Comprehensive protection method and system for mining belt conveyor

Abstract

The invention relates to a comprehensive protection method and system for a mining belt conveyor, and relates to the technical field of mining conveyor protection, and the method comprises the steps: extracting a first element, collecting the initial service data of the first element, taking an operation control parameter and a conveying load parameter as input, taking a preset state standard threshold as supervision, and training a first element standard threshold configurator; processing operation control and transmission load monitoring parameters through the configurator, and outputting a first state standard threshold value; taking the service duration of the first component and the transportation environment information as constraints, retrieving small sample first component health service data and performing statistics to obtain a second state standard threshold value; and obtaining a correction state standard threshold value by taking an intersection of the two, and executing comprehensive protection management. The method solves the problems that in protection of a traditional mining belt conveyor, due to the fact that equipment has different threshold states under different service states and operation states, due to the fact that a threshold value with fixed configuration is adopted, precision is insufficient, control errors are prone to occurring, and then safety production is affected.

Description

Technical Field

[0001] The present application relates to the field of mining conveyor protection, and in particular to a comprehensive protection method and system for mining belt conveyors. Background Art

[0002] With the increasing mechanization of mining operations, the stable operation of mining belt conveyors has become a key factor in ensuring production efficiency. Accurately implementing comprehensive protection is a key technical requirement for avoiding equipment failures. Currently, traditional mining belt conveyor protection methods use fixed thresholds for monitoring and control. This makes it difficult to adapt to the dynamic threshold states of the equipment at different service stages and under different operating loads, resulting in insufficient threshold accuracy and frequent control errors. Summary of the Invention

[0003] In order to solve the above technical problems, the present application provides a comprehensive protection method and system for mining belt conveyors, which improves the current situation in traditional protection where the fixed threshold cannot adapt to the threshold differences under different service and operating conditions of the equipment, resulting in insufficient threshold accuracy, frequent control errors, and impact on production.

[0004] The embodiments of this application disclose the following technical solutions: In a first aspect, an embodiment of the present application provides a comprehensive protection method for a mining belt conveyor, the method comprising: Extract the first element of the mining belt conveyor, collect the initial service data of the first element, use the operation control parameters and conveying load parameters as input, and use the preset state standard threshold as supervision to train the first element standard threshold configurator; Processing the operation control monitoring parameter and the transport load monitoring parameter through the first component standard threshold value configurator to output a first state standard threshold value; Using the service life of the first component and the transportation environment information as constraints, a small sample of the first component's healthy service data is retrieved, statistics are performed, and the second status standard threshold is obtained; Take the intersection of the first state standard threshold and the second state standard threshold to obtain a modified state standard threshold and perform comprehensive protection management.

[0005] In a second aspect, an embodiment of the present application provides a comprehensive protection system for a mining belt conveyor, the system comprising: a configurator training module, for extracting the first element of the mining belt conveyor, collecting the initial service data of the first element, using the operation control parameters and the conveying load parameters as input, and using the preset state standard threshold as supervision, to train the first element standard threshold configurator; a first threshold output module, configured to process the operation control monitoring parameter and the transport load monitoring parameter through a first component standard threshold configurator, and output a first state standard threshold; A second threshold statistics module is used to retrieve a small sample of healthy service data of the first component based on the service time of the first component and the transportation environment information, perform statistics, and obtain a second state standard threshold; The modified threshold comprehensive protection module is used to take the intersection of the first state standard threshold and the second state standard threshold, obtain the modified state standard threshold, and perform comprehensive protection management.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application proposes a comprehensive protection method and system for mining belt conveyors. It realizes accurate and efficient protection of mining belt conveyors by clustering components to integrate similar state data, constructing a dynamic threshold configurator to output real-time thresholds, combining service time with environmental information to calculate health thresholds, taking the intersection of the two to obtain a modified threshold, and performing collaborative operations of protection management. First, the first component of the belt conveyor is extracted, and the initial components are clustered in the same state based on the initial service data. The components are integrated to form a component cluster to expand the amount of training data and reduce computing power requirements; then the initial service data of the first component is collected, and a fully connected neural network is trained to obtain a first-level standard threshold configurator. After verification and optimization, it is determined to be the first component standard threshold configurator, which is used to process real-time monitoring parameters and output the first state standard threshold; at the same time, with the component service time and transportation environment as constraints, a small sample of healthy service data is retrieved, and the second state standard threshold is obtained through statistical analysis; finally, the intersection of the two thresholds is taken as the modified state standard threshold, and comprehensive protection management is performed accordingly.

