CN108857600A - One kind being based on light-catalysed cobalt-base alloys processing method and processing platform - Google Patents
One kind being based on light-catalysed cobalt-base alloys processing method and processing platform Download PDFInfo
- Publication number
- CN108857600A CN108857600A CN201810828984.0A CN201810828984A CN108857600A CN 108857600 A CN108857600 A CN 108857600A CN 201810828984 A CN201810828984 A CN 201810828984A CN 108857600 A CN108857600 A CN 108857600A
- Authority
- CN
- China
- Prior art keywords
- polishing
- value
- processing
- formula
- optimized
- 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.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B1/00—Processes of grinding or polishing; Use of auxiliary equipment in connection with such processes
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B41/00—Component parts such as frames, beds, carriages, headstocks
- B24B41/02—Frames; Beds; Carriages
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B41/00—Component parts such as frames, beds, carriages, headstocks
- B24B41/06—Work supports, e.g. adjustable steadies
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B57/00—Devices for feeding, applying, grading or recovering grinding, polishing or lapping agents
- B24B57/02—Devices for feeding, applying, grading or recovering grinding, polishing or lapping agents for feeding of fluid, sprayed, pulverised, or liquefied grinding, polishing or lapping agents
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Finish Polishing, Edge Sharpening, And Grinding By Specific Grinding Devices (AREA)
- Grinding And Polishing Of Tertiary Curved Surfaces And Surfaces With Complex Shapes (AREA)
Abstract
One kind being based on light-catalysed cobalt-base alloys processing method and processing platform, the processing method includes that chemical reaction process is optimized and revised and optimized and revised with mechanical arm polishing path, and chemical reaction process, which is optimized and revised, seeks the conditions such as optimal intensity of illumination and chemical addition agent proportion, temperature using establishing linear regression model (LRM) for different processing objects based on linear regression algorithm;Mechanical arm polishing path, which is optimized and revised, is adjusted polishing locus using tactful network;The processing platform includes:Robot module, polishing station, polishing fluid adding set, industrial personal computer and memory.The beneficial effects of the present invention are:The amount bad fixed problem of the ultraviolet ray intensity and chemical addition agent of chemical reaction process, can effectively promote quality of finish in effective solution photocatalysis process;Mechanical arm polishing path is trained for different processing objects, the flexible bring trajectory error of polishing disk has been can effectively solve, has promoted the quality and efficiency of polishing.
Description
Technical field
The present invention relates to one kind to be based on light-catalysed cobalt-base alloys processing method and processing platform.
Background technique
Cobalt-base alloys can be used for improving metal material surface characteristics as self-melting alloy material, form high rigidity, resistance to
High temperature and anti abrasive coating.This method is widely used in the industries such as mold manufacture, automobile industry, turbine making
In.Mold cobalt-base alloys coating, which not only has the complexity of free form surface, while also encountering high rigidity and high-wearing feature etc., newly asks
Topic so that Precision Machining is more difficult, standard machinery processing method is difficult to quickly remove, if can by specific chemical method,
So that such difficult-to-machine material is become easy processing material, then can reach the purpose for promoting processing efficiency.For cobalt member in cobalt-base alloys
The material property of element, configures effective reaction dissolvent, so that surface layer cobalt atom is generated reaction, forms product easy to remove, and combine
Ultraviolet catalytic effect, establishes the relationship of photon and quantum yield, adjusts the fracture of cobalt element atom key and energy release process, real
Now reaction is controllable.Since cobalt-base alloys is multielement alloy type material, needs conjunction with mechanical to act on, realize the material of insoluble product
Removal.
The processing platform material removal process is the collaborative processes an of chemical reaction, optics control and mechanical removal.
Therefore influence of the processing technology for processing result is bigger, in order to enable processing technology is optimal, is existed using artificial intelligence technology
Ability in continuity decision and searching optimum point, artificial intelligence technology and photocatalysis processing of robots platform are combined, mentioned
Rise the processing efficiency and processing quality of the polishing of photocatalysis robot.
Summary of the invention
The purpose of the present invention is to provide a kind of suitable photocatalysis processing platform based on artificial intelligence technology, the processing is flat
Platform can select optimal chemical reaction condition, and solution when polishing different processing objects by soft for different processing objects
Property polishing disk bring polishing path error problem.The processing platform can be directed to different processing pair based on artificial intelligence technology
As adjustment processing technology includes:Tactful network model pair is optimized and revised and is based on to chemical reaction process based on linear regression model (LRM)
Mechanical arm polishing path is optimized and revised.
