Method and system for detecting abnormal impact of hydraulic oil cylinder based on working condition clustering

Abstract

The invention relates to the technical field of hydraulic oil cylinder detection, in particular to a hydraulic oil cylinder abnormal impact detection method and system based on working condition clustering. The method comprises the steps that a hydraulic oil cylinder working condition information set is obtained, and based on the hydraulic oil cylinder working condition information set, working characteristics of different loads and action frequencies of an advancing jack in underground coal mining circulation are analyzed, and a working condition clustering information set is obtained; on the basis of the working condition clustering information set, the correlation mutual influence between abnormal impact and valve element movement nonlinearity, coal slime clamping stagnation and multi-cylinder synchronous deviation is analyzed, and a fault risk information set is obtained; and based on the fault risk information set, incentives and fault states of abnormal impact under different working conditions are analyzed, and a hydraulic oil cylinder abnormal impact detection log is generated and output. Impact inducements and fault states under different working conditions are accurately positioned, active detection and accurate study and judgment of abnormal impact are achieved, and equipment faults are avoided in advance.

Description

Technical Field

[0001] This application relates to the field of hydraulic cylinder testing technology, and in particular to a method and system for detecting abnormal impacts in hydraulic cylinders based on working condition clustering. Background Technology

[0002] Currently, abnormal impact detection of hydraulic cylinders mostly relies on threshold judgment and simple time series analysis of pressure, flow or vibration signals. However, in the underground coal mining environment, the load and operating frequency of hydraulic cylinders change dynamically with the coal mining cycle. Single signal threshold detection is difficult to distinguish between normal fluctuations in working conditions and real abnormal impacts. Furthermore, it rarely considers the coupling effects between factors such as valve core nonlinearity, coal slime jamming and multi-cylinder synchronization deviation, resulting in a high false alarm rate and ambiguous cause location in the detection results, making it difficult to meet the needs of accurate diagnosis under complex working conditions.

[0003] Existing detection methods fail to effectively link the multi-dimensional relationship between the dynamic working conditions of hydraulic cylinders and abnormal impacts. Especially in underground coal mining cycles, the load and operating frequency of the jacks exhibit periodic changes. Abnormal impacts are often caused by multiple factors, such as nonlinear valve core movement, coal slurry jamming, and multi-cylinder synchronization deviation, either individually or in combination. It is difficult to quantify the independent and coupled effects of multiple fault causes, resulting in insufficient accuracy in abnormal impact detection and difficulty in fault tracing, which limits the effectiveness of hydraulic cylinder condition assessment and maintenance decisions. Summary of the Invention

[0004] This application provides a method and system for detecting abnormal impacts in hydraulic cylinders based on working condition clustering, in order to solve the above-mentioned problems.

[0005] Firstly, this application provides a method for detecting abnormal impacts in hydraulic cylinders based on working condition clustering. The method includes: acquiring a set of hydraulic cylinder working condition information; analyzing the working characteristics of a jack under different loads and operating frequencies during underground coal mining cycles based on the hydraulic cylinder working condition information set to obtain a set of working condition clustering information; analyzing the correlation and mutual influence between abnormal impacts and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation based on the working condition clustering information set to obtain a set of fault risk information; and analyzing the causes and fault states of abnormal impacts under different working conditions based on the fault risk information set to generate and output a hydraulic cylinder abnormal impact detection log.

[0006] By utilizing the above technical solutions, the operating characteristics of hydraulic cylinders can be accurately analyzed through working condition clustering. This clarifies the correlation between abnormal impacts and valve core nonlinearity, coal slurry jamming, and multi-cylinder deviation. It can accurately locate the impact causes and fault states under different working conditions, enabling proactive detection and accurate judgment of abnormal impacts. This allows for early avoidance of equipment failures, reduces the operation and maintenance costs of underground coal mining equipment, minimizes downtime losses, ensures continuous and efficient coal mining operations, and provides a scientific basis for the operation, maintenance, and performance optimization of hydraulic cylinders.

[0007] Optionally, the step of analyzing the working characteristics of the pushing jack under different loads and operating frequencies in the underground coal mining cycle based on the hydraulic cylinder working condition information set to obtain a working condition clustering information set includes: the hydraulic cylinder working condition information set includes a coal mining machine operation information set and a hydraulic cylinder status information set; based on the coal mining machine operation information set, analyzing the operating state of the coal mining machine at different positions to obtain a hydraulic cylinder action sequence corresponding to the operating state; based on the hydraulic cylinder status information set, analyzing the load changes and action frequency changes of the hydraulic cylinder in different working stages to obtain a load-frequency dynamic curve; based on the hydraulic cylinder action sequence, combined with the load-frequency dynamic curve, analyzing the repeated load and action frequency combinations of the hydraulic cylinder in the underground coal mining cycle to obtain the working condition clustering information set characterizing different working modes.

[0008] Optionally, the process of constructing the load-frequency dynamic curve includes: analyzing the pressure timing information and action triggering timing information of the hydraulic cylinder based on the hydraulic cylinder state information set; analyzing the periodic fluctuation characteristics of the driving pressure of the hydraulic cylinder when performing pushing, pulling, or lifting actions based on the pressure timing information, and obtaining load stage division information characterized by multiple pressure peak intervals; analyzing the density and interval pattern of control commands received by the hydraulic cylinder per unit time based on the action triggering timing information, and obtaining action frequency stage division information; and associating and matching the load characteristics and action frequency characteristics in the same time interval based on the load stage division information and the action frequency stage division information, to obtain the load-frequency dynamic curve that synchronously describes the changes in load level and action frequency with time as the horizontal axis.

[0009] Optionally, the step of analyzing the repeated load and action frequency combinations of the hydraulic cylinders in the underground coal mining cycle based on the hydraulic cylinder action sequence and the load-frequency dynamic curve to obtain the working condition clustering information set representing different working modes includes: based on the hydraulic cylinder action sequence, analyzing the fixed sequence and duration of the pushing conveyor, pulling frame, and bottom lifting actions within a single coal mining cycle to obtain a working action stage sequence; based on the load-frequency dynamic curve and the working action stage sequence, analyzing the range of load level variation and the concentration interval of action frequency within each action stage to obtain a load-frequency feature interval; based on the load-frequency feature interval, analyzing the recurring feature information of the load-frequency feature interval corresponding to the same action stage in multiple consecutive coal mining cycles, and grouping the recurring feature intervals into a load-frequency combination; based on the load-frequency combination, merging combinations with similar load levels and similar action frequencies to obtain the working condition clustering information set representing different working modes.

[0010] Optionally, the step of analyzing the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slurry jamming, and multi-cylinder synchronization deviation based on the working condition clustering information set to obtain a fault risk information set includes: based on the working condition clustering information set, analyzing the triggering correspondence between abnormal impact and valve core motion nonlinearity according to the motion response characteristics of the hydraulic cylinder valve core to obtain valve core nonlinearity-induced danger characteristic information; based on the working condition clustering information set, analyzing the hindrance-induced relationship between abnormal impact and coal slurry jamming according to the operation and coordination characteristics of the internal components of the hydraulic cylinder to obtain coal slurry jamming-induced danger characteristic information; based on the working condition clustering information set, analyzing the cooperative imbalance relationship between abnormal impact and multi-cylinder synchronization deviation according to the synchronous action execution characteristics of multiple hydraulic cylinders to obtain synchronization deviation-induced danger characteristic information; and based on the valve core nonlinearity-induced danger characteristic information, the coal slurry jamming-induced danger characteristic information, and the synchronization deviation-induced danger characteristic information, analyzing the independent and coupled influence laws of the three types of danger characteristics on abnormal impact to obtain the fault risk information set.

[0011] Optionally, the process of constructing the coal sludge jamming hazard characteristic information includes: based on the working condition clustering information set, combined with the load-frequency combination, analyzing the variation of the fitting clearance and smoothness of the internal components of the hydraulic cylinder under different working conditions to obtain component operation fitting deviation information; based on the component operation fitting deviation information, analyzing the accumulation and movement obstruction characteristics of coal sludge in the component fitting clearance to obtain coal sludge accumulation and obstruction characteristic information; based on the coal sludge accumulation and obstruction characteristic information, analyzing the correlation between the forced obstruction and impact triggering of coal sludge jamming on the hydraulic cylinder movement to obtain the coal sludge jamming hazard characteristic information.

[0012] Optionally, the process of constructing the synchronous deviation-induced danger characteristic information includes: based on the working condition clustering information set and combined with the load-frequency combination, analyzing the synchronous action start-up timing differences of multiple hydraulic cylinders under the same working condition mode to obtain multi-cylinder action timing deviation information; based on the multi-cylinder action timing deviation information, analyzing the stroke matching deviation of multiple hydraulic cylinders when performing push-pull, pull-frame, or bottom-lifting actions to obtain multi-cylinder stroke coordination deviation information; based on the multi-cylinder stroke coordination deviation information, analyzing the superposition characteristics of uneven hydraulic cylinder force and action jamming caused by synchronous deviation to obtain coordination imbalance-induced impact characteristic information; based on the coordination imbalance-induced impact characteristic information, analyzing the induction correlation law between coordination imbalance and abnormal impact to obtain the synchronous deviation-induced danger characteristic information.