[0007] The technical solution of this application solves the control error problem caused by insufficient accuracy of fixed thresholds and inability to adapt to different service states and operating conditions in the protection of traditional mining belt conveyors through multiple steps such as integrating similar data through fusion element clustering, generating dynamic thresholds of neural network models, supplementing health thresholds with small sample statistics, and determining corrected state standard thresholds based on intersection. It realizes dynamic adaptation and precise control of protection thresholds, and provides technical support for ensuring the safe and stable operation of mining belt conveyors. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 This is a schematic diagram of the structure of the comprehensive protection system for mining belt conveyors provided in an embodiment of the present application.

[0010] Figure 2A schematic flow chart of the comprehensive protection method for a mining belt conveyor provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The present application provides a comprehensive protection method and system for mining belt conveyors, which is used to solve the technical problem in the prior art that mining belt conveyors have different threshold states in different service states and operating states, and the threshold accuracy fixed by the user is insufficient, resulting in control errors, which in turn affects production.

[0012] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0013] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically specified.

[0014] In the description of this application, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art will recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0015] Example 1, as shown in the attached Figure 1 and attached Figure 2 As shown, the present application provides a comprehensive protection system for a mining belt conveyor and a comprehensive protection method for a mining belt conveyor.

[0016] As attached Figure 2 As shown, the comprehensive protection method for mining belt conveyors includes the following steps: S110: Extracting the first component of the mining belt conveyor, collecting initial service data of the first component, using operation control parameters and conveying load parameters as input, and using a preset state standard threshold as supervision, to train a first component standard threshold configurator, where initial service indicates a service time of less than or equal to 3 months and no failures, and the preset state is a continuous variable; In the embodiment of the present application, in the scenario of comprehensive protection of mining belt conveyors, in order to realize the dynamic fluctuation of the threshold value with the control state and load parameters, it is necessary to construct a first element standard threshold configurator to accurately adapt to the protection requirements of the equipment under different states.

[0017] Specifically, the initial service data of the mining belt conveyor is first obtained, and the equipment is divided into the first to Nth initial components according to the parts table, where N is the total number of parts.

[0018] Among them, the initial service data of the mining belt conveyor contains the status record data of the corresponding mining belt conveyor. Using these status record data as the core basis for initial component clustering can accurately capture the status correlation of different components during operation.

[0019] Furthermore, these initial components are clustered in the same state based on the state record data. That is, the consistency of the state record data between the components is counted. When the ratio of the same state frequency to the total number of records meets the standard, they are clustered into one category to form the first to Qth clusters of initial components (Q is the total number of clusters). Any combination of cluster components constitutes the first component.

[0020] Furthermore, for the initial components in the first cluster of the first component, their initial service data sets (service time ≤ 3 months, no faults) are collected, and each data in the data set includes operation record control parameters, transmission record load parameters and status record data.

[0021] On this basis, the initial service data set is traversed, and the data with consistent operation record control parameters and transmission record load parameters are divided into the same group. The same state box plot analysis is performed on each group of status record data to obtain the status identification interval (i.e. the upper and lower limit constraint interval of the box), and the grouped status record data are associated with the status identification interval and stored to form the initial service data of the first component.

[0022] Furthermore, with the operation control parameters and the transport load parameters as inputs and the preset state standard threshold as supervision, a fully connected neural network is trained to obtain a first-level standard threshold configurator.

[0023] At the same time, the first-level standard threshold configurator is verified. If the modulus value of any one of the first to M-th state attribute verification residual vectors (M is the total number of state attributes) is ≥ the residual modulus threshold, the network topology is updated based on the residual vector and the training is cyclical; otherwise, the first-level standard threshold configurator is directly set as the first element standard threshold configurator.

[0024] This step provides the core configuration foundation for the subsequent generation of dynamic thresholds through the technical solutions of component clustering, data processing and neural network training, ensuring that the thresholds can be dynamically adjusted with the control status and load parameters, thereby ensuring the accuracy of the comprehensive protection of mining belt conveyors.

[0025] Step S110 of the method provided in the embodiment of the present application includes: Obtaining initial service data of a mining belt conveyor, wherein the initial service data of the mining belt conveyor includes state record data of the mining belt conveyor; According to the parts list, the mining belt conveyor is disassembled to obtain the first initial component to the Nth initial component, where N represents the total number of parts; Based on the mining belt conveyor state record data, clustering the first initial element to the Nth initial element in the same state to obtain the first cluster initial element to the Qth cluster initial element, where Q represents the total number of clusters; The first element is obtained by combining any cluster of elements from the first cluster of initial elements to the Qth cluster of initial elements.

