CN107807530A - A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm - Google Patents
A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm Download PDFInfo
- Publication number
- CN107807530A CN107807530A CN201711232473.4A CN201711232473A CN107807530A CN 107807530 A CN107807530 A CN 107807530A CN 201711232473 A CN201711232473 A CN 201711232473A CN 107807530 A CN107807530 A CN 107807530A
- Authority
- CN
- China
- Prior art keywords
- fuzzy
- algorithm
- neural network
- function
- hydraulic
- 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.)
- Pending
Links
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
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 discloses a kind of forcing press PID control system based on intelligent fuzzy neural network algorithm, straight side press is one strong time variation, nonlinear Large Hydraulic System, accurate mathematical modeling can not be obtained, so in general pid control algorithm is difficult to obtain preferable control effect.The present invention proposes a kind of intelligent fuzzy neural network algorithm to optimize the PID closed-loop controls of forcing press, wherein introduce the partial parameters that improved rapid particle swarm algorithm carrys out Optimization of Fuzzy neural network algorithm to optimize again, so that the overall performance of system is remarkable, certain reference can be provided by building emulation platform with the validity of verification algorithm for the improvement control of forcing press.
Description
Technical Field
The invention relates to a press PID control system based on an intelligent fuzzy neural network algorithm.
Background
In the prior art, an electro-hydraulic proportional servo control system is composed of an analog channel input and output element, a speed displacement sensor 9, a hydraulic cylinder 8, a pilot valve 6, a main valve and the like, as shown in fig. 1. The principle of the hydraulic control system is that a simulation channel collects speed and position information transmitted by a sensor and transmits the speed and position information to a CPU, the CPU controls a proportional electromagnet of a proportional pilot valve through various algorithms through a D/A module so as to control the opening of the main valve and control the hydraulic oil pressure of a hydraulic system, and the proportional electromagnet of a directional control valve controls the speed and position of a hydraulic cylinder in the same way.
Utilize the electro-hydraulic proportional control system can output accurate control system flow and pressure, can set up parameter such as speed, acceleration, stroke by oneself, more traditional press control system can realize more high-efficient control fast. Generally, PID closed-loop control is adopted to control the hydraulic oil, but the nonlinearity of a system, the hysteresis saturation of an electromagnetic valve, the compressibility of hydraulic oil and the like make the traditional PID control difficult to achieve the expected ideal control effect. Many scholars and experts also make a lot of researches on the method, such as a self-tuning control algorithm, and the algorithm usually needs a lot of data to support optimization of system parameters to obtain a good control effect, so that a lot of waste products are generated, and the performance and evaluation of a machine are seriously influenced; for another example, an adaptive control algorithm needs to accurately model a system, and the system is modeled by using perfect mathematical reasoning, but the actual system is often difficult to realize the modeling, and firstly, the modeling is often not realistic because different systems have different control requirements on different processes, and secondly, the coupling and nonlinearity of the system per se. The invention provides an intelligent fuzzy neural network algorithm, which utilizes the memory characteristics of a neural network to automatically select a control strategy of each control stage, wherein the fuzzy algorithm reduces the time-varying property and the coupling property of a system, and the self-learning function of the neural network can quickly and accurately adjust PID parameters in real time.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a press PID control system based on an intelligent fuzzy neural network algorithm.
