WO2011120200A1

WO2011120200A1 - Genetic optimization control technology for stacking machines

Info

Publication number: WO2011120200A1
Application number: PCT/CN2010/000424
Authority: WO
Inventors: 陆金桂; 韩绍军; 徐正林
Original assignee: 江苏六维物流设备实业有限公司
Priority date: 2010-04-01
Filing date: 2010-04-01
Publication date: 2011-10-06

Abstract

A genetic optimization control technology for stacking machines is provided. The technology enables speed optimization control during stacking machine operation by means of neural networks and genetic algorithms, so as to reduce the magnitude of vibrations produced by the stacking machine when operating at high speeds. On the basis of the main operating parameters of, and vibration data produced during operation of, a stacking machine, a neural network mapping model between stacking machine vibration states and said main operating parameters is established. On the basis of said established neural network model, optimization of operating speed with minimum vibratory magnitude during said machine operation is achieved by use of a genetic algorithm. Based on the stacking machine operating speed obtained through optimization, a speed control curve for machine operation is obtained. Speed control of the machine is achieved through use of a programmable logic controller and numerical conversion by a frequency-conversion controller.

Description

Genetic optimization control technology for stacker

The invention belongs to the field of stacker control, and particularly relates to a genetic optimization control technology for the operation process of a stacker. technical background:

Due to the advantages of high space utilization and high degree of automation, the automated three-dimensional warehouse has a wider application range. As a stacker for key equipment in a three-dimensional warehouse, its control technology is critical to the performance of the downtime. With the wide application of programmable logic controller (PLC), variable frequency speed regulation, laser ranging and other technologies in the field of stacker control, the operation speed of the stacker has been significantly improved. However, the increase in the running speed of the stacker brings about problems such as vibration, and therefore the speed of the stacker running process needs to be optimized to reduce the vibration of the stacker.

At present, some research work has been carried out on stacker control technology. For example, Chen Juan and Zhong Yongyan and others have discussed the hardware and software implementation of the stacker control system for the characteristics of stacker control in automated warehouses. Wang Yongjun and Zhou Qicai et al. used laser ranging technology as the means of accessing the roadway stacker, and applied the inverter stepless speed regulation closed-loop control method to study the control method of a new type of roadway stacker high speed operation. Lu Quanhai and Shen Minde et al. combined the laser ranging technology to explore the hardware and software implementation technology of the closed loop speed control of the stacker. Bi Wenqun and Yan Xingfeng et al. designed a PLC control system suitable for use in stacker walking, lifting and telescopic fork mechanisms. Ge Gaofeng analyzed the composition, structure and characteristics of the stacker control system, and discussed the stacker control system scheme using S7-300 PLC. Luo Zhiqing and Yu Shutian applied fuzzy control and predictive control to the stacker of the parcel stereo warehouse. The intelligent predictive fuzzy control method was studied and simulated according to the actual working conditions of the stacker movement. At present, the neural network and genetic optimization algorithm are used to carry out the stacking machine running speed g system to reduce the vibration of the stacker. The research work in this area has not been reported in the open literature.

The invention relates to a genetic optimization control technique for a stacker. The new control technology of the stacker proposed by the invention aims at reducing the amount of vibration generated during the high-speed operation of the stacker by means of genetic algorithm and neural network. The neural network is used to establish the mapping model between the vibration state of the stacker and the operating parameters of the stacker, and the genetic algorithm is used to optimize the running speed of the minimum vibration state of the stacker. On this basis, the control of the stacker running process is implemented to achieve the reduction. The target of the vibration amplitude of the small stacker operating process. The new technology of the genetic optimization optimization control of the stacker proposed by the invention is realized by using a neural network and a genetic algorithm. Therefore, the new control technology proposed by the invention belongs to the intelligent control method of the stacker. The genetic optimization control technology of the stacker proposed by the invention can reduce the vibration amount generated during the high-speed operation of the stacker, and provides an effective control means for ensuring efficient and safe operation of the stacker. Summary of the invention:

The object of the present invention is to optimize the speed of the stacker operation process by using a neural network and a genetic algorithm as a means to reduce the amount of vibration generated during the high-speed operation of the stacker. In order to achieve the above objectives, the technical solution adopted by the present invention is: based on the main operating parameters during the operation of the stacker and the vibration data generated during the operation of the stacker, establishing a vibration state between the stacker and the main operating parameters. Neural network mapping model; based on the established neural network model, using genetic algorithm to optimize the running speed of the vibration during the stacking machine operation; obtaining the speed of the stacking machine based on the optimized running speed of the stacker The control curve uses the numerical conversion of the programmable logic controller and the variable frequency controller to realize the speed control of the stacker.

The invention includes the establishment of sample data in the operation process of the stacker, the construction of a neural network model, the optimization of the running speed of the stacker, and the speed control of the stacker. The specific steps included in the present invention are as follows:

1) Establish sample data of stacker operation process

Firstly, the direct cause of vibration during the operation of the stacker is analyzed. On this basis, the main influencing factors related to the vibration of the stacker, the type of stacker, the height of the console, the load and the running speed are determined. According to the orthogonal test method, the experiment of the influence relationship between the type of stacker, the height of the console, the load, the running speed and the vibration of the stacker operation process is obtained, and the parameters such as the type of stacker and the vibration of the stacker are reflected. A large amount of data that affects relationships. Based on the obtained stacker operation data, the sample data of the stacker running process is constructed. Ben In the invention, the vibration amount of the stacker is used as the output of the sample data, and the parameters of the stacker type, the table height, the load, the running speed, and the like are input as sample data to form sample data. The sample data reflects the relationship between the vibration state of the stacker and the speed of the stacker, the type of stacker, the height of the console, and the load.

2) Building a neural network model

Based on the established sample data of the stacker running process, the neural network model is constructed. The sample data of the stacker operation process reflects the relationship between the vibration state of the stacker and the running speed of the stacker and the type of the stacker. Therefore, the sample learning is performed by using a multi-layer neural network or a radial basis function neural network. The relationship between the vibration of the stacker implicated in the sample data and the parameters of the stacker can be described by a neural network model.

Building a neural network model requires first determining the structure and properties of the neural network, including the number of layers of the neural network, the number of hidden layers, the number of elements per layer, the activation function settings for each layer, and the input and output parameters of the neural network model. . The number of neurons in the input layer of the neural network corresponds to the number of parameters in the input part of the sample, including the type of stacker, the height of the console, the weight of the load, and the running speed. The number of neuron output layer neurons corresponds to the number of parameters of the sample output portion, including the amount of vibration of the stacker. After determining the structure of the neural network, the sample data is processed to meet the requirements of the neural network learning needs, and a suitable learning algorithm is selected for the neural network learning. After completing the sample learning process of the neural network, a neural network model reflecting the relationship between the vibration of the stacker and the running speed of the stacker can be established.

3) Stacker running speed optimization

Based on the neural network model between the vibration amount of the stacker and the running speed of the stacker and the type of stacker, the genetic algorithm is used to optimize the running speed of the stacker with the minimum vibration amount as the optimization target. In the genetic optimization process, the amount of vibration involved in the optimization target will be calculated using the constructed neural network; the type of stacker, the height of the console, the weight of the load, and the speed value of the current design variable will be used as the neural network. The input parameters, the data processed to meet the requirements of the neural network prediction needs, the output of the neural network can be obtained; the output data of the neural network can be processed to calculate the vibration of the stacker. The optimal operation speed of the stacker for optimizing the design variables is optimized by genetic optimization, and the running speed corresponding to the minimum vibration amount of the stacker is obtained.

4) Stacker speed control

Based on the operating speed of the minimum vibration of the stacker obtained by the neural network prediction, the discrete velocity value curve is fitted, and the fitted velocity curve is stored in the data block of the programmable logic controller of the stacker as a heap. Prepare for the speed control of the downtime. In the actual speed control of the stacker, the programmable logic controller finds the corresponding speed value on the fitted speed curve according to the current position of the stacker, and drives the current vector through the analog quantity or the field bus mode, and Drive the stacker to run the AC asynchronous motor to complete the speed control of the stacker.

