A data management method and system for evaluating the maturity of digital transformation in the coal industry

Abstract

The application discloses a coal industry digital transformation maturity evaluation data management method and system, relates to the coal industry digital transformation maturity evaluation data management field, generates an event identifier containing a preset cause-effect chain for each event by responding to original event data collected in an industrial field, and actively pushes the event identifier to an associated receiving device to trigger a preset action, so that intelligent identification and response of event data are realized. More importantly, in the process of constructing an event cause-effect diagram, the method can intelligently detect conflicts between physical time stamps in original event data and cause-effect logic indicated by the preset cause-effect chain. The application can significantly improve the authenticity and reliability of coal industry digital transformation maturity evaluation data, provide solid data support for enterprise strategic decision-making, avoid incorrect decisions caused by data deviation, and has remarkable excellent technical effects.

Description

Technical Field

[0001] This application relates to the field of data management for assessing the maturity of digital transformation in the coal industry, and more specifically, to a method and system for managing data for assessing the maturity of digital transformation in the coal industry. Background Technology

[0002] In the digital transformation of the coal industry, enterprises urgently need a data management system that accurately reflects progress. Data time consistency across systems and environments has been a long-standing pain point—it involves not only technical differences but is also strongly related to the underground operating environment and complex integration.

[0003] In the early stages of the transformation, the company introduced heterogeneous systems from multiple vendors to quickly launch new functions: the production control system used millisecond-level timestamps, while the environmental monitoring system relied on the operating system for second-level recording. These differences in timing mechanisms created potential problems. In the harsh underground environment, the aging of the quartz crystal oscillators in the old system caused frequency drift, and the deviation between the timestamps and the standard source changed from constant to continuously increasing, exacerbating the inconsistency.

[0004] To reduce synchronization failure alerts, operations and maintenance (O&M) deliberately relaxed the "time lag tolerance range"—short-term alert reduction—effectively masking problems and allowing biased data to flow into the central database. Ultimately, these accumulated biases erupted during the transformation maturity assessment: event sequence was disordered, causal analysis failed, alerts were inaccurate, and even misleading decisions were made, compromising the reliability of the assessment report. Therefore, effectively managing and correcting complex time skewers and ensuring data accuracy has become a unique technical challenge that urgently needs to be addressed. Existing technologies urgently need improvement. Summary of the Invention

[0005] This application discloses a data management method and system for evaluating the maturity of digital transformation in the coal industry. It aims to solve the problem of inconsistent data timestamps caused by factors such as heterogeneous systems, hardware aging, and operation and maintenance strategies during the digital transformation of the coal industry, which affects the accuracy and reliability of the evaluation of the maturity of digital transformation.

[0006] The technical solution of this application is as follows:

[0007] Firstly, this application discloses a data management method for evaluating the maturity of digital transformation in the coal industry, the method comprising:

[0008] In response to raw event data collected from the industrial site, an event identifier containing a preset causal chain is generated for each event. This preset causal chain is predefined by event type association rules.

[0009] The event identifier of each event is actively pushed to the associated receiving device through the industrial network, and the receiving device is triggered to perform a preset action.

[0010] Based on the preset causal chain information contained in the event identifier, an event causal graph is constructed;

[0011] During the construction of the event causal graph, if a conflict is detected between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, the physical timestamp of the conflicting event is corrected according to the causal logic to obtain the logical calibration timestamp.

[0012] The raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operation status sensors. The data records contain the occurrence time, event type, and any one of the relevant parameter values.

[0013] Optionally, prior to the step of generating an event identifier containing a predefined causal chain for each event in response to raw event data collected from the industrial site, the method further includes:

[0014] It receives raw event data from sensor units in the industrial field, and for each sensor unit, it performs spatiotemporal correlation aggregation on all raw event data generated by the same sensor unit within a sliding time window;

[0015] For each event within the aggregation time window, the corresponding physical parameters are extracted, and the instantaneous change amplitude and rate of change of these physical parameters are calculated to obtain the instantaneous change characteristics;

[0016] Based on instantaneous change characteristics, key events with strong causal potential are identified. Combined with the rate of change, the causal transmission efficiency of key events to related events is quantified. And based on causal potential and transmission efficiency, a pre-defined causal chain is generated.

[0017] Furthermore, by analyzing the lead-lag relationship and the degree of intercorrelation between different physical parameters within the same aggregation time window, the temporal correlation between physical parameters can be obtained.

[0018] Optionally, when a conflict is detected, the step of correcting the physical timestamp of the conflict event according to causal logic to obtain a logically calibrated timestamp includes:

[0019] When the pre-defined causal chain description in the original event data is ambiguous or contradictory, the causal weight of each potential cause event is dynamically assigned based on the instantaneous change characteristics and temporal correlation.

[0020] Identify the potential causal event with the highest causal weight as the dominant cause, and correct the physical timestamp of the conflicting event according to the dominant cause to obtain the logically calibrated timestamp.

[0021] Optionally, when the pre-defined causal chain description in the original event data is ambiguous or contradictory, the step of dynamically assigning causal weights to each potential cause event based on instantaneous change characteristics and temporal correlation includes:

[0022] Real-time acquisition of current operating environment parameters;

[0023] Based on the current operating environment parameters, adjust the baseline value and change threshold used for calculating the time-varying characteristics, and adjust the time lag range and correlation threshold used for analyzing time-series correlations;

[0024] Based on the adjusted baseline value, change threshold, time lag range, and correlation threshold, calculate the instantaneous change characteristics and temporal correlation of physical parameters;

[0025] Based on instantaneous change characteristics and temporal correlation, causal weights are dynamically assigned to each potential cause event.

[0026] This technical solution allows for the adjustment of causal weight calculation parameters based on real-time environmental parameters, making causal weight allocation more environmentally adaptable and accurate, thereby improving the robustness of timestamp correction.

[0027] Optionally, when the pre-defined causal chain description in the original event data is ambiguous or contradictory, the step of dynamically assigning causal weights to each potential cause event based on instantaneous change characteristics and temporal correlation includes:

[0028] Real-time acquisition of the linkage response mode of physical parameters associated with multiple potential cause events;

[0029] Compare the differences in linkage response modes;

[0030] Identify the dominant causal event based on the differences;

[0031] Based on the identified dominant cause event, each potential cause event is dynamically assigned a causal weight.

[0032] Optionally, the steps for obtaining the linkage response pattern of physical parameters associated with multiple potential causal events in real time include:

[0033] The physical parameters are filtered for coal mine impulse noise to obtain the processed physical parameter data.

[0034] The processed physical parameter data is time-aligned to obtain time-aligned physical parameter data;

[0035] Based on preset linkage rules, extract the sequence of physical parameters associated with potential causal events from the time-aligned physical parameter data;

[0036] Based on the sequence of physical parameters, a linkage mode diagram of coordinated response of equipment groups is constructed.

[0037] Optionally, the steps of filtering the physical parameters for coal mine impulse noise to obtain the processed physical parameter data include:

[0038] Identify the noise type and outlier characteristics of physical parameters;

[0039] Based on the identified noise type and outlier characteristics, a filtering method is selected for the physical parameters;

[0040] By combining selected filtering methods, the physical parameters are filtered to obtain processed physical parameter data.

[0041] Optionally, the steps for identifying the noise type and outlier characteristics of the physical parameters include:

[0042] The physical parameters are decomposed into multiple frequency band components and residual components;

[0043] The energy distribution, statistical characteristics, and time correlation of multiple frequency band components and residual components are analyzed to distinguish different types of noise.

[0044] The system detects whether there are abnormal fluctuations in the residual components, and performs correlation analysis between the time of occurrence of abnormal fluctuations and the noise intensity changes in multiple frequency band components to obtain the analysis results.

[0045] Based on the analysis results, the types of composite noise and outlier characteristics present in the physical parameters were determined.

[0046] Optionally, the step of selecting a filtering method for the physical parameters based on the identified noise type and outlier characteristics includes:

[0047] Identify transient pulse noise caused by the start-up and shutdown of electromechanical equipment and low-frequency drift noise caused by changes in rock stress;

[0048] A cascaded adaptive filter is used to suppress interference components in different frequency bands in stages;

[0049] The signal fidelity is monitored in real time during the filtering process, and the filtering strategy is reorganized when a fault characteristic waveform of the equipment is detected.

