A method for tooth surface error online estimation and compensation for gear grinding process

Abstract

The application discloses a tooth surface error online estimation and compensation method for a gear grinding process, temperature change data of a contact area of a grinding wheel and a tooth surface is collected in a grinding process, a heat source distribution change pattern is constructed based on temperature difference direction change, a discontinuous heating block is identified, and the discontinuous heating block is associated with an offset trend of a tool axis in a grinding path, an interactive segment with heat disturbance and path change characteristics is extracted, and a difference state diagram structure is constructed. On this basis, combined with initial outer contour of a gear blank and adjacent tooth groove residual processing mark information, a tooth surface error contribution segment is identified and a segment mark index is generated; further, a control adjustment channel such as tool position compensation, cooling interval and spindle fine adjustment is linked, control instructions are combined and updated according to channel response time sequence, and continuous application across a grinding period is realized through a parameter buffer section, so that online estimation and compensation of the tooth surface error are realized.

Description

Technical Field

[0001] This invention relates to the field of gear manufacturing and intelligent control technology, specifically to a method for online estimation and compensation of tooth surface errors in the gear grinding process. Background Technology

[0002] As a core component of mechanical transmission systems, gears' tooth surface geometry accuracy and surface quality directly affect the overall machine's noise level, vibration characteristics, and service life. During the form grinding process, the grinding wheel and the workpiece tooth surface are in continuous contact and cutting action in a localized area, easily generating significant grinding heat in the contact zone. When heat is released in a concentrated manner within a short period of time or repeatedly accumulated in a localized area, it can easily cause thermal damage and thermally induced deformation of the tooth surface, leading to tooth profile errors and a decrease in meshing performance.

[0003] In existing technologies, compensation methods based on empirical models or feedback control are commonly used to address tooth surface errors generated during grinding. One type of method establishes a mapping relationship between temperature rise and thermal deformation in advance based on fixed empirical formulas, and then corrects the machining parameters accordingly. However, gear grinding paths are typically complex and discontinuous, and grinding heat sources exhibit significant non-uniform distribution characteristics in both spatial and temporal dimensions. Fixed empirical models struggle to accurately describe the heat accumulation behavior at different machining stages, thus limiting compensation accuracy under complex operating conditions.

[0004] Another approach is feedback control based on real-time temperature measurement signals, such as using infrared thermometry or thermocouples to monitor the temperature of the grinding contact area. However, in actual machining environments, coolant spraying, slag shielding, and limitations in the sensor's own response frequency often result in a large amount of noise in the measurement signal and a non-negligible time delay, making it difficult for feedback control based on single-point temperature measurement to reflect the true evolution of tooth surface thermal deformation in a timely manner.

[0005] Furthermore, in gear grinding machines, control mechanisms such as tool position compensation, coolant supply, and spindle operating parameter adjustment are typically implemented by multiple independent execution channels, each with significantly different physical response time constants. Existing control methods often assume that each control channel can respond to compensation commands synchronously, failing to adequately consider the timing differences between different channels. Under high-speed, continuous grinding conditions, the triggering time of the compensation command and the actual occurrence time of the error are difficult to match in the time dimension.

[0006] In view of this, the present invention provides an online estimation and compensation method for tooth surface error in gear grinding process, which solves the above problems. Summary of the Invention

[0007] The purpose of this invention is to provide an online estimation and compensation method for tooth surface errors in the gear grinding process, which solves the problem that tooth surface errors are difficult to identify and compensate in a timely and accurate manner due to discontinuous heat source distribution, evolution of processing path and asynchronous response of multiple control channels during gear grinding.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for online estimation and compensation of tooth surface error in gear grinding process, comprising the following steps:

[0009] S101. During gear grinding, temperature change data of the contact area between the grinding wheel and the gear surface are collected in multiple time slices. Based on the temperature difference direction change relationship of each spatial position in the temperature change data, a heat source distribution change pattern is constructed, and multiple discontinuous heating blocks are identified from the heat source distribution change pattern.

[0010] S102. The spatial offset trend of the discontinuous heating block and the tool position axis in the grinding path is analyzed to identify interactive segments that have both heat distribution change characteristics and tool position offset change characteristics. Based on the spatial adjacency relationship and temporal order relationship between the interactive segments, a difference state diagram structure containing the connection relationship of the interactive segments is constructed.

[0011] S103. Based on the difference state diagram structure, combined with the initial outer contour information of the gear blank and the residual machining mark information of adjacent tooth grooves, identify the set of error contribution segments that have a significant impact on the formation of tooth surface error, and generate a segment mark index for the set of error contribution segments.

[0012] S104. Based on the segment marker index, multiple preset control adjustment channels are invoked in a linked manner. According to the time-domain misalignment between the response order of each type of channel in the two most recent adjustments and the segment trigger sequence, the control commands corresponding to the tool position compensation, cooling interval and spindle running parameters are combined and updated. The updated control command combination is then sent to the parameter buffer segment for application in subsequent grinding cycles, thereby realizing online compensation for tooth surface errors.

[0013] As a preferred embodiment of the present invention, the process of identifying discontinuous heating blocks includes:

[0014] The heat source distribution change pattern during gear grinding is obtained, the temperature difference direction sequence in the heat source distribution change pattern is converted into a symbol change string, and the candidate boundary is determined based on the inversion position of the temperature difference direction sequence in the time dimension.

[0015] Based on the spatial relative position of each candidate boundary in the gear coordinate system, spatial grouping is performed. When the candidate boundaries in the same group form a continuous distribution structure in the gear axial or radial direction, a set of effective reversal segments of the heating block boundary is constructed, and the set of effective reversal segments is determined as a discontinuous heating block.