It is of the present invention a kind of based on light-catalysed cobalt-base alloys processing method, which is characterized in that include the following steps:
(1) chemical reaction process is optimized and revised;
Step 1) calls such processing of in-memory data library storage for industrial personal computer operating system described in processing object
Object sample number m, unit are a, job evaluation index Q, and unit is mm/s;Wherein processing object sample number m corresponding chemistry
Reaction material and condition include 40 solvent strength C1 of basic yellow, and unit is g/L, TiO2Concentration C 2, unit are g/L, ultraviolet lighting
Intensity I, unit are lx and temperature T, and unit is DEG C;Job evaluation index Q, unit are that mm/s includes machining accuracy a, and unit is
Mm, surface roughness r, unit are μm and processing duration h that unit is s, and wherein the bigger processing quality of Q value is better, and mathematics is public
Formula is as follows:
It is as follows that step 2) establishes linear regression model (LRM) function formula Q', w0、w1、w2、w3For weight parameter random initializtion:
Q'=w0+w1×C1+w2×C2+w3×I+w4×T (2);
Step 3) establishes feature vector x, and x is by basic yellow 40 solvent strength C1, TiO2Concentration C 2, ultraviolet light intensity I,
The vector of temperature T composition, x=[C1, C2, I, T];Establish loss function, penalty values L1Come measure obtained by formula (2) it is linear
Difference between regression model output Q ' and the actual job evaluation index Q obtained by formula (1), as a result formula is as follows, Qi
And Q'iRespectively i-th processing actual evaluation index value and prediction evaluation index value:
Step 4) updates weighted value using random batch gradient descent algorithm come iteration until reaching given number of iterations (in advance
Setting 1000 times), weight parameter value when penalty values minimum is calculated, weighted value is updated using following iterative formula, α
It is that learning rate generally takes 0.001, t to indicate iteration round, t is set as 0 when iteration starts, and first passes through program in advance and generates Gauss point
The random number of cloth initializes,It updates 1000 times altogether and namely needs to calculate 1000 formula
(4), it all needs for be calculated the last time each time In four weight parameter generations, go back to formula (4) and calculate
This wheel is updated
WhereinRepresent iterative calculation;
After step 5) completes iterationIt is exactly corresponding w0、w1、w2、w3Weighted value
Prediction model can be established out by substituting into formula (2), can predict different 40 solvent strengths of basic yellow by prediction model,
TiO2 concentration, ultraviolet light intensity, at a temperature of job evaluation refer to target value;
Step 6) is solved according to prediction model when Q ' value maximum, i.e., job evaluation index highest when, basic yellow 40
Solvent strength C1, TiO2Concentration C 2, ultraviolet light intensity L, the corresponding value of temperature T;Again by operating system by parameter instruction
Change, ultraviolet lamp and polishing fluid adding set is communicated to, to achieve the purpose that chemical reaction process is optimized and revised;
(2) mechanical arm polishing path is optimized and revised;
1. step is directed to the desired trajectory that industrial personal computer operating system described in processing object calls the storage of Cloud Server database
yk, it is made of k coordinate points, actual motion track is y' in tactful network training processk, establish error function ek, formula is such as
Under:
ek=yk-y'k (5)
Make e by the training of tactful network modelkValue becomes smaller as far as possible, that is, allow polishing locus error reduce under
Come;
2. step sets reward value R according to error amount, the smaller reward value of error is bigger, in order to allow tactful network preferably to manage
The following and potential reward of solution, used here as the reward of future decaying, it is assumed that action action is secondary, RkIt is obtained for kth
Reward, formula are as follows:
R=R1+γR2+γ2R3+...+γk-1Rk(γ is that attenuation coefficient takes 0.99) (6);
3. step constructs neural network model, input is the current coordinate value of mechanical arm, exports as mechanical arm towards each
Pro-bability value matrices p=[the p of azimuth motion1,p2,p3,...,pn] (n is determined by the orientation number set), it can gather around for selecting to go
It is dynamic;
4. step establishes loss function L2, using cross entropy loss function, formula is as follows:
L2=-(r*log p)/n (7)
So that the probability for obtaining the action of more reward value improves, the probability for obtaining lower reward value action is reduced;
5. step substitutes into loss function, be iterated more using gradient descent algorithm to the weighted value in neural network
Newly, the tactful network of optimal policy can be provided by finally obtaining one, can be carried out to mechanical arm polishing path based on tactful network excellent
Change adjustment.
