A business process mining-based enterprise digital transformation evaluation method and system

Abstract

The application discloses a kind of enterprise digital transformation evaluation method and system based on business process mining, it is related to data processing technical field, including: in response to scheduling parameter modification event, obtain the modification action sequence and production environment state sequence in system log;Modification action sequence and production environment state sequence are carried out time window alignment processing, obtain alignment state sequence;Intention proportion calculation is carried out to modification action sequence based on alignment state sequence and preset parameter constraint boundary, obtain the intention quantization value containing resistance intention value and rectification intention value;Based on intention quantization value, preset digital transformation initial score is corrected processing, obtains target transformation evaluation value.The application can distinguish whether artificial modification scheduling parameter is resistance or system is separated from reality, accurately quantifies modification intention, improves the accuracy of enterprise digital transformation evaluation.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for evaluating enterprise digital transformation based on business process mining. Background Technology

[0002] As enterprises continue to advance their digital transformation, manufacturing companies are widely introducing intelligent scheduling systems in their production scheduling processes, attempting to replace traditional manual experience-based scheduling with intelligent methods. During the actual operation of an intelligent scheduling system, the system automatically generates scheduling parameter suggestions based on production data.

[0003] However, workshop schedulers often lack trust in newly launched intelligent scheduling systems, frequently modifying the scheduling parameters generated by the system manually through the system interface. Existing business process analysis techniques, when addressing this issue, can only retrieve "scheduling parameters modified" records from the system logs, failing to reveal the reasons behind these modifications. During enterprise digital transformation assessments, if schedulers arbitrarily modify scheduling parameters based on past experience, it indicates resistance to the new tool, suggesting the intelligent scheduling system has not been truly adopted, significantly diminishing the effectiveness of the enterprise's digital transformation. Conversely, if schedulers modify parameters because the intelligent scheduling system's suggestions do not align with actual workshop production conditions, it indicates a design flaw in the intelligent scheduling system itself.

[0004] Existing business process mining technologies only analyze process logs and cannot distinguish whether manual modifications to automated results stem from human resistance to the new tool or from the tool itself being impractical. This leads to a lack of basis for evaluating enterprise digital transformation and fails to accurately reflect whether intelligent scheduling systems have played a real role in production scheduling. This blind spot in evaluation seriously hinders manufacturing enterprises from optimizing human-machine collaboration and improving intelligent scheduling systems. Summary of the Invention

[0005] In view of the aforementioned problems, this application is hereby filed.

[0006] Therefore, this application provides a method and system for evaluating enterprise digital transformation based on business process mining, which can solve the problem that existing process mining cannot distinguish whether manual modification of scheduling parameters is due to resistance or the system being out of touch with reality, resulting in distorted evaluation.

[0007] To solve the above-mentioned technical problems, this application provides the following technical solution: Firstly, this application provides a method for evaluating enterprise digital transformation based on business process mining, including: in response to the triggering of a production scheduling parameter modification event, obtaining a sequence of modification actions and a sequence of production environment statuses in the system log, wherein the sequence of modification actions includes a timestamp of manually modified parameters and the modified value; The modification action sequence and the production environment state sequence are aligned by time window to obtain an aligned state sequence, which contains the correspondence between modification actions and abnormal events in the production environment. Based on the alignment state sequence and preset parameter constraint boundaries, the intention ratio of the modification action sequence is calculated to obtain the intention quantification value, which includes the resistance intention value and the correction intention value. The initial score for digital transformation is corrected based on the intent quantification value to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process.

[0008] Preferably, the step of performing time window alignment processing on the modification action sequence and the production environment state sequence to obtain an aligned state sequence includes: Obtain the timestamp of the modified action in the modified action sequence as the modification time, which indicates the specific moment when the scheduler performs the parameter modification operation. The timestamps of abnormal events in the production environment state sequence are obtained as abnormal times, which indicate the specific moments when equipment failures or material shortages occur on the workshop floor. The time difference is calculated based on the modification time and the anomaly time. The time difference represents the time interval between the parameter modification operation and the occurrence of the on-site anomaly. Abnormal events with a time difference less than a preset time window are considered as associated abnormal events. Associated abnormal events indicate on-site abnormalities that have a potential causal relationship with parameter modification operations. The status labels of associated abnormal events are merged into the modification action sequence to form an aligned state sequence, which is used to provide the environmental context features of each modification action.

[0009] Preferably, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain the intent quantification value includes: In response to the absence of associated abnormal events in the alignment state sequence, the corresponding modification action in the modification action sequence is taken as a resisted modification action. The resisted modification action represents a manual modification operation without on-site abnormal support. Obtain the parameter modification range and preset parameter allowable fluctuation range of the resisted modification action. The parameter modification range represents the absolute difference between the modified value and the original value. The resistance intention value is calculated based on the ratio of the parameter modification magnitude of the resisted modification action to the preset parameter allowable fluctuation range. The resistance intention value represents the degree to which the dispatcher subjectively rejects the intelligent system. The resistance intention value is used as the intention quantification value, which is used to quantify the subjective resistance component in the manual modification operation.

[0010] Preferably, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain the intent quantification value includes: In response to the presence of associated abnormal events in the alignment state sequence, the corresponding modification actions in the modification action sequence are taken as potential correction actions. Potential correction actions indicate manual modification operations supported by on-site abnormalities. Extract the anomaly level of associated abnormal events and the parameter modification range of potential corrective actions. The anomaly level indicates the severity of the on-site anomaly. The preset parameter constraint boundary is determined based on the anomaly level. The preset parameter constraint boundary represents the maximum reasonable range of manual modification allowed under the corresponding anomaly level. In response to parameter modification magnitude being less than or equal to the preset parameter constraint boundary, potential correction actions are treated as reasonable correction actions, and a correction intention value is calculated. This correction intention value is used as the intention quantification value, which represents the reasonable correction component in the manual modification operation.

[0011] Preferably, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain the intent quantification value further includes: In response to parameter modification exceeding the preset parameter constraint boundary, potential correction actions are treated as mixed modification actions. Mixed modification actions indicate modification operations where the modification range exceeds the reasonable correction range. The modification amount corresponding to the preset parameter constraint boundary is obtained as the reasonable correction amount. The reasonable correction amount represents the minimum modification value required to deal with on-site anomalies. The amount of modification that exceeds the preset parameter constraint boundary is taken as the excessive resistance amount. The excessive resistance amount represents the additional modification value that exceeds the reasonable correction range. Based on the ratio of reasonable correction amount to excessive resistance amount, calculate the correction intention value and resistance intention value of the mixed modification action. The correction intention value and resistance intention value are used together as the intention quantification value. The intention quantification value is used to decompose the dual intention components in the mixed modification action.

[0012] Preferably, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value includes: In response to the presence of resisting intent values ​​in the quantitative intent values, the overall resistance index is obtained by statistically analyzing all resisting intent values. The overall resistance index represents the cumulative degree of subjective rejection of the intelligent system throughout the entire evaluation period. Obtain the initial score of human-machine collaboration in the preset initial score of digital transformation. The initial score of human-machine collaboration represents the collaboration baseline score before considering resistance to modification operations. The collaborative deduction amount is calculated by multiplying the overall resistance index by the preset collaborative deduction coefficient. The collaborative deduction amount represents the negative impact of the resistance operation on the degree of collaboration. Subtracting the collaboration deduction from the initial score of human-machine collaboration yields the target score of human-machine collaboration, which reflects the actual level of collaboration between humans and the system.

[0013] Preferably, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value further includes: In response to the presence of corrective intent values ​​in the quantitative intent values, the overall corrective index is obtained by statistically analyzing all corrective intent values. The overall corrective index represents the cumulative degree to which the intelligent system's suggestions deviate from reality throughout the entire evaluation period. Obtain the initial system maturity score from the preset initial score for digital transformation. The initial system maturity score represents the baseline score of system maturity before considering corrective modification operations. The maturity deduction amount is calculated by multiplying the overall correction index by the preset maturity deduction coefficient. The maturity deduction amount represents the negative impact of unreasonable system suggestions on maturity. Subtracting the maturity deduction from the initial system maturity score yields the system maturity target score, which reflects the actual applicability level of the intelligent scheduling system.

[0014] Preferably, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value further includes: Obtain the first preset evaluation coefficient corresponding to the target score of human-machine collaboration and the second preset evaluation coefficient corresponding to the target score of system maturity. The first preset evaluation coefficient and the second preset evaluation coefficient represent the importance of different evaluation dimensions in the overall evaluation. The target score of human-machine collaboration is weighted according to the first preset evaluation coefficient to obtain the collaboration evaluation component, which represents the weighted value of the actual collaboration level between human and system. The system maturity target score is weighted according to the second preset evaluation coefficient to obtain the maturity evaluation component, which represents the weighted value of the actual applicability level of the intelligent scheduling system. The collaborative assessment component and the maturity assessment component are summed to obtain the target transformation assessment value, which is used to comprehensively reflect the true effectiveness of the enterprise's digital transformation.

