Linear quadratic form control method for vibration recognition

Abstract

The invention relates to the technical field of industrial automation, in particular to a linear quadratic form control method for vibration recognition. According to the method, vibration data, controlled device data and corresponding linear quadratic form control parameter data are collected, the collected data are preprocessed, subsequent convolutional neural network training is facilitated, the reliability and accuracy of training are ensured, a corresponding virtual controlled model is constructed according to a controlled device, and the training efficiency is improved. Controlling the vibration of the virtual controlled model by adopting a linear quadratic form, training a convolutional neural network, optimizing control parameters, controlling a controlled device by taking the obtained optimized control parameters as parameters of linear quadratic form control, and finally updating the control parameters until the control parameters meet a set vibration threshold value; according to the method provided by the invention, optimal control under different vibration inputs can be realized by updating the control parameters, and the use vibration control requirements of the controlled device are met.

Description

Technical Field

[0001] The invention relates to the technical field of industrial automation, and in particular to a linear quadratic control method for vibration identification. Background Art

[0002] Vibration control refers to a series of technical measures to reduce, suppress or change the vibration in mechanical systems, structures or equipment. Vibration is harmful in many cases. For example, in precision machining, the vibration of machine tools can lead to a decrease in machining accuracy, increase the surface roughness of the workpiece, and exceed the dimensional accuracy. In the aerospace field, excessive vibration of aircraft structures may affect flight safety and cause fatigue damage to components. In the construction field, the vibration of buildings can affect their comfort and safety. For example, the vibration of high-rise buildings under strong winds or earthquakes may make residents feel uncomfortable or even endanger the safety of the structure.

[0003] Based on this, the present invention provides a linear quadratic control method for vibration identification to solve the above-mentioned technical problems. Summary of the invention

[0004] The purpose of the present invention is to provide a linear quadratic control method for vibration identification. The provided linear quadratic control method for vibration identification can achieve optimal control under different vibration inputs by updating control parameters, thereby meeting the vibration control requirements of the controlled device.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a linear quadratic control method for vibration identification, comprising the following steps:

[0007] S1, collecting vibration data, controlled device data and corresponding linear quadratic control parameter data, and preprocessing the collected data;

[0008] S2. constructing a corresponding virtual controlled model according to the controlled device, and using a linear quadratic form to control the vibration of the virtual controlled model;

[0009] S3, train the convolutional neural network and optimize the control parameters;

[0010] S4, using the obtained optimized control parameters as parameters of linear quadratic control to control the controlled device;

[0011] S5. Repeat steps S1 to S5 to update the control parameters to meet the set vibration threshold.

[0012] The present invention is further configured as follows: in step S1, the preprocessing process is as follows:

[0013] Screening of collected data;

[0014] Filter the filtered data;

[0015] Then the filtered data is normalized. In the formula, x min is the minimum value in the data set, x max is the maximum value in the data set, and x is the original data in the data set;

[0016] And classify and label the data according to the type of data.

[0017] The present invention is further configured such that: the screening is to eliminate data irrelevant to vibration.

[0018] The present invention is further configured as follows: in the step S2, it also includes constructing a data acquisition comparator for the virtual controlled model.

[0019] The present invention is further configured as follows: in step S3, the process of training the convolutional neural network is as follows:

[0020] Splitting the vibration data into data groups corresponding to the corresponding periods;

[0021] The data group in the corresponding period is used as the input of the convolutional neural network, and the convolutional neural network is input to train the convolutional neural network.

[0022] The present invention is further configured as follows: after completing the training of the convolutional neural network, the trained convolutional neural network is verified, and the process is as follows:

[0023] The vibration recognition performance of the convolutional neural network was evaluated using the test set data. The evaluation formula is: accuracy = number of correctly identified samples / total number of samples × 100%;

[0024] Compare the obtained accuracy with the preset threshold;

[0025] If the obtained accuracy is greater than the preset threshold, the training of the convolutional neural network is completed, otherwise, the convolutional neural network is retrained.

[0026] The present invention is further configured as follows: the convolutional neural network uses a cross entropy loss function for the classification of vibration recognition, wherein: In the formula, G is the number of categories, y i is the real classification mark, is the predicted probability.

[0027] The present invention is further configured as follows: in step S3, the process of optimizing the control parameters is as follows:

[0028] Input the real-time collected vibration data into the trained convolutional neural network to obtain the recognition result of the vibration pattern;

[0029] According to the identified vibration mode, taking the actual parameters of the controlled device under the vibration mode as the cardinality parameters, obtaining the adjustment parameter combination set of the cardinality parameters, and taking different adjustment parameter combination sets as the linear quadratic control parameters of the virtual controlled model respectively to control the vibration of the virtual controlled model;

[0030] Obtain the vibration data of the virtual controlled model, compare it using the data acquisition comparator, and optimize the control parameters.