[0013] Optionally, the step of analyzing the independent and coupled effects of the three types of risk-causing features on abnormal impacts based on the valve core nonlinearity risk-causing feature information, the coal slime jamming risk-causing feature information, and the synchronization deviation risk-causing feature information to obtain the fault risk information set includes: based on the valve core nonlinearity risk-causing feature information, the coal slime jamming risk-causing feature information, and the synchronization deviation risk-causing feature information, analyzing the correspondence between the hydraulic cylinder's own changes and the intensity and frequency of abnormal impacts when the other two types of risk-causing features are absent, to obtain independent risk-causing feature mapping information; based on the independent risk-causing feature mapping information, analyzing the amplification or induction law of the interaction between the two types of risk-causing features when two or three types of risk-causing features occur simultaneously, to obtain coupled risk-causing evolution information; based on the independent risk-causing feature mapping information, combined with the coupled risk-causing evolution information, constructing the fault risk information set of the probability and severity of impacts caused by single and multiple causes under different working conditions.

[0014] Optionally, the step of analyzing the causes and fault states of abnormal impacts under different working conditions based on the fault risk information set, and generating and outputting the abnormal impact detection log of the hydraulic cylinder, includes: based on the fault risk information set, analyzing the abnormal impact information dominated by the valve core nonlinearity-induced danger characteristic information, the coal slime jamming-induced danger characteristic information, and the synchronization deviation-induced danger characteristic information under different working conditions characterized by the load-frequency combination, to obtain the dominant cause discrimination information under each working condition; based on the dominant cause discrimination information, combined with the independent danger characteristic mapping information and the coupled danger evolution information, analyzing the severity and duration of hydraulic cylinder action jamming, pressure sudden change, and stroke deviation caused by single and coupled causes, to obtain fault state assessment information; based on the dominant cause discrimination information, combined with the fault state assessment information, according to the time sequence of the coal mining cycle, structurally associating and recording the abnormal impact causes identified under different working conditions, the corresponding fault state manifestations, and the corresponding load-frequency combinations, to generate and output the abnormal impact detection log of the hydraulic cylinder.

[0015] Secondly, this application provides a hydraulic cylinder abnormal impact detection system based on working condition clustering. The system includes: a clustering analysis module for acquiring a hydraulic cylinder working condition information set, and based on the hydraulic cylinder working condition information set, analyzing the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle to obtain a working condition clustering information set; a risk analysis module for analyzing the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation based on the working condition clustering information set to obtain a fault risk information set; and a cause status module for analyzing the causes and fault status of abnormal impact under different working conditions based on the fault risk information set, generating and outputting a hydraulic cylinder abnormal impact detection log. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of an application scenario provided in an embodiment of this application;

[0018] Figure 2 A flowchart illustrating a method for detecting abnormal impacts in hydraulic cylinders based on working condition clustering, provided as an embodiment of this application;

[0019] Figure 3 This is a schematic diagram of a hydraulic cylinder abnormal impact detection system based on working condition clustering, provided as an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0022] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0023] During the multi-condition operation of hydraulic cylinders in underground coal mining jacks, existing detection methods cannot establish a multi-dimensional correlation between the dynamic operating conditions of hydraulic cylinders and abnormal impacts. The operating conditions of jacks in underground coal mining change periodically, and multiple fault factors, either individually or in combination, can cause impacts. Because it is difficult to quantify the causal effects, the detection accuracy is insufficient and fault tracing is difficult, which restricts the effectiveness of cylinder condition assessment and maintenance.

[0024] Based on this, this application provides a method and system for detecting abnormal impacts in hydraulic cylinders based on working condition clustering. By analyzing the operating characteristics of hydraulic cylinders through working condition clustering, the correlation between abnormal impacts and valve core characteristics, coal slurry jamming, and multi-cylinder deviations is clarified, the cause of the fault is accurately located, and the impact is actively detected and predicted. This reduces the maintenance and downtime losses of underground coal mining equipment, ensures continuous operation, and provides a basis for cylinder optimization.

[0025] Figure 1 This application provides an application scenario diagram. During the multi-condition operation of the hydraulic cylinder of the underground coal mining jack, the method provided in this application is applied to analyze the operating characteristics of the hydraulic cylinder based on the working condition clustering, accurately identify the causes of abnormal impacts and actively warn of faults, effectively reduce the maintenance and downtime losses of underground equipment, provide a basis for the optimization of cylinder maintenance, and ensure continuous and efficient coal mining operations and stable equipment operation.

[0026] Specifically, the method provided in this application can be applied to any server. The server interacts with the hydraulic monitoring sensor to obtain the hydraulic cylinder operating condition information set provided by the hydraulic monitoring sensor, accurately locate the impact causes and fault states under different operating conditions, generate and output the hydraulic cylinder abnormal impact detection log to the equipment control personnel, and improve the overall operational stability of the coal mining equipment.

[0027] For specific implementation details, please refer to the following examples.

[0028] Figure 2 This is a flowchart illustrating a method for detecting abnormal impacts in hydraulic cylinders based on working condition clustering, as provided in one embodiment of this application. The method of this embodiment can be applied to servers in the above scenarios. Figure 2 As shown, the method includes:

[0029] S201. Obtain the hydraulic cylinder working condition information set. Based on the hydraulic cylinder working condition information set, analyze the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle, and obtain the working condition clustering information set.

[0030] The hydraulic cylinder operating condition information set can be a collection of parameters reflecting the characteristics of various operating conditions of the hydraulic cylinder during underground coal mining operations, with hydraulic monitoring sensors as the data source. The push jack can be a core application component of the hydraulic cylinder in underground coal mining equipment. The operating condition clustering information set can be a collection containing different operating condition types, operating characteristic parameters under each condition, and operating condition classification results.

[0031] Specifically, in the underground coal mining environment, the hydraulic cylinder, as the core component of the push jack, is significantly affected by the load and operating frequency. However, existing abnormal impact detection methods often ignore changes in operating conditions, resulting in insufficient detection accuracy. Therefore, it is necessary to identify different working modes through working condition clustering. By using cluster analysis, the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle can be analyzed, and similar features can be aggregated into classes to obtain a working condition clustering information set.

[0032] S202. Based on the working condition clustering information set, analyze the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation to obtain the fault risk information set.

[0033] A fault risk information set can be a collection of information such as fault type, risk level, and impact degree under different correlations. Abnormal impact can be an unexpected pressure surge, mechanical impact, or other abnormal phenomenon that occurs during the operation of a hydraulic cylinder. Valve core motion nonlinearity can be the phenomenon where the movement trajectory and speed of the valve core inside the hydraulic cylinder deviate from the preset linear law. Coal slurry jamming can refer to the phenomenon where coal slurry generated during underground coal mining enters the transmission and connection parts of the hydraulic cylinder, causing component jamming and obstructed movement. Multi-cylinder synchronization deviation can refer to the deviation in the action rhythm, stroke, and load distribution of each cylinder when multiple sets of hydraulic cylinders work together in coal mining operations.

[0034] Specifically, abnormal impacts in underground coal mining are often caused by the combined effects of multiple factors. However, existing technologies lack a comprehensive analysis of the correlation and mutual influence of these factors, making it difficult to fully assess the risk of failure. Therefore, it is necessary to explore their complex relationships in order to achieve accurate diagnosis. By using association rule mining and causal analysis methods, and based on the working condition clustering information set, we can analyze the correlation and mutual influence between abnormal impacts and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation to obtain a failure risk information set.

[0035] S203. Based on the fault risk information set, analyze the causes and fault states of abnormal impacts under different working conditions, and generate and output the abnormal impact detection log of the hydraulic cylinder.

[0036] The abnormal impact detection log of a hydraulic cylinder can be a document or data file that records the abnormal impact detection results, cause analysis, and fault status assessment.

[0037] Specifically, the causes and fault states of abnormal impacts vary under different operating conditions, and existing methods often cannot provide structured reports, leading to delays in maintenance decisions. Therefore, it is necessary to generate detailed detection logs to assist in timely intervention. By using pattern recognition and state assessment methods, the causes and fault states of abnormal impacts under different operating conditions are analyzed based on the fault risk information set, and the analysis results are integrated into structured detection log output using log generation technology.

[0038] By utilizing the above technical solutions, the operating characteristics of hydraulic cylinders can be accurately analyzed through working condition clustering. This clarifies the correlation between abnormal impacts and valve core nonlinearity, coal slurry jamming, and multi-cylinder deviation. It can accurately locate the impact causes and fault states under different working conditions, enabling proactive detection and accurate judgment of abnormal impacts. This allows for early avoidance of equipment failures, reduces the operation and maintenance costs of underground coal mining equipment, minimizes downtime losses, ensures continuous and efficient coal mining operations, and provides a scientific basis for the operation, maintenance, and performance optimization of hydraulic cylinders.