[0016] As a preferred embodiment of the present invention, the construction of the effective inversion segment set includes:

[0017] Bind the time position corresponding to each symbol change reversal segment to the spatial measurement point index to form a reversal event unit;

[0018] In the gear coordinate system, based on the adjacency relationship of the measuring points in the axial and radial directions, a spatial adjacency set of the reversal event unit is constructed;

[0019] In the spatial adjacency set, identify the combination of measurement points that simultaneously experience inversion events at the same or adjacent sampling times, and determine the combination of measurement points as a candidate spatial inversion cluster;

[0020] The corresponding inversion event is included in the set of valid inversion segments for subsequent segment boundary construction only when the candidate spatial inversion clusters form a continuous distribution structure in the axial or radial direction.

[0021] As a preferred embodiment of the present invention, the process of identifying interactive segments includes:

[0022] The discontinuous heating blocks are mapped to the grinding path reference trajectory, and the tool position axis offset trend of the trajectory segment in the grinding path is calculated.

[0023] When a trajectory segment is located within the coverage area of ​​a discontinuous heating block, and its offset trend changes by more than a preset threshold relative to adjacent trajectory segments, the trajectory segment is determined to be an interactive segment.

[0024] As a preferred embodiment of the present invention, the process of constructing the difference state diagram structure includes:

[0025] Each segment in the interactive segment group is mapped to the processing path reference trajectory, and the turning critical point is marked according to the change of local curvature of the trajectory. The turning critical point is used as the connection node to extract the effective connection pairs between segments and form the initial graph structure.

[0026] Directed edges are established based on the sequential relationship of the connection pairs on the time axis, and edge pairs that do not meet the boundary turning continuity rule are removed. The remaining edge pairs are constructed into a differential state graph structure, and each node is assigned a path span identifier and a segment continuous number index. All the filtered connection pairs form a directed graph structure, and each interaction segment is mapped to a graph node. The edge pairs reflect the continuous processing behavior in the processing path.

[0027] As a preferred embodiment of the present invention, the marking process of the turning critical point includes:

[0028] The position of each interactive segment on the processing reference trajectory is sampled at equal intervals. Local tangents are constructed based on adjacent sampling points, and the included angle between adjacent tangents is calculated to form a discrete tangential angle sequence arranged along the processing path.

[0029] In the discrete tangential angle sequence, candidate turning positions are identified based on the reversal relationship of the changing directions of adjacent angle values, and candidate positions with angle changes below a preset threshold are eliminated.

[0030] The selected candidate positions are identified as turning critical points, and the position index and corresponding processing time or feed position mark of each turning critical point in the interaction segment are recorded for subsequent construction of the association between interaction segments.

[0031] As a preferred embodiment of the present invention, the process of generating a fragment label index from the error contribution fragment set includes:

[0032] The segment path group that satisfies the boundary turning continuity rule is mapped to the initial outer contour area of ​​the gear blank, and the trajectory clustering area is identified based on the spatial clustering of the path nodes.

[0033] When the trajectory aggregation density exceeds a preset threshold and there is spatial overlap or adjacency with the adjacent tooth groove residual processing mark area, the corresponding segment is determined to be an error contribution segment;

[0034] A segment tag index is generated based on the error contribution segment. The segment tag index includes start and end times, processing feed positions, and intervention priority scores used to determine the order of control command calls.

[0035] As a preferred embodiment of the present invention, the intervention priority score is determined comprehensively based on the cumulative thermal disturbance degree corresponding to the error contribution segment, the spatial position sensitivity of the segment in the outer contour of the gear blank, and the influence level of residual machining marks in adjacent tooth grooves.

[0036] The intervention priority score of different error contribution segments is used to sort the order of control command calls in the segment label index, so that error contribution segments with higher intervention priority scores are given priority to trigger the corresponding compensation control commands in subsequent grinding cycles.

[0037] As a preferred embodiment of the present invention, the time-domain misalignment is controlled and updated through a prediction compensation algorithm, including:

[0038] Based on the segment marker index, the tool position compensation channel, cooling interval channel and spindle fine adjustment channel are retrieved in conjunction, and the trigger time of each error contribution segment and the actual response start time of the corresponding control adjustment channel are recorded respectively.

[0039] Based on the time difference between the trigger time and the response start time, the response time difference of each control adjustment channel relative to the error contribution segment is calculated, and a channel response misalignment set reflecting the time-domain misalignment relationship of multiple channels is constructed.

[0040] Based on the aforementioned channel response misalignment set, control and adjustment channels with smaller response time differences are marked as priority linkage channel candidates;

[0041] In multiple consecutive grinding cycles, the response time difference corresponding to the candidate priority linkage channels is statistically analyzed. When the fluctuation range of the response time difference of a certain control and adjustment channel in multiple grinding cycles is lower than a preset threshold, the control and adjustment channel is determined as a high-efficiency linkage channel.

[0042] As a preferred technical solution of the present invention, after determining the high-efficiency linkage channel, the calling order of the control and adjustment channels is reorganized based on the high-efficiency linkage channel, and the high-efficiency linkage channel is configured in the pre-execution position of the control command combination, and the remaining control and adjustment channels are arranged in order according to the response time difference corresponding to the previous grinding cycle.

[0043] The recombined control command combination and the corresponding channel call order are written into the parameter buffer segment as the initial call configuration of each control adjustment channel in the next grinding cycle, so as to realize the timing inheritance of control commands between consecutive grinding cycles.

