Three-coordinate measurement data intelligent analysis system

Abstract

The invention discloses a three-coordinate measurement data intelligent analysis system, relates to the technical field of three-coordinate measurement intelligent analysis, and is used for solving the problem of inaccurate probe wear error identification and compensation. According to the method, the wear rate is predicted by constructing a standardized feature set and combining an LSTM model, then the remaining life of the probe is calculated, whether the probe is in a critical state or not is judged, intelligent monitoring and early warning of the health state of the probe are achieved, the system generates a measuring point level credibility label by fusing the wear rate, the environment temperature and measurement time sequence information, and the reliability of the probe is improved. High-risk measurement data are effectively identified, when the probe enters a critical state, the system can automatically suspend a task and send an early warning prompt to assist in probe replacement or calibration, measurement error diffusion caused by probe degradation is avoided, and reliability and data credibility of the three-coordinate measurement process are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent analysis technology for coordinate measuring machines (CMMs), and more specifically, to an intelligent analysis system for CMM data. Background Technology

[0002] Currently, coordinate measuring machines (CMMs) are key devices in precision manufacturing quality control, widely used for measuring the geometric dimensions and form and position tolerances of complex parts, possessing high precision, high stability, and multi-axis spatial sensing capabilities. Traditional CMMs mostly employ contact probes, acquiring three-dimensional coordinate points by contacting the workpiece surface with the probe, and then calculating the workpiece's dimensional deviations, contour errors, and tolerance compliance. With the development of intelligent manufacturing, CMM tasks are gradually transitioning from single-piece inspection to batch continuous inspection, multi-machine linkage, and unmanned operation, placing higher demands on the long-term operational stability of the system and the reliability of measurement results.

[0003] The shortcomings of existing technologies are as follows: In continuous measurement tasks, the microscopic deformation of the probe or the wear of the spherical surface can cause the contact point position to shift, and the measurement data will show structural deviations. However, due to the lack of a modeling and state quantification mechanism for probe wear, the system cannot identify whether such deviations are caused by probe degradation, and still uses them as valid data in deviation evaluation. As a result, in the data processing stage, the systematic drift in the error distribution of the measurement point may be misinterpreted as workpiece error, misleading the analysis and judgment, resulting in misjudgment of non-conformity or omission of quality problems. Since no mapping relationship is established between the probe state and the measurement data deviation, the measurement system lacks self-diagnosis and early warning capabilities during long-term operation, making it difficult to adapt to the requirements of flexible manufacturing for high-frequency, multi-batch, and high-consistency measurement. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the following solution is proposed to solve the problem of inaccurate identification and compensation of probe wear error in the above-mentioned background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A coordinate measuring machine (CMM) intelligent analysis system includes a state parameter acquisition module, a feature extraction module, a wear rate modeling module, a probe life prediction module, and a state assessment linkage response module, with each module connected by a signal.

[0007] The status parameter acquisition module is used to acquire the operating status parameters of the probe in real time during the coordinate measuring machine task;

[0008] The feature extraction module is used to normalize, denoise, and align the running status parameters, and extract feature vectors that characterize the probe's usage status to construct a standardized state feature set.

[0009] The wear rate modeling module constructs a probe wear rate estimation model based on the state feature set and preset historical training samples, and outputs an estimated wear rate of the current probe.

[0010] The probe lifetime prediction module is used to estimate the remaining lifespan of the probe based on the wear rate and cumulative triggering behavior, and to determine whether the probe is in a preset critical state range.

[0011] The status assessment linkage response module is used to generate lifespan warning information, mark the credibility of relevant measurement data, and send status abnormality prompt signals to the operation terminal when the probe is in a critical state, so as to realize dynamic adjustment or suspension control of the measurement task.

[0012] Furthermore, the steps of the feature extraction module include:

[0013] Collect probe running status parameters from various dimensions, use a sliding time window mechanism to locally truncate the collected sequence, and construct a time-aligned state subsequence;

[0014] A nonlinear normalization method is used to compress the value range of various parameters, and the original parameters are uniformly mapped to the standard scoring interval.

[0015] For noise-sensitive parameters, use median filters or exponential smoothing to improve stability.

[0016] Finally, a standardized feature vector representing the current state of the probe is output, and a continuously evolving sequence of state features is constructed.

[0017] Furthermore, the wear rate modeling module is used to perform the following steps:

[0018] The historical probe is constructed using a sample set, and the sample features include the correspondence between state feature vectors and actual wear results;

[0019] The Long Short-Term Memory (LSTM) network is used as the state sequence learning structure to perform temporal modeling on the input state feature sequence.

[0020] Wear status labels are introduced as supervision signals to perform supervised training and generate wear rate output channels;

[0021] Generate an estimate of the wear rate at the current moment based on the model output.