[0015] Preferably, the acquisition of the parameter modification magnitude of the resisted modification action and the preset parameter allowable fluctuation range includes: Obtain the original scheduling parameter values ​​and manually modified parameter values ​​corresponding to the resisted modification actions. The original scheduling parameter values ​​represent the initial values ​​automatically generated by the intelligent scheduling system. The absolute difference between the original production scheduling parameter values ​​and the manually modified parameter values ​​is calculated. The absolute difference is used as the parameter modification range, which represents the absolute change of the initial value by the manual modification operation. Obtain the normal operating condition parameter range in the production environment state sequence, and use the normal operating condition parameter range as the preset parameter allowable fluctuation range. The normal operating condition parameter range represents the normal fluctuation range of the parameter when there are no abnormal events. The parameter modification range is compared with the preset parameter allowable fluctuation range. The resistance intention value is determined based on the comparison result. The comparison result is used to reflect the degree to which the manual modification operation deviates from the normal fluctuation range.

[0016] Secondly, this application also provides an enterprise digital transformation assessment system based on business process mining, including: an acquisition module, used to acquire the sequence of modification actions and the sequence of production environment status in the system log in response to the triggering of a production scheduling parameter modification event; the sequence of modification actions includes the timestamp and modification value of manually modified parameters. The alignment module is used to perform time window alignment processing on the sequence of modification actions and the sequence of production environment states to obtain an aligned state sequence, which contains the correspondence between modification actions and abnormal events in the production environment. The calculation module is used to calculate the intention ratio of the modified action sequence based on the alignment state sequence and preset parameter constraint boundaries, and obtain the intention quantification value, which includes the resistance intention value and the correction intention value. The correction module is used to correct the preset initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process.

[0017] Implementing this application will have the following beneficial effects: This application provides a method and system for evaluating enterprise digital transformation based on business process mining. In response to a production scheduling parameter modification event, it obtains a sequence of modification actions and a sequence of production environment states from the system logs. The method then aligns the modification action sequence and the production environment state sequence using time windows to obtain an aligned state sequence. Based on the aligned state sequence and preset parameter constraint boundaries, it calculates the intention ratio of the modification action sequence to obtain a quantitative intention value that includes resistance intention values ​​and corrective intention values. Finally, it corrects the preset initial score for digital transformation based on the quantitative intention value to obtain the target transformation evaluation value. This invention can distinguish between manual modification of production scheduling parameters due to resistance or the system being out of touch with reality, accurately quantifying the modification intention and improving the accuracy of enterprise digital transformation evaluation. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is an overall flowchart of an enterprise digital transformation assessment method based on business process mining, which is the subject of this application. Figure 2 This is a flowchart of the intent quantification and assessment correction judgment process of an enterprise digital transformation assessment method based on business process mining, which is involved in this application. Figure 3 This is an abstract scenario diagram of the intent mapping and evaluation ecosystem of an enterprise digital transformation assessment method based on business process mining, which is involved in this application. Figure 4 This application relates to an intent-quantifying data flow network diagram of an enterprise digital transformation assessment method based on business process mining. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0021] Example 1, referring to Figures 1-4 This is the first embodiment of the present invention, which provides a method for assessing enterprise digital transformation based on business process mining, including: This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to implement this enterprise digital transformation assessment method based on business process mining, using multiple embodiments. Figure 1 A flowchart illustrating a business process mining-based enterprise digital transformation assessment method is shown, including: Step A1: In response to the production scheduling parameter modification event, obtain the modification action sequence and production environment status sequence from the system log. The modification action sequence includes the timestamp and modified value of the manually modified parameters. In manufacturing scenarios of enterprise digital transformation, the scheduling parameters generated by the intelligent scheduling system are often subject to manual intervention by workshop schedulers. Behind this intervention lies the scheduler's attitude towards the intelligent system and the intelligent system's ability to adapt to the actual production environment.

[0022] Existing evaluation systems merely record process logs superficially, only showing the appearance of parameter modifications without reaching the deeper motivations behind them. To break this blind spot, it is essential to first establish a basic data collection channel, incorporating manual modification behavior and the status of the on-site production environment into the collection scope. This will provide original data support for subsequent intent analysis and evaluation correction, ensuring that the evaluation process remains aligned with the actual production environment. Consequently, this will lay a solid data foundation for accurately reflecting the collaborative level of enterprise digital transformation.

[0023] In this invention, the modification action sequence includes the timestamp and modification value of manually modified parameters, while the production environment status sequence includes real-time status information such as equipment load, material availability, order priority changes, and sudden abnormal events in the production site within the corresponding time interval. The two types of sequences are aligned and matched through a unified timestamp to ensure that the site environment when the modification action occurs can be completely restored, avoiding deviations in subsequent analysis due to misaligned data timing, and providing accurate and aligned raw data input for subsequent process mining.

[0024] Step A2: Perform time window alignment processing on the modification action sequence and the production environment state sequence to obtain an aligned state sequence. The aligned state sequence contains the correspondence between modification actions and abnormal events in the production environment. It should be noted that the manual modification operations of the dispatcher are often not generated out of thin air. Abnormal events such as equipment failure or material shortage in the workshop are very likely to be the direct cause of the modification action. Through alignment processing, modification actions that occur within a similar time range can be bound to abnormal events in a spatiotemporal dimension, so that each modification action has environmental context characteristics. The establishment of this correspondence is a prerequisite for distinguishing between resistance intentions and corrective intentions. Without the alignment of state sequences, the intention judgment will lose its factual basis, resulting in distorted evaluation results.

[0025] In some embodiments, the step of performing time window alignment processing on the modification action sequence and the production environment state sequence to obtain an aligned state sequence includes: Step A2-1: Obtain the timestamp of the modification action in the modification action sequence as the modification time. The modification time indicates the specific moment when the scheduler performs the parameter modification operation. This application provides a method for extracting time anchors, the method including accurately locating time scales from a sequence of modification actions recorded in system logs.

[0026] Furthermore, iterate through each data record in the modification action sequence, parse the time field of the data record, extract the year, month, day, hour, minute, and second information contained in the time field, convert it into a standard timestamp format, and mark it as the modification time.

[0027] Specifically, the modification time serves as the time anchor point for the action, accurately reflecting the absolute moment when the scheduler triggers the parameter modification operation on the intelligent production scheduling system interface. This absolute moment is the basic benchmark for subsequent time comparison with abnormal events in the production environment.

[0028] If there are multiple consecutive modification records in the modification action sequence, the modification time sequence is extracted according to the order of triggering to ensure the temporal continuity of the time anchor points.

[0029] In practice, to address the potential inconsistency in time formats within system logs, all extracted time information is uniformly converted to a millisecond representation since the standard epoch. This eliminates interference from format differences in subsequent time difference calculations, ensuring high precision and consistency of modification times. Furthermore, the extracted modification times are associated with their corresponding modification values, forming a key-value pair structure, providing a clear data index for subsequent alignment calculations.

[0030] Step A2-2: Obtain the timestamps of abnormal events in the production environment state sequence as abnormal times. Abnormal times indicate the specific moments when equipment failure or material shortages occur in the workshop. Specifically, extracting the occurrence time of abnormal events from the production environment state sequence is key to establishing the environmental context.

[0031] The production environment status sequence includes equipment operation data collected by various sensors on the workshop floor and inventory data output by the material management system. When equipment operation data exceeds the normal threshold or inventory data falls below the safe level, the system generates an abnormal event record.

[0032] The abnormal time is the timestamp information carried in the abnormal event record. This timestamp information accurately indicates the specific moment when the equipment shutdown or material supply interruption was detected.

[0033] It should be noted that obtaining abnormal times requires time uniform calibration of abnormal signals from different sources to eliminate time offsets caused by sensor acquisition delays or network transmission delays, and to ensure that the abnormal time and the modified time are under the same reference clock system.

[0034] For equipment malfunction anomalies, the timestamp of the alarm signal issued by the equipment control system is used as the anomaly time. For material shortage-related anomalies, the timestamp of the outbound failure record generated by the warehouse management system is used as the anomaly time, thereby ensuring that the anomaly time can accurately reflect the start time of the emergency on site.

[0035] It should be noted that the accurate extraction of anomaly times provides a reliable environmental state time anchor for the time interval between subsequent calculation operations and anomalies.

[0036] Step A2-3: Calculate the time difference based on the modification time and the anomaly time. The time difference represents the time interval between the parameter modification operation and the occurrence of the on-site anomaly. Understandably, the calculation of time difference is the quantitative basis for establishing the potential causal relationship between modification actions and abnormal events.

[0037] In this invention, the extracted modification time and abnormal time are placed on the same time axis for numerical comparison, and the absolute time span between the modification time and the abnormal time is calculated.

[0038] In some instances, suppose the timestamp is modified as Abnormal timestamps are Then the time difference The calculation formula is: This calculation process eliminates the influence of positive and negative signs caused by the order of time, and only retains the absolute length of the time interval.

[0039] It should be noted that the time difference The physical meaning lies in measuring the time delay of the dispatcher's response to on-site anomalies. If the time difference is... The smaller value indicates that the dispatcher made parameter modification operations within a very short time after the anomaly occurred, and the probability of a causal relationship between the two is extremely high. If time difference The large value indicates that the modification operation and the abnormal event were separated by a significant time interval, and there is no direct causal relationship between the two.

[0040] Time difference Transforming the proximity of time dimensions into quantifiable numerical indicators provides direct data criteria for determining whether an abnormal event constitutes a modification action within its environmental context. For example, in an automotive parts stamping workshop, the abnormal time of a die jamming anomaly on the stamping press... The time when the dispatcher modified the production scheduling parameters was 10:00:15 AM. The time difference is calculated as 10:00:45 AM. The 30-second time difference indicates that the modification action closely follows the abnormal event, suggesting a high degree of correlation.

