Multi-dimensional performance prediction method for color cutter coating

Abstract

The invention provides a multi-dimensional performance prediction method for a colored cutter coating, which belongs to the technical field of cutter coatings and comprises the following steps: firstly, preparing a colored cutter coating sample and collecting data; carrying out normalization processing on the collected data; setting a multi-layer sensing structure, and constructing an MLP neural network model; defining a loss function, and training an MLP neural network model by using an Adam optimizer; and evaluating the performance of the MLP neural network model for predicting a new coating. Compared with the traditional coating design, the method has the advantages that the performance of the color cutter coating is predicted through the MLP neural network, the time is shortened by more than 90% compared with the traditional coating technology, the research and development cycle of new materials is remarkably accelerated, and the research cost is reduced. Synchronous prediction of the color and the mechanical performance is achieved through the double-branch neural network, and the color prediction error and the mechanical performance prediction error are small and are remarkably improved compared with a traditional method.

Description

Technical Field

[0001] This invention belongs to the field of tool coatings, and specifically relates to a method for predicting the multidimensional performance of colored tool coatings. Background Technology

[0002] Colored tool coatings have wide applications in the manufacturing of cutting tools. Their core performance indicators mainly include color parameters (such as CIELAB color coordinates) and mechanical property parameters (such as hardness, wear resistance, and adhesion). These parameters play a decisive role in the performance and effectiveness of colored tool coatings in different scenarios.

[0003] However, current traditional colored tool coating design technologies suffer from several significant drawbacks that urgently need to be addressed: First, the development cycle is lengthy. Researchers must repeatedly optimize the process through "composition design → sample preparation → performance testing → parameter adjustment," with each iteration potentially taking days to weeks. If complex performance requirements are involved (such as balancing color stability, wear resistance, and corrosion resistance), dozens or even hundreds of adjustments may be required, resulting in an overall development cycle lasting months to years. Second, development costs are high. Coating materials often use precious metals (titanium, chromium, etc.), and a single formulation may waste several grams or tens of grams of material. High iteration costs and high energy consumption of coating equipment, coupled with frequent use, accelerate equipment aging and increase maintenance costs. Third, reusability is weak. Under traditional development models, historical research data cannot be effectively utilized, making it difficult to quickly reuse the same coating materials, leading to low development efficiency.

[0004] With the development of artificial intelligence, neural networks have shown unique advantages in predicting material properties, but existing research has not solved the aforementioned key problems. Therefore, there is an urgent need to develop an intelligent prediction method that can simultaneously predict the color space parameters and mechanical property parameters of colored tool coatings. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a multidimensional performance prediction method for colored cutting tool coatings. By constructing a multi-output MLP (multilayer perceptron) model, the method simultaneously predicts color and mechanical properties, thus solving the problems of low prediction efficiency and single prediction target in traditional methods.

[0006] This invention is achieved through the following technical solution: a method for predicting the multidimensional performance of colored cutting tool coatings: The method specifically includes the following steps: Step 1: Prepare colored tool coating samples and collect data; Step 2, data preprocessing, normalize the data collected in Step 1; Step 3: Set up a multilayer perceptron structure and construct an MLP neural network model; Step 4: Define the loss function and train the MLP neural network model using the Adam optimizer; Step 5: Evaluate the performance of the MLP neural network model for predicting the new coating.

[0007] Further, in step 1, A series of colored tool coating samples were prepared using physical vapor deposition (PVD) and chemical vapor deposition (CVD), and the coating composition ratio parameters, deposition process parameters, and coating structure parameters were recorded. Mechanical property parameters and color space parameters (CIElab) of the samples were tested and experimental data were collected. Furthermore, in step 2, The collected coating composition ratio parameters, deposition process parameters, coating structure parameters, mechanical property parameters, and color space parameters were normalized. The formula is: ; in, The value after normalization. and These are the maximum and minimum values ​​of the parameter, respectively. This is the original data.

[0008] Furthermore, in step 3, The MLP neural network model includes an input layer, an output layer, and hidden layers. The input layer is responsible for receiving the coating group allocation ratio parameters, deposition process parameters, and coating structure parameters, and the number of nodes in the input layer is determined according to the number of input parameters. The output layer adopts a dual-branch structure to predict mechanical performance parameters and color space parameters respectively: The coating color is represented by Lab values; the mechanical performance parameters are set with 3 nodes, corresponding to hardness, wear resistance coefficient and bonding strength respectively, and the activation function is a linear function; Establish 3-5 hidden layers, with the number of neurons in each layer decreasing sequentially, forming a funnel-shaped structure to extract higher-order features.

