number of hidden layers and hidden layer neurons of the neural network, when the accuracy of identifying the neural network inverse model reaches lambda/10, completing the identification of the new neural network inverse model corresponding to the load segment, continuing to execute the step 2.2-2.3, if the waveform distortion value after the re-control is not less than or equal to alpha, setting the accuracy of the neural network inverse model to 1/10 of the last identification, continuing to further refine the feature points, increasing the data volume of the training sample set, increasing the number of hidden layers and hidden layer neurons of the neural network, re-training the new neural network inverse model and re-executing the step 2.2-2.3, repeating the steps until the waveform distortion value of the output vibration acceleration signal is less than or equal to alpha, and finishing final identification of the load section neural network, so that the harmonic control of the electromagnetic vibration table in the load section meets the requirement.
The control system of the invention realizes that the harmonic wave of the vibration acceleration signal output by the electromagnetic vibration table is controlled to be within a set range in the whole frequency band needing to be applied with the harmonic wave control of the output vibration acceleration signal, and when the harmonic wave control in a certain load section does not reach the set requirement, the more accurate neural network inverse model under the corresponding load is re-identified, thereby realizing the higher-precision harmonic wave control under the iterative control.
Generally, the control system has the advantages of simple composition structure, simple and convenient operation process and wide applicable frequency range, can realize open-loop intelligent control on the output vibration acceleration signal harmonic waves of the load parameter change electromagnetic vibration table, and has higher control precision and control efficiency.
Drawings
FIG. 1 is a flow chart of the harmonic control of the present invention;
FIG. 2 is a schematic diagram of the training and identification of the inverse model set of the variable load neural network using the harmonic control system of the electromagnetic vibration table in the present invention.
Detailed Description
Taking the example that a vibration acceleration sensor is adopted in a vibration detection unit to detect an output vibration acceleration signal of an electromagnetic vibration table, as shown in fig. 2, the electromagnetic vibration table harmonic control system based on the neural network inverse model comprises a program control signal source, an electromagnetic vibration table system, a vibration detection unit, a data acquisition unit and a neural network inverse model; the neural network inverse model is obtained by MATLAB based on training sample set identification; the electromagnetic vibration table system comprises a power amplifier and an electromagnetic vibration table; the vibration detection unit comprises a vibration acceleration sensor and an amplifier or an adapter; the data acquisition unit comprises a data acquisition card and a computer.
The method comprises the steps that a program control signal source generates an expected output vibration acceleration signal, the expected output vibration acceleration signal is input into a neural network inverse model corresponding to a corresponding load to obtain an input vibration signal of an electromagnetic vibration table, the input vibration signal is amplified by a power amplifier and then drives the electromagnetic vibration table to generate an output vibration acceleration signal to be input into a computer, harmonic analysis software in the computer calculates amplitude and phase of fundamental frequency and harmonic of each order in the output vibration acceleration signal to obtain a corresponding waveform distortion degree, whether the waveform distortion degree of the output acceleration signal meets the requirement of being less than or equal to the set waveform distortion degree or not is judged, and if the waveform distortion degree meets the requirement, the control requirement of a neural network inverse model control system is met. If not, re-identifying the neural network inverse model until the waveform distortion degree value of the output acceleration signal is smaller than the set waveform distortion degree requirement.
The program control signal source is a sine signal generator and can be controlled by computer software to output vibration signals with set frequency, amplitude and phase.
In order to realize the harmonic control of the vibration acceleration signal output by the electromagnetic vibration table, firstly, a training identification schematic diagram of a neural network inverse model shown in fig. 2 is used for identifying and obtaining a neural network inverse model group corresponding to different load sections of the electromagnetic vibration table with variable load parameters, and the specific steps are as follows:
(a) the drivable load range of the electromagnetic vibration table is uniformly divided into x sections, the middle value of each section of load is used as the representative value of the section of load, vibration acceleration sensors with selected load values are sequentially installed on the electromagnetic vibration table, and the electromagnetic vibration table under different load working states is equivalent to represent electromagnetic vibration table model groups under different sections.