[0050] Secondly, this application also discloses a data management system for evaluating the maturity of digital transformation in the coal industry, the system comprising:

[0051] The identifier generation module is used to generate an event identifier for each event in response to raw event data collected in the industrial field, which contains a preset causal chain. The preset causal chain is predefined by event type association rules. The raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operation status sensors. The data records contain the occurrence time, as well as any one of the event type and related parameter values.

[0052] The data transmission module is used to actively push the event identifier of each event to the associated receiving device through the industrial network and trigger the receiving device to execute preset actions;

[0053] The causal construction module is used to construct an event causal graph based on the preset causal chain information contained in the event identifier;

[0054] The time adjustment module is used to correct the physical timestamps of conflicting events based on the causal logic indicated by the preset causal chain information when a conflict is detected in the original event data during the construction of the event causal graph, so as to obtain the logical calibration timestamps. Beneficial effects

[0055] The data management method for evaluating the maturity of digital transformation in the coal industry disclosed in this application generates an event identifier containing a preset causal chain for each event by responding to raw event data collected from the industrial site, and actively pushes it to associated receiving devices to trigger preset actions, thus achieving intelligent identification and response to event data. More importantly, in the process of constructing the event causal graph, this method can intelligently detect conflicts between the physical timestamps in the raw event data and the causal logic indicated by the preset causal chain. Once a conflict is detected, the system will correct the physical timestamps of the conflicting events according to the causal logic, thereby obtaining a logically calibrated timestamp.

[0056] Through the above technical solution, this application effectively solves the problem of inconsistent data timestamps caused by factors such as heterogeneous systems, hardware aging, and operation and maintenance strategies in existing technologies. Traditional methods often mask the problem by relaxing the tolerance range for time differences, leading to a decline in data quality and affecting the accuracy of digital transformation maturity assessment. This application, by introducing a pre-set causal chain and logical calibration mechanism, fundamentally ensures the accuracy of event sequences and logical relationships, overcoming the shortcomings of disordered timelines and unreliable causal analysis in existing technologies. Therefore, this application can significantly improve the authenticity and reliability of digital transformation maturity assessment data in the coal industry, providing solid data support for enterprise strategic decision-making, avoiding erroneous decisions due to data bias, and demonstrating significant and superior technical effects. Attached Figure Description

[0057] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0058] Figure 1 An exemplary flowchart illustrates a data management method for evaluating the maturity of digital transformation in the coal industry.

[0059] Figure 2 An exemplary schematic diagram of a data management system for evaluating the maturity of digital transformation in the coal industry is shown.

[0060] Figure reference numerals: 100, A data management system for evaluating the maturity of digital transformation in the coal industry; 10, Identifier generation module; 20, Data transmission module; 30, Causal construction module; 40, Time adjustment module. Detailed Implementation

[0061] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0062] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0063] Against the backdrop of digital transformation in the coal industry, enterprises have an urgent need for a data management system that can accurately reflect the progress of the transformation. However, ensuring that data from different systems and environments maintains a high degree of consistency over time to support reliable evaluation and decision-making remains a long-standing challenge in actual deployment and operation. Traditional data management methods often fall short when dealing with timestamp differences between heterogeneous systems, crystal oscillator drift caused by hardware aging, and lenient timestamp calibration rules set by maintenance personnel to reduce alarms. These problems collectively lead to frequent errors or significant deviations in evaluation models when establishing the sequence of events and logical relationships between data from different sources. This makes causal analysis based on event sequences unreliable, significantly reduces the accuracy of risk warnings, and seriously undermines the accuracy of digital transformation maturity assessment reports and their value as a basis for enterprise strategic decision-making.

[0064] like Figure 1 As shown, an exemplary flowchart illustrates a data management method for assessing the maturity of digital transformation in the coal industry. This application proposes a data management method for assessing the maturity of digital transformation in the coal industry, comprising:

[0065] S10, in response to raw event data collected from the industrial site, generates an event identifier for each event containing a preset causal chain, which is predefined by event type association rules.

[0066] Raw event data refers to the unprocessed, original data records collected from various sensors, control systems, and monitoring equipment at the coal industry site. This data typically includes information such as event type, occurrence time, and relevant parameter values, forming the basis for all subsequent analysis and processing.

[0067] In some embodiments, the raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operation status sensors. These data records contain the occurrence time, and any one of the event type and related parameter values. For example, the raw event data may include concentration values ​​recorded by a gas concentration sensor, start / stop signals of the belt conveyor operation status, or real-time data on the fan speed.

[0068] "Predefined causal chains" refer to predefined logical sequences that describe the causal relationships between different events. These causal chains are derived from domain expert knowledge, historical data analysis, or machine learning model training, and are used to guide the association and analysis of events.

[0069] S20: The event identifier of each event is actively pushed to the associated receiving device through the industrial network, and the receiving device is triggered to perform a preset action.

[0070] The event identifier is a unique identifier generated for each event, containing information about the pre-defined causal chain associated with that event. This identifier not only uniquely identifies an event but also carries its potential causal relationship, facilitating subsequent data transmission and processing.

[0071] Receiving devices are those in industrial networks that receive event identifiers and execute preset actions, such as PLCs (Programmable Logic Controllers), DCSs (Distributed Control Systems), SCADA (Supervisory Control and Data Acquisition) servers, or edge computing devices.

[0072] S30, construct an event causal graph based on the preset causal chain information contained in the event identifier.

[0073] S40, during the construction of the event causal graph, if a conflict is detected between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, the physical timestamp of the conflicting event is corrected according to the causal logic to obtain the logical calibration timestamp.

[0074] Among them, the logical calibration timestamp is a timestamp that conforms to causal logic and is obtained by correcting it using this method after detecting a conflict between the physical timestamp and causal logic.

[0075] This method is primarily implemented in a digital transformation maturity assessment system for coal industry sites. This system typically includes a data acquisition layer, a data transmission layer, a data processing and analysis layer, and an application layer. The data acquisition layer is responsible for acquiring raw event data from various sensors and devices; the data transmission layer transmits data to the processing and analysis layer via an industrial network; the data processing and analysis layer is responsible for generating event identifiers, constructing cause-effect graphs, and calibrating timestamps; the application layer uses the calibrated data for digital transformation maturity assessment, risk warning, and decision support. The entire system needs to possess high real-time performance, high reliability, and high data consistency to adapt to the complex and ever-changing environment underground in coal mines.

[0076] The core of the data management method for evaluating the maturity of digital transformation in the coal industry proposed in this application lies in effectively solving the problem of data time consistency in the digital transformation of the coal industry through a series of innovative steps. The following will elaborate on each of the main features.

[0077] First, this method responds to raw event data collected from the industrial site, generating an event identifier for each event that contains a predefined causal chain. This predefined causal chain is defined by event type association rules. In practice, raw event data may come from various sensors and devices, such as gas sensors, temperature sensors, pressure sensors, and conveyor belt operation status sensors. When these sensors collect data and form an event, such as "increased gas concentration," "conveyor belt start," or "abnormal fan speed," the system generates a unique event identifier for that event. This event identifier is not just a simple sequence number; it also embeds predefined causal chain information. For example, if "increased gas concentration" typically leads to "accelerated fan operation," then the identifier for the "increased gas concentration" event will contain causal chain information pointing to "accelerated fan operation." The definition of the predefined causal chain can be based on historical data analysis, expert experience, or machine learning models. For example, by analyzing a large amount of historical operating data, the frequent occurrence order and correlation patterns between different events can be identified, thereby establishing a set of event type association rules. These rules can be as simple as "if A occurs, then B is likely to occur in time T", or they can be complex causal inference rules based on probabilistic graphical models.

[0078] Secondly, this method proactively pushes the event identifier of each event to the associated receiving device via the industrial network, triggering the receiving device to execute preset actions. In coal industry sites, the industrial network is a critical infrastructure connecting various devices and systems. When an event identifier is generated, it does not passively wait for the receiving device to query it, but is proactively pushed to the pre-configured associated receiving devices. For example, if an event identifier indicates "gas concentration exceeds the limit," and its preset causal chain points to "starting the local ventilation fan," then this event identifier will be proactively pushed to the PLC or DCS system controlling the local ventilation fan. After receiving this event identifier, the receiving device will immediately parse the causal chain information and execute the corresponding operation according to the preset action rules. For example, the PLC may immediately issue a command to start the local ventilation fan, or the DCS system may adjust the fan speed to reduce the gas concentration. This proactive push mechanism ensures real-time transmission and response of event information, avoiding the delays that may be caused by traditional polling mechanisms.

