Electric spark machining efficiency analysis system based on historical working conditions

Abstract

The invention discloses an electric spark machining efficiency analysis system based on historical working conditions, and relates to the technical field of machining analys.The electric spark machining efficiency analysis system comprises the steps that an event account book is generated according to pulse events in the electric spark machining process, event spectrum characteristics are extracted, energy is decomposed into effective erosion, short-circuit loss and arc loss according to event types, and the energy supply effective rate is calculated; and chip removal retention, medium pollution and servo matching indexes are further generated, clearance pulse, rebound rhythm, flushing liquid duty ratio and servo gain switching parameters which can be issued are output in combination with historical prescription threshold values, re-check difference values of efficiency, cluster length and recovery quantile are given after execution, and attribution and correction closed loop of deep cavity section efficiency reduction are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of process analysis, in particular to an EDM efficiency analysis system based on historical working conditions. BACKGROUND

[0002] In deep cavity forming EDM, the machining efficiency is highly coupled with the gap discharge state, chip removal fluency, medium pollution, temperature rise, and servo feeding stability. On-site, the same program and parameters may cause normal performance in the front section and slow performance in the rear section, which is manifested as an increase in short circuit and arc, a slow recovery of the gap, and accelerated electrode wear. However, machine tool alarms and conventional statistics are difficult to provide clear attribution and executable correction. Existing methods mostly rely on multi-source sensor alignment, offline trial cutting, or experience-based parameter tuning, which are difficult to decompose efficiency loss into distinguishable mechanisms such as short circuit clustering, arc trailing, and servo mismatch without stopping the machine, and to deposit historical experience into transferable working condition prescriptions, resulting in unstable deep cavity section processing cycle, increased rework, and rising electrode consumption.

[0003] At present, the Chinese invention patent with application number CN202111096841.3 discloses a discharge electrode automatic trimming method and device and storage medium, which comprises: obtaining first point information before performing an EDM operation, obtaining second point information after performing the EDM operation, obtaining electrode wear parameters from the first point information and the second point information, comparing the electrode wear parameters with preset parameter thresholds to obtain a comparison result, and controlling a trimming mechanism to trim the discharge electrode. Through the implementation of the invention, the EDM CNC machine tool automatically analyzes the electrode wear parameters of the discharge electrode during the EDM operation by the first point information and the second point information after completing the EDM operation, correspondingly obtains the comparison result, and automatically trims the discharge electrode according to the comparison result. The entire automatic trimming process has low human involvement, and the trimming efficiency is improved by automatically performing trimming with the trimming mechanism integrated in the machine tool.

[0004] The above-mentioned technology is difficult to complete a verifiable loss attribution from the discharge event structure without stopping the machine, and output the correction action parameters and review results that can be directly issued. SUMMARY

[0005] The technical problem solved by the present application is that the existing technology is difficult to complete a verifiable loss attribution from the discharge event structure without stopping the machine, and output the correction action parameters and review results that can be directly issued.

[0006] To solve the above technical problems, the present application provides the following technical solutions: The application discloses a kind of based on historical working condition's electric spark machining efficiency analysis system, including event account generation module, event spectrum feature extraction module, energy decomposition calculation module, hidden cause index generation module, historical prescription matching module and action review module: The event account generation module is used to collect gap voltage sampling data, gap current sampling data, pulse switch state data, servo feed instruction data, flushing state data, bounce state data and step number data, and generate event window data based on the pulse switch state data, and generate discharge event type marker data and event integral energy data within the event window, encapsulate to form discharge event record data and gather as event account data; The event spectrum feature extraction module is used to generate short-circuit cluster data and recovery statistics data based on the event account data, and generate event spectrum feature data by summarizing; The energy decomposition calculation module is used to generate input energy accumulation data, effective ablation energy data, short-circuit loss energy data and arc loss energy data based on the event account data, and calculate energy supply efficiency data from the input energy accumulation data and effective ablation energy data; The hidden cause index generation module is used to generate chip removal retention index data and medium pollution index data based on the event spectrum feature data, and generate servo matching rate data based on the servo feed instruction data and recovery statistics data; The historical prescription matching module is used to output candidate prescription data from historical prescription data based on energy supply efficiency data, chip removal retention index data, medium pollution index data, servo matching rate data and step number data; The action review module is used to generate correction action sequence data based on the candidate prescription data and issue, and regenerate energy supply efficiency data and event spectrum feature data based on updated event account data after issuance, and output review result data.