In order to solve the technical problems, the invention provides the following technical scheme:
the invention relates to a press PID control system based on an intelligent fuzzy neural network algorithm, which is designed by the following steps:
1. designing a hydraulic control system model of a press machine:
q L ＝K q x V -K c p L (1)
wherein q is L The flow rate of the four-side slide valve; k c Pressure coefficients for different flow rates; k q The gain flow for the four-sided spool valve; p L Load pressure drop for the press; a. The p The sectional area of a hydraulic cylinder of the hydraulic system; x is the number of p The displacement of the piston in the hydraulic cylinder; c tp The leakage coefficient of the interior of the hydraulic cylinder is obtained; beta is a e Is the modulus of elasticity of the piston; v t Is the total compression volume of the hydraulic cylinder; m is a unit of t The piston of the hydraulic cylinder and the mass loaded on the piston; b is p Is the viscosity coefficient; k is the hooke coefficient of the load spring; f L Extra load force for external interference;
2. designing an intelligent fuzzy neural network algorithm
For an electro-hydraulic proportional control system, the primary purpose is to output stable and accurate hydraulic oil pressure, and timely flow control parameters are obtained, wherein the control on the hydraulic pressure is mainly researched, so that e is the control deviation ec of the pressure and is the change rate of the pressure;
the pressure of the hydraulic system at the moment nT is e (n), and the control is carried outDeviation, i.e. output of set quantity P REF The deviation from the feedback amount is ec (n);
using the fuzzy system to optimize parameters, firstly, the input quantity is fuzzified, here, the quantization level of the input link is defined as 7, which is described as the fuzzy language:
domain of E: [ -1,1]
Discourse domain of EC: [ -1,1]
E fuzzy rule: { NB, NM, NS, ZO, PS, PM, PB }
Fuzzy rule of EC: { NB, NM, NS, ZO, PS, PM, PB }
The number of nodes of the corresponding third tier is N =7 × 7= 49;
for the obfuscation layer, a node corresponds to an obfuscation language, such as NM, NS, ZO, etc., mentioned above; and (3) calculating each linguistic variable by using a Gaussian function as an evaluation standard to obtain the membership degree of each input variable, and then performing fuzzy reasoning:
wherein σ ij Width representing the membership of the ith input variable to the jth fuzzy aggregation function, c ij Representing that i input variables belong to the center position of the jth fuzzy set function; i =1,2, \8230;, n; j =1,2, \8230, L;
next, in the fuzzy inference layer, a node corresponds to a fuzzy inference rule, and in order to calculate the fitness of each rule to its respective node, each fuzzy rule is paired, that is, each node is a node in the fuzzy inference layer
In the formula (I), the compound is shown in the specification,wherein N is i Representing the fuzzy division number which is the fuzzy division number of the input i;
f 4 is k is p 、k i 、k d The setting result of (a), namely the output layer, is:
in the formula, W is a connection weight matrix, and i =1,2,3 is arranged between the fourth layer output layer and the third layer fuzzy inference layer; the controller is as follows:
Δu(k)＝f 4 ·xc＝k p xc(1)+k i xc(2)+k d xc(3) (8)
wherein: k is a radical of p ＝f 4 (1)，k i ＝f 4 (2)，k d ＝f 4 (3)，xc(1)＝e(k)，xc(2)＝e(k)-e(k-1)，xc(3)＝e(k)-2e(k-1)+e(k-2)；
An incremental PID algorithm which is easy to list expressions is selected as an algorithm of the hydraulic control system of the invention:
u(k)＝u(k-1)-Δu(k) (9)
an evaluation function is selected to score the control effect, and the accumulation of the deviation of the control quantity can be generally selected as an evaluation standard:
ec (t) is the hydraulic system pressure change rate; t is an integration time variable; t is t c The time when the algorithm is finished, namely the integral upper limit time;
get the connection right W ij Center function c ij And width function sigma ij Comprises the following steps:
wherein eta is the learning rate of the fuzzy neural network algorithm; α is an inertia coefficient, 0< α <1;
as the nonlinearity and the coupling ratio of the electro-hydraulic proportional control system of the press are stronger, and the simple optimization of PID parameters by the fuzzy neural network cannot achieve good effect, the invention introduces the particle swarm algorithm to c of the fuzzy neural network ij 、σ ij And the parameters eta and alpha are optimized again; different from the traditional particle swarm velocity and displacement formula, the velocity and displacement formula of the invention is as follows:
v i,j (t+1)＝wv i,j (t)+c 1 r 1 [p i,j -x i,j (t)]+c 2 r 2 [p g,j -x i,j (t)] (14)
x i,j (t+1)＝x i,j (t)+v i,j (t+1),j＝1,...,d (15)
r 1 、r 2 is a random number and takes values between (0, 1); i represents a particle numbered i; t is the iteration number of the population; d is the parameter to be optimized, and the dimension of the solution space in the algorithm; c. C 1 、c 2 Acceleration of a single particle and the entire population, respectively; for w is the improved decreasing power exponent inertial weight:
w＝w min +(w max -w min )×e -4Δ (16)
wherein the content of the first and second substances,f is an objective function value obtained after the algorithm is iterated; f. of avg Is the average of all particle objective functions of this iteration, f min Is the minimum value thereof;
3. designing an intelligent fuzzy neural network press PID controller:
step.1, initializing an algorithm, wherein fourteen-dimensional particles are endowed with respective initial values, which are local extreme values, the algorithm does not start iteration, a global extreme value cannot be obtained, and a fourteen-dimensional space is divided into two e and ec which respectively occupy seven dimensions; randomly initializing a Gaussian membership function, i.e. a central function c 0 And width function sigma 0 Selecting [0,0.5 ]]Assigning a random number therebetween to η and α;
step.2, running an algorithm, and carrying out online optimization according to the iterative algorithms (5) - (9) and (11) - (13) to obtain an output PID optimization result u of the fuzzy neural network;
step.3, calculating a central function c corresponding to each one-dimensional particle 0 And width function sigma 0 Fitness evaluation function f 3 ；
Step.4, optimally selecting the global optimal solution and the individual optimal solution to obtain a particle with the minimum fitness evaluation function, and replacing the speed and the displacement in the original formula with the speed and the displacement of the particle;
step.5 real-time optimization of the central value c in the fuzzy neural network using equations (14) to (15) ij And width σ ij ；
And step 6, calculating an evaluation function E of the system, screening the minimum eta and alpha of the target function, returning to the step 2, and repeating the algorithm until the specified iteration times are reached.
