Fault diagnosis method and device for hydraulic system of support equipment in fully mechanized coal mining and withdrawing triangular area

Abstract

The application relates to the technical field of hydraulic system control, and discloses a fault diagnosis method and device for a hydraulic system of a fully-mechanized coal mining and withdrawing triangular area supporting equipment. The method comprises the following steps: acquiring real-time operation data of each oil cylinder through a hydraulic system sensor, extracting the difference of the motion state of the oil cylinder by using a time sequence analysis and a principal component analysis method, and calculating the inertia force distribution of the oil cylinder and a quantitative index of uneven force distribution. When the quantitative index exceeds a preset threshold value, a clustering analysis method is used to identify an abnormal oil cylinder, the correlation between a positioning deviation and a seal wear is analyzed by using a linear regression model, and the occurrence probability of an early fault signal is calculated. A fault prediction grading result is generated based on the probability distribution, the control parameters of the oil cylinder are adjusted by using a decision tree algorithm, the operation parameters of the oil cylinder are updated, the dynamic adjustment of the hydraulic system and the optimization of the supporting structure are realized, the fault diagnosis precision of the hydraulic system is improved, and the stability and reliability of the supporting structure are ensured through real-time compensation control.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of hydraulic system control, in particular to a fault diagnosis method and device for a hydraulic system of a fully mechanized mining and withdrawing triangular area supporting device. BACKGROUND

[0002] The hydraulic system of a fully mechanized mining and withdrawing triangular area supporting device plays a crucial role in underground engineering such as mine exploitation and tunnel construction, especially in complex geological environments, where the hydraulic supporting device bears a huge safety pressure. The system usually works collaboratively with multiple oil cylinders, responsible for supporting and stabilizing the working face. However, during the collaborative control of multiple oil cylinders, due to changes in geological conditions and differences in oil cylinder operating states, the hydraulic system often has problems such as uneven oil cylinder movement and uneven force, which further affects the stability of the overall supporting structure. These problems not only lead to a decrease in equipment efficiency, but also may cause serious safety hazards.

[0003] Traditional fault diagnosis methods for hydraulic systems usually rely on low-frequency monitoring and manual judgment, which often cannot detect abnormalities in oil cylinder operation in real time, and it is difficult to accurately identify the root cause of the instability of the supporting structure. With the development of intelligent technology, fault diagnosis of hydraulic systems has gradually developed towards high-frequency data collection and intelligent analysis, but the existing diagnosis methods still have problems of insufficient accuracy and poor real-time performance. Especially in the scenario of multiple oil cylinders working collaboratively, how to accurately analyze the differences in oil cylinder movement states, uneven force distribution and other key indicators, and conduct dynamic compensation control based on real-time data, is still a technical problem to be solved.

[0004] The present application proposes a fault diagnosis method based on real-time data analysis and dynamic adjustment, aiming at the technical bottleneck of multiple oil cylinder collaborative control and fault diagnosis in hydraulic systems. This method can effectively improve the accuracy of oil cylinder collaborative control, ensure the stability of supporting equipment under complex working conditions through accurate diagnosis and dynamic compensation, and provide a new solution for the intelligent and refined management of hydraulic systems. SUMMARY

[0005] The present application provides a fault diagnosis method and device for a hydraulic system of a fully mechanized mining and withdrawing triangular area supporting device, aiming to solve the problems of fault diagnosis and dynamic compensation in the collaborative control of multiple oil cylinders in hydraulic supporting systems through high-frequency data collection and intelligent analysis technology, and thereby improve the stability and reliability of the supporting equipment.

[0006] In a first aspect, the present application provides a fault diagnosis method for a hydraulic system of a fully mechanized mining and withdrawing triangular area supporting device, which comprises:

[0007] Step 1, obtaining operation data of each oil cylinder from a hydraulic system sensor to obtain a real-time data sequence; Step 2, analyzing motion state differences of each oil cylinder according to the real-time data sequence to determine a deviation feature vector of multi-oil cylinder cooperation; Step 3, calculating an inertial force distribution of each oil cylinder based on the deviation feature vector and generating a quantitative index of uneven force distribution;

[0008] Step 4, if the quantitative index exceeds a preset threshold, identifying an abnormal oil cylinder and obtaining an identification set of the abnormal oil cylinder;

[0009] Step 5, analyzing the correlation between positioning deviation and fault based on the identification set and calculating a probability distribution of an early fault signal; Step 6, generating a hierarchical result of fault prediction analysis according to the probability distribution to determine an oil cylinder control parameter to be adjusted;

[0010] Step 7, generating a compensation control strategy according to the oil cylinder control parameter to obtain a dynamic adjustment instruction of each oil cylinder; Step 8, updating motion parameters of each oil cylinder based on the dynamic adjustment instruction to obtain an optimized state of a stable support structure.

[0011] In a second aspect, the present application provides a fault diagnosis device for a hydraulic system of a fully-mechanized mining and withdrawing triangular area support equipment, the device comprising:

[0012] A data acquisition module is configured to obtain operation data of each oil cylinder from a hydraulic system sensor to obtain a real-time data sequence;

[0013] A vector generation module is configured to analyze motion state differences of each oil cylinder according to the real-time data sequence to determine a deviation feature vector of multi-oil cylinder cooperation;

[0014] An index calculation module is configured to calculate an inertial force distribution of each oil cylinder according to the deviation feature vector and generate a quantitative index of uneven force distribution;

[0015] An abnormality identification module is configured to identify an abnormal oil cylinder and obtain an identification set of the abnormal oil cylinder when the quantitative index exceeds a preset threshold;

[0016] A probability calculation module is configured to analyze the correlation between positioning deviation and fault based on the identification set and calculate a probability distribution of an early fault signal;

[0017] A parameter determination module is configured to generate a hierarchical result of fault prediction analysis according to the probability distribution to determine an oil cylinder control parameter to be adjusted;

[0018] A strategy generation module is configured to generate a compensation control strategy according to the oil cylinder control parameter to obtain a dynamic adjustment instruction of each oil cylinder;

[0019] A parameter update module is configured to update motion parameters of each oil cylinder based on the dynamic adjustment instruction to obtain an optimized state of a stable support structure.

[0020] Compared with the prior art, the beneficial effects of the technical scheme of the present application are at least as follows:

[0021] 1. Through high-frequency data acquisition technology, the running state of each oil cylinder in the hydraulic system can be monitored in real time, and abnormalities in the operation of the oil cylinder can be found in a timely manner, significantly improving the response speed and accuracy of fault diagnosis.