[0007] Preferably, the event account generation module includes an acquisition unit, an event window generation unit, an event segmentation unit and an account writing unit. The acquisition unit is used to acquire gap voltage sampling data, gap current sampling data, pulse switch state data, servo feed instruction data, flushing state data, bounce state data and step number data. The event window generation unit is used to generate event window data based on the on-off edges of pulse switch state data. The event segmentation unit is used to generate discharge event type marker data, event integral energy data and event context marker data within the event window based on the event window data. The account writing unit is configured to encapsulate the step segment number data, the discharge event type marker data, the event integral energy data, and the event context marker data into discharge event record data and write the discharge event record data into the event account data.

[0008] Preferably, the event segmentation unit satisfies the following conditions when generating the discharge event record data: The event integral energy data is obtained by multiplying and accumulating the gap voltage sampling data and the gap current sampling data in the event window at a sampling interval. The discharge event type marker data includes normal discharge markers, short-circuit markers, arc markers, and open-circuit markers, and is obtained by window statistics of the gap voltage sampling data and the gap current sampling data in the event window according to a preset determination rule. The event context marker data is obtained by combining the servo feed instruction data, the flushing state data, and the bounce state data corresponding to the start time of the event window.

[0009] Preferably, the event spectrum feature extraction module includes a cluster construction unit and a recovery statistics unit. The cluster construction unit is configured to search for discharge event record data corresponding to short-circuit markers in the event account data, merge the discharge event record data into a same short-circuit cluster when the event start interval of adjacent two short-circuit marker discharge event record data is less than a cluster interval threshold data, and generate short-circuit cluster data, the short-circuit cluster data including cluster length data and cluster energy accumulation data. The recovery statistics unit is configured to generate event post-recovery duration data, the event post-recovery duration data being a time interval from the end of the discharge event record data with a short-circuit marker or an arc marker to the start of next normal discharge marker discharge event record data, and calculate recovery quantile data from the event post-recovery duration data. The short-circuit cluster data and the recovery quantile data are written into the event spectrum feature data.

[0010] Preferably, the energy decomposition calculation module includes an energy accumulation unit and an energy allocation unit. The energy accumulation unit is configured to accumulate the event integral energy data in the event account data to obtain input energy accumulation data. The energy allocation unit is configured to accumulate the event integral energy data according to the discharge event type marker data to obtain short-circuit loss energy data and arc loss energy data, and multiply the event integral energy data corresponding to the normal discharge marker by an etching correction coefficient data to obtain effective etching energy data. The energy supply efficiency data is calculated from the effective etching energy data and the input energy accumulation data.

[0011] Preferably, the hidden cause index generation module includes a retention index unit, a pollution index unit, and a servo matching unit. The retention index unit is used to generate chip retention index data by taking cluster length data, cluster energy accumulation data and recovery quantile data as inputs. The pollution index unit is used to generate medium pollution index data by taking the arc-marked event count percentage data, the arc-marked event integral energy percentage data, and the recovery quantile data as inputs. The servo matching unit is used to generate servo matching rate data by taking the window fluctuation data and recovery quantile data of the servo feed command data as input.