[0079] Furthermore, this method performs causal association analysis on events in the original event data based on the preset causal chain information contained in the event identifiers, constructing an event causal graph. The event causal graph is the core tool for time calibration and causal analysis in this method. After receiving a large amount of original event data and its associated event identifiers, the system extracts the preset causal chain information contained in each event identifier and constructs a complex event causal graph based on this. In this graph, each node represents an event, and each edge represents a causal relationship. For example, if "increased gas concentration" is the cause of "accelerated wind turbine operation," then there will be an edge in the causal graph pointing from "increased gas concentration" to "accelerated wind turbine operation." This causal graph not only shows the direct causal relationships between events but can also demonstrate more complex indirect causal relationships through multi-level causal chains. The process of constructing the causal graph can employ graph database technology, storing and managing events as nodes and causal relationships as edges.

[0080] Finally, during the construction of the event causal graph, if a conflict is detected between the physical timestamps in the original event data and the causal logic indicated by the preset causal chain information, the physical timestamps of the conflicting events are corrected according to the causal logic to obtain logically calibrated timestamps. This is a key step in solving the time consistency problem. When constructing the event causal graph, the system considers both the physical timestamps of events and their preset causal logic. For example, if the causal logic indicates that "Event A" is the cause of "Event B," then under normal circumstances, the physical timestamp of "Event A" should be earlier than the physical timestamp of "Event B." However, due to time deviations in heterogeneous systems, hardware aging, or transmission delays, the physical timestamp of "Event A" may be later than the physical timestamp of "Event B," or the time difference between the two may not conform to the preset time delay range in the causal chain. When such a conflict is detected, the system does not simply accept the original physical timestamps but corrects the physical timestamps of the conflicting events according to the preset causal logic. For example, if the physical timestamp of "Event A" is later than that of "Event B," but the causal logic explicitly states that "Event A" is the cause of "Event B," then the system might adjust the physical timestamp of "Event A" to before "Event B," or, based on a pre-defined time delay in the causal chain, adjust the physical timestamp of "Event B" to after "Event A," thus obtaining a logically calibrated timestamp. This correction process can employ heuristic algorithms or probabilistic model-based reasoning methods to ensure that the corrected timestamp conforms to causal logic while being as close as possible to the original physical timestamp.

[0081] The data management method for evaluating the maturity of digital transformation in the coal industry proposed in this application forms a complete and efficient data management process through the close coordination of the above-mentioned technical features, aiming to solve the long-standing problem of data time consistency in the digital transformation of the coal industry.

[0082] Specifically, when raw event data is collected at the industrial site, such as an abnormally high methane concentration, the system first responds to this data and generates an event identifier containing a predefined causal chain. This predefined causal chain, such as "high methane concentration" leading to "accelerated fan operation," is established based on predefined event type association rules. By embedding causal chain information into the event identifier, the system ensures that each event carries its potential causal relationship from the outset, laying the foundation for subsequent analysis.

[0083] Subsequently, this event identifier, containing causal chain information, is proactively pushed to associated receiving devices via the industrial network. For example, the PLC or DCS system controlling the fan will receive this identifier. Upon receiving the identifier, the receiving device will immediately trigger a preset action, such as starting a local ventilation fan or adjusting the fan speed. This proactive push mechanism ensures real-time transmission and response to event information, avoiding the delays that may occur with traditional polling mechanisms, thereby improving the system's response speed and efficiency to emergencies.

[0084] At the data processing and analysis layer, the system constructs a comprehensive cause-effect graph based on the pre-defined causal chain information contained in the event identifiers. This graph visually illustrates the causal relationships between different events, providing a macroscopic perspective for understanding complex industrial processes. For example, the graph clearly shows how an increase in gas concentration propagates layer by layer, ultimately affecting the operating status of the fan.

[0085] However, due to the complexity of industrial sites in the coal industry, including time deviations in heterogeneous systems, crystal oscillator drift caused by hardware aging, and transmission delays, the physical timestamps in the raw event data often exhibit inconsistencies. During the construction of the event causal graph, the system continuously detects whether there is a conflict between the physical timestamps in the raw event data and the causal logic indicated by the preset causal chain information. For example, if the causal logic indicates that "increased gas concentration" precedes "accelerated operation of the ventilation fan," but the actual collected physical timestamps show that "accelerated operation of the ventilation fan" precedes "increased gas concentration," this constitutes a conflict between the timestamps and the causal logic.

[0086] Once such a conflict is detected, this method does not simply accept the original physical timestamps. Instead, it corrects the physical timestamps of the conflicting events according to a pre-defined causal logic, thus obtaining logically calibrated timestamps. For example, if the physical timestamp of "increased gas concentration" is later than "accelerated fan operation," but the causal logic clearly indicates that the former is the cause of the latter, the system will adjust the physical timestamp of "increased gas concentration" to before "accelerated fan operation" according to the causal logic, or adjust the physical timestamp of "accelerated fan operation" according to a pre-defined time delay in the causal chain. This logical calibration ensures that the timestamps of all events conform to their inherent causal relationships, thereby eliminating causal analysis errors caused by inconsistent timestamps.

[0087] Through the aforementioned collaborative efforts, the data management method for assessing the maturity of digital transformation in the coal industry proposed in this application can effectively manage and correct complex and cumulative time biases, ensuring the authenticity and reliability of the assessment data. It not only improves the real-time performance and accuracy of data processing but also provides a solid foundation for causal analysis based on event sequences, thereby significantly enhancing the accuracy of the digital transformation maturity assessment report and its value as a basis for corporate strategic decision-making.

[0088] The core innovation of this application lies in the introduction of a "pre-defined causal chain" and a "logical calibration timestamp" mechanism. First, by generating an event identifier containing a pre-defined causal chain for each event, this application endows the event with inherent causal logical attributes at the source of data generation. This is fundamentally different from traditional methods that only record the timestamp of event occurrence; it ensures that data carries important semantic information throughout transmission and processing.

[0089] Secondly, this application employs an active push mechanism, which pushes event identifiers to associated receiving devices in real time via the industrial network and triggers preset actions. This real-time response capability is significantly superior to the polling delays or passive data synchronization that may exist in traditional methods, ensuring that critical events can be handled in a timely manner, thereby improving the overall efficiency and security of the system.

[0090] More importantly, this application can intelligently detect conflicts between the physical timestamps in the original event data and the causal logic indicated by the preset causal chain during the construction of the event causal graph. When such a conflict is detected, this application no longer simply accepts the original timestamps, but corrects the physical timestamps of the conflicting events according to the causal logic to obtain logically calibrated timestamps. This mechanism is the most innovative part of this application. It breaks through the limitations of traditional methods that rely solely on physical timestamps and introduces "causal logic" as a higher-level basis for time calibration. This means that even if there are deviations in the physical timestamps, as long as the causal logic is clear, the system can correct the time through logical reasoning, thereby ensuring the accuracy of the event sequence and the reliability of the causal relationship.

[0091] For example, in traditional methods, if a sensor clock drift causes the physical timestamp of "increased gas concentration" to be later than "accelerated fan operation," the system might incorrectly conclude that fan acceleration is unrelated to increased gas concentration, or that the causal relationship is reversed. However, in this application, because the pre-defined causal chain explicitly states that "increased gas concentration" is the cause of "accelerated fan operation," even if there is a conflict in the physical timestamps, the system will correct the timestamps according to causal logic, ensuring that "increased gas concentration" logically precedes "accelerated fan operation," thereby avoiding erroneous causal judgments.

[0092] Therefore, this application not only solves the technical challenge of data time consistency in the digital transformation of the coal industry, but also significantly improves the accuracy and reliability of data analysis by introducing a causal logic-driven time calibration mechanism. This method can provide a more realistic and reliable data foundation for evaluating the maturity of digital transformation, thereby providing stronger support for corporate strategic decision-making. Its technical contribution and creativity are self-evident.

[0093] In some embodiments, the above-described coal industry digital transformation maturity evaluation data management method further includes, before the step of generating an event identifier containing a preset causal chain for each event in response to the raw event data collected in the industrial field:

[0094] It receives raw event data from sensor units in the industrial field, and for each sensor unit, it performs spatiotemporal correlation aggregation on all raw event data generated by the same sensor unit within a sliding time window;

[0095] For each event within the aggregation time window, the corresponding physical parameters are extracted, and the instantaneous change amplitude and rate of change of these physical parameters are calculated to obtain the instantaneous change characteristics;

[0096] Based on the instantaneous change characteristics, key events with strong causal potential are identified. Combined with the change rate, the causal transmission efficiency of key events to related events is quantified. And based on the causal potential and transmission efficiency, a preset causal chain is generated.