[0022] Furthermore, based on the model output, an estimate of the wear rate at the current moment is generated, including:

[0023] The hidden state unit of the probe wear rate estimation model is used to perform feature fusion and correlation analysis on the changes between the current state and the historical state, and output an intermediate vector representation of the current probe health status.

[0024] The intermediate vector is input into the model output layer. Through a set of trained weights and activation functions, a standardized wear rate score is generated. The wear rate score is located in a closed interval between 0 and 1.

[0025] The wear rate score is inversely normalized to the known actual wear rate range in the training sample set to restore the wear rate estimate, which is used to indicate the degree of probe performance degradation per unit time.

[0026] Furthermore, the remaining service life of the probe is estimated based on wear rate and cumulative triggering behavior, including:

[0027] Count the number of triggers and total contact time of the current probe since the last replacement or calibration;

[0028] Construct a remaining lifetime function based on cumulative wear rate and total lifetime budget, mapping cumulative wear to a percentage of lifetime.

[0029] Furthermore, a remaining lifetime function based on cumulative wear rate and total lifetime budget is constructed, mapping cumulative wear to a percentage of the lifetime, including:

[0030] The probe was collected to estimate all historical wear rates since it was put into use, and a time series table of wear rates was constructed based on timestamps.

[0031] Perform a time-weighted cumulative operation on the wear rate time series list, multiply the wear rate of each time period by the duration of that time period and sum them up to obtain the cumulative wear rate at the current moment;

[0032] Obtain the probe's total lifetime budget value, which is the system's preset maximum allowable wear accumulation rate, used to reflect the probe's maximum usage threshold;

[0033] The ratio of the cumulative wear rate to the total lifespan budget is calculated to obtain the current probe lifespan percentage, which reflects the ratio between the probe's consumed lifespan and its remaining lifespan.

[0034] Furthermore, it determines whether the probe is within a preset critical state range, including:

[0035] Set a lifetime status level threshold to classify and judge the proportion of probe lifetime. The lifetime status level includes normal status, warning status and critical status.

[0036] Set a first threshold and a second threshold. The first threshold is used to define the boundary between normal and warning, and the second threshold is used to define the boundary between warning and critical. The two thresholds are configured based on the probe material type, structure model and usage scenario experience.

[0037] The calculated lifecycle percentage is compared with two thresholds one by one. When the lifecycle percentage is less than the first threshold, it is judged as a normal state; when the lifecycle percentage is greater than or equal to the first threshold and less than the second threshold, it is judged as a warning state; when the lifecycle percentage is greater than or equal to the second threshold, it is judged as a critical state.

[0038] The determination result is output as a lifetime status label and attached to the probe's current measurement task.

[0039] Furthermore, it is used to generate lifetime warning information when the probe is in a critical state, mark the reliability of relevant measurement data, and send an abnormal status prompt signal to the operation terminal, so as to realize dynamic adjustment or suspension control of the measurement task, including:

[0040] Determine the remaining life status label output by the life prediction module. If an early warning state is entered, record the current task measurement point number and add a medium confidence flag to the measurement data.

[0041] If a critical state is reached, a control signal command is generated to suspend the current task execution process and send a prompt message to the operation terminal, recommending to perform probe replacement, recalibration, or compensation calibration operations.

[0042] Furthermore, the steps for labeling the confidence level of relevant measurement data include:

[0043] Extract the probe status features, wear rate estimate, and lifetime status label associated with each measurement point; construct a scoring function that takes the current probe wear rate, ambient temperature change, and measurement time at the measurement point as combined inputs;

[0044] Generate a confidence score in the range of 0 to 1 from the combined input values;

[0045] Based on the credibility score, a grading strategy is set to divide the measurement points into three categories: high credibility, medium credibility, and low credibility, and the credibility labels are attached to the measurement data output structure.

[0046] The technical effects and advantages of the intelligent analysis system for coordinate measuring machine data of the present invention are as follows:

[0047] This invention collects multi-dimensional state parameters, including trigger frequency, contact force, trigger duration, displacement fluctuation trend, and ambient temperature, to construct a standardized feature set. Combined with an LSTM model, it models and predicts probe wear behavior. Based on the probe's wear rate and trigger behavior, it can dynamically estimate its remaining lifespan and determine whether it has entered a warning or critical state. This enables early identification and quantitative management of probe degradation trends, solving the problems of difficult early warning of probe errors and unquantifiable wear in traditional measurement systems, and improving the reliability of measurement data and the system's adaptability.