Processing platform according to the present invention based on the building of light-catalysed cobalt-base alloys processing method, feature exist
In:Including
Robot module, including industrial robot and flexible pneumatic burnishing device, flexible polishing device are mounted on industrial machine
On the mechanical arm of device people, for adjusting the location of workpiece;
Polishing station, including the light for holding polishing fluid polishing disk and for being irradiated to polishing fluid and workpiece
According to device, illumination apparatus is mounted in polishing disk, and the irradiation head of illumination apparatus is directed at polishing fluid;
Polishing fluid adding set, is set to by polishing station, and its filling opening is located at the polishing pond of polishing station
Surface, for adding polishing fluid into polishing station;
Industrial personal computer, is electrically connected or signal with the control terminal of robot module, polishing station and polishing fluid adding set
Connection realizes chemical reaction process optimization for controlling robot module, polishing station and the work of polishing fluid adding set
It adjusts and mechanical arm polishing path is optimized and revised;
And memory, signal transmission end are double by the signal transmission port of conducting wire or wireless network and the industrial personal computer
It connects or is electrically connected to signal, for storing machining information.
The polishing pan bottom and the rotary table that is driven by rotating electric machine are affixed, realization polishing disk around itself it is vertical in
Mandrel rotation;The elastic layer for being used to support abrasive grain, elastic film layer are successively arranged in the polishing pond from top to bottom and is used for
To the abrasive grain layer that die surface is processed, wherein the bottom end of elastic layer is connect with polishing pond inner bottom plating, the upper surface of elastic layer
It is laid with elastic film layer, and the inner wall in the edge of elastic film layer and polishing pond is tightly connected, so that elastic film layer and polishing
Pond inner wall surrounds the processing district for accommodating polishing fluid;Abrasive grain layer is laid on the surface of elastic film layer, fills and uses on abrasive grain layer
In the polishing fluid of lubrication and heat dissipation;The control terminal electrical connection corresponding with master controller of the control terminal of rotating electric machine.
The elastic layer is oval section volute spring array, and the oval section volute spring array includes several mutually independent cones
Shape helical spring, the small-caliber end of the oval section volute spring and the inner bottom surface in the polishing pond are affixed, the cone-type spiral bullet
The large diameter end of spring is connected with a hemispherical support, the arc curve of the hemispherical support and the following table face contact of elastic layer;
The upper end of the oval section volute spring is against elastic film layer lower surface.
The illumination apparatus is ultraviolet lamp, is fixed on rotary table center, wherein ultraviolet lamp is radiated at workpiece always and adds
Work area.
The chemical reaction process optimization is adjusted using linear regression algorithm.The mechanical arm polishing path
Optimization is adjusted polishing locus using tactful network.
The data stored in the database of the memory are based on history process data.
The model of neural network and the selection of parameter are adjusted according to actual processing in tactful network.
The beneficial effects of the present invention are:The present invention mutually ties artificial intelligence technology with photocatalysis processing of robots platform
It closes, the ultraviolet ray intensity of chemical reaction process and the amount of chemical addition agent are bad fixed in effective solution photocatalysis process
The problem of, it can effectively promote quality of finish;Mechanical arm polishing path is trained for different processing objects, can effectively solve
The flexible bring trajectory error of polishing disk, promotes the quality and efficiency of polishing.
Detailed description of the invention
Fig. 1 is that chemical reaction process of the invention optimizes and revises flow diagram.
Fig. 2 is that mechanical arm polishing path of the invention optimizes and revises flow diagram.
Fig. 3 is photocatalysis machining robot platform structure of the present invention.
Specific embodiment
The present invention is further illustrated with reference to the accompanying drawing.