[0041] Step A2-4: Abnormal events with a time difference less than a preset time window are identified as associated abnormal events. Associated abnormal events indicate on-site abnormalities that have a potential causal relationship with the parameter modification operation. In a preferred implementation, the preset time window is a time boundary threshold for measuring whether the modification action and the abnormal event have a causal relationship.

[0042] Specifically, in this embodiment, each calculated time difference value With preset time window Perform a one-by-one comparison, when the time difference satisfy When the time difference is determined, it is determined that the time difference has occurred. If an abnormal event and its corresponding modification action are sufficiently close in time, the abnormal event is marked as a related abnormal event.

[0043] Furthermore, the establishment of associated abnormal events means that on-site anomalies are regarded as the direct cause that triggers the dispatcher to modify actions, and a potential causal chain is constructed between the two.

[0044] Furthermore, preset time window An upper limit on the time frame for establishing a causal relationship has been set, and abnormal events exceeding this limit are considered unrelated to the modification action.

[0045] It should be pointed out that the preset time window The settings directly determine the stringency of the filtering for associated anomalies; too high a threshold will introduce irrelevant anomalies, while too low a threshold will miss genuine correlations. Specifically, the preset time window... The preset time window needs to be set according to the workshop production rhythm and the scheduler's response habits. The average response time of the scheduler to known anomalies is obtained by statistically analyzing the historical logs.

[0046] In one implementation, the aforementioned preset time window It can be obtained through the following specific calculation method, where:

[0047] In the formula, This indicates the preset time window, in seconds; This represents the average production cycle time in the workshop, in seconds. This represents the beat response coefficient, ranging from 0.5 to 2.0, and has no unit. This represents a communication delay compensation constant, measured in seconds, with a value ranging from 5 to 30. This represents the average reaction time of the scheduler in observing and making decisions based on the production rhythm, while This compensates for the fixed delays in the data acquisition and transmission process. The sum of the two ensures that the time window both includes normal response delays and excludes irrelevant accidental operations.

[0048] Step A2-5: Merge the status labels of associated abnormal events into the modification action sequence to form an aligned state sequence. The aligned state sequence is used to provide the environmental context features of each modification action. In this embodiment of the application, after obtaining the associated abnormal event, the status label of the associated abnormal event and the modification action sequence are fused together.

[0049] Accordingly, each modification action in the modification action sequence is traversed, and the associated abnormal event established by comparing the time difference with the modification action is found. The status label of the associated abnormal event is extracted, and the status label contains information on the abnormality type, abnormality level and abnormality duration.

[0050] Specifically, the extracted state tags are added as environmental context features to the data structure of the corresponding modification action, forming a comprehensive data record containing modification time, modification value, and environmental context features. All comprehensive data records are arranged in chronological order according to timestamps to form an aligned state sequence.

[0051] In some embodiments, if a modification action does not match any associated abnormal event, the environmental context feature of the modification action is set to null, indicating that the site was in normal working condition when the modification action occurred.

[0052] Furthermore, the aligned state sequence restores the originally isolated modification actions to the context of the original environment, making subsequent intent analysis not just theoretical but based on the real-world situation. The aligned state sequence, as the core input for intent quantification, directly supports the decomposition and determination of resistive and corrective intents.

[0053] Step A3: Based on the alignment state sequence and preset parameter constraint boundaries, calculate the intention ratio of the modified action sequence to obtain the intention quantification value, which includes the resistance intention value and the correction intention value. Understandably, after obtaining the aligned state sequence, simple correlation matching is still insufficient to accurately quantify the scheduler's modification intentions. Even in the presence of environmental anomalies, the scheduler's modifications may include elements of over-modification. Such over-modification reflects the scheduler's distrust of the intelligent system, while modifications in the absence of anomalies directly reflect subjective rejection.

[0054] Therefore, it is necessary to introduce preset parameter constraint boundaries as a benchmark to decompose the intent of each modification operation in the modification action sequence, accurately classify the modification behavior into resistance intent and correction intent, and present them in the form of quantitative values, thereby transforming the vague behavioral motivation into calculable data indicators, and providing precise punishment or exemption basis for subsequent evaluation and correction.

[0055] In some embodiments, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain an intent quantification value includes: Step A3-1: In response to the absence of associated abnormal events in the alignment state sequence, the corresponding modification action in the modification action sequence is taken as a resisted modification action. The resisted modification action represents a manual modification operation without on-site abnormality support. It should be noted that in the production scheduling stage of enterprise digital transformation, the scheduler's modification operations on the system interface are not always driven by unexpected situations on site. When the intelligent scheduling system is running normally and there are no equipment failures or material shortages on the workshop, if the scheduler still frequently modifies the scheduling parameters generated by the system, this unjustified intervention strongly suggests the scheduler's distrust and subjective resistance to the intelligent system's decision-making. Identifying and isolating such operations as resistance to modification is a key entry point for quantifying human-machine collaboration barriers, providing an accurate analytical object for subsequent calculation of the degree of subjective resistance, and avoiding confusing unjustified resistance with normal business adjustments.

[0056] Step A3-2: Obtain the parameter modification range and preset parameter allowable fluctuation range of the resisted modification action. The parameter modification range represents the absolute difference between the modified value and the original value. After identifying actions that resist modification, the frequency of these actions alone cannot accurately measure the intensity of the resistance. The magnitude of the dispatcher's parameter modifications directly reflects their determination to deviate from the intelligent system's recommendations. Minor adjustments may simply be fine-tuning habits, while significant alterations reveal a strong sense of resistance. Furthermore, different parameters have different inherent fluctuation attributes, necessitating the introduction of a preset allowable fluctuation range as a reference benchmark. By obtaining the magnitude of parameter modifications and the preset allowable fluctuation range, a calculation basis is provided for converting absolute modifications into relative resistance levels. This ensures that the quantification of resistance intentions considers both the absolute strength of the modification and the normal fluctuation characteristics of the parameters themselves, making the evaluation results more consistent with actual business logic.

[0057] In some embodiments, the acquisition of the parameter modification magnitude of the resisted modification action and the preset parameter allowable fluctuation range includes: Step A3-21: Obtain the original production scheduling parameter values ​​and the manually modified parameter values ​​corresponding to the resisted modification actions. The original production scheduling parameter values ​​represent the initial values ​​automatically generated by the intelligent production scheduling system. In some embodiments, extracting specific parameter values ​​associated with resisting modification actions from the alignment state sequence is the basis for calculating the modification magnitude.

[0058] For example, for a resisted modification action where the environmental context feature in the alignment state sequence is empty, trace the parameter field corresponding to the modification action in the system log, obtain the value that the intelligent scheduling system automatically calculated and issued by the algorithm before intervention, and record this value as the original scheduling parameter value.

[0059] Specifically, in this embodiment, the values ​​that the dispatcher manually inputs and confirms on the system interface are simultaneously obtained, and these values ​​are recorded as manually modified parameter values.

[0060] Furthermore, the original scheduling parameter values ​​represent the theoretical optimal solution derived by the intelligent algorithm based on global optimization, while the manually modified parameter values ​​represent the actual execution values ​​given by the scheduler based on personal experience or subjective judgment. The difference between the two in numerical space is the original material for assessing resistance intentions.

[0061] It is understandable that the extracted original production scheduling parameter values ​​and manually modified parameter values ​​must belong to the same parameter dimension and have the same unit; otherwise, unit conversion and alignment are required to ensure the correctness of the physical meaning of subsequent difference calculations.

[0062] In one implementation, the original scheduling parameter value and the manually modified parameter value corresponding to the resisted modification action can be obtained by parsing the operation record table of the system database. For example, in an automotive parts manufacturing workshop, the intelligent scheduling system automatically generates a stamping process with a planned processing quantity of 500 pieces, i.e., the original scheduling parameter value is 500. The scheduler manually changes it to 300 pieces, i.e., the manually modified parameter value is 300. These two values ​​will be extracted for subsequent magnitude calculations.

[0063] Step A3-22: Calculate the absolute difference based on the original production scheduling parameter values ​​and the manually modified parameter values. Use the absolute difference as the parameter modification range. The parameter modification range represents the absolute change of the initial value by the manual modification operation. This invention relates to the precise measurement of the magnitude of changes made by manual modification operations. It should be noted that the magnitude of parameter modification is a core indicator for assessing the strength of resistance intentions and must be obtained through rigorous numerical calculations.

[0064] It should be pointed out that if we only compare the magnitude of the original production scheduling parameter values ​​and the manually modified parameter values ​​without calculating the absolute difference, we will not be able to eliminate the impact of the modification direction. This is because whether the parameter is increased or decreased, as long as the degree of deviation from the system recommendation is the same, the degree of resistance it reflects should be equivalent.

[0065] In detail, calculate the absolute difference between the original production scheduling parameter value and the manually modified parameter value, remove the modification direction symbol, retain only the numerical value, and define the absolute difference as the parameter modification range.

[0066] In one implementation, the parameter modification range mentioned above can be obtained through the following specific calculation method, wherein:

[0067] In the formula, This indicates the magnitude of parameter modification, with the unit consistent with the original production scheduling parameter value. This indicates that the parameter value has been manually modified, and the unit is consistent with the original production scheduling parameter value. This represents the original production scheduling parameter value, and the unit is consistent with the original production scheduling parameter value unit; This indicates the absolute value operation. In other words, it represents the magnitude of parameter modification. The larger the value, the better the initial value given by the scheduler to the intelligent scheduling system. The more drastic the changes, the more obvious the subjective intention to alter the system's decision-making.