[0044] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0045] This invention constructs a structured model of the heat source distribution changes and machining path deviation behavior during gear grinding, transforming discrete thermal disturbance information into a traceable difference state diagram structure. Furthermore, it identifies a set of error segments that significantly contribute to the formation of tooth surface errors, thereby enabling early determination of the location and timing characteristics of tooth surface errors. Based on this, by constructing a segment labeling index and introducing response time difference analysis of multiple control adjustment channels, the invention dynamically adjusts the calling order of tool position compensation, cooling intervals, and spindle fine-tuning channels. This ensures that compensation commands match the error evolution process in the time dimension, avoiding the lag or redundant intervention problems present in traditional single compensation strategies. Simultaneously, by using parameter buffers to inherit control command combinations across cycles, the control strategy can be continuously modified as the machining state evolves, thereby improving the targeting and stability of tooth surface error compensation under complex thermal disturbance and path change conditions. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0047] Figure 1 This is a schematic diagram of the online estimation and compensation method for tooth surface error of the present invention;

[0048] Figure 2 This is a schematic diagram illustrating the identification of discontinuous heating blocks during gear grinding in this invention.

[0049] Figure 3 This is a schematic diagram of the multi-control adjustment channel linkage update and parameter cache segment inheritance of the present invention. Detailed Implementation

[0050] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art. The drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0051] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0052] Example 1

[0053] like Figure 1 As shown, this embodiment provides a method for online estimation and compensation of tooth surface errors in gear grinding processes, including the following steps:

[0054] S101. Obtain the heat source distribution change pattern during gear grinding. The heat source distribution change pattern is a set of heat-affected segments constructed based on the temperature difference direction change relationship of each position in the grinding wheel contact area under multiple time slices. Based on the difference relationship between adjacent sequences of the heat-affected segments in the spatial contour, extract the set of discontinuous heating block boundaries. The set of heating block boundaries consists of multiple sets of effective reversal segments that satisfy spatial continuity constraints. Each set of effective reversal segments corresponds to an independent discontinuous heating block.

[0055] In other words, the heat source distribution change pattern is achieved based on a multi-point infrared temperature acquisition unit arranged at the machining site. This unit is installed above the workpiece tooth surface directly opposite the grinding wheel, with multiple independent measuring points distributed along the gear's axial and radial directions. It is used to collect the temperature response at different spatial locations at different times during the grinding feed process. The temperature acquisition unit employs a high-sensitivity, fast-response thermopile sensor, with a sampling period controlled within 1 ms, capable of capturing minute temperature differences caused by cutting heat within the grinding contact area. The temperature response data is organized temporally according to the sampling sequence and spatially based on sensor identifiers and a preset coordinate system index, forming a two-dimensional spatiotemporal structured heat dataset.

[0056] Furthermore, such as Figure 2 As shown, the extraction of the set of boundaries for discontinuous heating blocks includes:

[0057] The temperature difference direction sequence in the heat source distribution change pattern is converted into a symbol change string; for each spatial measuring point, the positive and negative relationship of the temperature difference value is judged between two consecutive sampling times; when the temperature difference value is positive, the change is recorded as +1, when the temperature difference value is negative, the change is recorded as -1, and when the temperature difference value is zero, the change is recorded as 0, thus forming a symbol sequence arranged in chronological order.

[0058] Preferably, the symbol change string is constructed by sliding a fixed window over time, and the window length is set to be no less than 5 consecutive sampling times to ensure the temporal continuity of the inversion segment judgment.

[0059] Candidate boundaries of adjacent sign change reversal segments are identified, and the sign change reversal segments are grouped according to spatial index. When the sign change changes from +1 to -1 or from -1 to +1, the corresponding time position is marked as the boundary position of the sign change reversal segment. Subsequently, based on the spatial position relationship of each measuring point in the gear coordinate system, the boundaries of adjacent measuring points with the same or similar reversal positions are spatially grouped.

[0060] The reversed boundaries in each spatial group form a candidate set of segment boundaries. Each candidate set of segment boundaries corresponds to a spatial region on the tooth surface that may have discontinuous thermal disturbance distribution characteristics. This region reflects the local structural features where the direction of heat input changes during the grinding process.

[0061] The candidate set of segment boundaries is filtered to remove boundary segments with a spatial span lower than the set minimum boundary distance, retaining boundary structures with strong spatial continuity as the effective set of inversion segments for constructing the boundaries of the heating blocks, thus forming the final set of discontinuous heating block boundaries. This filtering process effectively reduces false boundary interference caused by local noise or instantaneous temperature fluctuations, improving the accuracy and stability of boundary recognition.

[0062] It should be noted that the minimum boundary distance of the segment is dynamically determined based on the product of the grinding feed rate and the sampling period of the temperature acquisition unit. Preferably, the minimum boundary distance of the segment is set to... ,in: The compensation coefficient (with a value range of 2 to 5); For grinding feed rate, The sampling period of the temperature acquisition unit is used to filter out false boundary features caused by random thermal fluctuations at a single sampling point of the sensor.

[0063] Furthermore, the construction of the set of valid inverted segments includes:

[0064] Bind the time position corresponding to each symbol change reversal segment to the spatial measurement point index to form a reversal event unit;

[0065] In the gear coordinate system, based on the adjacency relationship of the measuring points in the axial and radial directions, a spatial adjacency set of the reversal event unit is constructed;

[0066] In the spatial adjacency set, identify the combination of measurement points that simultaneously experience inversion events at the same or adjacent sampling times, and determine the combination of measurement points as a candidate spatial inversion cluster;

[0067] The corresponding inversion event is included in the set of valid inversion segments for subsequent segment boundary construction only when the candidate spatial inversion clusters form a continuous distribution structure in the axial or radial direction.

[0068] The continuous distribution structure can be constructed in the following ways:

[0069] In the axial direction, the measurement point indices are sorted in ascending order of number to determine whether there is a continuous increasing relationship between the measurement point numbers; if the difference between any adjacent measurement point numbers in the sequence does not exceed the set axial step size threshold (e.g., 1), then the inversion cluster is considered to be continuous in the axial direction.