[0022] 2. Through multi-dimensional data analysis and principal component analysis, the deviation characteristics in the cooperative operation of the oil cylinder can be accurately identified, the degree of uneven force distribution can be quantified, and abnormal oil cylinders can be effectively identified using clustering analysis methods, combined with the probability distribution of fault signals, to generate accurate fault prediction results.

[0023] 3. According to the fault prediction analysis results, the running parameters of the oil cylinder are dynamically adjusted to generate real-time compensation control strategies. Dynamic adjustment instructions can ensure the synchronization and stability of the oil cylinder operation, thereby improving the overall reliability of the hydraulic support system.

[0024] 4. Through dynamic adjustment of the running parameters of the oil cylinder, the stability and safety of the support structure are optimized, effectively avoiding the problem of unstable support structure caused by uneven operation of the oil cylinder, and ensuring efficient and safe operation of the hydraulic system.

[0025] 5. Intelligent management of the hydraulic support system is realized, and the running efficiency and fault warning capability of the equipment are improved using data analysis and intelligent decision-making technology, providing technical support for long-term stable operation of the hydraulic support equipment. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor based on these drawings.

[0027] Figure 1 The flowchart of the fault diagnosis method of the hydraulic system of the fully mechanized coal mining and withdrawing triangular area support equipment of the present application;

[0028] Figure 2 The schematic diagram of the fault diagnosis device of the hydraulic system of the fully mechanized coal mining and withdrawing triangular area support equipment of the present application. DETAILED DESCRIPTION

[0029] The terms "first", "second", "third", "fourth" etc. (if any) in the description and claims of this application and the above drawings are used for distinguishing between similar objects, not necessarily for describing a particular sequential or chronological order. It is to be understood that the use of data herein set forth by way of example is not deemed to be limiting, but that any and all applications, including those peripherally, oreven obliquely related to the applications herein, are contemplated and can be claimed by the Applicant forthwith. Additionally, the terms "comprising", "having", "including", and "containing" are to be construed as open-ended terms (i.e., meaning "including, but not limited to", "comprising, but not limited to", "having, but not limited to", or "including, but not limited to") unless otherwise noted. It is intended that the application encompass multiple combinations of the elements specified herein including combinations that do not exist per se but would be readily known by those skilled in the art in light of the disclosure. It is not intended that the application be limited in any way by any theory or mechanism of action.

[0030] For the purpose of facilitating understanding, the specific flow of the embodiments of the present application is described below, Figure 1 The flow chart shown is an embodiment of the fault diagnosis method of the hydraulic system of the fully mechanized coal mining and withdrawal triangular area support equipment provided by the present application, which specifically includes the following steps:

[0031] Step 1, obtaining the operation data of each oil cylinder from the hydraulic system sensor to obtain a real-time data sequence.

[0032] In a specific embodiment, the process of performing step 1 can specifically include the following steps: (1) obtaining the pressure, flow and displacement data of each oil cylinder from the hydraulic system sensor through high-frequency sampling technology; (2) generating an initial real-time data sequence of each oil cylinder based on the pressure, flow and displacement data; (3) pre-processing the real-time data sequence to obtain a data sequence filtered of noise, and defining it as a real-time data sequence.

[0033] Specifically, the pressure sensor, flow meter and displacement sensor deployed on the oil cylinder are used to synchronously collect the operation parameters, and the sampling frequency is set to 1000 Hz to capture the transient characteristics of the hydraulic system. The pressure sensor is installed at the inlet and outlet of the oil cylinder, the flow meter adopts a turbine structure and is configured with a temperature compensation module, and the displacement sensor is a magnetostrictive type with a measurement accuracy of 0.01 mm. The data of the three types of sensors are time-aligned through a hardware synchronization mechanism, forming a three-dimensional data vector containing a time stamp and arranged in time sequence as an initial real-time data sequence. The sequence is subjected to Butterworth low-pass filtering and adaptive notch processing to eliminate hydraulic pump pulsation noise and mechanical vibration interference, and finally outputs a real-time data sequence filtered of noise.

[0034] Preferably, the dynamic characteristics of each oil cylinder during operation are determined according to the real-time data sequence after noise filtering, including pressure dynamic characteristics, flow dynamic characteristics, displacement dynamic characteristics, etc. First, the pressure dynamic characteristics are extracted, and three statistical characteristic parameters of the pressure signal, including mean value, variance and peak factor, are calculated. Then, the flow dynamic characteristics are analyzed, the main frequency components and corresponding amplitudes of the flow signal are extracted using a frequency domain analysis method, the spectral characteristics of the flow signal can reflect the periodicity and abnormal oscillation of the oil cylinder movement, the power spectral density of the flow signal is calculated to quantify the energy distribution of different frequency components. Next, the displacement dynamic characteristics are determined, and the velocity and acceleration information of the displacement signal are calculated. Finally, a multi-dimensional dynamic characteristic vector is constructed, and the dynamic characteristic parameters of pressure, flow and displacement are combined into a comprehensive characteristic vector, covering the main dynamic characteristics of the three types of sensor data. The multi-dimensional dynamic characteristic vectors of each oil cylinder are organized in time series to form a real-time state data set. The data set is stored in matrix form, with rows representing time series and columns representing feature dimensions.

[0035] Step 2, analyze the motion state differences of each oil cylinder according to the real-time data sequence, and determine the deviation feature vector of multi-oil cylinder cooperation.

[0036] In a specific embodiment, the process of performing step 2 can specifically include the following steps: (1) based on the real-time data sequence, using time series analysis method to extract the motion state parameters of each oil cylinder at start-stop moment; (2) setting a reference oil cylinder, calculating the motion difference value of each oil cylinder at start-stop moment based on the motion state parameters;

[0037] (3) arranging the motion difference values of each oil cylinder in time series to construct an evaluation matrix, performing principal component analysis on the evaluation matrix, extracting the main deviation mode and determining its contribution weight, performing weighted summation on the main deviation mode to generate multi-dimensional deviation feature data; (4) normalizing the deviation feature data to obtain standardized deviation feature data;

[0038] (5) based on the standardized deviation feature data, generating the deviation feature vector of multi-oil cylinder cooperation.

[0039] Specifically, based on the real-time collected pressure, flow and displacement data sequence, the time series analysis method is used to extract the motion state parameters of each oil cylinder at start-stop moment, including velocity change rate, acceleration peak value and pressure rising slope, etc., which can comprehensively reflect the dynamic response characteristics of the oil cylinder at start-stop moment.

