A control system for servicing electrical equipment of machine tools

Abstract

The application discloses a control system for machine tool electrical equipment maintenance, and relates to the technical field of intelligent manufacturing. The control system comprises: an intention coding module, which decomposes a standard operation process into a sequence of operation intention elements; a real-time intention recognition module, which collects multi-modal operation data streams and recognizes the current operation intention through time sequence pattern matching; a multi-dimensional deviation detection module, which detects deviations from four dimensions of sequence, parameters, time sequence and integrity and generates a deviation vector; a rectification rule engine module, which generates rectification instructions according to the deviation vector and forcibly triggers re-verification; and a closed-loop confirmation module, which generates an operation log and updates a digital twin model after all intention elements are verified. The system applies the closed-loop feedback principle to the maintenance execution link, realizes real-time verification and self-adaptive rectification of the maintenance operation quality, solves the problem that maintenance operation deviations cannot be perceived and corrected in real time in the prior art, and can reduce the maintenance rework rate.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and more specifically, to a control system for the maintenance of machine tool electrical equipment. Background Technology

[0002] As industrial equipment becomes increasingly complex, the electrical systems of high-end equipment such as CNC machine tools are becoming more and more integrated. The maintenance quality and efficiency of key electrical components such as servo drives, programmable logic controllers, and encoders directly affect the overall efficiency of the equipment and the production costs of enterprises. Currently, intelligent operation and maintenance technology has evolved from traditional reactive maintenance to predictive maintenance, and intelligent operation and maintenance systems based on technologies such as 5G and edge computing have achieved results.

[0003] For example, Chinese invention patent CN120491535A discloses an intelligent operation and maintenance management system and method for industrial equipment based on 5G-MOM. Through a multi-source heterogeneous data acquisition layer, an edge computing node cluster, a cloud-based intelligent analysis platform, and a human-machine collaborative operation and maintenance terminal, it achieves real-time monitoring of equipment health status and predictive maintenance decision optimization. Chinese invention patent CN105446266A discloses a real-time remote monitoring, central control, dispatch, and automatic repair system for CNC machine tools. Through wireless signal authentication, a hierarchical architecture of sub-control and central control, an expert system, and a problem database, it achieves networked monitoring and automatic repair guidance for CNC machine tools.

[0004] Specifically, the existing technology has the following technical problems:

[0005] (1) The quality of maintenance operations cannot be verified in real time. There is a lack of systematic real-time quantitative evaluation methods to assess whether maintenance personnel have completed high-precision operations (such as servo drive parameter calibration, encoder phase adjustment, and electrical terminal torque tightening) according to standards.

[0006] (2) Maintenance deviations cannot be detected and corrected in time. Minor deviations are often only discovered after maintenance is completed through abnormal equipment status, resulting in rework and extended downtime.

[0007] (3) The digital twin model is disconnected from the maintenance execution process. The existing digital twin is mainly used for equipment status mapping and life prediction, and has failed to form a real-time interactive closed loop with maintenance operations. Summary of the Invention

[0008] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a control system for machine tool electrical equipment maintenance. The technical problem to be solved is that during the maintenance of machine tool electrical equipment, the deviation between the actual operation of the maintenance personnel and the standard operating procedure cannot be perceived and corrected by the system in real time, resulting in uncontrollable maintenance quality and high rework rate.

[0009] To achieve the above objectives, the present invention provides a control system for the maintenance of machine tool electrical equipment.

[0010] The system includes: an intent encoding module, a real-time intent recognition module, a multi-dimensional deviation detection module, a deviation correction rule engine module, and a closed-loop confirmation module.

[0011] The intent encoding module is used to decompose the standard operating procedure corresponding to the maintenance work order into a sequence of sequentially arranged operation intent elements. Each operation intent element contains at least the operation type, target parameter value, allowable deviation range, and time constraint.

[0012] The real-time intent recognition module is used to collect multimodal operation data streams in real time during maintenance operations. It extracts temporal features through a sliding window and matches them with the preset feature templates of each intent element to identify the currently executing operation intent element and record the actual parameter values.

[0013] The multi-dimensional deviation detection module compares the identified current operation intent element with the expected intent element, detecting deviations from four dimensions: sequence consistency, parameter deviation, timing compliance, and operation integrity, generating a four-dimensional deviation vector. Each component corresponds to the sequence deviation, parameter deviation, timing deviation, and integrity deviation, respectively.

