Intelligent Detection System for Mine Hoisting Steel Wire Rope

Abstract

The present invention relates to the fields of intelligent mining and machine learning, specifically an intelligent detection system for mine hoisting steel ropes, which includes an integrated sensor module, a data preprocessing module, a feature extraction module, a machine learning module, a real-time monitoring module, and a data analysis and maintenance recommendation module. The sensor module is responsible for collecting key data of the steel rope in real time. The data preprocessing module filters, calibrates, processes missing values, and standardizes the original data. The feature extraction module extracts key features from the processed data for subsequent machine learning analysis. The machine learning module uses training data to build a model to monitor the state of the steel rope and detect anomalies. The real-time monitoring module monitors and identifies any abnormal situations through real-time processing and a rule engine. The data analysis and maintenance recommendation module provides detailed data analysis and maintenance recommendations. It can comprehensively and intelligently monitor the state of mine hoisting steel ropes, helping to reduce maintenance costs and improve equipment reliability.

Description

Technical Field

[0001] The invention relates to the fields of smart mining and machine learning, and specifically to an intelligent detection system for mining hoisting wire ropes. Background Art

[0002] The modern mining industry plays an extremely important role worldwide, providing the necessary raw materials for the development of all walks of life. As a key base for resource mining, mines undertake the mission of excavating, extracting and processing ore. In the mining production process, mining hoisting equipment, as an important engineering machinery, has the key function of lifting ore and slag materials from deep to the surface. Its position in mining production is self-evident. However, the challenges that come with it are also increasingly prominent. Mining hoisting equipment has heavy work tasks and harsh working environments. In the high temperature, high humidity and dusty mining environment, these equipment need to maintain continuous and stable operation. Operation to ensure the continuity of production, however, due to the influence of these special environments, the wear, corrosion and fatigue problems of the equipment are becoming increasingly apparent, becoming one of the factors restricting the performance and safety of the equipment. In mining hoisting equipment, the wire rope plays a vital role as a transmission and support device. The wire rope needs to withstand extremely high tension and torsion, as well as frequent bending and wear. Therefore, the state of the wire rope has a direct impact on the safety and reliability of the equipment. Once the wire rope is broken, worn, or corroded, it may cause serious accident risks, leading to equipment failure, production interruption, and even endangering the lives of personnel.

[0003] However, the traditional method of monitoring the condition of mining hoisting equipment has a series of limitations. It usually adopts the method of regular inspection and maintenance. However, this method is not only discontinuous in time, but also easily ignores the transient problems of equipment status. In addition, it relies on the subjective judgment of the operator, which may be inconsistent and inaccurate. At the same time, regular inspection and maintenance require a lot of manpower and time, which increases operating costs. More importantly, this method cannot achieve real-time monitoring and response to equipment status, and thus cannot prevent accidents.

[0004] Therefore, the present invention comes into being, aiming to solve the problem of status monitoring and management of mining hoisting equipment in an innovative way. The goal of the present invention is to build an intelligent detection system to realize real-time monitoring, abnormality detection and predictive analysis of wire rope status, thereby improving the safety and reliability of equipment, reducing maintenance costs, and ensuring the continuity of mine production. Summary of the invention

[0005] In view of the above problems, the present invention aims to provide an intelligent detection system for mining hoisting wire ropes.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] Intelligent detection system for mine hoisting steel wire ropes, including an integrated sensor module, a data preprocessing module, a feature extraction module, a machine learning module, a real-time monitoring module, and a data analysis and maintenance advice module. The sensor module is responsible for collecting data on key parameters of the steel wire rope in real time, including tension, vibration, and temperature, providing the basic data for subsequent analysis. The data preprocessing module filters, calibrates, and normalizes the raw data, improving the quality and reliability of the data; the feature extraction module extracts useful features from the massive data through advanced signal processing techniques, providing support for machine learning modeling. The machine learning module trains the model using historical data, achieving anomaly detection and predictive analysis, thereby improving the accuracy of the steel wire rope status; the real-time monitoring module continuously monitors the steel wire rope status, promptly senses abnormal situations, and takes automated responses according to predetermined rules; the data analysis and maintenance advice module provides detailed data analysis and maintenance advice through in-depth analysis of the data, helping operation and maintenance personnel better understand the data, identify problems, and plan maintenance work, reducing the maintenance cost.

[0008] Furthermore, the sensor module collects key parameters and data during the operation of the steel wire rope so that the system can monitor the status of the steel wire rope in real time. The sensors involved in the present invention include a tension sensor, a vibration sensor, and a temperature sensor. Among them, the tension sensor is used to measure the tension of the steel wire rope to detect changes in tension; the vibration sensor is used to detect the vibration of the steel wire rope to identify changes in vibration frequency and amplitude; the temperature sensor is used to monitor the temperature of the steel wire rope to detect problems caused by temperature rise; the sensor module has the function of data acquisition, collecting data generated by the sensors in real time, providing high-precision and high-resolution data to ensure accurate monitoring of the steel wire rope status. Higher precision and resolution help the system better detect small changes and anomalies. In the intelligent detection system for mine hoisting steel wire ropes, the data accuracy needs to have the following characteristics: high precision, precision calibration, number of digits and accuracy, signal-to-noise ratio. Among them, high precision means that the sensor module should provide high-precision data, making the measured value of the sensor close to the actual physical quantity with small errors. High-precision sensors can better detect small changes in the steel wire rope, thereby improving the sensitivity of the system; precision calibration means that the sensor needs to be calibrated regularly, and by correcting the output of the sensor, it is kept consistent with the actual physical quantity; the number of digits and accuracy mean that the data resolution is usually expressed in the number of digits (bits). The sensors of the present invention are 16-bit, capable of representing 65,536 different values, thereby providing a larger measurement range and more accurate measurement results; the signal-to-noise ratio refers to the ratio of the signal to the noise. A high signal-to-noise ratio means that the sensor can better distinguish the signal from the noise, thereby improving the reliability of the data.