[0039] In some embodiments, the hydraulic cylinder operating condition information set includes a coal mining machine operation information set and a hydraulic cylinder status information set. Based on the coal mining machine operation information set, the operating status of the coal mining machine at different positions is analyzed to obtain the hydraulic cylinder action sequence corresponding to the operating status. Based on the hydraulic cylinder status information set, the load change and action frequency change of the hydraulic cylinder at different working stages are analyzed to obtain the load-frequency dynamic curve. Based on the hydraulic cylinder action sequence and combined with the load-frequency dynamic curve, the repeated load and action frequency combinations of the hydraulic cylinder in the underground coal mining cycle are analyzed to obtain the operating condition clustering information set characterizing different working modes.

[0040] The coal mining machine operation information set can be a collection of information reflecting various operating parameters and states of the coal mining machine during underground operations. The hydraulic cylinder status information set can be a collection of information including various core status parameters of the hydraulic cylinder during operation, reflecting its own operating status. The coal mining machine operation status can be the specific working state and operating mode exhibited by the coal mining machine at different underground locations. The hydraulic cylinder action sequence can be a combination of the sequence and frequency of hydraulic cylinder actions executed according to specific logic, matching different operating states of the coal mining machine. The hydraulic cylinder working stage can be the specific working periods during which the hydraulic cylinder performs different operating actions in conjunction with the coal mining machine to complete the coal mining cycle. Load change can be the dynamic fluctuation of the load force borne by the drive end of the hydraulic cylinder at different working stages. Action frequency change can be the dynamic variation of the number of control commands received and executed by the hydraulic cylinder per unit time and the command interval. The load-frequency dynamic curve can be a characteristic curve with time as the horizontal axis, synchronously showing the trend of hydraulic cylinder load level and action frequency changes over time. A coal mining cycle can be defined as a complete work cycle in which the coal mining machine starts cutting from one end of the working face, completes a full set of actions to the other end, and then returns to the starting point. The combination of load and action frequency can be a characteristic combination formed by the load level range and the concentrated range of action frequency of the hydraulic cylinders during a certain working stage of the coal mining cycle. A working mode can be a typical and standardized working state formed by the hydraulic cylinders based on recurring combinations of load and action frequency.

[0041] Specifically, in the underground coal mining cycle, if the working characteristics of the hydraulic cylinders are not analyzed and clustered based on the hydraulic cylinder working condition information set, it will be impossible to distinguish the working conditions of different loads and frequencies, resulting in a lack of specificity in subsequent abnormal impact detection, easy to miss or misdetect, and causing hydraulic cylinder failure. To address the aforementioned issues: Firstly, relying on hydraulic monitoring sensors deployed on-site, and using MEMS piezoresistive pressure sensors and magnetostrictive displacement sensors installed at the jack's oil circuit inlet to collect pressure and displacement signals, combined with the solenoid valve drive signals of the electro-hydraulic control system, a set of hydraulic cylinder state information is constructed. After obtaining the raw data, multi-source information fusion and temporal pattern recognition technologies are applied. For the generation of action sequences, a process logic parser based on a finite state machine is designed by analyzing the timing data of the coal mining machine's position. When the coal mining machine coordinates enter the preset "push-pull trigger zone" (e.g., several meters from the scraper conveyor head), the parser outputs a "push-pull" action label and timestamp, thereby converting continuous position information into a discrete, sequentially fixed hydraulic cylinder action sequence. For the construction of the load-frequency dynamic curve, it is necessary to coordinate the processing of pressure and action trigger timing. A sliding time window technique is used to perform peak search and interval fitting within the window on the pressure signal (e.g., using the find_peaks function to identify pressure peaks and determining the main peak distribution intervals through clustering), thereby quantifying the load stage; simultaneously, in the same The instantaneous operating frequency is calculated by counting the number of rising edges of the solenoid valve drive signal within a time window. Then, by aligning timestamps, the "average load level" (e.g., a certain MPa range) and "average operating frequency" (e.g., times / minute) corresponding to each time window are used as two-dimensional coordinate points and connected in chronological order to generate a load-frequency dynamic curve. Finally, during working condition clustering, a complete coal mining cycle is divided into sub-stages such as "pushing the conveyor," "pulling the support," and "lifting the bottom" based on the action sequence. Within the time range of each sub-stage, the corresponding values ​​from the aforementioned dynamic curve are extracted. All load-frequency data points are then analyzed using the DBSCAN density clustering algorithm. The algorithm automatically identifies dense regions of points in the data space, and each dense region is considered a recurring "load-frequency combination" within that action phase. For example, in the "support pulling" phase data of dozens of coal mining cycles, the DBSCAN algorithm might identify a cluster of points centered on "medium load (e.g., 20-25 MPa), high frequency (e.g., 5-8 times / minute)." This cluster is defined as a typical "medium load, high frequency support pulling" working condition pattern. After traversing all action phases and merging similar patterns, the final structured working condition clustering information set is generated.

[0042] By using the above technical solutions, different working modes of hydraulic cylinders can be accurately divided, providing a clear working condition dimension for subsequent correlation analysis of abnormal impacts and failure factors, improving the pertinence and accuracy of impact detection, and ensuring the stable and safe operation of coal mining.

[0043] In some embodiments, based on the hydraulic cylinder state information set, the pressure timing information and action triggering timing information of the hydraulic cylinder are analyzed; based on the pressure timing information, the periodic fluctuation characteristics of the driving pressure of the hydraulic cylinder when performing push, pull, or lift actions are analyzed to obtain load stage segmentation information characterized by multiple pressure peak intervals; based on the action triggering timing information, the density and interval pattern of control commands received by the hydraulic cylinder per unit time are analyzed to obtain action frequency stage segmentation information; based on the load stage segmentation information and combined with the action frequency stage segmentation information, the load characteristics and action frequency characteristics in the same time interval are correlated and matched to obtain a load-frequency dynamic curve that synchronously describes the changes in load level and action frequency with time as the horizontal axis.

[0044] Pressure timing information can be a continuous data sequence formed by the change of driving pressure over time during the operation of the hydraulic cylinder. Action triggering timing information can be timing data composed of the trigger time and command type of control commands received by the hydraulic cylinder, such as pushing, pulling, and lifting. Load phase segmentation information can be information on different load intensity phases characterized by multiple pressure peak intervals, based on the periodic fluctuation characteristics of the hydraulic cylinder's driving pressure. Pressure peak intervals can be the specific pressure value range where the peak driving pressure occurs when the hydraulic cylinder performs a specific action. Action frequency phase segmentation information can be information on different action frequency intensity phases, based on the density and interval pattern of control commands received by the hydraulic cylinder per unit time. Load characteristics can be various characteristic information that reflects the magnitude and fluctuation pattern of the hydraulic cylinder's working load. Action frequency characteristics can be various characteristic information that reflects the density and interval pattern of the hydraulic cylinder's action triggering.

[0045] Specifically, in the process of analyzing the working characteristics of hydraulic cylinders during underground coal mining cycles, without constructing a load-frequency dynamic curve, it is impossible to accurately integrate load and action frequency data. This leads to a one-sided analysis of working condition characteristics, a lack of reliable basis for subsequent clustering, and makes the investigation of abnormal impact causes deviate from actual working conditions, significantly reducing detection accuracy. To address these issues, industrial-grade piezoelectric pressure sensors (e.g., Kistler's series of sensors) are used to collect analog pressure signals from the working chamber of the cylinder. Anti-aliasing filtering and noise reduction are then performed using a signal conditioning module and a real-time digital filter based on a field-programmable gate array (FPGA) (such as an FIR low-pass filter designed using the window function method) to obtain high-fidelity pressure timing information. Simultaneously, by analyzing the CAN bus communication protocol of the hydraulic support electro-hydraulic control system (such as the German MARCO or Tianjin Huaning system), the switching command messages of the control solenoid valves are directly read. A high-precision timing module (such as GPS or the IEEE 1588 protocol) is used to assign a unified timestamp to each command, thereby obtaining accurate action trigger timing information. In the feature extraction stage, the preprocessed pressure data is processed using a peak detection algorithm based on a sliding window (e.g., using the find_peaks function and setting minimum peak height and distance thresholds) to automatically identify the pressure peak points in each work cycle. Then, clustering algorithms in unsupervised machine learning (e.g., using the DBSCAN density clustering algorithm in the scikit-learn library) are used to perform cluster analysis on these peak points, classifying peaks with similar pressure values ​​to divide stages representing different load levels such as "heavy load advancement" and "medium load adjustment". For action trigger sequences, a sliding window of fixed duration (e.g., 1 second) is used for counting, and the trigger frequency within the window is calculated. Then, a threshold is set using statistical methods (e.g., calculating the percentile of the frequency distribution) to divide the continuous time into action stages such as "high frequency", "medium frequency", and "low frequency". Finally, using timestamps as the unique alignment key, the load phase and the action frequency phase defined above are precisely matched and merged. Through data visualization techniques (such as using Python's Matplotlib library to draw a dual Y-axis time series plot), a load-frequency dynamic curve is generated. The horizontal axis of this curve is absolute time, the left vertical axis is pressure value (MPa), and the right vertical axis is instantaneous frequency (Hz), which intuitively and quantitatively presents the synchronous evolution and coupling relationship between the two in the coal mining cycle.

