CN108857600B - Cobalt-based alloy processing method and processing platform based on photocatalysis - Google Patents
Cobalt-based alloy processing method and processing platform based on photocatalysis Download PDFInfo
- Publication number
- CN108857600B CN108857600B CN201810828984.0A CN201810828984A CN108857600B CN 108857600 B CN108857600 B CN 108857600B CN 201810828984 A CN201810828984 A CN 201810828984A CN 108857600 B CN108857600 B CN 108857600B
- Authority
- CN
- China
- Prior art keywords
- polishing
- processing
- value
- formula
- layer
- 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.)
- Active
Links
Images
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)
- Grinding And Polishing Of Tertiary Curved Surfaces And Surfaces With Complex Shapes (AREA)
- Finish Polishing, Edge Sharpening, And Grinding By Specific Grinding Devices (AREA)
Abstract
The cobalt-based alloy processing method based on photocatalysis and the processing platform thereof comprise optimization adjustment of a chemical reaction process and optimization adjustment of a mechanical arm polishing path, wherein the optimization adjustment of the chemical reaction process adopts the conditions of establishing a linear regression model for different processing objects based on a linear regression algorithm to search for optimal illumination intensity, chemical additive proportion, temperature and the like; the mechanical arm polishing path optimization adjustment adopts a strategy network to adjust the polishing track; the processing platform includes: robot module, polishing workstation, polishing liquid interpolation device, industrial computer and memory. The invention has the beneficial effects that: the problems of poor ultraviolet intensity and poor amount of chemical additives in the chemical reaction process in the photocatalysis processing process are effectively solved, and the polishing quality can be effectively improved; the mechanical arm polishing paths are trained for different processing objects, so that the track error caused by the flexibility of the polishing disc can be effectively solved, and the polishing quality and efficiency are improved.
Description
Technical Field
The invention relates to a cobalt-based alloy processing method and a processing platform based on photocatalysis.
Background
Cobalt-based alloys can be used as self-fluxing alloy materials to improve the surface properties of metallic materials to form high hardness, high temperature and wear resistant coatings. The method is widely applied to the industries of mould manufacturing, automobile manufacturing industry, turbine manufacturing and the like. The cobalt-based alloy coating of the die not only has the complexity of a free curved surface, but also encounters new problems of high hardness, high wear resistance and the like, so that precision machining is more difficult, and a common machining method is difficult to quickly remove. Aiming at the material characteristics of cobalt element in cobalt-based alloy, an effective reaction solvent is configured to enable cobalt atoms on the surface layer to react to form a product which is easy to remove, and the relationship between photons and quantum yield is established by combining the ultraviolet light catalysis, so that the processes of cobalt element atomic bond breakage and energy release are regulated, and the reaction is controllable. Because the cobalt-based alloy is a multi-element alloy material, the material removal of insoluble products is realized by the cooperative mechanical action.
The process platform material removal process is a synergistic process of chemical reaction, optical control and mechanical removal. Therefore, the influence of the processing technology on the processing result is relatively large, in order to optimize the processing technology, the capability of the artificial intelligence technology in continuity decision and searching for the optimal point is utilized, the artificial intelligence technology is combined with a processing platform of the photocatalytic robot, and the processing efficiency and the processing quality of the polishing of the photocatalytic robot are improved.
Disclosure of Invention
The invention aims to provide a photocatalysis processing platform suitable for an artificial intelligence technology, which can select optimal chemical reaction conditions for different processing objects and solve the problem of polishing path errors caused by a flexible polishing disk when polishing different processing objects. The processing platform can adjust processing technology for different processing objects based on an artificial intelligence technology, and comprises the steps of optimizing and adjusting a chemical reaction process based on a linear regression model and optimizing and adjusting a mechanical arm polishing path based on a strategy network model.