[0068] It should be noted that, due to the original production scheduling parameter values In extreme cases where the value might be zero, directly using the ratio for calculation could lead to division by zero errors. Therefore, the absolute difference is calculated first as an intermediate result to ensure the robustness of the calculation process. The parameter modification range in this invention... The absolute difference is obtained by calculation using the formula described above.

[0069] In this invention, the original production scheduling parameter values ​​automatically generated by the system are affected by potential sensor data fluctuations on-site. The original production scheduling parameters were already abnormally deviating from the normal range. If the scheduler were to change the parameters back to normal values, it would be misinterpreted as resistance to modification. Therefore, the original production scheduling parameters should be... When the value exceeds three standard deviations from the historical normal distribution range, if the parameter value is manually modified... If the value falls within the historical normal distribution range, then the parameter modification range will be adjusted. Set to zero; if the parameter value is manually modified. If it still exceeds the historical normal distribution range, then the normal calculation parameter modification range is... If the parameter value is manually modified If the value falls within the historical normal distribution range but the modification exceeds the preset parameter constraint boundary, the excess portion will be included in the resistance intention value.

[0070] Step A3-23: Obtain the normal operating condition parameter range in the production environment state sequence, and use the normal operating condition parameter range as the preset parameter allowable fluctuation range. The normal operating condition parameter range represents the normal fluctuation range of the parameter when there are no abnormal events. In a preferred implementation, the setting of the allowable fluctuation range of the preset parameters needs to be based on objective production operation data, rather than subjective assumptions.

[0071] It should be noted that different production scheduling parameters have reasonable fluctuation range in normal production. For example, processing time may fluctuate slightly due to the worker's skill level. Such fluctuation is normal and should not be counted as an intention to resist.

[0072] In a straightforward manner, all historical periods without abnormal event markers are selected from the production environment status sequence, the set of values ​​for the corresponding production scheduling parameters within these historical periods is extracted, and the statistical distribution characteristics of this set of values ​​are calculated.

[0073] Specifically, the range between the upper and lower quartiles of the set of values, or the range formed by the mean of the set of values ​​plus or minus twice the standard deviation, is defined as the normal operating condition parameter range.

[0074] The normal operating condition parameter range represents the normal fluctuation range of production scheduling parameters under the premise that the workshop equipment is operating normally and the material supply is sufficient. The normal operating condition parameter range is directly mapped to the preset parameter allowable fluctuation range, so that the definition of the fluctuation boundary is rooted in the real production data. In this application, the preset parameter allowable fluctuation range is obtained by statistically analyzing historical data of normal operating conditions.

[0075] In some embodiments, if the production scheduling parameters have explicit process specification tolerances, the process specification tolerance range is used as the preset parameter allowable fluctuation range to ensure that the range setting conforms to the actual engineering situation.

[0076] Step A3-24: Compare the parameter modification range with the preset parameter allowable fluctuation range, and determine the resistance intention value based on the comparison result. The comparison result is used to reflect the degree to which the manual modification operation deviates from the normal fluctuation range. In detail, comparing the parameter modification range with the preset parameter fluctuation range is a key step in converting the absolute modification amount into a relative resistance index.

[0077] If the parameter modification falls entirely within the preset parameter fluctuation range, it indicates that the dispatcher's modification is still within a reasonable fine-tuning range and the resistance is weak. If the parameter modification exceeds the preset allowable fluctuation range, the excess represents a conflicting modification that breaks the normal pattern.

[0078] In some instances, the width of the allowed fluctuation range is defined as a preset parameter. The comparison result represents the parameter modification magnitude. Relative to interval width The overflow ratio. According to an embodiment of the present invention, when When the comparison result is no overflow, the resistance intention value is zero or a very small baseline value; when At that time, the comparison result was a positive overflow, and the overflow amount was... The overflow amount and The ratio is used as the basis for calculating the resistance intention value.

[0079] It should be noted that the comparison results, by quantifying the degree to which the fluctuations exceed the normal range, directly map the intensity of the dispatcher's subjective rejection of the intelligent system, transforming the vague behavioral pattern into a precise numerical indicator. In this embodiment, the comparison results are obtained by calculating the difference between the parameter modification range and the preset parameter allowable fluctuation range.

[0080] Step A3-3: Calculate the resistance intention value based on the ratio of the parameter modification range of the resistance modification action to the preset parameter allowable fluctuation range. The resistance intention value represents the degree to which the dispatcher subjectively rejects the intelligent system. In one alternative implementation, in order to normalize the severity of the resistance to modification, the ratio of the parameter modification magnitude to the preset parameter allowable fluctuation range is used as the core calculation logic.

[0081] In some embodiments, the parameter modification magnitude is calculated. The width of the range of fluctuation allowed by the preset parameters The ratio of [value] to [value] indicates that the modification has exceeded the normal fluctuation boundary. The excess portion of the ratio is extracted as the resistance intention value. The foundation.

[0082] Understandably, the larger this ratio, the more outrageous the changes the dispatcher makes to the parameters given by the intelligent system, and the stronger their subjective resistance.

[0083] Resistance Intent Value It is positively correlated with this ratio. The ratio calculation eliminates the influence of differences in the dimensions and magnitudes of different parameters, so that the resistance intention values ​​of different parameters can be compared and accumulated on the same scale.

[0084] Preferred resistance intention value The calculation needs to consider the nonlinear amplification effect. When the ratio exceeds a certain threshold, an exponential function is used to amplify the excess portion to highlight the penalty weight of extreme resistance behavior. In this invention, the resistance intention value... The value is obtained by calculating the ratio of the parameter modification range to the preset parameter allowable fluctuation range and then substituting it into a nonlinear function.

[0085] Step A3-4: Use the resistance intention value as the intention quantification value. The intention quantification value is used to quantify and evaluate the subjective resistance component in the manual modification operation. It should be noted that for resisted modification actions without associated abnormal events, the underlying motivation is solely attributed to the scheduler's subjective resistance. Therefore, the calculated resistance intention value directly represents the overall intention quantification value of the modification action.

[0086] In this scenario without environmental anomalies, the intent quantification value does not contain any corrective components, and the resistance intent value is the entirety of the intent quantification value.

[0087] The resistance intention value is assigned to the intention quantification value, so that the intention quantification value can accurately represent the degree of purely subjective resistance in this manual modification operation.

[0088] Specifically, the intent quantification value is used as a numerical label and bound to the corresponding modification action record, providing precise input parameters for the subsequent overall evaluation model's deduction mechanism.

[0089] In some optional implementations, the resistance intention value is stored as an intention quantification value in the evaluation database and associated with the dispatcher's identity information for subsequent generation of individual-specific collaboration analysis reports. In this embodiment, the intention quantification value is obtained by directly assigning the resistance intention value.

[0090] In some embodiments, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain an intent quantification value includes: Step A3-5: In response to the existence of associated abnormal events in the alignment state sequence, the corresponding modification actions in the modification action sequence are taken as potential correction actions. Potential correction actions indicate manual modification operations supported by on-site abnormalities. If there are associated abnormal events in the alignment state sequence, it means that the scheduler's modification action occurred within the time window of the on-site abnormality, and the modification action has a reasonable business motivation.

[0091] Furthermore, these types of modifications are marked as potential corrective actions. Potential corrective actions mean that the modification is highly likely to be a corrective action to address equipment failure or material shortage, rather than simply a subjective resistance.

[0092] When there are associated abnormal events in the alignment state sequence, the corresponding modification actions in the modification action sequence are regarded as potential corrective actions. This classification logic is based on the principle of temporal proximity of causal relationships, ensuring that modification actions supported by environmental context are not misjudged as resistant behaviors.

[0093] In some embodiments, the potential corrective action only provides a reasonable assumption for the modification action, and its specific intent still needs to be further broken down according to the reasonableness of the modification range. In other words, the potential corrective action is a prerequisite for subsequent determination of the corrective intent and the mixed intent. In this invention, the potential corrective action is obtained by matching associated abnormal events.

[0094] Steps A3-6: Extract the anomaly level of the associated abnormal events and the parameter modification range of potential corrective actions. The anomaly level indicates the severity of the on-site anomaly. Specifically, the anomaly level of the associated abnormal event is the key variable that determines the reasonable threshold for modification. The higher the anomaly level, the more urgent the situation on site, and the more reasonable it is for the dispatcher to make significant adjustments.

[0095] It should be noted that the level of anomaly is usually determined by the type and duration of the abnormal event. For example, the level of a complete equipment shutdown is higher than that of equipment running at reduced speed, and the level of a shortage of critical materials is higher than that of a shortage of auxiliary materials.

[0096] In one embodiment, the level labels carried by the abnormal events are extracted from the production environment state sequence, and the level labels are mapped to numerical abnormality level coefficients.

[0097] The anomaly level coefficient quantifies the severity of on-site anomalies, providing a basis for subsequent dynamic adjustment of preset parameter constraint boundaries.

[0098] Simultaneously extract the parameter modification range of potential corrective actions. Parameter modification range This reflects the actual adjustments made by the dispatcher, linking the anomaly level to the magnitude of parameter modifications. The paired extraction forms the core data pairs for subsequent evaluation of whether the modification behavior is appropriate. In this invention, the anomaly level is obtained by parsing the anomaly event label, and the parameter modification range is... It is obtained by calculating the absolute difference between the manually modified parameter values ​​and the original production scheduling parameter values.