[0097] Furthermore, by analyzing the lead-lag relationship and the degree of intercorrelation between different physical parameters within the same aggregation time window, the temporal correlation between physical parameters can be obtained.

[0098] Specifically, receiving raw event data from sensor units in an industrial field refers to acquiring continuous or discrete measurement data in real time from various sensors deployed at the coal mine site. This raw event data may contain a large amount of noise and redundant information. For each sensor unit, all raw event data generated by the same sensor unit within a sliding time window are spatiotemporally correlated and aggregated to perform preliminary cleaning and integration of the raw, discrete sensor data. The sliding time window can be understood as a preset time period, such as 5 seconds, 10 seconds, or longer. Within this time window, data from the same sensor are collected and aggregated using statistical methods (such as calculating the average, median, maximum, minimum, or trend analysis) to eliminate instantaneous noise and extract more stable data features. Spatiotemporal correlation aggregation aims to integrate data that are temporally close and spatially originate from the same source to form more representative event data blocks.

[0099] For each event within the aggregation time window, corresponding physical parameters are extracted, and the instantaneous amplitude and rate of change of these physical parameters are calculated to obtain instantaneous change characteristics. Physical parameters refer to the physical quantities actually measured by the sensor, including at least temperature, pressure, and gas concentration, and may also include vibration frequency. Instantaneous amplitude refers to the amount of numerical change of a physical parameter within a short period, while the rate of change refers to the degree of change of the physical parameter per unit time. For example, the amplitude of change can be obtained by calculating the difference between the current value and the average value within the previous aggregation time window, and the rate of change can be obtained by dividing the difference by the time interval. These instantaneous change characteristics reflect the dynamics and intensity of the event and are key indicators for identifying the importance of the event.

[0100] In practical applications, based on the instantaneous change characteristics, key events with strong causal potential are identified. Combined with the rate of change, the causal transmission efficiency of key events to related events is quantified. Based on the causal potential and transmission efficiency, a pre-defined causal chain is generated. Key events with strong causal potential refer to those whose changes can significantly affect other events or system states. For example, a sharp rise in gas concentration may trigger a gas over-limit alarm, and an abnormal temperature rise may indicate equipment overheating. Causal potential can be assessed by analyzing the significance, duration, and synchronicity or lag of instantaneous change characteristics with changes in other parameters. Causal transmission efficiency quantifies the strength and speed of the impact of a key event on related events; for example, how long and to what extent another related event responds after one event occurs. Through these analyses, pre-defined causal chains can be dynamically generated or optimized to more accurately reflect the causal relationships in actual industrial settings, rather than relying solely on static predefined rules.

[0101] Furthermore, analyzing the lead-lag relationships and intercorrelation degrees among changes in different physical parameters within the same aggregation time window yields the temporal correlations between physical parameters. A lead-lag relationship refers to the fact that a change in one physical parameter precedes or lags a change in another, which is crucial for determining causal relationships. For example, an increase in motor current may precede an increase in motor temperature. The degree of intercorrelation measures the consistency or correlation of the trends in the changes of two or more physical parameters. These relationships can be quantified, for example, using statistical methods such as cross-correlation functions and Granger causality tests. Temporal correlations provide deeper evidence for the generation and verification of causal chains, contributing to the construction of more accurate and robust event causal graphs.

[0102] This application's solution effectively addresses the problems of low quality raw event data and insufficient accuracy of pre-defined causal chains by introducing a series of data preprocessing and causal chain generation steps before generating event identifiers. First, by aggregating the spatiotemporal correlations of the raw event data, noise can be effectively filtered out, data redundancy reduced, and discrete data integrated, thus providing a cleaner and more representative data foundation for subsequent analysis. Second, by extracting the instantaneous magnitude and rate of change of physical parameters, the dynamic characteristics and intensity of events can be captured, which is crucial for identifying truly causally significant key events. It is precisely through in-depth analysis of these instantaneous change characteristics that key events with strong causal potential can be identified, and their causal transmission efficiency to related events can be quantified, thereby enabling the dynamic and accurate generation or optimization of pre-defined causal chains. Furthermore, by analyzing the lead-lag relationships and interrelationships between different physical parameter changes, the reliability and accuracy of the causal chains are further enhanced, providing a solid logical foundation for the subsequent construction of event causal graphs. This preprocessing and optimization mechanism allows the subsequently generated event identifiers to more accurately reflect the causal relationships of actual events, thereby improving the effectiveness of the entire data management method.

[0103] In some embodiments, if a conflict is detected between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, then the step of correcting the physical timestamp of the conflicting event according to the causal logic to obtain a logically calibrated timestamp includes:

[0104] When the preset causal chain description in the original event data is ambiguous or contradictory, each potential cause event is dynamically assigned a causal weight based on the instantaneous change characteristics and temporal correlation.

[0105] Identify the potential causal event with the highest causal weight as the dominant cause, and correct the physical timestamp of the conflicting event according to the dominant cause to obtain the logically calibrated timestamp.

[0106] Specifically, during the construction of the event causal graph, the system cannot clearly determine the direct causal relationship between events, or there may be multiple conflicting causal paths. This ambiguity or contradiction may stem from the uncertainty of sensor data, system noise, or the complex and ever-changing industrial environment. The instantaneous change characteristics can be understood as the instantaneous amplitude and rate of change of physical parameters extracted by the above methods, reflecting the dynamic intensity and trend of physical quantities at the time of the event. The temporal correlation refers to the lead-lag relationship and degree of interdependence between changes in different physical parameters within the same aggregation time window, revealing the interdependence of events in the time dimension. By comprehensively utilizing these characteristics, this application can more precisely assess the influence of each potential causal event on conflicting events.

[0107] Furthermore, based on instantaneous change characteristics and temporal correlation, the system assigns a quantified weight value to each potential causal event that may lead to a conflict. This weight value reflects its likelihood and influence as a causal event. For example, a potential causal event with a larger instantaneous change magnitude or a stronger temporal leading correlation will be assigned a higher causal weight.

[0108] Identifying the potential causal event with the highest causal weight as the dominant cause means selecting the event with the highest causal weight from all potential causal events as the most important and decisive cause. Its purpose is to provide a clear basis for correction when the causal chain is ambiguous or contradictory. Correcting the physical timestamp of the conflicting event based on the dominant cause specifically means, after determining the dominant cause, adjusting the physical timestamp of the conflicting event according to its occurrence time and its logical causal relationship with the conflicting event, making it conform to more reliable causal logic, thereby obtaining a logically calibrated timestamp.

[0109] This application's solution, when the pre-defined causal chain description is ambiguous or contradictory, no longer simply relies on potentially inaccurate pre-defined information, but instead delves into the instantaneous change characteristics and temporal correlations inherent in the original event data. It is precisely because these characteristics can more objectively reflect the actual dynamic connections and time dependencies between events that the system can dynamically assess the causal influence of each potential cause event and quantify it as a causal weight. By identifying the dominant cause with the highest causal weight, this application can find the most reliable causal basis in a highly uncertain scenario, thereby enabling more accurate and reasonable correction of the physical timestamps of conflicting events.

[0110] In some preferred embodiments, suppose that in a coal mine underground monitoring system, a sensor unit collects two raw event data points: event X (abnormal increase in gas concentration) and event Y (sudden drop in local ventilation fan speed). The physical timestamps in the raw event data show that event X occurred at 10:00:05, while event Y occurred at 10:00:03. However, pre-defined causal chain information might indicate that a "sudden drop in local ventilation fan speed" is typically the cause of an "abnormal increase in gas concentration." In this case, there is a conflict between the physical timestamps and the causal logic.

[0111] Furthermore, if the pre-defined causal chain description is ambiguous or contradictory for this specific scenario (e.g., the system cannot determine whether the increase in gas level is caused by a ventilator malfunction or whether the increase in gas level triggers a protective speed reduction of the ventilator), the solution in this application will be enabled.

[0112] Specifically, the system first extracts the instantaneous change characteristics of events X and Y based on their raw event data. For example, it analyzes the instantaneous rate and magnitude of the increase in gas concentration, and the instantaneous rate and magnitude of the decrease in fan speed. Simultaneously, it analyzes the temporal correlation of these two physical parameters in historical data; for example, how much time before the increase in gas concentration usually precedes a sudden drop in fan speed under similar operating conditions.