[0046] Through the above technical solution, the time dimension data of hydraulic cylinder load and action frequency are accurately integrated, clearly presenting its dynamic coupling characteristics, providing accurate data support for working condition clustering, and making subsequent impact detection fit the actual working scenario.

[0047] In some embodiments, based on the hydraulic cylinder action sequence, the fixed sequence and duration of the pushing conveyor, pulling frame, and lifting bottom actions within a single coal mining cycle are analyzed to obtain the working action stage sequence; based on the load-frequency dynamic curve, combined with the working action stage sequence, the range of load level variation and the concentration interval of action frequency within each action stage are analyzed to obtain the load-frequency characteristic interval; based on the load-frequency characteristic interval, the recurring characteristic information of the load-frequency characteristic interval corresponding to the same action stage in multiple consecutive coal mining cycles is analyzed, and the recurring characteristic intervals are grouped into a load-frequency combination; based on the load-frequency combination, combinations with similar load levels and similar action frequencies are merged to obtain a working condition clustering information set representing different working modes.

[0048] A coal mining cycle can be considered the complete process in underground coal mining, from the start-up of the coal mining machine to the completion of a series of continuous actions such as coal mining, conveyor pushing, support pulling, and floor lifting in a working face, and then returning to the initial working state. The load-frequency characteristic range can be the specific range of load level variation and the concentrated distribution range of the operating frequency of a hydraulic cylinder within a single action phase. A load-frequency combination can be the set of load-frequency characteristic ranges that repeatedly occur within the same action phase in multiple consecutive coal mining cycles.

[0049] Specifically, in the underground coal mining cycle, if the hydraulic cylinder action sequence and load-frequency dynamic curve are not combined to analyze the repeated load-frequency combination, it will be impossible to classify the working mode, resulting in the subsequent impact cause analysis being untargeted, the fault location deviation being large, the abnormal impact of the hydraulic cylinder occurring frequently, affecting the operation of coal mining equipment and even causing underground operation safety accidents. To address the aforementioned issues: First, for the hydraulic cylinder action sequence, a rule-based timing segmentation technique is employed. Specifically, preset action command codes (e.g., "0xA1" representing the start of the push conveyor, "0xA2" representing the end of the push conveyor) are used as boundary markers. Combined with a time window smoothing algorithm, the "working action stage sequence" within a single coal mining cycle is accurately parsed from continuous controller communication messages, clearly defining the start and end timestamps of each push conveyor, pull frame, or bottom-lifting action. Next, this sequence is spatiotemporally aligned with the load-frequency dynamic curve. For each action stage identified in the sequence, sliding window statistical and distribution analysis methods are applied. For example, a box plot of the pressure sampling values ​​within that time period is calculated to determine the robust variation range of the load (e.g., the interval between the 25th and 75th percentiles). Simultaneously, by calculating the reciprocal of the time interval between action trigger events and obtaining the mode, the generation of that stage is determined. The system identifies key frequency intervals for representative actions, generating a series of load-frequency feature intervals. Then, in the pattern mining stage, a sequence matching algorithm based on Dynamic Time Warping (DTW) or cosine similarity is used to retrieve all feature intervals for the same type of action phase (e.g., the "first frame-pulling phase") from a historical database spanning multiple coal mining cycles. The similarity between these intervals is calculated, and intervals meeting a preset similarity threshold (e.g., DTW distance less than a set value) are aggregated into a load-frequency combination. Finally, an unsupervised clustering model is used, for example, using the median of the load interval and the median of the frequency interval for each combination as a two-dimensional feature vector. This is input into a K-means clustering algorithm for automatic classification, merging multiple combinations that are geographically close in the feature vector space into the same cluster. This systematically outputs a set of working condition clustering information representing different working modes such as "light load, high frequency" and "heavy load, low frequency."

[0050] The above technical solution accurately extracts and clusters recurring load-frequency combinations in the coal mining cycle, and classifies the working condition patterns with clear characteristics, providing accurate working condition basis for subsequent abnormal impact fault cause analysis, and improving the efficiency and accuracy of fault location.

[0051] In some embodiments, based on the working condition clustering information set, the triggering correspondence between abnormal impact and valve core motion nonlinearity is analyzed according to the motion response characteristics of the hydraulic cylinder valve core, thus obtaining valve core nonlinearity-induced danger characteristic information; based on the working condition clustering information set, the hindrance-induced relationship between abnormal impact and coal slurry jamming is analyzed according to the operation and coordination characteristics of the internal components of the hydraulic cylinder, thus obtaining coal slurry jamming-induced danger characteristic information; based on the working condition clustering information set, the cooperative imbalance relationship between abnormal impact and multi-cylinder synchronization deviation is analyzed according to the synchronous action execution characteristics of multiple hydraulic cylinders, thus obtaining synchronization deviation-induced danger characteristic information; based on the valve core nonlinearity-induced danger characteristic information, coal slurry jamming-induced danger characteristic information, and synchronization deviation-induced danger characteristic information, the independent and coupled influence laws of the three types of danger characteristics on abnormal impact are analyzed, thus obtaining a fault risk information set.

[0052] The nonlinearity-induced hazard characteristics of the valve core can be a set of characteristic information representing the triggering correspondence between the nonlinearity of valve core motion and abnormal impact. The coal slurry jamming-induced hazard characteristics can be a set of characteristic information representing the hindering and inducing relationship between coal slurry jamming and abnormal impact. The synchronization deviation-induced hazard characteristics can be a set of characteristic information representing the coordinated imbalance relationship between multi-cylinder synchronization deviation and abnormal impact. Independent influence can be the sole effect of a single hazard characteristic on abnormal impact of the hydraulic cylinder when there is no interference from the other two types of hazard characteristics. Coupled influence law can be the law of amplification, superposition, or induction of abnormal impact by the interaction relationship when two or three types of hazard characteristics occur simultaneously. Valve core motion response characteristics can be the response characteristics of the hydraulic cylinder valve core in terms of motion speed, stroke, and timing of action after receiving control commands. The operational coordination characteristics of internal components can be the characteristics of the clearance, smoothness of action, and linkage coordination of the internal components of the hydraulic cylinder during operation. Synchronous action execution characteristics can be the characteristics of the starting sequence, stroke matching, and action coordination when multiple hydraulic cylinders perform synchronous actions such as pushing and pulling.

[0053] Specifically, in underground coal mining cycles, if the correlation between abnormal impacts and three types of fault factors is not analyzed when hydraulic cylinders operate under different conditions, it can easily lead to misjudgment of fault causes and underestimation of impact risks, resulting in cylinder failure, equipment shutdown, and even coal mining operation interruption and underground production safety accidents. To address these issues: Regarding the nonlinear characteristics of valve core motion, we use signals from the valve core displacement LVDT sensor and the pilot pressure sensor to quantify its nonlinearity by calculating the area of ​​its response hysteresis loop and the linear fitting residual (e.g., using least squares fitting based on measured data and comparing it with the ideal linear response curve). We then perform time-series correlation analysis between this index and the pressure impact peak value collected under this condition (e.g., using a sliding window cross-correlation algorithm). When a stable mode is identified where the increased nonlinearity (e.g., a 30% increase in hysteresis loop area) precedes the pressure peak value (e.g., 50-200 milliseconds earlier), it is considered stable. This is transformed into a "valve core nonlinearity-induced hazard characteristic information." Targeting the coal slime jamming characteristic, online oil particle count data (monitoring particle concentration greater than 5μm) is integrated with cylinder operation stability indicators (such as the variance of the piston rod speed signal). When the particle concentration shows an upward trend and the speed variance increases abnormally, pattern recognition is triggered. Short-time Fourier transform analysis is used to analyze the low-frequency energy changes of the pressure signal during the action initiation phase. If abnormal energy accumulation occurs in a specific frequency band (such as 2-10Hz), it is determined to be a "sticky-slippery motion" characteristic caused by coal slime obstruction. This constructs a coal slime jamming characteristic strongly correlated with the degree of accumulation and action jamming. Regarding the risk-causing characteristic information for multi-cylinder synchronization deviation, the key technologies lie in high-precision timing alignment and stroke curve similarity calculation. Utilizing the synchronous clock of the distributed data acquisition unit, the trigger time of the multi-cylinder action command is aligned with the start time of the displacement sensor, calculating their start-up delay difference. A dynamic time warping algorithm is then used to match subsequent stroke curves, calculating the real-time stroke deviation curve. Statistical analysis reveals that when the standard deviation of the multi-cylinder stroke deviation consistently exceeds a threshold (e.g., 3% of a single propulsion stroke), the high-frequency component energy of the main system pressure fluctuation significantly increases. This correlation is extracted as the risk-causing characteristic information of synchronization deviation. To analyze the coupling effects, the quantitative values ​​of features extracted in real time from the above three channels (such as nonlinearity index, jamming confidence, and synchronization deviation coefficient) are used as input vectors. A lightweight gradient boosting decision tree model is trained using historical fault data. It can not only output the overall risk probability of the impact under the current feature combination, but also quantify and reveal the coupling relationship by analyzing the importance and interaction effects of features. For example, the model may determine that when "high synchronization deviation" and "medium-level coal slime jamming" coexist, their joint risk contribution is much higher than the sum of their independent risks, thus accurately characterizing such coupling patterns in the fault risk information set.