[0026] obtaining an initial element in a first cluster of the first element, and collecting an initial service data set of the initial elements in the first cluster, wherein any piece of initial service data of the initial elements in the first cluster in the initial service data set includes an operation record control parameter, a transport record load parameter, and a status record data; Traversing the initial component initial service data set in the first cluster, dividing the initial component initial service data in the first cluster having the same operation record control parameters and the same transmission record load parameters into the same group, and obtaining a plurality of groups of initial component initial service data in the first cluster; Traversing the plurality of groups of state record data of the initial service data of the initial components in the plurality of first clusters to perform a same-state box plot analysis to obtain a plurality of state identification intervals, wherein the state identification intervals are upper and lower limit constraint intervals of the state record data box; The plurality of groups of initial component initial service data in the first cluster and the plurality of state identification intervals are associated and stored, and added to the first component initial service data.

[0027] Taking the operation control parameters and the transport load parameters as input and the preset state standard threshold as supervision, the fully connected neural network is trained to obtain the first-level standard threshold configurator; Performing verification on the primary standard threshold configurator to obtain a first state attribute verification residual vector up to an M-th state attribute verification residual vector, where M represents the total number of state attributes; When any residual vector modulus value of the first state attribute verification residual vector until the M-th state attribute verification residual vector is greater than or equal to the residual modulus threshold, updating the first-level standard threshold configurator network topology based on the first state attribute verification residual vector until the M-th state attribute verification residual vector, and performing cyclic training; Otherwise, the primary standard threshold configurator is set to the first component standard threshold configurator.

[0028] In the embodiment of the present application, in order to achieve dynamic adaptation of the protection threshold of the mining belt conveyor, it is necessary to construct a first component standard threshold configurator through component clustering, data integration and model training to accurately capture the threshold characteristics of the equipment in different states and overcome the problem of insufficient accuracy of the fixed threshold.

[0029] Specifically, the initial service data of the mining belt conveyor is first obtained. The initial service data includes the status record data of the corresponding mining belt conveyor, and the mining belt conveyor is divided into the first to Nth initial components according to the component list (N is the total number of components).

[0030] Among them, the accessories table is a list of components of the mining belt conveyor, which includes the name, model, specification and assembly relationship of each core component of the equipment. It is used as a standardized basis for equipment disassembly to ensure that the mining belt conveyor can be fully and accurately disassembled into independent initial components.

[0031] Furthermore, the first to Nth initial elements are clustered in the same state based on the obtained state record data to obtain the first cluster initial elements to the Qth cluster initial elements (Q is the total number of clusters).

[0032] In the method provided in an embodiment of the present application, the step of “clustering the first initial element to the Nth initial element in the same state based on the mining belt conveyor state record data to obtain the first cluster of initial elements to the Qth cluster of initial elements” includes: Extracting first mining belt conveyor state record data from the mining belt conveyor state record data, wherein the first mining belt conveyor state record data includes first initial component state record data and second initial component state record data; When the first initial element state record data and the second initial element state record data are consistent, the same state frequency of the first initial element and the second initial element is increased by one, otherwise the same state frequency remains unchanged, wherein the initial value of the same state frequency is equal to 0; Until the Yth mining belt conveyor state record data is extracted from the mining belt conveyor state record data, the same state frequency is updated, and the same state frequency is output, Y ≥ 30, where Y represents the total number of mining belt conveyor state record data; When the ratio of the same-state frequency to Y is greater than or equal to a ratio threshold, clustering the first initial component and the second initial component into one category; When all initial elements are clustered, the first cluster of initial elements is output until the Qth cluster of initial elements.

[0033] In an embodiment of the present application, in order to achieve accurate clustering of the initial components of the mining belt conveyor, it is necessary to perform same-state judgment based on the state record data, and clustering is achieved by statistically analyzing the state consistency frequency between the initial components to integrate the initial component data of similar states, laying the foundation for the subsequent training of the threshold configurator.

[0034] Specifically, first, the first mining belt conveyor state record data is extracted from the mining belt conveyor state record data, and the state record data includes state record information (such as temperature, vibration, operating parameters, etc.) of the first initial element and the second initial element.