[0012] Preferably, the historical prescription matching module includes a prescription inventory unit and a matching scoring unit; The prescription inventory unit stores historical prescription data, which includes at least applicable boundary data, event threshold data, corrective action template data, and erosion correction coefficient data. The applicable boundary data includes at least material type data, electrode type data, processing polarity data, and process step number data range; The event threshold data includes at least cluster interval threshold data, maximum cluster length threshold data, and recovery quantile threshold data.

[0013] Preferably, the matching scoring unit is used to first filter historical prescription data based on the work step segment number data and the applicable boundary data to obtain a boundary candidate set, then compare the cluster length data, recovery quantile data and arc mark event count ratio data in the event spectrum feature data with the event threshold data to obtain threshold deviation data, and generate prescription scoring data based on the threshold deviation data and the energy supply efficiency data, and output historical prescription data that meet the preset conditions from the boundary candidate set as candidate prescription data.

[0014] Preferably, the action verification module includes an action parameterization unit and a verification calculation unit; The action parameterization unit is used to parameterize the correction action template data in the candidate prescription data to generate gap pulse sequence parameter data, rebound rhythm parameter data, flushing duty cycle parameter data and servo gain switching parameter data, and combine the parameter data to generate correction action sequence data.

[0015] Preferably, the verification calculation unit is used to read the updated event ledger data after the corrective action sequence data is issued, regenerate the power supply efficiency data and event spectrum feature data, and output the verification result data; The verification results data include at least the power supply efficiency difference data, the maximum cluster length difference data, and the recovery quantile difference data.

[0016] The beneficial effects of this invention are as follows: This invention generates an event ledger based on pulse events in the electrical discharge machining process, extracts event spectrum features, and decomposes energy into effective erosion, short-circuit loss, and arc loss according to event type, calculates energy supply efficiency, and further generates chip retention, media contamination, and servo matching indicators. Combined with historical prescription thresholds, it outputs the gap clearing pulse, rebound rhythm, flushing duty cycle, and servo gain switching parameters that can be issued. After execution, it provides the verification difference between efficiency, cluster length, and recovery quantile, realizing the attribution and correction closed loop of efficiency decline in deep cavity sections. Attached Figure Description

[0017] Figure 1 This is a basic flowchart of an electrical discharge machining efficiency analysis system based on historical working conditions, provided as an embodiment of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example, refer to Figure 1 This paper presents an EDM efficiency analysis system based on historical operating conditions, including an event ledger generation module, an event spectrum feature extraction module, an energy decomposition calculation module, a latent factor index generation module, a historical prescription matching module, and an action verification module. The event ledger generation module is used to collect gap voltage sampling data, gap current sampling data, pulse switch status data, servo feed command data, flushing status data, rebound status data, and step segment number data. Based on the pulse switch status data, it generates event window data, and generates discharge event type marker data and event integral energy data within the event window. These are then encapsulated to form discharge event record data and aggregated into event ledger data.

[0020] The event spectrum feature extraction module is used to generate short-circuit cluster data and recovery statistics based on event ledger data, and to summarize and generate event spectrum feature data.

[0021] The energy decomposition calculation module is used to generate input energy accumulation data, effective erosion energy data, short-circuit loss energy data, and arc loss energy data based on event ledger data, and calculates the energy supply efficiency data from the input energy accumulation data and the effective erosion energy data.

[0022] The latent cause index generation module is used to generate chip retention index data and media contamination index data based on event spectrum feature data, and to generate servo matching rate data based on servo feed command data and recovery statistics data.

[0023] The historical prescription matching module is used to output candidate prescription data from historical prescription data based on energy supply efficiency data, chip retention index data, media contamination index data, servo matching rate data, and work step number data.

[0024] The action review module is used to generate and distribute corrective action sequence data based on candidate prescription data. After distribution, it regenerates energy supply efficiency data and event spectrum feature data based on updated event ledger data and outputs review result data.

[0025] The event ledger generation module includes a data acquisition unit, an event window generation unit, an event segmentation unit, and a ledger writing unit.