[0054] Through the above technical solutions, the correlation between the three types of fault factors and abnormal impacts can be accurately explored, and the characteristics of single-factor and coupled hazards can be clarified, providing a scientific basis for subsequent impact cause identification and fault status assessment.

[0055] In some embodiments, based on the working condition clustering information set and combined with the load-frequency combination, the characteristics of the fit clearance change and smoothness of the internal components of the hydraulic cylinder under different working conditions are analyzed to obtain the component operation fit deviation information; based on the component operation fit deviation information, the characteristics of coal slime accumulation and motion obstruction in the component fit clearance are analyzed to obtain coal slime accumulation obstruction characteristic information; based on the coal slime accumulation obstruction characteristic information, the correlation between coal slime jamming and the forced obstruction and impact triggering of hydraulic cylinder movement is analyzed to obtain coal slime jamming hazard characteristic information.

[0056] The component operation and fit deviation information can be the characteristic information of the change in fit clearance and smoothness of operation of the internal components of the hydraulic cylinder under different working conditions. The coal slime accumulation and obstruction characteristic information can be the accumulation and buildup pattern of coal slime in the fit clearance of the internal components of the hydraulic cylinder and the characteristic information of its obstruction of component movement.

[0057] Specifically, in underground coal mining operations, without analyzing the characteristics of coal slime jamming that could lead to accidents, it's impossible to clearly define its correlation with abnormal impacts. This can result in misjudging the causes of hydraulic cylinder failures, allowing the impacts caused by jamming to persist, exacerbating wear on cylinder components, and reducing equipment operational stability. To address these issues: First, dynamic parameter identification and signal feature extraction techniques are applied. Specifically, by analyzing the high-frequency pressure pulsation spectrum of the cylinder under specific load-frequency combinations (e.g., using Fast Fourier Transform to analyze energy changes in the 5-200Hz frequency band) and the smoothness index of the velocity curve (e.g., calculating the acceleration abrupt change point during the action), the equivalent dynamic fit clearance changes of key internal kinematic pairs (e.g., piston rod and guide sleeve) are inverted and inferred. Furthermore, a machine learning-based classifier (e.g., Support Vector Machine, SVM) is used to classify states such as "smooth," "slightly hindered," and "significantly jammed" under various operating conditions, thereby establishing a database of component operation fit deviations. Second, a computational fluid dynamics-discrete element coupled simulation model is introduced, using the identified abnormal clearance parameters as boundary conditions, combined with underground coal slime samples... Measured physical properties (such as particle size distribution (0.05-0.2 mm) and adhesion coefficient) are used to simulate the transport and deposition process of coal slime-oil two-phase flow within a specific gap. The accumulation growth curve and local resistance force prediction value are quantitatively output for different operating conditions and operating times, generating coal slime accumulation resistance characteristic information. Finally, a combination of fault tree analysis and data-driven correlation modeling is used to perform correlation mining between the simulated resistance force growth curve and the historical pressure peak time series data collected in the cylinder action cycle (e.g., applying Granger causality test or constructing a long short-term memory network LSTM prediction model). This establishes a correlation model of "accumulation degree - resistance force - impact intensity / probability", accurately depicting the critical conditions and risk evolution law of coal slime jamming triggering impact, and finally outputting structured coal slime jamming risk characteristic information.

[0058] The above technical solutions accurately uncover the patterns of abnormal impacts triggered by coal slime jamming, providing precise evidence for fault risk analysis, effectively identifying jamming-induced dangerous conditions, and reducing cylinder impact damage.

[0059] In some embodiments, based on the working condition clustering information set and combined with the load-frequency combination, the timing differences of the synchronous action start-up of multiple hydraulic cylinders under the same working condition mode are analyzed to obtain multi-cylinder action timing deviation information; based on the multi-cylinder action timing deviation information, the stroke matching deviation of multiple hydraulic cylinders when performing push-pull, pull-frame, or bottom-lifting actions is analyzed to obtain multi-cylinder stroke coordination deviation information; based on the multi-cylinder stroke coordination deviation information, the superposition characteristics of uneven hydraulic cylinder force and action jamming caused by synchronization deviation are analyzed to obtain coordination imbalance-induced impact characteristic information; based on the coordination imbalance-induced impact characteristic information, the induction correlation law between coordination imbalance and abnormal impact is analyzed to obtain synchronization deviation-induced danger characteristic information.

[0060] Multiple hydraulic cylinders can refer to multiple sets of hydraulic cylinder actuators that work together in underground coal mining to complete actions such as pushing conveyors, pulling supports, and raising the bottom. The same working condition mode can refer to multiple hydraulic cylinders operating in the same load-frequency combination. Synchronous action start-up timing differences can refer to the deviation in the timing of the start-up actions when multiple hydraulic cylinders perform the same coal mining action. Multi-cylinder action timing deviation information can be a set of information formed after quantifying and extracting features from the synchronous action start-up timing differences of multiple hydraulic cylinders. Pushing conveyors can be a core underground coal mining operation where hydraulic cylinders push the scraper conveyor forward. Pulling supports can be a core underground coal mining operation where hydraulic cylinders pull the hydraulic supports. Raising the bottom can be a core underground coal mining operation where hydraulic cylinders lift the base of the hydraulic support to adjust the support position. Stroke matching deviation can be the deviation between the actual stroke and the preset standard stroke, as well as the mutual deviation between the actual strokes of multiple cylinders when multiple hydraulic cylinders perform the same action. Multi-cylinder stroke coordination deviation information can be a set of information formed after statistically analyzing the characteristics of the stroke matching deviations of multiple hydraulic cylinders. Uneven force distribution can be a state where the load pressure borne by multiple hydraulic cylinders deviates from the preset equilibrium value due to synchronization deviation. The characteristic information of impact caused by coordinated imbalance can be a set of information formed after extracting the superposition law and features of uneven force distribution and motion jamming caused by synchronization deviation. Coordinated imbalance can be a state where the overall coordinated operation state of multiple hydraulic cylinders deviates from the preset standard due to deviations in action timing and stroke matching. The induced correlation law can be the inherent correspondence between the degree and manifestation of coordinated imbalance and the frequency, intensity, and duration of abnormal impacts.

[0061] Specifically, in the process of multi-cylinder coordinated operation in underground coal mining, if the characteristics of the danger caused by synchronization deviation are not analyzed, the superimposed hazards of uneven force and jamming caused by timing and stroke deviations cannot be identified, which will lead to inaccurate positioning of abnormal impact causes and frequent equipment failures. To address the above problems, by deploying industrial Ethernet modules supporting the IEEE 1588 Precision Time Protocol (PTP) to the command output terminal of each cylinder controller and high-response displacement sensors (such as magnetostrictive displacement sensors), it is ensured that all data points are marked with a uniform microsecond-level time stamp during acquisition, thereby accurately extracting the microsecond timing difference from the issuance of the command to the start of the cylinder action. For the multiple sets of displacement time series obtained in this way, in order to accurately quantify the stroke matching deviation, the Dynamic Time Warping (DTW) algorithm is adopted as a specific signal processing method. This algorithm constructs a cost matrix and finds the optimal bending path, which can adaptively stretch or compress the time axis, thereby effectively aligning the displacement curves of each cylinder with different action speeds due to different loads. Finally, it outputs a warped distance that does not depend on absolute time but only reflects the difference in relative position. This value directly quantifies the coordination deviation. To transform this geometric deviation into a mechanical risk indicator, a simplified multi-rigid-body static model is introduced as the analysis kernel. This model uses the geometric relationship of the support frame's CAD drawings as its topological foundation and takes the real-time collected displacements of each hydraulic cylinder as the driving input. Through the force and torque balance equations, it calculates the theoretical forces at each key hinge point of the support frame in real time. Once the model's calculation results show that the stress distribution at a certain point deviates significantly from the calculated value under the ideal coordinated state due to stroke deviation (e.g., the stress concentration factor exceeds a preset threshold), the state at that moment is immediately marked as a coordinated imbalance-induced impact characteristic. Finally, association rule learning techniques from data mining (e.g., ...) are used... The FP-Growth algorithm is used to perform batch analysis on historical feature markers from long-term operation and subsequent actual recorded impact events (determined by high-frequency pressure sensors). This process automatically discovers a strong statistical correlation between feature combination patterns such as "when the DTW distance between cylinders A and B is continuously greater than X millimeters, and the stress fluctuation variance at point C calculated by the model is higher than the Y value for N consecutive cycles" and the result "a sudden pressure change exceeding Z MPa occurs within the next M seconds". This kind of verified and interpretable "feature-result" rule chain is structured into the final synchronous deviation risk feature information database that can be directly used for online risk assessment.