[0099] Step A3-7: Determine the preset parameter constraint boundary based on the anomaly level. The preset parameter constraint boundary represents the maximum reasonable range of manual modification allowed under the corresponding anomaly level. In some optional implementations, the determination of the preset parameter constraint boundaries must be dynamically correlated with the anomaly level to reflect the differences in fault tolerance under different urgency levels.

[0100] When the anomaly level is low, the dispatcher only needs to make minor adjustments to cope with it, and the maximum reasonable range of manual modification is relatively small; when the anomaly level is high, the conventional parameters can no longer meet the production needs, and the maximum reasonable range of manual modification should be increased accordingly.

[0101] As an feasible approach, a mapping table between anomaly levels and maximum reasonable ranges can be established. The higher the anomaly level, the wider the corresponding preset parameter constraint boundary.

[0102] In this invention, the preset parameter constraint boundary is obtained by querying an anomaly level mapping table, which is constructed based on historical expert experience or simulation data.

[0103] It should be noted that the preset parameter constraint boundary sets a dynamic red line for judging the rationality of potential corrective actions. Only modifications within the red line are considered pure corrections, while those exceeding the red line are suspected of excessive intervention. This dynamic boundary setting ensures the adaptability and fairness of the evaluation criteria under different abnormal scenarios.

[0104] Step A3-8: In response to the parameter modification range being less than or equal to the preset parameter constraint boundary, the potential correction action is taken as a reasonable correction action, the correction intention value is calculated, and the correction intention value is taken as the intention quantification value. The correction intention value represents the reasonable correction component in the manual modification operation. When the parameter modification is less than or equal to the preset parameter constraint boundary, the dispatcher's modification operation is determined to be within the reasonable range for handling anomalies, with no trace of excessive modification. If the parameter modification is less than or equal to the preset parameter constraint boundary, the potential corrective action is marked as a reasonable corrective action, and the motivation for the modification action is confirmed as purely corrective.

[0105] Furthermore, the intention value for correction is calculated. The magnitude of the intention value for correction is directly proportional to the magnitude of parameter modification. The larger the magnitude of modification, the more thorough the correction action is and the greater the contribution to correcting the system's deviation from reality.

[0106] It should be noted that in this case, the intention quantification value only includes the corrective intention value and contains no resistance component; the corrective intention value is directly assigned to the intention quantification value. That is, the identification of reasonable corrective actions not only exempts the scheduler from suspicion of resistance, but also proves that the intelligent scheduling system's suggestions are indeed out of touch with reality and require manual intervention for correction. In this application, the corrective intention value is obtained by calculating the ratio of the parameter modification magnitude to the preset parameter constraint boundary.

[0107] In some embodiments, the step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain an intent quantification value further includes: Step A3-9: In response to the parameter modification range being greater than the preset parameter constraint boundary, the potential correction action is treated as a mixed modification action. The mixed modification action represents a modification operation whose modification range exceeds the reasonable correction range. To make it easier to understand, when the parameter modification exceeds the preset parameter constraint boundary, it means that although the scheduler made the modification under the guise of dealing with the anomaly, the extent of the modification far exceeds what is needed to resolve the current anomaly. This excessive modification partially exposes the scheduler's subjective intention to forcibly change the system's decision.

[0108] It should be noted that defining such potential corrective actions as mixed modification actions means that the action simultaneously includes reasonable corrective components to deal with anomalies and resistance components that exceed necessary limits.

[0109] Specifically, the identification of mixed modification actions reveals a common phenomenon of "piggybacking" modifications in human-computer interaction, namely, when the dispatcher corrects system errors, they may secretly adjust parameters to their preferred values.

[0110] In one alternative implementation, the portion of parameter modification that exceeds the preset parameter constraint boundary is separated as the object of analysis for resistance intent, while the portion within the preset parameter constraint boundary is retained as the object of analysis for correction intent. In other words, the determination of mixed modification actions provides a logical premise for the subsequent decomposition of dual intents. For example, when a minor fault in the stamping machine requires a reduction of 100 pieces in production, but the scheduler reduces production by 500 pieces, this excess modification of 400 pieces belongs to the resistance component in the mixed modification action.

[0111] Step A3-10: Obtain the modification amount corresponding to the preset parameter constraint boundary as the reasonable correction amount. The reasonable correction amount represents the minimum modification value required to deal with on-site anomalies. To accurately decompose the dual intentions in mixed modification actions, it is first necessary to define a reasonable proportion for responding to anomalies. This invention provides a method for extracting a reasonable correction amount to solve the problem of confusion between correction and resistance components in mixed modification actions.

[0112] Furthermore, the modification amount corresponding to the preset parameter constraint boundary is directly defined as the reasonable correction amount. The reasonable correction amount represents the minimum parameter adjustment value that the scheduler must make to maintain production under the current anomaly level.

[0113] Furthermore, the reasonable correction amount is the upper limit of reasonable modifications recognized by the system. Regardless of the actual amount modified by the scheduler, only the portion within the reasonable correction amount is considered a correction to the shortcomings of the intelligent system. In this embodiment, the reasonable correction amount is equal to the value of the preset parameter constraint boundary. This value is obtained through anomaly level mapping, reflecting the rigid demand for parameter adjustment due to objective production anomalies. In this invention, the reasonable correction amount is obtained by reading the value of the preset parameter constraint boundary.

[0114] Step A3-11: Obtain the amount of modification that exceeds the preset parameter constraint boundary as the excessive resistance amount. The excessive resistance amount represents the additional modification value that exceeds the reasonable correction range. Furthermore, based on determining a reasonable amount of correction, the redundant portion that exceeds the reasonable range in the actual modification is calculated.

[0115] Specifically, the difference between the parameter modification range and the preset parameter constraint boundary is defined as the excessive resistance amount. The excessive resistance amount represents the amount of modification that the dispatcher adds under the guise of an anomaly. This part of the modification is not necessary to deal with the on-site anomaly, but is an intervention that the dispatcher subjectively imposes on the system.

[0116] In one alternative implementation, the calculation of the excessive resistance amount must be based on the difference in absolute values ​​to ensure that no excess modification in any direction is missed.

[0117] The excessive resistance measure isolates the subjective resistance hidden in the mixed modification action, making the quantification of resistance intention more accurate and avoiding the evaluation bias of treating all modifications as corrections. The excessive resistance measure is obtained by calculating the difference between the parameter modification magnitude and the preset parameter constraint boundary.

[0118] Step A3-12: Based on the ratio of reasonable correction amount to excessive resistance amount, calculate the correction intention value and resistance intention value of the mixed modification action, and use the correction intention value and resistance intention value together as the intention quantification value. The intention quantification value is used to decompose the dual intention components in the mixed modification action. This application provides a method for decomposing mixed intentions, the method comprising allocating intentions to mixed modification actions according to the ratio of reasonable correction amount to excessive resistance amount.

[0119] Furthermore, the overall amount of modification in a mixed modification action is composed of two parts: a reasonable correction amount and an excessive resistance amount. The corresponding intention value needs to be calculated based on the proportion of each part.

[0120] Furthermore, the proportion of reasonable corrections in the total number of revisions determines the magnitude of the correction intention value, while the proportion of excessive resistance in the total number of revisions determines the magnitude of the resistance intention value.

[0121] Specifically, in one implementation, the aforementioned corrective intent value and resistance intent value can be obtained through the following specific calculation method, wherein:

[0122]

[0123] In the formula, Indicates the intended correction value; no unit is specified. Indicates the resistance intention value; no unit. This indicates the reasonable correction amount, and the unit is consistent with the unit of the parameter modification range; This indicates the magnitude of parameter modification, and the unit is the same as the unit of parameter modification magnitude. This indicates the magnitude of parameter modification, and the unit is the same as the unit of parameter modification magnitude. This indicates the weighting of the reasonable correction amount within the total amount of modifications. This represents the weight ratio of excessive resistance in the total amount of modification. The sum of the two is 1, ensuring the conservation of the intended dismantling.

[0124] In other words, through the above calculations, the mixed modification action is broken down into correction intention values. and resistance intention value Two independent components will be used to determine the intended correction value. and resistance intention value Together, they serve as the intent quantification value, which fully expresses the dual intent components in the mixed modification action, neither wrongly acknowledging reasonable corrections nor overlooking hidden resistance. In this application, the correction intent value and the resistance intent value are obtained by multiplying the aforementioned weight ratio with the parameter modification magnitude.

[0125] Step A4: Based on the intent quantification value, the preset initial score of digital transformation is corrected to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process. It should be noted that the calculation of intent quantification only completes the motivational analysis of underlying behaviors. If it is not used in the overall evaluation system, its impact on the effectiveness of transformation cannot be reflected. Existing digital transformation evaluations often give a static initial score, ignoring the dynamic friction in the human-computer interaction process. By using intent quantification, which includes resistance intent and corrective intent, as a correction factor, the preset initial score of digital transformation can be deducted or adjusted in a targeted manner. This can directly reflect the subjective resistance of the scheduler and the objective defects of the system in the final score, thereby breaking the evaluation bubble of false prosperity and obtaining a target transformation evaluation value that can truly reflect the collaborative level of the enterprise's digital transformation process, providing a basis for management to make decisions.

