Tool state monitoring and method, system and electronic device for constructing a model thereof

Abstract

The present disclosure provides a tool state monitoring method and a model construction method, system and electronic equipment thereof. The model construction method comprises: acquiring machining data under normal working conditions of a tool, the machining data comprising a power signal of a tool spindle; segmenting the power signal according to a segmentation line, and determining a machining period of the tool according to the segmentation result; the segmentation line is used to define the machining state and the non-machining state of the tool; extracting characteristic indexes of the machining data in each machining period; determining a tool state monitoring threshold of each machining period according to the characteristic values of the characteristic indexes, and constructing a tool state monitoring model according to the tool state monitoring threshold. The present disclosure can effectively eliminate the influence of non-machining period interference and cross-condition cross-coupling, so that the tool state monitoring model constructed has stronger working condition self-adaptive ability, realizes accurate evaluation and real-time monitoring of the tool state, and significantly improves the accuracy and timeliness of the monitoring result.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of industrial manufacturing technology, and particularly relates to a tool state monitoring method and system and an electronic device. BACKGROUND

[0002] With the development of manufacturing industry towards high-end and intelligence, high-end numerical control machine tools as the strategic cornerstone of supporting industrial upgrading, the intelligent level of its core machining process has become a key bottleneck restricting the performance and reliability of equipment. Therefore, it is urgent to carry out key technology research to improve the intelligent level of high-end numerical control equipment. Among them, the tool state online monitoring technology is one of the core links to realize the intelligent perception and control of the machining process. Through real-time perception and intelligent diagnosis of tool wear, damage and other states in the machining process, it can effectively guarantee the machining quality, optimize the machining efficiency and prevent equipment damage, which is the key to break through the current intelligent short board.

[0003] In the research and industrial application of tool state real-time monitoring technology, how to accurately and efficiently perceive and obtain tool state information in the machining process is the core prerequisite for reliable monitoring. According to the difference of data acquisition method, the existing technical system is mainly divided into two technical paths of direct measurement method and indirect measurement method.

[0004] The core idea of direct measurement method is to directly observe or measure the physical appearance of tool, such as using optical imaging method, contact probe measurement and other methods to directly obtain the state information of tool wear form, broken blade position, damage degree, etc. The advantage of this method is that the obtained state information is intuitive and the diagnosis result is clear, but in the implementation process, the machining process usually needs to be paused (i.e. stop detection) to complete the static contact measurement or imaging observation of the tool. This feature makes it unable to meet the high real-time online response demand of sudden tool damage, tool and workpiece collision and other emergency events in industrial scene, and the application scene is greatly limited.

[0005] Indirect measurement method is the mainstream research direction and industrial application scheme of current tool state real-time monitoring technology, its core principle is not to directly observe the tool itself, but to deploy various industrial sensors to collect the physical signals that are strongly related to the tool state and can be obtained online in real time, and to infer and evaluate the running state of the tool by analyzing the characteristic change rule of indirect signals. This method can realize continuous monitoring of tool state without interrupting the machining process, with the core advantages of no need to stop, strong real-time and continuous monitoring, and becomes the main data source and technical basis for building practical tool state intelligent monitoring system.

[0006] However, the current tool state monitoring method based on indirect signals still faces technical bottlenecks in actual industrial applications, which are specifically manifested as follows: the mapping relationship between the tool state and the signal is complex, different states of the tool wear, damage and the like have significant nonlinear and time-varying coupling characteristics with the multi-physical quantity signal, it is difficult to establish a precise and universal state diagnosis model, thereby seriously restricting the reliability, diagnosis accuracy and universality of the existing tool state monitoring, and leading to the difficulty in realizing large-scale industrial applications. SUMMARY

[0007] The technical problem to be solved by the present disclosure is to overcome the defects in the prior art that the tool state monitoring method based on indirect signals has significant nonlinear and time-varying coupling characteristics between the tool state and the physical quantity signal, and it is difficult to establish a precise and universal state diagnosis model, and to provide a tool state monitoring method and a system for constructing a tool state monitoring model.

[0008] The present disclosure solves the above technical problems by the following technical solutions:

[0009] In a first aspect, a method for constructing a tool state monitoring model is provided, comprising:

[0010] Obtaining machining data of the tool in a normal working condition, the machining data comprising a power signal of a tool spindle;

[0011] Segmenting the power signal according to a segmentation line, and determining machining periods of the tool according to the segmentation result; the segmentation line is used to define the machining state and the non-machining state of the tool;

[0012] Extracting feature indexes of the machining data in each machining period;

[0013] Determining tool state monitoring thresholds of each machining period according to characteristic values of the feature indexes, and constructing a tool state monitoring model according to the tool state monitoring thresholds; the tool state monitoring model is used to analyze and determine real-time machining data of the tool, so as to monitor the state of the tool.

[0014] Optionally, the method further comprises:

[0015] Smoothly processing the power signal by using a first sliding window with gradually increasing size to obtain a power sequence; each element of the power sequence corresponds to a calculation result of a first sliding window;

[0016] Performing quantile calculation on the power sequence, and determining the segmentation line according to the calculation result.

[0017] Optionally, before the step of performing quantile calculation on the power sequence, the method further comprises:

[0018] determining whether the number of elements in the power sequence reaches a target number;

[0019] in response to the number of elements not reaching the target number, padding the number of elements in the power sequence to the target number by using a last element in the power sequence;

[0020] the step of performing quantile calculation on the power sequence comprises:

[0021] performing quantile calculation on the padded power sequence.