[0040] The motion state parameter of the first oil cylinder is taken as a reference, and the parameter difference values of other oil cylinders and the reference oil cylinder at the same time are calculated to obtain the relative motion difference values of each oil cylinder. Through this way, the motion state difference of each oil cylinder at the start-stop moment is quantified, and data support is provided for constructing the evaluation matrix. The relative motion difference values of each oil cylinder are arranged in time sequence to construct a multi-oil cylinder motion synchronization evaluation matrix. The rows of the matrix represent different oil cylinder numbers, the columns represent time sampling points, and the matrix elements are the motion difference values of the corresponding oil cylinder at a specific time. In this way, a data matrix reflecting the overall collaborative state is formed.

[0041] The principal component analysis can identify the main change direction in the data. The principal component analysis method is used to reduce the dimension of the motion synchronization evaluation matrix, extract the main deviation modes affecting the multi-oil cylinder collaborative performance, such as overall lag deviation, local asynchronous deviation and periodic fluctuation deviation, etc., and determine the contribution weight of each deviation mode. The influence degree of different deviation modes is quantified by weighted summation to generate multi-dimensional deviation feature data.

[0042] For example, assume that we collect data of 4 oil cylinders (A, B, C, D) at 5 sampling time points (t1, t2, t3, t4, t5), and obtain the motion difference value matrix as shown in Table 1:

[0043] Table 1

[0044]

[0045] Next, an evaluation matrix is constructed. The motion difference values of each oil cylinder at different time points are arranged in time sequence to form an evaluation matrix as shown in Table 1 above. The matrix reflects the motion state difference of each oil cylinder at the start-stop moment. Then, principal component analysis (PCA) is performed. By performing PCA processing on the above evaluation matrix, the main deviation mode is extracted. PCA identifies the principal component that has the most influence on the coordination of the motion of the oil cylinder by calculating the eigenvalue and eigenvector. According to the PCA analysis, the following principal component contribution rates are obtained: the first principal component (PC1) explains 60% of the variance; the second principal component (PC2) explains 25% of the variance; the third principal component (PC3) explains 10% of the variance; and the fourth principal component (PC4) explains 5% of the variance. According to the above PCA results, the weight of each principal component is determined, wherein the weight of PC1 is 60%, the weight of PC2 is 25%, the weight of PC3 is 10%, and the weight of PC4 is 5%. Next, weighted summation is performed to generate multi-dimensional deviation feature data. The principal component score corresponding to each time point is multiplied by its weight, and then the weighted summation of all principal component scores is performed to obtain the multi-dimensional deviation feature data at each time. For example, at t1, the characteristic vector is calculated using the following formula: characteristic vector 1 = 0.2 x 0.6 + 0.1 x 0.25 + 0.05 x 0.1 + 0.02 x 0.05 = 0.16, and the characteristic vectors at other time points are calculated in the same way, as shown in Table 2:

[0046] Table 2

[0047]

[0048] The final multi-dimensional deviation feature data is a sequence of characteristic vectors, representing the coordinated deviation of the motion state of the oil cylinder.

[0049] The normalized deviation features are arranged in time order and importance, and a fixed-dimensional characteristic vector is constructed, wherein the dimension of the initial characteristic vector is set to 12, the first 4 dimensions correspond to the quantization values of the main deviation mode, the middle 4 dimensions represent the time distribution characteristics of the deviation, and the last 4 dimensions reflect the frequency domain characteristics of the deviation. The characteristic vector is sparsified to obtain a deviation feature vector. Specifically, elements with a value less than a threshold value of 0.1 are set to zero, non-significant deviation features are removed, and the most critical information is retained, thereby reducing the data storage space and improving the subsequent calculation efficiency. Step 3, calculate the inertia force distribution of each oil cylinder based on the deviation feature vector, and generate a quantization index of force distribution unevenness.

[0050] In a specific embodiment, the process of step 3 can specifically include the following steps: (1) calculating the inertia force distribution of each oil cylinder using a weighted average algorithm based on the deviation feature to obtain an inertia force distribution vector;

[0051] (2) Calculate the difference between each element of the inertial force distribution vector and the average of the vector to determine the force distribution deviation value of each oil cylinder, and obtain a force distribution deviation value vector;

[0052] (3) Calculate the standard deviation, mean, skewness and kurtosis of the force distribution deviation value vector, and define the ratio of the standard deviation to the mean as the coefficient of variation, and perform weighted summation on the coefficient of variation, skewness and kurtosis to obtain a quantitative index.

[0053] Specifically, based on the deviation characteristics of each oil cylinder, the inertial force distribution of each oil cylinder is calculated by a weighted average algorithm to obtain an inertial force distribution vector, which reflects the contribution of each oil cylinder to the inertial force at the start-stop moment. On this basis, the mass equivalent coefficient of the oil cylinder is calculated by combining the effective action area of the oil cylinder and the density of the hydraulic oil, which represents the equivalent mass characteristics of the oil cylinder in inertial motion. According to these data, the instantaneous inertial force value of each oil cylinder is further calculated based on Newton's second law, which reflects the inertial resistance generated by the oil cylinder during startup or shutdown. In order to accurately process these inertial force values, a time window moving average method is used for weighted average processing, with a time window length of 100 milliseconds and a sliding step of 10 milliseconds. When weighted averaging, the data is exponentially weighted according to the time distance, with data closer to the current time having a greater weight. Through this weighted average processing, the inertial force value of each oil cylinder at different time points is obtained, thereby forming an inertial force distribution vector.

[0054] Based on the inertial force distribution vector, the force distribution deviation value of each oil cylinder is calculated, which reflects the difference between the inertial force value of each oil cylinder and the average inertial force of the system, thereby revealing the difference in collaborative work between the oil cylinders. To further quantify the degree of uneven force distribution, the standard deviation, mean, skewness and kurtosis of the inertial force distribution deviation value are calculated, where the standard deviation represents the degree of inertial force difference between the oil cylinders, the mean reflects the overall inertial force level of the oil cylinders, and the skewness and kurtosis represent the symmetry and sharpness of the inertial force distribution, respectively. The concept of coefficient of variation is introduced, and the ratio of the standard deviation to the mean is taken as a normalized unevenness index, and the coefficient of variation eliminates the dimensional influence, making the unevenness degree of force distribution under different working conditions comparable.

[0055] The three statistical quantities of the coefficient of variation, skewness and kurtosis are combined to generate a comprehensive quantitative index using weighted summation. By way of example, the weight distribution is that the coefficient of variation weight is 0.6, the skewness weight is 0.25, and the kurtosis weight is 0.15.