The invention has the following beneficial effects:
the closed type press is a large hydraulic system with strong time variability and nonlinearity, and an accurate mathematical model cannot be obtained, so that an ideal control effect cannot be obtained by a general PID control algorithm. The invention provides an intelligent fuzzy neural network algorithm to optimize the PID closed-loop control of the press, wherein an improved rapid particle swarm algorithm is introduced to optimize partial parameters optimized by the fuzzy neural network algorithm, so that the overall performance of the system is excellent, the effectiveness of the algorithm can be verified by building a simulation platform, and a certain reference is provided for the improved control of the press.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a diagram of a prior art electro-hydraulic proportional control system for a press;
FIG. 2 is a diagram of a RBF fuzzy neural network architecture;
FIG. 3 is a graph of Griewank function fitness versus evolutionary algebra curves;
FIG. 4 is a graph of the initial control of the constant pressure side force of the intelligent neural network and the conventional PID;
FIG. 5 is a control diagram of the intelligent neural network and the self-correcting PID controller for small disturbance.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it should be understood that they are presented herein only to illustrate and explain the present invention and not to limit the present invention.
Model of hydraulic control system of press machine:
the hydraulic cylinder, the hydraulic valve and the load determine the dynamic state of the electro-hydraulic proportional hydraulic cylinder, and the system is nonlinear, so that the dynamic equation in the instantaneous micro state is considered, and the state of the system in the instantaneous state is basically in a linear state. Assuming that the hydraulic valve is a zero-opening four-side slide valve, and four throttle valves are symmetrically distributed at four corners; the hydraulic oil pipeline between the hydraulic cylinder and the valve is assumed to be large enough and symmetrically distributed, so that the hydraulic pipeline is basically in an ideal state, and the internal pressure and instantaneous dynamics can be basically ignored; the hydraulic cylinder is only force balanced with the load. This results in a total of three hydrodynamic and kinetic equations:
q L ＝K q x V -K c p L (1)
wherein q is L The flow rate of the four-side slide valve; k is c Pressure coefficients for different flow rates; k q The gain flow of the four-side slide valve; p L Load pressure drop for the press; a. The p The sectional area of a hydraulic cylinder of the hydraulic system; x is the number of p The displacement of the piston in the hydraulic cylinder; c tp The leakage coefficient of the interior of the hydraulic cylinder is obtained; beta is a e Is the modulus of elasticity of the piston; v t Is the total compression volume of the hydraulic cylinder; m is t The piston of the hydraulic cylinder and the mass loaded on the piston; b p Is the viscosity coefficient; k is the hooke's coefficient of the load spring; f L An additional load force for external interference.
The intelligent fuzzy neural network algorithm comprises the following steps:
FIG. 2 shows a fourth-order neural network system built by combining the RBF neural network and the fuzzy system T-S model.
Each layer is an input layer, a first hidden layer, a second hidden layer and an output layer in sequence. For an electro-hydraulic proportional control system, the primary purpose is to output stable and accurate hydraulic oil pressure, and certainly obtain timely flow control parameters, and the control on the hydraulic pressure is mainly studied here, so that e is the control deviation ec of the pressure, and is the change rate of the pressure.
The pressure of the hydraulic system at time nT is e (n), and the control deviation, i.e., the output set value P REF The deviation from the feedback amount is ec (n).