[0048] Building upon this foundation, the system further constructs a reliability scoring mechanism oriented towards the measurement point level. This mechanism integrates influencing factors such as current wear rate, measurement time at the measurement point, and environmental fluctuations to identify high-reliability, medium-reliability, and low-reliability data areas. This helps operators accurately identify risky measurement points. Through a state-driven linkage response module, when the probe is in a critical state, the system can automatically pause the task, issue an early warning, and suggest replacement or calibration, preventing the continuous propagation of abnormal measurement data and ensuring measurement quality and production safety. Compared to traditional methods relying on manual experience and fixed-cycle calibration, this invention achieves digital perception and dynamic feedback of probe health status, significantly improving the intelligence level and operational reliability of the coordinate measuring machine system. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the structure of a coordinate measuring data intelligent analysis system according to the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] In order to achieve the above objectives, Figure 1 A schematic diagram of the structure of a three-coordinate measurement data intelligent analysis system of the present invention is given, which specifically includes a state parameter acquisition module, a feature extraction module, a wear rate modeling module, a probe life prediction module, and a state assessment linkage response module. The modules are connected by signals.

[0052] The status parameter acquisition module is used to acquire the operating status parameters of the probe in real time during the coordinate measuring machine task;

[0053] The feature extraction module is used to normalize, denoise, and align the running status parameters, and extract feature vectors that characterize the probe's usage status to construct a standardized state feature set.

[0054] The wear rate modeling module constructs a probe wear rate estimation model based on the state feature set and preset historical training samples, and outputs an estimated wear rate of the current probe.

[0055] The probe lifetime prediction module is used to estimate the remaining lifespan of the probe based on the wear rate and cumulative triggering behavior, and to determine whether the probe is in a preset critical state range.

[0056] The status assessment linkage response module is used to generate lifespan warning information, mark the credibility of relevant measurement data, and send status abnormality prompt signals to the operation terminal when the probe is in a critical state, so as to realize dynamic adjustment or suspension control of the measurement task.

[0057] The status parameter acquisition module is used to acquire the operating status parameters of the probe in real time during the coordinate measuring machine (CMM) task. The specific implementation steps are as follows:

[0058] The operating status of the probe is collected in real time throughout the entire process without any omissions, so as to obtain multi-dimensional raw state parameter data reflecting the dynamic characteristics of the measurement process. Specifically, the state parameters include the following five categories: trigger frequency, contact force, trigger duration, displacement change trend and ambient temperature information. These parameters are all accurately collected through dedicated sensor components and synchronously calibrated with the system time base.

[0059] By integrating a miniature Hall effect sensor or photoelectric encoder module inside the probe, a trigger signal is generated the instant the probe contacts the workpiece. The system records the cumulative number of trigger events within a fixed time period (e.g., every 100 milliseconds), and the statistical value obtained is the trigger frequency per unit time. This frequency parameter is used to measure the contact density of the measurement task and the frequency of probe operation.

[0060] By utilizing a triaxial force-sensitive sensor array installed inside the probe, the magnitude of the contact force acting on the probe tip is recorded in real time each time the probe contacts the surface of the workpiece. The contact force is measured in Newtons, and the sampling period is 1 millisecond. This parameter reflects the force stability of the probe during the measurement process and has a direct impact on probe wear and measurement accuracy.

[0061] The system records the start and end timestamps of the trigger signal for each contact event and calculates the difference between them as the duration of that contact. The duration is in milliseconds and represents the length of time the probe remains in contact with the workpiece surface. This parameter is used to evaluate the probe's response adaptability under different surface morphologies.

[0062] By combining a laser displacement sensor with a high-resolution encoder, the movement trajectory of the probe in the XYZ directions in three-dimensional space is continuously recorded. The displacement increment vector sequence within each time slice is calculated using a set sliding time window (e.g., the most recent 200ms). The rate of change of displacement velocity per unit time is then obtained through differentiation, forming a sequence of data representing the displacement change trend. This parameter is used to characterize the continuity and stability of the probe motion and is highly sensitive to subsequent state modeling.

[0063] At least one temperature sensor is deployed around the probe to monitor the ambient temperature of the measurement area in real time. Temperature information is sampled in degrees Celsius and synchronized with the measurement time.

[0064] All the above parameters are timestamped using a unified sampling clock (the high-precision time base of the system's main control module) to ensure consistency of various data in the time dimension. The operating status parameter data output by the acquisition module, after signal amplification, analog-to-digital conversion, and format normalization, is transmitted to the feature extraction module in the form of structured data frames for subsequent standardized modeling. The entire acquisition system possesses high reliability and low error drift rate, and can adapt to precision 3D measurement tasks under various probe structures and complex working conditions.