[0014] The deviation correction rule engine module is used to generate a deviation correction instruction based on the deviation vector and call the corresponding deviation correction strategy. This instruction is then pushed to the human-computer interaction interface and forcibly triggered to re-execute the real-time intent recognition module and the multi-dimensional deviation detection module. The process continues until all components of the deviation vector of the current intent element are zero, at which point the intent element is marked as verified and proceeds to the next intent element.

[0015] The closed-loop verification module is used to generate an operation log and update the device's digital twin model after all intent elements have been verified.

[0016] Furthermore, the data structure of the operation intent element is a quintuple:

[0017] Where ID is the sequential number, Type is the operation type enumeration, Target is the target parameter value, Range is the allowable deviation range, and Time is the time constraint.

[0018] The enumeration of operation types includes torque tightening, electrical measurement, parameter adjustment, connector operation, and calibration; the time constraints include the longest allowable delay time and the shortest effective duration of the operation.

[0019] As a further improvement, the real-time intent recognition module employs a dynamic time warping algorithm to calculate the actual temporal feature sequence. With preset feature template sequence similarity between The similarity is defined as follows: ,in For dynamic time-warped distance; when Greater than the preset threshold It determines whether the current operation matches the intent element.

[0020] Furthermore, the multi-dimensional deviation detection module specifically includes:

[0021] The sequence deviation detection unit is used to compare whether the sequence number of the current intent element is equal to the sequence number of the expected intent element;

[0022] The parameter deviation detection unit is used to calculate the actual parameter values. The absolute value of the deviation from the target parameter value Target and judge Is it greater than the allowable deviation range (Range)?

[0023] The timing deviation detection unit is used to check whether the start delay time and duration of the operation meet the time constraints.

[0024] The integrity deviation detection unit is used to analyze whether the waveform of the operating signal contains a complete critical stage.

[0025] As a further improvement, the correction rule engine module incorporates a correction strategy based on a decision tree structure, the correction strategy including:

[0026] For sequence deviations, generate instructions to revert to the correct intent element;

[0027] For parameters with deviation and operation types of parameter adjustment or torque tightening, a compensation command is generated based on the difference between the target value and the actual value.

[0028] For parameters with deviation and operation type of electrical measurement, generate a remeasurement instruction;

[0029] For timing deviations, generate operation rhythm adjustment prompts;

[0030] For integrity deviations, generate instructions to continue execution until completion.

[0031] Furthermore, the system also includes a self-learning module for a correction knowledge base, used to record the type and amount of deviation for each correction event, and to count the frequency of occurrence of various deviations on each operational intent element. When the frequency of occurrence of any deviation pattern exceeds a preset threshold... When this happens, the allowable deviation range for the intent element is automatically tightened, and the update rule is as follows:

[0032] ,in This refers to the allowable deviation range before tightening. This is the tightening coefficient. The value range is from 0.8 to 0.95.

[0033] Furthermore, the operation log generated by the closed-loop confirmation module includes: comparison data of expected parameter values ​​and actual parameter values ​​of each intent element, records of deviation occurrences and history of correction instructions, comparison data of total operation time and standard operation time, and final verification conclusions.

[0034] The operation log is encrypted, stored, and linked to the device's maintenance history. Furthermore, the closed-loop confirmation module is used for secondary cross-validation after updating the digital twin model: it calls the cloud-based intelligent analysis platform to perform a short-term operational status assessment of the repaired device, cross-compares the assessment results with the verification conclusions in the operation log, and if they match, the repair is confirmed as complete; otherwise, a review process is triggered.

[0035] As a further improvement, the system described in this invention can be applied to parameter calibration scenarios of CNC machine tool servo drives or encoder installation and adjustment scenarios; the multimodal operation data stream originates from at least one of a smart torque wrench, a smart multimeter, a parameter debugger, or a machine tool controller.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] (1) Real-time closed-loop verification of maintenance operation quality has been achieved. In the prior art, the system is in an open-loop state after the maintenance command is issued, and it is impossible to sense whether the on-site operation meets the standard. This system decomposes the standard operation process into a machine-understandable sequence of intent elements, identifies the current intent in real time during the operation, detects deviations in multiple dimensions, dynamically generates correction commands and forces closed-loop re-verification, and applies the feedback principle in control theory to the execution of machine tool electrical maintenance, thus solving the technical problem that maintenance operation deviations cannot be perceived and corrected in real time.