The advantages of the invention: the genetic optimization control technology of the stacker can optimize and control the running speed under the minimum vibration state during the stacking machine operation, and reduce the vibration amount during the running process of the stacker. BRIEF DESCRIPTION OF THE DRAWINGS:

Figure 1 is a flow chart of the genetic optimization control of the stacker;

2 is a schematic diagram of a genetic optimization model of a stacker;

Figure 3 is a schematic diagram of the genetic optimization control principle of the stacker.

The genetic optimization control technology for the automated stereoscopic warehouse stacker proposed by the present invention will be described below with reference to the accompanying drawings. The specific implementation method is as follows - FIG. 1 is a flow chart of the evolution optimization control of the stacker, and the stacker control proposed by the present invention is new. The technology uses neural networks and genetic algorithms as the means to reduce the amount of vibration generated during the high-speed operation of the stacker, and optimizes the speed of the stacker operation process. In order to achieve the above objectives, the technical solution adopted by the present invention is: based on the main operating parameters during the operation of the stacker and the vibration data generated during the operation of the stacker, establishing a vibration state between the stacker and the main operating parameters. Neural network mapping model; Based on the network model, the genetic algorithm is used to optimize the running speed of the stacker during the operation of the stacker. The speed control curve of the stacker operation is obtained based on the optimized stacking machine running speed, and the programmable logic is used to control. The numerical conversion of the inverter and the variable frequency controller realizes the speed control of the stacker.

Fig. 2 is a schematic diagram of the evolution optimization model of the stacker. The genetic optimization model of the stacker in the present invention comprises two parts: a neural network prediction technique and a genetic algorithm optimization technique. In the invention, based on the established neural network model, the genetic algorithm is used to optimize the running speed of the vibration during the stacking operation.

1. The invention utilizes neural network technology to predict the running speed of the stacker. The constructed neural network is a BP neural network model with a three-layer structure. The detailed description is as follows:

1) Determine input and output variables

In the three-dimensional automated warehouse, the goods are usually placed on the operation platform of the stacker and transported by the movement of the stacker. During the running of the goods, due to the start, acceleration, hook speed and deceleration of the stacker until The stopped composite motion and the like easily cause harmful vibrations of the stacker body, which affects the safe and effective operation of the stacker. These factors mainly include the type of stacker, the height of the stacker shelf, the current load of the stacker, and the running speed of each section of the stacker on the running line. Therefore, several types of stacker type, shelf height, load capacity, and various speed values can be used as input variables of the stacker neural network intelligent control model, and the stacker is operated on each segment of the running line. Speed is used as the output variable of the intelligent control model of the stacker neural network.

2) Network structure description

In the stacker neural network control technology, we set the operation route of the stacker into N parts according to the needs, then the input layer of the neural network model in the invention contains N + 3, corresponding to the type of stacker, the height of the console, the current load, and the speed of each segment of the stacker on the run line ( _P V ₂ V _N) The number of output nodes of the neural network model is N Corresponding to the velocity value of each segment of the stacking machine (V ₂ V _N ); the middle layer is the hidden layer of the neural network model, and the number of neurons it contains is V2N + 3 +t Where t represents any integer between 0 and 10. In the neural network model of the present invention, the input layer, the intermediate layer (hidden layer) and the output layer, and the neurons between the upper and lower layers are connected, the same layer There is no connection between them, and the activation function of the hidden layer uses the Sigmoid function (S-type function). The function is described as follows:

/(v) =

1 + exp(-av)

Where a is the slope parameter of the Sigmoid function. By changing the parameter a, Sigmoid functions with different slopes can be obtained.

3) Network training

The training samples of the network are randomly collected by the stacker. The neural network uses a 3-layer neural network model. The structure uses N input neurons, > 2N + 3 + t hidden layer nodes (t Represents any integer between 0 and 10) and N output nodes. The process of neural network training is also a learning process. First, the input information is processed layer by layer through the input layer and the actual output value of each unit is calculated. Then, according to the difference between the expected output and the actual output, the output is output. Start adjusting the weights layer by layer.

