Numerical control lathe tool risk identification system and method based on data analysis

Abstract

The invention discloses a numerical control lathe tool risk identification system and method based on data analysis, and relates to the technical field of data analysis, and the identification method comprises the steps: collecting parameter information and wear data of a tool in a complete use process, and building a life cycle file with a time sequence structure; a wear sequence is constructed by extracting time sequence data of wear parameters, statistical analysis is carried out, a dynamic baseline function is established in combination with an adjustable coefficient, and a wear allowable interval changing along with time is formed; real-time wear data of the tool are obtained, the wear data are compared with the dynamic base line, and the deviation between the current wear data and the dynamic base line is recognized and analyzed; capturing abnormal time points deviating from the dynamic baseline, extracting wear features contained in all the abnormal time points, and generating an abnormal event; for the generated abnormal event, performing cross validation on the abnormal event from a plurality of dimensions, and outputting an abnormal analysis result; the reliability and practicability of the early warning system can be improved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, specifically a CNC lathe tool risk identification system and method based on data analysis. Background Technology

[0002] Life management and fault early warning of CNC lathe tools are key technologies in the field of intelligent manufacturing. At present, tool production lacks systematic tracking and quantitative management of the tool's entire life cycle status, making it difficult to achieve continuous and accurate prediction from initial wear to rapid failure, which affects machining quality and production efficiency.

[0003] Existing technical solutions typically focus on threshold alarms for instantaneous signals, failing to establish a dynamic baseline model that integrates multi-stage wear characteristics. Furthermore, they lack a comprehensive analysis of individual tool differences and machining history archives. This results in a high false alarm rate, poor adaptability, and an inability to perform multi-dimensional cross-validation and root cause tracing for abnormal events. Summary of the Invention

[0004] The purpose of this invention is to provide a data analysis-based CNC lathe tool risk identification system and method to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a data analysis-based method for identifying risks in CNC lathe cutting tools, the identification method comprising:

[0006] Collect parameter information and wear data of the cutting tool during its complete use process, and establish a life cycle archive with a time sequence structure;

[0007] By extracting time-series data of wear parameters, a wear sequence is constructed and statistical analysis is performed. A dynamic baseline function is established by combining an adjustable coefficient, forming a wear allowable range that changes over time.

[0008] Acquire real-time tool wear data, compare the wear data with a dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline;

[0009] Capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events;

[0010] For the generated abnormal events, cross-validation is performed on the abnormal events from several dimensions, and the abnormal analysis results are output.

[0011] Furthermore, lifecycle profile establishment includes:

[0012] Several physical parameters are preset for the cutting tool, and a monitoring system is pre-installed on the lathe to monitor the tool's usage process. At each unit time point, arbitrary physical parameters are collected to obtain their values ​​at each unit time point, and a set of parameter values ​​for each physical parameter is generated in chronological order. The set of parameter values ​​for each physical parameter during the complete use of the tool is then summarized to generate a corresponding usage record. The physical parameters of the cutting tool include cutting force, vibration acceleration, torque, cutting temperature, tool temperature, etc.

[0013] A pre-defined wear parameter database stores several wear parameters. For each wear parameter, a corresponding set of physical parameters and related data processing rules are matched. A wear parameter is arbitrarily selected from the database, and the data processing rules for that selected parameter are retrieved to process the relevant physical parameters in the usage record, resulting in wear data for the selected parameter. For example, for tool wear rate, the corresponding physical parameters are cutting temperature, vibration amplitude, and cutting force. The data processing rules assign appropriate coefficients to these three parameters, then perform logarithmic and square root operations on the parameters themselves, and finally sum the results using the coefficients to obtain the wear rate. Another example is the cutting edge degradation index, corresponding to acoustic emission signal amplitude, surface roughness, and spindle current fluctuation. The data processing rules normalize these three parameters and calculate their average value to obtain the degradation index. And so on.

[0014] If there are several wear data points for the selected wear parameters in the record, then sort the wear data points according to the order of the unit time points to obtain the wear data change process of the selected wear parameters.

[0015] The tool lifecycle archive is constructed by structuring the set of physical parameter values ​​and the wear data change process of each wear parameter in any usage record in chronological order.