[0038] (2) Multi-dimensional deviation detection and differentiated correction are achieved. The system detects operations from four dimensions: sequence consistency, parameter deviation, timing compliance, and operational integrity. It can identify different types of maintenance quality problems, such as sequence errors, parameter deviations, timing anomalies, and incomplete operations. The correction rule engine automatically selects correction strategies such as rollback, compensation, retesting, speed adjustment, or continuation based on the deviation type, making the correction instructions specific into executable operation guidelines, thereby improving the correction efficiency and first-time success rate of maintenance personnel.

[0039] (3) Possesses self-learning capability. The system uses a self-learning module in its error correction knowledge base to statistically analyze the frequency of various deviations occurring on each operational intent element. When a certain deviation pattern occurs frequently, it automatically tightens the allowable deviation range for that operation (e.g., from...). Narrow to This improves the sensitivity to high-frequency deviations. This mechanism allows the system to adapt to the degradation characteristics of specific equipment or the operating habits of maintenance personnel during long-term operation, reducing the workload of manual parameter adjustment. Attached Figure Description

[0040] Figure 1 This is an overall flowchart of the control system for machine tool electrical equipment maintenance described in this invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example 1

[0043] As attached Figure 1 The control system for maintaining machine tool electrical equipment, as shown, is implemented as follows:

[0044] The overall system architecture is as follows:

[0045] As one embodiment of the present invention, the control system for machine tool electrical equipment maintenance includes an intent encoding module, a real-time intent recognition module, a multi-dimensional deviation detection module, a deviation correction rule engine module, and a closed-loop confirmation module.

[0046] The system also includes a self-learning module for the knowledge base for error correction. This module is connected to the output end of the multi-dimensional deviation detection module and the configuration end of the intent encoding module, respectively, and is used to record deviation events and automatically tighten the allowable deviation range.

[0047] Data interaction between modules is achieved through an internal communication bus or message queue.

[0048] The system can be deployed on edge computing nodes or on cloud servers, and interacts with on-site smart tools and human-computer interaction terminals via 5G networks.

[0049] The intelligent tools include, but are not limited to, intelligent torque wrenches, intelligent multimeters, and parameter adjusters; the human-computer interaction terminals include, but are not limited to, AR glasses and tablet computers.

[0050] The specific implementation of the intent encoding module is as follows:

[0051] The intent encoding module is used to decompose the standard operating procedure corresponding to a maintenance work order into a sequential sequence of operational intent elements. In practice, this module first obtains structured or semi-structured standard operating procedure text from a Manufacturing Operations Management System (MOM) or a maintenance knowledge base, and then identifies each independent operational unit using built-in parsing rules. These parsing rules include action verb matching based on regular expressions and dependency parsing based on natural language processing.

[0052] For each identified operation unit, the intent encoding module encodes it into a 5-tuple data structure: .in:

[0053] ID is a positive integer, and its value is incremented according to the order in which the operation units appear in the process.

[0054] Type is an enumeration value for operation types, selected from a predefined set, which includes: TORQUE (torque fastening type), MEASURE (electrical measurement type), ADJUST (parameter adjustment type), CONNECT (connector operation type), and CALIBRATE (calibration type).

[0055] Target is the target parameter value. For operation types without continuous target values, Target is set to an empty value.

[0056] Range represents the allowable deviation range. For operation types without continuous target values, Range is set to an empty value; for operation types with continuous target values, Range is a positive number representing the upper limit of the allowable absolute deviation, i.e., the actual operation value is allowed to be within a certain range. Within the range.

[0057] The Time constraint contains two optional subfields: maxDelay (maximum allowed delay time in seconds) and minDuration (minimum valid duration in seconds). Set maxDelay when the operation requires it to start within a certain time after the previous operation is completed; set minDuration when the operation itself requires a certain duration. Both can exist simultaneously, or only one can exist, or both can be empty.

[0058] The intent encoding module outputs an ordered list of intent elements, denoted as . ,in This represents the total number of intent elements. This list is sent to the real-time intent recognition module as a standard for subsequent verification.

[0059] The specific implementation of the real-time intent recognition module is as follows:

[0060] The real-time intent recognition module is used to collect multimodal operation data streams in real time during maintenance operations and identify the currently executing operation intent element through time-series pattern matching.

[0061] (I) Data Acquisition and Preprocessing

[0062] This module uses a fixed sampling frequency. Receive time series data from multiple data sources. The preferred range is 50Hz-200Hz, and in this embodiment, it is taken as... The data sources include: the torque-time curve output by the smart torque wrench, voltage or current measurements output by the smart multimeter, driver parameter register values ​​read by the parameter debugger, and hand spatial trajectory coordinates acquired by the AR device. These data constitute a multidimensional time series, denoted as... ,in The number of data dimensions. For discrete sampling times.