[0022] Optionally, the feature indicators of the machining data in each machining period are extracted, comprising:

[0023] the machining data in the machining period is segmented and windowed by using a second sliding window, and the feature indicators of the machining data in each second sliding window are calculated; wherein the window length of the second sliding window is negatively correlated with the rotational speed of the spindle, and / or the step length of the second sliding window is negatively correlated with the rotational speed of the spindle.

[0024] Optionally, the window length of the second sliding window is determined by the following steps:

[0025] an initial window length is determined according to the rotational speed of the spindle;

[0026] it is determined whether the initial window length is greater than a preset minimum window length;

[0027] in response to the initial window length being greater than the minimum window length, the initial window length is used as the window length of the second sliding window;

[0028] in response to the initial window length being less than the minimum window length, the minimum window length is used as the window length of the second sliding window.

[0029] Optionally, the calculation of the feature indicators of the machining data in each second sliding window comprises:

[0030] the machining data in the second sliding window is subjected to fast Fourier transform;

[0031] the feature indicators of the machining data in an effective frequency band range are calculated.

[0032] Optionally, the machining data further comprises a vibration signal synchronously collected with the power signal, and the feature indicators comprise a root mean square of the vibration signal.

[0033] In a second aspect, a tool state monitoring method is provided, comprising:

[0034] acquire machining data of the tool in a normal working condition, the machining data comprising power signals and vibration signals of a tool spindle acquired synchronously;

[0035] segment the power signals according to a segmentation line, and determine machining periods of the tool according to a segmentation result; the segmentation line is used to define machining states and non-machining states of the tool;

[0036] extract feature indexes of the machining data in the machining periods;

[0037] determine tool state monitoring thresholds according to feature values of the feature indexes;

[0038] analyze and determine real-time machining data of the tool according to the tool state monitoring thresholds, so as to monitor a state of the tool.

[0039] In a third aspect, a system for constructing a tool state monitoring model is provided, comprising:

[0040] an acquisition module, configured to acquire machining data of the tool in a normal working condition, the machining data comprising power signals of a tool spindle;

[0041] a segmentation module, configured to segment the power signals according to a segmentation line, and determine machining periods of the tool according to a segmentation result; the segmentation line is used to define machining states and non-machining states of the tool;

[0042] an extraction module, configured to extract feature indexes of the machining data in the machining periods;

[0043] a monitoring module, configured to determine tool state monitoring thresholds of the machining periods according to feature values of the feature indexes, and construct a tool state monitoring model according to the tool state monitoring thresholds; the tool state monitoring model is used to analyze and determine real-time machining data of the tool, so as to monitor a state of the tool.

[0044] In a fourth aspect, a tool state monitoring system is provided, comprising:

[0045] an acquisition module, configured to acquire machining data of the tool in a normal working condition, the machining data comprising power signals and vibration signals of a tool spindle acquired synchronously;

[0046] a segmentation module, configured to segment the power signals according to a segmentation line, and determine machining periods of the tool according to a segmentation result; the segmentation line is used to define machining states and non-machining states of the tool;

[0047] an extraction module, configured to extract feature indexes of the machining data in the machining periods;

[0048] The monitoring module is configured to determine a tool state monitoring threshold according to the characteristic value of the characteristic index, and analyze and determine real-time machining data of the tool according to the tool state monitoring threshold, so as to monitor the state of the tool.

[0049] In a fifth aspect, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and used to run on the processor, and the processor executes the computer program to implement the method of any one of the above aspects.

[0050] In a sixth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the method of any one of the above aspects.

[0051] In a seventh aspect, a computer program product is provided, including a computer program, and the computer program is executed by a processor to implement the method of any one of the above aspects.

[0052] On the basis of common sense in the art, the above-mentioned preferred conditions can be combined arbitrarily, that is, to obtain each preferred example of the present disclosure.

[0053] The positive progress effect of the present disclosure is that the present disclosure first automatically segments the independent machining period according to the power signal, then analyzes the machining data in the machining period, extracts the characteristics of the machining data under normal working conditions, and constructs a tool state monitoring model, so that through the integrated design of "signal segmentation-feature extraction-state determination", on the one hand, the machining data is bound with the machining condition, effectively eliminating the influence of non-machining period interference and cross-condition cross-coupling, and the tool state monitoring model constructed on the basis of the data can accurately evaluate and monitor the tool state, significantly improving the accuracy and timeliness of the monitoring result; on the other hand, the technical pain points of insufficient model generalization ability and significant performance decline when the machining conditions change are effectively solved, the present disclosure gives the tool state monitoring model stronger working condition self-adaptive ability, ensures that it can still maintain stable and reliable monitoring performance when the machining conditions such as machine tool type, workpiece material and process parameters are dynamically adjusted, and greatly expands the application range. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 A flowchart of a tool state monitoring model construction method provided for an exemplary embodiment of the present disclosure is provided;

[0055] Figure 2 A flowchart of a tool state monitoring method provided for an exemplary embodiment of the present disclosure is provided;

[0056] Figure 3 A structural schematic diagram of an electronic device provided for an exemplary embodiment of the present disclosure is provided. DETAILED DESCRIPTION

[0057] The present disclosure will be further illustrated by the following examples, but the present disclosure is not limited to the scope of the examples.

[0058] The prefix words such as "first", "second" in the embodiments of the present disclosure are only used to distinguish different description objects, and do not have limiting effect on the position, order, priority, quantity or content of the described objects. The use of prefix words such as ordinal numbers in the embodiments of the present disclosure does not constitute limitation on the described objects, and the description of the described objects should refer to the description in the context of claims or embodiments, and should not constitute redundant limitation because of the use of such prefix words. In addition, in the description of the embodiments, unless otherwise specified, the meaning of "plurality" is two or more.