The invention discloses a cobalt-based alloy processing method based on photocatalysis, which is characterized by comprising the following steps of:
(1) Optimizing and adjusting the chemical reaction process;
step 1), calling the number m of samples of the processing object stored in a database in a memory aiming at the operating system of the industrial personal computer, wherein the units are a number Q and a processing evaluation index Q, and the units are mm/s; wherein the number of processing object samples is m corresponding to the chemical reaction materialsAnd the conditions include a basic yellow 40 solvent concentration C1 in g/L, tiO 2 Concentration C2, the unit is g/L, ultraviolet illumination intensity I, the unit is lx and temperature T, the unit is DEG C; the machining evaluation index Q is in mm/s and comprises machining precision a, mm, surface roughness r, mu m and machining duration h, and s, wherein the machining quality is better when the Q value is larger, and the mathematical formula is as follows:
step 2) establishing a linear regression model function formula Q' as follows, w 0 、w 1 、w 2 、w 3 Randomly initializing weight parameters:
Q'＝w 0 +w 1 ×C1+w 2 ×C2+w 3 ×I+w 4 ×T (2)；
step 3) establishing a characteristic vector x, wherein x is obtained by the concentration C1 and TiO of an alkaline yellow 40 solvent 2 Concentration C2, intensity of ultraviolet illumination I, vector of temperature T, x= [ C1, C2, I, T]The method comprises the steps of carrying out a first treatment on the surface of the Establishing a loss function, and a loss value L 1 To measure the difference between the linear regression model output Q' obtained from the formula (2) and the actual process evaluation index Q obtained from the formula (1), the result formula is as follows, Q i And Q' i The i-th machining actual evaluation index value and the predicted evaluation index value are respectively:
step 4) iteratively updating the weight value by using a random batch gradient descent algorithm until reaching the designated iteration times (preset 1000 times), calculating the weight parameter value with the minimum loss value, updating the weight value by adopting an iteration formula, wherein alpha is the learning rate and is generally 0.001, t represents iteration rounds, t is set to 0 at the beginning of iteration, random number initialization of Gaussian distribution is generated in advance through a program,
a total of 1000 updates, i.e. 1000 calculations of equation (4), are needed, each time the last calculated +.>
The four weight parameters are substituted back into formula (4) to calculate the updated +.>
Wherein the method comprises the steps of
All represent iterative computations;
step 5) after the iteration is completed
Is the corresponding w 0 、w 1 、w 2 、w 3 The weight value is substituted into the formula (2) to establish a prediction model, and the values of the processing evaluation indexes under different alkaline yellow 40 solvent concentrations, tiO2 concentrations, ultraviolet illumination intensities and temperatures can be predicted through the prediction model;
step 6) solving the concentration C1, tiO of the alkaline yellow 40 solvent when the Q' value is maximum, namely, the processing evaluation index is highest according to the prediction model 2 Concentration C2, ultraviolet irradiation intensity L and temperature T respectively correspond to the values; the operating system sends the instructions of parameters to the ultraviolet lamp and the polishing solution adding device, so that the aim of optimizing and adjusting the chemical reaction process is fulfilled;
(2) Optimizing and adjusting the polishing path of the mechanical arm;
step (1) calling an expected track y stored in a cloud server database by the industrial personal computer operating system aiming at a processing object k Consists of k coordinate points, and the actual motion trail in the strategy network training process is y' k Establishing an error function e k The formula is as follows:
e k ＝y k -y' k (5)
training e through strategic network models k The value becomes as small as possible, i.e., the error of the polishing track is reduced;
setting a reward value R according to the error value, wherein the smaller the error is, the larger the reward value is, and in order to enable the strategy network to better understand future and potential rewards, the future attenuated rewards are used, and the action is assumed to be the next time, R k The formula for the K-th prize is as follows:
R＝R 1 +γR 2 +γ 2 R 3 +...+γ k-1 R k (gamma is the attenuation coefficient to be 0.99) (6);
step (3) a neural network model is built, the input is the current coordinate value of the mechanical arm, and the output is a probability value matrix p= [ p ] of the mechanical arm moving towards each azimuth 1 ,p 2 ,p 3 ,...,p n ](n is determined by the set number of orientations) and can be used to select actions;
step (4) establishing a loss function L 2 The cross entropy loss function is used, and the formula is as follows:
L 2 ＝-(r*log p)/n (7)
the probability of obtaining actions with more rewards is improved, and the probability of obtaining actions with lower rewards is reduced;
and (5) substituting the loss function, and iteratively updating the weight value in the neural network by using a gradient descent algorithm to finally obtain a strategy network capable of providing an optimal strategy, wherein the polishing path of the mechanical arm can be optimally adjusted based on the strategy network.