Using the fuzzy system to optimize parameters, firstly, the input quantity is fuzzified, here, the quantization level of the input link is defined as 7, which is described as the fuzzy language:
domain of E: [ -1,1]
Discourse domain of EC: [ -1,1]
E fuzzy rule: { NB, NM, NS, ZO, PS, PM, PB }
Fuzzy rule of EC: { NB, NM, NS, ZO, PS, PM, PB }
The number of nodes of the corresponding third tier is N =7 × 7= 49.
For the fuzzification layer, a node corresponds to a fuzzy language, such as NM, NS, ZO, etc., as mentioned above. And (3) calculating each linguistic variable by using a Gaussian function as an evaluation standard to obtain the membership degree of each input variable, and then performing fuzzy reasoning:
wherein σ ij Representing the width of the i-th input variable subordinate to the j-th fuzzy aggregation function, c ij Indicating that i input variables are attached to the center position of the jth fuzzy aggregation function. i =1,2, \8230;, n; j =1,2, \8230;, L.
Next, in the fuzzy inference layer, a node corresponds to a fuzzy inference rule, and in order to calculate the fitness of each rule to its respective node, each fuzzy rule is paired, that is, each fuzzy rule is paired
In the formula (I), the compound is shown in the specification,wherein N is i The number of fuzzy partitions is indicated as the number of fuzzy partitions of the input i.
f 4 Is k p 、k i 、k d The setting result of (a), namely the output layer, is:
in the formula, W is a connection weight matrix, and i =1,2,3 is between the fourth layer output layer and the third layer fuzzy inference layer. The controller is as follows:
Δu(k)＝f 4 ·xc＝k p xc(1)+k i xc(2)+k d xc(3) (8)
wherein: k is a radical of p ＝f 4 (1)，k i ＝f 4 (2)，k d ＝f 4 (3)，xc(1)＝e(k)，xc(2)＝e(k)-e(k-1)，xc(3)＝e(k)-2e(k-1)+e(k-2)。
An incremental PID algorithm which is easy to list expressions is selected as the algorithm of the hydraulic control system of the invention:
u(k)＝u(k-1)-Δu(k) (9)
an evaluation function is selected to score the control effect, and the accumulation of the deviation of the control quantity can be generally selected as an evaluation standard:
ec (t) is the hydraulic system pressure change rate; t is an integral time variable; t is t c The time at the end of the algorithm, i.e. the integration ceiling time.
Get the connection right W ij Center function c ij And width function sigma ij Comprises the following steps:
wherein eta is the learning rate of the fuzzy neural network algorithm; α is the coefficient of inertia, 0< α <1.
As the nonlinearity and the coupling ratio of the electro-hydraulic proportional control system of the press are stronger, and the simple optimization of PID parameters by the fuzzy neural network cannot achieve good effect, the invention introduces the particle swarm algorithm to c of the fuzzy neural network ij 、σ ij And the parameters of eta and alpha are optimized again. Different from the traditional particle swarm velocity and displacement formula, the velocity and displacement formula of the invention is as follows:
v i,j (t+1)＝wv i,j (t)+c 1 r 1 [p i,j -x i,j (t)]+c 2 r 2 [p g,j -x i,j (t)] (14)
x i,j (t+1)＝x i,j (t)+v i,j (t+1),j＝1,...,d (15)
r 1 、r 2 is a random number and takes values between (0, 1); i represents a particle numbered i; t is the number of iterations of the population; d is the parameter to be optimized, and the dimension of the solution space in the algorithm; c. C 1 、c 2 Acceleration of a single particle and the whole population respectively; for w is the improved decreasing power exponent inertial weight:
w＝w min +(w max -w min )×e -4Δ (16)
wherein, the first and the second end of the pipe are connected with each other,f is an objective function value obtained after the algorithm is iterated; f. of avg Is the average of all particle objective functions of this iteration, f min Of which is the minimum value.
Here, the selection of the fitness objective function obviously has a crucial influence on the inertial weight of the particle swarm algorithm, the input of the system has pressure change and pressure change rate, the corresponding fuzzy rule has 7, so that the spatial dimension of the population is 14, each one-dimensional solution has respective membership central value and width, so f 3 Should be the center function c ij And width function σ ij The fitness evaluation function of (1) is obtained by optimizing the center function and the width functionThe fuzzy parameter is only the input quantity fuzzification process from the input layer to the fuzzification layer, so that the overfitting phenomenon cannot be generated.