[0070] In the radial direction, an adjacency linked list structure is constructed based on the radial coordinate index of the measurement points; if all measurement points in the linked list can find another measurement point belonging to the same inversion cluster at their adjacent positions, then a radial link structure is considered to be formed.

[0071] If a candidate inversion cluster satisfies either of the two structural conditions mentioned above, then the cluster is considered to constitute spatial continuity.

[0072] Furthermore, to support subsequent mapping and boundary positioning, the geometric center coordinates of the measurement points contained in each effective inversion segment set can be calculated as the spatial reference position of the thermal disturbance boundary.

[0073] To further explain, in order to achieve physical matching between heat source distribution data and grinding path information, this embodiment also includes spatiotemporal registration processing of the heat dataset, machining reference trajectory, and gear coordinate system:

[0074] (1) Spatial projection alignment: A mapping matrix from the sensor identifier coordinates to the gear global coordinate system is pre-established. The two-dimensional spatiotemporal structural heat dataset is projected onto the spatial topology mesh of the gear tooth surface according to the actual installation position and tilt angle of the sensor on the machine tool.

[0075] (2) Time sequence synchronization: Using the machine tool spindle synchronization pulse as the reference timestamp, the sampling timing of the thermopile sensor is aligned with the timing of the tool position controller of the CNC system to eliminate the microsecond-level clock offset caused by the signal transmission link.

[0076] (3) Dynamic contact area mapping: Real-time call the feed position information in the grinding path, calculate the tooth surface coordinate index corresponding to the grinding wheel contact point at the current moment, and map the identified heating block boundary and the tool axis offset vector to the same spatial measurement space.

[0077] S102. Based on the set of discontinuous heating block boundaries, cross-compare it with the tool axis offset trend information in the current grinding path to identify interactive segment groups with relative offset abrupt change indicators, and construct a difference state diagram structure containing the connection relationship of interactive segments.

[0078] In other words, the tool axis offset trend information in the grinding path is obtained by modeling the contact point change behavior of multiple consecutive trajectory segments in the tool path during the machining process. During tooth surface grinding, the actual contact point between the grinding wheel and the workpiece tooth surface has certain spatial offset characteristics due to factors such as path turning, load disturbance or uneven cooling. To characterize this offset trend, every three consecutive trajectory segments are used as an analysis window. The contact points in the middle trajectory segment are fitted with the trajectory, and the displacement direction of the maximum contact point position relative to the preceding and following segments is calculated. The above displacement direction is quantified into an offset trend vector.

[0079] If the angle change between the offset trend vector and its immediate neighbors exceeds a preset threshold (e.g., an angle change exceeding 15°), the corresponding trajectory segment is marked as an offset abrupt change segment. The offset abrupt change segment is cross-compared with the set of discontinuous boundaries of the heat source. If spatially overlapping offset abrupt change segments appear within a certain thermal disturbance block, the segment is considered to have thermal-mechanical coupling characteristics and is included in the interactive segment group. Based on the interactive segment group, a difference state diagram structure containing spatial turning relationships and temporal sequential connection relationships is established.

[0080] Specifically, the construction of the difference state diagram structure includes:

[0081] Each segment in the interactive segment group is mapped to a machining path reference trajectory, and turning critical points are marked based on changes in the local curvature of the trajectory. For each interactive segment, its spatial mapping position in the machining path reference trajectory is first obtained. By constructing tangents at equidistant sampling points and calculating the angle between adjacent tangents, a curvature change sequence is obtained. If the angle change between adjacent sampling points exceeds a set angle threshold (e.g., 15°), the position is marked as a turning critical point. Turning critical points are used to identify direction switching behavior in the trajectory, reflecting potential process disturbance nodes in the machining path.

[0082] Using turning critical points as connection nodes, valid connection pairs between segments are extracted to form the initial graph structure. After identifying turning critical points, valid connection pairs are constructed based on the spatial adjacency relationship and temporal adjacency between segments. Each connection pair consists of two turning critical points, indicating that the two interacting segments have processing continuity on the spatial processing path. Only segments whose time interval does not exceed a set time window (e.g., 10ms) and satisfy spatial adjacency constraints (e.g., axial or radial adjacent measurement points) are included in the set of valid connection pairs.

[0083] Directed edges are established based on the temporal relationship of connected pairs, and edge pairs that do not meet the boundary turning continuity rule are removed. Valid connected pairs are then constructed with consistent-direction edges according to their corresponding time markers. Subsequently, a filtering process is performed based on the boundary turning continuity rule: if there is an abnormal reverse angle between the two critical points of a connected pair (e.g., a reversal exceeding 90°), or if the path span exceeds a set physical limit (e.g., 5mm), it is determined to be a false connected edge pair and is removed. This rule is designed to prevent false connections caused by short-term high-frequency vibrations or temperature noise, thereby improving the stability and physical consistency of the graph structure.

[0084] The edge pairs are preserved to construct a differential state graph structure, and each node is assigned a path span identifier and a segment consecutive number index. All filtered connection pairs are then used to form a directed graph structure, with each interaction segment mapped to a graph node. The edge pairs reflect the continuous processing behavior in the processing path. Furthermore, attribute information is added to each node, including: the processing path span (unit length or angle) corresponding to the segment, its number index in the overall segment sequence, start and end timestamps, etc., which facilitates its use as input features in subsequent thermal drift analysis or lifetime prediction models.

[0085] Furthermore, the marker turning critical point includes:

[0086] Calculate the discrete tangential angle sequence of each interactive segment on the processing reference trajectory; sample the position of each interactive segment on the processing reference trajectory at equal intervals, select every three consecutive sampling points to construct two adjacent tangential vectors, and calculate their included angle value. The included angle value is obtained by calculating the vector dot product formula, forming a discrete tangential angle sequence arranged in order of trajectory position, which is used to characterize the degree of local turning change of the segment on the trajectory.