[0009] Further, the raw data collected from the sensor module is processed and optimized by the data preprocessing module to ensure that high-quality input can be obtained for data analysis and algorithms. The data preprocessing module uses filtering techniques to remove noise and interference in the sensor data. This noise comes from the sensor itself, electromagnetic interference, and environmental vibration factors. Based on the Bessel function, the present invention proposes a window function ω(n) applicable to the intelligent detection system of mine hoisting steel ropes, which can balance the main lobe width and sidelobe suppression and can achieve fine spectral control to overcome the noise or signal attenuation from the sensor itself, electromagnetic interference, and environmental vibration. The window function ω(n) proposed by the present invention is expressed as follows:

[0010]

[0011] where n is the sample index of the window function, α is a parameter for controlling the main lobe width and sidelobe suppression. If α→0, it will result in a narrower main lobe but increase the amplitude of the sidelobe. N is the window length, and k is the index number. If there is a calibration deviation in the sensor, the data preprocessing module corrects the sensor output by applying a calibration coefficient to ensure the consistency between the sensor measurement value and the actual physical quantity, and constructs a polynomial model:

[0012] D calibrated (t)=a 0 +a 1 L(t)+a 2 L 2 (t)+…+a k L k (t)

[0013] where D calibrated (t) is the sensor output value, a 0 ,a 1 ,a 2 ,…,a k are the calibration coefficients to be estimated, L is the actual load. To estimate the calibration coefficients, by collecting the sensor measurement values D filtered under the known load L, the data set under the actual load L is: L={(L 1 ,D filtered (L 1 )),(L 2 ,D filtered (L 2 ),…,(L K ,D filtered (L K ))}, where K is the number of data points collected, L 1 ,L 2 ,...,L K are the 1st, 2nd, …, Kth actual loads respectively, and the loss function is defined as To find the minimum value of the loss function Loss, a search factor set S is defined. The s-th search factor satisfies s ∈ S, and the position of the search factor is represented by x s and the next position is selected through a probability distribution The probability distribution of the present invention satisfies the following formula:

[0014]

[0015] where is the transition probability, x is the position in the search space, τ() is the data fermentation degree, η() is the heuristic factor, a and b are hyperparameters, length() is the length function used to find the length of the set. In each iteration, the data fermentation degree is updated according to the search path of the search factor and the loss function value:

[0016]

[0017] where γ is the data residue, L(x s ) is the loss function value of the search factor s at the position x, and γτ(x) takes into account the decay of the data fermentation degree. For the data fermentation degree, it should satisfy the information equivalence with the data entropy value. For τ(x s ) the data fermentation degree of the search factor s at the position x should satisfy:

[0018]

[0019] By the updated data fermentation degree, the search factor can iterate to the optimal solution faster to obtain the minimum value of the loss function. Sensor data may be lost or incomplete in some cases, and the data set L under the actual load is extended and updated to make it have timeliness:

[0020] L = {(t 1 , L 1 , D filtered (L 1 ))), (t 2 , L 2 , D filtered (L 2 ))), …, (t K , L K , D filtered (L K )))}

[0021] The data preprocessing module fills in the missing parts of the data through the following formula to maintain the continuity of the data:

[0022]

[0023] where L K-1 , LK+1 They are the (K - 1)-th and (K + 1)-th actual loads respectively, D^ is the missing data value, and t (+1) is the next time point, and t (-1) is the previous time point. For different application requirements, the data sampling rate needs to be adjusted. The data preprocessing module reduces or increases the data sampling rate to meet the requirements of subsequent processing steps, thereby reducing the computational load or improving the data resolution. The data preprocessing module stores the preprocessed data in the database and records logs to track data quality and analyze historical data.

[0024] Furthermore, the feature extraction module first needs to select appropriate features, including features of tension, vibration, and temperature, to best describe the state of the wire rope. Among them, tension has sample correlation in the field related to the present invention, vibration has time-frequency characteristics, and extending to frequency domain analysis can bring faster calculation results, while temperature has temporality. Wavelet transform is used to capture different frequency components and change trends in the temperature signal. In order to directly represent the best state of the features of tension, vibration, and temperature, the feature comprehensive value is defined as Φ, satisfying:

[0025]

[0026] where ZL, ZD, and WD represent tension, vibration, and temperature respectively, and f ZL is the tension function, f ZD is the vibration function, f WD is the temperature function, Δs is the increment of s, and w ZL is the tension weight, w ZD is the vibration weight, w WD is the temperature weight, i * is the imaginary factor, and C s,λ is the wavelet coefficient. The sorting function sort() is used to sort the feature comprehensive value Φ, and the feature comprehensive values from good to bad are obtained:

[0027] Φ * →sort(Φ(ZL, ZD, WD))

[0028] where Φ * is the feature comprehensive value from good to bad. The feature extraction module also needs to determine the dimension of the extracted features. High-dimensional features provide more information but may lead to overfitting. Therefore, a trade-off needs to be made between information gain and computational complexity, and at the same time, features must be able to be extracted quickly under real-time requirements in order to trigger an alarm or take measures in a timely manner.