[0026] The acquisition cycle of the acquisition unit is Collect gap voltage sampling data Gap current sampling data Pulse switch status data (0 indicates pulse on, 1 indicates pulse off), servo feed command data Fluid status data bounce status data and work step number data .

[0027] The event window generation unit is used to generate event window data based on the on / off edges of pulse switch state data.

[0028] The event window is divided by the edge of the pulse switch state: like and Then the starting point of the time window is generated. ; like and Then the end of the time window is generated. ; Output event window data .

[0029] The energy injection and discharge state of electrical discharge machining are strongly coupled with pulse on / off. By cutting the window according to the pulse edge, the continuous curve can be compressed into discrete event units. All subsequent indicators are based on event statistics, which is resistant to sampling drift and resistant to multiple sources of different frequencies.

[0030] The event segmentation unit is used to generate discharge event type marker data, event integral energy data, and event context marker data within the event window based on the event window data.

[0031] For each event window calculate: ; Output event integral energy data .

[0032] In the event window First, calculate the window statistics: ; ; and short-circuit voltage thresholds given by historical prescriptions or machine tool experience. Open-circuit voltage threshold Effective on-current threshold and arc current threshold .

[0033] When the event segmentation unit generates discharge event recording data, it satisfies the following: The event integral energy data is obtained by multiplying and accumulating the gap voltage sampling data and gap current sampling data within the event window according to the sampling interval.

[0034] The discharge event type labeling data includes normal discharge label, short circuit label, arc label, and open circuit label, and is obtained by window statistics of gap voltage sampling data and gap current sampling data within the event window according to preset judgment rules, namely: Short circuit marker: and ; Road markings: and ; Arc marking: and Falling within the arc voltage range ; Normal discharge marker: Those that do not meet the above criteria are classified as normal discharge markers; Output discharge type data .

[0035] The event context marker data is obtained by combining the servo feed command data, flushing status data and bounce status data corresponding to the start time of the event window.

[0036] At the start of the event window, retrieve the context values ​​of the servo feed command data, flushing status data, and bounce status data, and output the event context flag data. : The ledger writing unit is used to encapsulate the step segment number data, discharge event type marker data, event integral energy data, and event context marker data into discharge event record data. And write the event ledger data. .

[0037] The event spectrum feature extraction module includes a cluster construction unit and a recovery statistics unit.

[0038] The cluster construction unit is used to retrieve the discharge event record data corresponding to the short circuit marker in the event ledger data. When the event start interval of two adjacent short circuit marker discharge event record data is less than the cluster interval threshold data, they are merged into the same short circuit cluster and short circuit cluster data is generated. The short circuit cluster data includes cluster length data and cluster energy accumulation data.

[0039] In the same work step section number Within, all short-circuit times are sorted by their starting point: Let the cluster interval threshold data be... .

[0040] If adjacent short-circuit events satisfy: ; Then they are merged into the same cluster. For the first... Cluster calculation: Cluster length data: ; Cluster energy accumulation data : Obtain short-circuit cluster data .

[0041] In deep cavity forming EDM, poor chip removal often manifests not only as an increase in the number of short circuits, but also as a continuous accumulation of short circuit events over time, forming a clustered structure. This structure is difficult to express using single-point statistics such as average voltage and current or short circuit percentage. Therefore, this solution merges adjacent short circuit events in the event ledger according to the cluster interval threshold to generate short circuit cluster data (including cluster length data and cluster energy accumulation data), and uses it as a key input to the subsequent data processing link. On the one hand, it is used together with the recovery quantile data to calculate the chip retention index data; on the other hand, it participates in the generation of prescription threshold deviation data and prescription score data. At the same time, after the action is issued, it is used to calculate the maximum cluster length difference data and other verification differences. This clearly distinguishes between the working condition of many but scattered short circuits and the working condition of short circuit clusters accompanied by recovery tailing under the same caliber, and drives the selection and verification of correction action parameters such as gap clearing pulse sequence parameters and rebound rhythm parameters.