[0087] Furthermore, the calculation of the discrete tangential angle sequence includes:

[0088] Three consecutive trajectory points are sampled at equal intervals on each interactive segment to construct two adjacent tangent lines;

[0089] Specifically, within each interactive segment, samples are taken at equal intervals according to the trajectory length to obtain multiple sampling points distributed along the processing trajectory. For each group of three consecutive points... , , respectively and Construct adjacent tangent vectors to capture the local turning trends of the path.

[0090] Calculate the angle between adjacent tangents to form a discrete set of angles; calculate the angle between any set of adjacent tangents. The following calculation formula is used:

[0091] ;

[0092] in, , ;

[0093] included angle value This is used to measure the change in direction of the trajectory in space. A larger angle indicates that the trajectory has a significant deflection.

[0094] All included angles are arranged sequentially along the trajectory direction to form an included angle sequence, and then bound to the original trajectory segment number.

[0095] Arrange all the above included angle values ​​in the natural order of the trajectory points to form a discrete tangential included angle sequence corresponding to the interactive segment. ; This represents the trajectory segment number. Simultaneously, each angle value is bound to its corresponding trajectory segment number for subsequent steps to locate the position of angle abrupt changes and to ensure consistency between critical point mapping and node connection criteria.

[0096] In the angle sequence, the location of the angle reversal trend change is identified, and angle reversal points with amplitudes below a set threshold are filtered out. By traversing the angle sequence, the difference between each pair of adjacent angles is calculated, and it is determined whether the sign of the difference changes. If the angle changes from positive to negative or from negative to positive, it is considered that there is a reversal of the turning trend. It is further determined whether the absolute value of the reversal angle exceeds a set angle threshold (e.g., 10°). If it does not exceed the threshold, it is considered a local disturbance and is removed, thus retaining only valid angle reversal points with significant turning characteristics.

[0097] Specifically, the formula for the difference between adjacent angles is: ; Indicates the trajectory segment number.

[0098] The difference between adjacent angles The positive and negative relationships generate corresponding symbol strings for reversing trend changes at different angles. ,in:

[0099] like Then record ;

[0100] like Then record ;

[0101] like Then record .

[0102] This symbol string reflects the rising, falling, or stable trend of the included angle on the trajectory, serving as the basis for judging angle reversal; trend symbol string. .

[0103] In trending symbol strings In the process of scanning the symbols before and after each item, if any of the following reversal types occur: i.e., from +1 to -1; or from -1 to +1, then that position is considered a candidate angle reversal point, and its corresponding index in the included angle sequence is recorded. .

[0104] For multiple consecutive "0" states (stable intervals), a maximum jump tolerance can be set. The maximum jump tolerance is set based on the grinding fluid spray interference frequency, which effectively avoids the interruption of heat source tracking in complex rinsing environments. That is, it allows trend reversal detection within the range of no more than the number of jumps, so as to enhance the fault tolerance to small-scale fluctuations.

[0105] To reduce misjudgments caused by unstructured interference such as measurement noise and local trajectory jitter, at each candidate reversal point location... Calculate the actual change between the two included angle values:

[0106]

[0107] If the actual change range Below the set angle reversal threshold If the threshold value is too high, the candidate inversion point position is discarded; otherwise, it is retained. Preferably, the threshold value is... Set to 10-15° (adjustable) to ensure that only structural turning features are retained; this can be understood as the angle reversal threshold. The selection criterion is based on the contour envelope accuracy requirements of the grinding wheel. When the abnormal deflection of the grinding trajectory at the microscale exceeds this threshold, it usually corresponds to irreversible errors caused by insufficient machine tool rigidity or sudden changes in cutting force. This setting can filter out the angular noise generated by normal high-frequency micro-vibrations of the machine tool.

[0108] Ultimately, all angle reversal points that pass the amplitude threshold screening form the turning critical point, whose elements include: the index position in the included angle sequence, the corresponding original trajectory point number, the timestamp or feed position in the processing path, and the actual included angle change amplitude.

[0109] The remaining reversal points are used as turning critical points, and their corresponding position indices and timestamps in the segment are recorded. The angle reversal points that are retained after filtering are confirmed as turning critical points, and each critical point is assigned its sampling index position and corresponding timestamp in the segment trajectory. The position index is used to locate the specific contact point position when constructing the connection relationship later, and the timestamp is used for sorting and establishing direction edges to ensure temporal consistency.

[0110] This can be understood as binding the position index of each turning critical point in the discrete tangential angle sequence with its corresponding original trajectory sampling point number; recording the time marker or feed position marker corresponding to the turning critical point according to the time order of the trajectory sampling points in the processing path; and using the position index and time marker together as the structural attribute of the turning critical point for subsequent temporal judgment and spatial positioning of connection relationships.

[0111] By utilizing the aforementioned critical point locations to guide the connection node extraction process, the graph structure's ability to express turning patterns is enhanced. When constructing the interactive segment graph structure, these turning critical points are used as candidate connection nodes, prioritizing the establishment of edge connections originating from the critical points between spatially adjacent segments. This approach improves the graph structure's ability to express changes in actual processing paths, avoids structural deviations caused by noise or atypical contact behaviors, and provides structured support for subsequent thermal drift propagation path identification.