[0029] According to the intelligent detection system for mine hoisting steel ropes, it is characterized in that the machine learning module uses the data obtained from the sensors and the features extracted by the feature extraction module to train and deploy a machine learning model to monitor and identify the state of the steel ropes. For machine learning, the above-mentioned feature comprehensive values from excellent to poor and labels are required, and the data set L is further expanded to make it identifiable:

[0030] L

[0031] ={(t 1 ,L 1 ,D filtered (L 1 ),Φ 1 ,y 1 ),(t 2 ,L 2 ,D filtered (L 2 ),Φ 2 ,y 2 ),…,(t K ,L K ,D filtered (L K ),Φ K ,y K )}

[0032] Among them, Φ 1 , Φ 2 , Φ K are the first, second, and L-th feature comprehensive values respectively, and y 1 , y 2 , y K are the first, second, and K-th data labels respectively. Define the machine learning model parameter as θ, and it satisfies:

[0033]

[0034] Among them, in order to find the optimal machine learning model parameter θ * , define the number of iterations as g, and there is:

[0035]

[0036] Among them, θ g is the machine learning model parameter of the g-th iteration, θ g+1 is the machine learning model parameter of the g+1-th iteration, and α is the learning rate. The present invention uses the F2 score to evaluate the quality of the machine learning model, and defines:

[0037]

[0038] Where ε is the adjustment factor, TP is the positive class sample, FP is the negative class sample, Recall is the recall rate, and the larger ε is, the higher the model's emphasis on the recall rate.

[0039] Furthermore, the real-time monitoring module continuously monitors the state of the wire rope and triggers an alarm and takes necessary measures when an abnormal situation is detected. It receives the real-time data stream from the sensor module and the data preprocessing module. The data stream includes data of key parameters such as tension, vibration, and temperature. It quickly processes the data from the sensors to detect any abnormal patterns related to the state of the wire rope.

[0040] Furthermore, the real-time monitoring module of the present invention can define a series of thresholds and rules to determine whether the data exceeds the normal range. If a certain threshold is exceeded, the system issues an alarm. Once the real-time monitoring module detects an abnormal or rule-violating situation, it will trigger an alarm, and the alarm can be in various ways, including sound alarms, visual cues, text messages, and email notifications, so that the operation and maintenance personnel can take timely actions. And it has a user interface for the operation and maintenance personnel to monitor the state of the wire rope in real time. The interface usually provides visual charts, real-time images, and data trends to help the operators quickly identify problems.

[0041] Furthermore, the data analysis and maintenance advice module is used to analyze the monitoring data, generate reports, and provide maintenance advice. It deeply analyzes the data collected from the sensor module and the real-time monitoring module to identify potential problems and trends. The data analysis and maintenance advice module analyzes the data to identify long-term trends, and based on historical data, this module performs predictive analysis to help predict future maintenance needs and possible failures. The data analysis and maintenance advice module generates detailed reports, provides information about the state of the wire rope, can effectively manage and store a large amount of monitoring data and analysis results for subsequent query, retrieval, and report generation, and can also recommend maintenance and repair plans to help plan maintenance work in advance and reduce the risk of sudden failures.

[0042] Advantages of the present invention: Traditional regular inspection and maintenance methods often have difficulty in detecting transient problems in the equipment state, while the real-time monitoring module of the present invention can achieve continuous monitoring of the equipment state. Once the equipment shows abnormalities, the system can quickly take measures, reducing the equipment failure rate. By promptly discovering and solving problems, the present invention helps to improve the reliability of mine hoisting equipment and reduces the impact of equipment failures on production. Traditional maintenance methods usually rely on regular inspections and maintenance, which not only consume a large amount of manpower and time but also may increase costs due to over-frequent or insufficient maintenance. The present invention realizes more scientific and targeted maintenance through the intelligent monitoring and data analysis and maintenance advice module. The system can provide maintenance advice according to the actual situation. In addition, through preventive maintenance, the present invention can also avoid emergency repair costs caused by equipment failures, further reducing maintenance expenditures. Innovations of the present invention: In the sensor module, the window function proposed in the present invention uses α as the parameter for controlling the main lobe width and sidelobe suppression. If α→0, it will result in a narrower main lobe but increase the amplitude of the sidelobe. N is the window length, and the present invention continuously updates and expands the data set under the actual load during the improvement of the module, making it include the actual load, sensor output value, time, feature comprehensive value, and label. In the process of finding the minimum value of the loss function Loss, the transition probability proposed in the present invention is more directive compared to the transition probability of heuristic algorithms such as the roulette method. This is because the present invention has imposed constraints on the data fermentation degree, making it obey the function change. Since the mine hoisting steel wire rope has actual physical significance, the related tension, vibration, and temperature parameters are all closely related to the mine hoisting steel wire rope. And considering the information attenuation of the sensor transmission (such as multipath fading, hardware deviation), the present invention updates the data fermentation degree, making the search factor more directive in finding the optimal solution. In addition, the present invention constructs a feature comprehensive value, taking into account the influence of actual parameters on the mine hoisting steel wire rope. On the one hand, it reduces the computational load of the system for multiple parameters to best describe the state of the steel wire rope. Among them, tension has sample correlation in the field involved in the present invention, vibration has time-frequency characteristics, and faster calculation results can be obtained by extending to frequency domain analysis, while temperature has time series characteristics, and wavelet transform is used to capture different frequency components and change trends in the temperature signal. Finally, the present invention builds a machine learning module. For performance evaluation, the present invention proposes the F2 score. Compared with the traditional F1 score, the F2 score proposed in the present invention is more sensitive to unbalanced data sets. Especially in unbalanced data sets where the number of positive class samples is far less than the number of negative class samples, the F2 score can better reflect the model's attention to the minority class than the F1 score. It pays more attention to the performance of the model on the positive class and reduces the impact of false negatives.The F2 score achieves a trade - off between positive class samples, negative class samples, and recall by adjusting the ε value (the ε parameter of the F2 score is greater than 1). A larger ε value emphasizes recall more, while a smaller ε value emphasizes precision more, making the F2 score a flexible evaluation metric that can be adjusted according to application requirements. Through the real - time monitoring and automated response functions of the present invention, the system can take timely measures when abnormal situations are detected, avoiding potential accident risks and reducing the possibility of production interruptions. In addition, through intelligent maintenance suggestions, the system can minimize the production interruption time during maintenance, improve production efficiency, and ensure the smooth execution of the production plan. The present invention makes full use of big data analysis technology. By analyzing a large amount of data, valuable information is mined from it. This provides more data support for mine management and decision - making. Managers can formulate production plans, maintenance strategies, and resource allocations more scientifically based on the analysis results provided by the system. This helps improve the accuracy and efficiency of decision - making, provides a solid foundation for the long - term development and sustainability of the mine, leads the wave of intelligent production, enhances competitiveness, and promotes the development of the industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The invention will be further described with reference to the accompanying drawings. However, the embodiments in the drawings do not constitute any limitation to the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the following drawings.