[0113] Based on these instantaneous change characteristics and temporal correlations, the system dynamically assigns a higher causal weight to event Y (sudden drop in local fan speed), because it is physically more likely to be the dominant cause of the abnormal increase in gas concentration. For example, if historical data shows a stronger leading correlation between the instantaneous change characteristics of the sudden drop in fan speed and the increase in gas concentration, then the causal weight of event Y will be higher.

[0114] Ultimately, the system identifies event Y as the dominant cause and, based on the occurrence time of event Y (10:00:03) and the logic behind it as the dominant cause, corrects the physical timestamp of event X. For example, the logical calibration timestamp of event X is adjusted to 10:00:06 to ensure that it conforms to the causal logic of "a sudden drop in the speed of the local ventilation fan leading to an abnormal increase in gas concentration," thus obtaining the logical calibration timestamp.

[0115] In some embodiments, when the pre-defined causal chain description in the original event data is ambiguous or contradictory, the step of dynamically assigning causal weights to each potential cause event based on instantaneous change characteristics and temporal correlation specifically includes:

[0116] Real-time acquisition of current operating environment parameters;

[0117] Based on the current operating environment parameters, adjust the baseline value and change threshold used to calculate the instantaneous change characteristics, and adjust the time lag range and correlation threshold used to analyze the time series correlation.

[0118] Based on the adjusted baseline value, change threshold, time lag range, and correlation threshold, the instantaneous change characteristics of the physical parameters and the temporal correlation are calculated.

[0119] Based on the instantaneous change characteristics and the temporal correlation, each potential cause event is dynamically assigned a causal weight.

[0120] Specifically, various environmental sensors deployed at the industrial site continuously monitor and collect environmental status data of the current work area. These parameters reflect the external conditions of equipment operation and have a significant impact on the normal fluctuation range and correlation patterns of physical parameters. Adjusting baseline values, change thresholds, time lag ranges, and correlation thresholds based on the current working environment parameters can be understood as establishing a mapping relationship or adjustment rules between environmental parameters and these calculated parameters. For example, when the ambient temperature rises, the normal operating temperature baseline value of some equipment may also rise, and its vibration or current change thresholds may need to be adjusted accordingly; when dust concentration increases, the noise level of sensor signals may increase, which will affect the setting of correlation thresholds and time lag ranges in time-series correlation analysis. These adjustments can be based on pre-established empirical models, machine learning models, or expert knowledge bases.

[0121] Therefore, based on the adjusted baseline values, change thresholds, time lag ranges, and correlation thresholds, the instantaneous change characteristics and temporal correlations of physical parameters are recalculated to ensure that these characteristics and correlations more accurately reflect the actual physical processes under the current operating environment. For example, the calculation of instantaneous change characteristics will be based on baselines and thresholds that are more consistent with the current environment, and the temporal correlation analysis will also adopt time lag and correlation standards that are more adapted to the current environment. Finally, based on these instantaneous change characteristics and temporal correlations calibrated with environmental parameters, causal weights are dynamically assigned to each potential causal event. This dynamic weighting mechanism allows causal weights to adapt to environmental changes, thereby improving their accuracy and reliability.

[0122] This application's solution addresses the problem of inaccurate calculations of instantaneous change characteristics and temporal correlations in complex and ever-changing industrial environments where fixed parameters can lead to errors in these calculations. By dynamically adjusting the baseline values ​​and change thresholds used to calculate instantaneous change characteristics, as well as the time lag range and correlation thresholds used to analyze temporal correlations, the solution acquires real-time parameters of the current operating environment. Because these key parameters can adaptively adjust according to the actual environment, the calculated instantaneous change characteristics and temporal correlations of the physical parameters more accurately reflect the actual physical processes and intrinsic connections between events under the current operating conditions. This mechanism ensures that when the pre-defined causal chain description is ambiguous or contradictory, the assigned causal weights are based on the most accurate data analysis results, thereby improving the accuracy of identifying the dominant cause event and ensuring the reliability of correcting the physical timestamps of conflicting events.

[0123] For example, suppose a coal mining machine is operating underground in a coal mine. When the machine enters a geologically complex area, such as when the coal seam hardness suddenly increases or when it encounters a fault, its operating environment parameters will change significantly, such as increased motor load, intensified vibration, and changes in the stress of the surrounding rock strata. Specifically, the system will acquire operating environment parameters in real time, including the current motor load of the coal mining machine, vibration sensor data, and stress sensor data of the surrounding rock strata. When it detects a continuous increase in motor load and vibration frequency and amplitude exceeding the normal range, the system will dynamically adjust the baseline value and change threshold used to calculate the instantaneous change characteristics of motor vibration based on these environmental parameters. For example, during normal coal seam operation, the vibration baseline value may be low and the change threshold may be narrow; however, during hard coal seam operation, the system will increase the vibration baseline value and relax the change threshold to avoid misjudging normal hard coal seam vibration as an abnormal event.

[0124] Meanwhile, to analyze the temporal correlation between motor current and the cutting head speed of the coal mining machine, the system adjusts the time lag range and correlation threshold according to the current operating environment parameters. For example, under normal operating conditions, current changes may lag speed changes by a short period; however, under high load or drill jamming conditions, this lag relationship may change, and the system will adjust the time lag range accordingly to more accurately capture the causal relationship between the two. Through this dynamic adjustment, even when the preset causal chain (e.g., "abnormal motor current causes cutting head jamming") is vaguely described or shows contradictions under specific operating conditions, the system can more accurately calculate the causal weight of abnormal motor current on the cutting head jamming event based on the instantaneous change characteristics and temporal correlation calibrated by the environment. For example, when operating in hard coal seams, even if the current fluctuation is large, if its change pattern matches the adjusted characteristics and correlation, its causal weight may be assigned a lower value to avoid misjudgment; conversely, if the current fluctuation pattern is abnormal and highly correlated with the adjusted parameters, its causal weight will be assigned a higher value, thereby accurately identifying the dominant cause and precisely correcting the physical timestamps of related events.

[0125] In some embodiments, this application further proposes an optimization scheme for dynamically assigning causal weights to each potential cause event when the pre-defined causal chain description in the original event data is ambiguous or contradictory. The step of dynamically assigning causal weights to each potential cause event based on instantaneous change characteristics and temporal correlation when the pre-defined causal chain description in the original event data is ambiguous or contradictory includes:

[0126] Real-time acquisition of the linkage response mode of multiple physical parameters associated with the potential causal events;

[0127] Compare the differences between the aforementioned linkage response modes;

[0128] Based on the aforementioned differences, identify the dominant causal event;

[0129] According to the identified leading cause event, causally weight is dynamically assigned to each of the potential cause events.

[0130] Specifically, the system continuously monitors the changing trends and mutual influence modes of physical parameters (such as temperature, pressure, flow rate, vibration, etc.) related to each potential cause event before and after the event occurs. The changes of these physical parameters are not isolated. They often form specific linkage response patterns, reflecting the collaborative working mechanism within the device or system. For example, when a certain potential cause event occurs, multiple physical parameters directly related to it may show a specific combination of increase, decrease or fluctuation in a short period of time, while the physical parameters related to other potential cause events may show different response patterns or lagged responses.

[0131] By comparing the differences in the linkage response patterns, mainly comparing and analyzing the linkage response patterns of the physical parameters associated with different potential cause events. Such differences can be reflected in aspects such as the amplitude, duration, response speed of the response, and the co-variation relationship with other parameters. For example, a leading cause event may cause a group of physical parameters to change in a rapid and large-scale linkage manner, while a secondary cause event may only cause a local or slow response.

[0132] Furthermore, by analyzing the differences in the linkage response patterns, it is determined which potential cause event is the main and decisive triggering factor when a specific conflict event occurs. The leading cause event usually triggers the most significant, most direct and most logical linkage response pattern in line with the system operation.

[0133] Once the leading cause event is identified, a higher causal weight can be assigned to it based on its key role in the linkage response, while other potential cause events are assigned corresponding lower weights according to their degree of association with the leading cause event and the intensity of the linkage response. This dynamic weighting mechanism enables the causal weight to more accurately reflect the true causal relationship between events.