As for the learning rate eta and the inertia coefficient alpha of the output layer, the general algorithm is random number, the parameter is optimized by utilizing a particle swarm algorithm and is related to the actual output of the system, and E is the fitness evaluation function of the parameter.
In order to verify the convergence speed of the fast convergence particle swarm algorithm, a generalized evaluation function Griewank function is introduced to perform convergence verification on the algorithm for improving the inertial weight, the iteration is performed for 50 times, and a convergence effect graph is obtained as shown in the following figure, wherein a blue line is the convergence effect of a common particle swarm, and a black dotted line is the fast-falling inertial weight particle swarm algorithm. The convergence speed can be obviously seen, table 1 shows the convergence times of different evaluation functions converging to the lowest value, and it can be obviously seen through fig. 3 and table 1 that the algorithm convergence is very quick after the inertia weight is improved, which is beneficial to improving the real-time performance of the control of the press.
TABLE 1 number of Convergence iterations
| Optimization algorithm | Ackley | Sphere | Rastighin | Griewank |
| PSO | 432 | 1401 | 893 | 678 |
| IPSO | 252 | 608 | 393 | 199 |
3. Intelligent fuzzy neural network press PID controller:
step.1, initializing an algorithm, and assigning respective initial values to 14-dimensional particles, wherein the initial values are local extreme values, the algorithm does not start iteration, a global extreme value cannot be obtained, and a fourteen-dimensional space is divided into two e dimensions and two ec dimensions, wherein the e dimension and the ec dimension respectively occupy seven dimensions. Randomly initializing a Gaussian membership function, i.e. central function c 0 And width function sigma 0 Selecting [0,0.5 ]]A random number between is assigned to η and α.
Step.2, running an algorithm, and carrying out online optimization according to the iterative algorithms (5) - (9) and (11) - (13) to obtain an output PID optimization result u of the fuzzy neural network;
step.3, calculating a central function c corresponding to each one-dimensional particle 0 And width function sigma 0 Fitness evaluation function f 3 ；
Step.4, optimally selecting the global optimal solution and the individual optimal solution to obtain a particle with the minimum fitness evaluation function, and replacing the speed and the displacement in the original formula with the speed and the displacement of the particle;
step.5 real-time optimization of the central value c in the fuzzy neural network using equations (14) to (15) ij And width σ ij ；
And step 6, calculating an evaluation function E of the system, screening the minimum eta and alpha of the target function, returning to the step 2, and repeating the algorithm until the specified iteration times are reached.
4 simulation experiment and result analysis
Taking laplace transform of equations (1) - (3), and taking a die-free multipoint forming press YAM1 as an actual simulation model, the open-loop transfer function of the system can be obtained as follows:
and (4) simulating the formula on an MATLAB platform.
As shown in fig. 4, comparing the control effect of the press at the initial setting with the curve, it can be clearly found that the control effect of the system is much better than that of the self-correcting PID controller proposed in document [11], and the adjustment time and overshoot are significantly reduced.
Table 2 shows the control effect of the fixed blank holder force of the press, and it can also be found that the adjustment time and the control accuracy are significantly better than those of the conventional PID control algorithm, and along with the increase of the blank holder force, the adjustment time and the control accuracy of the system do not change greatly, and on the other hand, the stable output effect of the algorithm is also verified.