[0062] Through the above technical solutions, the correlation between synchronization deviation and abnormal impact can be accurately explored, the risk-causing pattern of multi-cylinder coordinated imbalance can be clearly characterized, accurate data can be provided for fault risk analysis, and the pertinence of abnormal impact cause identification and prevention can be improved.

[0063] In some embodiments, based on the nonlinearity-causing characteristic information of the valve core, the slurry jamming-causing characteristic information, and the synchronization deviation-causing characteristic information, the correspondence between the hydraulic cylinder's own changes and the intensity and frequency of abnormal impacts is analyzed when the other two types of hazard-causing characteristics are absent, to obtain independent hazard-causing characteristic mapping information; based on the independent hazard-causing characteristic mapping information, the amplification or induction law of the interaction between two or three types of hazard-causing characteristics when they occur simultaneously is analyzed, to obtain coupled hazard-causing evolution information; based on the independent hazard-causing characteristic mapping information and combined with the coupled hazard-causing evolution information, a fault risk information set of the probability and severity of impacts caused by single and multiple causes under different working conditions is constructed.

[0064] Independent hazard-causing feature mapping information can be the correspondence between the hydraulic cylinder's own changes and the intensity and frequency of abnormal impacts when a single hazard-causing feature is free from other interferences. Coupled hazard-causing evolution information can be the amplification or induction pattern of abnormal impacts due to the interaction of two or three types of hazard-causing features when they occur simultaneously.

[0065] Specifically, during the multi-condition operation of hydraulic cylinders in underground coal mining jacks, without analyzing the independent and coupled effects of the three types of hazardous characteristics, it is impossible to accurately assess the impact risk, which can easily lead to equipment jamming, sudden pressure changes, and even cylinder damage and coal mining operation interruption. Furthermore, inaccurate risk assessment can cause maintenance delays. To address these issues, the approach begins with in-depth analysis of the pre-processed time-series data of the three types of hazardous characteristics (valve core motion response current curves, cylinder pressure fluctuation signals, and multi-cylinder displacement sensor data). To achieve independent impact analysis, correlation analysis and regression modeling techniques are employed. For example, using Spearman's rank correlation coefficient combined with piecewise linear regression, historical periods in the data caused by only a single factor (such as valve core nonlinearity) can be isolated. The mapping function between the characteristic indicators of this factor (such as valve core hysteresis area) and the impact intensity indicators (such as peak pressure) can be quantitatively analyzed, thereby generating an independent database of risk-causing feature mappings. For coupled impact analysis, a combination of Bayesian networks and Granger causality tests is introduced for interaction modeling. First, domain knowledge is used to construct an initial Bayesian network topology with three types of risk-causing features as nodes and impact events as endpoints. The system first constructs a structure; then, Granger causality tests are used to calculate the synchronously collected multidimensional time-series data. For example, it analyzes whether the time series of "coal slime jamming force estimate" significantly leads and affects the change of the "valve core command-response deviation" sequence, thereby verifying the causal direction and strength between nodes, and then updating the network's conditional probability table to quantitatively characterize how the occurrence of one factor changes the probability of another factor causing an impact. This process generates coupled risk evolution information. Based on the above information database and network model, Monte Carlo simulation and risk matrix assessment techniques are used to construct the final fault risk information set. For a given working condition cluster, the typical value range and statistical distribution of the three types of risk-causing features under that working condition are used as input to drive the constructed Bayesian network to perform simulation and deduction. The frequency of various fault combinations leading to impact events is statistically estimated as a probability estimate, and the expected severity is calculated by combining the impact intensity function in the independent mapping library. Finally, the results are mapped to a risk matrix with "probability-severity" as the axis, and a structured risk rating and a list of dominant cause combinations are output for each working condition mode, thereby completing the dynamic construction and updating of the fault risk information set.

[0066] Through the above technical solutions, the effects of the three types of risk-causing characteristics on abnormal impacts can be accurately clarified, and the impact risk under different working conditions can be scientifically quantified. This provides an accurate basis for subsequent cause identification and condition assessment, effectively reduces the probability of impacts, and ensures the stable operation of coal mining equipment.

[0067] In some embodiments, based on the fault risk information set, the abnormal impact information dominated by valve core nonlinearity-induced fault characteristics, coal slime jamming-induced fault characteristics, and synchronization deviation-induced fault characteristics under different load-frequency combinations is analyzed to obtain the dominant cause discrimination information under each working condition. Based on the dominant cause discrimination information, combined with independent fault characteristic mapping information and coupled fault evolution information, the severity and duration of hydraulic cylinder action jamming, pressure sudden change, and stroke deviation caused by single and coupled causes are analyzed to obtain fault state assessment information. Based on the dominant cause discrimination information and the fault state assessment information, the abnormal impact causes identified under different working conditions, the corresponding fault state manifestations, and the corresponding load-frequency combinations are structurally correlated and recorded according to the time sequence of the coal mining cycle to generate and output the hydraulic cylinder abnormal impact detection log.

[0068] The dominant cause identification information can be the identification information obtained from the analysis of one or more dominant abnormal impacts among the three types of risk-causing characteristics under different operating conditions. The fault state assessment information can be the assessment information obtained after analyzing the severity and duration of cylinder movement jamming, sudden pressure changes, and stroke deviations caused by single and coupled causes. Movement jamming can be a fault state manifestation of stagnation or unevenness when the hydraulic cylinder performs actions such as pushing, pulling, or lifting. Sudden pressure changes can be a fault state manifestation of a sudden increase or decrease in the internal driving pressure of the hydraulic cylinder within a short period of time. Stroke deviation can be a fault state manifestation of a difference between the actual stroke of the hydraulic cylinder and the preset stroke.

[0069] Specifically, in various operating conditions of underground coal mining cycles, if the impact causes and fault states are not analyzed, problems such as valve core nonlinearity cannot be accurately identified, leading to hydraulic cylinder jamming, frequent sudden pressure changes, and interruption of the coal mining cycle. To address these issues, the fault risk information set (including risk probability and severity matrices of various risk-causing characteristics under different load-frequency combinations) is transformed into business rules (Production) that can be recognized by the rule engine. Rules), for example, a rule can be defined as: "IF the operating condition mode is 'high load-low frequency propulsion' AND a specific high-frequency flutter component is detected in the real-time pressure fluctuation spectrum THEN increase the confidence of 'valve core nonlinearity' to 0.8 and mark it as a high-priority dominant cause candidate". The rule engine uses pattern matching and conflict resolution strategies to quickly infer the real-time collected operating condition signals and historical risk characteristics, and outputs the most likely dominant cause identification information. Secondly, in the fault state assessment stage, a real-time calculation variant of the fault mode, effect, and hazard analysis model is introduced. This model takes the dominant cause identification information as input and calls independent hazard feature mapping information (such as the mathematical model of "coal slime stagnation-pressure slow rise rate") and coupled hazard characteristics. The system processes information on fault evolution (such as a regression model of "valve core nonlinearity and synchronization deviation coupling - impact peak"). In terms of details, it calculates in real-time the hysteresis index of the current action lag (e.g., calculated through the lag time window of piston rod displacement and command), key indicators of sudden pressure changes (e.g., the peak value and duration exceeding a set threshold (e.g., 35 MPa), and the synchronization deviation rate of multi-cylinder strokes). These quantitative indicators are input into a pre-trained lightweight evaluation classifier (e.g., a gradient boosting tree-based classification model). This classifier outputs the severity level of the fault state (e.g., "mild," "moderate," "severe") and a description of its potential impact. Finally, to generate structured logs, it utilizes the time-series database InfluxDB and a structured log framework (e.g., Log4j2). InfluxDB combines technologies such as custom layout to efficiently store raw indicator data including timestamped condition tags, cause identification results, and status assessments. Meanwhile, preset templates (template example: [TIMESTAMP][WORKING_CYCLE_ID][MODE: load-frequency combination][ROOT_CAUSE: dominant cause][STATUS: assessment level and key indicators]) pull data from the database and computing engine, instantiate it, and output it as human-readable and machine-parseable detection log entries, thus completing the final encapsulation from data to diagnostic knowledge.

[0070] Through the above technical solutions, the dominant causes of impact under different working conditions can be accurately identified, the fault status can be quantitatively assessed, and a structured detection log can be generated. This provides a basis for preventive maintenance of equipment, avoids coal mining interruptions and safety hazards, and ensures the continuous and efficient operation of underground coal mining.