[0059] Figure 1 A flowchart of a method for constructing a tool state monitoring model is provided for an exemplary embodiment of the present disclosure, and the method comprises the following steps:

[0060] In step 101, machining data under normal working conditions of the tool is acquired.

[0061] The machining data comprises a power signal of the tool spindle. A power sensor is arranged on the tool spindle, and the power signal is collected based on the power sensor.

[0062] In addition to the power signal, the machining data can also comprise signals reflecting the cutting force, wear and temperature of the tool, such as vibration signals, spindle current signals, servo motor loads, temperature signals, tool spindle speeds, acoustic emission signals, etc., which are synchronously collected with the power signal. These signals are collected by deploying corresponding types of sensors on the machine tool, such as arranging vibration sensors on the tool spindle to collect vibration signals based on the vibration sensors. The layout number and layout method of various sensors on the machine tool can be set according to actual conditions, and the embodiments of the present disclosure do not particularly limit this.

[0063] In step 101, the signals collected by the sensors can be directly received, or the signals can be acquired from the memory; the signals stored in the memory are the signals collected by the sensors and stored therein.

[0064] In the machining process of the numerical control machine tool, multiple tools and multiple machining processes can be used to machine the workpiece. In the machining data collection process, each collected data point is bound with the tool identification (Tn) used during machining and the machining process being executed (represented by the program number Pm). Therefore, when each workpiece is machined, a complete set of machining data can be obtained, and each piece of data is uniquely identified by the combination tag "Tn-Pm" for its process background.

[0065] The machining data of each tool to be monitored is acquired in step 101, and the machining data is bound with the tool identifier and the machining process. It should be noted that the user can specify the tool to be monitored, and the tool to be monitored can be all tools on the machine tool or part of the tools.

[0066] In step 102, the power signal is segmented according to the segmentation line, and the machining period of the tool is determined according to the segmentation result.

[0067] The segmentation line is used to define the machining state and the non-machining state of the tool, that is, the segmentation line represents the typical quantization level of the power signal in the machine tool idle running or standby state.

[0068] The power of the tool spindle directly reflects the total energy required to drive the tool to rotate and overcome the cutting resistance. When the tool starts to contact the workpiece, the load increases instantaneously, and the power of the tool spindle shows a clear upward trend. This power change is a one-to-one correspondence and a clear causal relationship with the conversion of the machining state, providing the most direct and reliable basis for determining the machining period. The embodiment automatically identifies the machining period through the power signal, realizes full-automatic and high-precision segmentation of continuous signals into independent machining segments with clear process meaning, effectively replaces the traditional manual setting or fixed time window segmentation mode, and improves the intelligent level of the machining process monitoring.

[0069] In step 103, the feature index of the machining data in the machining period is extracted.

[0070] In the embodiment, the machine tool idle interference signal in the non-machining period is removed through the machining period, the effective machining data feature index in the machining period is extracted, and the machining data feature index is used as the data basis for tool state monitoring. This method can effectively improve the mapping correlation between the signal and the tool state, and ultimately provide the tool state monitoring model with input data that is structurally uniform, high in signal purity, and comparable. This method solves the technical problem of insufficient model accuracy and universality caused by the nonlinearity, time-varying nature, and coupling characteristics of indirect signals.

[0071] In step 104, the tool state monitoring threshold is determined according to the feature value of the feature index, and the tool state monitoring model is constructed according to the tool state monitoring threshold.

[0072] The feature index is a quantitative representation that is strongly related to the tool state and is extracted from the machining data. The tool state monitoring threshold is determined based on the feature value of the feature index, so that the tool state monitoring threshold has dynamic adaptability and is updated synchronously with the tool life cycle and the machine tool state, providing a guarantee for accurate and stable tool state monitoring.

[0073] The tool state monitoring model represents tool state monitoring thresholds corresponding to each tool identifier, machining process and machining period. The tool state monitoring thresholds are accurately adapted to the cutting conditions and life cycle stages of the tool. The model is used for targeted analysis and determination of real-time machining data of the tool, thereby realizing online dynamic monitoring of the tool state. The specific implementation process is as follows: first, according to the identifier of the current monitoring tool, the machining process and the corresponding machining period, the tool state monitoring threshold matched from the tool state monitoring model; then, the collected real-time machining data of the tool is quantitatively compared with the matched tool state monitoring threshold; finally, the working state of the tool is determined according to the comparison result (such as whether the real-time data exceeds the threshold range, the exceeding range). The working state of the tool is, for example, normal cutting, slight wear, severe wear, and blade collapse.

[0074] In this embodiment, the independent machining period is first automatically segmented according to the power signal, and then the machining data in the machining period is used for targeted analysis to extract the characteristics of the machining data under normal working conditions, and a tool state monitoring model is constructed, so that through the integrated design of "signal segmentation-feature extraction-state determination", on the one hand, the machining data is bound with the machining conditions, effectively eliminating the influence of non-machining period interference and cross-coupling of cross-working conditions, realizing accurate evaluation and real-time monitoring of the tool state, and significantly improving the accuracy and timeliness of the monitoring result; on the other hand, the technical pain points of insufficient model generalization ability and significant performance decline when the machining conditions change in the prior art are effectively solved. The tool state monitoring model in this embodiment has stronger working condition self-adaptability, ensuring that it can maintain stable and reliable monitoring performance when the machining conditions such as machine tool type, workpiece material and process parameters are dynamically adjusted, and the application range is greatly expanded.