[0065] The feature extraction module is used to perform preprocessing and feature construction operations on the collected probe running status parameters, specifically including the following steps:

[0066] After the state parameter acquisition module completes the real-time acquisition of probe operating state parameters, the feature extraction module performs preliminary organization of probe operating state parameters in various dimensions, including but not limited to trigger frequency, contact force, trigger duration, displacement change trend, and ambient temperature information. The above parameters are arranged in the order of sampling time to form the original state parameter sequence. To improve the accuracy of subsequent analysis, a sliding time window mechanism is used to perform local truncation processing on the original sequence. The length of the sliding window can be configured according to the duration of the measurement task and the frequency of data fluctuation. For example, it can be set to continuous sampling data within 5 seconds with a step size of 1 second. The sampling points in each sliding window constitute a state subsequence, ensuring sufficient time coverage overlap between adjacent subsequences to support the continuity and stability of time series information modeling.

[0067] Secondly, to address the dimensional differences of different physical quantities, the feature extraction module performs nonlinear normalization on various state parameters. Specifically, for parameters with positive deviation trends, such as trigger frequency and contact force, the Sigmoid function is used for normalization, compressing their value range to a standard scoring interval between 0 and 1. For parameters such as displacement change trends, which may have negative values ​​or symmetrical distributions, the hyperbolic tangent (Tanh) function is used for value range mapping, maintaining their relative change trends while improving feature consistency. Through nonlinear normalization, the numerical dimension differences between different physical dimensions are eliminated, enabling subsequent model learning to have input features at a uniform scale.

[0068] A denoising mechanism is introduced for noise-sensitive parameters. For parameters such as contact force and displacement trends, which may be affected by minor mechanical vibrations, a median filter is used to remove extreme values ​​within a sliding window, filtering out abrupt changes. For parameters with gradual trends, such as trigger frequency and ambient temperature, an exponential smoothing method is adopted, introducing a smoothing factor to control the degree of integration between historical and current values, suppressing the interference of local abnormal fluctuations on the overall trend. The various parameters after normalization and denoising are combined into a unified feature vector, which serves as a standardized feature representation of the probe's current operating state. The feature vectors are stacked sequentially in chronological order to form a continuously evolving state feature sequence, used for subsequent modules to perform analysis tasks such as probe wear modeling, lifetime prediction, and state assessment. This standardized feature sequence not only ensures the numerical consistency of the input data but also preserves the temporal trends of various state parameters.

[0069] Based on the standardized state feature sequence output by the feature extraction module, the wear state of the probe is modeled and its rate is estimated. The specific steps are as follows:

[0070] The historical sample set required for probe wear modeling is constructed. The sample set comes from the probe operation data collected by the existing coordinate measuring system during actual use. Each sample data should include: a sequence of state feature vectors processed by the feature extraction module (including parameters such as trigger frequency, contact force, trigger duration, displacement change trend, and ambient temperature information), and the corresponding actual wear condition of the probe. The actual wear condition can be obtained from the physical detection data of the probe after regular maintenance, such as the change in geometric error measured by standardized measuring tools or the wear degree level marked in the replacement record. The sample labels can be further discretized into wear rate reference values ​​in the range of 0 to 1 for training target output.

[0071] Long Short-Term Memory (LSTM) neural networks were chosen as the temporal modeling structure. LSTMs possess the ability to model dependencies in state sequences over long time spans, making them suitable for recognizing slowly changing behavioral patterns during the gradual aging process of probes. The model input is a sequence of state feature vectors with continuous time steps, where the input dimension matches the feature vector length, for example, a five-dimensional vector (corresponding to five types of parameters). The internal structure of the LSTM includes an input gate, a forget gate, and an output gate. Through training, the influence of historical states on the current output can be adaptively controlled to construct a time-dependent model that matches the probe performance degradation process.

[0072] During training, wear status labels are introduced as supervision signals. The model's loss function adopts the form of mean squared error (MSE), aiming to minimize the difference between the predicted wear rate value and the label value in the real sample. The training data is divided into training set and validation set, and an iterative optimization strategy (such as Adam optimizer) is used to adjust the network weights so that the model has good generalization ability under different types of tasks and different working conditions.

[0073] After training, the model is used in a real-world task. The output current state feature sequence is input into the LSTM model, and the output generated by the model represents the current wear rate estimate of the probe. This estimate is an index of the rate of performance degradation per unit time, which is used for subsequent lifetime prediction and state assessment. In order to improve the interpretability of the system, the output results can also be normalized and inversely transformed to restore the wear rate reference value in the actual physical quantity, which is convenient for comparison and analysis with manual measurement or equipment maintenance records.

[0074] The specific process of generating the wear rate estimate for the current moment based on the model output is as follows:

[0075] The system first invokes the hidden state unit of the probe wear rate estimation model to perform feature fusion between the current probe state and historical states. This hidden state unit is constructed using a recurrent neural network structure, enabling it to capture probe state change trends across time dimensions. When the standardized state feature vector of the current moment is input into the model, it participates in state evolution modeling together with the state memory unit accumulated in the previous moment, generating an intermediate vector representation that integrates the features of the current and historical states. This intermediate vector representation reflects the stability, wear trend, and potential abnormal changes of the probe's recent operating state, representing a multidimensional compressed expression of the probe's health status. This intermediate vector representation is then input into the model's output layer to perform the wear score inference mapping process. The output layer uses a pre-trained set of weight parameters and activation functions, typically a single hidden layer fully connected structure. The activation function performs a non-linear mapping on the linearly transformed values, generating a normalized wear rate score. This wear rate score is located in a closed interval between zero and one, where zero indicates that the probe is in a completely healthy state with no obvious wear, while one indicates that the probe is in a severely worn state with significantly degraded performance.