Referring to attached drawing:
Embodiment 1 is of the present invention a kind of to be based on light-catalysed cobalt-base alloys processing method, which is characterized in that including with
Lower step:
(1) parameter needed for optimization chemically reacts;
Step 1) calls such processing of in-memory data library storage for industrial personal computer operating system described in processing object
Object sample number m (a), job evaluation index Q (mm/s);The wherein corresponding chemical reaction material of processing object sample number m (a)
It include 40 solvent strength C1 (g/L) of basic yellow, TiO with condition2Concentration C 2 (g/L), ultraviolet light intensity I (lx) and temperature T
(℃);Job evaluation index Q (mm/s) includes machining accuracy a (mm), surface roughness r (μm) and processes duration h (s),
The middle bigger processing quality of Q value is better, and mathematical formulae is as follows:
It is as follows that step 2) establishes linear regression model (LRM) function formula Q', w0、w1、w2、w3For weight parameter random initializtion:
Q'=w0+w1×C1+w2×C2+w3×I+w4×T (2);
Step 3) establishes feature vector x, and x is by basic yellow 40 solvent strength C1, TiO2Concentration C 2, ultraviolet light intensity I,
The vector of temperature T composition, x=[C1, C2, I, T];Establish loss function, penalty values L1To measure model output Q ' (formula (2)
Obtain) and practical Q (being obtained by formula (1)) between difference, as a result formula is as follows, QiAnd Q'iRespectively i-th processing is practical
Evaluation index value and prediction evaluation index value.
Step 4) updates weighted value using random batch gradient descent algorithm come iteration until reaching given number of iterations (in advance
Setting 1000 times), weight parameter value when penalty values minimum is calculated, weighted value is updated using following iterative formula, α
It is that learning rate generally takes 0.001, t to indicate iteration round, t is set as 0 when iteration starts, and first passes through program in advance and generates Gauss point
The random number of cloth initializes,It updates 1000 times altogether and namely needs to calculate 1000 formula
(4), it all needs for be calculated the last time each time In four weight parameter generations, go back to formula (4) and calculate
This wheel is updated
WhereinRepresent iterative calculation;
After step 5) completes iterationIt is exactly corresponding w0、w1、w2、w3Weighted value
Prediction model can be established out by substituting into formula (2), can predict different 40 solvent strengths of basic yellow by prediction model,
TiO2 concentration, ultraviolet light intensity, at a temperature of job evaluation refer to target value;
Step 6) is solved according to prediction model when Q ' value maximum, i.e., job evaluation index highest when, basic yellow 40
Solvent strength C1, TiO2 concentration C 2, ultraviolet light intensity L, the corresponding value of temperature T;Again by operating system by parameter instruction
Change, ultraviolet lamp and polishing fluid adding set is communicated to, to achieve the purpose that chemical reaction process is optimized and revised;
(2) mechanical arm polishing path is optimized and revised;
1. step is directed to the desired trajectory that industrial personal computer operating system described in processing object calls the storage of Cloud Server database
yk, it is made of k coordinate points, actual motion track is y' in tactful network training processk, establish error function ek, formula is such as
Under:
ek=yk-y'k (5)
Make e by the training of tactful network modelkValue becomes smaller as far as possible, that is, allow polishing locus error reduce under
Come;
2. step sets reward value R according to error amount, the smaller reward value of error is bigger, in order to allow tactful network preferably to manage
The following and potential reward of solution, used here as the reward of future decaying, it is assumed that action action is secondary, RkIt is obtained for kth
Reward, formula are as follows:
R=R1+γR2+γ2R3+...+γk-1Rk(γ is that attenuation coefficient takes 0.99) (6);
3. step constructs neural network model, input is the current coordinate value of mechanical arm, exports as mechanical arm towards each
Pro-bability value matrices p=[the p of azimuth motion1,p2,p3,...,pn] (n is determined by the orientation number set), it can gather around for selecting to go
It is dynamic;
4. step establishes loss function, using cross entropy loss function, formula is as follows:
L2=-(r*log p)/n (7)
So that the probability for obtaining the action of more reward value improves, the probability for obtaining lower reward value action is reduced;
5. step substitutes into loss function, be iterated more using gradient descent algorithm to the weighted value in neural network
Newly, the tactful network of optimal policy can be provided by finally obtaining one, can be carried out to mechanical arm polishing path based on tactful network excellent
Change adjustment.
The processing platform according to the present invention based on the building of light-catalysed cobalt-base alloys processing method of embodiment 2, packet
It includes
Robot module 1, including industrial robot and flexible pneumatic burnishing device, flexible polishing device are mounted on industrial machine
On the mechanical arm of device people, for adjusting the location of workpiece;
Polishing station 2, including for holding polishing fluid polishing disk and for being irradiated to polishing fluid and workpiece
Illumination apparatus, illumination apparatus are mounted in polishing disk, and the irradiation head of illumination apparatus is directed at polishing fluid;
Polishing fluid adding set 3, is set to by polishing station, and its filling opening is located at the polishing pond of polishing station
Surface, for adding polishing fluid into polishing station;
Industrial personal computer 4, is electrically connected or signal with the control terminal of robot module, polishing station and polishing fluid adding set
Connection realizes chemical reaction process optimization for controlling robot module, polishing station and the work of polishing fluid adding set
It adjusts and mechanical arm polishing path is optimized and revised;
And memory 5, the signal transmission port that signal transmission end passes through conducting wire or wireless network and the industrial personal computer
Two-way signaling connection or electrical connection, for storing machining information.