[0075] The following describes an implementation of determining the segmentation line:

[0076] S1, a first sliding window with gradually increasing size is used to smooth the power signal to obtain a power sequence; each element contained in the power sequence corresponds to a calculation result of the first sliding window.

[0077] The power signal is smoothed by using a first sliding window with gradually increasing size (i.e. a recursive sliding window) to suppress random noise introduced in the signal acquisition process, while maximizing the retention of the true trend characteristics of the power signal.

[0078] The following describes the specific process of smoothing:

[0079] 1) A one-dimensional discrete signal sequence U with a length of N is used to represent the power signal, U=[u0,u1,u2,…,u N-1 ], where u i ∈R, 0≤i≤N-1.

[0080] 2) To solve the boundary problem of the first sliding window not having enough historical data at the beginning of the power signal, the embodiment adopts the cumulative average method to smooth the data at the beginning. Let the window length of the first sliding window be K (K is an odd number), and for the first Kp = min(K, N) points of the sequence U, gradually increasing windows are used for averaging to avoid data mutation at the beginning. The calculation formula is:

[0081] where i = 0, 1, …, K p-1 , which ensures smooth transition of the signal at the beginning, and Yi represents the result of the smoothed data at the beginning.

[0082] 3) Calculate the sum of the first K elements of the sequence U as the initial window sum S K-1 , K-1 = , calculate the constant Cinv = 1 / K, and calculate the arithmetic mean of the data in the first complete window, and output: ;

[0083] 4) For i = K, K+1, …, N-1, update the window sum and calculate the output according to the following recursive formula:

[0084] S i =S i-1 -u i-K +u i ,

[0085] y i =C inv *S i ;

[0086] y i is the calculation result of each first sliding window.

[0087] In an embodiment, to match the length of the output power sequence with the complete period of the sliding window processing, based on the sliding window processing mechanism, the end of the signal is smoothed and closed. The target length of the smoothed sequence should be N + K- 1 to ensure that there is corresponding output during the entire data processing period.

[0088] In an embodiment, the end of the power sequence is smoothed and closed by the following steps:

[0089] S1-1, judge whether the number of elements contained in the power sequence reaches the target number.

[0090] S1-2, in response to the number of elements not reaching the target number, use the last element of the power sequence to pad the number of elements in the power sequence to the target number.

[0091] Step S1-2 is to calculate the result y of the N-1th point N-1 Copy backward K-1 times as a constant:

[0092] y i =y N-1 , where i = N, N+1, …, N+K-2;

[0093] Finally, the smoothed power sequence Y = [y0, y1, y2, …, y N+K-2 ] is obtained, where the first Kp points (y0 to y Kp-1 ) of the sequence Y are calculated by the formula in Step 2).

[0094] In this embodiment, a complete smoothing process is constructed, including gradual front processing, main recursive update, and tail backward retention. It can effectively suppress random noise in industrial time series signals, while completely preserving the original trend characteristics of the signal, and meeting the real-time processing requirements of industrial sites, thereby providing a high-quality data basis for subsequent signal analysis, feature extraction, state monitoring, etc.

[0095] S2, perform quantile calculation on the power sequence, and determine the split line according to the calculation result.

[0096] Next, an implementation of determining the split line is introduced:

[0097] The split line B = Q q (Y), where represents the quantile function. The power sequence Y or the smoothed power sequence Y is arranged in ascending order to obtain an ordered sequence Y sorted , and the value at index is taken as the split line B, where the quantile parameter q is a preset constant, and the value range is [0.1, 0.3], M is the length of the power signal, which is determined by the sampling frequency of the power signal and the time length of the power signal required for analysis, M = sampling frequency x power signal time length, and the moving window sequence should ensure to contain at least one complete machining cycle.

[0098] In this embodiment, the quantile is used to determine the split line, which can adaptively match the characteristic fluctuation law of different machining conditions, adapt to the inherent noise of the equipment, the slight differences of different tools, and the slow changes of the working conditions. It not only significantly improves the robustness and accuracy of signal segmentation, but also improves the accuracy of machining process tool anomaly monitoring, effectively reduces the production loss caused by false alarm events and unplanned downtime, greatly reduces the time and labor cost of machine tool site debugging, and finally provides key technical support for the construction of equipment predictive maintenance system.

[0099] In other implementations, the threshold method is used to determine the segmentation line, specifically: a fixed power threshold is set, and when the power signal exceeds the threshold for a certain duration, it is determined that the processing period starts. The first derivative (change rate) of the power signal can also be used, and when its absolute value exceeds the set threshold, it is identified as a state switching point.

[0100] In other implementations, a machine learning model is used to identify the state of the power time sequence signal, and automatically segment the "no load", "processing", "tool changing" and other state stages.

[0101] In other implementations, the double-peak threshold method is used to determine the segmentation line, specifically: the full working condition signal of the machine tool (including complete power signals of no load, standby, and processing) is collected, the frequency histogram / kernel density curve of the signal is drawn, and the double peak and intermediate valley are identified; the segmentation line is determined according to the signal value corresponding to the valley value.

[0102] In other implementations, the segmentation line is determined based on empirical values or simulation tests.

[0103] The following describes an implementation of determining the processing period:

[0104] After the effective interval segmentation based on the baseline B is completed for the power signal, the position indexes of n processing periods are obtained:

[0105] {[s1,e1],[s2,e2],...,[s n ,e n ]}, where s q and e q represent the start and end indexes of the qth processing period in the smoothed signal sequence Y. To convert the above position information into time coordinates with physical meaning, the sampling frequency fs of the power signal is introduced. The conversion relationship is as follows:

[0106] The start time t sq and end time t eq of the qth processing period are calculated as follows:

[0107] t sq = s q / fs, t eq = e q / fs;

[0108] Therefore, all processing periods can be represented as time intervals:

[0109] {[t s1 , t e1 ], [t s2 , t e2 ],...,[t sn , ten ]}.