[0035] For example, if the first initial element is a belt-driven roller, its state recording data includes a temperature of 42°C, a vibration amplitude of 0.6 mm / s, and a rotational speed of 980 r / min; the second initial element is a driven roller, its state recording data includes a temperature of 43°C, a vibration amplitude of 0.65 mm / s, and a rotational speed of 975 r / min.

[0036] Furthermore, the state record data of the first initial component and the state record data of the second initial component are compared. When the state record data of the first initial component and the state record data of the second initial component are consistent, the same state frequency of the first initial component and the second initial component is increased by one to continuously accumulate the number of times the two appear consistent in the state record data.

[0037] In the method provided in the embodiment of the present application, “when the first initial component state record data and the second initial component state record data are consistent” includes: Configure the state attribute consistency tolerance deviation for the state attribute set; When the state attribute deviations of the first initial component state record data and the second initial component state record data both meet the corresponding state attribute consistency fault tolerance deviation, the first initial component state record data and the second initial component state record data are considered to be consistent; Otherwise, it is considered that the first initial component state record data and the second initial component state record data are inconsistent.

[0038] In the embodiment of the present application, in order to achieve accurate judgment of the consistency of the initial component state of the mining belt conveyor, it is necessary to establish a quantitative comparison standard by configuring the consistency tolerance deviation of the state attributes to ensure the scientificity and accuracy of the same-state clustering, and provide a reliable basis for the formation of subsequent component clusters.

[0039] Specifically, the state attribute set of the first initial element and the second initial element is firstly clarified, and the state attribute set covers key monitoring indicators during the operation of each initial element, such as temperature, vibration amplitude, operating speed, pressure, etc.

[0040] Furthermore, for each state attribute, a corresponding consistency fault tolerance deviation is configured according to the physical characteristics, operation requirements and monitoring accuracy requirements of each initial component.

[0041] Specifically, the temperature attribute is usually configured with a tolerance range of ±2°C to adapt to ambient temperature fluctuations and sensor measurement errors; the vibration amplitude attribute is configured with a tolerance deviation of ±0.1mm / s to take into account the distinction between the normal vibration range and abnormal vibration of the equipment; the speed attribute is configured with a tolerance deviation of ±50r / min to adapt to small fluctuations in speed caused by load changes.

[0042] For example, the first initial component is a drive motor, whose state data includes a temperature of 45°C, a vibration amplitude of 0.7 mm / s, and a speed of 1000 r / min. The second initial component is a reduction gearbox, whose state data includes a temperature of 46°C, a vibration amplitude of 0.75 mm / s, and a speed of 980 r / min. The corresponding state attribute consistency tolerances are ±2°C for temperature, ±0.1 mm / s for vibration amplitude, and ±50 r / min for speed, respectively.

[0043] Furthermore, the fault tolerance deviation of each state attribute of the two initial components is calculated.

[0044] Specifically, the temperature deviation is 1°C (46°C-45°C), which is less than the fault tolerance range of ±2°C; the vibration amplitude deviation is 0.05mm / s (0.75mm / s-0.7mm / s), which is less than the fault tolerance range of ±0.1mm / s; the speed deviation is 20r / min (1000r / min-980r / min), which is less than the fault tolerance range of ±50r / min.

[0045] Since the fault tolerance deviations of all state attributes satisfy the corresponding consistency fault tolerance deviations, it is determined that the first initial component state record data and the second initial component state record data are consistent.

[0046] On the contrary, if the temperature record of the second initial component is 48°C, the temperature deviation is 3°C (48°C-45°C), which exceeds the tolerance range of ±2°C. At this time, regardless of whether the deviations of other attributes meet the standards, the status record data of the two are judged to be inconsistent.

[0047] This technical solution for consistency judgment based on multi-attribute fault-tolerant deviation can not only avoid misjudgments caused by minor data fluctuations, but also accurately identify true state differences to ensure that the same-state clustering results can truly reflect the operational correlation of components.

[0048] Furthermore, after the consistency determination of the state attributes of the first initial component and the second initial component is completed, the same-state frequency needs to be updated according to the determination result.

[0049] Specifically, if the state attribute consistency is determined to be consistent, the same-state frequency of the two is increased by 1 based on the current value (the initial value is 0); if the state attribute consistency is determined to be inconsistent, the same-state frequency remains unchanged.

[0050] For example, if the first state attribute consistency is determined to be consistent, the same state frequency changes from 0 to 1; if the second state attribute consistency is still determined to be consistent, the same state frequency changes from 1 to 2, and so on.