[0126] In some embodiments, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value includes: Step A4-1: In response to the presence of resisting intent values ​​in the quantified intent values, the overall resistance index is obtained by statistically analyzing all resisting intent values. The overall resistance index represents the cumulative degree of subjective rejection of the intelligent system throughout the entire evaluation period. In this embodiment, for all modification actions within the evaluation period, the resistance intent value contained in the intent quantification value is extracted. These scattered resistance intention values The overall resistance index is calculated by summing the results. In other words, the overall resistance index is the cumulative sum of the dispatchers' subjective resistance to the intelligent system throughout the entire evaluation period. A single, small amount of resistance may just be a difference in individual habits, but a rise in the overall resistance index clearly indicates the existence of systemic resistance at the workshop level.

[0127] Accordingly, the overall resistance index needs to cover all types of resistant modification actions and the resistant components in mixed modification actions to ensure that no rejection signals are missed.

[0128] It should be noted that the overall resistance index compresses discrete resistance behaviors over time into a macro-level indicator, providing a total basis for subsequent adjustments to the human-machine collaboration score.

[0129] Specifically, for example, during a one-month evaluation period, there were 10 instances of non-abnormal resistance modification actions with resistance intention values ​​of 0.2, 0.3, etc., and 5 instances of mixed modification actions with resistance intention values ​​of 0.1, 0.4, etc. The sum of these 15 resistance intention values ​​yields an overall resistance index of 3.5. This value directly reflects the cumulative degree of resistance to the system by the scheduler in that month. In this invention, the overall resistance index is obtained by accumulating all resistance intention values.

[0130] Step A4-2: Obtain the initial score of human-machine collaboration in the preset initial score of digital transformation. The initial score of human-machine collaboration represents the collaboration baseline score before considering resistance to modification operations. In one implementation, the preset initial score for digital transformation is an initial evaluation score calculated by the evaluation system based on static indicators such as system function coverage and data connectivity.

[0131] It is important to understand that the initial score for human-machine collaboration is a sub-dimension within the pre-set initial score for digital transformation, specifically used to evaluate the initial level of cooperation between humans and systems.

[0132] Furthermore, the initial score for human-machine collaboration represents the baseline score for collaboration before the introduction of a penalty mechanism for resistant behavior. This baseline score is usually given based on superficial indicators such as system uptime and function utilization, and is often overly optimistic.

[0133] In this embodiment, the initial score of human-machine collaboration is read from the database of preset initial scores for digital transformation and used as the base for subsequent deduction and correction.

[0134] It should be noted that the initial score for human-machine collaboration sets an ideal upper limit for human-machine collaboration. The actual level of collaboration must be reduced by deducting the losses caused by subjective resistance. The initial score for human-machine collaboration in this application is obtained by reading the preset initial score database for digital transformation.

[0135] Step A4-3: Calculate the collaboration deduction amount based on the product of the overall resistance index and the preset collaboration deduction coefficient. The collaboration deduction amount represents the negative impact of the resistance operation on the collaboration degree. This invention provides a method for calculating collaborative deduction to address the problem that static assessments cannot reflect dynamic resistance losses.

[0136] Preferably, mapping the overall resistance index to specific score deduction values ​​requires introducing a preset collaborative deduction coefficient as a conversion bridge.

[0137] It should be noted that the preset collaborative deduction coefficient determines the decrease in the initial score of human-machine collaboration corresponding to each unit of resistance intention value. The setting of the coefficient needs to be adjusted according to the company's tolerance for resistant behavior.

[0138] In a preferred embodiment, the aforementioned collaborative deduction amount can be obtained through the following specific calculation method, wherein:

[0139] In the formula, This indicates the amount of deductions, expressed in cents. This represents the overall resistance index, without units. This represents the preset collaborative deduction coefficient, expressed in points per unit of resistance index. This invention transforms the unitless cumulative resistance into a deduction value with scoring significance, intuitively quantifying the negative impact of resistance on cooperation. This invention relates to the precise calculation of cooperation deductions, ensuring that the penalty is strictly proportional to the degree of resistance, avoiding assessment failure due to excessively lenient deductions or assessment distortion due to excessively severe deductions. The cooperation deduction amount in this invention... By calculating the overall resistance index With preset collaborative deduction coefficient The product is obtained by presetting the collaborative deduction coefficient. Obtained through expert calibration or historical data regression analysis.

[0140] Step A4-4: Subtract the collaboration deduction from the initial human-machine collaboration score to obtain the target score for human-machine collaboration. The target score for human-machine collaboration is used to reflect the actual level of collaboration between humans and the system. This invention provides a method for calculating the target score of human-machine collaboration, in order to solve the problem that the initial score is too high and cannot reflect the true collaboration status.

[0141] Furthermore, using the initial score for human-machine collaboration as the starting point for evaluation, and eliminating the negative losses represented by the collaboration deduction, the remaining score becomes the target score for human-machine collaboration. It's important to understand that the target score for human-machine collaboration is no longer an optimistic estimate based on the system's online rate, but rather a true score verified through actual operational behavior.

[0142] In some instances, when the deduction for collaboration exceeds the initial score for human-machine collaboration, the target score for human-machine collaboration is set to zero to avoid negative scores, which is contrary to conventional scoring logic.

[0143] It should be noted that the target score for human-machine collaboration truly reflects the degree of cooperation between the scheduler and the intelligent production scheduling system. The larger the deduction, the lower the actual level of collaboration. In this invention, the target score for human-machine collaboration is obtained by calculating the difference between the initial score for human-machine collaboration and the collaboration deduction.

[0144] In some embodiments, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value further includes: Steps A4-5: In response to the presence of corrective intent values ​​in the quantitative intent values, all corrective intent values ​​are statistically analyzed to obtain the overall corrective index. The overall corrective index represents the cumulative degree to which the intelligent system's suggestions deviate from reality throughout the entire evaluation period. In this embodiment of the application, after obtaining the intent quantification value, the correction intent value is extracted. Accumulated statistics are performed. Accordingly, all reasonable corrective actions and mixed modification actions within the evaluation period are traversed, and the corresponding corrective intent values ​​are extracted. All correction intention values Summing them up yields the overall correction index.

[0145] Specifically, the overall correction index represents the cumulative degree of unreasonable suggestions given by the intelligent scheduling system throughout the entire evaluation cycle. Each manual correction implies an unrealistic error in the system algorithm. In this embodiment, the overall correction index aggregates scattered system errors into a macro-level quality evaluation indicator, providing data support for revising the system maturity score.

[0146] A higher overall correction index indicates that the intelligent scheduling system is less adapted to the actual workshop environment and requires manual intervention to maintain production. In this application, the overall correction index is calculated by summing all correction intention values. Obtain.

[0147] Step A4-6: Obtain the initial system maturity score from the preset initial score for digital transformation. The initial system maturity score represents the baseline score of system maturity before considering corrective modification operations. This invention relates to the extraction of the initial system maturity score. In other words, the initial system maturity score is another core sub-dimension in the preset initial score for digital transformation, which is specifically used to evaluate the maturity and applicability of the intelligent scheduling system algorithm model.

[0148] For ease of understanding, the initial system maturity score is a baseline score derived from the evaluation of the system's own attributes, such as algorithm advancement and data integrity, without taking into account situations where the system needs manual correction due to deviation from actual field conditions during operation.

[0149] In some embodiments, an initial system maturity score is extracted from a database of preset initial scores for digital transformation and used as the initial benchmark for subsequent maturity deduction and correction.

[0150] Specifically, the initial system maturity score reflects the theoretical design level of the system, while the corrective actions in actual operation will expose the gap between it and the design expectations. In this invention, the initial system maturity score is obtained by reading a preset digital transformation initial score database.

[0151] Step A4-7: Calculate the maturity deduction amount based on the product of the overall correction index and the preset maturity deduction coefficient. The maturity deduction amount represents the negative impact of unreasonable system suggestions on maturity. In this embodiment, converting the overall correction index into a deduction value for the initial system maturity score requires introducing a preset maturity deduction coefficient as a conversion factor.

[0152] Correspondingly, the preset maturity deduction coefficient determines the decrease in the initial system maturity score corresponding to each unit of corrective intent value, reflecting the severity of the assessment system's punishment for systems that deviate from reality.

[0153] In one implementation, the preset maturity deduction coefficient can be obtained by statistically analyzing the relationship between historical correction frequency and system version iteration.

[0154] Specifically, the overall correction index is multiplied by a preset maturity deduction coefficient, and the product is the maturity deduction amount. Furthermore, the maturity deduction amount quantifies the number and severity of the system's deviations from reality into specific score penalties. The more frequent the system errors and the more severe the deviations, the larger the maturity deduction amount, and the more severely it erodes the initial maturity score of the system. In this invention, the maturity deduction amount is obtained by calculating the product of the overall correction index and the preset maturity deduction coefficient.

[0155] In one implementation, the calculation of the maturity deduction amount can be achieved by a preset penalty function. For example, linear deduction is used when the overall correction index is at a low level, and exponential deduction is used when the overall correction index exceeds the warning threshold, so as to reflect a zero-tolerance attitude towards frequent system errors. In another implementation, the maturity deduction can also be achieved by looking up a table, dividing the overall correction index into multiple intervals, each interval corresponding to a fixed deduction value. For example, 2 points are deducted when the overall correction index is between 0 and 5, 5 points are deducted when it is between 5 and 10, and so on.