[0056] As a preferred technical solution of the present application, statistical analysis is performed on the quantitative index to obtain the distribution rule of uneven force distribution, and the mechanical contribution degree of each oil cylinder in collaborative operation is determined according to the distribution rule.

[0057] Specifically, the time series of the quantitative index is established, and the mean and slope of the quantitative index in different time windows are calculated by using the moving average method to identify the trend of uneven force distribution. In addition, through autocorrelation function analysis, the periodic characteristics of the quantitative index are further identified to determine whether there is a fixed fluctuation period. The probability density estimation method is used to fit the probability distribution function of the quantitative index, and then the quantile interval of the quantitative index is determined to form the judgment threshold of the four levels of slight, general, severe and extremely severe. Illustratively, after generating the continuous probability distribution curve of the quantitative index, the cumulative distribution function is calculated to determine the five key quantile points of 5%, 25%, 50%, 75% and 95%. Taking the measured data of a mine hydraulic support as an example, the distribution of the quantitative index presents right-skewed characteristics, and the quantile values are: Q5% = 0.22, Q25% = 0.35, Q50% = 0.48, Q75% = 0.61, and Q95% = 0.73. Taking the quantile points as the level boundaries, four-level judgment criteria are established: slight unevenness (normal working condition): index ≤ Q25% (0.35), general unevenness (attention state): Q25% < index ≤ Q50% (0.35-0.48), severe unevenness (warning state): Q50% < index ≤ Q75% (0.48-0.61), and extremely severe unevenness (fault state): index > Q75% (> 0.61).

[0058] Based on the inertia force distribution vector, the relative contribution ratio of each oil cylinder is calculated. The inertia force value of each oil cylinder is divided by the sum of the inertia forces of all oil cylinders to obtain the mechanical contribution percentage of each oil cylinder. A stability coefficient is introduced to correct the mechanical contribution of each oil cylinder. The stability coefficient is obtained by calculating the reciprocal of the coefficient of variation of the inertia force value of each oil cylinder. The smaller the coefficient of variation, the better the stability of the oil cylinder, and the corresponding weight of the mechanical contribution is increased. Considering the synergistic effect between the oil cylinders, a contribution correction matrix is established. By analyzing the correlation between the inertia forces of the oil cylinders, an n x n dimensional synergy coefficient matrix is constructed, with diagonal elements being 1 and non-diagonal elements being the mechanical synergy coefficients between the oil cylinders. The corrected mechanical contribution is sorted in descending order to determine the importance level of each oil cylinder in the synergistic operation. The oil cylinder with high contribution has a greater impact on the system stability, and needs to be monitored and maintained first.

[0059] Step 4, if the quantitative index exceeds the preset threshold, an abnormal oil cylinder is identified, and an identification set of the abnormal oil cylinder is obtained.

[0060] In a specific embodiment, the process of performing step 4 can specifically include the following steps: (1) when the quantitative index exceeds the preset threshold, a clustering analysis method is used to classify the operation data of each oil cylinder;

[0061] (2) Calculate the deviation degree of the feature vector of each cluster category from the normal operation benchmark, and identify the cluster category whose deviation degree exceeds the abnormality determination threshold, based on the identification result, obtain the abnormal oil cylinder causing positioning deviation, and then generate a preliminary abnormal oil cylinder list;

[0062] (3) Based on the preliminary abnormal oil cylinder list, generate an initial identification set;

[0063] (4) Verify the identification set to determine the accuracy of the abnormal oil cylinder, update the initial identification set based on the verification result, and obtain the identification set.

[0064] Specifically, when the quantitative index exceeds the preset threshold, it indicates that the oil cylinder has obvious uneven force or motion difference, which may be an early signal of failure. Identifying abnormal oil cylinders in a timely manner and taking appropriate measures can prevent the spread of failure and ensure the stability of the hydraulic system and the safe operation of the support equipment. The preset threshold is set according to the design specifications and historical operation data of the hydraulic system, and is usually set to 1.5 to 2 times the force distribution deviation in the normal operating state. When the quantitative index exceeds the threshold, the clustering analysis method is used to classify and process the operating data of each oil cylinder. The pressure fluctuation amplitude, flow rate change rate and displacement deviation of each oil cylinder are taken as characteristic parameters to construct a multi-dimensional feature vector, and the clustering analysis can classify similar features into the same category and identify the clustering category that exhibits abnormal behavior. Specifically, the pressure fluctuation amplitude is obtained by calculating the standard deviation of the pressure data at the start and stop moments, the flow rate change rate is obtained by calculating the difference between the flow data at adjacent time points, and the displacement deviation is the absolute difference between the actual displacement and the target displacement. The K-means clustering algorithm is used to classify these feature vectors, and the clustering number K value is set according to the number of oil cylinders and the complexity of the system, generally 1 / 3 to 1 / 2 of the total number of oil cylinders. The clustering algorithm calculates the Euclidean distance between each data point and the cluster center to classify similar oil cylinders into the same category. The feature value distribution of each cluster category can be determined by calculating the mean and variance of the pressure, flow and displacement dimensions of the cluster center, which will be used for subsequent identification of abnormal oil cylinders.

[0065] After classification, the deviation degree of the feature vector of each cluster category from the normal operation benchmark is calculated. The normal operation benchmark is obtained from the oil cylinder data during the stable operation period, which includes the stable range of pressure, flow and displacement. The deviation degree is calculated by Mahalanobis distance, which considers the correlation and weight of each feature parameter to evaluate the cluster category. If the deviation degree of the cluster category exceeds the abnormality determination threshold, it is marked as an abnormal category, and the abnormality determination threshold is set to 2 times the standard deviation of the normal operation benchmark deviation to ensure the accuracy and sensitivity of abnormality identification.

[0066] Based on the above analysis results, the cylinder numbers in the abnormal cluster category are extracted to form a preliminary abnormal cylinder list. The list includes the number of each abnormal cylinder, the cluster category it belongs to, and the deviation value. The cylinder numbers in the preliminary abnormal cylinder list are sorted and arranged according to the regional distribution and abnormality degree to form a structured abnormal cylinder identification set, which includes cylinder numbers, abnormal type identification, and priority level.