The processing platform constructed by the cobalt-based alloy processing method based on photocatalysis is characterized in that: comprising
The robot module comprises an industrial robot and a flexible pneumatic polishing device, wherein the flexible polishing device is arranged on a mechanical arm of the industrial robot and is used for adjusting the position of a workpiece;
the polishing workbench comprises a polishing disc for containing polishing liquid and an illumination device for illuminating the polishing liquid and a workpiece, wherein the illumination device is arranged in the polishing disc, and an illumination head of the illumination device is aligned to the polishing liquid;
the polishing liquid adding device is arranged beside the polishing workbench, and a liquid adding port of the polishing liquid adding device is positioned right above a polishing pool of the polishing workbench and is used for adding polishing liquid into the polishing workbench;
the industrial personal computer is electrically connected or in signal connection with the control ends of the robot module, the polishing workbench and the polishing liquid adding device and is used for controlling the robot module, the polishing workbench and the polishing liquid adding device to work so as to realize optimal adjustment of the chemical reaction process and optimal adjustment of the polishing path of the mechanical arm;
and the signal transmission end of the memory is in bidirectional signal connection or electric connection with the signal transmission port of the industrial personal computer through a wire or a wireless network and is used for storing processing information.
The bottom of the polishing disc is fixedly connected with a rotary workbench driven by a rotary motor, so that the polishing disc can rotate around a vertical central shaft of the polishing disc; the polishing device comprises a polishing pool, a polishing layer, an elastic film layer, a polishing layer and a polishing layer, wherein the elastic layer, the elastic film layer and the abrasive layer are sequentially arranged in the polishing pool from bottom to top and used for supporting abrasive particles, the abrasive layer is used for processing the surface of a die, the bottom end of the elastic layer is connected with a bottom plate in the polishing pool, the elastic film layer is paved on the upper surface of the elastic layer, and the edge of the elastic film layer is connected with the inner wall of the polishing pool in a sealing manner, so that a processing area for containing polishing liquid is formed by the elastic film layer and the inner wall of the polishing pool; the abrasive particle layer is paved on the surface of the elastic film layer, and polishing liquid for lubrication and heat dissipation is filled on the abrasive particle layer; the control end of the rotating motor is electrically connected with the control end corresponding to the master controller.
The elastic layer is a conical spiral spring array, the conical spiral spring array comprises a plurality of mutually independent conical spiral springs, the small-caliber end of each conical spiral spring is fixedly connected with the inner bottom surface of the polishing pool, the large-caliber end of each conical spiral spring is connected with a hemispherical support, and the arc-shaped curve of each hemispherical support is in contact with the lower surface of the elastic layer; the upper end of the conical spiral spring is abutted against the lower surface of the elastic film layer.
The illumination device is an ultraviolet lamp and is fixed at the center of the rotary workbench, wherein the ultraviolet lamp always irradiates the workpiece processing area.
The chemical reaction process optimization is adjusted by adopting a linear regression algorithm. The mechanical arm polishing path is optimized by adopting a strategy network to adjust the polishing track.
The data stored in the database of the memory is based on historical process data.
The model and the parameters of the neural network in the strategy network are selected and adjusted according to the actual processing conditions.
The invention has the beneficial effects that: according to the invention, an artificial intelligence technology is combined with a photocatalytic robot processing platform, so that the problem that the ultraviolet intensity and the amount of chemical additives in the chemical reaction process are not good in the photocatalytic processing process is effectively solved, and the polishing quality can be effectively improved; the mechanical arm polishing paths are trained for different processing objects, so that the track error caused by the flexibility of the polishing disc can be effectively solved, and the polishing quality and efficiency are improved.
Drawings
FIG. 1 is a schematic diagram of the chemical reaction process optimization adjustment flow of the invention.
Fig. 2 is a schematic diagram of a polishing path optimization adjustment process of a mechanical arm according to the present invention.
Fig. 3 is a view showing a platform structure of the photocatalytic machining robot according to the present invention.
Detailed Description
The invention is further described below with reference to the drawings.