(b) The method comprises the steps of taking an electromagnetic vibration table model under a first section of load as an example to identify a neural network inverse model of the electromagnetic vibration table, intensively selecting characteristic points in a set test frequency range and a vibration displacement range of the electromagnetic vibration table, selecting corresponding test frequency points and displacement points according to the characteristic points, generating input vibration signals of each frequency point and displacement stroke by a program control signal source, and driving the electromagnetic vibration table to generate vibration acceleration signals of each frequency point and displacement stroke after amplification by a power amplifier. The lower limit value of the test frequency range is 0.1 time of the lower limit frequency of the waveform distortion control to be applied to the electromagnetic vibration table, and the upper limit value of the test frequency range is 10 times of the upper limit frequency of the waveform distortion control to be applied to the electromagnetic vibration table; the vibration displacement range is a rated vibration displacement range of the electromagnetic vibration table.
(c) The vibration acceleration sensor of the vibration detection unit detects the vibration acceleration and the amplifier or the adapter generates output vibration acceleration signals of the electromagnetic vibration table corresponding to the displacement stroke at each frequency point, the data acquisition card synchronously acquires the output vibration acceleration signals and input vibration signals generated by the program control signal source, and then the output vibration acceleration signals and the input vibration signals are sent to the computer.
(d) Further obtaining an output vibration speed signal and an output vibration displacement signal corresponding to a first-order integral and a second-order integral of the output vibration acceleration signal in a computer, forming a neural network training sample set by taking the output vibration acceleration signal, the output vibration speed signal and the output vibration displacement signal as input layers and the input vibration signal as an output layer, carrying out normalization processing on the training sample set, and carrying out identification training on a neural network through a neural network identification tool to obtain a neural network inverse model of the load-segmented electromagnetic vibration table.
(e) The precision requirement of the neural network inverse model is set to be lambda, the identification precision of the neural network inverse model is verified by randomly selecting data in the training sample set, the neural network training algorithm is adjusted, and the neural network transmission function is changed, so that the neural network inverse model meets the precision requirement.
(f) And selecting a model of the electromagnetic vibration table under the second section of load, identifying the corresponding neural network inverse model according to the identification method of the neural network inverse model corresponding to the electromagnetic vibration table under the first section of load, and so on, identifying the neural network inverse models corresponding to the electromagnetic vibration tables of all the load sections, so that the model groups of the electromagnetic vibration table correspond to the neural network inverse model groups one by one.
As shown in fig. 1, the specific steps of the vibration harmonic control process of the neural network inverse model group of the electromagnetic vibration table based on load parameter variation are as follows:
1) selecting a corresponding neural network inverse model: and positioning the load section to which the actual load on the electromagnetic vibration table belongs, and selecting an electromagnetic vibration table model and a neural network inverse model corresponding to the load section.
2) Obtaining an input vibration signal: and the output of the neural network inverse model is connected with the input of the electromagnetic vibration table in series, and the expected output vibration acceleration, speed and displacement signals of the electromagnetic vibration table are input into the neural network inverse model through a program control signal source to obtain the input vibration signals of the electromagnetic vibration table.
3) Generating an output vibration acceleration signal: after the input vibration signal is amplified by the power amplifier, the electromagnetic vibration table is driven to generate an output vibration acceleration signal and the output vibration acceleration signal is input into a computer.
4) Detecting and outputting a vibration acceleration signal: the amplitude and phase of fundamental frequency and each order harmonic in the output vibration acceleration signal and the corresponding waveform distortion value are obtained by calculation of harmonic analysis software in a computer.
5) Judging whether the distortion degree meets the requirement: setting the maximum allowable value of the waveform distortion degree of the output acceleration signal of the electromagnetic vibration table as alpha, judging whether the waveform distortion degree value of the output vibration acceleration signal meets the condition that the waveform distortion degree value is less than or equal to alpha, and if so, indicating that the neural network inverse model controller meets the control requirement; if not, executing the steps b-e to re-identify the neural network inverse model corresponding to the load segment, simultaneously increasing the data volume of the training sample set by thinning the feature points, and increasing the number of hidden layers and the number of neurons of the neural network, when the accuracy of identifying the neural network inverse model reaches lambda/10, completing the identification of the new neural network inverse model corresponding to the load segment, continuously executing the steps 2-4, if the waveform distortion value after the re-control is not less than or equal to alpha, setting the accuracy of the neural network inverse model to be 1/10 of the last identification, simultaneously continuously thinning the feature points to increase the data volume of the training sample set, increasing the number of hidden layers and the number of neurons of the hidden layers of the neural network, re-training the new neural network inverse model and executing the steps 2-4 again, repeating the steps until the waveform distortion value of the output vibration acceleration signal is less than or equal to alpha, and finishing final identification of the load section neural network, so that the harmonic control of the electromagnetic vibration table in the load section meets the requirement.