[0112] In other words, between adjacent or temporally continuous interactive segments, their turning critical points are preferentially selected as potential connection start and end points. When the turning critical points of two interactive segments satisfy a temporal continuity relationship on the machining path and are spatially mapped to adjacent or reachable path segments, a valid connection is determined to exist between them. For segments that do not contain turning critical points, or whose turning critical points do not satisfy the continuity constraints in both time and space, no connection is established to avoid erroneous connections introduced by local noise or atypical trajectory disturbances. By using turning critical points as constraints on connection nodes, the difference state diagram structure can more accurately reflect the actual directional changes and offset evolution nodes in the grinding path, thereby enhancing the ability of the diagram structure to express changes in machining state.

[0113] S103. Take the segment path group that satisfies the boundary turning continuity rule in the difference state diagram structure as the input segment, combine the initial outer contour line of the gear blank and the residual marking information of adjacent tooth grooves, identify the error contribution segment set that constitutes the tooth surface error significant contribution area, and generate a segment marking index for the intervention time sequence for the node set.

[0114] Specifically, firstly, each path node in the segment path group is mapped to the initial outer contour region of the gear blank in the gear coordinate system. By matching the spatial position of each node with the grid points of the blank contour line, trajectory clustering regions near the contour boundary are identified. When the trajectory clustering density of a certain region exceeds a preset trajectory clustering density threshold, it is determined to be a potential error excitation region with significant superposition of processing heat effects, and is used as the first candidate segment set. The trajectory clustering density threshold is defined by the path node coverage rate per unit area. When the repetition rate of path nodes within the same coordinate region exceeds 60% in consecutive processing cycles, it is determined to be a heat effect superposition region. The trajectory clustering density threshold is combined with the thermal diffusivity of the material to distinguish between the instantaneously passing heat source region and the superposition region that produces deep thermal effects.

[0115] Furthermore, the distribution data of residual heat traces in adjacent tooth grooves obtained by the post-processing image detection system or high-precision 3D scanning system are invoked. The heat trace markers are mapped to the aforementioned trajectory path space according to the gear coordinate system to determine whether the trajectory clustering area has a spatial overlap or adjacency relationship with the residual heat trace area. Only when a certain area simultaneously meets the conditions of high trajectory clustering and residual heat trace coverage is it identified as a composite error contribution point caused by the coupling of thermal offset effect and material residual effect, and included in the final significant error candidate region, while isolated clustering segments unrelated to the residual area are excluded.

[0116] Within the aforementioned significant error candidate regions, all path nodes are aggregated, and a structured representation model is established. This set of error contribution segments is constructed as a directed graph based on the temporal relationships within the processing path. Each node contains at least the following attribute information: its associated path segment number, the index of the associated temperature reversal segment, the direction of local offset trend change, the number of adjacent heat trace residual regions, and the mapping position number in the blank outline. Through node-level attribute fusion and topology modeling, the error contribution region is transformed from the original segment clustering level to the level of controllable node behavior.

[0117] Finally, based on the aforementioned set of error contribution segments, a segment labeling index table for time-series intervention is generated for each segment path group. This index table includes: start and end time labels for each segment, machining feed position coordinates, corresponding contact point information, intervention priority score (which can be calculated by comprehensively considering the cumulative thermal disturbance amplitude of the segment, the sensitive spatial position coefficient, and the residual thermal trace level), and labeling status (such as whether error compensation has been implemented, whether it has been included in the lifetime prediction model, etc.). This labeling index table serves as a key input for lifetime prediction and active compensation path planning in subsequent steps, and is used to locate the key time periods and spatial targets for tooth surface error intervention.

[0118] To further explain, the statistical method for the intervention priority score is as follows: the cumulative thermal disturbance amount is used to characterize the temporal superposition effect of heat, the spatial sensitivity coefficient is used to characterize the geometric response difference of the tooth surface topology, and the residual thermal trace level of adjacent tooth grooves is combined to characterize the spatial coupling effect; through linear weighted fusion, the multi-source heterogeneous processing disturbance characteristics are mapped into a unified dimension intervention criterion, thereby guiding the temporal response priority division of the control channel under complex processing paths.

[0119] Preferably, the formula for calculating the intervention priority score is:

[0120]

[0121] in: To accumulate thermal disturbance, This is the spatial sensitivity coefficient (e.g., the tooth root or tooth tip region). The level is determined by thermal residue. , and These are the cumulative thermal disturbance weighting coefficients, spatial sensitivity weighting coefficients, and thermal trace residual level weighting coefficients.

[0122] It should also be noted that the segment tagging index table is not only used to locate intervention periods, but also to dynamically suppress and release control channel resources through intervention priority scoring. Specifically, when the intervention priority score of a segment exceeds a set high-risk threshold, the system automatically increases the priority of that segment in the parameter cache segment and forcibly shortens the response start window of each adjustment channel to offset the impact of physical time delay. Details are as follows:

[0123] Priority-driven instruction reconfiguration: Based on the intervention priority score, the system defines the segment sequence with the highest intervention priority score as the "core intervention anchor point" and prioritizes matching the efficient linkage channel with the least delay fluctuation in that area. If the intervention priority score is extremely high (e.g., in the tooth root sensitive area with severe heat accumulation), the system will skip the conventional delay ascending order and directly trigger the tool position compensation module and the cooling interval module with strong coupling.

[0124] Historical adjustments to scoring weights: Response latency fluctuations will have a reverse effect on intervention priority scoring. If, over three consecutive grinding cycles, the efficient linkage channel's adjustment effect on high-scoring segments (i.e., residual error) fails to meet expectations, the system will automatically increase the cumulative thermal disturbance weighting coefficient. or spatially sensitive weighting coefficient This allows the intervention logic to be triggered earlier in the next cycle, forming an adaptive update mechanism of "scoring guides control, and control corrects scoring".