[0051] Similarly, according to the above steps, the 2nd to Yth (Y ≥ 30) status record data are continuously extracted from the mining belt conveyor status record data, the state attribute consistency judgment is completed one by one, and the same-state frequency is updated. After traversing the Yth state record data, the final same-state frequency is output.

[0052] By traversing Y state records to obtain the final same-state frequency, we ensure sufficient statistical sample size and improve the reliability of the clustering results. For example, when Y = 30, 30 determinations and frequency updates are required to obtain the cumulative same-state frequency (e.g., 22).

[0053] On this basis, the ratio of the same-state frequency to Y is calculated. When the ratio is greater than or equal to a preset ratio threshold (such as 0.7), the first initial component and the second initial component are clustered into one category.

[0054] Among them, the preset ratio threshold is determined based on the historical operation correlation data of the initial components of the mining belt conveyor, the probability of coordinated fault occurrence, and equipment operation and maintenance experience, to ensure that the clustering results can effectively integrate components with closely correlated states and avoid incorrectly classifying components with weaker correlation into one category.

[0055] For example, if Y=30 and the frequency of the same state=22, the ratio of the two is 22 / 30≈0.73. Since 0.73>0.7 (preset ratio threshold), it is determined that the state attributes of the two are highly correlated and can be clustered into one category; if the frequency of the same state=20, the ratio of the two is 20 / 30≈0.67. Since 0.67<0.7, it is determined that the state attributes of the two are weakly correlated and no clustering is performed.

[0056] Similarly, following the same process, all initial components are clustered pairwise. First, the frequency statistics and clustering determination are performed for the first initial component and the third, fourth, and up to the Nth initial components. The same process is then repeated for the second, third, up to the N-1th initial components and any other unclustered components until all initial components are clustered.

[0057] Finally, after all initial components are clustered, the output is the initial components of the first cluster up to the Qth cluster, where Q represents the total number of clusters. For example, 15 initial components can be clustered into four clusters to integrate components with closely related state attributes, providing efficient units for subsequent data utilization and model training.

[0058] Furthermore, the first element can be obtained by combining any cluster of elements from the first cluster of initial elements to the Qth cluster of initial elements.

[0059] Among them, this cluster combination method can overcome the defect of less sample data of a single component. By integrating the service data of all components in the cluster, it can maximize the use of available data resources. At the same time, it can reduce the number of configurators that need to be trained, so that instead of training configurators for N initial components separately, after clustering, only configurators need to be trained for Q clusters (Q < N), which significantly reduces computing power requirements and improves operating efficiency.

[0060] Furthermore, after the first component is obtained, for the initial components in the first cluster contained therein, an initial service data set of the initial components in the cluster is collected.

[0061] The initial service data set must meet the conditions of an initial service time of ≤ 3 months and no failures. Each data entry includes operation record control parameters (such as motor operating voltage, belt operating speed, etc.), conveying record load parameters (such as hourly conveying volume, average material density, etc.), and status record data (such as component temperature, vibration frequency, etc.).

[0062] Furthermore, by traversing the initial service data set of the initial components in the first cluster, the data with consistent operation record control parameters and transmission record load parameters are divided into the same group to ensure that the data in the same group are under the same operating conditions and load conditions, providing a unified benchmark for the analysis of subsequent status record data.

[0063] For example, all data of "motor operating voltage 380V, hourly delivery volume 200 tons" are grouped together, and several groups of initial component initial service data in the first cluster are obtained.

[0064] On this basis, a same-state box plot analysis is performed on the status record data of each group of initial service data to extract the status distribution characteristics of this group of data under the same working conditions through statistical methods and determine the normal fluctuation range of the status record data.

[0065] Specifically, by calculating the quartiles of the initial service data, the upper and lower limits of the box are determined, which serve as the status identification range. This status identification range covers 75% of the status records in this set of initial service data and can effectively represent the health status benchmark of the component under this operating condition.

[0066] For example, the box plot analysis results of a certain set of temperature data show that the status identification interval is 35-48°C, which means that the initial service data of this set presents stable health status characteristics within this status identification interval.

[0067] Furthermore, the obtained initial service data of the initial components in the first cluster are associated with the corresponding state identification intervals and stored to form complete initial service data of the first components, providing structured and labeled basic data for subsequent model training.

[0068] On this basis, with the operation control parameters and transportation load parameters as input and the preset state standard threshold as supervision, a fully connected neural network is trained to obtain a first-level standard threshold configurator.

[0069] Specifically, a fully connected neural network framework consisting of an input layer, a hidden layer, and an output layer is first constructed. The number of neurons in the input layer matches the total dimension of the operating control parameters and the transport load parameters.