[0016] Furthermore, dynamic baseline establishment includes:

[0017] Based on wear data in the tool lifecycle archive, a sliding statistical analysis of the wear sequence is performed by constructing a time window, calculating the comprehensive wear mean and standard deviation at each time point, and constructing a dynamic baseline function with an adjustable coefficient to form the allowable wear range that changes over time.

[0018] Furthermore, dynamic baseline establishment also includes:

[0019] Select any usage record and retrieve its lifecycle file. Extract the wear data changes of each wear parameter from the lifecycle file. Select any wear parameter and generate a wear sequence over time from the wear data changes of the selected wear parameter. Let the wear sequence of the i-th wear parameter be S. i (t)={x i (1),x i (2),…,x i (t),…,x i (a)},x i (t) represents the wear data of the i-th wear parameter at the t-th unit time point, and a represents the number of unit time points contained in the life cycle file;

[0020] Construct a time window with a window length of L and a sliding step of one unit time point. Slide the time window starting from the first unit time point and arbitrarily select the t-th unit time point. Set the time length between any two adjacent unit time points to Δr. When Δr×(t-1)<L, statistically generate the mean and standard deviation of the wear data of the first t unit time points. When Δr×(t-1)≥L, statistically generate the mean and standard deviation of the wear data of all unit time points in the time window and set them as the mean and standard deviation of the t-th unit time point.

[0021] The mean and standard deviation of the i-th wear parameter in each usage record at the t-th unit time point are obtained, and the average value is calculated to obtain the comprehensive wear mean μ of the i-th wear parameter at the t-th unit time point. i (t) and the comprehensive wear standard deviation σ i (t); A preset adjustable coefficient α is used to construct the dynamic baseline function for the i-th wear parameter:

[0022] ;

[0023] The dynamic baseline function of each wear parameter is aligned and correlated with the time axis to form a wear allowable range that changes with the unit time point; the dynamic baseline is automatically adjusted with the usage time to adapt to factors such as individual differences of different tools and changes in machining materials, thereby reducing false alarms.

[0024] Further, wear data deviation identification includes:

[0025] The current tool usage process is set as a real-time usage record. Real-time wear data of the tool at the current time point is acquired, and wear data from previous unit time points during the usage process is also acquired to generate a real-time wear sequence for each wear parameter. The i-th wear parameter is arbitrarily selected, and the dynamic baseline function constructed using the i-th wear parameter is retrieved. The current time point in the real-time usage record is set as the t-th unit time point. The allowable wear interval for the t-th unit time point in the dynamic baseline function is obtained as [μ]. i (t)-α×σ i (t),μ i (t)+α×σ i [(t)], if the real-time wear data is within the allowable wear range, it is judged as normal wear; if the real-time wear data exceeds the allowable wear range, it is judged as abnormal deviation.

[0026] If there are abnormal deviations in the real-time wear data at the current time point, then according to the formula:

[0027] ;

[0028] Among them, y is set i (t) represents the real-time wear data at the current time point, μ i (t) represents the average wear value of the i-th wear parameter at the t-th unit time point, σ i (t) represents the comprehensive wear standard deviation of the i-th wear parameter at the t-th unit time point, and α is an adjustable coefficient; the wear deviation P of the i-th wear parameter at the current time point is calculated. i (t);

[0029] Let Q be the actual deviation at the current time point. i (t), if y i (t)<μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a negative abnormal deviation and Q i (t)=-P i (t), if y i (t)≥μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a positive abnormal deviation and Q i (t)=P i (t).

[0030] Further, anomaly event identification includes:

[0031] Extract abnormal time points from any usage record, obtain parameter values ​​and wear data for each abnormal time point, and compare them with preset thresholds to determine abnormal parameters; based on the occurrence of abnormal parameters in consecutive abnormal time points, distinguish between abnormal time intervals and independent abnormal points, and determine whether they constitute wear characteristics to identify abnormal events.

[0032] Furthermore, anomaly event identification also includes:

[0033] Select any tool usage record, extract each unit time point in the selected usage record that has abnormal deviations, and set it as an abnormal time point; select any abnormal time point, obtain the parameter values ​​of each physical parameter and the wear data of each wear parameter in the selected abnormal time point, preset the corresponding parameter thresholds for each physical parameter, and also preset the corresponding data thresholds for the wear data of each wear parameter.