Referring to the drawings:
(1) Optimizing parameters required by chemical reaction;
step 1) the work for the processing objectThe control computer operating system calls the number m (number) of the processing object samples and the processing evaluation index Q (mm/s) stored in the database in the memory; wherein the chemical reaction materials and conditions corresponding to the number m (m) of the processing object samples comprise the concentration C1 (g/L) of the alkaline yellow 40 solvent and TiO 2 Concentration C2 (g/L), ultraviolet illumination intensity I (lx) and temperature T (DEG C); the machining evaluation index Q (mm/s) comprises machining precision a (mm), surface roughness r (mu m) and machining duration h(s), wherein the machining quality is better as the Q value is larger, and the mathematical formula is as follows:
step 2) establishing a linear regression model function formula Q' as follows, w 0 、w 1 、w 2 、w 3 Randomly initializing weight parameters:
Q'＝w 0 +w 1 ×C1+w 2 ×C2+w 3 ×I+w 4 ×T (2)；
step 3) establishing a characteristic vector x, wherein x is obtained by the concentration C1 and TiO of an alkaline yellow 40 solvent 2 Concentration C2, intensity of ultraviolet illumination I, vector of temperature T, x= [ C1, C2, I, T]The method comprises the steps of carrying out a first treatment on the surface of the Establishing a loss function, and a loss value L 1 To measure the difference between the model output Q' (from equation (2)) and the actual Q (from equation (1)), the result equation is as follows, Q i And Q' i The i-th machining actual evaluation index value and the predicted evaluation index value are respectively.
Step 4) iteratively updating the weight value by using a random batch gradient descent algorithm until reaching the designated iteration times (preset 1000 times), calculating the weight parameter value with the minimum loss value, updating the weight value by adopting an iteration formula, wherein alpha is the learning rate and is generally 0.001, t represents iteration rounds, t is set to 0 at the beginning of iteration, random number initialization of Gaussian distribution is generated in advance through a program,
a total of 1000 updates, i.e. 1000 calculations of equation (4), are needed, each time the last calculated +.>
The four weight parameters are substituted back into formula (4) to calculate the updated +.>
Wherein the method comprises the steps of
All represent iterative computations;
step 5) after the iteration is completed
Is the corresponding w 0 、w 1 、w 2 、w 3 The weight value is substituted into the formula (2) to establish a prediction model, and the values of the processing evaluation indexes under different alkaline yellow 40 solvent concentrations, tiO2 concentrations, ultraviolet illumination intensities and temperatures can be predicted through the prediction model;
step 6), solving the corresponding values of the concentration C1 of the alkaline yellow 40 solvent, the concentration C2 of TiO2, the ultraviolet irradiation intensity L and the temperature T when the Q' value is maximum, namely when the processing evaluation index is highest, according to the prediction model; the operating system sends the instructions of parameters to the ultraviolet lamp and the polishing solution adding device, so that the aim of optimizing and adjusting the chemical reaction process is fulfilled;
(2) Optimizing and adjusting the polishing path of the mechanical arm;
step (1) calling an expected track y stored in a cloud server database by the industrial personal computer operating system aiming at a processing object k Consists of k coordinate points, and the actual motion trail in the strategy network training process is y' k Establishing an error function e k The formula is as follows:
e k ＝y k -y' k (5)
training e through strategic network models k The value becomes as small as possible, i.e., the error of the polishing track is reduced;
setting a reward value R according to the error value, wherein the smaller the error is, the larger the reward value is, and in order to enable the strategy network to better understand future and potential rewards, the future attenuated rewards are used, and the action is assumed to be the next time, R k The formula for the K-th prize is as follows:
R＝R 1 +γR 2 +γ 2 R 3 +...+γ k-1 R k (gamma is the attenuation coefficient to be 0.99) (6);
step (3) a neural network model is built, the input is the current coordinate value of the mechanical arm, and the output is a probability value matrix p= [ p ] of the mechanical arm moving towards each azimuth 1 ,p 2 ,p 3 ,...,p n ](n is determined by the set number of orientations) and can be used to select actions;
and (4) establishing a loss function, and adopting a cross entropy loss function, wherein the formula is as follows:
L 2 ＝-(r*log p)/n (7)
the probability of obtaining actions with more rewards is improved, and the probability of obtaining actions with lower rewards is reduced;
and (5) substituting the loss function, and iteratively updating the weight value in the neural network by using a gradient descent algorithm to finally obtain a strategy network capable of providing an optimal strategy, wherein the polishing path of the mechanical arm can be optimally adjusted based on the strategy network.