[0070] For example, if the input includes four parameters: motor speed, belt tension, material weight, and conveying speed, the input layer is set with 4 neurons; the hidden layer adopts a 2-3 layer structure, and the number of neurons in each layer is set to 1.5-2 times that of the input layer. The ReLU activation function is used to enhance the nonlinear fitting ability of the network to capture the complex mapping relationship between parameters and thresholds; the number of neurons in the output layer corresponds to the dimension of the preset state standard threshold, such as the threshold of state attributes such as temperature, vibration, and pressure, and directly outputs the threshold prediction value of each state attribute.

[0071] Furthermore, the initial service data of the first component is divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0072] For example, 400 groups of data are selected from 500 groups as training sets for model parameter learning, 50 groups are selected as validation sets for adjusting hyperparameters, and 50 groups are selected as test sets for final performance evaluation to ensure that the data distribution covers different combinations of operating control parameters and transport load parameters to improve the generalization ability of the model.

[0073] Furthermore, during the model training phase, the operational control parameters and transport load parameters from the training set were input into the network. The weights and bias parameters were optimized using a backpropagation algorithm to gradually reduce the mean square error between the output prediction threshold and the preset state standard threshold. The initial learning rate was set at 0.005, and the learning rate was decayed to half of the previous value every 100 iterations. The Adam optimizer was also used to accelerate convergence.

[0074] At the same time, the validation set is used to monitor model performance in real time. If the mean squared error of the validation set decreases by less than 0.01 (a threshold unit) for 20 consecutive rounds, the model is considered converged and training is stopped. For example, if the mean squared error of the validation set decreases from an initial 1.2 to 0.05 and remains between 0.048 and 0.052 for 20 consecutive rounds, the model is considered converged.

[0075] During the testing phase, the test set is input into the trained network. If the mean absolute error between the predicted threshold and the preset state standard threshold is less than 0.03 (threshold unit), the first-level standard threshold configurator is judged to be qualified.

[0076] For example, in a certain test sample, the preset temperature threshold is 50°C, the model output predicted temperature threshold is 49.8°C, and the deviation is 0.2°C; the preset vibration threshold is 1.0mm / s, the model output vibration threshold is 0.98mm / s, and the deviation is 0.02mm / s. The overall average deviation is 0.025, which meets the accuracy requirements.

[0077] Finally, the trained first-level standard threshold configurator can quickly output the corresponding state standard threshold prediction value based on the input real-time operation control parameters and transportation load parameters, providing basic model support for subsequent threshold verification and correction.

[0078] Furthermore, the constructed primary standard threshold configurator is verified to obtain the first state attribute verification residual vector until the Mth state attribute verification residual vector, where M is the total number of state attributes.

[0079] Specifically, after the operation control parameters and transport load parameters of the verification set are input into the first-level standard threshold configurator, the predicted threshold of each state attribute is output, and then it is subtracted from the preset state standard threshold to obtain the verification residual vector of each state attribute.

[0080] In the method provided in an embodiment of the present application, the step of “updating a primary standard threshold configurator network topology based on the first state attribute verification residual vector up to the Mth state attribute verification residual vector, and performing cyclic training when the modulus of any one of the residual vectors from the first state attribute verification residual vector to the Mth state attribute verification residual vector is greater than or equal to the residual modulus threshold” includes: When the modulus of the first state attribute verification residual vector is greater than or equal to the residual modulus threshold, the operation control parameter and the transport load parameter are used as input, and the first state attribute verification residual vector is used as supervision to train a fully connected neural network to obtain a first state attribute output corrector, which is integrated into the first state attribute output node of the primary standard threshold configurator. The integration rule is to add the outputs of the first state attribute output corrector and the first state attribute output node and use the sum as the final output of the first state attribute output node; Until the modulus value of the Mth state attribute verification residual vector is greater than or equal to the residual modulus threshold, the operation control parameter and the transport load parameter are used as input, and the Mth state attribute verification residual vector is used as supervision to train a fully connected neural network to obtain an Mth state attribute output corrector, which is integrated into the Mth state attribute output node of the primary standard threshold configurator, wherein the integration rule is to add the outputs of the Mth state attribute output corrector and the Mth state attribute output node as the final output of the Mth state attribute output node; Obtain the first-level standard threshold configurator network topology and perform cyclic training.