[0034] If the physical parameter value or wear data of the wear parameter exceeds the preset data threshold in the selected abnormal time point, the corresponding physical parameter or wear parameter is set as an abnormal parameter, and the abnormal parameter set of the selected abnormal time point is obtained.

[0035] The abnormal parameter set of each abnormal time point in the selected usage record is extracted. If a certain abnormal parameter exists in several adjacent and consecutive abnormal time points, an abnormal time interval for the certain abnormal parameter is generated. Otherwise, the abnormal time point is set as an independent abnormal point, thus obtaining the abnormal division of the certain abnormal parameter. A time interval threshold and an independent number threshold are preset. If the interval length of an abnormal time interval exceeds the time interval threshold, the certain abnormal parameter is set as a wear feature. If the abnormal time intervals do not exceed the time interval threshold, but the number of independent abnormal points exceeds the independent number threshold, the certain abnormal parameter is also set as a wear feature, thus generating several wear features for the selected usage record.

[0036] An abnormal event database is pre-set and stores several abnormal events. Each abnormal event is assigned a corresponding wear feature set. Several wear features of the selected usage record are compared with each wear feature set. The abnormal event corresponding to the wear feature set containing the several wear features is set as the abnormal event that causes the selected usage record to be abnormal.

[0037] Furthermore, the anomaly analysis results output includes:

[0038] Randomly select an abnormal event from the usage record, acquire each wear feature in the selected abnormal event, and randomly extract one wear feature to obtain the abnormal parameter corresponding to the extracted wear feature. At the same time, obtain the actual deviation degree of the corresponding abnormal parameter at each abnormal time point. Accumulate the actual deviation degree of each abnormal time point to obtain a reference value. If the reference value is positive, count the number of abnormal time points with positive actual deviation degree. If the reference value is negative, count the number of abnormal time points with negative actual deviation degree. Obtain the abnormal number percentage of the counted abnormal time points.

[0039] A threshold for the percentage of abnormal numbers is preset. If the percentage of abnormal numbers extracted from wear features exceeds the threshold, the extracted wear features are set as risk features, and risk anomaly warnings are issued for each risk feature.

[0040] A risk identification system for CNC lathe tools, comprising a periodic record acquisition module, a dynamic baseline establishment module, a baseline deviation analysis module, an abnormal event identification module, and an abnormal result verification module;

[0041] The lifecycle record acquisition module is used to collect parameter information and wear data of the tool during its complete use process and establish a lifecycle record with a time sequence structure.

[0042] The dynamic baseline establishment module is used to construct a wear sequence by extracting time-series data of wear parameters and perform statistical analysis. It combines adjustable coefficients to establish a dynamic baseline function, forming a wear allowable range that changes over time.

[0043] The baseline deviation analysis module is used to acquire real-time tool wear data, compare the wear data with the dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline.

[0044] The abnormal event identification module is used to capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events.

[0045] The anomaly result verification module is used to cross-validate the generated anomaly events from several dimensions and output the anomaly analysis results.

[0046] Furthermore, the dynamic baseline establishment module includes a wear data analysis unit and a dynamic baseline generation unit;

[0047] The wear data analysis unit is used to construct wear sequences and perform statistical analysis by extracting time-series data of wear parameters; the dynamic baseline generation unit is used to establish a dynamic baseline function by combining adjustable coefficients to form a wear allowable range that changes over time.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] 1. This invention enables dynamic tracking of the entire process of a tool from initial wear to failure by constructing a lifecycle archive. It can establish a dynamic baseline that changes over time based on historical data, significantly improving the estimation accuracy of the remaining tool life and reducing downtime losses caused by sudden failures.

[0050] 2. This invention employs a sliding time window statistical analysis of wear sequences to construct a dynamic baseline function, forming a wear tolerance range that varies with usage time. This overcomes the problem of poor adaptability to individual differences and changes in operating conditions, significantly reducing the false alarm rate and improving the reliability and practicality of the early warning system.

[0051] 3. By capturing abnormal time points that deviate from the dynamic baseline and extracting abnormal parameters and their time distribution characteristics, this invention can distinguish between continuous abnormal intervals and independent abnormal points, thereby identifying wear characteristics with practical significance and associating them with specific abnormal events. Attached Figure Description

[0052] Figure 1 A schematic diagram illustrating the steps of a data analysis-based method for identifying risks in CNC lathe cutting tools;

[0053] Figure 2 This is a schematic diagram of a data analysis-based CNC lathe tool risk identification system. Detailed Implementation

[0054] 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.

