Real-time temperature estimation and fault monitoring method for the unmeasurable temperature area of ​​electric spindle

Abstract

A method for real-time temperature estimation and fault monitoring in unmeasurable areas of an electric spindle is proposed. This method constructs a heat capacitance and resistance network model of the electric spindle of a machining center to be monitored and performs preliminary finite element simulation to obtain transient temperature field prediction data. The heat capacitance and resistance network model is then parameterized using a joint optimization algorithm based on the transient temperature field prediction data and local measured temperature data to obtain a real-time prediction model for the electric spindle's transient temperature field. After mesh refinement, a real-time electric spindle temperature distribution map is generated for real-time identification of abnormal heating faults. This method can provide real-time, complete, and accurate predictions of the electric spindle's transient temperature field under current operating conditions, effectively protecting the electric spindle.

Description

Technical Field

[0001] The present invention relates to a technology in the field of machine tool spindle control, specifically a method for real-time temperature estimation and fault monitoring in an unmeasurable temperature area of ​​an electric spindle based on mechanism-data fusion. Background Art

[0002] The lack of strong cooling measures in the unmeasurable temperature region inside a machine tool's electric spindle can cause irreversible spindle damage or destruction. Therefore, it is necessary to use an indirect temperature estimation method to predict the temperature field in this unmeasurable region in real time to achieve intelligent monitoring and fault detection of the production process. Summary of the Invention

[0003] In response to the shortcomings of existing electric spindles, such as the existence of temperature monitoring blind spots inside the spindle, which makes fault monitoring difficult and the temperature of the spatially measurable area can only be predicted in time, the present invention proposes a real-time temperature estimation and fault monitoring method for the unmeasurable temperature area of ​​the electric spindle. The method can predict the transient temperature field of the electric spindle under the current working conditions in real time, completely and accurately, and better protect the safety of the electric spindle.

[0004] The present invention is achieved through the following technical solutions:

[0005] The present invention relates to a method for real-time temperature estimation and fault monitoring in an unmeasurable temperature area of ​​an electric spindle. By constructing a heat capacitance and thermal resistance network model of an electric spindle of a machining center to be monitored and performing preliminary finite element simulation, transient temperature field prediction data is obtained. Parameters of the heat capacitance and thermal resistance network model are jointly optimized based on the transient temperature field prediction data and local measured temperature data to obtain a real-time prediction model for the transient temperature field of the electric spindle. After mesh refinement, a real-time electric spindle temperature distribution map is generated and used for real-time judgment of abnormal heating faults.

[0006] The thermal capacitance and resistance network model is derived by modifying the heat transfer coefficients at the flow coupling interface and the kinematic pair interface to dynamically change with spindle speed and real-time temperature, thereby adapting to complex and variable machine tool operating conditions. Ultimately, a thermal capacitance and resistance network model for the monitored machining center's electric spindle is established, using spindle operating conditions and ambient temperature as inputs and the temperatures at each node of the thermal capacitance and resistance network as outputs.

[0007] The local measured data refers to: arranging a number of temperature sensors in the temperature measurable area of ​​the electric spindle of the machining center to be monitored to record the dynamic temperature changes of the spindle under different time-varying working conditions.

[0008] The joint parameter optimization method includes: dividing the parameters of the heat capacitance and thermal resistance network model into constants, static parameters and time-varying functions, optimizing the model parameters through a two-step parameter estimation method based on a genetic algorithm, bringing the optimized model parameters into the heat capacitance and thermal resistance network model, and inputting the current spindle operating condition data and real-time monitored ambient temperature change data into the heat capacitance and thermal resistance network model through real-time communication means of the numerical control system, thereby realizing real-time temperature prediction at each node of the heat capacitance and thermal resistance network.

[0009] The grid refinement process includes: mapping the half-section of the electric spindle of the monitored machining center to a square grid space, thereby obtaining a subdivided grid model of a heat capacitance and thermal resistance network; assigning the node temperature predicted by the heat capacitance and thermal resistance network model to the equivalent temperature center corresponding to the subdivided grid area, obtaining the stable temperature of all unassigned grids through fuzzy operator loop iteration calculation, and symmetricizing the final temperature result of the subdivided grid along the center axis of the spindle to draw a thermal map of the temperature value of each grid, that is, a real-time electric spindle temperature distribution map under the current working conditions.

[0010] The real-time judgment of abnormal heating fault includes: collecting real-time measured temperature through an embedded temperature sensor arranged at the interface between the internal temperature unmeasurable area and the external temperature measurable area, obtaining the real-time predicted temperature at the position of the embedded temperature sensor through a heat capacitance and thermal resistance network model, and comparing the real-time measured temperature with the real-time predicted temperature. If the deviation between the real-time measured temperature and the real-time predicted temperature is greater than a set threshold, it is considered that an abnormal heating fault has occurred.

[0011] The present invention relates to a system for implementing the above-mentioned method, comprising: a sensing unit, a model unit, a display unit and an execution unit, wherein: the sensing unit obtains the current spindle working condition through real-time communication means of a numerical control system and monitors the ambient temperature change information in real time through an ambient temperature sensor; the model unit predicts the temperature at each node in real time through a heat capacitance and thermal resistance network model based on the information obtained by the sensing unit; the display unit performs grid refinement processing based on the temperature information at each node predicted by the model unit, obtains a real-time electric spindle temperature distribution diagram under the current working condition and displays it on a graphic interface; the execution unit calculates and performs threshold judgment processing with an embedded temperature sensor based on the temperature information at each node predicted by the model unit, obtains a fault judgment result and sends an execution signal to a subsequent fault handling process.

[0012] Technical Effects

[0013] This invention combines the advantages of finite element simulation, experimental data-driven modeling, and thermal capacitance and resistance network models to establish a real-time prediction model for the transient temperature field of an electric spindle, featuring complete spatial distribution, accurate temperature prediction, and excellent real-time performance. Compared to existing technologies, this invention can predict the transient temperature field of an electric spindle in real time with a complete spatial distribution and accurate numerical values, resulting in a high fault detection rate and fast time response characteristics, thus better protecting the safety of the electric spindle. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 Flowchart of the present invention;

[0015] In the figure: 1 spindle housing, 2 spindle motor stator, 3 spindle coolant, 4 spindle mounting base, 5 front bearing seat, 6 rear bearing, 7 spindle motor rotor, 8 rotating shaft, 9 front bearing, 10 tool holder tapered sleeve, 11 embedded temperature sensor, 12 temperature unmeasurable area;

[0016] Figure 2 Schematic diagram of the thermal capacitance and thermal resistance network model of the embodiment;

[0017] Figure 3 A real-time temperature distribution diagram under a certain working condition predicted by the embodiment;

[0018] Figure 4 A comparison chart of fault detection between this method and the traditional method under different working conditions and fault degrees of Monte Carlo simulation in the embodiment;

[0019] Figure 5 This is a comparison chart of the delay time of fault detection between this method and the traditional method when a severe fault reaching the damage risk temperature occurs in the Monte Carlo simulation of the embodiment. DETAILED DESCRIPTION

[0020] This embodiment relates to a method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region based on mechanism-data fusion, comprising:

[0021] The first step is to Figure 1 The thermal characteristics of a typical electric spindle of a machining center shown in the figure are analyzed. The dynamic heat generation power of the spindle motor and bearings under complex working conditions and the dynamic convection heat transfer characteristics of the spindle coolant and air are considered. Figure 2 The thermal capacitance and thermal resistance network model shown describes the heat generation, transfer and dissipation process of the electric spindle.

[0022] The thermal capacitance and thermal resistance network model includes: 25 thermal capacitors, 29 conductive thermal resistances, 13 basic convection resistances, 13 convection interfaces, 3 static convection heat transfer coefficients, 1 time-varying convection heat transfer coefficient and 4 time-varying heating powers of main axis heat sources.

[0023] like Figure 1The area inside the dotted box is the unmeasurable temperature area of ​​the spindle.

[0024] In the second step, a finite element transient thermal model is established for the electric spindle of the machining center, and a transient temperature field simulation is performed based on the empirically estimated finite element boundary conditions to obtain transient temperature field prediction data with complete spatial distribution, including the area where the spindle temperature cannot be measured, but the temperature values ​​are not accurate enough.

[0025] In the third step, several temperature sensors are arranged in the temperature measurable area of ​​the electric spindle of the machining center to record the dynamic temperature changes of the spindle under different time-varying working conditions, that is, the local measured temperature data with incomplete spatial distribution but accurate temperature values ​​excluding the unmeasurable area of ​​the spindle temperature.

[0026] The fourth step is to perform parameter joint optimization on the thermal capacitance and thermal resistance network model established in the first step based on the transient temperature field prediction data obtained in the second step and the local measured temperature data obtained in the third step, and establish a real-time prediction model for the transient temperature field of the electric spindle. Specifically, the following steps are performed:

[0027] 4.1) The parameters of the thermal capacitance and thermal resistance network model are divided into constants, static parameters and dynamic parameters;

[0028] The constant mentioned means that the heat capacity can be determined directly and accurately based on the structure and specific heat capacity of the spindle material. Therefore, the heat capacity is classified as a constant.

[0029] Static parameters are those that do not change with operating conditions. They include 29 conduction resistances, 13 basic convection resistances, 13 convection interface areas, and three static convection heat transfer coefficients. These static parameters primarily describe the heat transfer relationship between temperature nodes.

[0030] The dynamic parameters are: a time-varying convection heat transfer coefficient and a time-varying heating power of four spindle heat sources that vary with the spindle working condition and real-time temperature.

[0031] 4.2) Model parameter optimization is performed using a two-step parameter estimation method based on a genetic algorithm. Specifically, the dynamic parameters in the heat capacitance and thermal resistance network model are first replaced by empirically estimated finite element boundary conditions, and the static parameters are optimized to make them fit the transient temperature field prediction data of the finite element transient thermal model as closely as possible. Then, the optimized static parameters are substituted into the heat capacitance and thermal resistance network model, and the dynamic parameters are optimized to make them fit the local measured temperature data of the thermal characteristics experiment as closely as possible.

[0032] 4.3) The optimized model parameters are brought into the heat capacitance and thermal resistance network model. The current spindle operating condition data and the real-time monitored ambient temperature change data are input into the heat capacitance and thermal resistance network model through the real-time communication means of the CNC system. This can realize the real-time temperature prediction of each node of the heat capacitance and thermal resistance network.

[0033] In the fifth step, the temperature at each node of the heat capacitance and thermal resistance network obtained in the fourth step is further subdivided. The temperature of the subdivided grid is determined by the temperature between two adjacent nodes to solve the problem that the nodes of the heat capacitance and thermal resistance network are sparse and the predicted temperature only represents the equivalent center temperature of a certain area in the electric spindle. The temperature field data at a certain moment is obtained, which specifically includes:

[0034] 5.1) Map the half-section of the electric spindle to a 100*21 square grid space to obtain a subdivided grid model of the thermal capacitance and thermal resistance network. There are 1264 grids in total, and the initial grid temperature is set to 20℃.

[0035] 5.2) Assign the node temperature predicted by the thermal capacitance and thermal resistance network model to the equivalent temperature center corresponding to the subdivided grid area, and reduce the temperature difference of the subdivided grid through the Gaussian fuzzy algorithm.

[0036] 5.3) Repeat step 5.2 until the temperature of the subdivided grid stabilizes.

[0037] 5.4) The final temperature result of the subdivided grid is symmetrical along the main axis and plotted as follows: Figure 3 The thermal map of the temperature values ​​of each grid is shown, and the real-time temperature distribution map of the electric spindle under the current working conditions is obtained.

[0038] The sixth step is to use the temperature at each node of the thermal capacitance and thermal resistance network predicted in the fourth step for real-time monitoring of abnormal heating failure of the electric spindle, specifically including:

[0039] 6.1) Embed a temperature sensor in the spindle, closest to the unmeasurable area, near the outer rings of the front and rear bearings, and collect the embedded temperature sensor readings.

[0040] 6.2) Compare the embedded temperature sensor value with the real-time predicted temperature of the corresponding node of the thermal capacitance and thermal resistance network. When the deviation is greater than the set threshold, it is considered that an abnormal heating fault has occurred.

[0041] In the parameter range of spindle speed 0 to 20,000 rpm and fault heating power ratio 0 to 4 times of normal power, 1,000 Monte Carlo simulations of abnormal heating fault of electric spindle were carried out. The comparison chart of fault detection between this method and traditional method was obtained, as shown in the figure below. Figure 4 The present invention has a detection rate of 68.1% for all faults, including 100% for severe faults that could damage the spindle, while the conventional method has a detection rate of only 24.0% and 83.0%. Further analysis of the severe faults that could damage the spindle revealed a delay between the time when the two methods detected abnormal heating faults and the time when the bearing balls reached the grease's allowable limit temperature, as shown in Figure 2. Figure 5As shown in the figure, it can be seen that in most cases, this method can detect abnormal heating failures before the bearing ball temperature reaches the limit; while traditional methods generally detect abnormal heating failures only after the bearing ball temperature exceeds the limit.

[0042] Compared to existing fault monitoring methods that set a static temperature threshold and identify an abnormal heating fault as soon as the value collected in real time by the embedded temperature sensor exceeds this threshold, the present invention uses a dynamic threshold generated in real time by the model, theoretically resulting in a higher fault detection rate and faster time response. In summary, this method offers a higher fault detection rate and faster time response, better protecting the safety of the electric spindle.

[0043] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principles and purpose of the present invention. The scope of protection of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. All implementation schemes within its scope shall be subject to the constraints of the present invention.

Claims

1. A method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region, characterized in that: By constructing a heat capacitance and thermal resistance network model of the electric spindle of the machining center to be monitored and performing preliminary finite element simulation, transient temperature field prediction data is obtained. Based on the transient temperature field prediction data and local measured temperature data, the parameters of the heat capacitance and thermal resistance network model are jointly optimized to obtain a real-time prediction model for the electric spindle's transient temperature field. After mesh refinement, a real-time electric spindle temperature distribution map is generated and used for real-time judgment of abnormal heating faults. Specifically, the following steps are performed: The first step is to analyze the thermal characteristics of a typical electric spindle for a machining center. Considering the dynamic heat generation power of the spindle motor and bearings under complex operating conditions, as well as the dynamic convection heat transfer characteristics of the spindle coolant and air, a thermal capacitance and thermal resistance network model is established to describe the heat generation, transfer, and dissipation processes of the electric spindle. The second step is to establish a finite element transient thermal model for the machining center's electric spindle and perform transient temperature field simulation based on empirically estimated finite element boundary conditions. This yields transient temperature field prediction data with a complete spatial distribution, including areas where spindle temperature cannot be measured, but with inaccurate temperature values. The third step is to place several temperature sensors in the temperature-measurable area of ​​the machining center's electric spindle to record the dynamic temperature changes of the spindle under different time-varying working conditions. This means that the local measured temperature data, which does not include the incomplete spatial distribution of the spindle temperature in the unmeasurable area but has accurate temperature values, is obtained. In the fourth step, based on the transient temperature field prediction data obtained in the second step and the local measured temperature data obtained in the third step, the parameters of the heat capacity and thermal resistance network model established in the first step are jointly optimized to establish a real-time prediction model for the transient temperature field of the electric spindle; In the fifth step, the temperature at each node of the heat capacitance and thermal resistance network obtained in the fourth step is further subdivided. The temperature of the subdivided grid is determined by the temperature between two adjacent nodes. This solves the problem that the nodes of the heat capacitance and thermal resistance network are sparse and the predicted temperature only represents the equivalent center temperature of a certain area in the electric spindle. The temperature field data at a certain moment is obtained. The sixth step is to use the temperature at each node of the thermal capacitance and thermal resistance network predicted in the fourth step for real-time monitoring of abnormal heating failure of the electric spindle, specifically including: 6.1) Embed a temperature sensor in the spindle, closest to the unmeasurable area, near the outer rings of the front and rear bearings, and collect the embedded temperature sensor readings. 6.2) Compare the embedded temperature sensor value with the real-time predicted temperature of the corresponding node of the thermal capacitance and thermal resistance network. When the deviation is greater than the set threshold, it is considered that an abnormal heating fault has occurred.

2. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1 is characterized in that: The thermal capacitance and thermal resistance network model is obtained by improving the heat transfer coefficient of the flow coupling interface and the kinematic pair interface into a dynamic value that changes with the spindle speed and real-time temperature to adapt to the complex and changeable machine tool working conditions. Finally, a thermal capacitance and thermal resistance network model of the electric spindle of the machining center to be monitored is established, which takes the spindle working conditions and ambient temperature as input and the temperature at each node of the thermal capacitance and thermal resistance network as output.

3. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1 is characterized in that: The locally measured temperature data refers to: arranging a number of temperature sensors in the temperature measurable area of ​​the electric spindle of the machining center to be monitored to record the dynamic temperature changes of the spindle under different time-varying working conditions.

4. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1, wherein: The joint parameter optimization method includes: dividing the parameters of the heat capacitance and thermal resistance network model into constants, static parameters and time-varying functions, optimizing the model parameters through a two-step parameter estimation method based on a genetic algorithm, bringing the optimized model parameters into the heat capacitance and thermal resistance network model, and inputting the current spindle operating condition data and real-time monitored ambient temperature change data into the heat capacitance and thermal resistance network model through real-time communication means of the numerical control system, thereby realizing real-time temperature prediction at each node of the heat capacitance and thermal resistance network.

5. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1 is characterized in that: The grid refinement process includes: mapping the half-section of the electric spindle of the monitored machining center to a square grid space, thereby obtaining a subdivided grid model of a heat capacitance and thermal resistance network; assigning the node temperature predicted by the heat capacitance and thermal resistance network model to the equivalent temperature center corresponding to the subdivided grid area, obtaining the stable temperature of all unassigned grids through fuzzy operator loop iteration calculation, and symmetricizing the final temperature result of the subdivided grid along the center axis of the spindle to draw a thermal map of the temperature value of each grid, that is, a real-time electric spindle temperature distribution map under the current working conditions.

6. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1, wherein: The real-time judgment of abnormal heating fault includes: collecting real-time measured temperature through an embedded temperature sensor arranged at the interface between the internal temperature unmeasurable area and the external temperature measurable area, obtaining the real-time predicted temperature at the position of the embedded temperature sensor through a heat capacitance and thermal resistance network model, and comparing the real-time measured temperature with the real-time predicted temperature. If the deviation between the real-time measured temperature and the real-time predicted temperature is greater than a set threshold, it is considered that an abnormal heating fault has occurred.

7. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1, wherein: The fourth step specifically includes: 4.1) The parameters of the thermal capacitance and thermal resistance network model are divided into constants, static parameters and dynamic parameters; The constant mentioned means that the heat capacity can be determined directly and accurately based on the structure and specific heat capacity of the spindle material; therefore, the heat capacity is classified as a constant; The static parameters are parameters that do not change with operating conditions, including 29 conduction resistances, 13 basic convection resistances, 13 convection interface areas, and 3 static convection heat transfer coefficients. These static parameters mainly describe the heat transfer relationship between temperature nodes. The dynamic parameters are: a time-varying convection heat transfer coefficient and four time-varying heating powers of the spindle heat sources that vary with the spindle operating conditions and real-time temperature; 4.2) Optimizing model parameters using a two-step parameter estimation method based on a genetic algorithm. Specifically, the dynamic parameters in the heat capacitance and resistance network model are first replaced with empirically estimated finite element boundary conditions. The static parameters are then optimized to closely match the transient temperature field prediction data from the finite element transient thermal model. The optimized static parameters are then substituted into the heat capacitance and resistance network model, and the dynamic parameters are optimized to closely match the locally measured temperature data from the thermal characterization experiment. 4.3) The optimized model parameters are brought into the heat capacitance and thermal resistance network model. The current spindle operating condition data and the real-time monitored ambient temperature change data are input into the heat capacitance and thermal resistance network model through the real-time communication means of the CNC system. This can realize the real-time temperature prediction of each node of the heat capacitance and thermal resistance network.

8. The method for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region according to claim 1, wherein: The fifth step specifically includes: 5.1) Map the half-section of the motorized spindle to a 100*21 square grid space to obtain a subdivided grid model of the thermal capacitance and thermal resistance network. There are 1264 grids in total, and the initial grid temperature is set to 20°C. 5.2) Assign the node temperatures predicted by the thermal capacitance and thermal resistance network model to the equivalent temperature center corresponding to the subdivided grid area, and use the Gaussian fuzzy algorithm to reduce the temperature difference of the subdivided grid; 5.3) Repeat step 5.2 until the temperature of the subdivided grid is stable; 5.4) The final temperature results of the subdivided grids are symmetrically arranged along the spindle axis, and a thermal map of the temperature values ​​of each grid is drawn to obtain the real-time temperature distribution map of the electric spindle under the current working conditions.

9. A system for real-time temperature estimation and fault monitoring of an electric spindle in an unmeasurable temperature region for implementing the method of any one of claims 1 to 8, characterized in that: include: A sensing unit, a model unit, a display unit and an execution unit, wherein: the sensing unit obtains the current spindle working condition through the real-time communication means of the numerical control system and monitors the ambient temperature change information in real time through the ambient temperature sensor; the model unit predicts the temperature of each node in real time through the heat capacitance and thermal resistance network model based on the information obtained by the sensing unit; the display unit performs grid refinement processing based on the temperature information of each node predicted by the model unit, obtains the real-time electric spindle temperature distribution diagram under the current working condition and displays it on the image interface; the execution unit calculates the real-time predicted temperature of the corresponding node of the heat capacitance and thermal resistance network based on the temperature information of each node predicted by the model unit, and performs threshold judgment processing with the value of the embedded temperature sensor to obtain the fault judgment result and send an execution signal to the subsequent fault handling process.

Images