Example 2A processing platform constructed according to the photocatalytic cobalt-based alloy processing method of the present invention comprises
The robot module 1 comprises an industrial robot and a flexible pneumatic polishing device, wherein the flexible polishing device is arranged on a mechanical arm of the industrial robot and is used for adjusting the position of a workpiece;
the polishing workbench 2 comprises a polishing disc for containing polishing liquid and an illumination device for illuminating the polishing liquid and a workpiece, wherein the illumination device is arranged in the polishing disc, and an illumination head of the illumination device is aligned with the polishing liquid;
the polishing solution adding device 3 is arranged beside the polishing workbench, and a liquid adding port of the polishing solution adding device is positioned right above a polishing pool of the polishing workbench and is used for adding polishing solution into the polishing workbench;
the industrial personal computer 4 is electrically connected or in signal connection with the control ends of the robot module, the polishing workbench and the polishing liquid adding device and is used for controlling the robot module, the polishing workbench and the polishing liquid adding device to work so as to realize optimal adjustment of the chemical reaction process and optimal adjustment of the polishing path of the mechanical arm;
and the signal transmission end of the memory 5 is in bidirectional signal connection or electric connection with the signal transmission port of the industrial personal computer through a wire or a wireless network and is used for storing processing information.
The bottom of the polishing disc is fixedly connected with a rotary workbench driven by a rotary motor, so that the polishing disc can rotate around a vertical central shaft of the polishing disc; the polishing device comprises a polishing pool, a polishing layer, an elastic film layer, a polishing layer and a polishing layer, wherein the elastic layer, the elastic film layer and the abrasive layer are sequentially arranged in the polishing pool from bottom to top and used for supporting abrasive particles, the abrasive layer is used for processing the surface of a die, the bottom end of the elastic layer is connected with a bottom plate in the polishing pool, the elastic film layer is paved on the upper surface of the elastic layer, and the edge of the elastic film layer is connected with the inner wall of the polishing pool in a sealing manner, so that a processing area for containing polishing liquid is formed by the elastic film layer and the inner wall of the polishing pool; the abrasive particle layer is paved on the surface of the elastic film layer, and polishing liquid for lubrication and heat dissipation is filled on the abrasive particle layer; the control end of the rotating motor is electrically connected with the control end corresponding to the master controller.
The elastic layer is a conical spiral spring array, the conical spiral spring array comprises a plurality of mutually independent conical spiral springs, the small-caliber end of each conical spiral spring is fixedly connected with the inner bottom surface of the polishing pool, the large-caliber end of each conical spiral spring is connected with a hemispherical support, and the arc-shaped curve of each hemispherical support is in contact with the lower surface of the elastic layer; the upper end of the conical spiral spring is abutted against the lower surface of the elastic film layer.
The illumination device is an ultraviolet lamp and is fixed at the center of the rotary workbench, wherein the ultraviolet lamp always irradiates the workpiece processing area.
The chemical reaction process optimization is adjusted by adopting a linear regression algorithm. The mechanical arm polishing path is optimized by adopting a strategy network to adjust the polishing track.
The data stored in the database of the memory is based on historical process data.
The model and the parameters of the neural network in the strategy network are selected and adjusted according to the actual processing conditions.
The formula of the polishing solution is as follows: the polishing solution comprises alkaline yellow 40 solvent and TiO 2 Solvent and water, wherein the mass concentration of the alkaline yellow 40 is 80 g/L-100 g/L; the TiO 2 The mass concentration of (3) is 120 g/L-150 g/L. Each test uses different concentrations of basic yellow 40 and TiO 2 Preparing polishing solution.
Example 3
The polishing solution adding device stores alkaline yellow 40 and TiO with different concentrations 2 The solvent is controlled to flow out by an electromagnetic valve, the flow out of the solvent is randomly controlled in each processing test, the concentration of the solvent is calculated and recorded, the solvent is uniformly covered on the surface of a workpiece after being mixed, and then the workpiece is polished by a mechanical arm.
The solvent concentration data will be transmitted to a database of the server for storage for later recall.