[0055] Example: Figures 1 to 2 As shown, this invention provides a data analysis-based method for identifying risks in CNC lathe cutting tools. The identification method includes:

[0056] Collect parameter information and wear data of the cutting tool during its complete use process, and establish a life cycle archive with a time sequence structure;

[0057] The establishment of lifecycle profiles includes:

[0058] Several physical parameters are preset for the cutting tool, and a monitoring system is pre-installed on the lathe to monitor the tool's usage process. At each unit time point, the parameters of any physical parameter are collected to obtain the parameter values ​​of any physical parameter at each unit time point, and the parameter value set of each physical parameter is generated in the order of the unit time points. The parameter value set of each physical parameter is summarized during the complete use of the tool to generate a corresponding usage record.

[0059] A wear parameter database is pre-defined, which stores several wear parameters. For each wear parameter, a corresponding set of physical parameters and related data processing rules are matched. A wear parameter is arbitrarily selected from the wear parameter database, and the data processing rules for the selected wear parameter are retrieved to process the related physical parameters in the usage record to obtain the wear data of the selected wear parameter.

[0060] If there are several wear data points for the selected wear parameters in the record, then sort the wear data points according to the order of the unit time points to obtain the wear data change process of the selected wear parameters.

[0061] The tool lifecycle archive is constructed by structuring the set of parameter values ​​of each physical parameter and the wear data change process of each wear parameter in any usage record in chronological order.

[0062] Example 1: When a CNC lathe is machining stainless steel parts, it uses carbide cutting tools. The system's preset physical parameters include cutting force, vibration amplitude, tool temperature, and acoustic emission signal. The monitoring system continuously collects the values ​​of the above parameters every 5 seconds. After a complete machining cycle of 8 hours, the system obtains the time series data of each physical parameter, forming a usage record. A wear parameter database is preset, such as the wear parameter "flank wear". The corresponding physical parameters are "cutting force" and "vibration amplitude". The data processing rule is weighted summation. The time series of these two physical parameters are extracted from the usage record, and the wear data sequence of "flank wear" changing over time is calculated according to the rules.

[0063] By extracting time-series data of wear parameters, a wear sequence is constructed and statistical analysis is performed. A dynamic baseline function is established by combining an adjustable coefficient, forming a wear allowable range that changes over time.

[0064] The establishment of dynamic baselines includes:

[0065] Based on wear data in the tool lifecycle archive, a sliding statistical analysis of the wear sequence is performed by constructing a time window, calculating the comprehensive wear mean and standard deviation at each time point, and constructing a dynamic baseline function with an adjustable coefficient to form the allowable wear range that changes over time.

[0066] The establishment of dynamic baselines also includes:

[0067] Select any usage record and retrieve its lifecycle file. Extract the wear data changes of each wear parameter from the lifecycle file. Select any wear parameter and generate a wear sequence over time from the wear data changes of the selected wear parameter. Let the wear sequence of the i-th wear parameter be S. i (t)={x i (1),x i (2),…,x i (t),…,x i (a)},x i (t) represents the wear data of the i-th wear parameter at the t-th unit time point, and a represents the number of unit time points contained in the life cycle file;

[0068] Construct a time window with a window length of L and a sliding step of one unit time point. Slide the time window starting from the first unit time point and arbitrarily select the t-th unit time point. Set the time length between any two adjacent unit time points to Δr. When Δr×(t-1)<L, statistically generate the mean and standard deviation of the wear data of the first t unit time points. When Δr×(t-1)≥L, statistically generate the mean and standard deviation of the wear data of all unit time points in the time window and set them as the mean and standard deviation of the t-th unit time point.