[0134] The solution of this application solves the problem that it may be impossible to accurately identify the dominant cause event only based on the instantaneous change characteristics and temporal correlation when the preset causal chain in the original event data is vague or contradictory by introducing the analysis of the "linkage response mode". Specifically, when multiple potential cause events show similarities in instantaneous change characteristics and temporal correlation, making it difficult to distinguish their causal weights, this solution obtains and compares the linkage response modes of the physical parameters associated with these potential cause events in real time. The linkage response mode can more comprehensively reflect the co-variation and mutual influence within the system when an event occurs, thereby revealing deeper causal relationships. By analyzing the differences in these modes, the true dominant cause event can be more accurately identified because the dominant cause event usually triggers a more unique and significant system-level linkage response. Once the dominant cause event is identified, its causal weight will be correspondingly increased, making the subsequent correction of the physical timestamps of conflict events more accurate and reliable.

[0135] In some preferred embodiments, assume that an event of abnormal increase in gas concentration occurs in a coal mine underground, and the preset causal chain in the original event data may vaguely point to a ventilation system failure or mining operation disturbance. At this time, multiple potential cause events (for example, a decrease in the fan speed, abnormal readings of local gas sensors, an increase in the cutting operation intensity of the coal cutter) may all show instantaneous change characteristics and temporal correlation.

[0136] To more accurately dynamically assign causal weights, this solution will obtain the linkage response modes of the physical parameters associated with these potential cause events in real time. For example:

[0137] 1. If the decrease in the fan speed is the dominant cause, it may be observed that the readings of the ventilation volume sensors 400 in multiple areas underground generally decrease, while the readings of the gas concentration sensors increase synchronously at multiple points, and this change pattern highly coincides with the historical ventilation failure pattern.

[0138] 2. If the increase in the cutting operation intensity of the coal cutter is the dominant cause, it may be observed that the readings of the power consumption sensors of the coal cutter increase significantly, while the readings of the gas concentration sensors in the nearby areas increase rapidly locally, and the gas concentration changes in the areas far from the coal cutter are not obvious, and this pattern conforms to the historical mining disturbance pattern.

[0139] By comparing the differences in these co - response patterns, the system can identify which pattern best matches the true cause of the abnormal increase in gas concentration. For example, if the co - response of the first pattern is more significant, more extensive, and has a stronger causal logic with the increase in gas concentration, the decrease in fan speed is identified as the leading cause event. Subsequently, based on the leading cause of the decrease in fan speed, the system dynamically assigns a higher causal weight to it and adjusts the weights of other potential cause events accordingly. Finally, it is used to correct the physical timestamp of the abnormal increase in gas concentration event to make it consistent with the true causal logic.

[0140] In some embodiments, the step of obtaining in real - time the co - response patterns of multiple physical parameters associated with the potential cause events may include:

[0141] Perform filtering processing on the physical parameters for coal - mine impulse noise to obtain processed physical parameter data;

[0142] Perform time - series alignment on the processed physical parameter data to obtain time - series - aligned physical parameter data;

[0143] According to the preset co - response rules, extract the physical parameter sequences associated with the potential cause events from the time - series - aligned physical parameter data;

[0144] Construct a co - response pattern map of the coordinated response of the device group based on the physical parameter sequences.

[0145] Specifically, in the coal - mine industrial site, the physical parameter data collected by sensors are often disturbed by various noises, such as transient impulse noise caused by the start - stop of electromechanical equipment and low - frequency drift noise caused by rock - layer stress changes. Therefore, first, it is necessary to perform filtering processing on the physical parameters for coal - mine impulse noise to remove or suppress these noise components, so as to obtain more pure and reliable processed physical parameter data. This filtering processing aims to improve the signal - to - noise ratio of the data and lay a foundation for subsequent analysis.

[0146] Furthermore, due to differences in the data collection frequencies, transmission delays, etc. of different sensors or devices, the processed physical parameter data may have deviations on the time axis. In order to accurately analyze the co - response relationship between different physical parameters, it is necessary to perform time - series alignment on the processed physical parameter data. Time - series alignment aims to eliminate the time inconsistency and ensure that all relevant physical parameter data are compared and analyzed on the same time basis, so as to obtain time - series - aligned physical parameter data.

[0147] Building upon this, to identify the specific linkage between potential causal events and physical parameters, it is necessary to extract the sequence of physical parameters associated with the potential causal events from the time-aligned physical parameter data according to preset linkage rules. These linkage rules can be predefined based on domain expert knowledge, historical data analysis, or machine learning models, and are used to describe which physical parameters will change collaboratively and in what way when a specific event occurs. Through these rules, physical parameter data fragments closely related to the potential causal events can be accurately selected.

[0148] Finally, based on the sequence of physical parameters, a linkage pattern map of the coordinated response of the device group can be constructed. This map graphically illustrates how the physical parameters monitored by different devices or sensors interact and respond collaboratively under the influence of specific potential causal events. The map can contain nodes (representing physical parameters or devices) and edges (representing the linkage relationship, intensity, and time delay between parameters), thus intuitively presenting complex linkage mechanisms.

[0149] Through the aforementioned technical solutions, this application effectively addresses challenges such as poor data quality, inconsistent timing, and complex correlations in coal mine industrial field data, significantly improving the accuracy and reliability of acquiring physical parameter linkage response patterns. Specifically, filtering for coal mine impulse noise effectively removes interference and improves data purity; timing alignment ensures the synchronization of multi-source data, avoiding misjudgments caused by time deviations; and linkage rule extraction and pattern graph construction clearly and intuitively present the complex collaborative response relationships between potential causal events and physical parameters. Therefore, it provides high-quality, highly reliable linkage response pattern information for accurately identifying dominant causal events and dynamically assigning causal weights when the pre-defined causal chain description is ambiguous or contradictory, thereby improving the accuracy and decision support capabilities of data management for evaluating the maturity of digital transformation in the coal industry.

[0150] In some embodiments of this application described above, filtering the physical parameters to address coal mine impulse noise and obtaining processed physical parameter data is a crucial step in acquiring the linkage response mode. Specifically, the step of filtering the physical parameters to address coal mine impulse noise and obtaining processed physical parameter data can be further refined into the following sub-steps:

[0151] Identify the noise type and outlier characteristics of the physical parameters;

[0152] Based on the identified noise type and outlier characteristics, a filtering method is selected for the physical parameters;

[0153] The selected filtering methods are combined to filter the physical parameters, resulting in processed physical parameter data.

[0154] Specifically, in-depth analysis is conducted on raw physical parameter data collected from industrial sites to identify various interference components and outlier data points. This may include, but is not limited to, identifying high-frequency impulse noise, low-frequency drift, periodic interference, and anomalies such as transient spikes or drops. The aim is to comprehensively understand the nature of data contamination, providing a basis for subsequent precise filtering.

[0155] Selecting a filtering method for physical parameters based on the identified noise type and outlier characteristics can be understood as choosing the most suitable filtering algorithm or technique based on an in-depth analysis of the noise and outlier characteristics. For example, for high-frequency impulse noise, median filtering, wavelet denoising, or adaptive Wiener filtering can be chosen; for low-frequency drift, high-pass filtering or baseline correction methods can be used; and for instantaneous spikes, thresholding or outlier removal algorithms can be employed. The aim is to ensure that the selected method can efficiently and accurately remove specific types of interference while preserving the effective information of the original signal to the maximum extent possible.

[0156] In practical applications, combining selected filtering methods to filter the physical parameters and obtain processed physical parameter data specifically refers to integrating and applying multiple selected filtering methods in a certain order or in parallel. For example, one method can be used first to remove the main impulse noise, and then another method can be used to process the residual low-frequency drift, or a cascaded filter can be used to suppress interference components in different frequency bands in stages. The aim is to achieve comprehensive and refined processing of complex noisy environments through the synergistic effect of multiple stages and methods, thereby obtaining high-quality, noise-free physical parameter data.

[0157] This application's solution first identifies the types of noise and outlier characteristics present in the physical parameter data, enabling a comprehensive understanding of the nature of data contamination and avoiding under-filtering or filtered-out issues that may result from blindly applying general filtering methods. This precise understanding of noise characteristics allows for the targeted selection of the most suitable filtering methods, ensuring the efficiency and accuracy of the filtering process. Based on this, by combining multiple selected filtering methods, comprehensive and refined processing of complex noise environments can be achieved, effectively addressing multi-source and multi-type impulse noise and outliers in coal mine environments. This provides a high-quality and highly reliable data foundation for subsequent time-series alignment, physical parameter sequence extraction, and the construction of linkage pattern maps.