The embodiments described in the present specification are merely examples of implementation forms of the inventive concept, and the scope of protection of the present invention should not be construed as being limited to the specific forms set forth in the embodiments, but also equivalent technical means that can be conceived by those skilled in the art according to the inventive concept.
Claims (5)
1. The cobalt-based alloy processing method based on photocatalysis is characterized by comprising the following steps of:
(1) Optimizing and adjusting the chemical reaction process;
step 1), calling a processing object sample number m stored in a database in a memory aiming at an industrial personal computer operating system of a processing object, wherein the unit is a processing evaluation index Q and the unit is mm/s; wherein the number m of corresponding chemically reactive materials and conditions of the processing object sample include the concentration C1 of the basic yellow 40 solvent, the unit is g/L, tiO 2 Concentration C2, the unit is g/L, ultraviolet illumination intensity I, the unit is lx and temperature T, the unit is DEG C; the machining evaluation index Q is in mm/s and comprises machining precision a, mm, surface roughness r, mu m and machining duration h, and s, wherein the machining quality is better when the Q value is larger, and the mathematical formula is as follows:
step 2) establishing a linear regression model function formula Q' as follows, w 0 、w 1 、w 2 、w 3 Randomly initializing weight parameters:
Q'＝w 0 +w 1 ×C1+w 2 ×C2+w 3 ×L+w 4 ×T (2)
step 3) establishing a characteristic vector x, wherein x is a vector consisting of an alkaline yellow 40 solvent concentration C1, a TiO2 concentration C2, an ultraviolet illumination intensity I and a temperature T, and x= [ C1, C2, I and T ]]The method comprises the steps of carrying out a first treatment on the surface of the Establishing a loss function, and a loss value L 1 To measure the difference between the linear regression model output Q' obtained from the formula (2) and the actual process evaluation index Q obtained from the formula (1), the result formula is as follows, Q i And Q' i The i-th machining actual evaluation index value and the predicted evaluation index value are respectively:
step 4) utilizing a random batchThe weight value is updated iteratively by a quantity gradient descent algorithm until reaching the specified iteration times, 1000 times are preset, the weight parameter value with the minimum loss value is calculated, the weight value is updated by adopting an iteration formula, alpha is the learning rate and is generally 0.001, t represents iteration rounds, t is set to 0 at the beginning of iteration, random number initialization of Gaussian distribution is generated in advance through a program,
a total of 1000 updates, i.e. 1000 calculations of equation (4), are needed, each time the last calculated +.>
The four weight parameters are substituted back into formula (4) to calculate the updated +.>
Wherein the method comprises the steps of
All represent iterative computations;
step 5) after the iteration is completed
Is the corresponding w 0 、w 1 、w 2 、w 3 The weight value is substituted into the formula (2) to establish a prediction model, and the values of the processing evaluation indexes under different alkaline yellow 40 solvent concentrations, tiO2 concentrations, ultraviolet illumination intensities and temperatures can be predicted through the prediction model;
step 6), solving the corresponding values of the concentration C1 of the alkaline yellow 40 solvent, the concentration C2 of TiO2, the ultraviolet irradiation intensity L and the temperature T when the Q' value is maximum, namely when the processing evaluation index is highest, according to the prediction model; the operating system sends the instructions of parameters to the ultraviolet lamp and the polishing solution adding device, so that the aim of optimizing and adjusting the chemical reaction process is fulfilled;
(2) Optimizing and adjusting the polishing path of the mechanical arm;
step (1) calling an expected track y stored in a cloud server database by the industrial personal computer operating system aiming at a processing object k Consists of k coordinate points, and the actual motion trail in the strategy network training process is y' k Establishing an error function e k The formula is as follows:
e k ＝y k -y' k (5)
training e through strategic network models k The value becomes as small as possible, i.e., the error of the polishing track is reduced;
setting a reward value R according to the error value, wherein the smaller the error is, the larger the reward value is, and in order to enable the strategy network to better understand future and potential rewards, the future attenuated rewards are used, and the action is assumed to be the next time, R k The formula for the K-th prize is as follows:
R＝R 1 +γR 2 +γ 2 R 3 +...+γ k-1 R k (gamma is the attenuation coefficient to be 0.99) (6);
step (3) a neural network model is built, the input is the current coordinate value of the mechanical arm, and the output is a probability value matrix p= [ p ] of the mechanical arm moving towards each azimuth 1 ,p 2 ,p 3 ,...,p n ](n is determined by the set number of orientations) and can be used to select actions;
step (4) establishing a loss function L 2 The cross entropy loss function is adopted, and the formula is as follows:
L 2 ＝-(r*logp)/n (7)
The probability of obtaining actions with more rewards is improved, and the probability of obtaining actions with lower rewards is reduced;
and (5) substituting the loss function, and iteratively updating the weight value in the neural network by using a gradient descent algorithm to finally obtain a strategy network capable of providing an optimal strategy, wherein the polishing path of the mechanical arm can be optimally adjusted based on the strategy network.