[0071] Figure 3 A schematic diagram of a hydraulic cylinder abnormal impact detection system based on working condition clustering is provided in one embodiment of this application, as shown below. Figure 3 As shown, the hydraulic cylinder abnormal impact detection system 300 based on working condition clustering in this embodiment includes: a clustering analysis module 301, a risk analysis module 302, and a cause state module 303.

[0072] The clustering analysis module 301 is used to acquire the hydraulic cylinder operating condition information set, and based on the hydraulic cylinder operating condition information set, analyze the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle to obtain the operating condition clustering information set; the risk analysis module 302 is used to analyze the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slurry jamming, and multi-cylinder synchronization deviation based on the operating condition clustering information set to obtain the fault risk information set; the cause status module 303 is used to analyze the causes and fault status of abnormal impact under different operating conditions based on the fault risk information set, and generate and output the hydraulic cylinder abnormal impact detection log.

[0073] Optionally, when the clustering analysis module 301 analyzes the working characteristics of the pushing jack under different loads and operating frequencies in the underground coal mining cycle based on the hydraulic cylinder working condition information set to obtain the working condition clustering information set, it is specifically used for: the hydraulic cylinder working condition information set including the coal mining machine operation information set and the hydraulic cylinder status information set; based on the coal mining machine operation information set, analyzing the operating status of the coal mining machine at different positions to obtain the hydraulic cylinder action sequence corresponding to the operating status; based on the hydraulic cylinder status information set, analyzing the load change and action frequency change of the hydraulic cylinder in different working stages to obtain the load-frequency dynamic curve; based on the hydraulic cylinder action sequence, combined with the load-frequency dynamic curve, analyzing the repeated load and action frequency combinations of the hydraulic cylinder in the underground coal mining cycle to obtain the working condition clustering information set representing different working modes.

[0074] Optionally, the clustering analysis module 301, during the construction of the load-frequency dynamic curve, is specifically used for: analyzing the pressure timing information and action triggering timing information of the hydraulic cylinder based on the hydraulic cylinder state information set; analyzing the periodic fluctuation characteristics of the driving pressure of the hydraulic cylinder when performing push-pull, pull-frame, or bottom-lifting actions based on the pressure timing information, and obtaining load stage division information characterized by multiple pressure peak intervals; analyzing the density and interval pattern of control commands received by the hydraulic cylinder within a unit time based on the action triggering timing information, and obtaining action frequency stage division information; and, based on the load stage division information and combined with the action frequency stage division information, associating and matching the load characteristics and action frequency characteristics in the same time interval to obtain the load-frequency dynamic curve that synchronously describes the changes in load level and action frequency with time as the horizontal axis.

[0075] Optionally, when the clustering analysis module 301 analyzes the repeated load and action frequency combinations of the hydraulic cylinder in the underground coal mining cycle based on the hydraulic cylinder action sequence and combined with the load-frequency dynamic curve to obtain the working condition clustering information set representing different working modes, it is specifically used for: analyzing the fixed sequence and duration of the pushing conveyor, pulling frame, and lifting bottom actions within a single coal mining cycle based on the hydraulic cylinder action sequence to obtain the working action stage sequence; analyzing the range of load level variation and the concentration interval of action frequency within each action stage based on the load-frequency dynamic curve and combined with the working action stage sequence to obtain the load-frequency feature interval; analyzing the repeated feature information of the load-frequency feature interval corresponding to the same action stage in multiple consecutive coal mining cycles based on the load-frequency feature interval, and grouping the repeated feature intervals into a load-frequency combination; and merging combinations with similar load levels and similar action frequencies based on the load-frequency combination to obtain the working condition clustering information set representing different working modes.

[0076] Optionally, when the risk analysis module 302 analyzes the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slurry jamming, and multi-cylinder synchronization deviation based on the working condition clustering information set to obtain a fault risk information set, it is specifically used for: based on the working condition clustering information set, analyzing the triggering correspondence between abnormal impact and valve core motion nonlinearity according to the motion response characteristics of the hydraulic cylinder valve core, to obtain valve core nonlinearity-induced danger characteristic information; based on the working condition clustering information set, analyzing the hindrance-induced relationship between abnormal impact and coal slurry jamming according to the operation and coordination characteristics of the internal components of the hydraulic cylinder, to obtain coal slurry jamming-induced danger characteristic information; based on the working condition clustering information set, analyzing the cooperative imbalance relationship between abnormal impact and multi-cylinder synchronization deviation according to the synchronous action execution characteristics of multiple hydraulic cylinders, to obtain synchronization deviation-induced danger characteristic information; and based on the valve core nonlinearity-induced danger characteristic information, the coal slurry jamming-induced danger characteristic information, and the synchronization deviation-induced danger characteristic information, analyzing the independent and coupled influence laws of the three types of danger characteristics on abnormal impact, to obtain the fault risk information set.

[0077] Optionally, the risk analysis module 302, during the construction of the coal sludge jamming-induced danger characteristic information, is specifically used for: based on the working condition clustering information set, combined with the load-frequency combination, analyzing the changes in the fit clearance and smoothness of the hydraulic cylinder internal components under different working conditions, to obtain component operation fit deviation information; based on the component operation fit deviation information, analyzing the accumulation and movement obstruction characteristics of coal sludge in the component fit clearance, to obtain coal sludge accumulation and obstruction characteristic information; based on the coal sludge accumulation and obstruction characteristic information, analyzing the correlation between the forced obstruction and impact triggering of coal sludge jamming on the hydraulic cylinder movement, to obtain the coal sludge jamming-induced danger characteristic information.

[0078] Optionally, the risk analysis module 302, during the construction of the synchronous deviation-induced risk characteristic information, is specifically used for: analyzing the timing differences of synchronous action startup of multiple hydraulic cylinders under the same working condition mode based on the working condition clustering information set and the load-frequency combination, to obtain multi-cylinder action timing deviation information; analyzing the stroke matching deviation of multiple hydraulic cylinders when performing push-pull, pull-frame, or bottom-lifting actions based on the multi-cylinder action timing deviation information, to obtain multi-cylinder stroke coordination deviation information; analyzing the superposition characteristics of uneven hydraulic cylinder force and action jamming caused by synchronous deviation, to obtain coordination imbalance-induced impact characteristic information; and analyzing the induction correlation law between coordination imbalance and abnormal impact based on the coordination imbalance-induced impact characteristic information, to obtain the synchronous deviation-induced risk characteristic information.

[0079] Optionally, when the risk analysis module 302 analyzes the independent and coupled effects of the three types of risk-causing features on abnormal impacts based on the valve core nonlinear risk-causing feature information, the coal slime jamming risk-causing feature information, and the synchronization deviation risk-causing feature information to obtain the fault risk information set, it is specifically used to: analyze the correspondence between the hydraulic cylinder's own changes and the intensity and frequency of abnormal impacts when the other two types of risk-causing features are absent, based on the valve core nonlinear risk-causing feature information, the coal slime jamming risk-causing feature information, and the synchronization deviation risk-causing feature information, to obtain independent risk-causing feature mapping information; based on the independent risk-causing feature mapping information, analyze the amplification or induction law of the interaction between the two types of risk-causing features when they occur simultaneously, to obtain coupled risk-causing evolution information; and based on the independent risk-causing feature mapping information, combined with the coupled risk-causing evolution information, construct the fault risk information set of the probability and severity of impacts caused by single and multiple causes under different working conditions.

[0080] Optionally, when the cause status module 303 analyzes the causes and fault states of abnormal impacts under different working conditions based on the fault risk information set, and generates and outputs the abnormal impact detection log of the hydraulic cylinder, it is specifically used for: analyzing the abnormal impact information dominated by the valve core nonlinearity-causing characteristic information, the coal slime jamming-causing characteristic information, and the synchronization deviation-causing characteristic information under different working conditions characterized by the fault risk information set, based on the fault risk information set, to obtain the dominant cause discrimination information under each working condition; based on the dominant cause discrimination information, combined with the independent cause feature mapping information and the coupled cause evolution information, analyzing the severity and duration of hydraulic cylinder action jamming, pressure sudden change, and stroke deviation caused by single and coupled causes, to obtain fault state assessment information; based on the dominant cause discrimination information, combined with the fault state assessment information, according to the time sequence of the coal mining cycle, structurally associating and recording the abnormal impact causes identified under different working conditions, the corresponding fault state manifestations, and the corresponding load-frequency combinations, to generate and output the abnormal impact detection log of the hydraulic cylinder.

[0081] The system in this embodiment can be used to execute the methods of any of the above embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

Claims

1. A method for detecting abnormal impacts in hydraulic cylinders based on working condition clustering, characterized in that, include: Obtain a hydraulic cylinder working condition information set; based on the hydraulic cylinder working condition information set, analyze the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle, and obtain a working condition clustering information set. Based on the aforementioned working condition clustering information set, the correlation and mutual influence between abnormal impact and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation are analyzed to obtain a fault risk information set. Based on the fault risk information set, the causes and fault states of abnormal impacts under different working conditions are analyzed, and abnormal impact detection logs of hydraulic cylinders are generated and output.