[0081] In an embodiment of the present application, in order to further improve the prediction accuracy of the first-level standard threshold configurator, when there is a situation where the state attribute verification residual exceeds the standard, it is necessary to optimize the configurator by targeted training of the first state attribute output corrector and integrating it into the original network topology to ensure that the threshold value it outputs can more accurately adapt to the actual operating status of the device.

[0082] Specifically, when it is detected that the modulus value of the first state attribute verification residual vector is greater than or equal to a preset residual modulus threshold (such as 0.5), the operating control parameters and the transmission load parameters are first used as input data, and the first state attribute verification residual vector is used as a supervision signal to train a fully connected neural network to obtain a first state attribute output corrector.

[0083] Among them, the network structure of the first state attribute output corrector adopts a simplified design, that is, the number of neurons in the input layer is consistent with the dimension of the input parameters, the hidden layer is set to 1-2 layers (the number of neurons in each layer is 1-1.5 times that of the input layer), and the output layer has only 1 neuron, which is used to output the residual correction value of the corresponding state attribute.

[0084] At the same time, during the training process, the weight parameters of the corrector are continuously optimized through the back-propagation algorithm, so that the mean square error between the residual prediction value output by the corrector and the actual residual is gradually reduced until the model converges (for example, the error of the validation set decreases by less than 0.001 for 15 consecutive rounds).

[0085] After the training is completed, the first state attribute output modifier is integrated into the first state attribute output node of the first-level standard threshold configurator.

[0086] Specifically, the residual correction value output by the first state attribute output corrector is added to the prediction threshold of the original output node, and the obtained result is used as the final output threshold of the state attribute output node.

[0087] For example, the temperature threshold predicted by the original output node is 50°C, and the residual correction value output by the first state attribute output corrector is -0.3°C. The final output temperature threshold is 49.7°C (50°C-0.3°C), thereby achieving accurate correction of the original output node prediction value.

[0088] Similarly, following the same steps as above, the second to M-th state attribute verification residual vectors are checked in sequence.

[0089] Specifically, if the modulus value of the Kth (1<K≤M) state attribute verification residual vector is greater than or equal to the residual modulus threshold, the Kth state attribute output corrector is trained with the same input operation control parameters and transmission load parameters and the corresponding state attribute verification residual vector as supervision, and is integrated into the Kth state attribute output node of the first-level standard threshold configurator. The integration rule is also to add the residual correction value output by the Kth state attribute output corrector and the original node prediction threshold output as the final output threshold.

[0090] For example, in the case where the vibration attribute residual exceeds the standard, a vibration output corrector is trained, and the correction value output by the corrector is added to the original vibration threshold prediction value to serve as the final vibration threshold.

[0091] On this basis, when all state attributes with excessive residuals have completed the training and integration of the corresponding correctors, the updated first-level standard threshold configurator network topology is obtained.

[0092] Furthermore, the updated first-level standard threshold configurator network topology is retrained using the training set. This involves inputting the training set data into a fully connected neural network, and using a backpropagation algorithm to simultaneously optimize the parameters of the original configurator and each modifier, resulting in a continuous decrease in the mean squared error between the predicted threshold output by the overall model and the preset state standard threshold.

[0093] The termination condition of the cyclic training is set as follows: after 20 consecutive rounds of training, the modulus values ​​of all state attribute verification residual vectors are less than the residual modulus threshold. At this time, the training is stopped to obtain the optimized first-level standard threshold configurator.

[0094] For example, a certain level of standard threshold configurator finds during verification that the residual vector moduli of temperature (first state attribute) and vibration (second state attribute) are 0.6 and 0.55, respectively, both exceeding the residual modulus threshold of 0.5.

[0095] Furthermore, for the temperature attribute, a temperature output corrector was trained, taking parameters such as motor speed and material weight as input and outputting a temperature residual correction value. For the vibration attribute, a vibration output corrector was trained, taking the same parameters as input and outputting a vibration residual correction value. After integrating the two state attribute output correctors into their corresponding output nodes and retraining the configurator, the residual vector moduli for temperature and vibration were ultimately reduced to 0.3 and 0.25, respectively, meeting the accuracy requirements.

[0096] Through the targeted correction and cyclic training method of the above steps, the shortcomings of the original configurator in predicting specific state attributes can be effectively compensated, the overall accuracy of threshold prediction can be significantly improved, and a more reliable judgment basis can be provided for the dynamic protection of mining belt conveyors.

[0097] On the contrary, if all residual vector moduli values ​​from the first state attribute verification residual vector to the Mth state attribute verification residual vector are less than the residual modulus threshold, there is no need to update the network topology of the first-level standard threshold configurator, and the first-level standard threshold configurator can be directly set as the first element standard threshold configurator.