[0044] Figure 1 It is a schematic structural diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0045] The present invention will be further described in conjunction with the following examples.

[0046] See Figure 1, the intelligent detection system for mine hoisting steel ropes in this embodiment includes an integrated sensor module, a data preprocessing module, a feature extraction module, a machine learning module, a real-time monitoring module, and a data analysis and maintenance advice module. The sensor module is responsible for collecting data on the key parameters of the steel rope in real time, including tension, vibration, and temperature, providing the basic data for subsequent analysis. The data preprocessing module filters, calibrates, and standardizes the original data, improving the quality and reliability of the data; the feature extraction module extracts useful features from the massive data through advanced signal processing techniques, providing support for machine learning modeling; the machine learning module trains the model using historical data to achieve anomaly detection and predictive analysis, thereby improving the accuracy of the steel rope state; the real-time monitoring module continuously monitors the steel rope state, senses abnormal situations in a timely manner, and takes automated responses according to predetermined rules; the data analysis and maintenance advice module provides detailed data analysis and maintenance advice through in-depth analysis of the data, helping the operation and maintenance personnel better understand the data, identify problems, and plan maintenance work, reducing the maintenance cost.

[0047] Specifically, the sensor module collects the key parameters and data during the operation of the steel rope so that the system can monitor the state of the steel rope in real time. The sensors involved in the present invention include a tension sensor, a vibration sensor, and a temperature sensor. Among them, the tension sensor is used to measure the tension of the steel rope to detect changes in tension; the vibration sensor is used to detect the vibration of the steel rope to identify changes in vibration frequency and amplitude; the temperature sensor is used to monitor the temperature of the steel rope to detect problems caused by temperature rise; the sensor module has the function of data collection, collecting the data generated by the sensors in real time, providing high-precision and high-resolution data to ensure accurate monitoring of the steel rope state. Higher precision and resolution help the system better detect small changes and anomalies. In the intelligent detection system for mine hoisting steel ropes, the data accuracy needs to have the following characteristics: high precision, precision calibration, number of digits and accuracy, signal-to-noise ratio. Among them, high precision means that the sensor module should provide high-precision data, making the measured value of the sensor close to the actual physical quantity with small errors. High-precision sensors can better detect the tiny changes of the steel rope, thereby improving the sensitivity of the system; precision calibration means that the sensor needs to be calibrated regularly, and by correcting the output of the sensor, it is kept consistent with the actual physical quantity; the number of digits and accuracy mean that the data resolution is usually expressed in the number of digits (bits). The sensors of the present invention are 16-bit, which can represent 65,536 different values, thereby providing a larger measurement range and more accurate measurement results; the signal-to-noise ratio refers to the ratio of the signal to the noise. A high signal-to-noise ratio means that the sensor can better distinguish the signal from the noise, thereby improving the reliability of the data.

[0048] Specifically, the raw data collected from the sensor module is processed and optimized by the data preprocessing module to ensure that high-quality input can be obtained for data analysis and algorithms. The data preprocessing module uses filtering techniques to remove noise and interference in the sensor data. This noise comes from the sensor itself, electromagnetic interference, and environmental vibration factors. Based on the Bessel function, the present invention proposes a window function ω(n) applicable to the intelligent detection system of mine hoisting steel ropes, which can balance the main lobe width and sidelobe suppression and can achieve fine spectral control to overcome noise or signal attenuation from the sensor itself, electromagnetic interference, and environmental vibration. The window function ω(n) proposed by the present invention is expressed as follows:

[0049]