[0076] To obtain a physically meaningful wear rate estimate, the system further performs an inverse normalization operation. This operation is based on a linear mapping recovery of the actual wear rate range known in the model training sample set. For example, if the actual wear rate range of the training samples is between 0.5% and 2.5% per thousand triggers, the normalized score value is mapped to this range to obtain the actual performance degradation rate of the current probe per unit time or per unit number of triggers. The final output wear rate estimate is used as an important basis for subsequent life cycle estimation and early warning judgment.

[0077] For example, in a high-precision parts inspection task, the measuring machine runs continuously for 15 minutes. The system collects and inputs probe operating status parameters, including a trigger frequency of 0.70 times / second, a contact force of 1.31 N, a trigger duration of 100 milliseconds, a displacement change trend of 0.22 mm / second, and an ambient temperature of 27.2 degrees Celsius. The feature extraction module performs nonlinear normalization on these parameters, uniformly mapping them to between 0 and 1. It also uses exponential smoothing to process the contact force and trigger duration to resist interference, generating a standardized state feature vector. Subsequently, the past five sets of time-series feature vectors are used as input and fed into a wear rate estimation model built on a Long Short-Term Memory (LSTM) network. This model has been trained on 5000 historical labeled probe samples. The system model outputs a wear score of 0.68 for the current state, representing a health condition of moderate to high wear. Referring to the physical wear rate range (0.01 mm to 0.10 mm / thousand triggers) corresponding to a wear score of 0 to 1 in the reference training set, the system maps a score of 0.68 to a current wear rate of approximately 0.072 mm / thousand triggers. Combined with the current trigger frequency of 600 times per minute, this translates to a current probe performance degradation rate of approximately 0.0432 mm / min. This rate is higher than the standard wear level of the probe under design specifications (approximately 0.02 mm / min), indicating that the probe is currently in an accelerated wear phase. The system will further update the cumulative wear record based on this and incorporate it into the lifetime function calculation.

[0078] Based on the wear rate estimate and cumulative triggering behavior, the remaining service life of the probe is calculated, and it is determined whether the probe is in a preset critical state range. The specific steps include:

[0079] To accurately estimate the remaining lifespan of the probe, the system first performs statistical processing on the historical usage behavior of the current probe. Specifically, it calls the status parameter acquisition module to record the cumulative number of triggers and cumulative contact duration since the last replacement or calibration of the probe. The cumulative number of triggers refers to the total number of times the probe actually makes physical contact during the participation of coordinate measuring machine, which is accumulated by the trigger frequency sensor each time a trigger action is activated. The cumulative contact duration refers to the sum of the durations of all effective triggering processes, which is obtained by summing the trigger duration parameter through a time series. The above two statistical parameters will serve as the basic information on the current probe usage intensity.

[0080] By combining the current probe wear rate estimate with the output, a remaining lifetime function for lifetime estimation is constructed. This function is based on two parameters: cumulative wear rate and total lifetime budget. The cumulative wear rate is the value obtained by multiplying the estimated wear rate of the probe in each time period since it was put into use by the duration of that time period and then summing them up. This process is achieved by setting a fixed sampling time interval, obtaining the wear rate estimate in each interval, multiplying it with the time length of that interval, and then summing the product results of all intervals to obtain the current cumulative wear value.

[0081] The ratio of the cumulative wear value to the total lifespan budget is used to calculate the lifespan percentage, which measures the ratio between the current lifespan of the probe and its theoretical maximum lifespan. For example, if a probe currently has a cumulative wear value of 20 units and a total lifespan budget of 100 units, the lifespan percentage is 20%, meaning that the probe has consumed approximately one-fifth of its lifespan. This percentage serves as the basis for subsequent status assessment and early warning response, supporting tiered lifespan classification to ensure the reliability and accuracy of coordinate measuring machine (CMM) tasks.

[0082] The specific construction process of the remaining lifetime function is as follows:

[0083] The wear rate estimates continuously output by the block throughout the measurement task are arranged sequentially according to the timestamps associated with each estimate, constructing a wear rate time series table. Each list element contains two fields: one is the wear rate estimate of the probe during that time period, which is a standardized output result, usually between zero and one; the other is the duration of the corresponding time interval, representing the period during which the wear rate continuously affects the measurement task, and the unit can be set to seconds or minutes.