[0009] Furthermore, in the MLP neural network model, the connections from the input layer to the hidden layer and from the hidden layer group to the output layer branch are all fully connected; Each layer of the calculation process , where W is the weight matrix and b is the bias vector, which is passed to the next layer after passing through the activation function; A Dropout layer is inserted between the hidden layers, and L2 regularization is applied to constrain the norm of the weight matrix to prevent overfitting.

[0010] Furthermore, in step 4, The loss function is defined using weighted multi-task mean squared error (MSE). The formula is: ,in The weights of each performance metric, and These are the predicted value and the actual value, respectively. The Adam optimizer is used with an initial learning rate of 0.001. The learning rate scheduling adopts a cosine annealing strategy and an early stopping strategy.

[0011] Furthermore, in step 5, The root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination were used. Evaluate the model performance; use the optimized MLP model to predict the performance of the new coating.

[0012] A multidimensional performance prediction system for colored cutting tool coatings; The system includes a preparation and acquisition module, a preprocessing module, a construction and training module, and a prediction module; The preparation and acquisition module is used to prepare colored tool coating samples and acquire data; The preprocessing module normalizes the collected data. The construction and training module sets up a multi-layer perceptron structure, constructs an MLP neural network model, defines a loss function, and trains it using the Adam optimizer. The prediction module evaluates the performance of the MLP neural network model for predicting new coatings.

[0013] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0014] A computer-readable storage medium for storing computer instructions that, when executed by a processor, implement the steps of the above-described method.

[0015] Beneficial effects of the invention In traditional methods, color and mechanical performance testing are independent and require separate optimization. This invention achieves end-to-end prediction of "input parameters - multiple performance outputs," breaking through the conventional approach of "single performance optimization." The hidden layer adopts a funnel-shaped decreasing structure (256→128→64) and is equipped with the ReLU activation function to solve the gradient vanishing problem. The output layer has two branches, using Sigmoid (color) and a linear function (mechanical performance) respectively, to match the nonlinear characteristics of different indicators. This structural design requires combining coating performance characteristics with neural network theory.

[0016] Traditional coating design relies on a trial-and-error approach, requiring repeated experiments to adjust composition and processes, resulting in long development cycles (months to years) and high costs (waste of precious metals, high equipment energy consumption). Existing neural network applications do not achieve "simultaneous prediction" of color and mechanical properties, often focusing on a single property (such as predicting only mechanical properties or only color), and lack a dual-branch output structure and weighted loss function. This invention, however, integrates coating composition ratio parameters, deposition process parameters, and coating structure parameters in the input layer, while the output layer simultaneously predicts color space parameters (Lab values) and mechanical properties (hardness, wear resistance coefficient, and bonding strength) through independent branches. Existing technologies lack similar multi-task collaborative prediction schemes. Furthermore, by combining material science parameters (such as deposition temperature and coating thickness) with AI algorithms, and optimizing the training process through regularization (Dropout+L2) and cosine annealing, prediction accuracy is improved.

[0017] The technical solution details the data acquisition (PVD / CVD sample preparation, mechanical property and color testing), preprocessing (normalization formula), model building (input / hidden / output layer parameters), training (loss function, optimizer), and evaluation process. The steps are clearly defined, and the MLP neural network and Adam optimizer involved are mature technologies in the AI ​​field. The coating composition and process parameters can be obtained through conventional experiments. The technical solution of this invention can be implemented in industry. By replacing some experiments with predictions, it reduces the number of "trial and error" attempts, reduces waste of precious metal materials (such as titanium and chromium) and equipment energy consumption, meets the rapid R&D needs in intelligent manufacturing, and can produce positive technical and application effects. The technical solution of this invention can be implemented in industry, significantly shortening the R&D cycle and reducing costs.

[0018] In summary, compared with the prior art, this invention proposes a method for predicting the color and mechanical properties of colored cutting tool coatings based on multilayer perceptron (MLP) neural networks; (1) Cost reduction: Compared with traditional coating design, this method uses MLP neural network to predict the performance of colored tool coating, which shortens the time by more than 90% compared with traditional coating technology, significantly accelerates the research and development cycle of new materials, and reduces research costs.