[0067] During the verification process, the running trend of each cylinder in the identification set over the past period of time is analyzed through the historical data backtracking method. The change trend of pressure, flow, and displacement data is used for backtracking, the linear regression slope of the change trend is calculated, and the proportion of data points exceeding the normal range is counted to evaluate the running stability of the cylinder. In addition, a cross-validation mechanism is used to compare the running states of adjacent cylinders to identify whether there is a regional abnormal pattern. When the correlation between adjacent cylinders is lower than a predetermined threshold, it can be marked as a potential abnormal cylinder. Finally, a multiple verification scoring mechanism is used to weight and sum the historical data backtracking results and cross-validation results to calculate the verification score. If the score exceeds 0.75, the cylinder is confirmed as an abnormal cylinder; if the score is between 0.5 and 0.75, it is marked as a suspected abnormality; and if the score is less than 0.5, the cylinder is removed from the identification set.

[0068] Through the above steps, abnormal cylinders caused by uneven force distribution can be effectively identified, and the final abnormal cylinder identification set can be generated to provide a basis for subsequent fault prediction and control parameter adjustment. In specific applications, such as mine hydraulic support systems, by collecting the running data of each cylinder and performing the above analysis, the cylinders that may cause instability of the support equipment can be quickly located and processed, thereby improving the running safety and maintenance efficiency of the system.

[0069] Step 5, based on the identification set, analyze the correlation between the deviation and the fault, and calculate the probability distribution of the early fault signal.

[0070] In a specific embodiment, the process of performing step 5 can specifically include the following steps:

[0071] (1) Based on the identification set, obtain the running parameters of each abnormal cylinder;

[0072] (2) Based on the running parameters, obtain the positioning deviation data and the wear characteristic data of the sealing element, use a linear regression model to establish the relationship between the wear characteristic and the positioning deviation, and calculate the regression coefficient and the Pearson correlation coefficient to quantify the correlation between the positioning deviation and the wear of the sealing element;

[0073] (3) By setting the correlation threshold interval, the Pearson correlation coefficient is divided into grades, the conditional probability table is constructed based on the historical failure data, the posterior probability is derived by using the Bayes theorem combined with the prior probability and the current correlation value, and the posterior probability is taken as the occurrence probability of the early failure signal; (4) According to the occurrence probability of each abnormal oil cylinder, the initial probability distribution of the early failure signal is generated; (5) The initial probability distribution is normalized to obtain the standardized probability distribution.

[0074] Specifically, the operating parameters of the oil cylinder include pressure, flow, displacement, acceleration, response time, oil temperature and working load, etc., which are the basic data for analyzing the correlation of oil cylinder performance, positioning deviation and seal wear. The pressure change curve, flow fluctuation data and displacement offset of the abnormal oil cylinder are read from the hydraulic system sensor, the response time, pressure peak and flow stability index of each abnormal oil cylinder at the start-stop moment are extracted, and the operating parameter data set is formed. Based on the operating parameters, the positioning deviation data and the wear characteristic data of the seal are obtained, a positioning deviation quantization matrix is established, the displacement offset of each abnormal oil cylinder is arranged in time sequence to form a deviation vector, and the oil cylinder number is taken as the row index, the sampling time point is taken as the column index, and the matrix element is the displacement deviation value of the corresponding oil cylinder at a specific time; a seal wear feature vector is constructed, and the wear degree of the seal is quantified by analyzing the pressure decay rate, flow leakage coefficient and response delay time of the oil cylinder. The least square method is used to establish a linear regression equation, the seal wear feature vector is taken as the independent variable, and the positioning deviation quantization matrix is taken as the dependent variable, the regression coefficients and the correlation coefficients are calculated, the linear regression equation is in the form of deviation value equal to wear coefficient multiplied by wear characteristic value plus constant term, and the best fitting parameters are determined by iteration optimization. Exemplarily, the linear regression equation is: deviation value = β0+ β1× pressure decay rate + β2× flow leakage coefficient + β3× response delay time, and the regression coefficients β0, β1, β2 and β3 are calculated by the least square method (OLS). The linear correlation strength between the positioning deviation and the seal wear is evaluated by the Pearson correlation coefficient, and the correlation numerical index is obtained. The value range of the Pearson correlation coefficient is from negative one to positive one, and the absolute value closer to one indicates stronger correlation. A positive value indicates a positive correlation, and a negative value indicates a negative correlation.

[0075] A correlation threshold interval is set, and the Pearson correlation coefficient is divided into three levels of high correlation, medium correlation and low correlation according to the intensity. For example, high correlation corresponds to an absolute value of the correlation coefficient greater than 0.8, medium correlation corresponds to 0.5 to 0.8, and low correlation corresponds to less than 0.5. A conditional probability table is established based on historical failure data, and the frequency proportion of actual failure under different correlation levels is counted. The conditional probability table records the probability value of the oil cylinder failing in a future preset time window under the condition of a specific correlation level. The occurrence probability of the current abnormal oil cylinder is calculated by using Bayes theorem, and the posterior probability is obtained by combining the prior probability and the likelihood function. The prior probability is derived from the historical failure statistics of the same type of oil cylinder, and the likelihood function is based on the observed correlation value. The posterior probability is the occurrence probability of the early failure signal.

[0076] The failure occurrence probabilities of each abnormal oil cylinder are sorted according to the numerical value to construct a probability sequence. The kernel density estimation method is used to fit the probability distribution function, and the Gaussian kernel function is selected as the smoothing kernel. The goodness-of-fit of the fitted distribution is verified, and the Kolmogorov-Smirnov test is used to evaluate the fitting effect. The test compares the maximum difference between the empirical distribution function and the theoretical distribution function to determine whether the fitted distribution can effectively describe the actual data characteristics. Based on the fitted probability distribution function, the complete probability distribution of the early failure signal is generated, including the probability density function and the cumulative distribution function. Through normalization processing, the sum of all probabilities is ensured to be 1, and the standardized failure probability distribution is obtained. This standardized distribution reflects the probability of the oil cylinder failing in a future time window.

[0077] The technical scheme of the present application can accurately identify potential failure oil cylinders in the hydraulic system and provide quantitative failure risk assessment for the maintenance team. This not only can detect potential failures caused by seal wear in advance, but also can improve the accuracy of failure warning and reduce the probability of failure, providing strong technical support for the stability and safety of the hydraulic support system.

[0078] Step 6, generating a hierarchical result of failure prediction analysis according to the probability distribution, and determining the oil cylinder control parameters that need to be adjusted.