[0069] The mean and standard deviation of the i-th wear parameter in each usage record at the t-th unit time point are obtained, and the average value is calculated to obtain the comprehensive wear mean μ of the i-th wear parameter at the t-th unit time point. i (t) and the comprehensive wear standard deviation σ i (t); A preset adjustable coefficient α is used for the dynamic baseline function B of the i-th wear parameter. i (t) is used for construction:

[0070] ;

[0071] The dynamic baseline functions of each wear parameter are aligned and correlated with the time axis to form a wear allowable range that varies with a unit time point;

[0072] Example 2: Extract the wear sequence of "flank wear" from the above lifecycle archive, set the time window length L=60s, and the sliding step size is 1s; perform sliding statistics on the wear data within the time window, and calculate the mean μ(t) and standard deviation σ(t) at each time point t. For example: when t≤60, use the first t data points for calculation; when t>60, use the most recent 60 data points for calculation; assuming 10 historical usage records are collected, the system averages μ(t) and σ(t) at time t in each record to obtain the comprehensive wear mean μ. i (t) = 0.2 mm and the comprehensive wear standard deviation σ i (t) = 0.02 mm; with the adjustable coefficient α = 2, the dynamic baseline function is calculated as B. i (t)=[0.2-2×0.02,0.2+2×0.02]=[0.16,0.24].

[0073] Acquire real-time tool wear data, compare the wear data with a dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline;

[0074] Among them, wear data deviation identification includes:

[0075] The current tool usage process is set as a real-time usage record. Real-time wear data of the tool at the current time point is acquired, and wear data from previous unit time points during the usage process is also acquired to generate a real-time wear sequence for each wear parameter. The i-th wear parameter is arbitrarily selected, and the dynamic baseline function constructed using the i-th wear parameter is retrieved. The current time point in the real-time usage record is set as the t-th unit time point. The allowable wear interval for the t-th unit time point in the dynamic baseline function is obtained as [μ]. i (t)-α×σ i (t),μ i (t)+α×σ i [(t)], if the real-time wear data is within the allowable wear range, it is judged as normal wear; if the real-time wear data exceeds the allowable wear range, it is judged as abnormal deviation.

[0076] If there are abnormal deviations in the real-time wear data at the current time point, then according to the formula:

[0077] ;

[0078] Among them, y is set i (t) represents the real-time wear data at the current time point, μ i (t) represents the average wear value of the i-th wear parameter at the t-th unit time point, σ i(t) represents the comprehensive wear standard deviation of the i-th wear parameter at the t-th unit time point, and α is an adjustable coefficient; the wear deviation P of the i-th wear parameter at the current time point is calculated. i (t);

[0079] Let Q be the actual deviation at the current time point. i (t), if y i (t)<μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a negative abnormal deviation and Q i (t)=-P i (t), if y i (t)≥μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a positive abnormal deviation and Q i (t)=P i (t);

[0080] Example 3: The tool is currently machining. The system collects the back face wear at t=500 in real time, which is 0.25mm. The dynamic baseline allowable range at this time is retrieved as [0.16mm, 0.24mm]. The wear deviation is calculated as P=(0.25-0.2-2×0.02) / 0.04=0.25. The actual deviation is Q=P=0.25 and is a positive abnormal deviation.

[0081] Capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events;

[0082] Among them, abnormal event identification includes:

[0083] Extract abnormal time points from any usage record, obtain parameter values ​​and wear data for each abnormal time point, and compare them with preset thresholds to determine abnormal parameters; based on the occurrence of abnormal parameters in consecutive abnormal time points, distinguish between abnormal time intervals and independent abnormal points, and determine whether they constitute wear characteristics to identify abnormal events.

[0084] Anomaly identification also includes:

[0085] Select any tool usage record, extract each unit time point in the selected usage record that has abnormal deviations, and set it as an abnormal time point; select any abnormal time point, obtain the parameter values ​​of each physical parameter and the wear data of each wear parameter in the selected abnormal time point, preset the corresponding parameter thresholds for each physical parameter, and also preset the corresponding data thresholds for the wear data of each wear parameter.

[0086] If the physical parameter value or wear data of the wear parameter exceeds the preset data threshold in the selected abnormal time point, the corresponding physical parameter or wear parameter is set as an abnormal parameter, and the abnormal parameter set of the selected abnormal time point is obtained.