[0050] where n is the sample index of the window function, α is a parameter for controlling the main lobe width and sidelobe suppression. If α→0, it will result in a narrower main lobe but increase the amplitude of the sidelobe. N is the window length, and k is the index number. If there is a calibration deviation in the sensor, the data preprocessing module corrects the sensor output by applying a calibration coefficient to ensure the consistency between the sensor measurement value and the actual physical quantity, and constructs a polynomial model:

[0051] D calibrated (t)=a 0 +a 1 L(t)+a 2 L 2 (t)+…+a k L k (t)

[0052] where D calibrated (t) is the sensor output value, a 0 ,a 1 ,a 2 ,…,a k are the calibration coefficients to be estimated, L is the actual load. To estimate the calibration coefficients, by collecting the sensor measurement values D filtered under the known load L, the data set under the actual load L is: L={(L 1 ,D filtered (L 1 )),(L 2 ,D filtered (L 2 )),…,(L K ,D filtered (L K ))}, where K is the number of data points collected, L 1 ,L 2 ,...,L K are the 1st, 2nd, …, Kth actual loads respectively, and the loss function is defined as To find the minimum value of the loss function Loss, a search factor set S is defined. The s-th search factor satisfies s ∈ S, and the position of the search factor is represented by x s and the next position is selected through a probability distribution The probability distribution of the present invention satisfies the following formula:

[0053]

[0054] where is the transition probability, x is the position in the search space, τ() is the data fermentation degree, η() is the heuristic factor, a and b are hyperparameters, length() is the length function used to calculate the length of the set. In each iteration, the data fermentation degree is updated according to the search path of the search factor and the loss function value:

[0055]

[0056] where γ is the data residue degree, L(x s ) is the loss function value of the search factor s at the position x, and γτ(x) takes into account the decay of the data fermentation degree. For the data fermentation degree, it should satisfy the equivalence with the data entropy value. For τ(x s ) the data fermentation degree of the search factor s at the position x should satisfy:

[0057]

[0058] Through the updated data fermentation degree, the search factor can iterate to the optimal solution faster and obtain the minimum value of the loss function. Sensor data may be lost or incomplete in some cases. Expand and update the data set L under the actual load to make it have timeliness:

[0059] L = {(t 1 , L 1 , D filtered (L 1 ))), (t 2 , L 2 , D filtered (L 2 ))), …, (t K , L K , D filtered (L K ))}

[0060] The data preprocessing module fills in the missing parts of the data through the following formula to maintain the continuity of the data:

[0061]

[0062] where L K-1 , LK+1 They are the (K - 1)-th and (K + 1)-th actual loads respectively, D^ is the missing data value, and t (+1) is the next time point, and t (-1) is the previous time point. For different application requirements, the data sampling rate needs to be adjusted. The data preprocessing module reduces or increases the data sampling rate to meet the requirements of subsequent processing steps, thereby reducing the computational load or improving the data resolution. The data preprocessing module stores the preprocessed data in the database and records logs to track the data quality and analyze historical data.

[0063] Specifically, the feature extraction module first needs to select appropriate features, including features of tension, vibration, and temperature, to best describe the state of the wire rope. Among them, tension has sample correlation in the field related to the present invention, vibration has time-frequency characteristics, and extending to frequency-domain analysis can bring faster calculation results, while temperature has temporal characteristics, and wavelet transform is used to capture different frequency components and change trends in the temperature signal. In order to directly represent the best state of the features of tension, vibration, and temperature, the feature comprehensive value is defined as Φ, which satisfies:

[0064]

[0065] where ZL, ZD, and WD represent tension, vibration, and temperature respectively, and f ZL is the tension function, f ZD is the vibration function, f WD is the temperature function, Δs is the increment of s, and w ZL is the tension weight, w ZD is the vibration weight, w WD is the temperature weight, i * is the imaginary factor, and C s,λ is the wavelet coefficient. The sorting function sort() is used to sort the feature comprehensive value Φ to obtain the feature comprehensive values from best to worst:

[0066] Φ * →sort(Φ(ZL, ZD, WD))

[0067] where Φ * is the feature comprehensive value from best to worst. The feature extraction module also needs to determine the dimension of the extracted features. High-dimensional features provide more information but may lead to overfitting. Therefore, a trade-off needs to be made between information gain and computational complexity, and at the same time, it must be able to quickly extract features under real-time requirements to trigger an alarm or take measures in a timely manner.

[0068] Preferably, the data obtained by the machine learning module from the sensor and the features extracted by the feature extraction module are used to train and deploy a machine learning model to monitor and identify the state of the wire rope. For machine learning, the above-mentioned feature comprehensive values from good to bad and labels are required, and the data set L is further expanded to make it identifiable:

[0069] L

[0070] ={(t 1 ,L 1 ,D filtered (L 1 ),Φ 1 ,y 1 ),(t 2 ,L 2 ,D filtered (L 2 ),Φ 2 ,y 2 ),…,(t K ,L K ,D filtered (L K ),Φ K ,y K )}

[0071] Among them, Φ 1 , Φ 2 , Φ K are the first, second, and Kth feature comprehensive values respectively, and y 1 , y 2 , y K are the first, second, and Kth data labels respectively. Define the machine learning model parameter as θ, and satisfy:

[0072]

[0073] Among them, in order to find the optimal machine learning model parameter θ * , define the number of iterations as g, and there is:

[0074]

[0075] Among them, θ g is the machine learning model parameter of the gth iteration, θ g+1 is the machine learning model parameter of the (g + 1)th iteration, and α is the learning rate. In the present invention, the F2 score is used to evaluate the quality of the machine learning model, and it is defined as:

[0076]

[0077] Among them, ε is a regulation factor, TP is a positive class sample, FP is a negative class sample, Recall is a recall rate, and the larger ε is, the higher the degree of attention of the model to the recall rate is.