[0125] Dynamic elimination of temporal misalignment: By incorporating intervention priority scores into the calculation logic of the temporal misalignment vector, segments exhibiting "high scores but lagging regulation" due to slow physical response are identified. Within the parameter buffer, for these segments, a predictive compensation algorithm is used to adjust the trigger points of control commands. , , Relative fragment triggering time Controlled advance can be implemented to eliminate temporal misalignment at the physical level.

[0126] S104. Based on the segment marker index, the tool position compensation module, cooling interval module and spindle fine adjustment module in the set control adjustment channel group are retrieved in conjunction. According to the response order of each type of channel in the last two adjustments and the time domain misalignment between the segment trigger sequence, the control command combination is updated. The update result is sent to the parameter buffer segment of the next grinding cycle.

[0127] This can be understood as: such as Figure 3 As shown, by optimizing the control channels through segment marking index execution, the grinding system's real-time adjustment capability for segments contributing to tooth surface errors can be improved, redundant interventions reduced, and response synchronization enhanced. The control adjustment channel group includes control adjustment channels corresponding to the tool position compensation module, cooling interval module, and spindle fine-tuning module. All three control adjustment channels are set in parallel, each responding to control calls from segment trigger events; that is to say:

[0128] Tool position compensation channel, including tool position compensation module;

[0129] Cooling intermittent channel, including cooling intermittent module;

[0130] Spindle fine-tuning channel, including spindle fine-tuning module.

[0131] Specifically, the time-domain misalignment is updated under the control of a prediction compensation algorithm, including:

[0132] Record the actual time series triggered by the previous segment and the start time of the three-channel response; that is: the actual time series is: ,in: Index the fragment. It is a positive integer; and in the fragment The corresponding response start times of the tool position compensation channel, cooling interval channel, and spindle fine-tuning channel are as follows: , , ;

[0133] Calculate the response time difference between the start point of each channel response and the corresponding trigger segment, and form a channel response misalignment vector;

[0134] For each segment Calculate the response delay and response time difference for each of the three control channels. Always positive (i.e., the response is later than the trigger), defining the response time difference. The formula is:

[0135] ;

[0136] ;

[0137] ;

[0138] In other words, the channel response misalignment vector is obtained as follows: .

[0139] Because mechanical actuators have inherent physical time delays, in order to achieve synchronous response, when performing forecast compensation, the historical data is used for controlled early triggering. Therefore, the command triggering time of the next cycle should be set as follows: the command triggering time of the next cycle is the time when the forecast error occurs minus the historical response delay time.

[0140] The control channel corresponding to the minimum value in the channel response misalignment vector is marked as a priority linkage channel candidate.

[0141] The response delay fluctuations of the candidate priority linkage channels over the past three grinding cycles are statistically analyzed. If the channel response consistently remains at the lowest or second-lowest latency level within consecutive cycles, and the standard deviation is below a preset threshold (e.g., 30ms), then the channel is confirmed as the efficient linkage channel for the current segment trigger sequence.

[0142] This can be understood as follows: if the response lag of the cooling intermittent channel is minimized, the cooling intermittent module will be triggered first in the next cycle. At the same time, the start windows of the tool position compensation module and the spindle fine-tuning module will be adjusted to achieve optimal linkage. In the multi-cycle linkage adjustment, a control channel sorting evolution path driven by "response history" is formed to adapt to the adjustment needs of different segment types, improve the adjustment coverage and linkage efficiency of high-temperature disturbance areas or path abrupt segment. If the standard deviation of the response delay of all channels is higher than the preset threshold, the system will automatically trigger the safe machining mode, reduce the feed rate, and use the most recent historical average delay as the default compensation bias until the system fluctuation returns to the normal threshold range.

[0143] In other words, the established high-efficiency linkage channels are reorganized in the following manner to reorder the control call sequence:

[0144] A channel priority execution queue is constructed, placing the high-efficiency linkage channel at the head; the remaining two channels are arranged in ascending order of response delay from the previous cycle. The call windows of each channel are synchronously adjusted to ensure the response trigger point is as close as possible to the segment trigger moment. The reorganized control instruction combination structure is written into the parameter cache segment as the call priority template for the next grinding cycle. Each template record includes the priority queue, control window adjustment amount, and response synchronization evaluation index. During multi-cycle evolution, a response history-driven control adjustment channel sorting path is formed, realizing dynamic linkage optimization between segment trigger events and control modules, thereby improving the response efficiency and coverage capability for thermal disturbance excitation regions and trajectory mutation segments, forming an update mechanism.