[0042] The recovery statistics unit is used to generate post-event recovery duration data, which is the time interval from the end of the discharge event record data of the short-circuit marker or arc marker to the start of the next normal discharge event record data, and calculates the recovery quantile data for the post-event recovery duration data.

[0043] Short-circuit cluster data and recovery quantile data are written into event spectrum feature data. .

[0044] For each short circuit or arc event Define the recovery time data after the event: ; in, This indicates the next normal discharge marker event.

[0045] right Find the recovered quantile data: ; in, The quantile parameter is set to 0.9 in this embodiment. The quantile is used here because the recovery time usually has a long tail, and the mean will be skewed by extreme values. The quantile is more stable and can quantify the recovery tail in a stable manner, which is suitable for distinguishing the long recovery caused by arc tail.

[0046] The energy decomposition calculation module includes an energy accumulation unit and an energy distribution unit.

[0047] The energy accumulation unit is used to accumulate the event integral energy data in the event ledger data to obtain the input energy accumulation data. .

[0048] The energy allocation unit is used to classify and accumulate event integral energy data according to the discharge event type to obtain short-circuit loss energy data and arc loss energy data. It also multiplies the event integral energy data corresponding to the normal discharge mark with the erosion correction coefficient data to obtain the effective erosion energy data. .

[0049] Energy supply efficiency data It is calculated from the effective erosion energy data and the cumulative input energy data.

[0050] The etching correction coefficient data incorporates the differences in etching conversion under different materials, polarities, and electrode conditions for the same normal discharge energy into a unified standard. If the ratio of the event integral energy data corresponding to the normal discharge mark to the cumulative input energy data is used directly, the differences in etching efficiency under different material conditions will be masked, leading to distortion of the prescription migration.

[0051] The latent cause index generation module includes a retention index unit, a pollution index unit, and a servo matching unit.

[0052] The retention index unit is used to generate chip retention index data by taking cluster length data, cluster energy accumulation data and recovery quantile data as input.

[0053] Take the maximum value of the cluster length data Within the statistical range corresponding to the same work step number, the number of discharge event records corresponding to the short-circuit marker in the discharge event type marker data is divided by the total number of discharge event records corresponding to the non-open-circuit markers (i.e., normal discharge markers, short-circuit markers, and arc markers) in the discharge event type marker data to obtain the short-circuit event percentage data. .

[0054] Calculate the chip retention index data: ; in, , and The preset weighting coefficients, As a preset constant, For the Sigmoid function, and This is event threshold data derived from historical prescriptions.

[0055] Since chip residue buildup can lead to short circuit clustering and prolonged recovery, the Sigmoid function is used to compress multiple indicators into a risk scale of 0 to 1, which facilitates prescription matching and action gating. It can differentiate between conditions with a high proportion of short circuits but no clustering and conditions with short circuit clustering and prolonged recovery on a single index.

[0056] The pollution index unit is used to generate medium pollution index data by taking the percentage data of arc-marked event counts, the percentage data of arc-marked event integral energy, and the recovery quantile data as inputs.

[0057] Within the statistical range corresponding to the same work step number, the number of discharge event records corresponding to the arc marker in the discharge event type marker data is divided by the total number of discharge event records corresponding to the non-open circuit markers (i.e., normal discharge markers, short circuit markers, and arc markers) in the discharge event type marker data to obtain the arc event percentage data. .

[0058] Calculate media contamination index data: ; in, , and The preset weighting coefficients, This is a preset constant.

[0059] Contamination of the medium and temperature rise make it easier for electric arcs to form and lengthen during recovery, with a higher energy percentage. The percentage of arcs is a better indicator of heat loss intensity. Under the same number of arcs, the condition with higher arc energy will be identified, avoiding misjudgment.

[0060] The servo matching unit is used to generate servo matching rate data by taking the window fluctuation data and recovery quantile data of the servo feed command data as input.