The polishing pan bottom and the rotary table that is driven by rotating electric machine are affixed, realization polishing disk around itself it is vertical in
Mandrel rotation;The elastic layer for being used to support abrasive grain, elastic film layer are successively arranged in the polishing pond from top to bottom and is used for
To the abrasive grain layer that die surface is processed, wherein the bottom end of elastic layer is connect with polishing pond inner bottom plating, the upper surface of elastic layer
It is laid with elastic film layer, and the inner wall in the edge of elastic film layer and polishing pond is tightly connected, so that elastic film layer and polishing
Pond inner wall surrounds the processing district for accommodating polishing fluid;Abrasive grain layer is laid on the surface of elastic film layer, fills and uses on abrasive grain layer
In the polishing fluid of lubrication and heat dissipation;The control terminal electrical connection corresponding with master controller of the control terminal of rotating electric machine.
The elastic layer is oval section volute spring array, and the oval section volute spring array includes several mutually independent cones
Shape helical spring, the small-caliber end of the oval section volute spring and the inner bottom surface in the polishing pond are affixed, the cone-type spiral bullet
The large diameter end of spring is connected with a hemispherical support, the arc curve of the hemispherical support and the following table face contact of elastic layer;
The upper end of the oval section volute spring is against elastic film layer lower surface.
The illumination apparatus is ultraviolet lamp, is fixed on rotary table center, wherein ultraviolet lamp is radiated at workpiece always and adds
Work area.
The chemical reaction process optimization is adjusted using linear regression algorithm.The mechanical arm polishing path
Optimization is adjusted polishing locus using tactful network.
The data stored in the database of the memory are based on history process data.
The model of neural network and the selection of parameter are adjusted according to actual processing in tactful network.
The formula of the polishing fluid is:The ingredient of polishing fluid includes 40 solvent of basic yellow and TiO2Solvent and water, wherein
The mass concentration of basic yellow 40 is 80g/L~100g/L;The TiO2Mass concentration be 120g/L~150g/L.Test every time
All use the basic yellow 40 and TiO of various concentration2It is made into polishing fluid.
Case study on implementation 3
The basic yellow 40 and TiO of various concentration are stored in the polishing fluid adding set2Solvent, it is molten by solenoid valve control
The outflow of agent, all outflows of STOCHASTIC CONTROL solvent of each processing experiment simultaneously calculate and record the concentration for having used solvent, mix it
After be uniformly covered on work piece surface, then polished by mechanical arm.
The solvent strength data will be transmitted in the database of server be stored for after call.
Content described in this specification embodiment is only enumerating to the way of realization of inventive concept, protection of the invention
Range should not be construed as being limited to the specific forms stated in the embodiments, and protection scope of the present invention also includes art technology
Personnel conceive according to the present invention it is conceivable that equivalent technologies mean.