[0031] The present invention is further configured as follows: the process of using the data acquisition comparator to compare and optimize the control parameters is as follows:

[0032] Comparing the acquired vibration data of the virtual controlled model with the vibration data collected in real time;

[0033] Filter out the vibration data of the virtual controlled model that is better than the vibration data collected in real time, and the corresponding adjustment parameter combination set;

[0034] The adjustment parameter combination set is used as the parameters of the linear quadratic control, that is, the optimized control parameters.

[0035] The present invention is further configured as follows: in the step S5, the updated control parameters are also stored.

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

[0037] The present invention collects vibration data, controlled device data and corresponding linear quadratic control parameter data, and pre-processes the collected data to facilitate subsequent convolutional neural network training to ensure the reliability and accuracy of the training, then constructs a corresponding virtual controlled model according to the controlled device, adopts a linear quadratic control to control the vibration of the virtual controlled model, trains the convolutional neural network, and optimizes the control parameters, uses the obtained optimized control parameters as the parameters of the linear quadratic control, controls the controlled device, and finally updates the control parameters to meet the set vibration threshold; the linear quadratic control method for vibration identification provided by the present invention can achieve optimal control under different vibration inputs by updating the control parameters, and meet the vibration control requirements of the controlled device; the linear quadratic control method for vibration identification provided by the present invention has broad application prospects and is worthy of promotion. DETAILED DESCRIPTION

[0038] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Example:

[0040] This embodiment provides a linear quadratic control method for vibration recognition, comprising the following steps:

[0041] S1. Collect vibration data, controlled device data and corresponding linear quadratic control parameter data, and pre-process the collected data.

[0042] The preprocessing process is as follows:

[0043] Screening of collected data;

[0044] Filter the filtered data;

[0045] Then the filtered data is normalized. In the formula, x min is the minimum value in the data set, x max is the maximum value in the data set, and x is the original data in the data set;

[0046] And classify and label the data according to the type of data.

[0047] Furthermore, screening is to eliminate data that are not related to vibration.

[0048] In this embodiment, it should be noted that the data filtering method includes but is not limited to moving average filtering, low-pass filtering, etc. As an implementation method, the moving average filtering is to calculate the average value of the data within a certain window size as the filtered value. If the window size is set to 5, then the filter value for the third data point is the average value of the two data points before it, itself, and the two data points after it (a total of 5 data points). This embodiment can facilitate the subsequent convolutional neural network training by preprocessing the data to ensure the reliability and accuracy of the training.

[0049] S2. Construct a corresponding virtual controlled model according to the controlled device, and use a linear quadratic form to control the vibration of the virtual controlled model.

[0050] It also includes building a data acquisition comparator for the virtual controlled model.

[0051] In this embodiment, it should be noted that the constructed virtual controlled model is used for subsequent quadratic control parameter optimization, and a data acquisition comparator is used to compare the vibration data of the virtual controlled model obtained subsequently with the vibration data collected in real time, so as to obtain an adjustment parameter combination set.

[0052] S3. Train the convolutional neural network and optimize the control parameters.

[0053] Among them, the process of training convolutional neural network is as follows:

[0054] Splitting the vibration data into data groups corresponding to the corresponding periods;

[0055] The data group in the corresponding period is used as the input of the convolutional neural network, and the convolutional neural network is input to train the convolutional neural network.

[0056] Furthermore, after completing the training of the convolutional neural network, the trained convolutional neural network is verified, and the process is as follows:

[0057] The vibration recognition performance of the convolutional neural network was evaluated using the test set data. The evaluation formula is: accuracy = number of correctly identified samples / total number of samples × 100%;

[0058] Compare the obtained accuracy with the preset threshold;

[0059] If the obtained accuracy is greater than the preset threshold, the training of the convolutional neural network is completed, otherwise, the convolutional neural network is retrained.

[0060] In addition, the convolutional neural network uses the cross entropy loss function for the classification of vibration recognition, where In the formula, G is the number of categories, y i is the real classification mark, is the predicted probability.

[0061] The process of optimizing control parameters is as follows:

[0062] Input the real-time collected vibration data into the trained convolutional neural network to obtain the recognition result of the vibration pattern;

[0063] According to the identified vibration mode, taking the actual parameters of the controlled device under the vibration mode as the cardinality parameters, obtaining the adjustment parameter combination set of the cardinality parameters, and taking different adjustment parameter combination sets as the linear quadratic control parameters of the virtual controlled model respectively to control the vibration of the virtual controlled model;

[0064] Obtain the vibration data of the virtual controlled model, compare it using the data acquisition comparator, and optimize the control parameters.