[0063] (ii) Sliding window settings

[0064] This module maintains a length of A sliding window. The value of is related to the sampling frequency. ,in The time window length (in seconds). The preferred range is 1 second to 3 seconds. In this embodiment, we take... seconds, then The data within the window is recorded as follows: As new sampling points arrive, the window slides forward, discarding the oldest data points and adding the newest ones.

[0065] (III) Feature Extraction

[0066] For the data within the window, the module extracts the corresponding temporal feature vectors based on the type of operation that may be performed at the moment. Feature extraction employs different algorithms depending on the type of operation:

[0067] For TORQUE type operations, the extracted features include: torque rise slope (obtained by fitting the rising segment data using the least squares method), peak torque (maximum value within the window), and plateau duration (torque value remains near the peak). (Time length within the range), and the slope of the falling edge.

[0068] For MEASURE type operations, the extracted features include: mean value during the stabilization period (the average value of data when the signal variance is lower than a preset threshold), variance during the stabilization period, and probe contact mark (determined based on the switching signal or impedance change of the contact sensor).

[0069] For ADJUST type operations, the extracted features include: parameter change trajectory (parameter value sequence at each sampling point), final value (stable value at the end of the window), and adjustment rate (parameter change per unit time).

[0070] For CONNECT type operations, the extracted features include: the inserted stroke curve (displacement sensor data sequence) and the locking feedback signal (position switch signal of the locking mechanism).

[0071] For CALIBRATE type operations, the extracted features include: calibration status flag sequence (status word read from the controller) and completion confirmation signal.

[0072] (iv) Template matching

[0073] The module pre-stores standard feature templates for each intent element. The template is generated as follows: during the system initialization phase, it is executed by skilled maintenance personnel according to standard operating procedures. Substandard operation ( The system records the feature sequence of each operation, and then obtains a virtual sequence through the dynamic time warping centroid algorithm. , making Minimum. This template is stored in a configuration file.

[0074] The module uses a dynamic time warping algorithm to calculate the actual feature sequence. With template Similarity between them. Dynamic time-warped distance. This is obtained by solving the following optimization problem:

[0075] ;

[0076] in, To align the path, the sequence and Pair the points in the data; This represents the total number of pairs in the alignment path; and They represent the first Taken from each pair and The index.

[0077] Similarity Defined as:

[0078] ;

[0079] The similarity value range is: A higher value indicates a higher degree of matching. The module presets a matching threshold. , The preferred value range is 0.80-0.90. This threshold is automatically determined during system initialization using a cross-validation method: a set of known standard operations is taken, and their similarity to the template is calculated; the minimum value is taken as the threshold. Alternatively, the operator can manually adjust the setting based on the actual false alarm rate. In this embodiment, the setting is... .

[0080] (v) Intent determination

[0081] when At that time, the module determines whether the operation within the current window matches the intent element corresponding to the template. The module outputs the identifier ID of the intent element and extracts the actual parameter value from the data. (For ADJUST and TORQUE operations, take the average value of the stable period;)

[0082] For the MEASURE operation, take the reading during the stable period; for the CONNECT and CALIBRATE operations, (Set as a status flag value). If multiple intent elements have a matching degree exceeding the threshold simultaneously, the one with the highest matching degree is taken as the output.

[0083] If no intent element matches the threshold, the recognition result from the previous moment is maintained or an "unrecognized" state is output, and data is collected in the next window.

[0084] The specific implementation of the multi-dimensional deviation detection module is as follows:

[0085] The multi-dimensional deviation detection module is used to compare the identified current operation intent element with the expected intent element and detect deviation from four dimensions.

[0086] This module contains four sub-units: sequence deviation detection unit, parameter deviation detection unit, timing deviation detection unit, and integrity deviation detection unit.

[0087] During implementation, this module maintains a progress pointer. The initial pointer points to the first intent element in the intent element sequence. When the real-time intent recognition module outputs a current intent element... (Including its identifier ID and actual parameter value) When the four sub-units are in operation, the following checks are performed in parallel:

[0088] (a) Sequence Deviation Detection

[0089] Sequence deviation detection unit reads and read the expected intent element. The ID of (i.e., Sequence[p]). If the two are equal, the order is correct, and it is set. If they are not equal, then the order is incorrect, and the value should be set to zero. And record the intent meta ID that was actually executed.