[0110] In one embodiment, in order to convert the continuous time-varying signal into a stationary segment sequence suitable for short-time frequency domain analysis, to extract features that can reflect the dynamic changes of the tool machining state, the machining data in the machining period is segmented by using a second sliding window, specifically:

[0111] The machining data in the machining period is segmented and windowed by using the second sliding window, and the feature indicators of the machining data in each second sliding window are calculated.

[0112] The window length of the second sliding window is negatively correlated with the speed of the spindle, and / or the step length of the second sliding window is negatively correlated with the speed of the spindle. The window length < the step length, and both the window length and the step length are expressed in points.

[0113] In this embodiment, the window length of the second sliding window is inversely proportional to the speed of the tool spindle. When the tool spindle is in a high-speed machining state, the window length of the second sliding window is shortened synchronously to effectively capture the high-frequency characteristics of machining.

[0114] In one embodiment, a reference window length T0 and a reference speed R0 of the tool spindle are defined, and the window length of the second sliding window is determined by the following formula:

[0115] T = T0*(R0 / R).

[0116] In this embodiment, the window length and the step length of the second sliding window are adaptively and dynamically adjusted according to the specific physical characteristics of the workpiece machining process, so as to effectively suppress the signal noise caused by the running of the machine tool, environmental vibration and electromagnetic interference, greatly reduce the difficulty of signal feature extraction, and guarantee the accuracy and reliability of the feature parameters.

[0117] In one embodiment, the window length of the second sliding window is determined by the following steps:

[0118] Determine the initial window length according to the speed of the spindle;

[0119] Determine whether the initial window length is greater than the preset minimum window length;

[0120] In response to the initial window length being greater than the minimum window length, the initial window length is taken as the window length of the second sliding window;

[0121] In response to the initial window length being less than the minimum window length, the minimum window length is taken as the window length of the second sliding window.

[0122] In this embodiment, in order to ensure the feasibility of frequency domain analysis, a minimum window length constraint N min, N min ) is adopted. The step size S is adaptively set according to the window length, and an overlap coefficient x (0 < x < 1) is set, S = W * x. The value of x is set according to actual conditions.

[0123] In an embodiment, the feature index is calculated based on frequency domain analysis. Specifically, the step of calculating the feature index of the machining data in each second sliding window comprises: performing fast Fourier transform on the machining data in the second sliding window, and calculating the feature index of the machining data in the effective frequency band range.

[0124] The effective frequency band range [F min , F max ] is determined according to the rotational speed of the tool spindle and its harmonics. The effective frequency band range [F min , F max ] is determined as follows: taking the rotational frequency f r =R / 60 of the spindle and its first n harmonics (n*f r , where n is a preset harmonic number) as the center frequency; taking each center frequency as the half-bandwidth of Δf above and below, and collectively forming a plurality of target frequency bands. The half-bandwidth Δf is usually half of the rotational frequency f r , that is, Δf=f r / 2; and the final effective frequency band [F min , F max ] is the union of these target frequency bands. The final feature index is the sum of the frequency band feature indexes calculated for all target frequency bands.

[0125] In this embodiment, the effective frequency band is determined by the rotational speed of the spindle and its harmonics, and the feature index of the machining data in the effective frequency band is calculated, which can realize accurate extraction of machining features, effectively suppress the noise influence of machine tool operation, environmental vibration and electromagnetic interference, adapt to different machining conditions of different spindle speeds, guarantee the consistency and reliability of the feature index, and meet the demand of real-time monitoring of the machining process.

[0126] In other implementations, the statistical quantities such as the effective value, peak value, kurtosis, etc. of the machining data are calculated as the feature index.

[0127] In an embodiment, the feature index includes the root mean square of the vibration signal. For each machining time period, the root mean square of the vibration signal of each second sliding window is calculated with a step size S sliding window. For the qth machining period, the feature value sequence F q =[f q1 ,f q2 ,...,f qmi ] is obtained, where m i represents the total number of second sliding windows in the machining period, and the feature value sequence Fq Each element in the equation corresponds to the root mean square of the vibration signal in a second sliding window.

[0128] Vibration signals are the direct physical response to stress waves and impacts during the contact between the tool and the workpiece. Even minute changes in the tool will generate detectable transient impacts or spectral changes in the vibration signal. Therefore, in this embodiment, independent machining periods are first automatically segmented based on the power signal, thus providing a precise framework for targeted analysis of the vibration signal. Then, within this framework, the vibration signal is used to evaluate and monitor indicators such as tool condition and machining process stability in real time.

[0129] In this embodiment, by conducting collaborative analysis of power signals and vibration signals and configuring an efficient data smoothing method, the automatic and accurate segmentation of the processing period is achieved, while real-time status monitoring of the processing process is completed.

[0130] In one embodiment, the tool condition monitoring threshold is determined based on the mean and standard deviation of the feature value sequence. Specifically: for each feature index to be monitored in the Tn-Pm combination, denoted as F, the baseline mean μ of the feature value sequence over all machining periods is calculated. F and standard deviation σ F The baseline mean μ of the eigenvalue sequence F and standard deviation σ F Characterizes health benchmarks.

[0131] Based on statistical principles, a dynamic early warning threshold T for this feature is set. W With dynamic alarm threshold T A And based on the threshold T W With T A Construct a tool condition monitoring model. Dynamic early warning threshold T. W With dynamic alarm threshold T A It is expressed as follows:

[0132] T W = μ F ± K1σ F , T A = μ F ± K2σ F ;

[0133] Wherein, K1 and K2 represent the set warning coefficient and alarm coefficient, respectively.