Claims (5)
1. one kind is based on light-catalysed cobalt-base alloys processing method, which is characterized in that include the following steps:
(1) chemical reaction process is optimized and revised;
Step 1) calls such processing object of in-memory data library storage for industrial personal computer operating system described in processing object
Sample number m, unit are a, job evaluation index Q, and unit is mm/s;Wherein processing object sample number m corresponding chemical reactions
Material and condition include 40 solvent strength C1 of basic yellow, and unit is g/L, TiO2Concentration C 2, unit are g/L, ultraviolet light intensity
I, unit are lx and temperature T, and unit is DEG C;Job evaluation index Q, it includes machining accuracy a that unit, which is mm/s, unit be mm,
Surface roughness r, unit are μm and processing duration h that unit is s, and wherein the bigger processing quality of Q value is better, mathematical formulae
It is as follows:
It is as follows that step 2) establishes linear regression model (LRM) function formula Q', w0、w1、w2、w3For weight parameter random initializtion:
Q'=w0+w1×C1+w2×C2+w3×L+w4×T(2)
Step 3) establishes feature vector x, and x is by 40 solvent strength C1, TiO2 concentration C 2 of basic yellow, ultraviolet light intensity I, temperature
The vector of T composition, x=[C1, C2, I, T];Establish loss function, penalty values L1To measure the linear regression obtained by formula (2)
Difference between model output Q ' and the actual job evaluation index Q obtained by formula (1), as a result formula is as follows, QiAnd Q'i
Respectively i-th processing actual evaluation index value and prediction evaluation index value:
Step 4) updates weighted value until reaching given number of iterations come iteration using random batch gradient descent algorithm, pre-sets
1000 times, weight parameter value when penalty values minimum is calculated, weighted value is updated using following iterative formula, α is to learn
Habit rate generally takes 0.001, t to indicate iteration round, and t is set as 0 when iteration starts, and first passes through program in advance and generates Gaussian Profile
Random number initialization,It updates 1000 times altogether and namely needs to calculate 1000 formula (4),
It each time all needs for be calculated the last time In four weight parameter generations, go back to formula (4) and calculate this
It takes turns updated
Wherein Represent iterative calculation;
After step 5) completes iterationIt is exactly corresponding w0、w1、w2、w3Weighted value substitutes into
Formula (2) can establish out prediction model, can predict different 40 solvent strengths of basic yellow, TiO2 by prediction model
Concentration, ultraviolet light intensity, at a temperature of job evaluation refer to target value;
Step 6) is solved according to prediction model when Q ' value maximum, i.e., job evaluation index highest when, 40 solvent of basic yellow
Concentration C 1, TiO2 concentration C 2, ultraviolet light intensity L, the corresponding value of temperature T;Parameter instruction is passed by operating system again
Up to ultraviolet lamp and polishing fluid adding set, to achieve the purpose that chemical reaction process is optimized and revised;
(2) mechanical arm polishing path is optimized and revised;
1. step is directed to the desired trajectory y that industrial personal computer operating system described in processing object calls the storage of Cloud Server databasek, by k
A coordinate points are constituted, and actual motion track is y' in tactful network training processk, establish error function ek, formula is as follows:
ek=yk-y'k (5)
Make e by the training of tactful network modelkValue becomes smaller as far as possible, that is, the error of polishing locus is allowed to reduce;
2. step sets reward value R according to error amount, the smaller reward value of error is bigger, in order to allow tactful network to better understand not
Come with potentially reward, used here as future decaying reward, it is assumed that action action be secondary, RkFor kth obtain reward,
Its formula is as follows:
R=R1+γR2+γ2R3+...+γk-1Rk(γ is that attenuation coefficient takes 0.99) (6);
3. step constructs neural network model, input is the current coordinate value of mechanical arm, exports as mechanical arm towards each orientation
Pro-bability value matrices p=[the p of movement1,p2,p3,...,pn] (n is determined by the orientation number set), it can gather around for selecting to take action;
4. step establishes loss function L2, using cross entropy loss function, formula is as follows:
L2=-(r*logp)/n (7)
So that the probability for obtaining the action of more reward value improves, the probability for obtaining lower reward value action is reduced;
5. step substitutes into loss function, be iterated update to the weighted value in neural network using gradient descent algorithm, most
The tactful network of optimal policy can be provided by obtaining one eventually, and tune can be optimized to mechanical arm polishing path based on tactful network
It is whole.
2. the processing platform as described in claim 1 based on the building of light-catalysed cobalt-base alloys processing method, it is characterised in that:
Including
Robot module, including industrial robot and flexible pneumatic burnishing device, flexible polishing device are mounted on industrial robot
Mechanical arm on, for adjusting the location of workpiece;
Polishing station, including the illumination dress for holding polishing fluid polishing disk and for being irradiated to polishing fluid and workpiece
It sets, illumination apparatus is mounted in polishing disk, and the irradiation head of illumination apparatus is directed at polishing fluid;
Polishing fluid adding set, is set to by polishing station, and its filling opening is located at the polishing Chi Zhengshang of polishing station
Side, for adding polishing fluid into polishing station;
Industrial personal computer is connect with the electrical connection of the control terminal of robot module, polishing station and polishing fluid adding set or signal,
For controlling robot module, polishing station and the work of polishing fluid adding set, realize that chemical reaction process is optimized and revised
It is optimized and revised with to mechanical arm polishing path;
And memory, signal transmission end pass through conducting wire or the two-way letter of signal transmission port of wireless network and the industrial personal computer
Number connection or electrical connection, for storing machining information.