The control system of the invention realizes that the harmonic wave of the vibration acceleration signal output by the electromagnetic vibration table is controlled to be within a set range in the whole frequency band needing to be applied with the harmonic wave control of the output vibration acceleration signal, and when the harmonic wave control in a certain load section does not reach the set requirement, the more accurate neural network inverse model under the corresponding load is re-identified, thereby realizing the higher-precision harmonic wave control under the iterative control.
Generally, the control system has the advantages of simple composition structure, simple and convenient operation process and wide applicable frequency range, can realize open-loop intelligent control on the output vibration acceleration signal harmonic waves of the load parameter change electromagnetic vibration table, and has higher control precision and control efficiency.
Claims (2)
1. A harmonic control system of an electromagnetic vibration table based on a neural network inverse model is characterized in that: the system comprises a program control signal source, an electromagnetic vibration table system, a vibration detection unit, a data acquisition unit and a neural network inverse model; the program control signal source is used for generating an input vibration signal; the electromagnetic vibration table system comprises a power amplifier and an electromagnetic vibration table; the vibration detection unit comprises a vibration acceleration sensor and an amplifier or an adapter; the data acquisition unit comprises a data acquisition card and a computer; the neural network inverse model is obtained by processing an output vibration acceleration signal and input vibration signal data of the electromagnetic vibration table by data processing software installed in a computer to form a neural network training sample set and identifying the neural network training sample set.
2. A harmonic control method of an electromagnetic vibration table based on a neural network inverse model is characterized by comprising the following steps: comprises the following steps:
step 1, obtaining a variable load electromagnetic vibration table neural network inverse model group:
1.1 in order to make the electromagnetic vibration table meet the requirement of accurately controlling and outputting vibration acceleration signal harmonic waves under different loads, a load subsection equivalence method is provided, a load-drivable range of the electromagnetic vibration table is evenly divided into a plurality of small sections, a middle value of each section of load is used as a representative value of the section of load, vibration acceleration sensors with selected load values are sequentially installed on the electromagnetic vibration table, and the electromagnetic vibration table under different load working states is equivalent to represent an electromagnetic vibration table model group under different load subsections;
1.2 identifying a neural network inverse model of the electromagnetic vibration table under the first section of load by taking the electromagnetic vibration table model as an example, densely selecting characteristic points in a set test frequency range and a vibration displacement range of the electromagnetic vibration table, selecting corresponding test frequency points and displacement points according to the characteristic points, generating input vibration signals of each frequency point and displacement stroke by a program control signal source, and driving the electromagnetic vibration table to generate vibration acceleration signals of each frequency point and displacement stroke after amplification by a power amplifier; the lower limit value of the test frequency range is 0.1 time of the lower limit frequency of the electromagnetic vibration table to which the waveform distortion control needs to be applied, and the upper limit value of the test frequency range is 10 times of the upper limit frequency of the electromagnetic vibration table to which the waveform distortion control needs to be applied; the vibration displacement range is a rated vibration displacement range of the electromagnetic vibration table;
1.3 detecting by a vibration acceleration sensor of a vibration detection unit, generating an output vibration acceleration signal of the electromagnetic vibration table corresponding to a displacement stroke at each frequency point by an amplifier or an adapter, synchronously acquiring the output vibration acceleration signal and an input vibration signal generated by a program control signal source by a data acquisition card, and then sending the output vibration acceleration signal and the input vibration signal to a computer;