[0134] In this embodiment, an independent monitoring benchmark is established for each processing period, and dynamic statistical threshold is used for tool state monitoring, which can adaptively match the characteristic fluctuation law of different processing conditions, effectively distinguish normal fluctuation from real abnormality, significantly improve the accuracy of tool abnormality monitoring in the processing process, effectively reduce the production loss caused by false alarm and unplanned downtime, greatly reduce the time and labor cost of machine tool site debugging, and finally provide key technical support for the construction of equipment predictive maintenance system.

[0135] In the subsequent workpiece processing process, the new processing process is monitored and judged in real time. When each processing period is completed, the Tn-Pm label is automatically identified, and the mean value μ curr of the characteristic value sequence in the current processing period in the real-time processing data is calculated in the same way as the construction of the tool state monitoring model.

[0136] In one embodiment, taking K1=2 and K2=3 as an example, if the mean value μ curr falls within the range [μ W -2σ F , μ F +2σ F ] determined according to T F , a warning prompt is given. If the mean value μ curr falls within the range [μ A -3σ F , μ F +3σ F ] determined according to T F , an alarm prompt is given.

[0137] In one embodiment, the standardized deviation D of the current characteristic value relative to the health benchmark (characteristic value sequence) is calculated:

[0138] D =∣μ curr - μ F ∣ / σ F , according to the size of D to judge the processing state of this processing period:

[0139] If D≤K1, it indicates that the process characteristics of the tool are within the normal fluctuation range;

[0140] If K1<D≤K2, it indicates that the process characteristics of the tool have deviated significantly, a warning prompt is given, such as recording logs and prompting inspection;

[0141] If D>3, it indicates that the process characteristics of the tool have exceeded the normal fluctuation limit, an alarm prompt is given, such as indicating a high probability of abnormality, and observing whether the tool has phenomena such as tool breakage, edge collapse, wear, etc.

[0142] The embodiment of the present disclosure further provides a tool state monitoring method, referring to Figure 2 The tool state monitoring method comprises the following steps:

[0143] Step 201, acquiring machining data under normal working conditions of a tool, wherein the machining data comprises power signals and vibration signals of a tool spindle acquired synchronously;

[0144] Step 202, dividing the power signals according to a division line, and determining machining time periods of the tool according to the division result; the division line is used to define machining states and non-machining states of the tool;

[0145] Step 203, extracting characteristic indexes of the machining data in the machining time periods;

[0146] Step 204, determining tool state monitoring thresholds according to characteristic values of the characteristic indexes;

[0147] Step 205, analyzing and determining real-time machining data of the tool according to the tool state monitoring thresholds, so as to monitor the state of the tool.

[0148] The implementation process of steps 201-204 is the same as that of steps 101-104, and will not be described here.

[0149] Corresponding to the tool state monitoring model construction method and the tool state monitoring method, the present disclosure further provides a tool state monitoring model construction system and a tool state monitoring system.

[0150] The embodiment of the present disclosure provides a tool state monitoring model construction system, which is used to implement the tool state monitoring model construction method provided in any of the above embodiments, and comprises:

[0151] An acquisition module is configured to acquire machining data under normal working conditions of a tool, wherein the machining data comprises power signals of a tool spindle;

[0152] A division module is configured to divide the power signals according to a division line, and determine machining time periods of the tool according to the division result; the division line is used to define machining states and non-machining states of the tool;

[0153] An extraction module is configured to extract characteristic indexes of the machining data in each machining time period;

[0154] A monitoring module is configured to determine tool state monitoring thresholds of each machining time period according to characteristic values of the characteristic indexes, and construct a tool state monitoring model according to the tool state monitoring thresholds; the tool state monitoring model is used to analyze and determine real-time machining data of the tool, so as to monitor the state of the tool.

[0155] Optionally, the system further comprises:

[0156] a smoothing module configured to smooth the power signal using a first sliding window with gradually increasing length to obtain a power sequence, wherein each element of the power sequence corresponds to a calculation result of a first sliding window;

[0157] a calculation module configured to perform quantile calculation on the power sequence, and determine the split line according to a calculation result.

[0158] Optionally, the system further comprises:

[0159] a judgment module configured to judge whether a number of elements contained in the power sequence reaches a target number;

[0160] in response to the number of elements not reaching the target number, the judgment module invokes a padding module to pad the number of elements of the power sequence to the target number using a last element contained in the power sequence;

[0161] the calculation module is specifically configured to perform quantile calculation on the padded power sequence.

[0162] Optionally, the extraction module is specifically configured to:

[0163] perform segmented sliding window processing on the machining data in the machining period using a second sliding window, and calculate a feature index of the machining data in each second sliding window; wherein a window length of the second sliding window is negatively correlated with the rotating speed of the main shaft, and / or a step length of the second sliding window is negatively correlated with the rotating speed of the main shaft.

[0164] Optionally, the system comprises a window length determination module of the second sliding window, which is specifically configured to:

[0165] determine an initial window length according to the rotating speed of the main shaft;

[0166] judge whether the initial window length is greater than a preset minimum window length;

[0167] in response to the initial window length being greater than the minimum window length, the initial window length is taken as the window length of the second sliding window;

[0168] in response to the initial window length being less than the minimum window length, the minimum window length is taken as the window length of the second sliding window.

[0169] Optionally, when calculating the feature index of the machining data in each second sliding window, the extraction module is specifically configured to:

[0170] perform fast Fourier transform on the machining data in the second sliding window;

[0171] The characteristic index of the processing data in the effective frequency band range is calculated.