[0078] Specifically, the real-time monitoring module continuously monitors the status of the wire rope, triggers an alarm and takes necessary measures when an abnormal situation is detected, receives real-time data streams from the sensor module and the data preprocessing module, and the data streams include data of key parameters such as tension, vibration, and temperature. It quickly processes the data from the sensors to detect any abnormal patterns related to the status of the wire rope.

[0079] Specifically, the real-time monitoring module of the present invention can define a series of thresholds and rules to determine whether the data exceeds the normal range. If a certain threshold is exceeded, the system issues an alarm. Once the real-time monitoring module detects an abnormality or a violation of the rules, it will trigger an alarm, and the alarm can be in various ways, including sound alarms, visual cues, text messages, and email notifications, so that the operation and maintenance personnel can take timely actions. And it has a user interface for the operation and maintenance personnel to monitor the status of the wire rope in real time. The interface usually provides visual charts, real-time images, and data trends to help the operators quickly identify problems.

[0080] Specifically, the data analysis and maintenance advice module is used to analyze the monitoring data, generate reports, and provide maintenance advice. It deeply analyzes the data collected from the sensor module and the real-time monitoring module to identify potential problems and trends. The data analysis and maintenance advice module analyzes the data to identify long-term trends, and based on historical data, this module performs predictive analysis to help predict future maintenance needs and possible failures. The data analysis and maintenance advice module generates detailed reports, provides information about the status of the wire rope, can effectively manage and store a large amount of monitoring data and analysis results for subsequent query, retrieval, and report generation, and can also recommend maintenance and repair plans to help plan maintenance work in advance and reduce the risk of sudden failures.

