CN102682139A - Method for forming shell plate curve of ship body - Google Patents

Abstract

The invention relates to the technical field of ship engineering, in particular to a method for forming a shell plate curve of a ship body. The method includes that experience knowledge of skilled operation staff is recorded through a large amount of actual operation of a line heating numerical control device, a large amount of machining process parameters are stored in a data base, and a knowledge base of an expert system is formed; when experiment data is accumulated to a certain degree, a mechanism model of line heating is built through a neural network, and an inference engine of the expert system is formed; by means of the inference engine of the expert system, the line heating numerical control device automatically gives machining tracks, and total automation and digitalization of the line heating are achieved. By means of the method, labor intensity of shipbuilding enterprise workers can be greatly reduced, working conditions of the workers are improved, and production efficiency and yield rate of products are improved.

Description

A kind of method of ship hull plate curve generating

Technical field

The present invention relates to the Marine engineering technical field, be specifically related to a kind of method of ship hull plate curve generating.

Background technology

Hull is to be made up of the space curved surface that complicacy can not be opened up, and ship plate is processed into the curve form of ship hull plate, and the method for shipyard, countries in the world employing at present mostly is a wire extreme misery deck of boat curve generating processing technology.The principle of this technology is to utilize the steel plate part to receive moulding change of thermoelastic that produces after the high temperature cooling and the flexural deformation that reaches steel plate integral body.Its job operation is to adopt oxy-acetylene flame or other thermals source that surface of steel plate is heated, and with cold water it is cooled off fast then, makes steel plate produce bigger thermal stress and diastrophic process.The shipyard all is to rely on experienced master worker's manual operations to realize this technology both at home and abroad for a long time.Along with the development of Modern Shipbuilding technology and the transformation of Shipbuilding Mode; Manual empirical technology pattern is on speed and all satisfied not the needs that Modern Shipbuilding is produced qualitatively far away, and this has become " bottleneck " problem of restriction ship building period and quality.

Summary of the invention

The method that the purpose of this invention is to provide a kind of ship hull plate curve generating can alleviate shipbuilding enterprise working strength of workers significantly through this method, improves workman's condition of work, the yield rate with product of enhancing productivity.

Method of the present invention is a large amount of practical operations through the flame forming plate numerical control device, and record actuator's experimental knowledge is stored into database with a large amount of working process parameters, forms the knowledge base of expert system; After empirical data runs up to a certain degree, set up the mechanism model of flame forming plate through neural network, form the inference machine of expert system; Through the inference machine of expert system, the flame forming plate numerical control device provides machining locus automatically, realizes the whole robotizations and the digitizing of flame forming plate.

In the real process of the knowledge base that forms expert system, its technological process comprises:

A, the new deck of boat Tribon data of importing, man-machine interface shows the ship hull plate three-dimensional data;

B, measurement deck of boat drift angle data;

Machined parameters is confirmed in c, operative employee's line;

D, numerical control device are processed along track;

E, recording track;

F, laser measurement deck of boat curved surface data;

If the g curved surface data meets accuracy requirement, then get into next step, if curved surface data does not meet accuracy requirement, then turn back to flow process c, line, confirm machined parameters;

H, record machining locus and working process parameter.

In the inference machine process that forms expert system; Adopting one type of neural network is SVMs (Support Vector Machine; Be called for short SVM) modeling; The SVM input comprises environmental factor, the deck of boat self parameter and working process parameter, is output as machining locus, and wherein: environmental factor comprises the temperature and humidity that adds man-hour; The deck of boat self parameter comprises deck of boat physical dimension, thickness of slab and material; Working process parameter comprises flue width, flame power, heating-up temperature, extreme misery torch and discharge, and neural network is set up the mechanism model of ship hull plate curve generating through the self study of mass data.

Description of drawings

Fig. 1 is a practical operation process flow diagram of the present invention;

Fig. 2 is a neural network structure block diagram of the present invention.

Embodiment

As shown in Figure 1; Enforcement of the present invention need be through a large amount of practical operations; Store a large amount of machining locus and working process parameter, i.e. the knowledge base of expert system, the concrete technological process of practical operation each time is: 1. derive ship hull plate data to be processed from hull design Tribon software; 2. carry out deck of boat 3-D display; 3. measure deck of boat drift angle data; 4. the operative employee relies on working experience for many years on ship hull plate, to mark the heating reference line, confirms machined parameters; 5. the numerical control device record heats reference line and carries out extreme misery processing; 6. write down machining locus; 7. adopt laser measurement deck of boat curved surface data, the data of measurement data and Tribon software compare, as have certain deviation, and then the operator provides the heating reference line once more, and this numerical control device repeats the aforesaid operations step, satisfies accuracy requirement until deviation at last.

