Method for online reconstruction of temperature field of cutting area of temperature measuring end mill

Abstract

The invention discloses a method for online reconstruction of a temperature field of a cutting area of a temperature measurement end mill. The method comprises the following steps: constructing a milling cutter three-dimensional model; establishing a milling heat conduction model; finite element analysis software is used for conducting heat conduction positive problem simulation under different heat flow densities; extracting a temperature value of each node and calculating a temperature sensitivity coefficient; acquiring a temperature data sequence at the temperature measurement position; establishing and training a heat conduction inverse problem algorithm model; acquiring real-time temperature data and predicting heat flux density through a temperature measurement module; and on-line reconstruction of the temperature field of the cutting area is realized in combination with the temperature sensitivity coefficient and the predicted heat flux density. The method can monitor the cutting temperature in real time, optimize the machining parameters, prolong the service life of the cutter and improve the machining precision.

Description

Technical Field

[0001] The present invention belongs to the technical field of mechanical cutting processing, and particularly relates to a method for online reconstruction of the temperature field in the cutting area of a temperature-measuring end mill. Background Art

[0002] In recent years, high-speed milling technology, which is characterized by high machining accuracy and low tool wear, has rapidly emerged, and milling has become an important research topic in the field of metal processing. Due to its multi-condition adaptability and high efficiency, milling can complete the machining of workpieces with different shape requirements in a short time and can ensure good machining accuracy and surface quality. It has become the main means of metal material removal and has been widely used in manufacturing fields such as aerospace, military weapons, and civilian production.

[0003] During high-speed metal milling, the material of the workpiece cutting layer undergoes plastic deformation under extremely high normal and tangential pressures. Coupled with the cutting heat generated by the friction between the tool and the chip and between the tool and the workpiece, the cutting temperature of the tool will rise rapidly. The severe temperature rise will cause local overheating of the cutting edge, reduce the strength of the cutting edge, increase the wear and even deformation of the cutting edge, and seriously reduce the surface quality and machining accuracy of the workpiece. Different from turning, milling is an intermittent machining process. Under periodic thermal shock, the milling cutter is more likely to crack due to the action of thermal stress, and even damage the workpiece and endanger the milling machine, causing a large amount of economic losses. The workpiece will also have a local temperature rise due to the cutting heat, and thermal damage may occur on the workpiece surface, such as surface oxidation and the formation of brittle phases. The local temperature rise will also cause thermal expansion of the workpiece, resulting in dimensional changes and deformation. The schematic diagrams of the three deformation zones in milling are shown as Figure 4 shown. The thermal expansion and deformation caused by the cutting heat may lead to stress concentration areas on the workpiece surface, resulting in residual stresses. These residual stresses may cause problems such as cracks and deformation during the use of the workpiece, posing a huge safety hazard.

[0004] The most direct and mature method for measuring the temperature in the cutting area of the tool is still the physical measurement method. According to the principle, it can be divided into a contact temperature measurement method based on heat conduction and a non-contact temperature measurement method based on the principle of thermal radiation. Among the contact measurement methods, the temperature measurement methods based on natural thermocouples and artificial thermocouples are economical, convenient, and technically mature, but have poor measurement accuracy, slow response, and cannot be close enough to the cutting area; the temperature measurement method based on thin-film thermocouples has a high time resolution and can measure the temperature close to the cutting area, but has a high cost and a short service life; the temperature measurement technology based on thermal radiation can achieve a high spatial resolution, but this technology cannot measure the temperature field inside the cutting area, and the installation, calibration, and debugging work are relatively complex.

[0005] In addition to measuring the tool temperature by physical measurement methods, there are also analytical models and numerical simulation models based on cutting mechanisms, as well as a solution model based on the inverse heat conduction problem that combines physical measurement methods and model solving. These methods can obtain the temperature field distribution in any cutting area. The analytical model is very flexible in temperature field modeling, but it also simplifies the heat conduction process of real metal cutting too much. These simplifications usually do not hold in the actual model, so there is a large error between the predicted result and the actual measurement result. Therefore, it will not be used alone in practical applications and is usually complementary to numerical analysis methods. Numerical modeling significantly reduces the simplifying assumptions, but a large number of input parameters need to be determined when performing numerical modeling, such as material model parameters, friction model parameters, and separation criteria, etc. These input parameters need to be determined through a large number of experiments. Therefore, the output results of numerical modeling may deviate greatly from the actual situation. At the same time, numerical modeling requires a long calculation cycle and low computational efficiency, and it is impossible to achieve real-time reconstruction of the temperature field.

[0006] The inverse heat conduction problem is defined relative to the direct heat conduction problem. For the cutting process, when the heat transfer equation, boundary conditions, initial conditions, thermal physical properties parameters, geometric shape, and heat flux density of heat conduction are known, and then the temperature field in the machining cutting area is simulated and calculated, this solution mode is called the direct heat conduction problem. For example, the numerical method is a solution method for the classical direct heat conduction problem. The inverse heat conduction problem usually inversely calculates the boundary conditions, initial conditions, thermal physical properties parameters, geometric shape, and heat flux density, etc. in the heat conduction process based on partial measurement information inside or on the surface of the heat conduction system (which is the solution to the direct heat conduction problem). In recent years, the prediction technology based on the inverse heat conduction problem has been widely studied and applied in multiple disciplinary fields. This method combines physical measurement methods and model solving methods, calculates the internal heat flux density distribution in the tool cutting area by measuring partial temperature information of the tool with thermocouples, and obtains the temperature field distribution information by combining the numerical solution of the heat conduction model. There have been many studies on the inverse heat conduction problem of tools, but the current research is difficult to achieve high spatio-temporal resolution, high-precision, and time-lag-free reconstruction of the temperature field in the tool cutting area. The main reason is that the existing solution methods for the heat conduction model cannot achieve rapid solution of the tool heat conduction model itself. The metal cutting working conditions have the characteristics of strong time-variation, so algorithms with time lag are difficult to meet the requirements of online monitoring. Summary of the Invention

[0007] The present invention proposes a method for online reconstruction of the temperature field in the cutting area of a temperature-measuring end mill to solve the problems existing in the above-mentioned prior art.

[0008] To achieve the above object, the present invention provides a method for online reconstruction of the temperature field in the cutting area of a temperature-measuring end mill, including the following steps:

[0009] Construct a three-dimensional model of the end mill according to the tool structure;

[0010] Construct a milling heat conduction model based on the three-dimensional model of the milling cutter;

[0011] Perform a forward finite element heat conduction simulation under different heat flux densities through finite element analysis software;

[0012] Obtain the temperature values of each node in the tool division in the forward finite element heat conduction simulation, and calculate the temperature sensitivity coefficient of each node according to the temperature values of each node;

[0013] Obtain the temperature data sequence at the temperature measurement position of the milling cutter in the forward problem;

[0014] Establish an inverse heat conduction algorithm model and train it through the temperature data sequence;

[0015] Obtain the temperature data at the temperature measurement position through the milling cutter temperature measurement module, and input the temperature data into the inverse heat conduction algorithm model to predict the actual heat flux density value;

[0016] According to the temperature data, select the temperature sensitivity coefficient from the temperature sensitivity coefficients of each node, and combine the predicted actual heat flux density value to perform online reconstruction of the temperature field in the cutting area.

[0017] Preferably, the three-dimensional model of the milling cutter includes:

[0018] The tool surface includes regular planes and irregular planes or curved surfaces;

[0019] The tool has a number of cutting teeth, and each cutting tooth has a side edge and a bottom edge at the same time;

[0020] The shape of the cutting area is irregular;

[0021] The shank of the milling cutter is hollow and is used for the temperature measurement module to measure temperature.

[0022] Preferably, the construction of the milling heat conduction model includes:

[0023] The heat conduction part includes the friction area between the bottom edge rake angle and the chip, the surface area between the shank and the cutting area of the tool, and the shank surface area;

[0024] The cutting area of the tool is marked as the heat flux density input area, the surface area between the shank and the cutting area of the tool is marked as the area exposed to the environment, and the shank surface area is marked as the area in contact with the fixture;

[0025] The boundary conditions of the model include the heat flux density in the cutting area, the convective heat transfer coefficient exposed to the environment, and the convective heat transfer coefficient in the area where the shank is in contact with the fixture.

[0026] Preferably, the forward finite element heat conduction simulation includes:

[0027] Apply intermittent heat flux density to the cutting area of the milling cutter through the transient heat module of ANSYS, where the heat flux density is constant during the cutting-in time and zero during the cutting-out time;

[0028] The simulated temperature field covers the temperature range of the actual working conditions, and simulates the temperature field under different combinations of milling parameters.

[0029] Preferably, the calculation of the temperature sensitivity coefficient of each node includes:

[0030] Obtain the temperature values of each node in the tool division through the APDL module, perform discretization processing according to the Duhamel superposition principle, and calculate the temperature sensitivity coefficient;

[0031] According to the intermittent characteristics of milling, regard the heat flux density during the cutting-out time within one cycle of the original intermittent cutting as the heat flux density input during the cutting-in time for equivalent processing.

[0032] Preferably, the discretization processing expression according to the Duhamel superposition principle is:

[0033]

[0034] In the formula, T i，k is the temperature of node i at time k, T i，0 is the initial temperature, q n * is the equivalent heat flux density input, and Δζ i,n is the temperature sensitivity coefficient.

[0035] Preferably, the establishment of the heat conduction inverse problem algorithm model includes:

[0036] Establish a heat conduction inverse problem algorithm model according to the Bi-LSTM neural network for inverse problem prediction; the input of the heat conduction inverse problem algorithm model is temperature data, and the output is heat flux density.

[0037] Preferably, the on-line reconstruction of the temperature field in the cutting area includes

[0038] According to the Duhamel superposition principle, combined with the selected temperature sensitivity coefficient and the predicted actual heat flux density value, reconstruct the temperature field in the cutting area.

[0039] The present invention also proposes a computer device, including a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps of the method.

[0040] The present invention also proposes a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method are implemented.

[0041] Compared with the prior art, the present invention has the following advantages and technical effects:

[0042] The present invention discloses a method for on-line reconstruction of the temperature field in the cutting area of a temperature-measuring end mill, comprising the following steps: constructing a three-dimensional model of the milling cutter; establishing a milling heat conduction model; performing a forward heat conduction problem simulation under different heat flux densities by using finite element analysis software; extracting the temperature values of each node and calculating the temperature sensitivity coefficient; obtaining the temperature data sequence at the temperature-measuring position; establishing and training a heat conduction inverse problem algorithm model; obtaining real-time temperature data through a temperature-measuring module and predicting the heat flux density; and realizing the on-line reconstruction of the temperature field in the cutting area by combining the temperature sensitivity coefficient and the predicted heat flux density. By constructing a three-dimensional model of the milling cutter and a milling heat conduction model, and combining finite element simulation with a heat conduction inverse problem algorithm, the present invention realizes the on-line reconstruction of the temperature field in the cutting area. Compared with the traditional method, the present invention can monitor the temperature change in the cutting process in real time, accurately predict the heat flux density, and dynamically reconstruct the temperature field. This technological breakthrough significantly improves the intelligent level of cutting processing, optimizes the processing parameters, prolongs the tool life, improves the processing accuracy and efficiency, and provides strong technical support for intelligent manufacturing and high-efficiency cutting. Description of the Drawings

[0043] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0044] Figure 1 is a schematic flow chart of the method according to an embodiment of the present invention;

[0045] Figure 2 is a schematic diagram of the temperature measurement method and the temperature measurement point position according to an embodiment of the present invention;

[0046] Figure 3 is an overall structure diagram of the temperature-measuring end mill according to an embodiment of the present invention;

[0047] Figure 4 is a schematic diagram of three deformation zones in the background art of the present invention;

[0048] Figure 5 is a schematic flow chart of sequentially solving the model temperature field by using FEM according to an embodiment of the present invention;

[0049] Figure 6 is a schematic diagram of the boundary conditions according to an embodiment of the present invention;

[0050] Figure 7 is a schematic diagram of the simulated heat flux density in the milling area according to an embodiment of the present invention;

[0051] Figure 8The temperature sensitivity coefficient diagram of three points in the cutting area when the heat conduction model of the embodiment of the present invention is linear (parameters at 20°C);

[0052] Figure 9 The schematic diagram of the simulated true values of the predicted values of different models based on noise data (σ = 20K) in the embodiment of the present invention, where (a) is Bi-LSTM, (b) is Stacked LSTM, (c) is GRU, and (d) is RNN neural network;

[0053] Figure 10 The schematic diagram of selecting the optimal temperature sensitivity coefficient in the embodiment of the present invention;

[0054] Figure 11 The flow chart of temperature field reconstruction in the embodiment of the present invention;

[0055] Figure 12 The assembly drawing of the milling temperature measurement platform in the embodiment of the present invention;

[0056] Figure 13 The results of the reconstructed temperature field and temperature gradient distribution at 6 moments in the experiment of the embodiment of the present invention;

[0057] Figure 14 The comparison diagram of the observed temperature at the infrared temperature measurement position and the reconstructed temperature at the corresponding position in the embodiment of the present invention;

[0058] Figure 15 The comparison diagram of the measured temperature at the thermosensitive ceramic temperature measurement point position and the reconstructed temperature at the corresponding position in the embodiment of the present invention. Detailed implementation manners

[0059] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0060] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0061] Embodiment 1

[0062] As Figure 1 shown, in this embodiment, a method for online reconstruction of the temperature field in the cutting area of a temperature measurement end milling cutter is provided, including the following steps:

[0063] Construct a three-dimensional model of the milling cutter according to the tool structure;

[0064] Construct a milling heat conduction model according to the three-dimensional model of the milling cutter;

[0065] Perform the forward problem simulation of finite element heat conduction under different heat flux densities using finite element analysis software. The schematic diagram of the heat flux density in the milling area is shown as Figure 7 shown;

[0066] Obtain the temperature values of each node in the tool division during the forward problem simulation of finite element heat conduction, and calculate the temperature sensitivity coefficient of each node based on the temperature values of each node. When the heat conduction model is linear (parameters at 20 °C), the temperature sensitivity coefficients of three points in the cutting area are shown as Figure 8 shown;

[0067] Obtain the temperature data sequence at the temperature measurement position of the milling cutter in the forward problem;

[0068] Establish an algorithm model for the inverse heat conduction problem and train it through the temperature data sequence;

[0069] Obtain the temperature data at the temperature measurement position through the milling cutter temperature measurement module, and input the temperature data into the algorithm model of the inverse heat conduction problem to predict the actual heat flux density value;

[0070] According to the temperature data, select the temperature sensitivity coefficient from the temperature sensitivity coefficients of each node. The schematic diagram of selecting the optimal temperature sensitivity coefficient is shown as Figure 10 shown. Combine the predicted actual heat flux density value to perform online reconstruction of the temperature field in the cutting area. According to Figure 11 it can be known that the temperature field reconstruction process is as follows.

[0071] Specifically, it includes the following steps:

[0072] Step 1: Establish an accurate three-dimensional model of the milling cutter.

[0073] In Step 1, establishing an accurate three-dimensional model of the milling cutter is mainly manifested in several aspects: 1) The tool surface not only includes regular planes but also many irregular planes or curved surfaces; 2) The tool has multiple cutting teeth, and each tooth has both side edges and bottom edges; 3) The shape of the cutting area is irregular, neither a standard rectangle nor a circle; 4) The shank of the milling cutter is hollow to facilitate temperature measurement by the temperature measurement module.

[0074] Step 2: Establish a complete milling heat conduction model.

[0075] In Step 2, establish a complete milling heat conduction model. The heat conduction part includes the friction area between the bottom edge rake angle and the chip, the surface area between the shank and the tool cutting area, and the shank surface area. The tool cutting area is marked as S 1 , and due to the friction between the bottom edge rake angle and the chip in this area, an unknown time-varying heat flux density q(t) is generated. The surface between the shank and the tool cutting area is marked as S 2 , and this part of the area is exposed to air or cutting fluid. The shank surface area is marked as S3 This part of the area is connected to the fixture. Each part of the cutting tool is cooled by convection with the environment during the milling process.

[0076] The boundary conditions are as Figure 6 shown, and the boundary conditions are as follows:

[0077]

[0078] Equation (10) is the boundary condition of the cutting area S of the cutting tool 1 , where represents the partial derivative of the temperature gradient along the η direction, where η is one of the coordinate directions in x, y, z or a normal direction. The heat flux density q(t) is a function that varies with time.

[0079] T ∞ is the temperature of the environment. In this study, the environmental temperature is set to 22 °C. Equation (11) is the boundary condition of the area S exposed to the environment 2 , and h 1 is the convective heat transfer coefficient between the cutting tool and the environment. Equation (12) is the boundary condition of the contact area S between the tool shank and the fixture 3 , and h 2 is the convective heat transfer coefficient between the cutting tool and the environment.

[0080] The three-dimensional transient heat conduction control equation of the model is shown in the following equation (13):

[0081]

[0082] Here, T(x, y, z, t) is the temperature at any position at each time; ρ is the mass density; c(T) and k(T) are the specific heat coefficient and thermal conductivity related to temperature. Due to the nonlinearity of both, this heat conduction model becomes a nonlinear problem.

[0083] The initial condition is set as:

[0084] T(x, y, z, 0) = T 0 (14)

[0085] Step 3: Use ANSYS to simulate the forward problem of finite element heat conduction under different heat flux densities.

[0086] In Step 3, after establishing the milling heat conduction model, use the transient heat module of ANSYS to apply the intermittent heat flux density q(t) to the cutting area of the milling cutter.

[0087] Within one cycle, the heat flux density function q(t) is not a constant value and will change with the cutting edge entering and exiting. Under the premise of determining the milling parameters and workpiece material, within a long milling time, the cutting-in time t inInside, the heat flux density input value can be considered as a constant h (h = 1, 2, 3…), with the unit of W·mm -2 , during this period, the temperature of the cutting edge will rise sharply. During the cutting-out time, the cutting edge is not in direct contact with the workpiece, and the heat flux density input value is 0.

[0088]

[0089] Within one cycle, the cutting edge is in two states: cutting-in and cutting-out. t in is the cutting-in time, and t out is the cutting-out time. The cutting-in and cutting-out times are as follows:

[0090]

[0091]

[0092] t out = T z - t in (18)

[0093] Among them, l h is the contact arc length between the milling cutter and the workpiece, v is the cutting speed, and D is the diameter of the milling cutter.

[0094] The value of the heat flux density h during the cutting-in time t in is h = 1, 2, 3···N. Each value corresponds to a set of simulations, and a total of N sets of simulations are carried out. These simulated temperature fields need to cover the possible temperature range of the actual working conditions to simulate the temperature fields under different milling parameter combinations.

[0095] Step 4: Use the APDL module to obtain the temperature values of each node in the tool division and calculate the temperature sensitivity coefficient for the fast temperature reconstruction algorithm.

[0096] In Step 4, in the heat conduction system, if the heat flux density function q(t) input on the boundary is known, then the temperature change at any coordinate point (x i , y i , z i ) in the heat conduction model can be accurately calculated according to the Duhamel superposition principle. The calculation formula is as follows:

[0097]

[0098] In Equation (19), the subscript i represents the spatial index identifier of the position (x i , y i , z i ) where the temperature is to be calculated, and ζ(x i , y i , z i, t) represents the unit step response of the temperature at point i in the heat conduction model with respect to the heat flux density function q(t). In the linear heat conduction model, it can be understood as the temperature response when the heat flux density function q(t) = 1, t > 0.

[0099] In practical applications, for the convenience of calculation, it is necessary to discretize the above Duhamel superposition principle. In the discrete system, the Duhamel superposition principle can be expressed as:

[0100]

[0101]

[0102] In Equation (20), the subscript n represents the time index, k represents the time index at the current moment, and the entire continuous time domain is uniformly discretized into continuous time points with Δt as the unit: Δt, 2Δt,..., kΔt, t n represents the following time period: (n - 1)·Δt ≤ t n <n·Δt, t n It can also be understood as the nth time step, n ∈ [1, k]. T i,0 represents the initial temperature at point i. Δζ i,n represents the temperature sensitivity coefficient at point i at t n time period, ζ i,n is the temperature step response, representing the temperature change at point i caused by the step heat flux density during the t n time period.

[0103] The conventional Duhamel superposition principle is applicable to the reconstruction of the tool temperature field in the stable and continuous machining mode of heat flux density, such as turning. For the machining form with intermittent characteristics like milling, certain improvements and corrections need to be made to the original formula.

[0104] In the original Duhamel superposition principle formula, the calculation of the temperature sensitivity coefficient Δζ involves the assignment of the heat flux density. If the heat flux density is zero, the temperature sensitivity coefficient loses its meaning. Therefore, when actually using the Duhamel superposition formula for the reconstruction of the milling cutter temperature field, it is necessary to perform equivalent processing on the input heat flux density to make the temperature sensitivity coefficient meaningful.

[0105] In the equivalent processing, the heat flux density during the cutting-out time t z within one cycle T out of the original interrupted cutting is still regarded as the heat flux density input at the cutting-in time t in . At this time, it can be understood that within the entire cutting cycle, only the influence of the tool heat flux density input during the cutting-in time t in on the tool temperature throughout the cycle is considered. Therefore, the equivalent heat flux density input is the cutting-in time tin Heat flux density input at that time.

[0106]

[0107] In the above formula (22), q(t) is the actual intermittent heat flux density, and q(t) * is the equivalent heat flux density. The equivalence of the heat flux density does not change the calculation result, and the intermittency and periodicity of the actual heat flux density will be reflected in the temperature sensitivity coefficient Δζ.

[0108] The discretized expression of Duhamel superposition principle for milling can be written as the following mathematical expression:

[0109]

[0110] In the above formula (23), q n * is the equivalent heat flux density input, and Δζ i,n is the temperature sensitivity coefficient.

[0111] Step 5: Separately obtain the temperature data series at the temperature measurement position of the milling cutter in the simulation model under different heat flux densities in the forward problem.

[0112] Step 6: Establish and train the heat conduction inverse problem algorithm model, input the temperature data and output the heat flux density.

[0113] In the heat conduction inverse problem, the Bi-LSTM neural network is used for inverse problem prediction. The training data comes from Step 5. By applying different intermittent heat flux density inputs q(t) = h (h = 1, 2, 3…N) in the cutting area, N simulation models with different thermal steady-state temperatures are created to simulate the milling temperature under different working conditions. The derived time interval Δτ is determined according to the actual milling cycle. Using the APDL parametric design module in ANSYS, the temperature data of all nodes in the milling cutter model at all moments with an interval of Δτ are obtained through the programming language, and the simulated temperature data T c (h n )(n = 1, 2,..., N) are used as the input data for training the model, and the corresponding heat flux density q(t) is used as the output value.

[0114] In a linear heat conduction system, the temperature sensitivity coefficient Δζ can also be calculated from the temperature response ζ. In a linear heat conduction system, when the intermittent heat flux density q(t) = 1 is input, the temperature step response ζ i,n = [ζ i,1 , ζ i,2 , … ζ i,k of a certain point i at each time step is obtained, and the temperature sensitivity coefficient Δζ of point i at each time step can be calculated through formula (21).

[0115] Under linear conditions, let the input intermittent heat flux density q(t) = h (h = 1, 2, 3, …) during the milling cutting-in time, and use Ansys transient thermal simulation to obtain the temperature corresponding to the heat flux density, and the equivalent temperature sensitivity coefficient Δζ * can be calculated by the following formula:

[0116]

[0117] q n * is the equivalent heat flux density input, q n * = h. Calculate the value of the temperature sensitivity coefficient Δζ using the above formula. Regardless of the value of the heat flux density q n = h (h = 1, 2, 3, …) during the milling cutting-in time, the result of Δζ is the same.

[0118] For the non-linear heat conduction model, the temperature sensitivity coefficient needs to be calculated at each time step. If the FEM is directly used for online calculation at each step, the solution efficiency will be very low, and it is difficult to achieve the rapid solution of the model. The flow chart of the sequential solution of the model temperature field using FEM is as Figure 5 shown.

[0119] Therefore, an offline method is proposed to obtain the offline temperature sensitivity coefficient database. This method has two key steps:

[0120] Key step 1: Simulate N heat conduction models with different thermal steady-state temperatures through ANSYS finite element simulation. These simulated temperature fields need to cover the possible temperature range of the actual working conditions to simulate the temperature fields under different milling parameter combinations. This can be achieved by loading intermittent heat flux density on the cutting area of the simulation model, that is, applying the heat flux density q(t) = h n = n (n = 1, 2, …, N) during the cutting-in time until the temperature field reaches a steady state.

[0121] Key step 2: Apply a temperature step input to the model, that is, let q(t) = h n +1, and obtain the corresponding temperature step response Δζ(h n ). Obtain the temperature sensitivity coefficient Δζ(h n ) through Equation (21). Record the temperature sensitivity coefficients of each node under different heat flux density inputs for later reconstruction and call.

[0122] Step 7: Obtain the temperature data at the temperature measurement position through the milling cutter temperature measurement module, and substitute it into Step 6 to predict the actual heat flux density value.

[0123] Step 8: Select the appropriate temperature sensitivity coefficient according to the measured temperature, and combine with Step 7 to realize the reconstruction of the temperature field in the cutting area using the Duhamel principle.

[0124] Verification is as follows:

[0125] The contact time between the milling tool and the workpiece is short and there are frequent alternate cut-ins and cut-outs. This periodic thermal shock causes rapid changes in the temperature field, making it difficult to capture the instantaneous temperature peak. Moreover, the high-speed rotating milling cutter makes it extremely difficult to install static sensors, and the sensors may not work stably or provide reliable readings at high speeds. To solve the above problems, a milling temperature measurement platform was built. The schematic diagram of the temperature measurement method and the position of the temperature measurement points is as Figure 2 shown, and the overall structure diagram of the temperature measurement milling cutter is as Figure 3 shown, and the assembly diagram of the milling temperature measurement platform is as Figure 12 shown. Verify the effectiveness of the proposed online reconstruction method for the temperature field of the milling cutter based on the inverse heat conduction problem (Bi-LSTM) and the Duhamel superposition principle and the accuracy of the reconstructed temperature field. The schematic diagram of the simulated true values of the predicted values of different models based on noisy data (σ = 20K) is as Figure 9 shown.

[0126] The workpiece material is shown in Table 1:

[0127] Table 1

[0128]

[0129] The final architecture and hyperparameters of the Bi-LSTM model are shown in Table 2:

[0130] Table 2

[0131]

[0132] The relative errors of the observed temperature at the infrared temperature measurement position and the reconstructed temperature at the corresponding position at six moments are shown in Table 3. The comparison between the observed temperature at the infrared temperature measurement position and the reconstructed temperature at the corresponding position is as Figure 14 shown.

[0133] Table 3

[0134]

[0135] The relative errors of the measured temperature at the thermosensitive ceramic temperature measurement point position and the reconstructed temperature at the corresponding position at six moments are shown in Table 4. The results of the reconstructed temperature field and the temperature gradient distribution at six moments in the experiment are as Figure 13 shown, and the comparison between the measured temperature at the thermosensitive ceramic temperature measurement point position and the reconstructed temperature at the corresponding position is as Figure 15 shown.

[0136] Table 4

[0137]

[0138] Through two verification methods, it can be found that the trend of the actually measured or observed temperature data basically coincides with the temperature data calculated by the reconstruction algorithm. The results show that the proposed online reconstruction method of the milling cutter temperature field based on the inverse heat conduction problem (Bi-LSTM) and the Duhamel superposition principle is effective and the accuracy of the reconstructed temperature field is relatively high.

[0139] This embodiment also proposes a computer device, including a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps of the method.

[0140] This embodiment also proposes a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method are implemented.

[0141] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for online reconstruction of the temperature field of a temperature measuring end mill cutting zone, characterized in that: The following steps are involved: Construct a three-dimensional model of the milling cutter according to the tool structure; Constructing a milling heat conduction model according to the three-dimensional model of the milling cutter; Finite element analysis software is used to simulate the finite element heat conduction problem under different heat flux densities; Obtain the temperature value of each node in the tool partition in the finite element heat conduction forward problem simulation, and calculate the temperature sensitivity coefficient of each node according to the temperature value of each node; Obtain the temperature data sequence at the temperature measurement position of the milling cutter in the positive problem; Establishing a heat conduction inverse problem algorithm model and training it through the temperature data sequence; Acquire temperature data of the temperature measurement position through the milling cutter temperature measurement module, and input the temperature data into the heat conduction inverse problem algorithm model to predict the actual heat flux density value; According to the temperature data, a temperature sensitivity coefficient is selected from the temperature sensitivity coefficients of each node, and the temperature field of the cutting zone is reconstructed online in combination with the predicted actual heat flux density value.

2. The method according to claim 1, characterized in that The milling cutter three-dimensional model comprises: The tool surface includes regular planes and irregular planes or curved surfaces; The tool has a plurality of cutting teeth, each of which has a side edge and a bottom edge; The shape of the cutting area is irregular; The shank of the milling cutter is hollow and is used for temperature measurement by the temperature measuring module.

3. The method according to claim 1, characterized in that The construction of the milling heat conduction model comprises: The heat conduction part includes the friction area between the bottom edge rake angle and the chip, the surface area between the tool holder and the tool cutting area, and the surface area of ​​the tool holder; The tool cutting area is marked as the heat flux input area, the surface area between the tool holder and the tool cutting area is marked as the area exposed to the environment, and the surface area of ​​the tool holder is marked as the area in contact with the fixture; The boundary conditions of the model include the heat flux density in the cutting zone, the convective heat transfer coefficient exposed to the environment, and the convective heat transfer coefficient in the contact area between the tool holder and the fixture.

4. The method according to claim 1, characterized in that: The finite element heat conduction forward problem simulation includes: The transient thermal module of ANSYS is used to apply intermittent heat flux density to the milling cutter cutting area, where the heat flux density is constant during the cutting-in time and zero during the cutting-out time; The simulated temperature field covers the temperature range of actual working conditions and simulates the temperature field under different milling parameter combinations.

5. The method according to claim 1, characterized in that The calculation of the temperature sensitivity coefficient of each node includes: The temperature value of each node in the tool division is obtained through the APDL module, and the temperature sensitivity coefficient is calculated by discretization according to the Duhamel superposition principle; According to the intermittent characteristics of milling, the heat flux density at the cut-out time in one cycle of the original intermittent cutting is regarded as the heat flux density input at the cut-in time and equivalent treatment is performed.

6. The method according to claim 5, characterized in that The discretization expression according to Duhamel superposition principle is: Where, T i，k is the temperature of node i at time k, T i，0 is the initial temperature, q n * is the equivalent heat flux input, Δζ i,n is the temperature sensitivity coefficient.

7. The method according to claim 1, characterized in that The establishment of the heat conduction inverse problem algorithm model comprises: A heat conduction inverse problem algorithm model is established based on the Bi-LSTM neural network to perform inverse problem prediction; the input of the heat conduction inverse problem algorithm model is temperature data, and the output is heat flux density.

8. The method according to claim 1, characterized in that The online reconstruction of the temperature field in the cutting zone includes: According to the Duhamel superposition principle, the temperature field in the cutting zone is reconstructed by combining the selected temperature sensitivity coefficient and the predicted actual heat flux density value.

9. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images