[0019] (2) High accuracy of multi-performance collaborative prediction: The color and mechanical performance are predicted synchronously through a dual-branch neural network. The color prediction error and mechanical performance prediction error are low, which is a significant improvement over traditional methods. Attached Figure Description

[0020] Figure 1 This is the overall flowchart of the present invention.

[0021] Figure 2 This is a schematic diagram of each layer of the neural network of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Unless otherwise specified, the experimental methods used in the following examples are conventional methods. Unless otherwise specified, the materials, reagents, methods, and instruments used are all conventional materials, reagents, methods, and instruments in the art, and can be obtained commercially by those skilled in the art.

[0024] Combination Figure 1 and Figure 2 A method for predicting the multidimensional performance of colored cutting tool coatings: The method specifically includes the following steps: Step 1: Prepare colored tool coating samples and collect data; A series of colored tool coating samples were prepared using physical vapor deposition (PVD) and chemical vapor deposition (CVD), and data such as coating composition ratio, deposition process parameters and structural parameters were recorded. Mechanical property tests (such as microhardness test, scratch test, wear and corrosion resistance test, etc.) and color space parameter measurements (Lab color parameters are obtained using a spectrophotometer) were performed on the samples to collect experimental data.

[0025] Step 2, data preprocessing, normalize the data collected in Step 1; The collected input parameters (coating composition ratio parameters, deposition process parameters, and coating structure parameters) and output parameters (mechanical property parameters and color parameters) are normalized. The formula is: ; in, The value after normalization. and These are the maximum and minimum values ​​of the parameter, respectively. This is the original data.

[0026] Step 3: Set up a multilayer perceptron structure and construct an MLP neural network model; The MLP neural network model includes an input layer, an output layer, and hidden layers. The input layer is responsible for receiving data such as coating composition ratio parameters (material types and proportions), process parameters (deposition temperature, sputtering power, working gas pressure, deposition time), and structural parameters (coating thickness). The number of nodes in the input layer is determined according to the number of input parameters (e.g., coating composition ratio parameter 3 + process parameter 4 + structural parameter 1 = 8 nodes). The output layer adopts a dual-branch structure to predict color space parameters and mechanical performance parameters respectively: Coating color: Represented by Lab values, requiring the setting of 3 nodes, each corresponding to a Lab color coordinate. The activation function is the Simgmoid function. ); Mechanical properties: Three nodes are set, corresponding to hardness (HV), wear resistance coefficient (HV), etc. The activation function uses a linear function to determine the binding strength (N) and the binding strength (N). Establish 3-5 hidden layers, with the number of neurons in each layer decreasing sequentially (e.g., 256→128→64), forming a funnel-shaped structure to extract higher-order features; Each layer is followed by a ReLU activation function. ), through formula (α=0.01) achieves nonlinear transformation, alleviating the gradient vanishing problem; The number of hidden layer nodes is determined by the formula. Determine, where p is the number of input layer nodes, q is the number of output layer nodes, and c is a random number between 1 and 10.

[0027] Data flow: The connection from the input layer to the hidden layer is fully connected (with weight parameters), and the connection from the hidden layer group to the output layer branch is also fully connected.

[0028] Each layer of the calculation process , where W is the weight matrix and b is the bias vector, which is passed to the next layer after passing through the activation function; Regularization: Insert a Dropout layer (with a deactivation rate of 0.2-0.3) between hidden layers and apply L2 regularization (λ=0.001) to constrain the norm of the weight matrix to prevent overfitting.

[0029] Step 4: Define the loss function and train the MLP neural network model using the Adam optimizer; The loss function is defined using weighted multi-task mean squared error (MSE). The formula is: ,in Assign weights to each performance indicator (set according to actual requirements, such as 0.6 for mechanical performance parameters and 0.4 for color space parameters). and These are the predicted value and the actual value, respectively. The Adam optimizer is used with an initial learning rate of 0.001, and cosine annealing is employed for learning rate scheduling. Batch size 32-64, training rounds 200-300, early stopping strategy (stop if the validation set loss does not decrease after 5 rounds).

[0030] Step 5: Evaluate the performance of the MLP neural network model for predicting the new coating.

[0031] The root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination were used. The model performance was evaluated; the optimized MLP model was used to predict the performance of new coatings, providing a reference for coating composition design and process parameter optimization.