[0087] The abnormal parameter set of each abnormal time point in the selected usage record is extracted. If a certain abnormal parameter exists in several adjacent and consecutive abnormal time points, an abnormal time interval for the certain abnormal parameter is generated. Otherwise, the abnormal time point is set as an independent abnormal point, thus obtaining the abnormal division of the certain abnormal parameter. A time interval threshold and an independent number threshold are preset. If the interval length of an abnormal time interval exceeds the time interval threshold, the certain abnormal parameter is set as a wear feature. If the abnormal time intervals do not exceed the time interval threshold, but the number of independent abnormal points exceeds the independent number threshold, the certain abnormal parameter is also set as a wear feature, thus generating several wear features for the selected usage record.

[0088] An abnormal event database is pre-set and stores several abnormal events. Each abnormal event is assigned a corresponding wear feature set. Several wear features of the selected usage record are compared with each wear feature set. The abnormal event corresponding to the wear feature set containing the several wear features is set as the abnormal event that causes the selected usage record to be abnormal.

[0089] For the generated abnormal events, cross-validation is performed on the abnormal events from several dimensions, and the abnormality analysis results are output.

[0090] The anomaly analysis results output includes:

[0091] Randomly select an abnormal event from the usage record, acquire each wear feature in the selected abnormal event, and randomly extract one wear feature to obtain the abnormal parameter corresponding to the extracted wear feature. At the same time, obtain the actual deviation degree of the corresponding abnormal parameter at each abnormal time point. Accumulate the actual deviation degree of each abnormal time point to obtain a reference value. If the reference value is positive, count the number of abnormal time points with positive actual deviation degree. If the reference value is negative, count the number of abnormal time points with negative actual deviation degree. Obtain the abnormal number percentage of the counted abnormal time points.

[0092] A threshold for the percentage of abnormal numbers is preset. If the percentage of abnormal numbers extracted from wear features exceeds the threshold, the extracted wear features are set as risk features, and risk anomaly warnings are issued for each risk feature.

[0093] A risk identification system for CNC lathe tools, comprising a periodic record acquisition module, a dynamic baseline establishment module, a baseline deviation analysis module, an abnormal event identification module, and an abnormal result verification module;

[0094] The lifecycle record acquisition module is used to collect parameter information and wear data of the tool during its complete use process and establish a lifecycle record with a time sequence structure.

[0095] The dynamic baseline establishment module is used to construct a wear sequence by extracting time-series data of wear parameters and perform statistical analysis. It combines adjustable coefficients to establish a dynamic baseline function, forming a wear allowable range that changes over time.

[0096] The baseline deviation analysis module is used to acquire real-time tool wear data, compare the wear data with the dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline.

[0097] The abnormal event identification module is used to capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events.

[0098] The anomaly result verification module is used to cross-validate the generated anomaly events from several dimensions and output the anomaly analysis results.

[0099] The dynamic baseline establishment module includes a wear data analysis unit and a dynamic baseline generation unit.

[0100] The wear data analysis unit is used to construct wear sequences and perform statistical analysis by extracting time-series data of wear parameters; the dynamic baseline generation unit is used to establish a dynamic baseline function by combining adjustable coefficients to form a wear allowable range that changes over time.

[0101] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A data analysis-based method for identifying risks in CNC lathe cutting tools, characterized in that: The identification method includes: Collect parameter information and wear data of the cutting tool during its complete use process, and establish a life cycle archive with a time sequence structure; By extracting time-series data of wear parameters, a wear sequence is constructed and statistical analysis is performed. A dynamic baseline function is established by combining an adjustable coefficient, forming a wear allowable range that changes over time. Acquire real-time tool wear data, compare the wear data with a dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline; Capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events; For the generated abnormal events, cross-validation is performed on the abnormal events from several dimensions, and the abnormal analysis results are output.

2. The method for identifying CNC lathe tool risks through data analysis according to claim 1, characterized in that: Lifecycle profile creation includes: Several physical parameters are preset for the cutting tool, and a monitoring system is pre-installed on the lathe to monitor the tool's usage process. At each unit time point, the parameters of any physical parameter are collected to obtain the parameter values ​​of any physical parameter at each unit time point, and the parameter value set of each physical parameter is generated in the order of the unit time points. The parameter value set of each physical parameter is summarized during the complete use of the tool to generate a corresponding usage record. A wear parameter database is pre-defined, which stores several wear parameters. For each wear parameter, a corresponding set of physical parameters and related data processing rules are matched. A wear parameter is arbitrarily selected from the wear parameter database, and the data processing rules for the selected wear parameter are retrieved to process the related physical parameters in the usage record to obtain the wear data of the selected wear parameter. If there are several wear data points for the selected wear parameters in the record, then sort the wear data points according to the order of the unit time points to obtain the wear data change process of the selected wear parameters. The tool lifecycle archive is constructed by structuring the set of physical parameter values ​​and the wear data change process of each wear parameter in any usage record in chronological order.