[0090] (ii) Parameter deviation detection

[0091] The parameter deviation detection unit first determines Does it belong to the type with continuous target values ​​(ADJUST or TORQUE)? If so, calculate the absolute value of the deviation. Then compare. and :

[0092] like If the parameters are qualified, set... ;

[0093] like If the parameter is out of tolerance, set... And record the deviation. .

[0094] like If the type has no continuous target value (MEASURE, CONNECT, CALIBRATE), then set it directly. .

[0095] (III) Timing Deviation Detection

[0096] Timing Deviation Detection Unit Inspection constraint:

[0097] like If the `maxDelay` field is included, the start time of the current operation is calculated. Completion time of the previous intention element The difference .

[0098] Completion time of the previous intention element This is recorded by this module when the previous intent element verification passed. If If so, it is determined that the delay exceeds the limit, and the setting is... If the timeout is not recorded, then the sub-item is considered qualified.

[0099] like If the minDuration field is included, the duration of the current operation is calculated. (The time interval from the start to the end of the operation). The start and end times of the operation are automatically marked by the real-time intent recognition module based on the characteristic points of the signal waveform. If If the time limit is insufficient, it will be set to... (If the limit has already been exceeded due to delay, keep it at 1), and record the insufficient amount; otherwise, the sub-item is qualified.

[0100] like If the field is empty or does not exist, the corresponding sub-item will automatically qualify.

[0101] in, The method for determining the value is as follows: starting from the starting point of the feature sequence, calculate the rate of change of adjacent sampling points, and mark it as the first time the rate of change exceeds a preset threshold. ; The method for determining it is as follows:

[0102] For the TORQUE operation, it is marked as such when the torque value drops below 10% of the peak torque. ;

[0103] For the ADJUST operation, a condition is marked as such when the rate of change of the parameter is less than 5% of the maximum rate of change for five consecutive sampling points. ; .

[0104] (iv) Integrity Deviation Detection

[0105] The integrity deviation detection unit analyzes whether the waveform of the operation signal contains a complete critical stage. The specific judgment rules are as follows:

[0106] For the TORQUE operation: detect whether the torque waveform exhibits a three-segment pattern of "rise - plateau - fall".

[0107] The criteria for judgment are: the rising segment has a clear positive slope interval (slope greater than a preset threshold), and the plateau segment has at least continuous... sampling points ( Pick (Where 0.5 seconds is an empirical value set based on the minimum effective duration of a typical torque operating plateau, which can be adjusted according to actual process requirements) the rate of change of torque value is less than the peak torque. The descending segment has a clear negative slope interval. If all three segments exist, the condition is considered complete, and the condition is set to... Otherwise, set .

[0108] For MEASURE operation: Check whether the probe contact signal changes from "not in contact" to "in contact" and remains stable for at least 0.5 seconds, and complete the reading during this period. If this condition is met, the operation is considered complete.

[0109] For the ADJUST operation: Detect whether the parameter change trajectory monotonically approaches the target value from the initial value and eventually stabilizes within the allowable deviation range, and remains stable for at least 0.5 seconds after stabilization. If this condition is met, the operation is considered complete.

[0110] For the CONNECT operation: check if the insertion stroke curve shows a clear inflection point of slope change, and if the final stroke reaches the lock position. If these conditions are met, the operation is considered complete.

[0111] For CALIBRATE operations: Check if the calibration completion flag changes from 0 to 1, or if a calibration confirmation signal appears. If both conditions are met, the operation is considered complete.

[0112] If the waveform shape is complete, then set If a key stage is missing, then set aside .

[0113] (v) Deviation Vector Output and Progress Update

[0114] The outputs of the four sub-units are combined into a four-dimensional bias vector. This vector, along with details of the deviations in each dimension (such as the amount of deviation), (Excess amount, insufficient amount, etc.) are all output to the correction rule engine module.

[0115] when When the time is right, it indicates that the current intent element has no deviation. The multi-dimensional deviation detection module will move the progress pointer... Increase by 1 ( Record the completion time of the current intent element. And continue to wait for the real-time intent recognition module to output the recognition result of the next intent element. When When the time is reached, it indicates that all intent elements have passed verification, and the closed-loop confirmation module is triggered.

[0116] when At that time, the progress pointer The module remains unchanged and waits for the correction rule engine module to process it before re-detecting.