TABLE 2 comparison of control effects of actual systems
FIG. 5 shows that in the small disturbance state during the press blank holding operation, the feedback control is performed by the system, and the adjustment time and overshoot are still that the intelligent fuzzy neural network PID controller is obviously smaller than the self-correcting PID closed-loop control, which is enough to explain the effectiveness of the algorithm.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (1)
1. The utility model provides a press PID control system based on intelligence fuzzy neural network algorithm which characterized in that designs through following steps and forms:
1. designing a hydraulic control system model of the press:
q L ＝K q x V -K c p L (1)
wherein q is L The flow rate of the four-side slide valve; k is c Pressure coefficients for different flow rates; k is q The gain flow for the four-sided spool valve; p L Load pressure drop for the press; a. The p The sectional area of a hydraulic cylinder of the hydraulic system; x is the number of p The displacement of the piston in the hydraulic cylinder; c tp The leakage coefficient of the interior of the hydraulic cylinder; beta is a e Is the modulus of elasticity of the piston; v t Is the total compression volume of the hydraulic cylinder; m is t The piston of the hydraulic cylinder and the mass loaded on the piston; b p Is the viscosity coefficient; k is the hooke's coefficient of the load spring; f L Extra load force for external interference;
2. designing an intelligent fuzzy neural network algorithm
For an electro-hydraulic proportional control system, the primary purpose is to output stable and accurate hydraulic oil pressure, and timely flow control parameters are obtained, wherein the control on the hydraulic pressure is mainly researched, so that e is the control deviation ec of the pressure and is the change rate of the pressure;
the pressure of the hydraulic system at the moment nT is e (n), and the deviation is controlledI.e. output set amount P REF The deviation from the feedback amount is ec (n);
using the fuzzy system to optimize parameters, firstly, the input quantity is fuzzified, here, the quantization level of the input link is defined as 7, which is described as the fuzzy language:
domain of E: [ -1,1]
Discourse domain of EC: [ -1,1]
Fuzzy rule of E: { NB, NM, NS, ZO, PS, PM, PB }
Fuzzy rules of EC: { NB, NM, NS, ZO, PS, PM, PB }
The number of nodes of the corresponding third tier is N =7 × 7= 49;
for the fuzzification layer, a node corresponds to a fuzzy language, such as NM, NS, ZO, etc. mentioned above; and (3) calculating each linguistic variable by using a Gaussian function as an evaluation standard to obtain the membership degree of each input variable, and then performing fuzzy reasoning:
wherein σ ij Width representing the membership of the ith input variable to the jth fuzzy aggregation function, c ij Representing that i input variables belong to the center position of the jth fuzzy set function; i =1,2, \8230;, n; j =1,2, \8230, L;
next, in the fuzzy inference layer, a node corresponds to a fuzzy inference rule, and in order to calculate the fitness of each rule to its respective node, each fuzzy rule is paired, that is, each node is a node in the fuzzy inference layer
In the formula (I), the compound is shown in the specification,wherein N is i Representing the fuzzy division number which is the fuzzy division number of the input i;
f 4 is k p 、k i 、k d The setting result of (c), that is, the output layer, namely:
in the formula, W is a connection weight matrix, and i =1,2,3 is arranged between the fourth layer output layer and the third layer fuzzy inference layer; the controller is as follows:
△u(k)＝f 4 ·xc＝k p xc(1)+k i xc(2)
+k d xc(3) (8)
wherein: k is a radical of formula p ＝f 4 (1)，k i ＝f 4 (2)，k d ＝f 4 (3)，xc(1)＝e(k)，xc(2)＝e(k)-e(k-1)，xc(3)＝e(k)-2e(k-1)+e(k-2)；
An incremental PID algorithm which is easy to list expressions is selected as the algorithm of the hydraulic control system of the invention:
u(k)＝u(k-1)-△u(k) (9)
an evaluation function is selected to score the control effect, and the accumulation of the deviation of the control quantity can be generally selected as an evaluation standard:
ec (t) is the hydraulic system pressure change rate; t is an integration time variable; t is t c The time when the algorithm is finished, namely the integration upper limit time;
get the connection right W ij C center function ij And width function sigma ij Comprises the following steps:
wherein eta is the learning rate of the fuzzy neural network algorithm; α is an inertia coefficient, 0< α <1;