[0081] Advantages of the present invention: Traditional regular inspection and maintenance methods often have difficulty in detecting transient problems in the equipment state. However, the real-time monitoring module of the present invention can achieve continuous monitoring of the equipment state. Once the equipment shows abnormalities, the system can quickly take measures, reducing the equipment failure rate. By promptly discovering and solving problems, the present invention helps to improve the reliability of mine hoisting equipment and reduces the impact of equipment failures on production. Traditional maintenance methods usually rely on regular inspections and maintenance, which not only consume a large amount of manpower and time but also may increase costs due to over-frequent or insufficient maintenance. The present invention realizes more scientific and targeted maintenance through intelligent monitoring, data analysis, and maintenance advice modules. The system can provide maintenance advice according to the actual situation. In addition, through preventive maintenance, the present invention can also avoid emergency repair costs caused by equipment failures, further reducing maintenance expenditures. Innovations of the present invention: In the sensor module, the window function proposed in the present invention uses α as a parameter to control the main lobe width and sidelobe suppression. If α → 0, it will result in a narrower main lobe but increase the amplitude of the sidelobe. N is the window length, and the present invention continuously updates and expands the data set under actual load in the improvement of the module, making it include actual load, sensor output value, time, feature comprehensive value, and label. In the process of finding the minimum value of the loss function Loss, the transition probability proposed in the present invention is more directive compared to the transition probability of heuristic algorithms such as the roulette method. This is because the present invention constrains the data fermentation degree to make it obey the function change. Since the mine hoisting steel wire rope has practical physical significance, the related tension, vibration, and temperature parameters are all closely related to the mine hoisting steel wire rope. And considering the information attenuation of the sensor transmission (such as multipath fading, hardware deviation), the present invention updates the data fermentation degree to make the search factor more directive in finding the optimal solution; in addition, the present invention constructs a feature comprehensive value, considering the influence of actual parameters on the mine hoisting steel wire rope, on the one hand, reducing the calculation load of the system for multiple parameters to best describe the state of the steel wire rope. Among them, tension has sample correlation in the field involved in the present invention, vibration has time-frequency characteristics, and faster calculation results can be obtained by extending to frequency domain analysis, while temperature has temporality, and wavelet transform is used to capture different frequency components and change trends in the temperature signal; finally, the present invention builds a machine learning module. For performance evaluation, the present invention proposes the F2 score. Compared with the traditional F1 score, the F2 score proposed in the present invention is more sensitive to unbalanced data sets. Especially in unbalanced data sets where the number of positive class samples is far less than the number of negative class samples, the F2 score can better reflect the model's attention to the minority class than the F1 score. It attaches more importance to the performance of the model in the positive class and reduces the impact of false negatives.The F2 score achieves a trade-off among positive class samples, negative class samples, and recall by adjusting the ε value (the ε parameter of the F2 score is greater than 1). A larger ε value emphasizes recall more, while a smaller ε value emphasizes precision more, making the F2 score a flexible evaluation metric that can be adjusted according to application requirements. Through the real-time monitoring and automated response functions of the present invention, the system can take timely measures when abnormal situations are detected, avoiding potential accident risks and reducing the possibility of production interruptions. In addition, through intelligent maintenance suggestions, the system can minimize the production interruption time during maintenance, improve production efficiency, and ensure the smooth execution of the production plan. The present invention makes full use of big data analysis technology. By analyzing a large amount of data, valuable information can be mined from it. This provides more data support for mine management and decision-making. Managers can formulate production plans, maintenance strategies, and resource allocations more scientifically based on the analysis results provided by the system. This helps improve the accuracy and efficiency of decision-making, provides a solid foundation for the long-term development and sustainability of the mine, leads the wave of intelligent production, improves competitiveness, and promotes the development of the industry.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting the protection scope of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. Intelligent detection system for mine hoisting steel ropes, characterized in that, it includes an integrated sensor module, a data preprocessing module, a feature extraction module, a machine learning module, a real-time monitoring module, and a data analysis and maintenance recommendation module. The sensor module is responsible for collecting data on key parameters of the steel rope in real time, including tension, vibration, and temperature, providing the basic data for subsequent analysis. The data preprocessing module filters, calibrates, and normalizes the raw data, improving the quality and reliability of the data; the feature extraction module extracts useful features from the massive data through advanced signal processing techniques, providing support for machine learning modeling. The machine learning module trains the model using historical data to achieve anomaly detection and predictive analysis, thus improving the accuracy of the steel rope state; the real-time monitoring module continuously monitors the steel rope state, promptly senses abnormal situations, and takes automated responses according to predetermined rules; the data analysis and maintenance recommendation module provides detailed data analysis and maintenance recommendations through in-depth analysis of the data, helping the operation and maintenance personnel better understand the data, identify problems, and plan maintenance work, reducing the maintenance cost; The raw data collected from the sensor module is processed and optimized by the data preprocessing module to ensure that high-quality input can be obtained for data analysis and algorithms. The data preprocessing module uses filtering techniques to remove noise and interference in the sensor data. This noise comes from the sensor itself, electromagnetic interference, and environmental vibration factors. Based on the Bessel function, a window function ω(n) suitable for the intelligent detection system of mine hoisting steel ropes is proposed, which can balance the main lobe width and sidelobe suppression and can achieve fine spectral control to overcome the noise or signal attenuation from the sensor itself, electromagnetic interference, and environmental vibration. The proposed window function ω(n) is expressed as follows: where n is the sample index of the window function, α is the parameter for controlling the main lobe width and sidelobe suppression. If α→0, it will result in a narrower main lobe but increase the amplitude of the sidelobe. N is the window length, and k is the index number; if there is a calibration deviation in the sensor, the data preprocessing module corrects the sensor output by applying a calibration coefficient to ensure the consistency between the sensor measurement value and the actual physical quantity, and constructs a polynomial model: D calibrated D(t) = a 0 + a 1 L(t) + a 2 L 2 (t) +... + a k L k (t); Among them, D calibrated (t) is the sensor output value, a 0 , a 1 , a 2 ,..., a k are calibration coefficients to be estimated, L is the actual load. To estimate the calibration coefficients, by collecting the sensor measurement values D filtered under the known load L, the data set under the actual load is: L = {(L 1 , D filtered (L 1 ))), (L 2 , D filtered (L 2 ))),..., (L K , D filtered (L K )))} where K is the number of data points collected, L 1 , L 2 ,..., L K are the 1st, 2nd,..., Kth actual loads respectively. Define the loss function To find the minimum value of the loss function Loss, define the search factor set S. The s-th search factor satisfies s ∈ S. The position of the search factor is represented by x s . Select the next position through a probability distribution The probability distribution satisfies the following formula: Among them, is the transition probability, x is the position in the search space, τ() is the data fermentation degree, η() is the heuristic factor, a and b are hyperparameters, length() is the length function used to calculate the length of a set. In each iteration, the data fermentation degree is updated according to the search path of the search factor and the loss function value: Among them, γ is the data residue degree, L(x s ) is the loss function value of the search factor s at the position x. γτ(x) takes into account the decay of the data fermentation degree. For the data fermentation degree, it should satisfy the information equivalence with the data entropy value. For τ(x s ) the data fermentation degree of the search factor s at the position x should satisfy: Through the updated data fermentation degree, the search factor can iterate to the optimal solution faster, obtaining the minimum value of the loss function. In some cases, sensor data may be lost or incomplete. Expand and update the data set L under the actual load to make it have time series: L = {(t 1 , L 1 , D filtered (L 1 ))), (t 2 , L 2 , D filtered (L 2 ))),..., (t K , L K , D filtered (L K )))}; The data preprocessing module fills in the missing part of the data through the following formula to maintain the continuity of the data: where L K-1 and L K+1 are the (K - 1)-th and (K + 1)-th actual loads respectively, D^ is the missing data value, t (+1) is the next time point, and t (-1) is the previous time point. For different application requirements, the data sampling rate needs to be adjusted. The data preprocessing module reduces or increases the data sampling rate to meet the requirements of subsequent processing steps, thereby reducing the computational load or increasing the data resolution. The data preprocessing module stores the preprocessed data in the database and records logs to track data quality and analyze historical data; The feature extraction module first needs to select appropriate features, including features of tension, vibration, and temperature, to best describe the state of the steel rope. Among them, tension has sample correlation in the involved field, vibration has time-frequency characteristics, and faster calculation results can be obtained by extending to frequency domain analysis, while temperature has time series. Wavelet transform is used to capture different frequency components and change trends in the temperature signal. In order to directly represent the best state of the features of tension, vibration, and temperature, the feature comprehensive value is defined as Φ, satisfying: Among them, ZL, ZD, and WD represent tension, vibration, and temperature respectively, and fz L is the tension function, fZ D is the vibration function, f WD is the temperature function, Δs is the increment of s, w ZL is the tension weight, w ZD is the vibration weight, w WD is the temperature weight, i * is the imaginary factor, C s，λ is the wavelet coefficient. The sorting function sort() is used to sort the feature comprehensive value Φ, and the feature comprehensive values from good to bad are obtained: Φ * →sort(Φ(ZL, ZD, WD)); Among them, Φ * is the comprehensive feature value from excellent to poor. The feature extraction module also needs to determine the dimension of the extracted features. High-dimensional features provide more information but may lead to overfitting. Therefore, it is necessary to balance between information gain and computational complexity, and at the same time, it must be able to quickly extract features under real-time requirements to trigger an alarm or take measures in a timely manner; The data obtained by the machine learning module from the sensors and the features extracted by the feature extraction module are used to train and deploy a machine learning model to monitor and identify the state of the wire rope. For machine learning, the above-mentioned feature comprehensive values from excellent to poor and labels are required, and the data set L is further expanded to make it identifiable: L＝ (t 1 , L 1 , D filtered (L 1 ), Φ 1 , y 1 ), (t 2 , L 2 , D filtered (L 2 ), Φ 2 , y 2 ),..., (t K , L K , D filtered (L K ), Φ K , y K )}; Among them, Φ 1 , Φ 2 , Φ K are the comprehensive feature values of the 1st, 2nd, and Kth features respectively, y 1 , y 2 , y K are the data labels of the 1st, 2nd, and Kth respectively. Define the machine learning model parameter as θ, satisfying: Among them, in order to find the optimal machine learning model parameter θ * , define the number of iterations as g, and there is: where θ g is the machine learning model parameter for the g-th iteration, and θ g+1 is the machine learning model parameter for the (g + 1)-th iteration. α is the learning rate. The goodness of the machine learning model is evaluated using the F2 score, and it is defined as: where ε is an adjustment factor, TP is the positive class sample, FP is the negative class sample, and Recall is the recall rate. The larger ε is, the higher the model's emphasis on the recall rate.