3. The method for risk identification of CNC lathe tools based on data analysis according to claim 1, characterized in that: Dynamic baseline establishment includes: Based on wear data in the tool lifecycle archive, a sliding statistical analysis of the wear sequence is performed by constructing a time window, calculating the comprehensive wear mean and standard deviation at each time point, and constructing a dynamic baseline function with an adjustable coefficient to form the allowable wear range that changes over time.

4. The method for identifying CNC lathe tool risks through data analysis according to claim 3, characterized in that: Dynamic baseline establishment also includes: Select any usage record and retrieve its lifecycle file. Extract the wear data changes of each wear parameter from the lifecycle file. Select any wear parameter and generate a wear sequence over time from the wear data changes of the selected wear parameter. Let the wear sequence of the i-th wear parameter be S. i (t)={x i (1),x i (2),…,x i (t),…,x i (a)},x i (t) represents the wear data of the i-th wear parameter at the t-th unit time point, and a represents the number of unit time points contained in the life cycle file; Construct a time window with a window length of L and a sliding step of one unit time point. Slide the time window starting from the first unit time point and arbitrarily select the t-th unit time point. Set the time length between any two adjacent unit time points to Δr. When Δr×(t-1)<L, statistically generate the mean and standard deviation of the wear data of the first t unit time points. When Δr×(t-1)≥L, statistically generate the mean and standard deviation of the wear data of all unit time points in the time window and set them as the mean and standard deviation of the t-th unit time point. The mean and standard deviation of the i-th wear parameter in each usage record at the t-th unit time point are obtained, and the average value is calculated to obtain the comprehensive wear mean μ of the i-th wear parameter at the t-th unit time point. i (t) and the comprehensive wear standard deviation σ i (t); A preset adjustable coefficient α is used to construct the dynamic baseline function for the i-th wear parameter: ； The dynamic baseline functions of each wear parameter are aligned and correlated with the time axis to form a wear allowable range that varies with a unit time point.

5. The method for identifying CNC lathe tool risks through data analysis according to claim 1, characterized in that: Wear data deviation identification includes: The current tool usage process is set as a real-time usage record. Real-time wear data of the tool at the current time point is acquired, and wear data from previous unit time points during the usage process is also acquired to generate a real-time wear sequence for each wear parameter. The i-th wear parameter is arbitrarily selected, and the dynamic baseline function constructed using the i-th wear parameter is retrieved. The current time point in the real-time usage record is set as the t-th unit time point. The allowable wear interval for the t-th unit time point in the dynamic baseline function is obtained as [μ]. i (t)-α×σ i (t),μ i (t)+α×σ i [(t)], if the real-time wear data is within the allowable wear range, it is judged as normal wear; if the real-time wear data exceeds the allowable wear range, it is judged as abnormal deviation. If there are abnormal deviations in the real-time wear data at the current time point, then according to the formula: ； Among them, y is set i (t) represents the real-time wear data at the current time point, μ i (t) represents the average wear value of the i-th wear parameter at the t-th unit time point, σ i (t) represents the comprehensive wear standard deviation of the i-th wear parameter at the t-th unit time point, and α is an adjustable coefficient; the wear deviation P of the i-th wear parameter at the current time point is calculated. i (t); Let Q be the actual deviation at the current time point. i (t), if y i (t)<μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a negative abnormal deviation and Q i (t)=-P i (t), if y i (t)≥μ i (t)-α×σ i (t), then the abnormal deviation is determined to be a positive abnormal deviation and Q i (t)=P i (t).

6. The method for identifying CNC lathe tool risks through data analysis according to claim 1, characterized in that: Abnormal event identification, including: Extract abnormal time points from any usage record, obtain parameter values ​​and wear data for each abnormal time point, and compare them with preset thresholds to determine abnormal parameters; based on the occurrence of abnormal parameters in consecutive abnormal time points, distinguish between abnormal time intervals and independent abnormal points, and determine whether they constitute wear characteristics to identify abnormal events.