[0117] The specific implementation of the error correction rule engine module is as follows:

[0118] The error correction rule engine module is used to determine the error vector. It generates correction instructions based on deviation details and forces a re-verification.

[0119] During implementation, this module has a built-in rule set with a decision tree structure. The rule set is expressed as a bias vector. The combination of components serves as the condition, and the correction instruction serves as the conclusion. The specific rules are as follows:

[0120] Rule 1 (Prioritize Order Deviation): If If so, regardless of other components, a rollback instruction will be generated first.

[0121] The data structure of this instruction includes: a Type field with the value "ROLLBACK", and a TargetID field with the value... The Message field reads, "The current operation is not as expected. Please cancel the current operation and return to step [step 1]." "(in Replace with ).

[0122] At the same time, this command triggers the system to adjust the progress pointer. Reset to the position corresponding to the expected ID and clear the verification status of the current intent element.

[0123] Rule 2 (Parameter Deviation Handling): If and According to Further differentiated processing:

[0124] like If the value is ADJUST or TORQUE, a compensation instruction is generated. Compensation amount. .

[0125] Command content: Type="COMPENSATE", Delta=" Message="Current value is The target value is Target. Please add / release C units. Use "Append" for positive numbers and "Release" for negative numbers. If the absolute value is less than the minimum adjustable resolution, the message "Please fine-tune to the target value" will be displayed.

[0126] like If MEASURE is selected, a retest command will be generated. Command content: Type="REDO", Message="Measurement deviation exceeds the limit, please remeasure".

[0127] Rule 3 (Timing Deviation Handling): If and Then a speed adjustment command is generated. The specific type is determined based on the deviation details:

[0128] If the deviation details include a timeout, then Message = "Operation start delay is too long, please speed up the operation".

[0129] If the deviation details include insufficient quantity, then Message = "Operation duration is insufficient, please continue the action for at least minDuration seconds" (where minDuration is taken as...). (value).

[0130] Rule 4 (Handling Integrity Deviations): If and and If so, a continuation instruction will be generated. Message = "Operation not completed, please continue execution until complete".

[0131] When multiple deviations exist simultaneously, the rule engine processes them sequentially according to the rules mentioned above, generating only the highest priority correction instruction each time.

[0132] After the correction instruction is generated, the module pushes it to the human-computer interaction interface. The push method includes text display and voice broadcast. At the same time, the module sends a forced re-verification signal to the real-time intent recognition module through the feedback output terminal. This signal causes the real-time intent recognition module to clear the data cache of the current sliding window, reset the internal state machine to the state of waiting for a new operation to start, and immediately start re-collecting data.

[0133] At the same time, the forced retest signal is also sent to the reset terminal of the multi-dimensional deviation detection module, causing the multi-dimensional deviation detection module to clear the deviation state of the current intent element, reset the internal detection flag, and wait for the real-time intent recognition module to output a new recognition result before re-performing deviation detection.

[0134] This closed-loop process is repeated until the deviation vector of the current intent element is reached. Only after a certain time is the progress pointer allowed to move forward.

[0135] The specific implementation of the self-learning module of the knowledge base for error correction is as follows:

[0136] As a further implementation, the system also includes a self-learning module for a correction knowledge base, which learns from historical correction events and automatically optimizes the allowable deviation range.

[0137] (a) Record of corrective events

[0138] This module maintains a deviation event database. After each correction is completed, the system records a structured event record containing the following fields: Intent Meta ID. Deviation type (values ​​are SEQ, PARAM, TIMING, COMPLETE), deviation amount Delta, applied correction strategy Correction type, and whether the correction was verified (Boolean value).

[0139] (II) Frequency Statistics and High-Incidence Pattern Mining

[0140] This module uses a sliding time window as the statistical period. The switching logic for the sliding time window is as follows:

[0141] Simultaneously maintain the event count window and the calendar time window. Each time a maintenance is completed, the event count is incremented by 1; when the event count reaches 100, statistics are immediately triggered and the window is reset.

[0142] If the event count does not reach 100 for an extended period, a statistical analysis will be triggered every 30 calendar days (based on the time of the last statistical analysis) and the event count will be reset.

[0143] Suppose a certain intent element is inside the window The total number of times it was executed was (Statistics from the closed-loop confirmation module), including occurrence types The number of deviations is Then frequency Preset a frequency threshold. , The preferred value range is 0.15-0.25, and in this embodiment, it is taken as... .when At that time, the deviation pattern (intent element) Deviation type () is marked as "high incidence deviation pattern".