[0079] In a specific embodiment, the process of performing step 6 can specifically include the following steps:

[0080] (1) According to the probability distribution, a decision tree algorithm is used to calculate the information gain of the probability distribution characteristics, select the feature with the maximum information gain as the splitting attribute of the decision tree, and generate a hierarchical result of failure prediction;

[0081] (2) Determine the fault risk level of each oil cylinder through the classification result; (3) Extract the control parameters of the oil cylinder that need to be adjusted according to the fault risk level.

[0082] Specifically, the probability distribution of the early fault signal is taken as the input feature vector of the decision tree, wherein the probability distribution features include the probability distribution of the seal wear probability, the oil cylinder internal leakage probability, and the displacement deviation probability. These features generally reflect the probability distribution of the occurrence of an event in different states of the oil cylinder, which is usually a probability value or a probability interval. The decision tree algorithm calculates the information gain of each probability dimension through information entropy. The information gain is a measure of the influence of a feature on the classification effect of the decision tree. The greater the information gain, the greater the contribution of the feature to the classification. The probability dimension with the maximum information gain is selected as the splitting attribute of the root node. For example, when the information gain of the seal wear probability is 0.85, the information gain of the oil cylinder internal leakage probability is 0.72, and the information gain of the displacement deviation probability is 0.63, the seal wear probability is selected as the splitting attribute of the root node.

[0083] Next, based on the selected splitting attribute, a probability threshold is set to split the nodes, and the probability distribution is divided into different subsets. For example, for samples with a seal wear probability greater than 0.6, they are divided into the left subtree; for samples with a seal wear probability less than or equal to 0.6, they are divided into the right subtree. In each split subtree, the decision tree continues to calculate the information gain of the remaining attributes until a predetermined stopping condition is met. The stopping conditions include that the number of samples in the subset is less than a set minimum value (e.g., 5 samples), the information gain is less than a certain threshold (e.g., 0.1), or the depth of the tree exceeds a predetermined maximum value (e.g., 6 layers). This process enables the decision tree to gradually optimize the division and ultimately generate a classification tree that conforms to the actual data.

[0084] After generating the decision tree, the fault prediction classification standard is set according to the leaf nodes of the tree, and the fault level is determined according to the sample features reaching the leaf nodes. The mean value of each leaf node is calculated according to the sample features contained in the leaf node, and the fault level is divided according to the relationship between the mean value and the fault level. For example, if the mean value of the seal wear probability is greater than 0.8 and the mean value of the oil cylinder internal leakage probability is greater than 0.7 in the leaf node, the leaf node corresponds to the severe fault level; if the mean value of the seal wear probability is between 0.4 and 0.8 or the mean value of the oil cylinder internal leakage probability is between 0.3 and 0.7, the leaf node corresponds to the moderate fault level; if all probability mean values are less than 0.4, the leaf node corresponds to the slight fault level. By traversing the decision tree path, the corresponding fault prediction classification result is generated for each abnormal oil cylinder, thereby effectively providing a clear risk level indication for fault warning.

[0085] According to the above classification results, the next step is to determine the fault risk level of each oil cylinder. Specifically, according to the fault prediction classification results, each level is mapped to a specific risk level value, and the severe fault level corresponds to the risk level 9-10, the moderate fault level corresponds to the risk level 4-8, and the slight fault level corresponds to the risk level 1-3. On this basis, the importance of the oil cylinder in the support structure also needs to be considered, and the risk level of the oil cylinder in the key load-bearing position is adjusted, to ensure that the risk assessment of important oil cylinders is more accurate.

[0086] According to the determined fault risk level, further extraction of the oil cylinder control parameters to be adjusted. A mapping relationship table between fault risk level and control parameter is established, and different risk levels correspond to different parameter adjustment requirements. For example, the oil cylinder with risk level 9-10 needs to adjust three types of parameters: pressure control parameter, flow control parameter and displacement compensation parameter; the oil cylinder with risk level 6-8 needs to adjust two types of parameters: pressure control parameter and flow control parameter; the oil cylinder with risk level 3-5 only needs to adjust one type of parameter: pressure control parameter; and the oil cylinder with risk level 1-2 does not need to adjust the control parameter.

[0087] For the oil cylinder that needs to adjust the pressure control parameter, the current pressure set value, pressure response time and pressure stability coefficient are extracted as the parameters to be adjusted, and the adjustment direction and amplitude are determined according to the relationship between these parameters and the fault risk level. For the flow control parameter, the focus is on adjusting the flow regulating valve opening, flow response delay and flow distribution ratio to optimize the oil cylinder operating state. And the adjustment of displacement compensation parameters considers displacement deviation threshold, compensation gain coefficient and compensation response frequency to ensure the accuracy of oil cylinder movement and the stability of support structure.

[0088] The technical scheme of the present application combines decision tree algorithm and risk level mapping, which can not only accurately predict oil cylinder failure, but also develop the optimal oil cylinder control strategy according to the fault risk, thereby improving the stability and reliability of the hydraulic system, reducing the possibility of failure, and providing technical support for efficient operation of the support equipment.

[0089] Step 7, generating compensation control strategy according to oil cylinder control parameters to obtain dynamic adjustment instructions of each oil cylinder.

[0090] Specifically, by performing priority assessment on each control parameter, combining the importance of the parameter, the adjustment urgency and the adjustment complexity, the comprehensive priority score of each control parameter is obtained, and the adjustment priority sequence of the control parameter is generated according to the score from high to low. Exemplarily, the comprehensive score is equal to the parameter importance score multiplied by 0.5 plus the adjustment urgency score multiplied by 0.3 minus the adjustment complexity score multiplied by 0.2. The priority sorting and adjustment of these control parameters not only depend on their influence on the performance of the oil cylinder, but also consider the specific working environment and running task of each oil cylinder. For example, if an oil cylinder is located at a key support position and its failure risk level is relatively high, the pressure control parameter adjustment of the oil cylinder will be prioritized to ensure the stability of the hydraulic system. Subsequently, according to the adjustment sequence, the control parameter optimization scheme of each oil cylinder is determined. Following the control parameter adjustment sequence, specific parameter optimization values and adjustment timing are developed for each oil cylinder. The optimization scheme includes the target value, adjustment step and adjustment period of parameter adjustment, ensuring that the parameter adjustment process is smooth and controllable, and avoiding adverse effects on the stability of the support structure.

[0091] Through the implementation of this control strategy, the hydraulic system can respond to fault signals and abnormal states in real time, quickly and accurately perform dynamic adjustment, and reduce the risk of accidents caused by unstable hydraulic systems. Specifically, by continuously monitoring the operating state of the oil cylinder and combining the real-time calculated failure risk level, each oil cylinder is ensured to be properly controlled and adjusted under its risk level. In particular, the key oil cylinders are prioritized to ensure the safety and stability of the support structure under extreme working conditions.