2. The method according to claim 1, characterized in that, Based on the hydraulic cylinder operating condition information set, the working characteristics of the pushing jack under different loads and operating frequencies in the underground coal mining cycle are analyzed to obtain an operating condition clustering information set, including: The hydraulic cylinder operating condition information set includes the coal mining machine operation information set and the hydraulic cylinder status information set. Based on the coal mining machine operation information set, the operation status of the coal mining machine at different positions is analyzed to obtain the hydraulic cylinder action sequence corresponding to the operation status. Based on the hydraulic cylinder state information set, the load change and action frequency change of the hydraulic cylinder in different working stages are analyzed to obtain the load-frequency dynamic curve. Based on the hydraulic cylinder action sequence and the load-frequency dynamic curve, the combination of load and action frequency that repeatedly occurs in the underground coal mining cycle is analyzed to obtain the working condition clustering information set that characterizes different working modes.

3. The method according to claim 2, characterized in that, The process of constructing the load-frequency dynamic curve includes: Based on the hydraulic cylinder state information set, analyze the pressure timing information and action triggering timing information of the hydraulic cylinder; Based on the pressure timing information, the periodic fluctuation characteristics of the driving pressure of the hydraulic cylinder when performing pushing, pulling, or lifting actions are analyzed to obtain load stage division information characterized by multiple pressure peak intervals. Based on the action triggering timing information, the density and interval of control commands received by the hydraulic cylinder within a unit time are analyzed to obtain action frequency stage division information. Based on the load phase segmentation information and the action frequency phase segmentation information, load characteristics and action frequency characteristics in the same time interval are associated and matched to obtain the load-frequency dynamic curve that synchronously describes the changes in load level and action frequency with time as the horizontal axis.

4. The method according to claim 2, characterized in that, Based on the hydraulic cylinder action sequence and combined with the load-frequency dynamic curve, the repeated load and action frequency combinations of the hydraulic cylinder in the underground coal mining cycle are analyzed to obtain the working condition clustering information set characterizing different working modes, including: Based on the hydraulic cylinder action sequence, the fixed sequence and duration of the pushing conveyor, pulling frame, and lifting bottom actions within a single coal mining cycle are analyzed to obtain the working action stage sequence. Based on the load-frequency dynamic curve and the sequence of working action stages, the range of load level variation and the concentration range of action frequency within each action stage are analyzed to obtain the load-frequency characteristic range. Based on the load-frequency characteristic interval, the characteristic information of repeated occurrence of the load-frequency characteristic interval corresponding to the same action stage in multiple consecutive coal mining cycles is analyzed, and the repeated characteristic intervals are grouped into a load-frequency combination. Based on the load-frequency combination, combinations with similar load levels and similar operating frequencies are merged to obtain the working condition clustering information set that represents different working modes.

5. The method according to claim 4, characterized in that, Based on the aforementioned operating condition clustering information set, the correlation and mutual influence between abnormal impacts and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation are analyzed to obtain a fault risk information set, including: Based on the aforementioned working condition clustering information set, and according to the motion response characteristics of the hydraulic cylinder valve core, the triggering correspondence between abnormal impact and valve core motion nonlinearity is analyzed to obtain valve core nonlinearity-induced danger characteristic information. Based on the aforementioned working condition clustering information set, and according to the operational coordination characteristics of the internal components of the hydraulic cylinder, the hindrance-induced relationship between abnormal impact and coal slime jamming is analyzed to obtain coal slime jamming-induced danger characteristic information. Based on the aforementioned working condition clustering information set, and according to the synchronous action execution characteristics of multiple hydraulic cylinders, the collaborative imbalance relationship between abnormal impact and multi-cylinder synchronization deviation is analyzed to obtain the characteristic information of synchronization deviation causing danger. Based on the nonlinear hazard-causing characteristic information of the valve core, the hazard-causing characteristic information of coal slime stagnation, and the hazard-causing characteristic information of synchronization deviation, the independent and coupled influence laws of the three types of hazard-causing characteristics on abnormal impacts are analyzed to obtain the fault risk information set.

6. The method according to claim 5, characterized in that, The process of constructing the characteristic information of coal slime entrapment leading to danger includes: Based on the working condition clustering information set, combined with the load-frequency combination, the characteristics of the fit clearance change and smoothness of the internal components of the hydraulic cylinder under different working conditions are analyzed to obtain the component operation fit deviation information. Based on the component operation and fit deviation information, the characteristics of coal slime accumulation and movement obstruction in the component fit gap are analyzed to obtain coal slime accumulation and hindrance characteristic information. Based on the coal slime accumulation and obstruction characteristics, the correlation between the forced obstruction and impact triggering of the hydraulic cylinder action caused by coal slime jamming is analyzed to obtain the dangerous characteristics of coal slime jamming.

7. The method according to claim 6, characterized in that, The process of constructing the synchronization deviation risk-causing feature information includes: Based on the working condition clustering information set and combined with the load-frequency combination, the synchronous action start-up timing difference of multiple hydraulic cylinders under the same working condition mode is analyzed to obtain multi-cylinder action timing deviation information. Based on the multi-cylinder action timing deviation information, the stroke matching deviation of multiple hydraulic cylinders when performing push, pull, or lift actions is analyzed to obtain multi-cylinder stroke coordination deviation information. Based on the multi-cylinder stroke coordination deviation information, the superposition characteristics of uneven hydraulic cylinder force and action jam caused by synchronization deviation are analyzed to obtain the coordination imbalance-induced impact characteristic information. Based on the aforementioned characteristics of coordinated imbalance leading to shocks, the correlation between coordinated imbalance and abnormal shocks is analyzed to obtain the characteristics of synchronous deviation leading to risks.

8. The method according to claim 7, characterized in that, Based on the nonlinear hazard-causing characteristic information of the valve core, the hazard-causing characteristic information of coal slime stagnation, and the hazard-causing characteristic information of synchronization deviation, the independent and coupled influence patterns of the three types of hazard-causing characteristics on abnormal impacts are analyzed to obtain the fault risk information set, including: Based on the nonlinear hazard-causing characteristic information of the valve core, the hazard-causing characteristic information of the coal slime jamming and the hazard-causing characteristic information of the synchronization deviation, the correspondence between the hydraulic cylinder's own changes and the abnormal impact intensity and occurrence frequency when the other two types of hazard-causing characteristics are absent is analyzed to obtain independent hazard-causing characteristic mapping information. Based on the independent risk-causing feature mapping information, the interaction between two or three types of risk-causing features when they appear simultaneously is analyzed to determine the amplification or induction pattern of abnormal shocks, thereby obtaining coupled risk-causing evolution information. Based on the independent risk-causing feature mapping information and combined with the coupled risk-causing evolution information, a set of fault risk information is constructed to represent the probability and severity of impacts caused by single or multiple causes under different operating conditions.

9. The method according to claim 8, characterized in that, Based on the fault risk information set, the causes and fault states of abnormal impacts under different working conditions are analyzed, and a hydraulic cylinder abnormal impact detection log is generated and output, including: Based on the fault risk information set, the abnormal impact information dominated by the valve core nonlinearity-induced danger characteristic information, the coal slime jamming-induced danger characteristic information, and the synchronization deviation-induced danger characteristic information under different load-frequency combinations is analyzed to obtain the dominant cause discrimination information under each working condition. Based on the dominant cause discrimination information, combined with the independent risk-causing feature mapping information and the coupled risk-causing evolution information, the severity and duration of hydraulic cylinder motion jamming, pressure sudden change and stroke deviation caused by single and coupled causes are analyzed to obtain fault status assessment information. Based on the dominant cause discrimination information and the fault status assessment information, the abnormal impact causes identified under different working conditions, the corresponding fault status manifestations, and the corresponding load-frequency combinations are structurally associated and recorded according to the time sequence of the coal mining cycle, and the abnormal impact detection log of the hydraulic cylinder is generated and output.

10. A hydraulic cylinder abnormal impact detection system based on working condition clustering, characterized in that, The method applied to any one of claims 1-9 includes: The clustering analysis module is used to obtain the hydraulic cylinder working condition information set. Based on the hydraulic cylinder working condition information set, the working characteristics of the push jack under different loads and operating frequencies in the underground coal mining cycle are analyzed to obtain the working condition clustering information set. The risk analysis module is used to analyze the correlation and mutual influence between abnormal impacts and valve core motion nonlinearity, coal slime jamming, and multi-cylinder synchronization deviation based on the working condition clustering information set, and to obtain a fault risk information set. The cause status module is used to analyze the causes and fault status of abnormal impacts under different working conditions based on the fault risk information set, and generate and output the abnormal impact detection log of the hydraulic cylinder.

Images