[0061] Calculate servo command fluctuations in each event window: ; Average the values ​​for short circuit and arc events. ; Calculate servo matching rate data: ; in, Based on event threshold data from historical prescriptions, servo mismatch can cause frequent and intense adjustments accompanied by recovery tailing. Servo matching rate data can further distinguish between short circuit clusters caused by chip removal problems and short circuits and arcs caused by servo oscillations.

[0062] The historical prescription matching module includes a prescription inventory unit and a matching scoring unit.

[0063] The prescription inventory unit stores historical prescription data, which includes at least applicable boundary data, event threshold data, corrective action template data, and erosion correction coefficient data.

[0064] The applicable boundary data should include at least the material type data, electrode type data, processing polarity data, and process step number range.

[0065] Event threshold data includes at least cluster interval threshold data, maximum cluster length threshold data, and recovery quantile threshold data.

[0066] The correction action template data includes at least the gap clearing pulse sequence parameter field, the rebound rhythm parameter field, the flushing duty cycle parameter field, and the servo gain switching parameter field.

[0067] The learning logic for the erosion correction coefficient data is as follows: For historical tasks within the same prescription boundary, the event integral energy data and segment-level actual removal amount corresponding to the normal discharge marker of each time window are taken as monitoring signals. If the segment-level progress (e.g., the percentage of step completion over time) can be obtained from the machine tool steps on-site, a removal agent per unit time is defined. (The change in completion rate can be used to represent the same work step), and the least squares fit is used: ; Under the same materials, electrodes and polarities, the normal discharge energy and the removal agent amount are approximately linear. The erosion correction coefficient data solidifies the effective conversion ratio into the formulation parameters, making the energy supply efficiency comparable across batches and the formulation migration more stable.

[0068] The matching scoring unit first filters historical prescription data based on work step number data and applicable boundary data to obtain a boundary candidate set. Then, it compares the cluster length data, recovery quantile data, and arc marker event count percentage data in the event spectrum feature data with the event threshold data to obtain threshold deviation data. The threshold deviation data includes cluster deviation. Recovery deviation and arc deviation And generate prescription scores based on threshold deviation data and energy supply efficiency data. The prescription with the highest prescription score and consistent boundary is selected as candidate prescription data. Historical prescription data with prescription score data that meet the preset conditions are output from the boundary candidate set as candidate prescription data. The preset conditions are that the action review module includes an action parameterization unit and a review calculation unit.

[0069] The action parameterization unit is used to parameterize the correction action template data in the candidate prescription data to generate gap pulse sequence parameter data, rebound rhythm parameter data, flushing duty cycle parameter data and servo gain switching parameter data, and combine the parameter data to generate correction action sequence data.

[0070] Fields are extracted from the correction action template data of the candidate prescription data and parameterized into gap pulse sequence parameter data, rebound rhythm parameter data, flushing duty cycle parameter data and servo gain switching parameter data. These are then assembled into correction action sequence data and sent out.

[0071] The verification calculation unit is used to read the updated event ledger data after the corrective action sequence data is issued, regenerate the power supply efficiency data and event spectrum characteristic data, and output the verification result data.

[0072] Put each sliding window Feature composition vector: ; Define tag: If the future Within a window Then the label data corresponding to the current window will be... Set to 1, otherwise set the label data corresponding to the current window. Set to 0.

[0073] Training with logistic regression: ; ; in, For the first The probability of efficiency degradation occurring for each sliding statistical window; The event spectrum features first decompose the mechanism, then use a probabilistic model for early warning, when In this way, the matching of candidate prescriptions and the issuance of actions can be triggered directly, which can intervene before the efficiency drops significantly and reduce the problem of deep cavity segments becoming slower and slower as they are processed.

[0074] This invention generates an event window based on the pulse switch state, and encapsulates the gap voltage sampling data, gap current sampling data, servo feed command data, flushing state data, rebound state data, and step segment number data into discharge event record data and forms event ledger data. This transforms the expression of operating conditions from continuous curve statistics to a discrete event structure, reducing the dependence on sampling frequency consistency and alignment processing.