[0158] In some embodiments of this application described above, in order to perform filtering processing on physical parameters for coal mine impulse noise, it is necessary to identify the noise type and outlier characteristics of the physical parameters. Specifically, the steps for identifying the noise type and outlier characteristics of the physical parameters include:

[0159] The physical parameters are decomposed into multiple frequency band components and residual components;

[0160] The energy distribution, statistical characteristics, and temporal correlation of the multiple frequency band components and the residual components are analyzed to distinguish different types of noise.

[0161] The system detects whether there are abnormal fluctuations in the residual components, and performs correlation analysis between the time of occurrence of the abnormal fluctuations and the noise intensity changes in the multiple frequency band components to obtain the analysis results.

[0162] Based on the analysis results, the types of composite noise and outlier characteristics present in the physical parameters are determined.

[0163] Specifically, a preliminary analysis of the physical parameters is performed to quickly identify potential anomalies in the data. This may include calculating basic statistical measures of the data, such as mean, variance, skewness, and kurtosis, and combining them with preset thresholds for judgment. For example, when a data point suddenly exceeds the normal fluctuation range, or exhibits sustained high-frequency oscillations, it is identified as a potential fluctuation, instantaneous spike, or drop.

[0164] Furthermore, decomposing the physical parameters into multiple frequency band components and residual components is fundamental to in-depth analysis of noise and anomalies. This decomposition process can employ various signal processing techniques, such as wavelet decomposition, Fourier transform, empirical mode decomposition (EMD), or variational mode decomposition (VMD). The frequency band components characterize the signal components within different frequency ranges, while the residual components typically contain non-periodic or transient information in the signal that cannot be explained by the primary frequency band model, such as sudden noise or anomalous events.

[0165] Based on this, the energy distribution, statistical characteristics, and temporal correlation of the multiple frequency band components and the residual components are analyzed to finely distinguish different types of noise. For example, the energy of high-frequency noise is usually concentrated in the high-frequency components, while low-frequency drift noise is mainly reflected in the low-frequency components. By analyzing the statistical characteristics of each component (such as mean, variance, skewness, and kurtosis), Gaussian white noise, impulse noise, etc., can be distinguished. At the same time, temporal correlation analysis (such as autocorrelation function) can reveal the duration, periodicity, and degree of correlation with other signals of the noise.

[0166] The process involves detecting abnormal fluctuations in the residual components and correlating these fluctuations with changes in noise intensity across multiple frequency bands. The aim is to more accurately identify transient anomalies and compound noise. Abnormal fluctuations in the residual components may indicate sensor malfunctions, data transmission errors, or sudden industrial events. By correlating their occurrence time with changes in noise intensity in the corresponding frequency bands, it's possible to determine whether the abnormal fluctuation is related to the operating state of specific equipment or environmental changes, thus distinguishing between random noise and anomalies with specific physical significance.

[0167] Therefore, based on the analysis results, the types of composite noise and outlier characteristics present in the physical parameters are determined. This step is a comprehensive judgment of the foregoing analyses, aiming to identify multiple types of noise that may exist in the data (e.g., the simultaneous presence of high-frequency impulse noise and low-frequency drift noise) and the specific manifestations of outliers (e.g., whether they are instantaneous spikes, continuous drifts, or step changes).

[0168] The above technical solution enables accurate identification of complex noises and outliers in the physical parameters of coal mine industrial sites. Compared to traditional single-filtering methods, this solution effectively addresses common composite noises in coal mine environments (such as transient impulse noise caused by the start-up and shutdown of electromechanical equipment, and low-frequency drift noise caused by changes in rock stress) as well as various outliers (such as spikes or drops caused by sensor failures and data transmission errors). This refined identification capability significantly improves the accuracy and robustness of data preprocessing, providing high-quality input data for subsequent causal chain analysis and event causal graph construction, thereby enhancing the reliability and effectiveness of the data management method for evaluating the maturity of digital transformation in the entire coal industry.

[0169] In some embodiments, the step of selecting a filtering method for the physical parameters based on the identified noise type and outlier characteristics includes:

[0170] Identify transient pulse noise caused by the start-up and shutdown of electromechanical equipment and low-frequency drift noise caused by changes in rock stress;

[0171] A cascaded adaptive filter is used to suppress interference components in different frequency bands in stages;

[0172] The signal fidelity is monitored in real time during the filtering process, and the filtering strategy is reorganized when a fault characteristic waveform of the equipment is detected.

[0173] Specifically, identifying transient impulse noise caused by the start-up and shutdown of electromechanical equipment and low-frequency drift noise caused by changes in rock stress involves in-depth analysis of physical parameter data collected from industrial sites to distinguish noise components with different sources and characteristics. For example, the sudden start-up or shutdown of electromechanical equipment generates short-duration, high-amplitude transient impulse noise, while slow changes in rock stress or geological activity may lead to long-term, slow low-frequency drift noise in physical parameters. Accurately identifying these noise types is the foundation for subsequently selecting targeted filtering strategies.

[0174] The use of cascaded adaptive filters to suppress interference components in different frequency bands in stages can be understood as designing and applying a multi-stage, dynamically adjustable filter structure based on the identified noise type. For example, for transient impulse noise, median filtering and wavelet thresholding can be used for initial suppression; for low-frequency drift noise, high-pass filtering or adaptive Kalman filtering can be used. The cascaded structure allows different filters to function for specific frequency bands or noise characteristics, while the adaptive nature allows the filter parameters to be adjusted according to real-time signal characteristics to achieve optimal suppression. The aim is to achieve refined and efficient removal of complex composite noise.

[0175] In practical applications, signal fidelity is monitored in real time during the filtering process. When a fault characteristic waveform is detected, the filtering strategy is reorganized. Specifically, this means that while processing the data, the quality of the filtered data and the retention of effective information are continuously evaluated. Signal fidelity can be measured by comparing the energy, waveform similarity, or integrity of specific features of the signals before and after filtering. When a preset fault characteristic waveform is detected in the filtered data—for example, some sensor data exhibiting abnormal periodic fluctuations or abrupt changes under specific operating conditions—this may indicate that the current filtering strategy may have mistakenly affected important fault signals or failed to effectively suppress fault-related noise. In this case, the system will trigger a reorganization of the filtering strategy, such as adjusting filter parameters, switching filtering algorithms, or temporarily bypassing certain filtering stages, to ensure that fault characteristics can be accurately retained or extracted, avoiding the loss of key information due to over-filtering. The goal is to remove noise while maximizing the retention of effective information in the original signal, especially key features related to equipment status and fault diagnosis.

[0176] Through the above technical solution, this application can more accurately identify and process various complex noises in coal mine industrial sites, such as transient pulse noise caused by the start-up and shutdown of electromechanical equipment and low-frequency drift noise caused by changes in rock stress. By employing a cascaded adaptive filter to suppress interference components in different frequency bands in stages, the efficiency and effectiveness of noise removal are significantly improved, resulting in cleaner physical parameter data after processing. More importantly, by monitoring signal fidelity in real time and triggering filter strategy reconfiguration when equipment fault characteristic waveforms are detected, this application ensures that while effectively suppressing noise, it retains the key information in the original signal to the maximum extent, especially features related to equipment operating status and potential faults, avoiding the problem of losing important information due to over-filtering. This not only improves the robustness and adaptability of data preprocessing but also provides a higher quality and more reliable data foundation for subsequent event causal analysis and digital transformation maturity evaluation.

[0177] like Figure 2 As shown, an exemplary data management system for assessing the maturity of digital transformation in the coal industry is illustrated. A specific embodiment of this application discloses a data management system 100 for assessing the maturity of digital transformation in the coal industry, which includes:

[0178] The identifier generation module 10 is used to generate an event identifier containing a preset causal chain for each event in response to the raw event data collected in the industrial field. The preset causal chain is predefined by the event type association rule. The raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operation status sensors. The data records include the occurrence time, and any one of the event type and related parameter values.

[0179] The data transmission module 20 is used to actively push the event identifier of each event to the associated receiving device through the industrial network and trigger the receiving device to execute a preset action;

[0180] The causal construction module 30 is used to construct an event causal chain based on the preset causal chain information contained in the event identifier;

[0181] The time adjustment module 40 is used to, during the construction of the event causal graph, detect a conflict between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, and then correct the physical timestamp of the conflicting event according to the causal logic to obtain a logical calibration timestamp.