As shown in Figure 2; Set up the mechanism model of flame forming plate, i.e. the inference machine of expert system adopts in the neural network one type to be SVMs (Support Vector Machine; SVM is made in letter) modeling; The SVM input comprises environmental factor, the deck of boat self parameter and working process parameter, is output as machining locus, and wherein: environmental factor comprises the temperature and humidity that adds man-hour; The deck of boat self parameter comprises deck of boat physical dimension, thickness of slab and material; Working process parameter comprises flue width, flame power, heating-up temperature, extreme misery torch and discharge.The mechanism model step of specifically setting up flame forming plate is following:

Adopt the very fast and generalization ability of speed of convergence preferably the method for v-SVM algorithm and mixed nucleus function set up the ship hull plate mechanism model.

Given sample set is { (x _i, y _i), i=1,2 ... L), x wherein _i∈ R ^NBe input value, y _i∈ R is corresponding desired value, and l is a sample number.The fitting function form is:

f(x)＝w·φ(x)+b w，φ(x)∈R ^N，b∈R (1)

Wherein: w is the Argument List vector, and φ () is a function row vector, and it is mapped to feature space to the input sample from the input space, and b often is worth deviation.

Representative overall kernel function (Polynomial kernel function) and local kernel function (RBF kernel function) are constituted a kind of mixed nucleus function S VM; Set up the ship hull plate mechanism model in conjunction with v-SVM; This method has good model fitting precision; And can effectively suppress the caused prediction output pulsation of karyomerite function, solved the problem of system modelling.That is: Polynomial kernel function: K (x, x _i)=[(xx _i)+1] ^q

The RBF kernel function: $K(x,x_i)=\mathrm{Exp}[-\frac{{|x-x_i|}^2}{{2\mathrm{\σ}}^2}];$

Mixed function: K _Mix=ρ K _Poly+ (1-ρ) K _RBF(2)

Wherein: K _PolyAnd K _RBFBe respectively Polynomial kernel function and RBF kernel function, coefficient ρ (0≤ρ≤1) is for regulating the constant of two kinds of kernel function effect sizes.

Can get SVM output fitting function by (1) formula and (2) formula:

$f\left(x\right)=\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({{\mathrm{\α}}_i}^{*}-{\mathrm{\α}}_i)K(x_i,x)+b---\left(3\right)$

Wherein: α, α ^*Be the Lagrange factor.

Neural network is set up the mechanism model of ship hull plate curve generating through the self study of mass data, i.e. the inference machine of expert system provides the processing reference line of any ship hull plate automatically, realizes the robotization and the digitizing of flame forming plate.

Claims (3)

1. the method for a ship hull plate curve generating, this method is a large amount of practical operations through the flame forming plate numerical control device, record actuator's experimental knowledge is stored into database with a large amount of working process parameters, forms the knowledge base of expert system; After empirical data runs up to a certain degree, set up the mechanism model of flame forming plate through neural network, form the inference machine of expert system; Through the inference machine of expert system, the flame forming plate numerical control device provides machining locus automatically, realizes the whole robotizations and the digitizing of flame forming plate.

2. the method for a kind of ship hull plate curve generating according to claim 1 is characterized in that: in the real process of the knowledge base that forms expert system, its technological process comprises:

A, the new deck of boat Tribon data of importing, display shows the ship hull plate three-dimensional data;

B, measurement deck of boat drift angle data;

Machined parameters is confirmed in c, operative employee's line;

D, numerical control device are processed along track;

E, recording track;

F, laser measurement deck of boat curved surface data;

If the g curved surface data meets accuracy requirement, then get into next step, if curved surface data does not meet accuracy requirement, then turn back to flow process c, line, confirm machined parameters;

H, record machining locus and working process parameter.

3. the method for a kind of ship hull plate curve generating according to claim 1; It is characterized in that: in the inference machine process that forms expert system; Adopting one type of neural network is SVMs (Support Vector Machine; Be called for short SVM) modeling, the SVM input comprises environmental factor, the deck of boat self parameter and working process parameter, is output as machining locus; Wherein: environmental factor comprises that the temperature and humidity, the deck of boat self parameter that add man-hour comprise that deck of boat physical dimension, thickness of slab and material, working process parameter comprise flue width, flame power, heating-up temperature, extreme misery torch and discharge; Neural network is set up the mechanism model of ship hull plate curve generating through the self study of mass data.

Images