[0144] (iii) The allowable deviation range is automatically tightened.

[0145] For parameter deviations in high-frequency deviation patterns ( This module automatically triggers a tightening of the allowable deviation range. The update rule is:

[0146] ;

[0147] in This refers to the allowable deviation range before tightening. This is the tightening coefficient. The preferred value range is 0.8 to 0.95, and in this embodiment, it is taken as... After tightening The value is written back to the Range field of the corresponding intent element in the intent encoding module and persisted. Simultaneously, the module records the time, original value, new value, and triggering reason for this adjustment in the system log.

[0148] (iv) Lower limit protection

[0149] To prevent excessive tightening from causing frequent misjudgments of normal operations, the module has a lower limit protection mechanism. The system pre-stores the minimum allowable tolerance (MinRange) for each intent element, which is read from the equipment manual or process specification; if not provided, it is set to 50% of the original Range by default. If the minimum range is reached, the lower limit MinRange will be used as the standard, and the range will not be tightened further. An alarm will be recorded.

[0150] The specific implementation of the closed-loop confirmation module is as follows:

[0151] The closed-loop confirmation module is used to generate an operation log and update the device's digital twin model after all intent elements have been verified.

[0152] During implementation, this module continuously monitors the progress pointer. The state. When the multi-dimensional deviation detection module issues... When the signal ( (Total number of intent elements), determining that all intent elements have passed verification. The module performs the following operations sequentially:

[0153] (a) Generate operation log

[0154] Log content should be in a structured format and should contain at least the following fields:

[0155] Device ID;

[0156] WorkOrderID (Work Order Number)

[0157] Operation timestamp (accurate to milliseconds);

[0158] Operator ID;

[0159] Intent Element Verification Record List: An array where each element corresponds to an intent element, containing: Expected parameter value (Target) and Actual parameter value. , deviation amount Validation results in four dimensions ( Vector components), correction instruction history (timestamp of each correction, instruction type, instruction content, and response result).

[0160] TotalTime (the time interval from the start of the first intent to the successful verification of the last intent)

[0161] Standard Time (the average time calculated from historical data or the standard duration specified by the process).

[0162] Final verification conclusion (value: "PASS")

[0163] After the operation log is generated, the module encrypts and stores it in a local database or cloud storage, and associates the log identifier with the device's maintenance history.

[0164] (ii) Updating the digital twin model

[0165] The module synchronizes the actual execution data of this maintenance operation to the digital twin model of the equipment via an API interface. The digital twin model uses this data to perform the following updates: update the equipment health status baseline, update the maintenance history, and adjust the input parameters of the life prediction model.

[0166] (iii) Secondary cross-validation

[0167] In a preferred implementation, after updating the digital twin model, the module invokes a cloud-based intelligent analysis platform to perform a short-term operational status assessment of the repaired equipment. The specific process is as follows: the platform issues a command to put the equipment into trial operation mode, collects key operating parameters (such as vibration amplitude, current waveform, and temperature distribution), compares them with the equipment's health baseline, and generates an operational status score. The score is obtained based on the Mahalanobis distance normalized from the baseline, with 80 points corresponding to a distance... The module cross-references the score with the validation conclusions from the operation logs.

[0168] If the operation log conclusion is PASS and the running status score is... Once the score is reached, the repair is confirmed as complete, and the work order is closed.

[0169] If the operation log conclusion is PASS but the running status score is... If the score is low, the review process will be triggered, and an engineer will be notified to conduct an inspection.

[0170] If the operation log concludes as FAIL, then the exception will be recorded for subsequent analysis.

[0171] After the secondary verification is completed, the module sends a work order completion signal to the manufacturing operations management system, and the system returns to standby mode, waiting for the next maintenance work order.

[0172] As a further application implementation, the system described in this invention is applied to the parameter calibration scenario of a CNC machine tool servo drive. In this scenario, the standard operating procedure includes steps such as reading drive parameters, modifying parameter values, and saving parameters. The multimodal operation data stream originates from a parameter debugger and a smart multimeter.

[0173] As another application implementation, the system described in this invention is applied to the installation and adjustment of encoders on CNC machine tools. In this scenario, the standard operating procedure includes steps such as installing the encoder, adjusting the air gap, tightening the screws, and measuring the feedback signal. The multimodal operation data stream originates from a displacement sensor, a smart torque wrench, and an oscilloscope.

[0174] In the above scenarios, the multimodal operation data stream originates from at least one or more of the following devices: smart torque wrench, smart multimeter, parameter debugger, and machine tool controller.

