CN115648227B - Robot motion trajectory neural network fuzzy control optimization method - Google Patents

Abstract

The present invention proposes a neural network fuzzy control optimization method for robot motion trajectory, including: step S1, generating a robot motion reference trajectory that meets the continuity requirement

; Step S2, designing a neural network fuzzy controller, wherein the neural network fuzzy controller includes: fuzzification, fuzzy reasoning and defuzzification three parts, and then the robot motion reference trajectory in the step 1

input into the neural network fuzzy controller; Step S3, optimize the adjustment parameters of the neural fuzzy controller by genetic algorithm, so as to realize the rapid adjustment of the controller parameters and realize the optimization.

Description

A neural network fuzzy control optimization method for robot motion trajectory

Technical Field

The invention relates to the technical field of industrial robots, and in particular to a neural network fuzzy control optimization method for robot motion trajectory.

Background Art

In the past few decades, industrial robot control has been mostly based on the principle of three-loop controllers. In order to improve the output accuracy of robot control systems, a variety of robot control systems have emerged. Among them, the neural network fuzzy controller has a better control effect on the robot output error. In the past, most tests were conducted under no-load conditions of the neural network fuzzy controller. Later, it was found that when moving under load conditions, it was easily disturbed by load factors, resulting in a large output error of the robot.

Summary of the invention

The object of the present invention is to solve at least one of the technical drawbacks.

To this end, the purpose of the present invention is to propose a neural network fuzzy control optimization method for robot motion trajectory.

In order to achieve the above object, an embodiment of the present invention provides a robot motion trajectory neural network fuzzy control optimization method, comprising the following steps:

；Step S1, generate a robot motion reference trajectory that meets the continuity requirements

;

输入至所述神经网络模糊控制器中；Step S2, designing a neural network fuzzy controller, wherein the neural network fuzzy controller includes three parts: fuzzification, fuzzy reasoning and defuzzification, and then converting the robot motion reference trajectory in step S1 into

Input into the neural network fuzzy controller;

Step S3, optimizing the adjustment parameters of the neuro-fuzzy controller by using a genetic algorithm to achieve rapid adjustment of the controller parameters and achieve optimization.

为：Furthermore, in step S1, the robot motion trajectory

for:

为参考轨迹;q₀ 、q'₀ 、q_f 、q'_f 为初始和最终位置和速度；a₀、 a₁、 a₂、 a₃ 分别是三阶多项式的系数。Among them, t ₀ and t _f are the initial and final movement moments;

is the reference trajectory; q ₀ , q' ₀ , q _f , q' _f are the initial and final positions and velocities; a ₀ , a ₁ , a ₂ , a ₃ are the coefficients of the third-order polynomial respectively.

进行输入归一化增益、模糊子集形式、模糊推理、去模糊化和输出增益，然后传输至工业机器人，得到工业机器人的运动轨迹q。Further, in step S2, the input robot motion reference trajectory

The input normalized gain, fuzzy subset form, fuzzy reasoning, defuzzification and output gain are performed and then transmitted to the industrial robot to obtain the motion trajectory q of the industrial robot.

Furthermore, in step S3, six parameters are optimized using a genetic algorithm, including two fuzzified input normalization gains and four defuzzified output gains of decentralized regulators, and the objective function is the absolute sum of all joint errors on the entire trajectory.

According to the robot motion trajectory neural network fuzzy control optimization method of the embodiment of the present invention, the neural network fuzzy control algorithm and the genetic algorithm are used to optimize the controller parameters. When the robot is loaded, after the genetic algorithm is used for optimization, the robot motion joint angular displacement tracking error is significantly reduced, and the genetic algorithm can be used to optimize the controller adjustment parameters, thereby quickly adjusting the controller parameters to achieve the purpose of reducing the error; the calculation process is less time-consuming and has high real-time performance.

Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and/or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

1 is a flow chart of a method for optimizing a robot motion trajectory through a neural network fuzzy control according to an embodiment of the present invention;

FIG2 is a schematic diagram of a method for optimizing a robot motion trajectory using a neural network fuzzy control method according to an embodiment of the present invention;

FIG3 is a structural diagram of a regulator in the form of a neural network according to an embodiment of the present invention.

DETAILED DESCRIPTION

Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

The present invention proposes a robot motion trajectory neural network fuzzy control optimization method, which uses a genetic algorithm to optimize the neural network fuzzy controller and improves the robot control accuracy under load conditions by selecting appropriate optimization parameters and objective functions.

As shown in FIG. 1 and FIG. 2 , the robot motion trajectory neural network fuzzy control optimization method according to the embodiment of the present invention comprises the following steps:

。Step S1, generate a robot motion reference trajectory that meets the continuity requirements

.

In this step, the trajectory is generated by a third-order polynomial to ensure the continuity requirement of the robot.

为：Robot motion trajectory

for:

为参考轨迹;q₀ 、q'₀ 、q_f 、q'_f 为初始和最终位置和速度；a₀、 a₁、 a₂、 a₃ 分别是三阶多项式的系数。Among them, t ₀ and t _f are the initial and final movement moments;

is the reference trajectory; q ₀ , q' ₀ , q _f , q' _f are the initial and final positions and velocities; a ₀ , a ₁ , a ₂ , a ₃ are the coefficients of the third-order polynomial respectively.