[0065] The process of using the data acquisition comparator to compare and optimize the control parameters is as follows:

[0066] Comparing the acquired vibration data of the virtual controlled model with the vibration data collected in real time;

[0067] Filter out the vibration data of the virtual controlled model that is better than the vibration data collected in real time, and the corresponding adjustment parameter combination set;

[0068] The adjustment parameter combination set is used as the parameters of the linear quadratic control, that is, the optimized control parameters.

[0069] In this embodiment, it should be noted that by modifying the number of convolutional layers and pooling layers in the convolutional neural network model to change the model architecture, different convolutional neural network models are obtained, and the accuracy is used as a measure of the classification effect of the convolutional neural network model to select the convolutional neural network with the best comprehensive training result. In this embodiment, according to the identified vibration mode, the actual parameters of the controlled device under the vibration mode are used as the cardinality parameters to obtain the adjustment parameter combination set of the cardinality parameters, and different adjustment parameter combination sets are used as the linear quadratic control parameters of the virtual controlled model to control the vibration of the virtual controlled model, that is, the cardinality parameters are used as the origin, and the cardinality parameters are scaled to obtain a set of corresponding scaling parameters, which are used as the parameters of the linear quadratic control.

[0070] S4. Using the obtained optimized control parameters as parameters of linear quadratic control to control the controlled device.

[0071] In this embodiment, it should be noted that the vibration control effect is verified respectively through the obtained adjustment parameter combination set, and the corresponding vibration data is obtained for subsequent data collection and parameter update.

[0072] S5. Repeat steps S1 to S5 to update the control parameters to meet the set vibration threshold.

[0073] In addition, the method also includes storing the updated control parameters.

[0074] In this embodiment, it should be noted that when the acquired vibration data meets the set vibration threshold, the updating of the control parameters is stopped, and the obtained control parameters are the control parameters of the linear quadratic control, otherwise steps S1 to S5 are repeated.

[0075] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0076] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A linear quadratic control method for vibration identification, characterized in that: The following steps are involved: S1, collecting vibration data, controlled device data and corresponding linear quadratic control parameter data, and preprocessing the collected data; S2. constructing a corresponding virtual controlled model according to the controlled device, and using a linear quadratic form to control the vibration of the virtual controlled model; S3, train the convolutional neural network and optimize the control parameters; S4, using the obtained optimized control parameters as parameters of linear quadratic control to control the controlled device; S5. Repeat steps S1 to S5 to update the control parameters to meet the set vibration threshold.

2. The linear quadratic control method for vibration identification according to claim 1, characterized in that: In step S1, the pre-processing process is as follows: Screening of collected data; Filter the filtered data; Then the filtered data is normalized. In the formula, x min is the minimum value in the data set, x max is the maximum value in the data set, and x is the original data in the data set; And classify and label the data according to the type of data.

3. The linear quadratic control method for vibration identification according to claim 2, characterized in that: The screening is to eliminate data irrelevant to vibration.

4. The linear quadratic control method for vibration identification according to claim 1, characterized in that: In the step S2, it also includes constructing a data acquisition comparator of the virtual controlled model.

5. The linear quadratic control method for vibration identification according to claim 1, characterized in that: In step S3, the process of training the convolutional neural network is as follows: Splitting the vibration data into data groups corresponding to the corresponding periods; The data group in the corresponding period is used as the input of the convolutional neural network, and the convolutional neural network is input to train the convolutional neural network.

6. The linear quadratic control method for vibration identification according to claim 1, characterized in that: In step S3, the process of optimizing the control parameters is as follows: Input the real-time collected vibration data into the trained convolutional neural network to obtain the recognition result of the vibration pattern; According to the identified vibration mode, taking the actual parameters of the controlled device under the vibration mode as the cardinality parameters, obtaining the adjustment parameter combination set of the cardinality parameters, and taking different adjustment parameter combination sets as the linear quadratic control parameters of the virtual controlled model respectively to control the vibration of the virtual controlled model; The vibration data of the virtual controlled model is obtained and compared using a data acquisition comparator to optimize control parameters.

7. The linear quadratic control method for vibration identification according to claim 6, characterized in that: The process of using the data acquisition comparator to compare and optimize the control parameters is as follows: Comparing the acquired vibration data of the virtual controlled model with the vibration data collected in real time; Filter out the vibration data of the virtual controlled model that is better than the vibration data collected in real time, and the corresponding adjustment parameter combination set; The adjustment parameter combination set is used as the parameters of the linear quadratic control, that is, the optimized control parameters.

8. The linear quadratic control method for vibration identification according to claim 1, characterized in that: In the step S5, the updated control parameters are also stored.

Images