[0145] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for online estimation and compensation of tooth surface error in gear grinding process, characterized in that, Includes the following steps: S101. During gear grinding, temperature change data of the contact area between the grinding wheel and the gear surface are collected in multiple time slices. Based on the temperature difference direction change relationship of each spatial position in the temperature change data, a heat source distribution change pattern is constructed, and multiple discontinuous heating blocks are identified from the heat source distribution change pattern. S102. The spatial offset trend of the discontinuous heating block and the tool position axis in the grinding path is analyzed to identify interactive segments that have both heat distribution change characteristics and tool position offset change characteristics. Based on the spatial adjacency relationship and temporal order relationship between the interactive segments, a difference state diagram structure containing the connection relationship of the interactive segments is constructed. S103. Based on the difference state diagram structure, combined with the initial outer contour information of the gear blank and the residual machining mark information of adjacent tooth grooves, identify the set of error contribution segments that have a significant impact on the formation of tooth surface error, and generate a segment mark index for the set of error contribution segments. S104. Based on the segment marker index, multiple preset control adjustment channels are linked and called. Based on the time domain misalignment between the response order of each type of channel in the two most recent adjustments and the segment trigger sequence, the control commands corresponding to the tool position compensation, cooling interval and spindle running parameters are combined and updated. The updated control command combination is sent to the parameter buffer segment for application in subsequent grinding cycles, thereby realizing online compensation for tooth surface error. The process of generating a fragment label index from the set of error contribution fragments includes: The segment path group that satisfies the boundary turning continuity rule is mapped to the initial outer contour area of ​​the gear blank, and the trajectory clustering area is identified based on the spatial clustering of the path nodes. When the trajectory aggregation density exceeds a preset threshold and there is spatial overlap or adjacency with the adjacent tooth groove residual processing mark area, the corresponding segment is determined to be an error contribution segment; A segment labeling index is generated based on the error contribution segment. The segment labeling index includes start and end times, processing feed position, and intervention priority score for determining the order of control command calls. The time-domain misalignment is controlled and updated through a prediction compensation algorithm, including: Based on the segment marker index, the tool position compensation channel, cooling interval channel and spindle fine adjustment channel are retrieved in conjunction, and the trigger time of each error contribution segment and the actual response start time of the corresponding control adjustment channel are recorded respectively. Based on the time difference between the trigger time and the response start time, the response time difference of each control adjustment channel relative to the error contribution segment is calculated, and a channel response misalignment set reflecting the time-domain misalignment relationship of multiple channels is constructed. Based on the aforementioned channel response misalignment set, control and adjustment channels with smaller response time differences are marked as priority linkage channel candidates; In multiple consecutive grinding cycles, the response time differences corresponding to the candidate priority linkage channels are statistically analyzed. When the fluctuation range of the response time difference of a certain control adjustment channel in multiple grinding cycles is lower than a preset threshold, the channel is... The control and adjustment channel is determined to be a high-efficiency linkage channel.

2. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 1, characterized in that, The process of identifying discontinuous heating blocks includes: The heat source distribution change pattern during gear grinding is obtained, the temperature difference direction sequence in the heat source distribution change pattern is converted into a symbol change string, and the candidate boundary is determined based on the inversion position of the temperature difference direction sequence in the time dimension. Based on the spatial relative position of each candidate boundary in the gear coordinate system, spatial grouping is performed. When the candidate boundaries in the same group form a continuous distribution structure in the gear axial or radial direction, a set of effective reversal segments of the heating block boundary is constructed, and the set of effective reversal segments is determined as a discontinuous heating block.

3. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 2, characterized in that, The construction of the set of valid reverse segments includes: Bind the time position corresponding to each symbol change reversal segment to the spatial measurement point index to form a reversal event unit; In the gear coordinate system, based on the adjacency relationship of the measuring points in the axial and radial directions, a spatial adjacency set of the reversal event unit is constructed; In the spatial adjacency set, identify the combination of measurement points that simultaneously experience inversion events at the same or adjacent sampling times, and determine the combination of measurement points as a candidate spatial inversion cluster; The corresponding inversion event is included in the set of valid inversion segments for subsequent segment boundary construction only when the candidate spatial inversion clusters form a continuous distribution structure in the axial or radial direction.

4. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 1, characterized in that, The process of identifying the interaction segment includes: The discontinuous heating blocks are mapped to the grinding path reference trajectory, and the tool position axis offset trend of the trajectory segment in the grinding path is calculated. When a trajectory segment is located within the coverage area of ​​a discontinuous heating block, and its offset trend changes by more than a preset threshold relative to adjacent trajectory segments, the trajectory segment is determined to be an interactive segment.

5. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 4, characterized in that, The process of constructing the difference state diagram structure includes: Each segment in the interactive segment group is mapped to the processing path reference trajectory, and the turning critical point is marked according to the change of local curvature of the trajectory. The turning critical point is used as the connection node to extract the effective connection pairs between segments and form the initial graph structure. Directed edges are established based on the sequential relationship of the connection pairs on the time axis, and edge pairs that do not meet the boundary turning continuity rule are removed. The remaining edge pairs are constructed into a differential state graph structure, and each node is assigned a path span identifier and a segment continuous number index. All the filtered connection pairs form a directed graph structure, and each interaction segment is mapped to a graph node. The edge pairs reflect the continuous processing behavior in the processing path.

6. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 5, characterized in that, The process of marking the turning critical point includes: The position of each interactive segment on the processing reference trajectory is sampled at equal intervals. Local tangents are constructed based on adjacent sampling points, and the included angle between adjacent tangents is calculated to form a discrete tangential angle sequence arranged along the processing path. In the discrete tangential angle sequence, candidate turning positions are identified based on the reversal relationship of the changing directions of adjacent angle values, and candidate positions with angle changes below a preset threshold are eliminated. The selected candidate positions are identified as turning critical points, and the position index and corresponding processing time or feed position mark of each turning critical point in the interaction segment are recorded for subsequent construction of the association between interaction segments.

7. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 1, characterized in that, The intervention priority score is determined comprehensively based on the cumulative thermal disturbance degree corresponding to the error contribution segment, the spatial position sensitivity of the segment in the outer contour of the gear blank, and the influence level of the residual machining marks in the adjacent tooth groove. The intervention priority scores of different error contribution segments are used to sort the order of control command calls in the segment mark index, so that the error contribution segments with higher intervention priority scores are preferentially triggered with corresponding compensation control commands in subsequent grinding cycles.

8. The method for online estimation and compensation of tooth surface error in gear grinding process according to claim 1, characterized in that, After determining the high-efficiency linkage channel, the calling order of the control and adjustment channels is reorganized based on the high-efficiency linkage channel. The high-efficiency linkage channel is configured as the pre-execution position of the control command combination, and the remaining control and adjustment channels are arranged in order according to the response time difference corresponding to the previous grinding cycle. The recombined control command combination and the corresponding channel call order are written into the parameter buffer segment as the initial call configuration of each control adjustment channel in the next grinding cycle, so as to realize the timing inheritance of control commands between consecutive grinding cycles.

Images