2. The invention utilizes a genetic algorithm to optimize the running speed of the stacker with the minimum vibration amount as the optimization target:

1) Design optimization variables

During the operation of the stacker, the speed of the stacker is mainly related to the type T of the stacker. During the operation, the height of the rack is H. The current load weight is G, so the optimization variable can be recorded as:

In the above formula, ^ denotes the running speed of the stacker in the i-th sub-segment, indicating the shelf height in the i-th sub-segment, indicating the load weight of the stacker at the time of the i-th sub-segment. i = \,2— N , W is the number of sub-segments divided by the stacker in one operation.

2) Design optimization objective function For the operation process and working principle of the stacker in the automated warehouse, the type of stacker is Γ, the height of the rack during the operation is H, the current load weight is G, and the stacker is in each sub-work. The vibration quantities of the road sections are respectively, , ... A _n (n=l, 2... ... ) Combined with various factors affecting the vibration of the stacker during the operation of the stacker, the stacker operation is designed. The design variables of speed optimization mainly include: type of stacker, height of cargo rack G, and comprehensive vibration of stacker in one operation. The comprehensive vibration amount of stacker in one operation is calculated as follows:

4 = (ί3⁄4 * 4 +. 2 * 2 ten...+

Among them, " ₂ ...^^ is the weighting coefficient of the vibration amount of each sub-segment in one operation of the stacker, N=l, 2, ..., which is the number of sub-segments divided according to needs.

3) Design optimization constraint function

During the operation of the stacker, the speed of the stacker should not exceed the specified maximum safe speed. At the same time, the vibration of the stacker should be minimized during the operation, and the constraint function of the intelligent optimization of the stacker can be determined accordingly.

r mm ^r i, *" max rmn A _i = g(V _i ,T, H _i ,G _i )

4) Optimization calculation

The invention adopts the genetic algorithm to carry out the intelligent optimization calculation of the stacker. In the optimization process of the stacker, the optimization operation of the stacker is carried out by the operation of the genetic algorithm, and finally the accurate optimization result is obtained.

Fig. 3 is a schematic diagram of the control principle of the neural network control of the stacker. In the present invention, the neural network control algorithm of the stacker is mainly completed by the programmable controller PLC. Through the training of the neural network, the running speed value of each section of the stacker on a single working path is predicted. The function block outputs the speed control values of each period. These speed values are some discrete points, how to directly use the control, due to Each point is discontinuous, which has a certain impact on the stacking opportunity. In order to make the stacker run smoothly, here we fit these discrete points into the operation control curve of the stacker to take the points on the curve as the speed setting. The fixed value is sent to the inverter controller.

The present invention uses closed-loop control of the speed, and the speed of the stacker can be calculated by the rate of change of the horizontal position of the stacker detected by the laser range finder. The speed control amount after the calculation is converted into a current signal by the D/A module of the PLC for controlling the frequency converter. In the present invention, the walking, lifting and fork motor are respectively controlled by two frequency converters to realize the stepless speed regulation, and the frequency converter Then control the speed of the motor by changing the power frequency.

Claims

Claim

1. The genetic optimization control technology of stacker is characterized by genetic algorithm and neural network as the means to reduce the vibration generated during the high-speed operation of the stacker, and the speed control of the stacker running process, including the following Steps:

Establish sample data of the stacker running process;

Genetic optimization of the speed of the stacker operating process;

Construct a neural network model;

Stacker running speed prediction;

Stacker speed control.

2. The genetic optimization control technology of a stacker according to claim 1, wherein: the genetic optimization of the speed value of the stacker running process is realized by optimizing the sample data by using a genetic algorithm.

3. The genetic optimization control technology of the stacker according to claim 1, wherein: the prediction of the running speed value of the stacker is based on the genetic optimization of the speed value of the stacking machine operation process, and the neural network is established. The network model is implemented by training the network.