as the nonlinearity and the coupling of the electro-hydraulic proportional control system of the press are stronger, and the simple PID parameter optimization of the fuzzy neural network cannot obtain good effect, the invention introduces the particle swarm algorithm to the c of the fuzzy neural network ij 、σ ij And the parameters eta and alpha are optimized again; different from the traditional particle swarm velocity and displacement formula, the velocity and displacement formula of the invention is as follows:
v i,j (t+1)＝wv i,j (t)+c 1 r 1 [p i,j -x i,j (t)]
+c 2 r 2 [p g,j -x i,j (t)] (14)
x i,j (t+1)＝x i,j (t)+v i,j (t+1),j＝1,...,d (15)
r 1 、r 2 is a random number and takes values between (0, 1); i represents a particle numbered i; t is the iteration number of the population; d is the parameter to be optimized, and the dimension of the solution space in the algorithm; c. C 1 、c 2 Acceleration of a single particle and the entire population, respectively; for w is the improved decreasing power exponent inertial weight:
w＝w min +(w max -w min )×e -4△ (16)
wherein the content of the first and second substances,f is an objective function value obtained after the algorithm is iterated; f. of avg Is the average of all particle objective functions of this iteration, f min Is the minimum value thereof;
3. designing an intelligent fuzzy neural network press PID controller:
step.1 initialization algorithm, fourteenThe dimension particles are endowed with respective initial values, the initial values are local extreme values, the algorithm does not start iteration, a global extreme value cannot be obtained, and the fourteen-dimensional space is divided into two e dimensions and two ec dimensions which are seven dimensions respectively; randomly initializing a Gaussian membership function, i.e. a central function c 0 And width function sigma 0 Selecting [0,0.5 ]]Assigning a random number therebetween to η and α;
step.2, running an algorithm, and carrying out online optimization according to the iterative algorithms (5) - (9) and (11) - (13) to obtain an output PID optimization result u of the fuzzy neural network;
step.3, calculating a central function c corresponding to each one-dimensional particle 0 And width function sigma 0 Fitness evaluation function f 3 ；
Step.4, optimally selecting the global optimal solution and the individual optimal solution to obtain a particle with the minimum fitness evaluation function, and replacing the speed and the displacement in the original formula with the speed and the displacement of the particle;
step.5 real-time optimization of the central value c in the fuzzy neural network using equations (14) to (15) ij And width σ ij ；
And step 6, calculating an evaluation function E of the system, screening the minimum eta and alpha of the target function, returning to the step 2, and repeating the algorithm until the specified iteration times are reached.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711232473.4A CN107807530A (en) | 2017-11-30 | 2017-11-30 | A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711232473.4A CN107807530A (en) | 2017-11-30 | 2017-11-30 | A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107807530A true CN107807530A (en) | 2018-03-16 |
Family
ID=61582167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711232473.4A Pending CN107807530A (en) | 2017-11-30 | 2017-11-30 | A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107807530A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109272037A (en) * | 2018-09-17 | 2019-01-25 | 江南大学 | A kind of self-organizing TS pattern paste network modeling method applied to infra red flame identification |
| CN110617152A (en) * | 2019-10-18 | 2019-12-27 | 上海格陆博实业有限公司 | Throttle control system based on fuzzy PID control |
| CN113885609A (en) * | 2021-10-25 | 2022-01-04 | 四川虹美智能科技有限公司 | Box body temperature control method and device of vehicle-mounted refrigerator and vehicle-mounted refrigerator |
| CN114412883A (en) * | 2022-01-14 | 2022-04-29 | 西安建筑科技大学 | Hydraulic system control method, device and system and storage medium |
| CN114839860A (en) * | 2022-04-11 | 2022-08-02 | 哈尔滨工程大学 | Fuzzy PID fuel injection quantity closed-loop control method based on pressure change monitoring of high-pressure natural gas injector inlet |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104133372A (en) * | 2014-07-09 | 2014-11-05 | 河海大学常州校区 | Room temperature control algorithm based on fuzzy neural network |
| CN105136469A (en) * | 2015-07-23 | 2015-12-09 | 江苏大学 | Unmanned vehicle speed control method based on PSO and RBF neutral network |
| CN105751470A (en) * | 2016-03-23 | 2016-07-13 | 广西科技大学 | Real-time temperature control method for injection molding machine |
| CN106371321A (en) * | 2016-12-06 | 2017-02-01 | 杭州电子科技大学 | PID control method for fuzzy network optimization of coking-furnace hearth pressure system |
| CN106678546A (en) * | 2017-01-05 | 2017-05-17 | 中国石油大学(华东) | Method and system for controlling outlet pressure of centrifugal compressor of natural gas pipeline |