[0084] Perform a time-weighted cumulative operation on the time series list. Specifically, iterate through all time periods in the list, multiply the estimated wear rate of each time period by its duration to represent the actual accumulated wear within that time period. Add the wear rates of all time periods together to obtain the cumulative wear value of the probe at the current moment. For example, if a probe has a wear rate of 0.015 in the first time period (30 seconds) and 0.010 in the second time period (20 seconds), the cumulative wear value is 0.015 × 30 plus 0.010 × 20, resulting in 0.65 units.

[0085] Load the total life budget value corresponding to this probe model. This value is a reference value preset by the equipment manufacturer or user during the system initialization phase. It represents the maximum cumulative wear of the probe within the acceptable error range. It is usually determined based on a large amount of offline experimental data, actual working condition tolerance, and probe material characteristics. For example, for a ball-head probe made of tungsten carbide, the system can set its total life budget value to 50 units.

[0086] Finally, the ratio of the current cumulative wear value to the total lifespan budget is calculated to obtain the probe's lifespan percentage. This percentage represents the proportion of the probe's lifespan already used within its designed total lifespan. For example, if the current cumulative wear value is 10 units and the total lifespan budget is 50 units, then the lifespan percentage is 20%. This value not only visually reflects the extent to which the probe has used up its lifespan but can also be used to determine the amount of measurement work corresponding to the remaining lifespan.

[0087] To classify the current wear status of the probe, based on the aforementioned calculation results of the lifespan percentage, a lifespan status classification mechanism is constructed, and the corresponding lifespan status label is output. The process includes the following steps:

[0088] The system presets a set of lifespan status thresholds to divide the probe's lifespan into three levels: normal state, warning state, and critical state. Each level corresponds to a different degree of wear and a different level of risk. The threshold settings are based on factors such as the material used by the probe (e.g., ceramic, steel alloy, tungsten carbide), the probe structure type (e.g., ball-head, cylindrical, or irregularly shaped probe), and the measurement scenario (e.g., high-precision mold measurement or routine workpiece inspection).

[0089] Two key threshold values ​​are defined: a first threshold and a second threshold. The first threshold distinguishes between normal and warning states, while the second threshold distinguishes between warning and critical states. For example, for a certain type of probe with a total lifespan budget of 50 units, the system can set the first threshold to 0.6 (i.e., 60% of the lifespan) and the second threshold to 0.85 (i.e., 85% of the lifespan). These values ​​can be provided by the equipment manufacturer or set by the user based on their experience with actual measurement tasks.

[0090] The currently measured lifecycle percentage is used as an input parameter and compared sequentially with the two thresholds mentioned above:

[0091] When the lifetime percentage is less than the first threshold, the system determines that the probe is in normal condition and can continue to perform measurement tasks. When the lifetime percentage is greater than or equal to the first threshold but less than the second threshold, the system determines that the probe is in an early warning state, prompting the user to pay attention to the probe wear. When the lifetime percentage is greater than or equal to the second threshold, the system determines that the probe is in a critical state, prompting the user to replace or calibrate it as soon as possible to prevent measurement errors caused by probe performance degradation.

[0092] The above judgment result is output as a life status tag. The tag can be represented in the form of a level identifier (such as Level 1: normal, Level 2: warning, Level 3: critical). The tag is then bound to the measurement task record in which the current probe participates, serving as a reference for data reliability assessment, task scheduling optimization, or subsequent quality traceability.

[0093] This mechanism not only enables quantitative assessment and real-time grading of probe lifetime status, but also establishes a threshold-based judgment logic, improving the safety, accuracy, and intelligent operation and maintenance of coordinate measuring machine (CMM) tasks.

[0094] When the probe is in a critical state, a lifetime warning message is generated, the reliability label of the relevant measurement data is marked, and an abnormal status prompt signal is sent to the operation terminal, realizing dynamic adjustment or suspension control of the measurement task. The specific steps include:

[0095] The system receives the output lifetime status tag, which, based on the lifetime percentage analysis, has been categorized into normal, warning, or critical states. When the current lifetime status is determined to be a warning state, the system assumes the probe can still perform its task, but its wear may have already interfered with some precision measurement processes.

[0096] To this end, the system automatically records the current task measurement point number that is performing measurements under warning status and adds a medium confidence flag to the original measurement data of that measurement point. This flag is used to provide prompts during subsequent data analysis, test report generation, or result review, reminding technicians to pay close attention to or repeat the verification of the measurement point data to enhance the reliability of the measurement results.

[0097] When the system detects that the probe is in a critical state, it triggers a higher-level linkage control strategy. Specifically, the system immediately generates a set of control signal commands to suspend the execution of the current measurement task. These control signals are transmitted to the device controller via the data link, forcibly interrupting the scanning task of subsequent measurement points to avoid probe failure leading to increased measurement errors or equipment damage.