[0092] Step 8, updating the motion parameters of each oil cylinder based on the dynamic adjustment instruction to obtain an optimized state of the support structure.

[0093] In a specific embodiment, the process of performing step 8 can specifically include the following steps: (1) generating motion parameter update values for each oil cylinder through the hydraulic system controller according to the dynamic adjustment instruction; (2) adjusting the operating parameters of each oil cylinder based on the motion parameter update values; (3) monitoring the stability of the support structure in real time according to the adjusted operating parameters, and calculating a comprehensive stability score; (4) generating optimization state data of the support structure based on the comprehensive stability score, combined with historical change trends and state levels.

[0094] Specifically, first, according to the received dynamic adjustment instruction, the hydraulic system controller parses the cylinder identification, adjustment amplitude and adjustment direction information contained therein, and calculates the motion parameter update value of each cylinder by using the parameter mapping table built-in the controller. The mapping table establishes the corresponding relationship between the adjustment instruction and the specific motion parameter. For example, for the pressure adjustment instruction, the controller calculates the pressure increment or decrement based on the current cylinder pressure value and the target pressure difference value; for the flow adjustment instruction, the controller determines the adjustment amount of the valve opening according to the opening characteristic curve of the flow control valve; for the displacement adjustment instruction, the controller generates a displacement increment value in combination with the stroke range and current position of the cylinder. These calculation results provide accurate update basis for subsequent parameter adjustment.

[0095] Secondly, for nonlinear adjustment requirements, the controller uses interpolation algorithm, especially cubic spline interpolation method, to ensure the generation of smooth transition values between known parameter control points. This process is crucial for controlling the adjustment of the cylinder beyond the linear range, which can avoid system oscillation caused by too sharp adjustment. At the same time, the controller considers the coupling effect between cylinders when processing each cylinder parameter. That is, when the parameter of a certain cylinder is adjusted, the system automatically calculates the compensation parameters of other related cylinders to ensure the coordination between cylinders and avoid the instability of the overall system caused by single cylinder adjustment.

[0096] After completing the update of the motion parameters, the controller will attach a timestamp and priority identifier to each update value for subsequent parameter adjustment sorting and execution. Then, the controller sends adjustment instructions to each cylinder in priority order to ensure that critical parameters are adjusted first. For pressure parameters, adjustment is made through proportional pressure reducing valves, flow parameters are controlled through electro-hydraulic proportional flow valves, and displacement parameters are adjusted through servo valves for precise positioning to ensure that the operating parameters of each cylinder are accurately adjusted.

[0097] On the basis of the adjusted operating parameters, real-time monitoring of the stability of the support structure is an important part of the entire optimization process. Acceleration sensors, strain sensors, and displacement sensors deployed at key nodes of the support structure synchronously collect vibration, stress, and deformation data. These data are transmitted in real time to the monitoring unit through the field bus. The monitoring unit uses the Kalman filter algorithm to fuse and process these multi-sensor data, estimates the current state of the structure through the prediction step, and updates the state using sensor measurements. This process effectively eliminates sensor noise and environmental interference, improving the accuracy of structure state estimation. Based on the fused sensor data, the stability of the support structure is calculated as a comprehensive index in multiple dimensions, including structure vibration amplitude, stress distribution uniformity, and deformation compatibility. These indicators generate a stability score for the support structure through a multi-index comprehensive evaluation method, with a score range of 0 to 1. The closer the value is to 1, the better the stability. According to the comprehensive score, the state of the support structure is further classified, generating different optimization strategies and control parameters. According to the trend of the score, the optimization state data generation module uses time series analysis to predict the future development direction of the support structure, and combines the current state level to generate optimization state data containing state description, risk level, recommended measures, and warning information.

[0098] Finally, by comparing the optimization state data with the preset performance benchmark, it is verified whether the support structure meets the design requirements, confirming the effectiveness of parameter adjustment, ensuring that the overall performance of the support structure reaches the expected target. Through this series of optimization and adjustment, not only can the running state of the oil cylinder in the hydraulic system be monitored and adjusted in real time, but also the stability of the support structure can be ensured, improving the safety and reliability of the hydraulic system.

[0099] By accurately controlling the update of oil cylinder motion parameters, real-time monitoring of the stability of the support structure, and optimization using intelligent algorithms, the hydraulic system can adaptively adjust in actual application, significantly improving fault diagnosis accuracy and system response speed, thereby ensuring the safe and stable operation of the fully mechanized retreat triangular area support equipment.

[0100] The above describes the fault diagnosis method of the hydraulic system of the fully mechanized retreat triangular area support equipment in the embodiment of the present application. The following describes the fault diagnosis device of the hydraulic system of the fully mechanized retreat triangular area support equipment in the embodiment of the present application. Please refer to Figure 2 An embodiment of the fault diagnosis device of the hydraulic system of the fully mechanized retreat triangular area support equipment provided by the present application is provided. The device comprises:

[0101] The data acquisition module 10 is used to obtain the running data of each oil cylinder from the hydraulic system sensor to obtain a real-time data sequence.

[0102] The vector generation module 20 is configured to analyze the motion state difference of each oil cylinder according to the real-time data sequence, and determine a deviation feature vector of the multi-oil-cylinder cooperation.

[0103] The index calculation module 30 is configured to calculate the inertia force distribution of each oil cylinder according to the deviation feature vector, and generate a quantitative index of force distribution unevenness.

[0104] The anomaly recognition module 40 is configured to recognize an abnormal oil cylinder and obtain an identification set of the abnormal oil cylinder when the quantitative index exceeds a preset threshold.

[0105] The probability calculation module 50 is configured to analyze the correlation between positioning deviation and failure according to the identification set, and calculate the probability distribution of the early failure signal.

[0106] The parameter determination module 60 is configured to generate a hierarchical result of failure prediction analysis according to the probability distribution, and determine an oil cylinder control parameter that needs to be adjusted.

[0107] The strategy generation module 70 is configured to generate a compensation control strategy according to the oil cylinder control parameter, and obtain a dynamic adjustment instruction of each oil cylinder.