-
2017
- 2017-11-30 CN CN201711232473.4A patent/CN107807530A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104133372A (en) * | 2014-07-09 | 2014-11-05 | 河海大学常州校区 | Room temperature control algorithm based on fuzzy neural network |
| CN105136469A (en) * | 2015-07-23 | 2015-12-09 | 江苏大学 | Unmanned vehicle speed control method based on PSO and RBF neutral network |
| CN105751470A (en) * | 2016-03-23 | 2016-07-13 | 广西科技大学 | Real-time temperature control method for injection molding machine |
| CN106371321A (en) * | 2016-12-06 | 2017-02-01 | 杭州电子科技大学 | PID control method for fuzzy network optimization of coking-furnace hearth pressure system |
| CN106678546A (en) * | 2017-01-05 | 2017-05-17 | 中国石油大学(华东) | Method and system for controlling outlet pressure of centrifugal compressor of natural gas pipeline |
Non-Patent Citations (5)
| Title |
|---|
| 宫向东: "压力机电液比例位置控制系统", 《万方》 * |
| 杨达飞 等: "混沌模糊神经网络算法在注塑机温度实时控制中的应用", 《塑料工业》 * |
| 梁德坚 等: "改进粒子群模糊控制算法在注塑机电液比例控制系统的应用", 《塑料》 * |
| 王彦 等: "基于改进粒子群算法的模糊神经网络PID控制器设计", 《控制工程》 * |
| 赵俊 等: "混沌粒子群优化的模糊神经PID控制器设计", 《西安电子科技大学学报(自然科学版)》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109272037A (en) * | 2018-09-17 | 2019-01-25 | 江南大学 | A kind of self-organizing TS pattern paste network modeling method applied to infra red flame identification |
| CN109272037B (en) * | 2018-09-17 | 2020-10-09 | 江南大学 | Self-organizing TS type fuzzy network modeling method applied to infrared flame identification |
| CN110617152A (en) * | 2019-10-18 | 2019-12-27 | 上海格陆博实业有限公司 | Throttle control system based on fuzzy PID control |
| CN113885609A (en) * | 2021-10-25 | 2022-01-04 | 四川虹美智能科技有限公司 | Box body temperature control method and device of vehicle-mounted refrigerator and vehicle-mounted refrigerator |
| CN114412883A (en) * | 2022-01-14 | 2022-04-29 | 西安建筑科技大学 | Hydraulic system control method, device and system and storage medium |
| CN114839860A (en) * | 2022-04-11 | 2022-08-02 | 哈尔滨工程大学 | Fuzzy PID fuel injection quantity closed-loop control method based on pressure change monitoring of high-pressure natural gas injector inlet |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107807530A (en) | A kind of forcing press PID control system based on intelligent fuzzy neural network algorithm | |
| US6701236B2 (en) | Intelligent mechatronic control suspension system based on soft computing | |
| Papageorgiou et al. | Fuzzy cognitive map learning based on nonlinear Hebbian rule | |
| CN101441441B (en) | Design method of intelligent swing-proof control system of crane | |
| US7219087B2 (en) | Soft computing optimizer of intelligent control system structures | |
| Detiček et al. | An Intelligent Electro-Hydraulic Servo Drive Positioning. | |
| Wani et al. | A critical review on control strategies for structural vibration control | |
| CN112904718B (en) | Magneto-rheological damper control system and method based on Hammerstein model | |
| CN112929205A (en) | Swarm unmanned aerial vehicle fault propagation method based on cellular automaton | |
| CN112462611A (en) | Sliding friction modeling method for precise electromechanical system | |
| CN105911865A (en) | Control method of PID controller | |
| Ho et al. | Design of an adaptive fuzzy observer-based fault tolerant controller for pneumatic active suspension with displacement constraint | |
| Zhao et al. | H∞ tracking control for nonlinear multivariable systems using wavelet-type TSK fuzzy brain emotional learning with particle swarm optimization | |
| CN106371321A (en) | PID control method for fuzzy network optimization of coking-furnace hearth pressure system | |
| Patel et al. | Servo actuating system control using optimal fuzzy approach based on particle swarm optimization | |
| Shen et al. | Vibration control of flexible structures using fuzzy logic and genetic algorithms | |
| CN110806693B (en) | Gray wolf prediction control method for time lag of plate heat exchanger | |
| Kharroubi et al. | Soft computing based control approach applied to an under actuated system | |
| Bartyś | Dynamics of Single-Action Pneumatic Actuators | |
| JP7329845B2 (en) | Control system design method | |
| CN115167150B (en) | Batch process two-dimensional off-orbit strategy staggered Q learning optimal tracking control method with unknown system dynamics | |
| Yousefi et al. | Adaptive neural network in compensation the dynamics and position control of a servo-hydraulic system with a flexible load | |
| CN112836818B (en) | Dynamic response artificial neural network suitable for process control modeling | |
| Zha | Soft computing in engineering design: A fuzzy neural network for virtual product design | |
| Ustinov et al. | A Hybrid Model for Fast and Efficient Simulation of Fluid Power Circuits With Small Volumes Utilizing a Recurrent Neural Network |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180316 |