[0175] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any person skilled in the art can make many possible variations and modifications to the technical solution of the present invention, or modify it into equivalent embodiments, without departing from the scope of the present invention. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention, without departing from the scope of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A control system for maintaining machine tool electrical equipment, characterized in that, include: The intent encoding module is used to decompose the standard operating procedure corresponding to the maintenance work order into a sequence of sequentially arranged operation intent elements. Each operation intent element contains at least the operation type, target parameter value, allowable deviation range, and time constraint. The real-time intent recognition module is used to collect multimodal operation data streams in real time during maintenance operations. It extracts temporal features through a sliding window and matches them with the preset feature templates of each intent element to identify the currently executing operation intent element and record the actual parameter values. The multi-dimensional deviation detection module is used to compare the identified current operation intention element with the expected intention element, and detect deviations from four dimensions: sequence consistency, parameter deviation, timing compliance and operation integrity, and generate a four-dimensional deviation vector, with each component corresponding to sequence deviation, parameter deviation, timing deviation and integrity deviation respectively. The deviation correction rule engine module is used to generate a deviation correction instruction based on the deviation vector by calling the corresponding deviation correction strategy, push it to the human-computer interaction interface, and force the real-time intent recognition module and the multi-dimensional deviation detection module to re-execute until all components of the deviation vector of the current intent element are zero, at which point the intent element is marked as verified and enters the next intent element. The closed-loop confirmation module is used to generate an operation log and update the device's digital twin model after all intent elements have been verified.

2. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, The data structure of the operation intent element is a quintuple: Where ID is the sequential number, Type is the operation type enumeration, Target is the target parameter value, Range is the allowable deviation range, and Time is the time constraint; the operation type enumeration includes torque tightening, electrical measurement, parameter adjustment, connector operation, and calibration; the time constraint includes the longest allowable delay time and the shortest effective duration of the operation.

3. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, The real-time intent recognition module uses a dynamic time warping algorithm to calculate the actual time-series feature sequence. With preset feature template sequence similarity between The similarity is defined as: ； in For dynamic time-warped distance; when Greater than the preset threshold It determines whether the current operation matches the intent element.

4. The control system for machine tool electrical equipment maintenance according to claim 1 or 2, characterized in that, The multi-dimensional deviation detection module includes: The sequence deviation detection unit is used to compare whether the sequence number of the current intent element is equal to the sequence number of the expected intent element; The parameter deviation detection unit is used to calculate the actual parameter values. The absolute value of the deviation from the target parameter value Target and judge Is it greater than the allowable deviation range (Range)? The timing deviation detection unit is used to check whether the start delay time and duration of the operation meet the time constraints. The integrity deviation detection unit is used to analyze whether the waveform of the operating signal contains a complete critical stage.

5. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, The error correction rule engine module has a built-in error correction strategy based on a decision tree structure, and the error correction strategy includes: For sequence deviations, generate instructions to revert to the correct intent element; For parameters with deviation and operation types of parameter adjustment or torque tightening, a compensation command is generated based on the difference between the target value and the actual value. For parameters with deviation and operation type of electrical measurement, generate a remeasurement instruction; For timing deviations, generate operation rhythm adjustment prompts; For integrity deviations, generate instructions to continue execution until completion.

6. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, It also includes a self-learning module for the error correction knowledge base, used to record the type and amount of deviation for each error correction event, and to count the frequency of occurrence of various types of deviations on each operational intent element. When the frequency of occurrence of any deviation pattern exceeds a preset threshold... When this happens, the allowable deviation range for the intent element is automatically tightened, and the update rule is as follows: ； in This refers to the allowable deviation range before tightening. This is the tightening coefficient. The value range is from 0.8 to 0.

95.

7. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, The operation log generated by the closed-loop confirmation module includes: comparison data of expected parameter values ​​and actual parameter values ​​of each intent element, records of deviation occurrences and history of correction instructions, comparison data of total operation time and standard operation time, and final verification conclusions; the operation log is encrypted, stored, and associated with the equipment's maintenance history.

8. The control system for machine tool electrical equipment maintenance according to claim 1, characterized in that, The closed-loop confirmation module is also used to call the cloud-based intelligent analysis platform to perform a short-term operational status assessment of the repaired equipment after updating the digital twin model, and to cross-compare the assessment results with the verification conclusions of the operation log. If they match, the repair is confirmed to be complete; if they do not match, a review process is triggered.

Images