[0098] Simultaneously, the system sends status anomaly alerts to the operating terminal via the communication interface. These alerts include the current probe number, a description of the critical state, and suggested operational instructions. Suggested operations include, but are not limited to: executing the probe replacement procedure, with manual or robotic arm replacement; recalibrating the probe to correct its spatial positioning error; and activating the measurement compensation module to compensate for geometric deviations in the measured data using software algorithms.

[0099] In addition, the system can embed current probe usage history, life assessment records, and measurement point reliability summaries into the prompt messages to provide operators with decision-making support.

[0100] The steps for labeling the confidence level of relevant measurement data include:

[0101] After each measurement point is completed, the system automatically extracts the probe status characteristics bound to that measurement point, including: the estimated current wear rate of the probe, which comes from the output of the wear rate modeling module and represents the degree of decline in probe detection performance per unit time; the probe's lifespan status label, which identifies the probe's current lifespan status (such as normal, warning, or critical); the ambient temperature change value recorded during the measurement process, which indicates the deviation of the temperature at the measurement point from the calibration reference (such as a change of ±3℃); and the measurement timestamp information of the measurement point, which reflects the relative position of the measurement point during the task execution process (such as the point collected 185 seconds after the start of the task).

[0102] The second step involves the system taking parameters from multiple dimensions as input to construct a credibility scoring function. This function is a multi-factor fusion model that outputs a standardized credibility score within a closed interval of 0 to 1. The scoring function internally employs non-linear mapping methods, such as piecewise linear decay and the sigmoid function, to adjust factor sensitivity.

[0103] For example, suppose the estimated wear rate of the probe bound to a certain measuring point is 0.65 (after normalization), the ambient temperature change is +2.5℃, and the measuring point is collected in the middle of the task. The scoring function combines the three factors according to the preset weights and generates a confidence score of 0.72. This score indicates that the measuring point has a medium to high level of data confidence under the current equipment status and environmental conditions.

[0104] The third step involves setting a credibility grading strategy, dividing the score into three levels: high credibility ([0.85, 1.00]), medium credibility ([0.60, 0.85]), and low credibility ([0.00, 0.60]). Based on the above scoring results, a credibility score ≥ 0.85 is marked as high credibility; a score between 0.60 and 0.85 is marked as medium credibility; and a score below 0.60 is marked as low credibility.

[0105] Taking the measurement point with a score of 0.72 in this example as an example, its confidence level is classified as medium confidence. The system will attach this label to the measurement point data structure and record it in the measurement data output file or database field.

[0106] Through the above processing flow, the quantitative evaluation and marking mechanism of the data quality of each measuring point in the coordinate measuring machine task can provide a basis for subsequent process control, result review or defect location, and effectively improve the availability, interpretability and safety fault tolerance of the system data output.

[0107] It should be noted that the total lifespan budget value is a reference value preset by the system at the factory or during the initial calibration stage, representing the theoretical maximum cumulative wear limit that the probe of this model can withstand. This budget value can be set according to the probe material type, structural strength, safety redundancy factor, and measurement conditions. For example, the total lifespan budget value of a certain model of tungsten carbide ball probe can be set to the cumulative wear level after one million trigger cycles.

[0108] The above formulas are all dimensionless calculations. The formulas are derived from software simulations using a large amount of collected data, and are the closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0109] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0110] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0113] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A three-coordinate measurement data intelligent analysis system, characterized in that: It includes a status parameter acquisition module, a feature extraction module, a wear rate modeling module, a probe life prediction module, and a status assessment linkage response module, and the modules are connected by signals. The status parameter acquisition module is used to acquire the operating status parameters of the probe in real time during the coordinate measuring machine task; The feature extraction module is used to normalize, denoise, and align the running status parameters, and extract feature vectors that characterize the probe's usage status to construct a standardized state feature set. The wear rate modeling module constructs a probe wear rate estimation model based on the state feature set and preset historical training samples, and outputs an estimated wear rate of the current probe. The probe lifetime prediction module is used to estimate the remaining lifespan of the probe based on the wear rate and cumulative triggering behavior, and to determine whether the probe is in a preset critical state range. The status assessment linkage response module is used to generate lifespan warning information, mark the credibility of relevant measurement data, and send status abnormality prompt signals to the operation terminal when the probe is in a critical state, so as to realize dynamic adjustment or suspension control of the measurement task.

2. The intelligent analysis system for coordinate measuring data according to claim 1, characterized in that: Operating status parameters include trigger frequency, contact force, trigger duration, displacement change trend, and ambient temperature information.