3. processing platform as claimed in claim 2, it is characterised in that:The polishing pan bottom and the rotation driven by rotating electric machine
Revolving worktable is affixed, realizes that polishing disk is rotated around itself vertical center axis;It is successively arranged and is used for from top to bottom in the polishing pond
Elastic layer, elastic film layer and the abrasive grain layer for being processed to die surface of abrasive grain are supported, wherein the bottom of elastic layer
End is connect with polishing pond inner bottom plating, the upper surface laying elastic film layer of elastic layer, and the edge of elastic film layer and polishing pond
Inner wall be tightly connected so that elastic film layer and polishing pond inner wall surround the processing district for accommodating polishing fluid;Abrasive grain layer paving
It is located at the surface of elastic film layer, polishing fluid of the filling for lubricating and radiating on abrasive grain layer;The control terminal of rotating electric machine and total
The corresponding control terminal electrical connection of controller.
4. processing platform as claimed in claim 3, it is characterised in that:The elastic layer is oval section volute spring array, described
Oval section volute spring array includes several mutually independent oval section volute springs, the small-caliber end of the oval section volute spring and institute
The inner bottom surface for stating polishing pond is affixed, and the large diameter end of the oval section volute spring is connected with a hemispherical support, the hemispherical
The arc curve of bracket and the following table face contact of elastic layer;The upper end of the oval section volute spring is against elastic film layer following table
Face.
5. processing platform as claimed in claim 3, it is characterised in that:The illumination apparatus is ultraviolet lamp, is fixed on rotation work
Make platform center, wherein ultraviolet lamp is radiated at work pieces process area always.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810828984.0A CN108857600B (en) | 2018-07-25 | 2018-07-25 | Cobalt-based alloy processing method and processing platform based on photocatalysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810828984.0A CN108857600B (en) | 2018-07-25 | 2018-07-25 | Cobalt-based alloy processing method and processing platform based on photocatalysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108857600A true CN108857600A (en) | 2018-11-23 |
CN108857600B CN108857600B (en) | 2023-07-14 |
Family
ID=64305297
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810828984.0A Active CN108857600B (en) | 2018-07-25 | 2018-07-25 | Cobalt-based alloy processing method and processing platform based on photocatalysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108857600B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110421574A (en) * | 2019-07-22 | 2019-11-08 | 北京理工大学 | A kind of robot intention movement assessment generation system |
CN110842761A (en) * | 2019-11-19 | 2020-02-28 | 湖南大学 | Photocatalysis high-energy field assisted chemical mechanical composite fine grinding method |
CN114406890A (en) * | 2021-12-20 | 2022-04-29 | 浙江工业大学 | Ultraviolet light response self-assembly cooperative dielectrophoresis polishing method and device |
CN116372781A (en) * | 2023-04-20 | 2023-07-04 | 山东欣立得光电科技有限公司 | Automatic cleaning and polishing system for LED screen substrate |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020187728A1 (en) * | 2000-01-31 | 2002-12-12 | Etsuo Kiuchi | Polishing device and method |
US20040166686A1 (en) * | 2003-02-20 | 2004-08-26 | Taiwan Semiconductor Manufacturing Co., Ltd. | Dynamically adjustable slurry feed arm for wafer edge profile improvement in CMP |
CN1606713A (en) * | 2001-11-15 | 2005-04-13 | 霍尼韦尔国际公司 | Spincoating antireflection paint for photolithography |
CN104669075A (en) * | 2014-12-08 | 2015-06-03 | 沈阳工业大学 | Photocatalytic auxiliary cutter sharpening method and device for diamond cutter |
CN106881659A (en) * | 2017-03-16 | 2017-06-23 | 浙江工业大学 | A kind of burnishing device with profiling abrasive particle group |
CN206614394U (en) * | 2017-03-16 | 2017-11-07 | 浙江工业大学 | A kind of burnishing device with profiling abrasive particle group |
CN107745324A (en) * | 2017-09-07 | 2018-03-02 | 天津大学 | A kind of optical glass surface forming method |
JP2018051636A (en) * | 2016-09-26 | 2018-04-05 | 三菱マテリアルテクノ株式会社 | Soft sheet manufacturing method and soft sheet polishing device |
-
2018