输入至神经网络模糊控制器中。Step S2, designing a neural network fuzzy controller, wherein the neural network fuzzy controller includes three parts: fuzzification, fuzzy reasoning and defuzzification, and then converting the robot motion reference trajectory in step S1 into

Input into the neural network fuzzy controller.

进行输入归一化增益、模糊子集形式、模糊推理、去模糊化和输出增益，然后传输至工业机器人，得到工业机器人的运动轨迹q。Specifically, the input robot motion reference trajectory

The input normalized gain, fuzzy subset form, fuzzy reasoning, defuzzification and output gain are performed and then transmitted to the industrial robot to obtain the motion trajectory q of the industrial robot.

和误差变化

作为输入，以及每个关节的驱动力矩

作为输出。The neuro-fuzzy inference system method is used to adjust the controller parameters by online learning of each joint to obtain good control performance. The structure of the regulator in the form of a neural network is shown in Figure 3. It contains the continuous error of the i-th joint

and error variation

As input, and the driving torque of each joint

as output.

与

是关节误差和误差变化，第一列A₁,A₂,B₁,B₂对应归一化增益函数，第二列对应模糊子集的计算，第三列对应模糊推理，第四列对应去模糊化处理，第5列对应输出增益计算，即输出关节力矩。Specifically, Figure 3 is a specific implementation diagram of neural fuzzy control.

and

are the joint errors and error changes. The first column A ₁ , A ₂ , B ₁ , B ₂ corresponds to the normalized gain function, the second column corresponds to the calculation of fuzzy subsets, the third column corresponds to fuzzy reasoning, the fourth column corresponds to defuzzification processing, and the fifth column corresponds to the output gain calculation, that is, the output joint torque.

Step S3, using genetic algorithm to optimize the adjustment parameters of the neuro-fuzzy controller to achieve rapid adjustment of the controller parameters and achieve optimization.

与q的差值，然后通过神经模糊控制器的5个步骤输出机器人关节力矩，这5个步骤分别是计算归一化增益，构建模糊子集，进行模糊推理计算，进行去模糊计算，计算输出增益。其中计算归一化增益，构建模糊子集，计算输出增益这三步有可调节参数参与计算，因此与遗传算法相关，即图中所述。Referring to Figure 2, the input of the entire system is the joint position error, and

The difference between q and q is then output through the five steps of the neuro-fuzzy controller, which are to calculate the normalized gain, construct the fuzzy subset, perform the fuzzy reasoning calculation, perform the defuzzification calculation, and calculate the output gain. Among them, the three steps of calculating the normalized gain, constructing the fuzzy subset, and calculating the output gain have adjustable parameters involved in the calculation, so they are related to the genetic algorithm, as described in the figure.

According to the robot motion trajectory neural network fuzzy control optimization method of the embodiment of the present invention, the neural network fuzzy control algorithm and the genetic algorithm are used to optimize the controller parameters. When the robot is loaded, after the genetic algorithm is used for optimization, the robot motion joint angular displacement tracking error is significantly reduced, and the genetic algorithm can be used to optimize the controller adjustment parameters, thereby quickly adjusting the controller parameters to achieve the purpose of reducing the error; the calculation process is less time-consuming and has high real-time performance.

In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" 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.

Although the embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. Those skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention without departing from the principles and purpose of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims (1)

1. A robot motion trajectory neural network fuzzy control optimization method is characterized by comprising the following steps:

step S1, generating a robot motion reference track meeting the continuity requirement

(ii) a The motion track of the robot

Comprises the following steps:

(ii) a Wherein, t ₀ 、t _f For an initial and final movement moment;)>

Is a reference track q ₀ 、q' ₀ 、q _f 、q' _f Initial and final positions and velocities; a is a ₀ 、 a ₁ 、 a ₂ 、 a ₃ Are the coefficients of a third order polynomial, respectively;

s2, designing a neural network fuzzy controller, wherein the neural network fuzzy controller comprises: fuzzification, fuzzy inference and defuzzification, and then referring the motion reference track of the robot in the step S1

Inputting the data into the neural network fuzzy controller; based on the input robot movement reference track->

Performing input normalization gain, fuzzy subset form, fuzzy reasoning, defuzzification and output gain, and then transmitting to the industrial robot to obtain a motion trail q of the industrial robot; the method adopts a neural fuzzy inference system method, and adjusts the parameters of the controller by learning each joint on line so as to obtain good control performance; regulator architecture in the form of a neural network comprising successive errors/of the i-th joint>

And error variation

As input, and the drive torque of each joint>

As an output; input->

And/or>

Is the joint error and error variation, first column A ₁ ,A ₂ ,B ₁ ,B ₂ Corresponding to a normalized gain function, calculating a fuzzy subset corresponding to a second column, performing fuzzy reasoning corresponding to a third column, performing defuzzification processing corresponding to a fourth column, and calculating output gain corresponding to a 5 th column, namely outputting joint torque;

s3, optimizing the adjustment parameters of the neural fuzzy controller by using a genetic algorithm to realize rapid adjustment of the parameters of the controller and realize optimization; wherein 6 parameters optimized using a genetic algorithm include: the objective function of the 2 blurred input normalized gains and the deblurred output gains of the 4 dispersion adjusters is the absolute sum of all joint errors over the entire trajectory.

Images