[0172] Optionally, the processing data further comprises a vibration signal synchronously collected with the power signal, and the characteristic index comprises a root mean square of the vibration signal.

[0173] The disclosure also provides a tool state monitoring system for implementing the tool state monitoring method of any of the above embodiments, and the system comprises:

[0174] An acquisition module is configured to acquire processing data of the tool in a normal working condition, wherein the processing data comprises a power signal and a vibration signal of a tool spindle synchronously collected;

[0175] A segmentation module is configured to segment the power signal according to a segmentation line, and determine a processing period of the tool according to a segmentation result; the segmentation line is used to define a processing state and a non-processing state of the tool;

[0176] An extraction module is configured to extract a characteristic index of the processing data in the processing period;

[0177] A monitoring module is configured to determine a tool state monitoring threshold according to a characteristic value of the characteristic index, and analyze and determine real-time processing data of the tool according to the tool state monitoring threshold, so as to monitor the state of the tool.

[0178] For the system embodiment, since it basically corresponds to the method embodiment, the related parts can be referred to the part of the method embodiment. The system embodiment described above is only illustrative, wherein the units described as separate components can or can not be physically separated, and the components of the unit can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. According to actual needs, some or all of the modules can be selected to achieve the purpose of the disclosure.

[0179] Figure 3 A structural schematic diagram of an electronic device is shown for an example embodiment of the disclosure, which comprises a memory, a processor, and a computer program stored in the memory and used for running on the processor, wherein the processor implements the method of any of the above embodiments when executing the computer program. Figure 3 The electronic device 30 shown is only an example, and should not limit the functions and use range of the embodiments of the disclosure.

[0180] As Figure 3As shown, the electronic device 30 can be in the form of a general computing device, for example, it can be a server device. The components of the electronic device 30 can include, but are not limited to, the at least one processor 31 described above, the at least one memory 32 described above, a bus 33 that connects the different system components, including the memory 32 and the processor 31.

[0181] The bus 33 includes a data bus, an address bus, and a control bus.

[0182] The memory 32 can include volatile memory, such as random access memory (RAM) 321 and / or cache memory 322, and can further include non-volatile memory, such as read-only memory (ROM) 323.

[0183] The memory 32 can also include a program tool 325 (or utility tool) having a set of (at least one) program modules 324, such as an operating system, one or more application programs, other program modules, and program data, and can include an implementation of a network environment, individually or in some combination.

[0184] The processor 31 performs various function applications and data processing by running the computer programs stored in the memory 32, such as the method provided by any of the embodiments described above.

[0185] The electronic device 30 can also communicate with one or more external devices 34 (such as a keyboard or a pointing device, etc.) via an input / output (I / O) interface 35. Also, the electronic device 30 can communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 36. As shown, the network adapter 36 communicates with the other modules of the electronic device 30 via the bus 33. It should be appreciated that although not shown, other hardware and / or software modules can be used in conjunction with the electronic device 30, including, but not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID (Redundant Array of Independent Disks) systems, tape drives, and data archival storage systems, etc.

[0186] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the foregoing detailed description, such division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules.

[0187] The embodiment of the present disclosure further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the method provided by any of the above embodiments.

[0188] More specifically, the readable storage medium can include, but is not limited to, a portable disc, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0189] The embodiment of the present disclosure further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the method described in any of the above.

[0190] The program code for executing the computer program product of the present disclosure can be written in any combination of one or more programming languages, and can be executed completely on a user device, partially on a user device, as a separate software package, partially on a user device and partially on a remote device, or completely on a remote device.

[0191] Although the specific embodiments of the present disclosure are described above, those skilled in the art should understand that this is only an example, and the protection scope of the present disclosure is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of the present disclosure, and these changes and modifications all fall within the protection scope of the present disclosure.

Claims

1. A method of constructing a tool condition monitoring model, characterized by, The method comprises the following steps: obtaining machining data of the tool in a normal working condition, the machining data comprising a power signal of a tool spindle; smoothing the power signal by using a gradually increasing first sliding window to obtain a power sequence; each element of the power sequence corresponds to an arithmetic mean value of data in a first sliding window; determining whether the number of elements contained in the power sequence reaches a target number; in response to the number of elements not reaching the target number, using the last element of the power sequence to fill the number of elements of the power sequence to the target number; The quantile calculation is performed on the padded power sequence, and a calculation result is used as a segmentation line; wherein the calculation result is a value corresponding to a target index value in an ordered sequence, the calculation formula of the target index value is , the ordered sequence is a sequence obtained by arranging the padded power sequence in ascending order, q is a quantile parameter, and M is the length of the power signal processing sequence, and the power signal processing sequence contains at least one complete processing cycle. segmenting the power signal according to a segmentation line, and determining a machining period of the tool according to the segmentation result; the segmentation line is used to define the machining state and non-machining state of the tool; extracting feature indicators of the machining data in each machining period; determining tool state monitoring thresholds of each machining period according to the feature values of the feature indicators, and constructing a tool state monitoring model according to the tool state monitoring thresholds; the tool state monitoring model is used to analyze and determine real-time machining data of the tool to monitor the state of the tool.

2. The construction method according to claim 1, characterized in that, The method of extracting feature indicators of the machining data in each machining period comprises the following steps: segmenting and sliding window processing the machining data in each machining period by using a second sliding window, and calculating feature indicators of the machining data in each second sliding window; wherein the window length of the second sliding window is negatively correlated with the rotating speed of the spindle, and / or the step length of the second sliding window is negatively correlated with the rotating speed of the spindle.