7. The method for identifying CNC lathe tool risks through data analysis according to claim 6, characterized in that: Abnormal event identification also includes: Select any tool usage record, extract each unit time point in the selected usage record that has abnormal deviations, and set it as an abnormal time point; select any abnormal time point, obtain the parameter values ​​of each physical parameter and the wear data of each wear parameter in the selected abnormal time point, preset the corresponding parameter thresholds for each physical parameter, and also preset the corresponding data thresholds for the wear data of each wear parameter. If the physical parameter value or wear data of the wear parameter exceeds the preset data threshold in the selected abnormal time point, the corresponding physical parameter or wear parameter is set as an abnormal parameter, and the abnormal parameter set of the selected abnormal time point is obtained. The abnormal parameter set of each abnormal time point in the selected usage record is extracted. If a certain abnormal parameter exists in several adjacent and consecutive abnormal time points, an abnormal time interval for the certain abnormal parameter is generated. Otherwise, the abnormal time point is set as an independent abnormal point, thus obtaining the abnormal division of the certain abnormal parameter. A time interval threshold and an independent number threshold are preset. If the interval length of an abnormal time interval exceeds the time interval threshold, the certain abnormal parameter is set as a wear feature. If the abnormal time intervals do not exceed the time interval threshold, but the number of independent abnormal points exceeds the independent number threshold, the certain abnormal parameter is also set as a wear feature, thus generating several wear features for the selected usage record. An abnormal event database is pre-set and stores several abnormal events. Each abnormal event is assigned a corresponding wear feature set. Several wear features of the selected usage record are compared with each wear feature set. The abnormal event corresponding to the wear feature set containing the several wear features is set as the abnormal event that causes the selected usage record to be abnormal.

8. The method for risk identification of CNC lathe cutting tools based on data analysis according to claim 1, characterized in that: The anomaly analysis results output includes: Randomly select an abnormal event from the usage record, acquire each wear feature in the selected abnormal event, and randomly extract one wear feature to obtain the abnormal parameter corresponding to the extracted wear feature. At the same time, obtain the actual deviation degree of the corresponding abnormal parameter at each abnormal time point. Accumulate the actual deviation degree of each abnormal time point to obtain a reference value. If the reference value is positive, count the number of abnormal time points with positive actual deviation degree. If the reference value is negative, count the number of abnormal time points with negative actual deviation degree. Obtain the abnormal number percentage of the counted abnormal time points. A threshold for the percentage of abnormal numbers is preset. If the percentage of abnormal numbers extracted from wear features exceeds the threshold, the extracted wear features are set as risk features, and risk anomaly warnings are issued for each risk feature.

9. A CNC lathe tool risk identification system, used to perform a CNC lathe tool risk identification method based on data analysis according to any one of claims 1-8, characterized in that: The identification system includes a periodic archive acquisition module, a dynamic baseline establishment module, a baseline deviation analysis module, an abnormal event identification module, and an abnormal result verification module. The lifecycle file acquisition module is used to collect parameter information and wear data of the tool during its complete use process and establish a lifecycle file with a time sequence structure. The dynamic baseline establishment module is used to construct a wear sequence by extracting time-series data of wear parameters and perform statistical analysis, and to establish a dynamic baseline function by combining an adjustable coefficient to form a wear allowable range that changes over time. The baseline deviation analysis module is used to acquire real-time tool wear data, compare the wear data with the dynamic baseline, and identify and analyze the deviation between the current wear data and the dynamic baseline. The abnormal event identification module is used to capture abnormal time points that deviate from the dynamic baseline, extract the wear features contained in all abnormal time points, and generate abnormal events. The anomaly result verification module is used to cross-validate the generated anomaly events from several dimensions and output the anomaly analysis results.

10. A CNC lathe tool risk identification system according to claim 9, characterized in that: The dynamic baseline establishment module includes a wear data analysis unit and a dynamic baseline generation unit; The wear data analysis unit is used to construct a wear sequence by extracting time-series data of wear parameters and perform statistical analysis; the dynamic baseline generation unit is used to establish a dynamic baseline function by combining an adjustable coefficient to form a wear allowable range that changes over time.

Images