[0182] Therefore, the system in this application not only solves the technical challenge of data time consistency in the digital transformation of the coal industry, but also significantly improves the accuracy and reliability of data analysis by introducing a causal logic-driven time calibration mechanism. This system can provide a more realistic and reliable data foundation for evaluating the maturity of digital transformation, thereby providing stronger support for corporate strategic decision-making. Its technological contribution and innovation are self-evident.

[0183] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A data management method for evaluating the maturity of digital transformation in the coal industry, characterized in that, include: In response to raw event data collected from the industrial site, an event identifier containing a preset causal chain is generated for each event, wherein the preset causal chain is generated based on event type association rules; the event identifier of each event is actively pushed to the associated receiving device through the industrial network, and the receiving device is triggered to execute a preset action; Based on the preset causal chain information contained in the event identifier, an event causal graph is constructed; During the construction of the event causal graph, if a conflict is detected between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, the physical timestamp of the conflicting event is corrected according to the causal logic to obtain a logical calibration timestamp. The raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operating status sensors. The data records include the occurrence time, event type, and any one of the relevant parameter values. Prior to the step of generating an event identifier containing a preset causal chain for each event in response to raw event data collected in the industrial field, the method further includes: It receives raw event data from sensor units in the industrial field, and for each sensor unit, it performs spatiotemporal correlation aggregation on all raw event data generated by the same sensor unit within a sliding time window; For each event within the aggregation time window, the corresponding physical parameters are extracted, and the instantaneous change amplitude and rate of change of these physical parameters are calculated to obtain the instantaneous change characteristics; Based on the instantaneous change characteristics, key events with strong causal potential are identified. Combined with the change rate, the causal transmission efficiency of key events to related events is quantified. And based on the causal potential and transmission efficiency, a preset causal chain is generated. Furthermore, the lead-lag relationship and inter-correlation degree between the changes of different physical parameters within the same aggregation time window are analyzed to obtain the temporal correlation between physical parameters; The step of detecting a conflict between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, and then correcting the physical timestamp of the conflicting event according to the causal logic to obtain a logical calibration timestamp includes: When the preset causal chain description in the original event data is ambiguous or contradictory, each potential cause event is dynamically assigned a causal weight based on the instantaneous change characteristics and the temporal correlation. Identify the potential causal event with the highest causal weight as the dominant cause, and correct the physical timestamp of the conflicting event according to the dominant cause to obtain the logically calibrated timestamp; When the preset causal chain description in the original event data is ambiguous or contradictory, the step of dynamically assigning causal weights to each potential cause event based on the instantaneous change characteristics and the temporal correlation includes: Real-time acquisition of the linkage response mode of multiple physical parameters associated with the potential causal events; Compare the differences between the aforementioned linkage response modes; Based on the aforementioned differences, identify the dominant causal event; Based on the identified dominant cause event, each potential cause event is dynamically assigned a causal weight.

2. The data management method for evaluating the maturity of digital transformation in the coal industry according to claim 1, characterized in that, When the preset causal chain description in the original event data is ambiguous or contradictory, the step of dynamically assigning causal weights to each potential cause event based on the instantaneous change characteristics and the temporal correlation includes: Real-time acquisition of current operating environment parameters; Based on the current operating environment parameters, adjust the baseline value and change threshold used to calculate the instantaneous change characteristics, and adjust the time lag range and correlation threshold used to analyze the time series correlation. Based on the adjusted baseline value, change threshold, time lag range, and correlation threshold, the instantaneous change characteristics of the physical parameters and the temporal correlation are calculated. Based on the instantaneous change characteristics and the temporal correlation, a causal weight is dynamically assigned to each potential cause event.

3. The data management method for evaluating the maturity of digital transformation in the coal industry according to claim 1, characterized in that, The step of acquiring the linkage response mode of multiple physical parameters associated with the potential causal events in real time includes: The physical parameters are filtered for coal mine impulse noise to obtain processed physical parameter data. The processed physical parameter data is time-aligned to obtain time-aligned physical parameter data; According to the preset linkage rules, extract the physical parameter sequence associated with the potential cause event from the time-aligned physical parameter data; Based on the physical parameter sequence, a linkage mode map of the coordinated response of the equipment group is constructed.

4. The data management method for evaluating the maturity of digital transformation in the coal industry according to claim 3, characterized in that, The step of filtering the physical parameters for coal mine impulse noise to obtain processed physical parameter data includes: Identify the noise type and outlier characteristics of the physical parameters; Based on the identified noise type and outlier characteristics, a filtering method is selected for the physical parameters; The selected filtering methods are combined to filter the physical parameters, resulting in processed physical parameter data.

5. The data management method for evaluating the maturity of digital transformation in the coal industry according to claim 4, characterized in that, The steps for identifying the noise type and outlier characteristics of the physical parameters include: The physical parameters are decomposed into multiple frequency band components and residual components; The energy distribution, statistical characteristics, and temporal correlation of the multiple frequency band components and the residual components are analyzed to distinguish different types of noise. The system detects whether there are abnormal fluctuations in the residual components, and performs correlation analysis between the time of occurrence of the abnormal fluctuations and the noise intensity changes in the multiple frequency band components to obtain the analysis results. Based on the analysis results, the types of composite noise and outlier characteristics present in the physical parameters are determined.

6. The data management method for evaluating the maturity of digital transformation in the coal industry according to claim 4, characterized in that, The step of selecting a filtering method for the physical parameters based on the identified noise type and outlier characteristics includes: Identify transient pulse noise caused by the start-up and shutdown of electromechanical equipment and low-frequency drift noise caused by changes in rock stress; A cascaded adaptive filter is used to suppress interference components in different frequency bands in stages; The signal fidelity is monitored in real time during the filtering process, and the filtering strategy is reorganized when a fault characteristic waveform of the equipment is detected.

7. A data management system for evaluating the maturity of digital transformation in the coal industry, characterized in that, The system includes: The identifier generation module is used to generate an event identifier containing a preset causal chain for each event in response to raw event data collected in the industrial field. The preset causal chain is generated based on event type association rules. The raw event data includes at least unprocessed data records collected by gas sensors, temperature sensors, pressure sensors, and belt conveyor operation status sensors. The data records include the occurrence time, and any one of the event type and related parameter values. The data transmission module is used to actively push the event identifier of each event to the associated receiving device through the industrial network and trigger the receiving device to execute preset actions; The causal construction module is used to construct an event causal graph based on the preset causal chain information contained in the event identifier; The time adjustment module is used to, during the construction of the event causal graph, detect a conflict between the physical timestamp in the original event data and the causal logic indicated by the preset causal chain information, and then correct the physical timestamp of the conflicting event according to the causal logic to obtain a logical calibration timestamp. Before generating an event identifier containing a preset causal chain for each event in response to raw event data collected in the industrial field, the method further includes: It receives raw event data from sensor units in the industrial field, and for each sensor unit, it performs spatiotemporal correlation aggregation on all raw event data generated by the same sensor unit within a sliding time window; For each event within the aggregation time window, the corresponding physical parameters are extracted, and the instantaneous change amplitude and rate of change of these physical parameters are calculated to obtain the instantaneous change characteristics; Based on the instantaneous change characteristics, key events with strong causal potential are identified. Combined with the change rate, the causal transmission efficiency of key events to related events is quantified. And based on the causal potential and transmission efficiency, a preset causal chain is generated. Furthermore, the lead-lag relationship and inter-correlation degree between the changes of different physical parameters within the same aggregation time window are analyzed to obtain the temporal correlation between physical parameters; The time adjustment module is also used for: When the preset causal chain description in the original event data is ambiguous or contradictory, each potential cause event is dynamically assigned a causal weight based on the instantaneous change characteristics and the temporal correlation. Identify the potential causal event with the highest causal weight as the dominant cause, and correct the physical timestamp of the conflicting event according to the dominant cause to obtain the logically calibrated timestamp; When the preset causal chain description in the original event data is ambiguous or contradictory, the dynamic assignment of causal weights to each potential cause event based on the instantaneous change characteristics and the temporal correlation includes: Real-time acquisition of the linkage response mode of multiple physical parameters associated with the potential causal events; Compare the differences between the aforementioned linkage response modes; Based on the aforementioned differences, identify the dominant causal event; Based on the identified dominant cause event, each potential cause event is dynamically assigned a causal weight.

Images