[0075] By constructing event spectrum feature data using short-circuit cluster data and recovery quantile data, different efficiency decline patterns, such as multiple but dispersed short circuits versus short circuit clusters and arc tails leading to prolonged recovery, can be distinguished under the same caliber, avoiding the mechanistic confusion caused by relying solely on mean and frequency.

[0076] Based on the event integral energy data, the data is categorized by discharge event type to form input energy accumulation data, effective erosion energy data, short-circuit loss energy data, and arc loss energy data. The energy supply efficiency data is also calculated so that efficiency changes can be mapped to specific loss sources.

[0077] Data on chip retention index, media contamination index, and servo matching rate are generated and compared with event threshold data and applicable boundary data in historical prescription data to obtain threshold deviation data and prescription score data, thereby outputting candidate prescription data to replace generalized recommendations based solely on similarity retrieval.

[0078] The correction action template data in the candidate prescription data is parameterized to generate gap clearing pulse sequence parameter data, rebound rhythm parameter data, flushing duty cycle parameter data and servo gain switching parameter data to form correction action sequence data. After being issued, the power supply efficiency data and event spectrum feature data are regenerated, and the power supply efficiency difference data, maximum cluster length difference data and recovery quantile difference data are output to achieve closed-loop verification, improve the stability of the deep cavity section processing cycle and reduce the risk of abnormal electrode wear and rework.

[0079] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A system for analyzing electrical discharge machining efficiency based on historical working conditions, characterized in that, It includes an event ledger generation module, an event spectrum feature extraction module, an energy decomposition calculation module, a latent cause index generation module, a historical prescription matching module, and an action verification module. The event ledger generation module is used to collect gap voltage sampling data, gap current sampling data, pulse switch status data, servo feed command data, flushing status data, rebound status data and step segment number data, and generate event window data based on the pulse switch status data. Within the event window, it generates discharge event type marker data and event integral energy data, encapsulates them to form discharge event record data, and collects them into event ledger data. The event spectrum feature extraction module is used to generate short-circuit cluster data and recovery statistics based on event ledger data, and to summarize and generate event spectrum feature data. The energy decomposition calculation module is used to generate input energy accumulation data, effective erosion energy data, short-circuit loss energy data, and arc loss energy data based on event ledger data, and calculates energy supply efficiency data from the input energy accumulation data and effective erosion energy data. The latent cause index generation module is used to generate chip retention index data and media contamination index data based on event spectrum feature data, and to generate servo matching rate data based on servo feed command data and recovery statistics data. The historical prescription matching module is used to output candidate prescription data from historical prescription data based on energy supply efficiency data, chip retention index data, media contamination index data, servo matching rate data, and work step number data. The action verification module is used to generate and distribute corrective action sequence data based on candidate prescription data, and after distribution, regenerate energy supply efficiency data and event spectrum feature data based on updated event ledger data, and output verification result data.

2. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 1, characterized in that, The event ledger generation module includes a data acquisition unit, an event window generation unit, an event segmentation unit, and a ledger writing unit; The acquisition unit is used to acquire gap voltage sampling data, gap current sampling data, pulse switch status data, servo feed command data, flushing status data, rebound status data, and step segment number data; The event window generation unit is used to generate event window data based on the on / off edges of pulse switch state data. The event segmentation unit is used to generate discharge event type marker data, event integral energy data, and event context marker data within the event window based on the event window data. The ledger writing unit is used to encapsulate the work step number data, the discharge event type marker data, the event integral energy data, and the event context marker data into discharge event record data and write them into the event ledger data.

3. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 2, characterized in that, When the event segmentation unit generates discharge event record data, it satisfies the following: The event integral energy data is obtained by multiplying and accumulating the gap voltage sampling data and gap current sampling data within the event window according to the sampling interval; The discharge event type marking data includes normal discharge marking, short circuit marking, arc marking and open circuit marking, and is obtained by window statistics of gap voltage sampling data and gap current sampling data within the event window according to preset judgment rules; The event context marker data is obtained by combining the servo feed command data, flushing status data, and bounce status data corresponding to the start time of the event window.

4. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 3, characterized in that, The event spectrum feature extraction module includes a cluster construction unit and a recovery statistics unit; The cluster construction unit is used to retrieve discharge event record data corresponding to short circuit markers in the event ledger data. When the event start interval of two adjacent short circuit marker discharge event record data is less than the cluster interval threshold data, they are merged into the same short circuit cluster and short circuit cluster data is generated. The short circuit cluster data includes cluster length data and cluster energy accumulation data. The recovery statistics unit is used to generate post-event recovery time data, which is the time interval from the end of the discharge event record data of the short-circuit mark or arc mark to the start of the next normal discharge event record data, and calculates the recovery quantile data for the post-event recovery time data; The short-circuit cluster data and the recovery quantile data are written into the event spectrum feature data.

5. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 4, characterized in that, The energy decomposition calculation module includes an energy accumulation unit and an energy distribution unit; The energy accumulation unit is used to accumulate the event integral energy data in the event ledger data to obtain the input energy accumulation data; The energy distribution unit is used to classify and accumulate event integral energy data according to the discharge event type label data to obtain short-circuit loss energy data and arc loss energy data, and multiply the event integral energy data corresponding to the normal discharge label with the erosion correction coefficient data to obtain effective erosion energy data. The energy supply efficiency data is calculated from the effective erosion energy data and the cumulative input energy data.

6. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 5, characterized in that, The latent cause index generation module includes a retention index unit, a pollution index unit, and a servo matching unit. The retention index unit is used to generate chip retention index data by taking cluster length data, cluster energy accumulation data and recovery quantile data as inputs. The pollution index unit is used to generate medium pollution index data by taking the arc-marked event count percentage data, the arc-marked event integral energy percentage data, and the recovery quantile data as inputs. The servo matching unit is used to generate servo matching rate data by taking the window fluctuation data and recovery quantile data of the servo feed command data as input.

7. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 6, characterized in that, The historical prescription matching module includes a prescription inventory unit and a matching scoring unit; The prescription inventory unit stores historical prescription data, which includes at least applicable boundary data, event threshold data, corrective action template data, and erosion correction coefficient data. The applicable boundary data includes at least material type data, electrode type data, processing polarity data, and process step number data range; The event threshold data includes at least cluster interval threshold data, maximum cluster length threshold data, and recovery quantile threshold data.

8. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 7, characterized in that, The matching and scoring unit is used to first filter historical prescription data based on the work step number data and the applicable boundary data to obtain a boundary candidate set, then compare the cluster length data, recovery quantile data and arc mark event count ratio data in the event spectrum feature data with the event threshold data to obtain threshold deviation data, and generate prescription scoring data based on the threshold deviation data and the energy supply efficiency data, and output historical prescription data that meet the preset conditions from the boundary candidate set as candidate prescription data.

9. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 8, characterized in that, The action verification module includes an action parameterization unit and a verification calculation unit; The action parameterization unit is used to parameterize the correction action template data in the candidate prescription data to generate gap pulse sequence parameter data, rebound rhythm parameter data, flushing duty cycle parameter data and servo gain switching parameter data, and combine the parameter data to generate correction action sequence data.

10. The electrical discharge machining efficiency analysis system based on historical working conditions as described in claim 9, characterized in that, The verification calculation unit is used to read the updated event ledger data after the corrective action sequence data is issued, regenerate the power supply efficiency data and event spectrum feature data, and output the verification result data. The verification results data include at least the power supply efficiency difference data, the maximum cluster length difference data, and the recovery quantile difference data.

Images