2. The processing platform constructed by the photocatalytic cobalt-based alloy processing method according to claim 1, wherein: comprising
The robot module comprises an industrial robot and a flexible pneumatic polishing device, wherein the flexible polishing device is arranged on a mechanical arm of the industrial robot and is used for adjusting the position of a workpiece;
the polishing workbench comprises a polishing disc for containing polishing liquid and an illumination device for illuminating the polishing liquid and a workpiece, wherein the illumination device is arranged in the polishing disc, and an illumination head of the illumination device is aligned to the polishing liquid;
the polishing liquid adding device is arranged beside the polishing workbench, and a liquid adding port of the polishing liquid adding device is positioned right above a polishing pool of the polishing workbench and is used for adding polishing liquid into the polishing workbench;
the industrial personal computer is electrically connected or in signal connection with the control ends of the robot module, the polishing workbench and the polishing liquid adding device and is used for controlling the robot module, the polishing workbench and the polishing liquid adding device to work so as to realize optimal adjustment of the chemical reaction process and optimal adjustment of the polishing path of the mechanical arm;
and the signal transmission end of the memory is in bidirectional signal connection or electric connection with the signal transmission port of the industrial personal computer through a wire or a wireless network and is used for storing processing information.
3. The processing platform of claim 2, wherein: the bottom of the polishing disc is fixedly connected with a rotary workbench driven by a rotary motor, so that the polishing disc can rotate around a vertical central shaft of the polishing disc; the polishing device comprises a polishing pool, a polishing layer, an elastic film layer, a polishing layer and a polishing layer, wherein the elastic layer, the elastic film layer and the abrasive layer are sequentially arranged in the polishing pool from bottom to top and used for supporting abrasive particles, the abrasive layer is used for processing the surface of a die, the bottom end of the elastic layer is connected with a bottom plate in the polishing pool, the elastic film layer is paved on the upper surface of the elastic layer, and the edge of the elastic film layer is connected with the inner wall of the polishing pool in a sealing manner, so that a processing area for containing polishing liquid is formed by the elastic film layer and the inner wall of the polishing pool; the abrasive particle layer is paved on the surface of the elastic film layer, and polishing liquid for lubrication and heat dissipation is filled on the abrasive particle layer; the control end of the rotating motor is electrically connected with the control end corresponding to the master controller.
4. A processing platform as claimed in claim 3, wherein: the elastic layer is a conical spiral spring array, the conical spiral spring array comprises a plurality of mutually independent conical spiral springs, the small-caliber end of each conical spiral spring is fixedly connected with the inner bottom surface of the polishing pool, the large-caliber end of each conical spiral spring is connected with a hemispherical support, and the arc-shaped curve of each hemispherical support is in contact with the lower surface of the elastic layer; the upper end of the conical spiral spring is abutted against the lower surface of the elastic film layer.
5. A processing platform as claimed in claim 3, wherein: the illumination device is an ultraviolet lamp and is fixed at the center of the rotary workbench, wherein the ultraviolet lamp always irradiates the workpiece processing area.