[0032] A multidimensional performance prediction system for colored cutting tool coatings, The system includes a preparation and acquisition module, a preprocessing module, a construction and training module, and a prediction module; The preparation and acquisition module is used to prepare colored tool coating samples and acquire data; The preprocessing module normalizes the collected data. The construction and training module sets up a multi-layer perceptron structure, constructs an MLP neural network model, defines a loss function, and trains it using the Adam optimizer. The prediction module evaluates the performance of the MLP neural network model for predicting new coatings.

[0033] An electronic device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above method.

[0034] A computer-readable storage medium for storing computer instructions that, when executed by a processor, implement the steps of the above-described method.

[0035] The memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory of the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0036] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means such as coaxial cable, optical fiber, digital subscriber line, DSL, or wireless means such as infrared, wireless, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium such as a floppy disk, hard disk, magnetic tape; an optical medium such as a high-density digital video disc, DVD; or a semiconductor medium such as a solid-state disk, SSD, etc.

[0037] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0038] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as execution by a hardware decoding processor, or as execution by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0039] The above provides a detailed description of the multidimensional performance prediction method for colored cutting tool coatings proposed in this invention, and elucidates the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for predicting the multidimensional performance of colored cutting tool coatings, characterized in that: The method specifically includes the following steps: Step 1: Prepare colored tool coating samples and collect data; Step 2, data preprocessing, normalize the data collected in Step 1; Step 3: Set up a multilayer perceptron structure and construct an MLP neural network model; Step 4: Define the loss function and train the MLP neural network model using the Adam optimizer; Step 5: Evaluate the performance of the MLP neural network model for predicting the new coating.

2. The prediction method according to claim 1, characterized in that: In step 1, A series of colored tool coating samples were prepared using physical vapor deposition (PVD) and chemical vapor deposition (CVD), and the coating composition ratio parameters, deposition process parameters, and coating structure parameters were recorded. Mechanical property parameters and color space parameters of the samples were tested and experimental data were collected.

3. The prediction method according to claim 2, characterized in that: In step 2, The collected coating composition ratio parameters, deposition process parameters, coating structure parameters, mechanical property parameters, and color space parameters were normalized. The formula is: ; in, The value after normalization. and These are the maximum and minimum values ​​of the parameter, respectively. This is the original data.

4. The prediction method according to claim 3, characterized in that: In step 3, The MLP neural network model includes an input layer, an output layer, and hidden layers. The input layer is responsible for receiving coating group allocation parameters, deposition process parameters and coating structure parameters, and determining the number of nodes in the input layer according to the number of input parameters. The output layer adopts a dual-branch structure to predict mechanical performance parameters and color space parameters respectively: The coating color is represented by Lab values; the mechanical performance parameters are set with 3 nodes, corresponding to hardness, wear resistance coefficient and bonding strength respectively, and the activation function is a linear function; Establish 3-5 hidden layers, with the number of neurons in each layer decreasing sequentially, forming a funnel-shaped structure to extract higher-order features.

5. The prediction method according to claim 4, characterized in that: In the MLP neural network model, the connections from the input layer to the hidden layer and from the hidden layer group to the output layer branch are all fully connected; Each layer of the calculation process , where W is the weight matrix and b is the bias vector, which is passed to the next layer after passing through the activation function; A Dropout layer is inserted between the hidden layers, and L2 regularization is applied to constrain the norm of the weight matrix to prevent overfitting.

6. The prediction method according to claim 5, characterized in that: In step 4, The loss function is defined using weighted multi-task mean squared error (MSE). The formula is: ,in The weights of each performance metric, and These are the predicted value and the actual value, respectively. The Adam optimizer is used with an initial learning rate of 0.

001. The learning rate scheduling adopts a cosine annealing strategy and an early stopping strategy.

7. The prediction method according to claim 6, characterized in that: In step 5, The root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination were used. Evaluate the model performance; use the optimized MLP model to predict the performance of the new coating.

8. A multidimensional performance prediction system for colored cutting tool coatings, characterized in that: The system is used to perform the steps of the method according to any one of claims 1 to 7; The system includes a preparation and acquisition module, a preprocessing module, a construction and training module, and a prediction module; The preparation and acquisition module is used to prepare colored tool coating samples and acquire data; The preprocessing module normalizes the collected data. The construction and training module sets up a multi-layer perceptron structure, constructs an MLP neural network model, defines a loss function, and trains it using the Adam optimizer. The prediction module evaluates the performance of the MLP neural network model for predicting new coatings.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images