3. The construction method of claim 2, wherein, The window length of the second sliding window is determined by the following steps: determining an initial window length according to the rotating speed of the spindle; determining whether the initial window length is greater than a preset minimum window length; in response to the initial window length being greater than the minimum window length, taking the initial window length as the window length of the second sliding window; in response to the initial window length being less than the minimum window length, taking the minimum window length as the window length of the second sliding window.

4. The construction method of claim 2, wherein, The method of calculating feature indicators of the machining data in each second sliding window comprises the following steps: performing fast Fourier transform on the machining data in the second sliding window; calculating feature indicators of the machining data in an effective frequency band range.

5. The method of construction of claim 2, wherein, The machining data further comprises a vibration signal collected synchronously with the power signal, and the feature indicators comprise a root mean square of the vibration signal.

6. A tool condition monitoring method characterized by, The method comprises the following steps: obtaining machining data of the tool in a normal working condition, the machining data comprising a power signal of a tool spindle and a vibration signal collected synchronously; smoothing the power signal by using a gradually increasing first sliding window to obtain a power sequence; each element of the power sequence corresponds to an arithmetic mean value of data in a first sliding window; determining whether the number of elements contained in the power sequence reaches a target number; in response to the number of elements not reaching the target number, using the last element of the power sequence to fill the number of elements of the power sequence to the target number; The quantile calculation is performed on the padded power sequence, and a calculation result is used as a segmentation line; wherein the calculation result is a value corresponding to a target index value in an ordered sequence, the calculation formula of the target index value is , the ordered sequence is a sequence obtained by arranging the padded power sequence in ascending order, q is a quantile parameter, and M is the length of the power signal processing sequence, and the power signal processing sequence contains at least one complete processing cycle. The power signal is segmented according to a segmentation line, and a machining period of the tool is determined according to a segmentation result; the segmentation line is used to define a machining state and a non-machining state of the tool; characteristic indexes of machining data in the machining period are extracted; a tool state monitoring threshold is determined according to characteristic values of the characteristic indexes; real-time machining data of the tool is analyzed and determined according to the tool state monitoring threshold, so as to monitor a state of the tool.

7. A system for constructing a tool condition monitoring model, characterized by Comprise: an acquisition module, configured to acquire machining data under a normal working condition of a tool, the machining data comprising a power signal of a tool spindle; a smoothing module, configured to perform smoothing processing on the power signal by using a gradually increasing first sliding window, to obtain a power sequence; each element contained in the power sequence corresponds to an arithmetic mean value of data in a first sliding window; a judgment module, configured to judge whether a number of elements contained in the power sequence reaches a target number; in response to the number of elements not reaching the target number, the judgment module invokes a padding module to pad the number of elements of the power sequence to the target number by using a last element contained in the power sequence; A calculating module is configured to perform quantile calculation on the padded power sequence, and take a calculation result as a segmentation line; wherein the calculation result is a value corresponding to a target index value in an ordered sequence, the target index value is calculated according to the formula , the ordered sequence is a sequence obtained by arranging the padded power sequence in ascending order, q is a quantile parameter, and M is the length of the power signal processing sequence, the power signal processing sequence containing at least one complete processing cycle. a segmentation module, configured to segment the power signal according to a segmentation line, and to determine a machining period of the tool according to a segmentation result; the segmentation line is used to define a machining state and a non-machining state of the tool; an extraction module, configured to extract characteristic indexes of machining data in each machining period; a monitoring module, configured to determine a tool state monitoring threshold of each machining period according to characteristic values of the characteristic indexes, and to construct a tool state monitoring model according to the tool state monitoring threshold; the tool state monitoring model is used to analyze and determine real-time machining data of the tool, so as to monitor a state of the tool.

8. A tool condition monitoring system characterised in that, Comprise: an acquisition module, configured to acquire machining data under a normal working condition of a tool, the machining data comprising a power signal of a tool spindle; a smoothing module, configured to perform smoothing processing on the power signal by using a gradually increasing first sliding window, to obtain a power sequence; each element contained in the power sequence corresponds to an arithmetic mean value of data in a first sliding window; a judgment module, configured to judge whether a number of elements contained in the power sequence reaches a target number; in response to the number of elements not reaching the target number, the judgment module invokes a padding module to pad the number of elements of the power sequence to the target number by using a last element contained in the power sequence; A calculating module is configured to perform quantile calculation on the padded power sequence, and take a calculation result as a segmentation line; wherein the calculation result is a value corresponding to a target index value in an ordered sequence, the target index value is calculated according to the formula , the ordered sequence is a sequence obtained by arranging the padded power sequence in ascending order, q is a quantile parameter, and M is the length of the power signal processing sequence, the power signal processing sequence containing at least one complete processing cycle. a segmentation module, configured to segment the power signal according to a segmentation line, and to determine a machining period of the tool according to a segmentation result; the segmentation line is used to define a machining state and a non-machining state of the tool; an extraction module, configured to extract characteristic indexes of machining data in each machining period; a monitoring module, configured to determine a tool state monitoring threshold of each machining period according to characteristic values of the characteristic indexes, and to construct a tool state monitoring model according to the tool state monitoring threshold; the tool state monitoring model is used to analyze and determine real-time machining data of the tool, so as to monitor a state of the tool.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory for running on the processor, characterized in that, The processor executes the computer program to implement the method in any one of claims 1 to 6.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, which when executed by a processor, implements the method of any one of claims 1 to 6.

11. A computer program product comprising a computer program, characterized in that, The computer program, which when executed by a processor, implements the method of any one of claims 1 to 6.

Images