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 CN108857600A (en) | 2018-11-23 |
| CN108857600B true 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) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110421574B (en) * | 2019-07-22 | 2020-07-03 | 北京理工大学 | Robot creative action evaluation generation system |
| CN110842761B (en) * | 2019-11-19 | 2021-07-09 | 湖南大学 | Photocatalysis high-energy field assisted chemical mechanical composite fine grinding method |
| CN114406890B (en) * | 2021-12-20 | 2023-04-07 | 浙江工业大学 | Ultraviolet light response self-assembly cooperative dielectrophoresis polishing method and device |
| 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 |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107745324A (en) * | 2017-09-07 | 2018-03-02 | 天津大学 | A kind of optical glass surface forming method |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1197293B1 (en) * | 2000-01-31 | 2007-06-06 | Shin-Etsu Handotai Company Limited | Polishing device and method |
| CN1606713B (en) * | 2001-11-15 | 2011-07-06 | 霍尼韦尔国际公司 | Spincoating antireflection paint for photolithography |
| US6821895B2 (en) * | 2003-02-20 | 2004-11-23 | Taiwan Semiconductor Manufacturing Co., Ltd | Dynamically adjustable slurry feed arm for wafer edge profile improvement in CMP |
| CN104669075B (en) * | 2014-12-08 | 2017-08-04 | 沈阳工业大学 | Diamond cutter photocatalysis aids in grinding method and device |
| 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 |
| CN206614394U (en) * | 2017-03-16 | 2017-11-07 | 浙江工业大学 | A kind of burnishing device with profiling abrasive particle group |
| CN106881659B (en) * | 2017-03-16 | 2023-03-31 | 浙江工业大学 | Polishing device with profile modeling abrasive particle group |
-
2018
- 2018-07-25 CN CN201810828984.0A patent/CN108857600B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107745324A (en) * | 2017-09-07 | 2018-03-02 | 天津大学 | A kind of optical glass surface forming method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108857600A (en) | 2018-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108857600B (en) | Cobalt-based alloy processing method and processing platform based on photocatalysis | |
| CN109828532B (en) | GA-GBRT-based surface roughness prediction method and technological parameter optimization method | |
| Yuzgec et al. | Dynamic neural-network-based model-predictive control of an industrial baker's yeast drying process | |
| CN112642619A (en) | Intelligent spraying robot system and spraying method thereof | |
| EP2247974B1 (en) | Devices, methods, and systems for control of heliostats | |
| JP2003534460A (en) | Tuning electrode used in a reactor for electrochemically processing microelectronic workpieces | |
| CN113382807A (en) | Paint repair method according to scenes | |
| US12111638B2 (en) | Adaptive production system | |
| CN108563121A (en) | Microwave heating temperature field intelligent control method based on historical data | |
| CN111438635B (en) | Method for improving polishing surface uniformity of free-form surface | |
| CN114488801B (en) | Bridge crane model prediction control method based on data driving | |
| CN108614422A (en) | Dissolved oxygen optimal control method, apparatus and system in the cultivation of continental rise plant circulation water | |
| Kumar et al. | Developing digital twin design for enhanced productivity of an automated anodizing industry and process prediction using hybrid deep neural network | |
| Bracino et al. | Optimization of biofilter size for aquaponics using genetic algorithm | |
| CN109972093B (en) | High polymer bionic configuration photothermal conversion material and preparation method and application thereof | |
| Yang et al. | Process optimization and quality prediction of laser aided additive manufacturing SS 420 based on RSM and WOA-Bi-LSTM | |
| Yoo et al. | Bespoke Metal Nanoparticle Synthesis at Room Temperature and Discovery of Chemical Knowledge on Nanoparticle Growth via Autonomous Experimentations | |
| Liao et al. | Hybrid-biotaxonomy-like machine learning enables an anticipated surface plasmon resonance of Au/Ag nanoparticles assembled on ZnO nanorods | |
| US20240077849A1 (en) | Systems and Methods for Using Wire Printing Process Data to Predict Material Properties and Part Quality | |
| RU78785U1 (en) | AUTOMATED INSTALLATION FOR FORMING THIN FILM COATINGS OF NANOSIZED THICKNESS BY THE METHOD OF MOLECULAR LAYERING | |
| CN116976192A (en) | JS-BP model-based die forging defect accurate repair process parameter decision method | |
| Schneider et al. | LSTM-based Inverse Dynamics Learning for Franka Emika Robot | |
| Song et al. | A hybrid metaheuristic method for dispatching automated guided vehicles in container terminals | |
| US20200192328A1 (en) | Methods of forming electroformed components and related system | |
| CN118672325B (en) | Self-adaptive regulating method and system for synthesis reaction temperature of organosilicon auxiliary agent |
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 |