2. The intelligent detection system for mine hoisting wire ropes according to claim 1, characterized in that, The sensor module collects key parameters and data during the operation of the wire rope so that the system can monitor the state of the wire rope in real time. The sensors involved include a tension sensor, a vibration sensor, and a temperature sensor. The tension sensor is used to measure the tension of the wire rope to detect changes in tension; The vibration sensor is used to detect the vibration of the wire rope to identify changes in vibration frequency and amplitude; the temperature sensor is used to monitor the temperature of the wire rope to detect problems caused by temperature rise; The sensor module has the function of data acquisition, and can collect the data generated by the sensors in real time, providing high-precision and high-resolution data to ensure accurate monitoring of the state of the wire rope. Higher precision and resolution help the system better detect small changes and anomalies. In the intelligent detection system for mine hoisting wire ropes, the data precision needs to have the following characteristics: high precision, precision calibration, number of digits and accuracy, signal-to-noise ratio. High precision means that the sensor module should provide high-precision data, making the measured value of the sensor close to the actual physical quantity with small errors. High-precision sensors can better detect small changes in the wire rope, thereby improving the sensitivity of the system; Precision calibration means that the sensor needs to be calibrated regularly, and the output of the sensor is corrected to keep it consistent with the actual physical quantity; The number of digits and accuracy mean that the data resolution is usually expressed in the number of digits. The sensor is 16 bits and can represent 65,536 different values, thereby providing a larger measurement range and more accurate measurement results; The signal-to-noise ratio refers to the ratio of the signal to the noise. A high signal-to-noise ratio means that the sensor can better distinguish the signal from the noise, thereby improving the reliability of the data.

3. The intelligent detection system for mine hoisting wire ropes according to claim 1, characterized in that, The real-time monitoring module continuously monitors the state of the wire rope and triggers an alarm and takes necessary measures when an abnormal situation is detected. It receives the real-time data stream from the sensor module and the data preprocessing module. The data stream includes data of key parameters such as tension, vibration, and temperature, and quickly processes the data from the sensors to detect any abnormal patterns related to the state of the wire rope.

4. The intelligent detection system for mine hoisting wire ropes according to claim 3, characterized in that, The described real-time monitoring module can define a series of thresholds and rules to determine whether the data exceeds the normal range. If a certain threshold is exceeded, the system issues an alarm. Once the real-time monitoring module detects an abnormality or a violation of the rules, it will trigger an alarm, and the alarm can be in various ways, including audible alarms, visual cues, text messages, and email notifications, so that the operation and maintenance personnel can take timely actions. And it has a user interface for the operation and maintenance personnel to monitor the status of the wire rope in real time. The interface usually provides visual charts, real-time images, and data trends to help the operators quickly identify problems.

5. The intelligent detection system for mine hoisting wire ropes according to claim 1, characterized in that the data analysis and maintenance recommendation module is used to analyze the monitoring data, generate reports, and provide maintenance recommendations. It deeply analyzes the data collected from the sensor module and the real-time monitoring module to identify potential problems and trends. The data analysis and maintenance recommendation module analyzes the data to identify long-term trends, and based on historical data, this module performs predictive analysis to help predict future maintenance needs and possible failures. The data analysis and maintenance recommendation module generates detailed reports, provides information about the status of the wire rope, can effectively manage and store a large amount of monitoring data and analysis results for subsequent query, retrieval, and report generation, and can also recommend maintenance and repair plans to help plan maintenance work in advance and reduce the risk of sudden failures.

Images