3. The intelligent analysis system for coordinate measuring data according to claim 2, characterized in that: The steps of the feature extraction module include: Collect probe running status parameters from various dimensions, use a sliding time window mechanism to locally truncate the collected sequence, and construct a time-aligned state subsequence; A nonlinear normalization method is used to compress the value range of various parameters, and the original parameters are uniformly mapped to the standard scoring interval. For noise-sensitive parameters, use median filters or exponential smoothing to improve stability. Finally, a standardized feature vector representing the current state of the probe is output, and a continuously evolving sequence of state features is constructed.

4. The intelligent analysis system for coordinate measuring data according to claim 3, characterized in that: The wear rate modeling module is used to perform the following steps: The historical probe is constructed using a sample set, and the sample features include the correspondence between state feature vectors and actual wear results; The Long Short-Term Memory (LSTM) network is used as the state sequence learning structure to perform temporal modeling on the input state feature sequence. Wear status labels are introduced as supervision signals to perform supervised training and generate wear rate output channels; Generate an estimate of the wear rate at the current moment based on the model output.

5. The intelligent analysis system for coordinate measuring data according to claim 4, characterized in that: Based on the model output, generate an estimate of the wear rate at the current moment, including: The hidden state unit of the probe wear rate estimation model is used to perform feature fusion and correlation analysis on the changes between the current state and the historical state, and output an intermediate vector representation of the current probe health status. The intermediate vector is input into the model output layer. Through a set of trained weights and activation functions, a standardized wear rate score is generated. The wear rate score is located in a closed interval between 0 and 1. The wear rate score is inversely normalized to the known actual wear rate range in the training sample set to restore the wear rate estimate, which is used to indicate the degree of probe performance degradation per unit time.

6. The intelligent analysis system for coordinate measuring data according to claim 5, characterized in that: Used to estimate the remaining lifespan of a probe based on wear rate and cumulative triggering behavior, including: Count the number of triggers and total contact time of the current probe since the last replacement or calibration; Construct a remaining lifetime function based on cumulative wear rate and total lifetime budget, mapping cumulative wear to a percentage of lifetime.

7. The intelligent analysis system for coordinate measuring data according to claim 6, characterized in that: Construct a remaining lifetime function based on cumulative wear rate and total lifetime budget, mapping cumulative wear to a percentage of lifetime, including: The probe was collected to estimate all historical wear rates since it was put into use, and a time series table of wear rates was constructed based on timestamps. Perform a time-weighted cumulative operation on the wear rate time series list, multiply the wear rate of each time period by the duration of that time period and sum them up to obtain the cumulative wear rate at the current moment; Obtain the probe's total lifetime budget value, which is the system's preset maximum allowable wear accumulation rate, used to reflect the probe's maximum usage threshold; The ratio of the cumulative wear rate to the total lifespan budget is calculated to obtain the current probe lifespan percentage, which reflects the ratio between the probe's consumed lifespan and its remaining lifespan.

8. The intelligent analysis system for coordinate measuring data according to claim 7, characterized in that: And determine whether the probe is in a preset critical state range, including: Set a lifetime status level threshold to classify and judge the proportion of probe lifetime. The lifetime status level includes normal status, warning status and critical status. Set a first threshold and a second threshold. The first threshold is used to define the boundary between normal and warning, and the second threshold is used to define the boundary between warning and critical. The two thresholds are configured based on the probe material type, structure model and usage scenario experience. The calculated lifecycle percentage is compared with two thresholds one by one. When the lifecycle percentage is less than the first threshold, it is judged as a normal state; when the lifecycle percentage is greater than or equal to the first threshold and less than the second threshold, it is judged as a warning state; when the lifecycle percentage is greater than or equal to the second threshold, it is judged as a critical state. The determination result is output as a lifetime status label and attached to the probe's current measurement task.

9. The intelligent analysis system for coordinate measuring data according to claim 8, characterized in that: This system is used to generate lifetime warning information when the probe is in a critical state, mark the reliability of relevant measurement data, and send an abnormal status prompt signal to the operation terminal, enabling dynamic adjustment or suspension control of the measurement task, including: Determine the remaining life status label output by the life prediction module. If an early warning state is entered, record the current task measurement point number and add a medium confidence flag to the measurement data. If a critical state is reached, a control signal command is generated to suspend the current task execution process and send a prompt message to the operation terminal, recommending to perform probe replacement, recalibration, or compensation calibration operations.

10. The intelligent analysis system for coordinate measuring data according to claim 9, characterized in that: The steps for labeling the confidence level of relevant measurement data include: Extract the probe status features, wear rate estimate, and lifetime status label associated with each measurement point; construct a scoring function that takes the current probe wear rate, ambient temperature change, and measurement time at the measurement point as combined inputs; Generate a confidence score in the range of 0 to 1 from the combined input values; Based on the credibility score, a grading strategy is set to divide the measurement points into three categories: high credibility, medium credibility, and low credibility, and the credibility labels are attached to the measurement data output structure.

Images