1.4 further solving an output vibration speed signal and an output vibration displacement signal corresponding to a first-order integral and a second-order integral of the output vibration acceleration signal in a computer, establishing a neural network training sample set for an output layer by taking the output vibration acceleration signal, the output vibration speed signal and the output vibration displacement signal as input layers and taking the input vibration signal as the output layer, carrying out normalization processing on the training sample set, and carrying out identification training on a neural network through a neural network identification tool to obtain a neural network inverse model of the electromagnetic vibration table with the load segment;
1.5, setting the precision requirement of the neural network inverse model as lambda, verifying the identification precision of the neural network inverse model by randomly selecting data in a training sample set, adjusting a neural network training algorithm and changing a neural network transmission function, so that the neural network inverse model meets the precision requirement;
1.6 selecting a second section of electromagnetic vibration table model under load, identifying the corresponding neural network inverse model according to the neural network inverse model identification method corresponding to the electromagnetic vibration table under the first section of load, and so on, identifying the neural network inverse models corresponding to all the load section electromagnetic vibration tables, so that the electromagnetic vibration table model groups correspond to the neural network inverse model groups one by one;
step 2, harmonic control of the variable-load electromagnetic vibration table based on the neural network inverse model:
2.1, determining the load section of the electromagnetic vibration table load in the step 1, and selecting an electromagnetic vibration table model and a neural network inverse model corresponding to the load section;
2.2, the output of the neural network inverse model is connected with the input of the electromagnetic vibration table in series, and the expected output vibration acceleration, speed and displacement signals of the electromagnetic vibration table are input into the neural network inverse model through a program control signal source to obtain the input vibration signals of the electromagnetic vibration table; then, inputting the input vibration signal into an electromagnetic vibration table, and exciting the electromagnetic vibration table to output a vibration acceleration signal;
2.3, calculating the output vibration acceleration signals obtained in the step 2.2 by adopting harmonic analysis software to obtain the frequency, amplitude, phase and corresponding waveform distortion value of fundamental frequency and harmonic components of each order contained in the signals;
2.4 setting the maximum allowable value of the waveform distortion degree of the output acceleration signal of the electromagnetic vibration table as alpha, judging whether the waveform distortion degree value of the output vibration acceleration signal meets the condition that the waveform distortion degree value is less than or equal to alpha, and if so, indicating that the neural network inverse model controller meets the control requirement; if not, executing the step 1.2-1.5 to re-identify the neural network inverse model corresponding to the load segment, simultaneously increasing the data volume of the training sample set by thinning the characteristic points, and increasing the number of the hidden layers and the number of the neurons of the neural network, when the accuracy of identifying the neural network inverse model reaches lambda/10, completing the identification of the new neural network inverse model corresponding to the load segment, continuously executing the step 2.2-2.3, if the waveform distortion value after the re-control is not less than or equal to alpha, setting the accuracy of the neural network inverse model to 1/10 of the last identification, simultaneously continuing to further thin the characteristic points, increasing the data volume of the training sample set, increasing the number of the hidden layers and the number of the neurons of the hidden layers of the neural network, re-training the new neural network inverse model and executing the step 2.2-2.3 again, repeating the steps until the waveform distortion value of the output vibration acceleration signal is less than or equal to alpha, and finishing final identification of the load section neural network, so that the harmonic control of the electromagnetic vibration table in the load section meets the requirement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210684067.6A CN114779650B (en) | 2022-06-17 | 2022-06-17 | Electromagnetic vibration table harmonic control system and method based on neural network inverse model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210684067.6A CN114779650B (en) | 2022-06-17 | 2022-06-17 | Electromagnetic vibration table harmonic control system and method based on neural network inverse model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114779650A true CN114779650A (en) | 2022-07-22 |
| CN114779650B CN114779650B (en) | 2022-09-02 |