- 2018-07-25 CN CN201810828984.0A patent/CN108857600B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020187728A1 (en) * | 2000-01-31 | 2002-12-12 | Etsuo Kiuchi | Polishing device and method |
CN1606713A (en) * | 2001-11-15 | 2005-04-13 | 霍尼韦尔国际公司 | Spincoating antireflection paint for photolithography |
US20040166686A1 (en) * | 2003-02-20 | 2004-08-26 | Taiwan Semiconductor Manufacturing Co., Ltd. | Dynamically adjustable slurry feed arm for wafer edge profile improvement in CMP |
CN104669075A (en) * | 2014-12-08 | 2015-06-03 | 沈阳工业大学 | Photocatalytic auxiliary cutter sharpening method and device for diamond cutter |
JP2018051636A (en) * | 2016-09-26 | 2018-04-05 | 三菱マテリアルテクノ株式会社 | Soft sheet manufacturing method and soft sheet polishing device |
CN106881659A (en) * | 2017-03-16 | 2017-06-23 | 浙江工业大学 | A kind of burnishing device with profiling abrasive particle group |
CN206614394U (en) * | 2017-03-16 | 2017-11-07 | 浙江工业大学 | A kind of burnishing device with profiling abrasive particle group |
CN107745324A (en) * | 2017-09-07 | 2018-03-02 | 天津大学 | A kind of optical glass surface forming method |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110421574A (en) * | 2019-07-22 | 2019-11-08 | 北京理工大学 | A kind of robot intention movement assessment generation system |
CN110842761A (en) * | 2019-11-19 | 2020-02-28 | 湖南大学 | Photocatalysis high-energy field assisted chemical mechanical composite fine grinding method |
CN114406890A (en) * | 2021-12-20 | 2022-04-29 | 浙江工业大学 | Ultraviolet light response self-assembly cooperative dielectrophoresis polishing method and device |
CN116372781A (en) * | 2023-04-20 | 2023-07-04 | 山东欣立得光电科技有限公司 | Automatic cleaning and polishing system for LED screen substrate |
CN116372781B (en) * | 2023-04-20 | 2023-11-07 | 山东欣立得光电科技有限公司 | Automatic cleaning and polishing system for LED screen substrate |
Also Published As
Publication number | Publication date |
---|---|
CN108857600B (en) | 2023-07-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108857600A (en) | One kind being based on light-catalysed cobalt-base alloys processing method and processing platform | |
JP2003534460A (en) | Tuning electrode used in a reactor for electrochemically processing microelectronic workpieces | |
US8571706B2 (en) | Method and system for optimizing the layout of a robot work cell | |
CN109202904B (en) | Method and system for determining motion path of mechanical arm | |
CN104375478B (en) | A kind of method and device of Rolling production process product quality on-line prediction and optimization | |
CN105976122B (en) | Multi-target resource allocation system | |
AU2018267755A1 (en) | Array element arrangement method for L-type array antenna based on inheritance of acquired characteristics | |
CN108563121A (en) | Microwave heating temperature field intelligent control method based on historical data | |
CN106022601B (en) | Multi-target resource allocation method | |
CN114386364B (en) | Electronic surface mounting welding quality prediction and process parameter optimization method | |
CN112536797A (en) | Comprehensive compensation method for position and attitude errors of industrial robot | |
CN104281722A (en) | Automatic discharging method | |
US20220207330A1 (en) | Operational neural networks and self-organized operational neural networks with generative neurons | |
Singh et al. | Modelling of direct metal laser sintering of EOS DM20 bronze using neural networks and genetic algorithms | |
CN110045613A (en) | MIXED INTEGER optimum control method of value solving based on Quantum annealing | |
CN107623171A (en) | The preparation method and mobile device of a kind of antenna | |
US20200324367A1 (en) | Method for aligning a plurality of laser lines | |
Mishra et al. | An improved hybrid flower pollination algorithm for assembly sequence optimization | |
Wu et al. | Applications of AI techniques to generation planning and investment | |
Zhang et al. | Surface quality optimization of laser cladding based on surface response and genetic neural network model | |
Keung et al. | A genetic algorithm approach to the multiple machine tool selection problem | |
CN109112466A (en) | Plasma spraying Ni Superalloy Coating high throughput preparation method | |
CN106312250B (en) | A kind of electric arc fuse increasing material manufacturing method of disc supporting item | |
CN116151589A (en) | Green repairable workshop production scheduling method based on improved moth fire suppression algorithm | |
CN110533151A (en) | A kind of firefly optimization algorithm based on the law of universal gravitation |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |