A thermal error compensation method and system for an electric spindle

Abstract

The application discloses a kind of thermal error compensation method and system of electric spindle, it is related to thermal error compensation technical field, the method includes: multiple types temperature sensor is laid out, to construct temperature sensing array;According to the real-time monitoring data returned by temperature sensing array, generate spindle three-dimensional temperature cloud picture;Identify the spatial thermal error area of temperature gradient exceeding preset threshold value;From electric spindle log, call the runtime data of the running condition of spindle;Thermal error traceability is carried out, and error source type feature is obtained;Thermal error compensation optimization is carried out, and collaborative compensation parameter is output;Real-time thermal error dynamic compensation is carried out to electric spindle.The present application solves the technical problems that electric spindle thermal error is difficult to accurately identify and locate in the prior art, and the compensation efficiency is low, which leads to poor running accuracy and insufficient stability of electric spindle, achieves accurate identification, positioning and efficient compensation of electric spindle thermal error, and improves the technical effects of running accuracy and stability of electric spindle.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of thermal error compensation, in particular to a thermal error compensation method and system of an electric spindle. BACKGROUND

[0002] In the field of modern precision machining, the electric spindle as the core component of the numerical control machine tool directly affects the quality of the processed products. However, during the high-speed operation of the electric spindle, factors such as motor heating and bearing friction can cause thermal deformation and generate thermal errors, which in turn lead to a decrease in machining precision. In the prior art, the monitoring of the thermal error of the electric spindle often relies on a single-point temperature sensor, which cannot fully capture the distribution of the thermal error; in terms of error identification and positioning, there is a lack of accurate and effective methods, making it difficult to quickly determine the error source; and the compensation strategy is mostly based on experience, resulting in low compensation efficiency and unstable effects.

[0003] The prior art has the technical problems of inaccurate identification and positioning of the thermal error of the electric spindle, low compensation efficiency, poor running precision of the electric spindle, and insufficient stability. SUMMARY

[0004] The present application provides a thermal error compensation method and system of an electric spindle, which is used to solve the technical problems of inaccurate identification and positioning of the thermal error of the electric spindle, low compensation efficiency, poor running precision of the electric spindle, and insufficient stability in the prior art.

[0005] In view of the above problems, the present application provides a thermal error compensation method and system of an electric spindle.

[0006] In a first aspect of the present application, a thermal error compensation method of an electric spindle is provided, which comprises:

[0007] By arranging multiple types of temperature sensors on the electric spindle to construct a temperature sensor array covering the global heat source area of the electric spindle, a spindle three-dimensional temperature cloud map is generated by a Kriging spatial interpolation algorithm according to the real-time monitoring data returned by the temperature sensor array; a spatial thermal error area with a temperature gradient exceeding a preset threshold is identified in the spindle three-dimensional temperature cloud map; the spatial boundary of the spatial thermal error area is used as a spatial constraint for calling the running log, and the running time sequence data that meets the spindle running condition is called from the electric spindle log; the error source type characteristics are obtained by tracing the thermal error according to the running time sequence data; after matching the compensation strategy priority according to the error source type characteristics, the thermal error compensation optimization is performed with the real-time compensation strategy priority as the constraint, and the collaborative compensation parameters are output; the PID adjustment module of the electric spindle is run with the collaborative compensation parameters to dynamically compensate the real-time thermal error of the electric spindle.

[0008] In a second aspect of the present application, a thermal error compensation system of an electric spindle is provided, and the system comprises:

[0009] a temperature sensor array construction module configured to construct a temperature sensor array covering a global heat source area of the electric spindle by arranging multiple types of temperature sensors on the electric spindle; a spindle three-dimensional temperature cloud map generation module configured to generate a spindle three-dimensional temperature cloud map by a Kriging spatial interpolation algorithm according to real-time monitoring data returned by the temperature sensor array; a spatial thermal error area identification module configured to identify a spatial thermal error area with a temperature gradient exceeding a preset threshold in the spindle three-dimensional temperature cloud map; a runtime sequence data calling module configured to call runtime sequence data meeting a spindle operation condition from a log of the electric spindle by taking a spatial boundary of the spatial thermal error area as a spatial constraint of a runtime log call; an error source type feature acquisition module configured to trace a thermal error source according to the runtime sequence data to obtain an error source type feature; a collaborative compensation parameter output module configured to perform thermal error compensation optimization by taking a real-time compensation strategy priority as a constraint after matching a compensation strategy priority according to the error source type feature, and output a collaborative compensation parameter; and a real-time thermal error dynamic compensation module configured to perform real-time thermal error dynamic compensation on the electric spindle by adopting the collaborative compensation parameter to run a PID adjustment module of the electric spindle.

[0010] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0011] The multiple types of temperature sensors are arranged on the electric spindle to construct a temperature sensor array, a spindle three-dimensional temperature cloud map is generated according to real-time monitoring data returned by the temperature sensor array, a spatial thermal error area with a temperature gradient exceeding a preset threshold is identified in the spindle three-dimensional temperature cloud map, runtime sequence data meeting a spindle operation condition is called from a log of the electric spindle, a thermal error source is traced according to the runtime sequence data to obtain an error source type feature, thermal error compensation optimization is performed by taking a real-time compensation strategy priority as a constraint after matching a compensation strategy priority according to the error source type feature, and a collaborative compensation parameter is output, and real-time thermal error dynamic compensation is performed on the electric spindle. The technical effect of accurately identifying, locating and efficiently compensating thermal errors of the electric spindle is achieved, and the operation precision and stability of the electric spindle are improved. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0013] Figure 1This is a schematic flowchart of a thermal error compensation method for an electric spindle provided in an embodiment of this application;

[0014] Figure 2 This is a schematic diagram of a thermal error compensation system for an electric spindle provided in an embodiment of this application.

[0015] Figure labeling: Temperature sensor array construction module 10, main axis 3D temperature cloud map generation module 20, spatial thermal error region identification module 30, runtime sequence data retrieval module 40, error source type feature acquisition module 50, collaborative compensation parameter output module 60, real-time thermal error dynamic compensation module 70. Detailed Implementation

[0016] This application provides a thermal error compensation method and system for electric spindles, which addresses the technical problems in the prior art where thermal errors in electric spindles are difficult to accurately identify and locate, resulting in low compensation efficiency and poor operating accuracy and stability of electric spindles.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, this application provides a method for thermal error compensation of an electric spindle, the method comprising:

[0019] Step S100: By deploying multiple types of temperature sensors on the electric spindle, a temperature sensing array is constructed to cover the entire heat source region of the electric spindle.

[0020] Specifically, in the global heat source areas of the electric spindle, such as bearings, motors, and lead screws, various types of temperature sensors, including thermocouples, infrared temperature sensors, and fiber Bragg grating temperature sensors, are deployed based on the heat source distribution characteristics and heat conduction paths. Thermocouples are used for precise measurement of local point temperatures (accuracy up to ±0.5℃), infrared sensors cover a large area for temperature monitoring, and fiber Bragg grating sensors are suitable for distributed measurement in environments with high electromagnetic interference. By optimizing the sensor layout (e.g., four groups evenly distributed circumferentially at 10cm intervals along the spindle axis), a temperature sensing array covering the entire electric spindle is constructed. This enables real-time synchronous acquisition of multi-dimensional heat source signals, such as bearing temperature rise, motor winding temperature, and housing surface temperature field, providing high-density, high-reliability basic data support for subsequent three-dimensional temperature field modeling and thermal error analysis.

[0021] Step S200: generating a three-dimensional temperature cloud map of the spindle according to real-time monitoring data returned by the temperature sensing array through a Kriging spatial interpolation algorithm.

[0022] Specifically, the multi-type temperature data (including point data of thermocouples, surface data of infrared sensors, etc.) returned by the temperature sensing array in real time is first subjected to spatio-temporal synchronization preprocessing to eliminate time misalignment and sensor installation position errors caused by sampling frequency differences. Subsequently, the Kriging spatial interpolation algorithm is used to perform three-dimensional spatial fitting on the discrete temperature data: taking the geometric axis of the motorized spindle as the reference, a cylindrical coordinate system (radial R, axial Z, and circumferential θ) is established, each sensor measurement point is taken as a spatial sample point, the spatial correlation between sample points is calculated based on a semi-variogram function model, and the temperature values of unmeasured points inside and on the surface of the spindle are predicted. Through iterative optimization of interpolation parameters (such as the range and nugget effect), a three-dimensional temperature cloud map of the spindle with a resolution of 0.5 cm x 0.5 cm x 0.5 cm is generated, which intuitively presents the temperature distribution gradient (such as the bearing area high-temperature core area ≥ 80℃, and the motor tail heat dissipation area 50-60℃) and the heat flow direction, providing a visual spatial temperature field model for subsequent fine identification of thermal error areas.

[0023] Step S300: identifying a spatial thermal error area with a temperature gradient exceeding a preset threshold value in the three-dimensional temperature cloud map of the spindle.

[0024] Specifically, first, based on the geometric dimensions (such as total length L and diameter D) of the motorized spindle, an initial grid side length (for example, the initial side length is L / 10) is set, and the three-dimensional temperature cloud map of the spindle is divided into multiple initial spatial temperature grids. For each initial grid, the geometric center point is extracted as a voxel, the temperature gradient amplitude (unit: ℃ / cm) and gradient direction standard deviation of adjacent grids are calculated, and P initial grids with dramatic temperature changes are filtered and selected through predefined temperature gradient amplitude deviation thresholds (such as ≥ 5℃ / cm) and gradient direction standard deviation deviation thresholds (such as ≥ 15°). Subsequently, 1 / K (such as K=2) of the initial grid side length is taken as the first updated grid side length, the P initial grids are subjected to secondary subdivision, sub-grids are generated, and the temperature gradient vector consistency analysis of adjacent sub-grids is repeated to locate F first updated grids with more concentrated temperature gradients. According to this logic, the layers are refined and segmented for a preset number of iterations (such as 3 times), and finally a plurality of thermal error sub-grids are obtained. Three-dimensional morphological closing operation (small cavities are eliminated through dilation-erosion operation, and adjacent areas are connected) is performed on these sub-grids to generate a closed and continuous spatial thermal error area, which accurately outlines the high-risk area with a temperature gradient exceeding the preset threshold value (such as ≥ 3℃ / cm), providing a clear spatial positioning basis for subsequent error tracing and compensation.

[0025] Step S400: Taking the spatial boundary of the spatial thermal error region as a spatial constraint of a running log call, running time sequence data satisfying the spindle running condition is called from the spindle log.

[0026] Specifically, first, the three-dimensional boundary vertex coordinates of the spatial thermal error region (such as axial Z1-Z2, radial R1-R2, and circumferential θ1-θ2) are extracted, and a spatial grid interval containing parameters such as spatial position and volume proportion is constructed. According to the proportion of the region in the overall space of the electric spindle (such as 20% of the length of the spindle), a corresponding backtracking time window (such as 20% of the backtracking data for the last 2 hours) is matched. At the same time, based on the device structure of the electric spindle (such as bearings, motors, cooling systems, and M component parts), an independent component grid interval is constructed for each component. By calculating the coverage ratio of the spatial grid interval to each component grid interval (such as a bearing component coverage ratio of 60%), the top N component parts (such as N=3) are screened out according to the coverage ratio. Finally, taking the spatial grid interval and the backtracking time window as the spatial constraint, and the top N components as the component constraint, the running log backtracking call is executed under the double constraints, the running time sequence data containing parameters such as speed, load current, and cooling flow are obtained, and the data are strongly related to the spatial position of the thermal error region and the corresponding component working condition, which provides accurate data support for subsequent error tracing.

[0027] Step S500: According to the running time sequence data, the error source type feature is obtained.

[0028] Specifically, first, M sets of component running working condition baselines (including speed-temperature curves, load-current thresholds, etc.) of M component parts (such as bearings, motors, leadscrews, etc.) under multiple typical working conditions (such as high-speed idling and heavy-load cutting) are obtained interactively. The running time sequence data is processed into a time sequence image, and real-time running working condition curves (such as bearing temperature fluctuation curve and motor current change curve) of the top N related components are generated. According to the current running condition attribution (such as spindle speed 3000 r / min and feed rate 0.1 mm / r), N component working condition baselines are called from the M sets of baselines. By comparing and analyzing the deviation of the real-time curve and the baseline (such as temperature deviation ΔT≥10℃ and current deviation ΔI≥5%), if the deviation of all components is within the preset scale, it is determined that the error source is systematic thermal load accumulation; if the deviation of a component (such as the first component bearing) exceeds the threshold, it is determined that the component function is degraded, and the degradation severity level is matched according to the deviation amplitude (such as 1st mild degradation and 3rd severe degradation). Finally, the error source type feature is output in the form of component name and degradation level, which provides a clear direction for compensation strategy matching.

[0029] Step S600: After matching the compensation strategy priority according to the error source type characteristics, the real-time compensation strategy priority is used as a constraint to optimize the thermal error compensation, and the cooperative compensation parameters are output.

[0030] Specifically, based on historical thermal error compensation data, compensation strategies with high-frequency effects are aggregated according to error source types (such as component degradation and systematic accumulation), and a thermal error compensation strategy library containing various sample error source types and their corresponding compensation priorities is constructed (for example, bearing degradation triggers cooling PID parameter adjustment, and systematic accumulation triggers global parameter adjustment strategy). By traversing the strategy library through error source type characteristics, the real-time compensation strategy priority is matched and output, and is divided into hard constraint priority (such as immediate cooling flow threshold adjustment) and soft constraint priority (such as motor power fine-tuning that can be delayed). With the hard constraint priority as the boundary condition, the minimum energy consumption compensation parameter that meets the lower limit of the thermal error suppression rate (such as a suppression rate ≥ 70%) is solved in the energy consumption model as the hard constraint compensation parameter, and the remaining thermal error margin is calculated. If the margin exceeds the preset tolerance range (such as > 0.02 mm), the soft constraint compensation parameter (such as cooperative adjustment of cooling temperature and spindle speed) for hedging the margin is solved within the resource capacity of the remaining cooling power, motor load, etc. with the soft constraint priority as the optimization target. Finally, the hard constraint parameter is executed first, and the soft constraint parameter is supplemented cooperatively, and the cooperative compensation parameters including cooling PID coefficients, component parameter adjustment thresholds, etc. are output in sequence, achieving a balance between thermal error suppression and energy optimization.

[0031] Step S700: The PID adjustment module of the electric spindle is run with the cooperative compensation parameters to dynamically compensate for the real-time thermal error of the electric spindle.

[0032] Specifically, the synergistic compensation parameters are decomposed into multiple independent thermal error compensation control parameters, such as the proportional coefficient (Kp), integral time (Ti), and differential time (Td) of the cooling system PID adjustment, as well as motor power adjustment parameters and spindle speed correction values, each of which is assigned an execution priority (such as cooling flow adjustment with the highest priority) and an execution time limit identifier (such as immediate execution / delayed for 5 minutes). According to the priority and time limit, the parameters are assigned corresponding PID adjustment weights (such as high-priority parameter weight accounting for 70%), and the parameters are converted into input variables (such as temperature deviation ΔT and speed deviation Δn) recognizable by the PID adjustment module. The PID adjustment module dynamically responds to multiple variable inputs according to the weights, first executes the hard constraint compensation parameters (such as immediately increasing the cooling flow from 5 L / min to 8 L / min), quickly suppresses the dominant thermal error source; at the same time, gradually adjusts the synergistic parameters according to the soft constraint priority (such as fine-tuning the motor current threshold by 0.5 A every 10 minutes), continuously offsets the remaining thermal error. By real-time monitoring of the gradient change of the temperature cloud map after compensation, the PID parameter weight is dynamically adjusted to form a closed-loop control of monitoring-compensation-reoptimization, realizing dynamic adaptive compensation of the thermal error of the motorized spindle and ensuring that the spindle running accuracy is maintained within the preset threshold (such as thermal deformation ≤0.01 mm).

[0033] In one possible implementation manner, step S300 further includes:

[0034] Step S310: setting an initial grid side length based on the spindle geometric size of the motorized spindle.

[0035] Step S320: dividing the three-dimensional temperature cloud map space of the spindle into a plurality of initial spatial temperature grids using the initial grid side length.

[0036] Step S330: performing adjacent grid temperature gradient vector consistency analysis on the plurality of initial spatial temperature grids to filter and select P initial spatial temperature grids.

[0037] Step S340: taking 1 / K of the initial grid side length as a first updated grid side length, performing adjacent sub-grid temperature gradient vector consistency analysis after secondary subdivision of the P initial spatial temperature grids, and locating F first updated spatial temperature grids.

[0038] Step S350: iteratively refining and segmenting the three-dimensional temperature cloud map of the spindle layer by layer for a preset number of times, and locating a plurality of thermal error sub-grids.

[0039] Step S360: performing three-dimensional morphological closing operation on the plurality of thermal error sub-grids to generate a closed continuous spatial thermal error region.

[0040] Specifically, according to the geometric parameters of the motorized spindle (such as the total length, diameter, bearing spacing, etc.), combined with the accuracy requirements of the thermal field monitoring, the initial grid length is set. For example, for a motorized spindle with a total length of 500 mm, if the axial thermal field resolution of every 50 mm is required, the initial grid length can be set to 50 mm, so that the three-dimensional temperature cloud map forms a uniform spatial grid division in the axial, radial and circumferential directions, providing an initial scale constraint for subsequent layer-by-layer refinement analysis of the thermal error region.

[0041] With the set initial grid length as the spatial division scale, the continuous space of the spindle three-dimensional temperature cloud map is discretized into a plurality of regular hexahedral initial space temperature grids. For example, if the initial grid length is 50 mm, the 500 mm long motorized spindle can be divided into 10 grid units in the axial direction, and the radial and circumferential directions are equally divided according to the diameter and angular resolution (such as 4 sectors in the circumferential direction of 360°). A three-dimensional grid matrix is formed in the axial x radial x circumferential direction. Each grid corresponds to a specific space region of the spindle (such as the bearing mounting section, the motor winding area), stores the average temperature value and spatial coordinates of the region, and provides a discretized space unit basis for subsequent temperature gradient analysis of adjacent grids.

[0042] The geometric center point of each initial space temperature grid is taken as a feature voxel, the average temperature value of the grid is extracted, and the consistency of heat conduction between adjacent grids is evaluated by calculating the temperature gradient vector (including gradient amplitude and direction) between the center points of adjacent grids. Specifically, first, the temperature gradient amplitude (unit: ℃ / mm) of adjacent grids is calculated, that is, the ratio of temperature difference to spatial distance, and the standard deviation of the gradient direction is calculated to measure the dispersion degree of the heat flow direction between adjacent grids. If the temperature gradient amplitude of a certain grid exceeds the preset threshold (such as ≥3℃ / mm) and the direction standard deviation is greater than the allowed range (such as ≥20°), it is determined that the grid is a thermal error sensitive region and is retained; otherwise, it is filtered out. Through the above analysis, P grids with rapid temperature changes and inconsistent heat flow directions are selected from all initial grids as candidate regions for subsequent refinement analysis, thereby focusing on high-risk thermal error regions and reducing invalid calculation.

[0043] The P initial space temperature grids screened out are taken as objects for refined analysis, and 1 / K (such as K=2, the edge length is reduced to 50% of the original edge length) of the initial grid edge length is taken as the first updated grid edge length, and secondary space subdivision is performed on each initial grid. For example, if the initial grid edge length is 50 mm, the updated edge length is 25 mm, each initial grid is divided into 2x2x2=8 sub-grids, and a finer three-dimensional space grid is formed. For each sub-grid, the geometric center point temperature value is extracted again, the temperature gradient amplitude and direction standard deviation between adjacent sub-grids are calculated, the consistency analysis logic of step S330 is repeated, and a more stringent screening threshold (such as temperature gradient amplitude ≥4℃ / mm, direction standard deviation ≥15°) is set, and F first updated space temperature grids with more concentrated temperature gradient and higher thermal flow direction dispersion are located from them. This process realizes local amplification analysis of the thermal error area by reducing the grid scale, accurately captures the temperature mutation details in the initial grid, and provides more accurate space units for subsequent iterative refinement.

[0044] The updated space temperature grids screened out in the previous round are processed by cyclic iteration: after each subdivision, the current grid edge length is reduced to 1 / K of the previous round (such as from 25 mm to 12.5 mm), and gradient vector consistency analysis of adjacent sub-grids is performed based on a more stringent temperature gradient threshold (such as increasing 1℃ / mm for each iteration), and areas with more intense temperature changes are filtered out layer by layer. For example, when the preset iteration number is 3 times, the initial grid (50 mm) is 25 mm after the first subdivision, 12.5 mm after the second subdivision, and 6.25 mm after the third subdivision. After each subdivision, low-risk areas are gradually excluded through double screening of gradient amplitude and direction standard deviation, and multiple thermal error sub-grids with a size of only 1 / K n (n is the number of iterations) of the initial grid are finally located. These sub-grids are densely distributed in the temperature gradient peak area and accurately represent the thermal error core nodes inside the electric spindle, providing high-density space feature units for subsequent morphological processing.

[0045] For the multiple discrete thermal error sub-grids obtained by multi-layer iterative subdivision and screening, a three-dimensional morphological closing operation is used for spatial topology optimization. First, an inflation operation is performed on all thermal error sub-grids, and by expanding the sub-grid boundary by a preset pixel unit (such as 1 voxel), adjacent discrete sub-grids are connected, and small gaps caused by sparse sensor distribution or interpolation error are filled. Then, an erosion operation is performed to remove isolated pixel points on the sub-grid boundary and eliminate small protrusions or cavities caused by noise interference. Through the combined operation of inflation and erosion, the dispersed thermal error sub-grids are integrated into a closed and continuous spatial thermal error region with smooth boundaries. This region completely covers the high-risk area where the temperature gradient exceeds the preset threshold, and the geometric shape conforms to the actual heat conduction path of the spindle (such as the annular high-temperature area distributed along the bearing circumference and the heat flow band extending axially along the motor shaft), providing an accurate physical space range for subsequent retrieval of operating data and matching compensation strategies according to the spatial boundary of the region.

[0046] In one possible implementation, step S330 further includes:

[0047] Step S331: Taking the geometric center points of the multiple initial spatial temperature grids as spatial representative voxels, a plurality of initial grid temperature values are extracted.

[0048] Step S332: Performing adjacent grid temperature gradient vector calculation on the multiple initial spatial temperature grids according to the multiple initial grid temperature values, outputting multiple initial temperature gradient amplitudes and multiple initial gradient direction standard deviations.

[0049] Step S333: Filtering and outputting P initial spatial temperature grids from the multiple initial spatial temperature grids according to the multiple initial temperature gradient amplitudes and multiple gradient direction standard deviations.

[0050] Specifically, for the divided multiple initial spatial temperature grids, the geometric center point of each grid is selected as a representative voxel representing the spatial temperature characteristics of the grid. For example, in a three-dimensional grid matrix uniformly distributed in the axial, radial, and circumferential directions, the geometric center point coordinates of each hexahedral grid can be calculated by averaging the grid boundary coordinates (such as the axial midpoint Z=(Z min +Z max ) / 2, the radial midpoint R=(R min +R max ) / 2, and the circumferential midpoint θ=(θ min +θ max ) / 2). By extracting the temperature values corresponding to each center point (which can be obtained by weighted averaging of all interpolation point temperatures in the grid), an initial grid temperature value set corresponding to each initial grid is formed, providing a core data basis for subsequent temperature gradient vector calculation between adjacent grids, ensuring that the temperature characteristics of each grid can be quantitatively analyzed through a single and spatially representative voxel value.

[0051] According to the temperature values of each initial spatial temperature grid obtained, for each grid, the temperature gradient vector between it and the adjacent grids (the grids directly adjacent in the axial, radial, and circumferential directions) is calculated. First, the temperature gradient amplitude between adjacent grids is calculated, that is, the absolute value of the difference between the temperature values of the two grids divided by the spatial distance between their center points, to obtain the temperature change in a unit space, which measures the degree of temperature change. At the same time, the direction of the temperature gradient between adjacent grids is calculated, which is represented by the spatial direction angle (such as the direction angle along the coordinate axis) to represent the direction of heat flow transfer. Then, for each grid, the temperature gradient direction data of all adjacent grids is counted, and the standard deviation of these directions is calculated to reflect the consistency of the heat flow direction: the smaller the standard deviation, the more concentrated the heat flow direction; the larger the standard deviation, the more dispersed the heat flow direction, and there may be complex heat conduction paths. Finally, two key parameters are output for each initial grid: the temperature gradient amplitude and the gradient direction standard deviation, which serve as quantitative basis for determining whether the grid belongs to a thermal error sensitive region.

[0052] Based on the temperature gradient amplitude and the gradient direction standard deviation of each initial spatial temperature grid obtained, double filtering thresholds are set to screen the grids: first, a temperature gradient amplitude threshold (such as 3°C / mm) and a gradient direction standard deviation threshold (such as 15°) are preset. If the temperature gradient amplitude of a certain grid is greater than or equal to the amplitude threshold, and the gradient direction standard deviation is greater than or equal to the direction standard deviation threshold, it is determined that the grid is in a thermal error sensitive region, and is retained as an effective grid. Otherwise, if either the amplitude or the standard deviation does not meet the threshold, it is considered that the grid has a gentle temperature change and consistent heat flow direction, and is filtered out. Through the above screening mechanism, P high-risk grids with sharp temperature changes and dispersed heat flow directions are accurately extracted from all initial grids, excluding the interference of low-risk regions, focusing on the key spatial units for subsequent analysis, and laying a foundation for the fine positioning of the thermal error region.

[0053] In one possible implementation manner, the step S333 further includes:

[0054] The step S3331: predefining a grid filtering screening condition, which is composed of a temperature gradient amplitude deviation threshold and a gradient direction standard deviation deviation threshold.

[0055] Specifically, specific conditions for screening and filtering grid data are set in advance, which are mainly composed of two key parameters. One is the temperature gradient amplitude deviation threshold, which is used to measure the allowed range of temperature change amplitude between adjacent grids. By setting upper and lower thresholds, it is judged whether the temperature gradient amplitude between grids is within the normal range. If it exceeds the threshold, it indicates that the temperature change in this area is abnormal. The second is the gradient direction standard deviation threshold, which is used to evaluate the dispersion degree of the temperature gradient direction of the grid. The standard deviation reflects the fluctuation of the gradient direction. By setting the corresponding deviation threshold, grids with excessive or abnormal temperature gradient direction fluctuations can be screened out. The grid filtering and screening conditions composed of the two parameters lay the foundation for subsequent accurate identification of key heat source areas of the motorized spindle and elimination of interference data.

[0056] In one possible implementation manner, step S400 further includes:

[0057] Step S410: Extracting the three-dimensional boundary vertex coordinates of the spatial thermal error area to construct a spatial grid interval.

[0058] Step S420: According to the spatial grid interval, match the backtracking time window according to the spatial proportion of the motorized spindle.

[0059] Step S430: According to the device structure of the motorized spindle, construct M component grid intervals of M constituent components in the motorized spindle.

[0060] Step S440: According to the coverage proportion characteristics of the spatial grid interval to the M component grid intervals, screen the first N constituent components.

[0061] Step S450: Take the backtracking time window and N constituent components as dual spindle operation condition constraints, call the operation log backtracking, and obtain the runtime sequence data.

[0062] Specifically, for the generated closed continuous space thermal error region, its boundary features are extracted through three-dimensional space discretization processing: first, the thermal error region is octree decomposed, and the irregular three-dimensional region is recursively divided into smaller cubic units until the temperature gradient change in each unit is less than a preset threshold (such as 0.5°C / mm); then the coordinates of the cubic unit vertices at the region boundary are extracted, which constitute the discretized three-dimensional boundary of the thermal error region. Next, the space grid interval is constructed based on these boundary vertex coordinates: the maximum and minimum values of the vertex coordinates are determined along the X, Y, and Z coordinate axes respectively, forming the six faces of the space cube; in each coordinate axis direction, equidistant grid lines are generated according to a preset step size (such as 1 mm), and the distance between adjacent grid lines constitutes the grid cell size; finally, a three-dimensional orthogonal grid interval covering the entire thermal error region is formed, which not only completely contains the spatial range of the thermal error region, but also controls the calculation accuracy of subsequent analysis through the grid cell size, providing a standardized geometric model for subsequent space ratio calculation and component matching.

[0063] After obtaining the constructed space grid interval, first calculate the ratio of the grid interval in the overall three-dimensional space of the motorized spindle, by dividing the volume of the space grid interval by the volume of the three-dimensional space occupied by the motorized spindle, to obtain the space ratio value. According to the preset ratio threshold rule, match the backtracking time window: if the space ratio exceeds a higher threshold (such as 30%), it indicates that the thermal error problem of the motorized spindle is relatively serious and the influence range is wide, at this time a longer backtracking time window is matched, for example, the running data of the past 24 hours is matched, in order to comprehensively analyze the correlation between the motorized spindle running condition and the thermal error generated in a long time; if the space ratio is lower than a lower threshold (such as 10%), it indicates that the thermal error impact is small, and a shorter backtracking time window is matched, such as the running data of the past 1 hour, so as to quickly locate the factors causing the thermal error in the recent period. In this way, according to the space ratio of the space thermal error region, the appropriate backtracking time window is matched accurately, ensuring that the obtained running data can meet the analysis requirements, avoiding data redundancy, and improving the analysis efficiency.

[0064] Based on the CAD model or physical structure parameters of the motorized spindle, it is decomposed into M constituent components (such as bearings, stators, rotors, cooling systems, etc.), and an independent three-dimensional grid interval is constructed for each component. First, according to the geometric characteristics (such as cylinder, cuboid) and spatial position relationship of the components, the boundary parameters (such as center coordinates, radius, length) of each component are defined in the global coordinate system. Then, spatial discretization is performed on each component: for regular geometric bodies (such as bearing inner ring), structured grid division is adopted, and uniform grids are generated along the axial, radial, and circumferential directions with a preset step size (such as 0.5 mm); for irregular components (such as cooling pipes), unstructured tetrahedral grid division is adopted to ensure that the grid accurately fits the actual outline of the component. Each component grid interval contains spatial position information (such as coordinate origin offset) and topological relationship (such as adjacent grid connection method), and is associated with the physical component through a unique identifier. The M component grid intervals formed ultimately maintain spatial independence and establish relative position relationships through the global coordinate system, providing an accurate geometric model basis for subsequent analysis of the influence of thermal error regions on each component.

[0065] On the basis of having constructed the spatial grid interval and the M component grid intervals, the coverage ratio of the spatial grid interval to each component grid interval is calculated one by one. Through three-dimensional spatial geometric algorithms, the coincidence of grid cells in the spatial grid interval with grid cells in each component grid interval is judged, and the ratio of the number of coincident grid cells to the total number of grid cells in the component grid interval is the coverage ratio of the component. For example, if a component grid interval has a total of 100 grid cells, of which 60 coincide with the spatial grid interval, then its coverage ratio is 60%. According to the coverage ratio from high to low, the top N constituent components with the highest coverage ratio are selected. These selected components mean that they have a higher degree of coincidence with the spatial thermal error region and are more likely to be affected by the spatial thermal error, and are potential key factors leading to thermal errors of the motorized spindle, which will be the focus of in-depth analysis and research in the future.

[0066] With the determined backtracking time window and the screened N constituent components as double constraint conditions, the electric spindle operation log database is called back accurately. First, the time range of data retrieval is determined according to the backtracking time window, such as the operation records in the past 24 hours or 1 hour; at the same time, the operation data related to the key components are screened out according to the unique identifiers of the N constituent components, including but not limited to bearing temperature, motor power, spindle speed and other monitoring indicators. Through this double screening mechanism, redundant data unrelated to thermal error analysis is eliminated, and only the operation records related to key components in a specific time range are extracted, and the operation time series data are arranged in chronological order. These data record the working condition changes of key components within the backtracking time window, providing strong and accurate data support for subsequent in-depth analysis of the causes of thermal error, establishment of thermal error prediction model and development of compensation strategy.

[0067] In one possible implementation manner, step S500 further includes:

[0068] Step S510: interactively obtaining M sets of component operation condition baselines of the M constituent components under multiple spindle operation conditions.

[0069] Step S520: time series image processing of the operation time series data to obtain N component operation condition curves of the N constituent components.

[0070] Step S530: calling N component operation condition baselines from the M sets of component operation condition baselines according to the condition attribution of the operation time series data and the N constituent components.

[0071] Step S540: change deviation analysis on the N component operation condition baselines and N component operation condition curves, and outputting the error source type characteristics.

[0072] Specifically, by monitoring the system operation interface, interact with the motorized spindle test platform, in the no-load, light load cutting, heavy load machining and other typical operating conditions, the M components of the motorized spindle (such as bearings, motors, spindle sleeves, etc.) are systematically tested. During the operation of each condition, the temperature, vibration, and current of each component are synchronously collected, and the high-precision sensor is used to capture the parameter change data over time in real time. For example, the infrared thermal imager is used to monitor the surface temperature distribution of the component, the vibration sensor is used to obtain the vibration amplitude and frequency in different directions, and the current transformer is used to collect the motor current fluctuation. The collected data is transmitted to the data analysis system in real time, and the temperature, vibration, and current data of each component at the same time are correlated and integrated, with time as the horizontal axis, to draw three-dimensional coupling curves, and the dynamic changes and mutual influence of the three parameters are intuitively presented. For each component under different conditions, the above data collection and curve drawing work is completed, and finally M sets of temperature-vibration-current coupling curves of component operating conditions are formed, covering M components and multiple conditions, providing multi-dimensional reference standards for subsequent accurate judgment of component operating state.

[0073] With time as the horizontal coordinate, the temperature, vibration, current and other performance parameters related to N key components (such as bearings, motors, spindles, etc.) in the running time sequence data are plotted as vertical coordinates. For example, for the bearing component, the temperature values monitored at different time points are connected in sequence to form a curve of the bearing temperature change over time; for the motor component, the current parameter fluctuation over time is plotted into a current curve. Through the smoothing filter algorithm, the original data is processed to remove noise interference, so that the curve can more clearly and accurately reflect the change trend of the performance parameters of each component. Finally, the corresponding operating condition curve is generated for each key component, and the operating condition evolution of N components during operation is intuitively presented in a visual manner, which is convenient for subsequent comparison and analysis with the standard operating condition baseline.

[0074] The acquired runtime sequence data corresponding to the working condition type is identified, such as the working condition attribution information of no-load, light-load cutting, heavy-load machining and the like. Meanwhile, combined with the screened N key constituent components, accurate retrieval is performed in the M groups of component operation working condition baseline databases established in step S510. Each group of baselines in the database is marked with applicable working conditions and corresponding component identifiers, and according to the working condition label of the runtime sequence data, all component operation working condition baselines under the working condition are quickly located; then according to the unique identifiers of the N key constituent components, the N component operation working condition baselines matched therewith are screened from the located baselines. For example, if the runtime sequence data shows that the current working condition is heavy-load machining, then in the baseline set of heavy-load machining, the baseline data corresponding to the N key components (such as main shaft bearing, motor rotor, cooling pump, etc.) is extracted. Through this double matching mechanism, it is ensured that the called baseline data is highly consistent with the actual operation working condition and key components, and a reliable reference standard is provided for subsequent deviation analysis.

[0075] The called N component operation working condition baselines and the obtained N component operation working condition curves are compared and analyzed in all directions. By calculating the difference between the actual working condition curve parameter value and the baseline parameter value at each corresponding time node, quantitative deviation data is obtained, such as the temperature difference between the actual value and the baseline value of the bearing temperature, the deviation degree between the actual fluctuation range and the baseline range of the motor current, etc. At the same time, by using a trend analysis algorithm, the change trend of the working condition curve and the baseline curve is compared and judged whether the parameter change is linear deviation, periodic fluctuation anomaly or sudden change. Combined with the preset error determination rule, when abnormal conditions such as temperature deviation exceeding 5℃ and vibration amplitude deviation exceeding 20% occur, the error source type characteristics report containing the component name, deviation parameter, deviation degree, change trend and possible cause is analyzed by comprehensively analyzing the deviation value, change trend and component characteristics. For example, if the bearing temperature is continuously higher than the baseline and the vibration amplitude is simultaneously increased, combined with the stable motor current, it can be determined that the error source is bearing lubrication failure or assembly abnormality; if the motor current fluctuates greatly and the temperature changes little, it may indicate motor winding failure. Finally, the error source type characteristics report is output, which provides clear guidance for subsequent targeted solution of the electric spindle thermal error problem.

[0076] In one possible implementation manner, step S540 further includes:

[0077] Step S541: If the change deviations of the N component operation working condition baselines and the N component operation working condition curves all satisfy the preset deviation scale, it is determined that the error source type characteristics is systematic thermal load accumulation.

[0078] Step S542: If the first change deviation of the first component operation working condition baseline and the first component operation working condition curve does not satisfy the preset deviation scale, it is determined that the error source type characteristics is component functional degradation.

[0079] Step S543: match a first degradation severity level according to the first change deviation, and identify the error source type feature by using the first component and the first degradation severity level.

[0080] Specifically, the N-component operating condition baseline of the call is compared with the obtained N-component operating condition curve. By calculating the deviation value of the actual operating condition curve and the baseline at the corresponding time node on the key performance parameters such as temperature, vibration, current, etc., and comparing these deviation values with the preset deviation scale one by one. For example, the preset temperature deviation scale is ±3℃, the vibration amplitude deviation scale is ±15%, and the current deviation scale is ±10%. If the change deviation of N components in all monitoring parameters is within the preset scale range of each component, and none of the components appears significant abnormality, it is determined that the current thermal error of the electric spindle is caused by systematic thermal load accumulation. This indicates that during the operation of the electric spindle, due to the influence of overall working condition, environment, etc., heat is continuously accumulated among components and cannot be effectively dissipated, eventually leading to thermal error, rather than the failure of a single component.

[0081] When comparing the N-component operating condition baseline with the operating condition curve, if it is found that the actual operating curve of a certain component (i.e. the first component) has significant deviation from the baseline on key parameters such as temperature, vibration, current, etc., and the deviation exceeds the preset normal fluctuation range (for example, the temperature deviation exceeds ±5℃, and the vibration amplitude increases by 30%), it is determined that the error source type feature is component function degradation. This degradation may manifest as lubrication performance degradation of bearings due to long-term friction, abnormal current fluctuation caused by motor winding aging, or deformation of spindle sleeve due to thermal fatigue, etc. By identifying the abnormal change of this single component, the error source is located to the specific functionally degraded component, rather than the overall system problem, providing accurate basis for subsequent targeted maintenance or adjustment.

[0082] When it is determined that the first component has functional degradation, the first change deviation between the first component operating condition curve and the baseline is matched with the corresponding first degradation severity level according to the specific value, change trend, etc. of the first change deviation. Different deviation intervals and degradation severity levels are preset to correspond, for example, if the temperature deviation is 5℃-10℃ and the vibration amplitude deviation is 30%-50%, it is determined to be mild degradation; if the temperature deviation exceeds 10℃ and the vibration amplitude deviation is greater than 50%, it is determined to be severe degradation. After determining the first degradation severity level, the name of the first component with functional degradation and the first degradation severity level are combined to clearly identify the error source type feature. For example, when it is determined that the first degradation severity level of the spindle bearing is severe, the error source type feature of the spindle bearing-severe degradation is finally output, so that maintenance personnel and technicians can intuitively understand the faulty component and its damage degree, and thus quickly develop corresponding repair schemes or compensation strategies.

[0083] In one possible implementation manner, the step S600 further includes:

[0084] The step S610: based on the error source type, the historical thermal error compensation data is aggregated, and by extracting high-frequency effective strategies, a plurality of sample compensation priorities of a plurality of sample error source types are obtained.

[0085] The step S620: the plurality of sample compensation priorities of the plurality of sample error source types are associated and stored, and a thermal error compensation strategy library is constructed.

[0086] The step S630: the error source type characteristics are used to traverse the thermal error compensation strategy library, and the real-time compensation strategy priority is matched and output.

[0087] The step S640: the real-time compensation strategy priority is decomposed, and a hard constraint priority and a soft constraint priority are output.

[0088] The step S650: the hard constraint priority is used as a boundary condition, a compensation parameter with the lowest energy consumption that meets a lower limit of a thermal error suppression rate is solved as a hard constraint compensation parameter, and a residual thermal error margin is calculated.

[0089] The step S660: if the residual thermal error margin exceeds a preset tolerance range, the soft constraint priority is used as an optimization target, and a soft constraint compensation parameter that offsets the residual thermal error margin within a remaining resource capacity is solved.

[0090] The step S670: the hard constraint compensation parameter and the soft constraint compensation parameter are sequentially connected, and the collaborative compensation parameter is output.

[0091] Specifically, according to the output error source type characteristics (such as bearing-severe degradation, systematic thermal load accumulation, etc.), the thermal error compensation data in the historical database is classified and aggregated. The compensation cases of the same error source type are collected into a group, for example, all compensation records caused by bearing-severe degradation are integrated together. Then, each group of compensation data is analyzed in depth, and the use frequency and effectiveness of each compensation strategy (such as replacing bearings, adjusting cooling flow, optimizing lubrication, etc.) are counted. By calculating the success rate of the strategy in solving a specific error source problem, the high-frequency effective strategy is identified. For example, in the case of bearing-severe degradation, if the effectiveness frequency of the bearing replacement strategy reaches 80%, and the effectiveness frequency of the optimized lubrication strategy is only 30%, the former is identified as a high-frequency effective strategy. Finally, according to the effectiveness frequency, compensation efficiency and implementation cost and other multi-dimensional indexes of the strategy, a plurality of sample compensation priorities are determined for each sample error source type, and a priority sequence such as replacing bearings > adjusting cooling flow > optimizing lubrication is formed, which provides data support for subsequent construction of a compensation strategy library.

[0092] After obtaining the determined multiple sample error source types and their corresponding multiple sample compensation priorities, a structured data storage method is used to associate and integrate the error source types, compensation strategies and priorities. Taking the error source type as the index primary key, the high-frequency effective compensation strategies and their priorities under each error source type are stored in the database in the form of key-value pairs. For example, for the error source type of bearing-severe degradation, the strategies of replacing the bearing (high priority) and optimizing lubrication (low priority) are associated and stored. At the same time, in order to enable the strategy library to have efficient retrieval capability, a similarity calculation algorithm based on semantics and feature vectors is introduced, and the feature description of each error source type (such as component name, degradation degree, parameter deviation, etc.) is vectorized to construct an error source feature vector space. When the user inputs the feature of a new error source type, the cosine similarity between the feature vectors is calculated to quickly retrieve the most similar historical error source type and the corresponding compensation strategy priority, thereby providing quick and accurate compensation strategy reference for different but similar thermal error problems, and finally forming a functional complete and efficient thermal error compensation strategy library.

[0093] The derived error source type feature (such as main shaft bearing-severe degradation) is used as a retrieval keyword to perform a comprehensive traversal of the constructed thermal error compensation strategy library. By comparing the error source type feature with the various sample error source types stored in the strategy library, the record with the highest similarity or complete match is found. If there is a completely matched error source type, the pre-set sample compensation priority of this type is directly retrieved, for example, for main shaft bearing-severe degradation, the compensation strategy priority sequence of replacing the bearing (high priority) and optimizing lubrication (low priority) is directly obtained. If no complete match is found, the similarity between the input feature and the error source types in the library is calculated based on the error source feature similarity retrieval function supported by the strategy library, and the error source type with the highest similarity and its corresponding compensation strategy priority are selected as the reference. Finally, combined with the real-time running parameters, working conditions and other information of the current electric spindle, the retrieved compensation strategy priority is adaptively adjusted, and the real-time compensation strategy priority suitable for the current situation is output, providing clear strategy execution order guidance for subsequent thermal error compensation.

[0094] The priority of the real-time compensation strategy for the output is analyzed in depth, and according to the characteristics and execution requirements of the compensation strategy, it is finely divided into hard constraint priority and soft constraint priority. In terms of hard constraint priority, the adjustment amount of the strategy parameter of priority 1 should not be lower than 80% of the historical successful case, such as the cooling flow increase should be ≥ benchmark value × 0.8, which ensures that the key compensation strategy can effectively play a role; At the same time, set the lower limit of thermal error suppression rate (≥ 85%), to ensure that the compensation effect meets the core index requirements. In terms of soft constraint priority, it is stipulated that the strategy parameter of priority 2 is executed when the resources allow, such as the number of vibration suppression opening times ≤ 3 times / minute, and the upper limit of compensation resource occupation is clearly defined, such as the remaining pump power ≥ 20% of the rated value, and the vibration suppression channel occupation ≤ 2, which realizes the compensation target while reasonably allocating equipment resources, and clearly defines the constraint conditions at different levels.

[0095] The hard constraint priority is determined as the key boundary condition. Based on the thermal error model and energy consumption model, the lower limit of the compensation parameter adjustment amount stipulated by the hard constraint priority (such as cooling flow increase ≥ 80% of the historical average), the lower limit of the thermal error suppression rate (≥ 85%) and other requirements are input as constraint conditions. Through optimization algorithm (such as genetic algorithm), search and calculation are carried out in the parameter feasible region. Among the many compensation parameter combinations that meet the lower limit of the thermal error suppression rate, the one with the lowest energy consumption is selected as the hard constraint compensation parameter. For example, when adjusting and optimizing the related parameters of the cooling system and the lubrication system, the influence of each parameter on the thermal error suppression and energy consumption is considered comprehensively. After the hard constraint compensation parameter is determined, it is applied to the electric spindle thermal error model to calculate the remaining thermal error margin after executing this group of parameters, so as to evaluate the suppression effect of the hard constraint compensation parameter on the thermal error, and provide a basis for whether further compensation is needed.

[0096] When the calculated residual thermal error margin exceeds the pre-set tolerance range, further operation is carried out according to the determined soft constraint priority. At this time, the upper limit of the compensation resource occupation stipulated in the soft constraint priority (such as the remaining pump power ≥ 20% of the rated value, and the vibration suppression channel occupation ≤ 2) and other conditions are used as restrictions, combined with the current remaining resource capacity of the equipment, such as the remaining cooling capacity, power adjustment range, etc. Optimization algorithm (such as linear programming, dynamic programming algorithm, etc.) is used to search in the range allowed by the resources to find a compensation parameter combination that can offset the residual thermal error. For example, by adjusting the number of vibration suppression opening times (≤ 3 times / minute), optimizing the related parameters of the soft constraint strategy such as lubrication strategy, etc., the residual thermal error is reduced as much as possible without exceeding the upper limit of resource occupation, and the appropriate soft constraint compensation parameter is determined to further improve the thermal error compensation effect.

[0097] The obtained hard constraint compensation parameters are integrated with the obtained soft constraint compensation parameters. According to the priority order of the compensation strategy, the execution order and specific values of the hard constraint compensation parameters are determined first, because the hard constraint compensation parameters play a key decisive role in thermal error suppression and need to be executed first. Then, the soft constraint compensation parameters are sequentially connected after the hard constraint compensation parameters according to their priority and actual working condition requirements. For example, key operations such as bearing replacement (hard constraint compensation parameters) are performed first, and then auxiliary operations such as optimized lubrication (soft constraint compensation parameters) are performed according to the remaining thermal error and resource conditions. Through this sequential connection, a complete set of collaborative compensation parameters is formed, covering a comprehensive compensation scheme from key compensation to auxiliary optimization. Finally, the collaborative compensation parameters are output, providing systematic and effective parameter support for the thermal error compensation of the motorized spindle, ensuring that the thermal error suppression requirements are met while achieving rational use of resources and optimization of compensation effect.

[0098] In one possible implementation manner, the step S700 further includes:

[0099] Step S710: Taking the geometric center points of the plurality of initial spatial temperature grids as spatial representative voxels, a plurality of initial grid temperature values are extracted.

[0100] Step S720: Performing adjacent grid temperature gradient vector calculation on the plurality of initial spatial temperature grids according to the plurality of initial grid temperature values, and outputting a plurality of initial temperature gradient amplitudes and a plurality of initial gradient direction standard deviations.

[0101] Step S730: Filtering and outputting P initial spatial temperature grids from the plurality of initial spatial temperature grids according to the plurality of initial temperature gradient amplitudes and the plurality of gradient direction standard deviations.

[0102] Specifically, for the plurality of initial spatial temperature grids obtained by division, the geometric center point of each grid is selected as a spatial representative voxel. Based on a three-dimensional spatial coordinate system, the center point position is accurately determined by calculating the mean value of the boundary coordinates of each dimension of the grid (for example, in a rectangular coordinate system, the center point coordinates are one-half of the sum of the minimum and maximum values of the axial, radial, and circumferential boundary coordinates). Then, the temperature value corresponding to the center point is extracted, which can be an interpolated temperature at the center point position or a weighted average of all temperature data in the grid. In this way, a representative temperature value is obtained for each initial spatial temperature grid, forming a data set containing a plurality of initial grid temperature values, and providing an accurate and spatially representative data basis for subsequent calculation of temperature gradient vectors between adjacent grids.

[0103] Based on the obtained temperature values of each initial spatial temperature grid, for each grid, the temperature gradient vector between it and its adjacent grids is calculated. First, determine the adjacent grids of each grid in three-dimensional space (usually 6 grids directly connected along the X, Y, Z three directions). For each pair of adjacent grids, calculate their temperature difference and divide by the distance between the two center points to get the size of the temperature gradient (i.e. the temperature gradient amplitude), which reflects the intensity of temperature change per unit distance. At the same time, determine the direction of the temperature gradient, that is, the direction from the grid with lower temperature to the grid with higher temperature. For each grid, count the temperature gradient directions of all adjacent grids, and calculate the standard deviation of these directions (i.e. the gradient direction standard deviation), which is used to measure the consistency of the heat flow direction: the larger the standard deviation, the more dispersed the heat flow direction, indicating complex heat conduction paths or heat source distribution. Finally, output two key indicators for each initial grid: temperature gradient amplitude and gradient direction standard deviation, which together constitute a quantitative description of the heat conduction characteristics, used in subsequent steps to identify areas where thermal errors may exist.

[0104] Based on the calculated temperature gradient amplitude and gradient direction standard deviation of each initial spatial temperature grid, filtering is performed by setting double thresholds. First, preset the temperature gradient amplitude threshold (e.g. 3℃ / mm) and the gradient direction standard deviation threshold (e.g. 15°), if the temperature gradient amplitude of a certain grid is greater than or equal to the amplitude threshold, and its gradient direction standard deviation is greater than or equal to the direction standard deviation threshold, then it is determined that the grid is a thermal error sensitive area and is retained; otherwise, if either indicator does not meet the threshold, it is considered that the grid has a gentle temperature change or the heat flow direction is relatively consistent, and it is a low-risk area that is filtered out. Through this screening mechanism, P grids that simultaneously satisfy high temperature gradient amplitude and high direction dispersion are accurately identified from all initial spatial temperature grids, which represent areas with extremely abnormal heat conduction and complex direction inside the motorized spindle, serving as the focus of subsequent detailed analysis, effectively reducing the amount of redundant data processing, and improving the efficiency and accuracy of thermal error area positioning.

[0105] In one possible implementation, step S700 further includes:

[0106] Step S740: decompose the cooperative compensation parameter to obtain a plurality of thermal error compensation control parameters, wherein the thermal error compensation control parameter has an execution priority and an execution time limit identifier.

[0107] Step S750: assign a plurality of PID adjustment weights to the plurality of thermal error compensation control parameters according to the execution priority and the execution time limit identifier.

[0108] Step S760: convert the plurality of thermal error compensation control parameters into a plurality of PID thermal error input variables.

[0109] Step S770: According to the PID adjustment weight of the plurality of PIDs, the PID adjustment module is controlled to respond to the plurality of PID thermal error input variables for real-time thermal error dynamic compensation of the electric spindle.

[0110] Specifically, the output cooperative compensation parameter is systematically disassembled and refined into a plurality of specific thermal error compensation control parameters, such as temperature adjustment parameters of the cooling system, power correction parameters of the motor, and speed compensation parameters of the spindle. Based on the key degree and response timeliness demand of each parameter to the thermal error suppression of the electric spindle, each thermal error compensation control parameter is given a dual attribute identification: on the one hand, according to the urgency and importance of the parameter to the thermal error suppression effect, the execution priority is divided, such as setting the parameter involving direct cooling as high priority and setting the auxiliary power fine-tuning as low priority; on the other hand, according to the working condition demand and compensation mechanism, the execution time limit is set to clearly indicate whether the parameter needs to be executed immediately to quickly respond to sudden thermal error or can be delayed and executed at a specific working condition conversion, thereby laying a foundation for the orderly implementation of subsequent compensation strategies.

[0111] According to the execution priority and execution time limit identification of the thermal error compensation control parameter, each parameter is precisely allocated a PID adjustment weight. For control parameters with high execution priority and immediate execution, such as flow regulation parameters of the cooling system, since they can quickly suppress the temperature rise of key parts of the electric spindle and have a significant impact on thermal error compensation effect, they are given a higher PID adjustment weight (e.g. 70% of the total weight); while parameters with lower execution priority and relatively loose execution time limit, such as motor power fine-tuning parameters, are allocated a lower weight (e.g. 30% of the total weight). The execution time limit identification also affects the weight allocation, for parameters requiring immediate execution, the weight proportion in PID adjustment will be further increased to ensure their priority in action; for parameters that can be delayed for execution, the weight proportion will be correspondingly reduced. In this way, the differential management of different thermal error compensation control parameters in the PID adjustment process is realized, ensuring the accuracy and timeliness of the thermal error compensation of the electric spindle.

[0112] To enable the thermal error compensation control parameters to be recognized and processed by the PID adjustment module, it is necessary to convert them into PID thermal error input variables. The specific conversion process is based on the mapping of parameter physical meaning and PID control logic: for parameters with clear physical dimensions, such as cooling system flow regulation parameters, they are converted into temperature deviation variables (ΔT) through a thermal resistance model; motor power correction parameters are converted into speed deviation variables (Δn) according to the spindle thermal characteristic curve. For execution priority and execution time limit identification, they are mapped into weight coefficients and time lag parameters of PID input variables, respectively, with high priority parameters corresponding to larger weight coefficients and parameters that need to be executed immediately set as non-time-lag inputs. Through this conversion, the control parameters containing compensation strategies, priorities and time limit information are converted into error variables (such as temperature deviation and speed deviation) and their weight configurations that can be directly processed by the PID adjustment module, ensuring the effectiveness and real-time performance of subsequent PID control.

[0113] Based on the determined multiple PID adjustment weights, the PID adjustment module drives real-time response to the multiple PID thermal error input variables converted. In the running process, the PID adjustment module multiplies each input variable by the corresponding weight, and preferentially processes high-weight input variables, for example, for high-weight variables such as cooling system temperature deviation, the adjustment module will quickly output a larger control amount to drive the cooling system to increase the cooling liquid flow or adjust the cooling intensity; for low-weight motor power fine-tuning variables, a smaller control amount is output for progressive adjustment. At the same time, the PID adjustment module continuously monitors the thermal error changes of the electric spindle, and continuously optimizes the adjustment weights of each variable according to the feedback data, dynamically adjusts the control output, forms a closed-loop control mechanism, and realizes precise and efficient compensation of the thermal error of the electric spindle, ensuring that the electric spindle always maintains a high-precision operating state during the machining process.

[0114] Embodiment two, based on the same inventive concept as the thermal error compensation method of one of the foregoing embodiments, as shown in Figure 2 The present application provides a thermal error compensation system for an electric spindle, and the system and method embodiments in the present application are based on the same inventive concept. Among them, the system comprises:

[0115] A temperature sensor array construction module 10 is used to construct a temperature sensor array covering the global heat source area of the electric spindle by arranging multiple types of temperature sensors on the electric spindle.

[0116] A spindle three-dimensional temperature cloud map generation module 20 is used to generate a spindle three-dimensional temperature cloud map through a Kriging spatial interpolation algorithm according to the real-time monitoring data returned by the temperature sensor array.

[0117] A spatial thermal error area identification module 30 is used to identify a spatial thermal error area with a temperature gradient exceeding a preset threshold in the spindle three-dimensional temperature cloud map.

[0118] The runtime sequence data calling module 40 is configured to call the runtime sequence data of the spindle from the spindle log call to meet the spindle operation condition as a spatial constraint of the spatial boundary of the spatial thermal error area.

[0119] The error source type feature acquisition module 50 is configured to trace the thermal error according to the runtime sequence data to obtain the error source type feature.

[0120] The collaborative compensation parameter output module 60 is configured to perform thermal error compensation optimization with the real-time compensation strategy priority as a constraint after matching the compensation strategy priority according to the error source type feature, and output the collaborative compensation parameter.

[0121] The real-time thermal error dynamic compensation module 70 is configured to use the collaborative compensation parameter to perform real-time thermal error dynamic compensation on the spindle by using the PID adjustment module of the spindle.

[0122] Further, the system is also used for the following functions:

[0123] Based on the spindle geometric size of the spindle, an initial grid side length is set, and the spindle three-dimensional temperature cloud space is divided into a plurality of initial space temperature grids using the initial grid side length. The adjacent grid temperature gradient vector consistency analysis is performed on the plurality of initial space temperature grids to filter and screen out P initial space temperature grids. The first update grid side length is set as 1 / K of the initial grid side length, and after the P initial space temperature grids are subdivided, the adjacent sub-grid temperature gradient vector consistency analysis is performed to locate F first update space temperature grids. In this way, the spindle three-dimensional temperature cloud is refined and segmented layer by layer and the grid is screened for a preset number of iterations to locate a plurality of thermal error sub-grids. The three-dimensional morphological closing operation is performed on the plurality of thermal error sub-grids to generate a closed continuous spatial thermal error area.

[0124] Further, the system is also used for the following functions:

[0125] The geometric center points of the plurality of initial space temperature grids are taken as space representative voxels to extract a plurality of initial grid temperature values. The adjacent grid temperature gradient vector calculation is performed on the plurality of initial space temperature grids according to the plurality of initial grid temperature values to output a plurality of initial temperature gradient amplitudes and a plurality of initial gradient direction standard deviations. The P initial space temperature grids are filtered and output from the plurality of initial space temperature grids according to the plurality of initial temperature gradient amplitudes and the plurality of gradient direction standard deviations.

[0126] Further, the system is also used for the following functions:

[0127] Decompose the cooperative compensation parameter to obtain a plurality of thermal error compensation control parameters, wherein the thermal error compensation control parameters have execution priority and execution time limit identification; distribute a plurality of PID adjustment weights to the plurality of thermal error compensation control parameters according to the execution priority and the execution time limit identification; convert the plurality of thermal error compensation control parameters into a plurality of PID thermal error input variables; control the PID adjustment module to respond to the plurality of PID thermal error input variables for real-time thermal error dynamic compensation of the electric spindle according to the plurality of PID adjustment weights.

[0128] Further, the system is also used for the following functions:

[0129] Extract the three-dimensional boundary vertex coordinates of the spatial thermal error area to construct a spatial grid interval; match a backtracking time window according to the spatial proportion of the spatial grid interval in the electric spindle; construct M component grid intervals of M constituent components in the electric spindle according to the device structure of the electric spindle; filter the first N constituent components according to the coverage proportion characteristics of the spatial grid interval on the M component grid intervals; take the backtracking time window and the N constituent components as dual spindle operating condition constraints to perform operation log backtracking calling to obtain the runtime sequence data.

[0130] Further, the system is also used for the following functions:

[0131] Interactively obtain M sets of component operating condition baselines of the M constituent components under a plurality of spindle operating conditions; time-series image process the runtime sequence data to obtain N component operating condition curves of the N constituent components; call N component operating condition baselines from the M sets of component operating condition baselines according to the operating condition attribution of the runtime sequence data and the N constituent components; perform change deviation analysis on the N component operating condition baselines and the N component operating condition curves to output the error source type characteristics.

[0132] Further, the system is also used for the following functions:

[0133] If the change deviations of the N component operating condition baselines and the N component operating condition curves all satisfy a preset deviation scale, it is determined that the error source type characteristics are systematic thermal load accumulation; if a first change deviation of a first component operating condition baseline and a first component operating condition curve does not satisfy the preset deviation scale, it is determined that the error source type characteristics are component function degradation; match a first degradation severity level according to the first change deviation, and identify the error source type characteristics by using a first constituent component and the first degradation severity level.

[0134] Further, the system is also used for the following functions:

[0135] In the historical thermal error compensation data based on the error source type, a plurality of sample compensation priorities of a plurality of sample error source types are obtained by extracting a high-frequency effective strategy; a plurality of sample compensation priorities of the plurality of sample error source types are associated and stored to construct a thermal error compensation strategy library; the error source type characteristics are used to traverse the thermal error compensation strategy library, and the real-time compensation strategy priority is matched and output; the real-time compensation strategy priority is decomposed to output a hard constraint priority and a soft constraint priority; the hard constraint priority is used as a boundary condition, the compensation parameter with the lowest energy consumption that meets the lower limit of the thermal error suppression rate is solved as a hard constraint compensation parameter, and the residual thermal error margin is calculated; if the residual thermal error margin exceeds the preset tolerance range, the soft constraint compensation parameter that offsets the residual thermal error margin is solved in the remaining resource capacity by taking the soft constraint priority as an optimization target; the hard constraint compensation parameter and the soft constraint compensation parameter are sequentially connected to output the collaborative compensation parameter.

[0136] Further, the system is also used for the following functions:

[0137] A predefined grid filtering screening condition is composed of a temperature gradient amplitude deviation threshold and a gradient direction standard deviation deviation threshold.

[0138] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes a specific embodiment of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0139] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0140] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalent technology, the present application intends to include these modifications and changes.

Claims

1. A thermal error compensation method of an electric spindle, characterized by, The method comprises: By arranging multiple types of temperature sensors on the electric spindle, a temperature sensor array covering the global heat source area of the electric spindle is constructed; According to the real-time monitoring data returned by the temperature sensor array, a spindle three-dimensional temperature cloud map is generated by Kriging spatial interpolation algorithm; Identify the spatial thermal error area with temperature gradient exceeding the preset threshold in the spindle three-dimensional temperature cloud map; Take the spatial boundary of the spatial thermal error area as the spatial constraint of the operation log, and call the runtime sequence data of the electric spindle that meets the spindle operation condition from the electric spindle log; According to the running time sequence data, the error source type characteristics are obtained by tracing the error source; After matching the compensation strategy priority according to the error source type characteristics, the real-time compensation strategy priority is taken as the constraint to perform thermal error compensation optimization, and the collaborative compensation parameters are output; Run the PID adjustment module of the electric spindle with the collaborative compensation parameters to dynamically compensate the real-time thermal error of the electric spindle.

2. A thermal error compensation method of an electric spindle as claimed in claim 1, characterized in that, In the spindle three-dimensional temperature cloud map, identify the spatial thermal error area with temperature gradient exceeding the preset threshold, the method comprises: Based on the spindle geometric size of the electric spindle, set the initial grid length; Divide the spindle three-dimensional temperature cloud space into multiple initial space temperature grids using the initial grid length; Perform adjacent grid temperature gradient vector consistency analysis on the multiple initial space temperature grids to filter out P initial space temperature grids; Take 1 / K of the initial grid length as the first updated grid length, and after secondary subdivision of the P initial space temperature grids, perform adjacent sub-grid temperature gradient vector consistency analysis to locate F first updated space temperature grids; By analogy, the spindle three-dimensional temperature cloud map is segmented and screened layer by layer by preset iteration times, and multiple thermal error sub-grids are located; Perform three-dimensional morphological closing operation on the multiple thermal error sub-grids to generate a closed continuous spatial thermal error area.

3. A thermal error compensation method of an electric spindle as claimed in claim 2, characterized in that, Perform adjacent grid temperature gradient vector consistency analysis on the multiple initial space temperature grids to filter out P initial space temperature grids, the method comprises: Take the geometric center points of the multiple initial space temperature grids as the space representative voxels, and extract multiple initial grid temperature values; According to the multiple initial grid temperature values, perform adjacent grid temperature gradient vector calculation on the multiple initial space temperature grids, output multiple initial temperature gradient amplitudes and multiple initial gradient direction standard deviations; According to the multiple initial temperature gradient amplitudes and multiple gradient direction standard deviations, filter and output P initial space temperature grids from the multiple initial space temperature grids.

4. A thermal error compensation method of an electric spindle as claimed in claim 1, characterized in that, Run the PID adjustment module of the electric spindle with the collaborative compensation parameters to dynamically compensate the real-time thermal error of the electric spindle, the method comprises: Decompose the collaborative compensation parameters to obtain multiple thermal error compensation control parameters, wherein the thermal error compensation control parameters have execution priority and execution time limit identifier; According to the execution priority and execution time limit identifier, assign multiple PID adjustment weights to the multiple thermal error compensation control parameters; convert the plurality of thermal error compensation control parameters into a plurality of PID thermal error input variables; control the PID adjustment module to dynamically compensate for the real-time thermal error of the electric spindle in response to the plurality of PID thermal error input variables according to the plurality of PID adjustment weights.

5. A thermal error compensation method of an electric spindle as claimed in claim 1, characterized in that, The method comprises: extracting the three-dimensional boundary vertex coordinates of the spatial thermal error region to construct a spatial grid interval; According to the space proportion matching of the space grid interval in the electric spindle, the backtracking time window is matched; According to the device structure of the electric spindle, M component grid intervals of M components in the electric spindle are constructed; According to the coverage ratio characteristics of the space grid interval to the M component grid intervals, the first N components are screened; The backtracking time window and the N components are used as double spindle operation condition constraints to call the operation log backtracking, and the operation time sequence data is obtained.

6. A thermal error compensation method of an electric spindle as claimed in claim 5, characterized in that, According to the operation time sequence data, the error source type characteristics are obtained, and the method comprises: interactively obtaining M sets of component operation condition baselines of the M components under multiple spindle operation conditions; time sequence image processing the operation time sequence data to obtain N component operation condition curves of the N components; According to the operation condition attribution of the operation time sequence data and the N components, N component operation condition baselines are called from the M sets of component operation condition baselines; The N component operation condition baselines and the N component operation condition curves are analyzed for change deviation, and the error source type characteristics are output.

7. A thermal error compensation method of an electric spindle as claimed in claim 6, characterized in that, The N component operation condition baselines and the N component operation condition curves are analyzed for change deviation, and the error source type characteristics are output, and the method comprises: If the change deviation of the N component operation condition baselines and the N component operation condition curves meets the preset deviation scale, it is determined that the error source type characteristics are systematic thermal load accumulation; If the first change deviation of the first component operation condition baseline and the first component operation condition curve does not meet the preset deviation scale, it is determined that the error source type characteristics are component function degradation; According to the first change deviation, a first degradation severity level is matched, and the first component and the first degradation severity level are used to identify the error source type characteristics.

8. A thermal error compensation method of an electric spindle as claimed in claim 1, characterized by, According to the error source type characteristics, the real-time compensation strategy priority is matched, and the thermal error compensation optimization is performed with the real-time compensation strategy priority as a constraint to output the cooperative compensation parameters, and the method comprises: Based on the error source type, the historical thermal error compensation data is aggregated, a high-frequency effective strategy is extracted, and a plurality of sample compensation priorities of a plurality of sample error source types are obtained; The plurality of sample compensation priorities of the plurality of sample error source types are associated and stored to construct a thermal error compensation strategy library; The error source type characteristics are used to traverse the thermal error compensation strategy library to match and output the real-time compensation strategy priority; The real-time compensation strategy priority is decomposed to output a hard constraint priority and a soft constraint priority; Solve the compensation parameter with the lowest energy consumption that meets the lower limit of the thermal error suppression rate as the hard constraint compensation parameter, and calculate the residual thermal error margin, with the hard constraint priority as the boundary condition; If the residual thermal error margin exceeds the preset tolerance range, solve the soft constraint compensation parameter that offsets the residual thermal error margin within the remaining resource capacity, with the soft constraint priority as the optimization target; Sequentially concatenate the hard constraint compensation parameter and the soft constraint compensation parameter, and output the collaborative compensation parameter.

9. A thermal error compensation method of an electric spindle as claimed in claim 3, characterized in that, Predefine a grid filtering screening condition composed of a temperature gradient amplitude deviation threshold and a gradient direction standard deviation threshold.

10. A thermal error compensation system for an electric spindle, characterized by The system is used to implement the thermal error compensation method of the electric spindle according to any one of claims 1-9, and the system comprises: A temperature sensor array construction module is configured to construct a temperature sensor array covering the global heat source area of the electric spindle by arranging multiple types of temperature sensors on the electric spindle; A spindle three-dimensional temperature cloud map generation module is configured to generate a spindle three-dimensional temperature cloud map by Kriging spatial interpolation algorithm according to the real-time monitoring data returned by the temperature sensor array; A spatial thermal error area identification module is configured to identify a spatial thermal error area with a temperature gradient exceeding a preset threshold in the spindle three-dimensional temperature cloud map; A runtime sequence data calling module is configured to call the runtime sequence data that meets the spindle operating condition from the electric spindle log, with the spatial boundary of the spatial thermal error area as the spatial constraint of the running log; An error source type feature acquisition module is configured to trace the thermal error source according to the runtime sequence data to obtain the error source type feature; A collaborative compensation parameter output module is configured to perform thermal error compensation optimization with the real-time compensation strategy priority as the constraint after matching the compensation strategy priority according to the error source type feature, and output the collaborative compensation parameter; A real-time thermal error dynamic compensation module is configured to run the PID adjustment module of the electric spindle to perform real-time thermal error dynamic compensation on the electric spindle with the collaborative compensation parameter.

Images