Family
ID=82422042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210684067.6A Active CN114779650B (en) | 2022-06-17 | 2022-06-17 | Electromagnetic vibration table harmonic control system and method based on neural network inverse model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114779650B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2019526A (en) * | 1978-04-19 | 1979-10-31 | Hughes Aircraft Co | Controlling Vibrations of Vibration-Testing Shaker Tables |
| CN101741318A (en) * | 2009-12-31 | 2010-06-16 | 中国航空工业集团公司北京长城计量测试技术研究所 | Method for controlling degree of distortion of sine vibrating table |
| CN101917163A (en) * | 2010-07-29 | 2010-12-15 | 大连理工大学 | Method for improving electrohydraulic vibration exciting control waveform of non-sinusoidal periodic signal |
| CN109062036A (en) * | 2018-06-12 | 2018-12-21 | 太原理工大学 | Vibration harmonics iterative control system based on transmission function |
| CN111975976A (en) * | 2020-08-17 | 2020-11-24 | 太原理工大学 | Method for synchronously controlling vibration forming acceleration of ballastless track slab |
-
2022
- 2022-06-17 CN CN202210684067.6A patent/CN114779650B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2019526A (en) * | 1978-04-19 | 1979-10-31 | Hughes Aircraft Co | Controlling Vibrations of Vibration-Testing Shaker Tables |
| CN101741318A (en) * | 2009-12-31 | 2010-06-16 | 中国航空工业集团公司北京长城计量测试技术研究所 | Method for controlling degree of distortion of sine vibrating table |
| CN101917163A (en) * | 2010-07-29 | 2010-12-15 | 大连理工大学 | Method for improving electrohydraulic vibration exciting control waveform of non-sinusoidal periodic signal |
| CN109062036A (en) * | 2018-06-12 | 2018-12-21 | 太原理工大学 | Vibration harmonics iterative control system based on transmission function |
| CN111975976A (en) * | 2020-08-17 | 2020-11-24 | 太原理工大学 | Method for synchronously controlling vibration forming acceleration of ballastless track slab |
Non-Patent Citations (2)
| Title |
|---|
| YAO,JIANJUN等: "Harmonic identification based on RBF neural network", 《2016 35TH CHINESE CONTROL CONFERENCE (CCC)》 * |
| 胡天恩等: "簧片式电磁振动台波形失真特性分析", 《振动与冲击》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114779650B (en) | 2022-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11630929B2 (en) | Linearization identification method for hysteresis model of piezoceramics based on Koopman operators | |
| CN111931411B (en) | Duhem dynamic hysteresis modeling method for piezoelectric driving micro-positioning platform | |
| CN106125574B (en) | Piezoelectric ceramics mini positioning platform modeling method based on DPI model | |
| CN106707760A (en) | Nonlinear inverse control method used for dynamic hysteresis compensation of piezoelectric actuator | |
| CN110245430B (en) | Improved Bouc-Wen model lag modeling method | |
| CN104991997B (en) | The broad sense rate correlation P-I hysteresis modeling methods of adaptive differential evolution algorithm optimization | |
| CN106292550B (en) | The Servocontrol device of function with vehicle air-conditioning gain | |
| CN109372832B (en) | Energy consumption optimization method for bivariate hydraulic system under working condition change | |
| CN101741318A (en) | Method for controlling degree of distortion of sine vibrating table | |
| CN109284541A (en) | A kind of more Method of Physical Modeling of neural network for microwave passive component | |
| CN114779650B (en) | Electromagnetic vibration table harmonic control system and method based on neural network inverse model | |
| CN112861429A (en) | Wind turbine engine room transfer function calculation method | |
| CN115793461A (en) | Seismic simulation vibrating table three-parameter control parameter setting method based on deep learning | |
| CN108845594B (en) | Vibration harmonic iterative control method based on transfer function | |
| CN106406100A (en) | Rotor dynamic balancing control system based on fuzzy self-tuning single neure PID control and method thereof | |
| CN116771655B (en) | Intelligent control system and intelligent control method for water pump | |
| CN106563645A (en) | Intelligent piezoelectric film sensor sorting method based on tensor decomposition | |
| CN104834271B (en) | Axle servo parameter optimization method is directly fed based on dynamic stiffness evaluation | |
| CN104571086B (en) | Temperature controller emulation test method based on transmission function | |
| CN109062036B (en) | Vibration harmonic iterative control system based on transfer function | |
| CN107944552A (en) | A kind of industrial Internet of Things parameter prediction method based on Elman neutral nets | |
| US7870180B2 (en) | Method and apparatus for nonlinear signal and data aggregation | |
| CN114690642A (en) | Coupling control method of piezoelectric driving system | |
| CN108875268A (en) | A kind of design method of the linear motor PID controller based on attenuation curve method | |
| CN106324279B (en) | A kind of real-time tracing method of ultrasonic aerovane cavity resonance frequency |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |