ZAMAK is a family of alloys with a primary base metal of zinc and alloying elements of aluminium, magnesium and copper. Zamak alloys are part of the zinc aluminium alloy family. ZAMAK alloys are commonly used for Zinc Die Casting applications.It is applied to castings that do not require high mechanical strength, such as toys, lamps, decorations, and some electrical components. Die Casting is a metal casting process characterized by the use of a mold cavity to apply high pressure to the molten metal. Molds are usually machined from higher strength alloys, some of which are similar to injection molding.
Zamak Die Casting,Pressure Die Casting,High Pressure Die Casting,Low Pressure Die Casting Chongqing Jin Te Rui Machine Co.,Ltd , https://www.cncjtr.com
1 Introduction. Modeling identification and control of nonlinear systems is an important application direction. It is clear that for nonlinear systems, traditional analytical methods can only be applied to specific applications without a universally applicable method. Artificial neural networks provide excellent knowledge of nonlinear system modeling with its excellent nonlinear mapping approximation ability and self-learning ability. It can also adjust parameters according to a given data set to obtain a good fuzzy model. In recent years, how to integrate the fuzzy system through the network has many forms. 314. We can use some neuro-fuzzy systems to model the control objects in order to obtain better results than the simple application techniques. In this paper, we will model a nonlinear motor system using an adaptive neuro-fuzzy inference system based on the model of the Takagi Taki model. The experimental results were compared to a back propagation network.
In this study, the Chinese Academy of Sciences key project fund support, the section gives the experimental results and analysis of the model; finally give the conclusion.
2,5 Introduction by Miao said, the adaptive network-based fuzzy inference system proposed by Ton Ton, also known as adaptive neuro-fuzzy inference system, is a fuzzy inference system based on Takagi Sagano model, Takagi SugenoModel. When the fuzzy set adopts the non-trapezoidal nonlinear membership function, the Dingye 35å©63 fuzzy system is more economical than the 4ca 13 fuzzy system, that is, the required fuzzy rules and the number of input fuzzy sets are less.
2.1 The structure of six å²3 is simple Ding 38816 with two rules, the fuzzy system six 18 structure is as follows: the corresponding ANFIS structure l only the signal flow direction, no weight is associated with it; square node with adjustable parameters Node, a circular node with no tunable nodes. From the parameters. The function of each layer is as follows: the first layer obscures the input variable, and outputs the membership degree of the corresponding fuzzy set. The transfer function of each node can be according to the form of the selected membership function, and the corresponding parameter set can be obtained, which is called the condition parameter. . For example, the usual membership function is a set of conditional parameters for all.
The second layer implements the operation of the fuzzy set of the conditional part, and outputs the applicability of each rule corresponding to the formula 1, usually by multiplication.
Level 3 normalizes the applicability of each rule. The transfer function of each node of CO layer 4 is a linear function, a local linear model, and the output of each rule is calculated.
The set of parameters consisting of all, 6, is called the conclusion parameter.
The fifth layer calculates the sum of the outputs of all the rules. From the input-output relationship of the above network, it can be seen that the network is completely equivalent to the fuzzy inference system of the formula. The learning of fuzzy inference system comes down to the adjustment of the nonlinear parameters of conditional parameters and the linear parameters of the conclusion parameters. 2.2 Hybrid learning algorithm For all parameters, the gradient descent based backpropagation algorithm can be used to adjust the parameters; however, the hybrid algorithm can be used. Improve the speed of learning. The conditional parameters in the hybrid algorithm still use the backpropagation algorithm, while the conclusion parameters use the linear least squares estimation algorithm to adjust the parameters. The conclusion part can adopt the least-multiply estimation algorithm because the final output of ANFIS is f-conclusion. The linearity of l is still ff. The output is in each iteration of the hybrid learning algorithm. The input signal is along the network forward. Passing until the fourth layer, at this time fixed condition parameters, the minimum multiplication estimation algorithm is used to adjust the conclusion parameters; then, the signal continues to pass along the network forward until the output layer is the fifth layer. Thereafter, the obtained error signal is propagated back in the network so that the condition parameters can be adjusted.
Using the hybrid learning algorithm, for the given condition parameters, the global best of the conclusion parameters can be obtained, which not only can reduce the dimension of the search space in the gradient method, but also can greatly improve the convergence speed of the parameters. A detailed description of the hybrid learning algorithm is referenced 51.
3 Nonlinear motor system modeling Below, we model a DC motor system with nonlinear friction effects. The actual control system used to collect input and output signals includes a table, 6 payouts, 1200 microcomputer blocks built into the computer's 12-bit, conversion board power amplifier circuit DC torque motor and DC tachogenerator for speed feedback. The analog voltage input range and output control voltage range are both 5,5 volts, and the digital range after analog-to-digital conversion is 2048, 2048. For convenience, the input and output units are digital. The model of the controlled system includes a collection of all components except the computer.
At a sampling period of 5 milliseconds, the input signal lasts 10 seconds, ie 2000 sample periods. In order to adapt the model to different frequencies, the training input signal is a composite of multiple frequency component sinusoidal signals.
The 9-type signal is used as the input of the actual system to obtain the real output of the motor, that is, the speed signal. Actual motor input and output 2.
3.1 Modeling based on 15 In Section 2, the structure and algorithm of 5 have been introduced, but the initialization of the parameters in 18 is not mentioned, and the division of the input space is determined, thereby determining the number of fuzzy rules. Usually the input space can be divided by the average segmentation method or Tian Kan, the Julong method first, the pseudo motor model is the order system, that is, the input and output relationship can be used for the average segmentation method, and the input ranges for 1 and 1 will be the same. It is divided into three parts, and the fuzzy membership function is Gaussian. Thus, the two inputs form a total of nine combinations, that is, there are nine fuzzy rules. According to the description in Section 2, it can be seen that there are 12 32+32 conditional parameters and 27 conclusions. The error index is the mean square error MSE of the actual output and the model output. After 100 iterations, the ANFIS identification results are 3 , the mean square error of the training results, = 4.9854.
3.2 Identification of the identification model Firstly, a multi-frequency component sinusoidal signal with changes in amplitude and frequency is also used as the test signal to test the nonlinear model. The test result 4, the mean square error is 5=6.5836.
Second, we test the two typical nonlinear characteristics of the modeled system with saturation and dead zone. We use two single-frequency sinusoidal signals as test signals 123 and 12, which are input to the model and input to the actual system. The output and input signals obtained from the actual system and the built test model are respectively 56.
It can be seen from the output of the actual system and the output of the model. The sum of the squares of the errors at 2000 sampling points is less than 7, but the accuracy is high. It can be seen that the actual output and the model output almost coincide with each other.
It can be seen from the above test that the established ANFIS model can not only adapt to the changes of amplitude and frequency, but also can well contain the nonlinear characteristics of the motor system. The established model has strong adaptability to the change of amplitude and frequency within the range of the training signal, and achieves the purpose of dynamic and accurate modeling.
4 summarize the dynamic model. By comparing with Japan, the main advantage of the description 15 is that the convergence speed is very fast, and the convergence time only needs to be 8 tenths. When faced with a complex system, the parameters required for modeling increase sharply, and the convergence speed is crucial. At this time, Can 15 showed greater superiority than the 8 measurement. At the same time, in the analysis of the experimental results, it is clarified that ANFIS guarantees faster convergence speed than BPNN from both initialization and learning algorithms. In addition, force. In practical applications. ANFIS provides a powerful tool for modeling identification and time series analysis of nonlinear systems.
Zhao Zhenyu Xu Yongyu, the basis and application of fuzzy theory and neural network. Tsinghua University Press, 996.6.
University Press, 3337, 1998.7.
Modeling of Nonlinear Motor System Based on ANFIS
Next Article
How to apply for a special car on the card
Prev Article
Improved design of YG-94 (051A) chain saw gearbox