[0156] Step A4-8: Subtract the maturity deduction from the initial system maturity score to obtain the system maturity target score. The system maturity target score is used to reflect the actual applicability level of the intelligent scheduling system. In one optional implementation, the initial system maturity score is used as the evaluation score under ideal conditions. The maturity deduction caused by the system deviating from reality is removed, and the remaining value is the target system maturity score. It can be understood that the target system maturity score is no longer simply a paper score based on the algorithm model, but rather an evaluation of the actual applicability level after on-site practical testing and deducting the cost of manual correction.

[0157] In other words, the larger the maturity deduction, the more defects the system exposes in actual application, and the lower the system maturity target score. For example, if the initial system maturity score is 85 points and the maturity deduction is 20 points, then the system maturity target score is 65 points. This 65 points truly reflects the actual applicability of the intelligent scheduling system in the current workshop environment. According to an embodiment of the present invention, when the calculated system maturity target score is lower than the passing mark, a system refactoring warning signal is triggered, prompting management to re-optimize the intelligent scheduling algorithm. In this application, the system maturity target score is obtained by calculating the difference between the initial system maturity score and the maturity deduction.

[0158] In some embodiments, the step of correcting the preset initial score for digital transformation based on the intent quantification value to obtain the target transformation evaluation value further includes: Step A4-9: Obtain the first preset evaluation coefficient corresponding to the target score of human-machine collaboration and the second preset evaluation coefficient corresponding to the target score of system maturity. The first preset evaluation coefficient and the second preset evaluation coefficient represent the importance of different evaluation dimensions in the overall evaluation. Furthermore, the target scores for human-machine collaboration and system maturity reflect human acceptance and system reliability during the digital transformation process, respectively, and they hold different strategic positions in the overall assessment.

[0159] Specifically, during the preparation phase, based on the enterprise's strategic direction for digital transformation, the weight of the human-machine collaboration dimension is determined as the first preset evaluation coefficient, and the weight of the system maturity dimension is determined as the second preset evaluation coefficient. During the execution phase, the first and second preset evaluation coefficients are read from the configuration center to ensure that the sum of the two coefficients is 1, thereby guaranteeing the normalization of weight allocation.

[0160] During the verification phase, check whether the first and second preset evaluation coefficients match the preset strategy. If the company's current focus is on promoting system application, the first preset evaluation coefficient should be greater than the second preset evaluation coefficient; if the focus is on improving the system's intelligence level, the second preset evaluation coefficient should be greater than the first preset evaluation coefficient.

[0161] It should be noted that the first and second preset evaluation coefficients are not arbitrarily set, but are deeply aligned with the core needs of the enterprise at the current stage. In this invention, the first and second preset evaluation coefficients are obtained through input via the management strategic configuration interface.

[0162] Step A4-10: The target score of human-machine collaboration is weighted according to the first preset evaluation coefficient to obtain the collaboration evaluation component. The collaboration evaluation component represents the weighted value of the actual collaboration level between human and system. It should be pointed out that directly adding the target score for human-machine collaboration with the target score for system maturity will ignore the differences in magnitude and importance between the dimensions. By weighting, the absolute score can be transformed into a relative contribution.

[0163] In a preferred embodiment, the target score for human-machine collaboration is multiplied by a first preset evaluation coefficient, and the product is the collaboration evaluation component. This invention relates to dimensional compression of evaluation components; that is, the collaboration evaluation component maps the absolute value of the target score for human-machine collaboration to a weighted contribution value within the overall evaluation framework, eliminating the dimensional coupling of the original scores. To achieve accurate aggregation of the overall evaluation value, the collaboration evaluation component only represents the portion of the final score belonging to the human-machine collaboration dimension. For example, if the target score for human-machine collaboration is 80 points and the first preset evaluation coefficient is 0.6, then the collaboration evaluation component is 48 points. This 48 points is the weighted value of the actual level of human-system collaboration. In this invention, the collaboration evaluation component is obtained by calculating the product of the target score for human-machine collaboration and the first preset evaluation coefficient.

[0164] Step A4-11: The system maturity target score is weighted according to the second preset evaluation coefficient to obtain the maturity evaluation component. The maturity evaluation component represents the actual applicability level of the weighted intelligent scheduling system. The second preset evaluation coefficient represents the weight ratio of system maturity in the overall evaluation. Multiplying the system maturity target score by the second preset evaluation coefficient will scale the original value of the system maturity target score to the standard contribution value according to the weight ratio.

[0165] Specifically, the calculation process of the maturity assessment score and the weighting process of the human-machine collaboration target score are independent of each other, but logically symmetrical, ensuring the fairness of the assessment system.

[0166] Furthermore, the maturity assessment component transforms the absolute evaluation of the system maturity target score into a relative contribution, stripping away dimensional information that is irrelevant to the overall assessment.

[0167] In some embodiments, if the second preset evaluation coefficient is set to 0, the maturity evaluation component is 0, indicating that the system maturity dimension is not considered under the current evaluation strategy. This flexible mapping rule supports the dynamic adjustment of the evaluation model. In this invention, the maturity evaluation component is obtained by calculating the product of the system maturity target score and the second preset evaluation coefficient.

[0168] In one implementation, the maturity assessment component can be implemented by directly calling the weight multiplication function. For example, the system maturity target score is passed into the function as a variable, and the function reads the second preset assessment coefficient and returns the product result. In another implementation, maturity assessment components can also be implemented through lookup table interpolation. A mapping table between system maturity target scores and maturity assessment components is pre-established. The corresponding maturity assessment component is found based on the input system maturity target score. If an exact match cannot be found, linear interpolation is used for calculation.

[0169] Step A4-12: Sum the collaborative evaluation component and the maturity evaluation component to obtain the target transformation evaluation value. The target transformation evaluation value is used to comprehensively reflect the true effectiveness of the enterprise's digital transformation.

[0170] Furthermore, the collaborative evaluation component and the maturity evaluation component are placed in the same evaluation space. These two components represent the true levels of human factors and systemic factors in the transformation process after weighted adjustment.

[0171] Furthermore, the collaborative evaluation component and the maturity evaluation component are numerically added together, and the sum is the target transformation evaluation value.

[0172] It is important to understand that the target transformation assessment value integrates the actual friction loss of human-machine collaboration and the objective disconnection frequency of the system algorithm, completely abandoning the superficial assessment logic that only looks at the system online rate.

[0173] It should be noted that the target transformation assessment value not only penalizes the scheduler's subjective resistance but also deducts for design flaws in the intelligent system. The final value can accurately reveal the true level of collaboration and actual implementation effectiveness of enterprise digital transformation in the production scheduling process, providing a reliable quantitative basis for management to optimize human-machine relationships and improve system algorithms. In this invention, the target transformation assessment value is obtained by calculating the sum of the collaboration assessment component and the maturity assessment component.

[0174] Figure 2This is the core intent quantification and scoring correction logic in the enterprise digital transformation assessment method based on business process mining. First, the system simultaneously acquires the sequence of modification actions and the production environment state sequence. After time window alignment, it proceeds to determine the existence of associated anomalies. If no associated anomalies exist, the modification action is judged as a resisted modification action, and a resisted intent value is calculated. If associated anomalies exist, the system proceeds to parameter modification magnitude extraction and limit judgment. Actions within the limits are judged as reasonable corrective actions, and a corrective intent value is calculated; actions exceeding the limits are judged as mixed modification actions, requiring further decomposition of the dual intent components. After all intent quantification values ​​are aggregated and temporarily stored, the overall resistivity index and the overall corrective index are calculated separately. Finally, a weighted sum is used to obtain the target transformation assessment value reflecting the true level of collaboration. This logic strictly distinguishes between subjective resistivity and objective corrective action, ensuring the authenticity of the assessment results.

[0175] Figure 3 This section discusses the mapping of intent and the relationship between the assessment ecosystem and the evaluation process in enterprise digital transformation. When a scheduler modifies scheduling parameters suggested by the intelligent scheduling system, the intervention of production site anomalies differentiates into two forms: modifications without anomaly support and modifications with anomaly support. Modifications without anomaly support directly map into the resistance intent domain, revealing human resistance to the system; modifications with anomaly support, constrained by preset parameter boundaries, map into the corrective intent domain, revealing the system's unrealistic shortcomings. These two types of intent domains respectively erode human-machine collaboration and system maturity measurements, ultimately converging in the comprehensive transformation assessment index, objectively reflecting the dual forces of resistance and corrective action within the human-machine interaction ecosystem.

[0176] Figure 4 This is a data flow network for quantifying intent during enterprise digital transformation assessment. System log persistence and environment state persistence serve as the original data sources, parsing out modification action timestamp streams and abnormal event timestamp streams, respectively. These streams converge into the time window alignment calculation node. The generated associated abnormal state queue and preset parameter constraint boundaries flow together into the intent proportion calculation node. The calculation results are split into resistive intent quantification data and corrective intent quantification data, converging into an overall resistive index stream and an overall corrective index stream, respectively. These two index streams drive collaborative deduction calculation and mature deduction calculation. Simultaneously, the assessment benchmark library provides initial scores, and finally, the target transformation assessment value is output through deduction calculation, achieving a closed-loop data chain from raw logs to assessment scores.