[0108] The parameter update module 80 is configured to update the motion parameter of each oil cylinder according to the dynamic adjustment instruction, and obtain an optimized state of the support structure stability.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, system and unit can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0110] The integrated unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0111] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A fault diagnosis method for a hydraulic system of a support device in a fully mechanized retreat triangular area, characterized in that, The method comprises the following steps: Step 1, obtaining the operation data of each oil cylinder from the hydraulic system sensor to obtain a real-time data sequence; Step 2, analyzing the motion state difference of each oil cylinder according to the real-time data sequence to determine the deviation feature vector of multi-oil cylinder cooperation; Step 3, calculating the inertia force distribution of each oil cylinder based on the deviation feature vector and generating a quantitative index of uneven force distribution; Step 4, if the quantitative index exceeds the preset threshold, identifying the abnormal oil cylinder and obtaining an identification set of the abnormal oil cylinder; Step 5, based on the identification set, analyzing the correlation between positioning deviation and fault, and calculating the probability distribution of early fault signal; Step 6, generating a hierarchical result of fault prediction analysis according to the probability distribution to determine the oil cylinder control parameter that needs to be adjusted; Step 7, generating a compensation control strategy according to the oil cylinder control parameter to obtain a dynamic adjustment instruction of each oil cylinder; Step 8, updating the motion parameters of each oil cylinder based on the dynamic adjustment instruction to obtain an optimized state of the supporting structure stability; The step 2 comprises: based on the real-time data sequence, using a time series analysis method to extract the motion state parameters of each oil cylinder at the start-stop moment; setting a reference oil cylinder, and calculating the motion difference value of each oil cylinder at the start-stop moment based on the motion state parameters; Arranging the motion difference values of each oil cylinder in time sequence to construct an evaluation matrix, performing principal component analysis on the evaluation matrix, extracting the main deviation mode and determining its contribution weight, performing weighted summation on the main deviation mode to generate multi-dimensional deviation feature data; performing normalization processing on the deviation feature data to obtain standardized deviation feature data; Based on the standardized deviation feature data, the deviation feature vector of multi-oil cylinder cooperation is generated; The step 3 comprises: according to the deviation feature, the inertia force distribution of each oil cylinder is calculated by using a weighted average algorithm to obtain an inertia force distribution vector; The difference between each element in the inertia force distribution vector and the mean value of the vector is calculated to determine the force distribution deviation value of each oil cylinder to obtain a force distribution deviation value vector; The standard deviation, mean value, skewness and kurtosis of the force distribution deviation value vector are calculated, and the ratio of the standard deviation to the mean value is defined as the coefficient of variation, and the coefficient of variation, the skewness and the kurtosis are weighted and summed to obtain the quantitative index.

2. The method of claim 1, wherein, The step 1 comprises: through high-frequency sampling technology, obtaining the pressure, flow and displacement data of each oil cylinder from the hydraulic system sensor; based on the pressure, flow and displacement data, generating an initial real-time data sequence of each oil cylinder respectively; pre-processing the real-time data sequence to obtain a data sequence filtered from noise, and defining it as the real-time data sequence.

3. The method of claim 1, wherein, The step 4 comprises: when the quantitative index exceeds the preset threshold, using a clustering analysis method to classify the operation data of each oil cylinder; The deviation degree of the feature vector of each cluster category from the normal operation reference is calculated, and the cluster category whose deviation degree exceeds the abnormal judgment threshold is identified, based on the identification result, the abnormal oil cylinder causing the positioning deviation is obtained, and then a preliminary abnormal oil cylinder list is generated; Based on the preliminary abnormal oil cylinder list, an initial identification set is generated; The initial identification set is verified to determine the accuracy of the abnormal oil cylinder, the initial identification set is updated based on the verification result, and the identification set is obtained.

4. The method of claim 1, wherein, The step 5 includes: Based on the identification set, the operating parameters of each abnormal oil cylinder are obtained; Based on the operating parameters, positioning deviation data and wear characteristic data of the sealing element are obtained, a linear regression model is used to establish the relationship between wear characteristics and positioning deviation, and regression coefficients and Pearson correlation coefficients are calculated to quantify the correlation between positioning deviation and sealing element wear; By setting the correlation threshold interval, the Pearson correlation coefficient level is divided, the conditional probability table is constructed based on historical failure data, the posterior probability is derived by combining the prior probability and the current correlation value using Bayes' theorem, and the posterior probability is taken as the occurrence probability of the early failure signal; according to the occurrence probability of each abnormal oil cylinder, an initial probability distribution of the early failure signal is generated; the initial probability distribution is normalized to obtain the standardized probability distribution.

5. The method of claim 1, wherein, The step 6 includes: According to the probability distribution, a decision tree algorithm is used to calculate the information gain of the probability distribution characteristics, select the feature with the maximum information gain as the split attribute of the decision tree, and generate a hierarchical result of fault prediction; Through the hierarchical result, the fault risk level of each oil cylinder is determined; according to the fault risk level, the oil cylinder control parameters that need to be adjusted are extracted.

6. The method of claim 1, wherein, The step 8 includes: according to the dynamic adjustment instruction, the motion parameter update value of each oil cylinder is generated by the hydraulic system controller; the operating parameters of each oil cylinder are adjusted based on the motion parameter update value; the stability of the support structure is monitored in real time according to the adjusted operating parameters, and a comprehensive stability score is calculated; based on the comprehensive stability score, the historical change trend and the state level are combined to generate optimization state data of the support structure.

7. A fault diagnosis device for a hydraulic system of a support device in a fully mechanized retreat triangle area, for implementing the method according to any one of claims 1 to 6, characterized in that, The device includes: A data acquisition module is configured to obtain operating data of each oil cylinder from a hydraulic system sensor to obtain a real-time data sequence; A vector generation module is configured to analyze the motion state difference of each oil cylinder based on the real-time data sequence to determine a deviation feature vector of multi-oil cylinder cooperation; An index calculation module is configured to calculate the inertia force distribution of each oil cylinder based on the deviation feature vector and generate a quantitative index of uneven force distribution; An abnormal identification module is configured to identify an abnormal oil cylinder and obtain an identification set of the abnormal oil cylinder when the quantitative index exceeds a preset threshold; A probability calculation module is configured to analyze the correlation between positioning deviation and failure based on the identification set and calculate the probability distribution of the early failure signal; A parameter determination module is configured to generate a hierarchical result of fault prediction analysis based on the probability distribution and determine the oil cylinder control parameters that need to be adjusted; A strategy generation module is configured to generate a compensation control strategy based on the oil cylinder control parameters to obtain a dynamic adjustment instruction for each oil cylinder; A parameter update module is configured to update the motion parameters of each oil cylinder according to the dynamic adjustment instruction to obtain an optimized state of the support structure.

Images