[0177] Example 2, this example also provides an enterprise digital transformation assessment system based on business process mining, including: an acquisition module, used to acquire the modification action sequence and production environment status sequence in the system log in response to the triggering of the production scheduling parameter modification event, the modification action sequence including the timestamp and modification value of the manually modified parameters; The alignment module is used to perform time window alignment processing on the sequence of modification actions and the sequence of production environment states to obtain an aligned state sequence, which contains the correspondence between modification actions and abnormal events in the production environment. The calculation module is used to calculate the intention ratio of the modified action sequence based on the alignment state sequence and preset parameter constraint boundaries, and obtain the intention quantification value, which includes the resistance intention value and the correction intention value. The correction module is used to correct the preset initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process.

[0178] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0179] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0180] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A business process mining based enterprise digital transformation assessment method, characterized by, include: In response to the triggering of the production scheduling parameter modification event, obtain the modification action sequence and production environment status sequence from the system log. The modification action sequence includes the timestamp and modified value of the manually modified parameters. The modification action sequence and the production environment state sequence are aligned by time window to obtain an aligned state sequence, which contains the correspondence between modification actions and abnormal events in the production environment. Based on the alignment state sequence and preset parameter constraint boundaries, the intention ratio of the modification action sequence is calculated to obtain the intention quantification value, which includes the resistance intention value and the correction intention value. The initial score for digital transformation is corrected based on the intent quantification value to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process.

2. The business process mining based enterprise digital transformation assessment method as claimed in claim 1 wherein: The step of aligning the modification action sequence and the production environment state sequence using a time window to obtain an aligned state sequence includes: Obtain the timestamp of the modified action in the modified action sequence as the modification time, which indicates the specific moment when the scheduler performs the parameter modification operation. The timestamps of abnormal events in the production environment state sequence are obtained as abnormal times, which indicate the specific moments when equipment failures or material shortages occur on the workshop floor. The time difference is calculated based on the modification time and the anomaly time. The time difference represents the time interval between the parameter modification operation and the occurrence of the on-site anomaly. Abnormal events with a time difference less than a preset time window are considered as associated abnormal events. Associated abnormal events indicate on-site abnormalities that have a potential causal relationship with parameter modification operations. The status labels of associated abnormal events are merged into the modification action sequence to form an aligned state sequence, which is used to provide the environmental context features of each modification action.

3. The business process mining based enterprise digital transformation assessment method as claimed in claim 2 wherein: The intention proportion calculation based on the aligned state sequence and preset parameter constraint boundaries is used to obtain the intention quantification value, including: In response to the absence of associated abnormal events in the alignment state sequence, the corresponding modification action in the modification action sequence is taken as a resisted modification action. The resisted modification action represents a manual modification operation without on-site abnormal support. Obtain the parameter modification range and preset parameter allowable fluctuation range of the resisted modification action. The parameter modification range represents the absolute difference between the modified value and the original value. The resistance intention value is calculated based on the ratio of the parameter modification range of the resisted modification action to the preset parameter allowable fluctuation range. The resistance intention value represents the degree to which the dispatcher subjectively rejects the intelligent system. The resistance intention value is used as the intention quantification value, which is used to quantify the subjective resistance component in the manual modification operation.

4. The business process mining based enterprise digital transformation assessment method as claimed in claim 2 wherein: The intention proportion calculation based on the aligned state sequence and preset parameter constraint boundaries is used to obtain the intention quantification value, including: In response to the presence of associated abnormal events in the alignment state sequence, the corresponding modification actions in the modification action sequence are taken as potential correction actions. Potential correction actions indicate manual modification operations supported by on-site abnormalities. Extract the anomaly level of associated abnormal events and the parameter modification range of potential corrective actions. The anomaly level indicates the severity of the on-site anomaly. The preset parameter constraint boundary is determined based on the anomaly level. The preset parameter constraint boundary represents the maximum reasonable range of manual modification allowed under the corresponding anomaly level. In response to parameter modification magnitude being less than or equal to the preset parameter constraint boundary, potential correction actions are treated as reasonable correction actions, and a correction intention value is calculated. This correction intention value is used as the intention quantification value, which represents the reasonable correction component in the manual modification operation.

5. The enterprise digital transformation assessment method based on business process mining as described in claim 4, characterized in that: The step of calculating the intent proportion of the modified action sequence based on the aligned state sequence and preset parameter constraint boundaries to obtain the intent quantification value also includes: In response to parameter modification exceeding the preset parameter constraint boundary, potential correction actions are treated as mixed modification actions. Mixed modification actions indicate modification operations where the modification range exceeds the reasonable correction range. The modification amount corresponding to the preset parameter constraint boundary is obtained as the reasonable correction amount. The reasonable correction amount represents the minimum modification value required to deal with on-site anomalies. The amount of modification that exceeds the preset parameter constraint boundary is taken as the excessive resistance amount. The excessive resistance amount represents the additional modification value that exceeds the reasonable correction range. Based on the ratio of reasonable correction amount to excessive resistance amount, calculate the correction intention value and resistance intention value of the mixed modification action. The correction intention value and resistance intention value are used together as the intention quantification value. The intention quantification value is used to decompose the dual intention components in the mixed modification action.

6. The enterprise digital transformation assessment method based on business process mining as described in claim 5, characterized in that: The process of correcting the initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value includes: In response to the presence of resisting intent values ​​in the quantitative intent values, the overall resistance index is obtained by statistically analyzing all resisting intent values. The overall resistance index represents the cumulative degree of subjective rejection of the intelligent system throughout the entire evaluation period. Obtain the initial score of human-machine collaboration in the preset initial score of digital transformation. The initial score of human-machine collaboration represents the collaboration baseline score before considering resistance to modification operations. The collaborative deduction amount is calculated by multiplying the overall resistance index by the preset collaborative deduction coefficient. The collaborative deduction amount represents the negative impact of the resistance operation on the degree of collaboration. Subtracting the collaboration deduction from the initial score of human-machine collaboration yields the target score of human-machine collaboration, which reflects the actual level of collaboration between humans and the system.

7. The enterprise digital transformation assessment method based on business process mining as described in claim 6, characterized in that: The process of correcting the initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value also includes: In response to the presence of corrective intent values ​​in the quantitative intent values, the overall corrective index is obtained by statistically analyzing all corrective intent values. The overall corrective index represents the cumulative degree to which the intelligent system's suggestions deviate from reality throughout the entire evaluation period. Obtain the initial system maturity score from the preset initial score for digital transformation. The initial system maturity score represents the baseline score of system maturity before considering corrective modification operations. The maturity deduction amount is calculated by multiplying the overall correction index by the preset maturity deduction coefficient. The maturity deduction amount represents the negative impact of unreasonable system suggestions on maturity. Subtracting the maturity deduction from the initial system maturity score yields the system maturity target score, which reflects the actual applicability level of the intelligent scheduling system.

8. The enterprise digital transformation assessment method based on business process mining as described in claim 7, characterized in that: The process of correcting the initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value also includes: Obtain the first preset evaluation coefficient corresponding to the target score of human-machine collaboration and the second preset evaluation coefficient corresponding to the target score of system maturity. The first preset evaluation coefficient and the second preset evaluation coefficient represent the importance of different evaluation dimensions in the overall evaluation. The target score of human-machine collaboration is weighted according to the first preset evaluation coefficient to obtain the collaboration evaluation component, which represents the weighted value of the actual collaboration level between human and system. The system maturity target score is weighted according to the second preset evaluation coefficient to obtain the maturity evaluation component, which represents the weighted value of the actual applicability level of the intelligent scheduling system. The collaborative assessment component and the maturity assessment component are summed to obtain the target transformation assessment value, which is used to comprehensively reflect the true effectiveness of the enterprise's digital transformation.

9. The enterprise digital transformation assessment method based on business process mining as described in claim 3, characterized in that: The acquisition of the parameter modification range and preset parameter allowable fluctuation range for resisting modification actions includes: Obtain the original scheduling parameter values ​​and manually modified parameter values ​​corresponding to the resisted modification actions. The original scheduling parameter values ​​represent the initial values ​​automatically generated by the intelligent scheduling system. The absolute difference between the original production scheduling parameter values ​​and the manually modified parameter values ​​is calculated. The absolute difference is used as the parameter modification range, which represents the absolute change of the initial value by the manual modification operation. Obtain the normal operating condition parameter range in the production environment state sequence, and use the normal operating condition parameter range as the preset parameter allowable fluctuation range. The normal operating condition parameter range represents the normal fluctuation range of the parameter when there are no abnormal events. The parameter modification range is compared with the preset parameter allowable fluctuation range. The resistance intention value is determined based on the comparison result. The comparison result is used to reflect the degree to which the manual modification operation deviates from the normal fluctuation range.

10. A business process mining-based enterprise digital transformation assessment system, employing the business process mining-based enterprise digital transformation assessment method as described in any one of claims 1 to 9, characterized in that, include: The acquisition module is used to respond to the triggering of production scheduling parameter modification events, and to acquire the modification action sequence and production environment status sequence in the system log. The modification action sequence includes the timestamp and modification value of manually modified parameters. The alignment module is used to perform time window alignment processing on the sequence of modification actions and the sequence of production environment states to obtain an aligned state sequence, which contains the correspondence between modification actions and abnormal events in the production environment. The calculation module is used to calculate the intention ratio of the modified action sequence based on the alignment state sequence and preset parameter constraint boundaries, and obtain the intention quantification value, which includes the resistance intention value and the correction intention value. The correction module is used to correct the preset initial score of digital transformation based on the intent quantification value to obtain the target transformation evaluation value, which is used to reflect the true level of collaboration in the enterprise's digital transformation process.

Images