[0098] At this point, the first-element standard threshold configurator has a high prediction accuracy and can accurately output the standard threshold of the current state of the adaptation equipment based on the input operation control parameters and conveying load parameters, providing a stable and reliable judgment basis for the comprehensive protection of mining belt conveyors.

[0099] For example, it has been verified that the residual vector modulus values ​​of all state attributes such as temperature and vibration of a certain level of standard threshold configurator are 0.3, which is less than the residual modulus threshold of 0.5. Therefore, it is directly determined as the first element standard threshold configurator, which can provide accurate threshold reference for protecting mining belt conveyors in real time.

[0100] S120: Processing the operation control monitoring parameter and the transport load monitoring parameter by the first component standard threshold value configurator, and outputting a first state standard threshold value; In the embodiment of the present application, in the scenario of comprehensive protection of mining belt conveyors, the first element standard threshold configurator serves as the core module for dynamic threshold generation. Its core function is to output the first state standard threshold adapted to the current working conditions based on the real-time monitoring of equipment operating parameters, thereby providing an accurate judgment basis for equipment protection actions.

[0101] Specifically, when the mining belt conveyor is in actual operation, various sensors (such as deviation, coal accumulation, tearing, tension, speed, smoke, vibration, temperature, flame, etc.) will collect the equipment's operation control monitoring parameters (including motor speed, belt tension, drive voltage, etc.) and conveying load monitoring parameters (including material conveying volume, instantaneous load, material density, etc.) in real time, and input these parameters into the first element standard threshold configurator according to preset standardized values.

[0102] Among them, the first component standard threshold configurator has a built-in trained and optimized fully connected neural network model, which has mastered the complex mapping relationship between operating control parameters, transmission load parameters and state standard thresholds through training and learning.

[0103] Furthermore, when the real-time monitored equipment operation control parameters are input into the first element standard threshold configurator, the built-in fully connected neural network model will quickly start multi-layer neuron operations, that is, receiving parameters through the input layer and passing them to the hidden layer. The hidden layer deeply extracts features through nonlinear activation functions, and finally the output layer outputs the first state standard threshold corresponding to each state attribute.

[0104] For example, when the equipment operation control parameters are monitored as "motor speed 1200r / min, belt tension 600N", and the conveying load parameters are "material conveying volume 300t / h, instantaneous load peak 250kN", the first element standard threshold configurator will calculate through the built-in model and output the corresponding first state standard threshold: temperature threshold 55°C, vibration threshold 1.2mm / s, speed deviation threshold ±50r / min, etc. These thresholds can accurately adapt to the current high-load and high-speed operating conditions.

[0105] Among them, unlike the fixed threshold in the prior art, the first state standard threshold has dynamic adaptability, that is, when the equipment operation control monitoring parameters and the transmission load monitoring parameters change, the first element standard threshold configurator will adjust the output first state standard threshold in real time to avoid misjudgment or protection lag caused by the fixed threshold.

[0106] For example, when the conveying load drops to 100 t / h, the first element standard threshold configurator will automatically adjust the temperature threshold down to 50° C. to match the change in heat dissipation capacity of the mining belt conveyor under low load.

[0107] In addition, the first-element standard threshold configurator also has a parameter self-verification function, that is, while outputting the first-state standard threshold, it will synchronously record the correspondence between the input monitoring parameters and the output threshold, providing original data support for subsequent threshold corrections to ensure the reliability and traceability of the output threshold.

[0108] Through this step, the transition from a static fixed threshold to a dynamic adaptive threshold is achieved, so that the protection threshold of the mining belt conveyor can closely follow the changes in the equipment's operating status, improving the accuracy and flexibility of protection, and effectively reducing the risk of production impact or equipment damage caused by threshold mismatch.

[0109] S130: Retrieving a small sample of healthy service data of the first component based on the service life of the first component and the transportation environment information, performing statistics, and obtaining a second state standard threshold; In the embodiment of the present application, in order to further improve the reliability and adaptability of the protection threshold of the mining belt conveyor, it is necessary to combine the service history and environmental characteristics of the equipment, and form a second state standard threshold through statistics of small sample healthy service data, which complements the first state standard threshold to provide a multi-dimensional reference for the final correction threshold.

[0110] Specifically, the service life and transportation environment information of the first component are first determined and used as search constraints.

[0111] Furthermore, a small sample of healthy service data of the first component with a service time similar to that of the current first component and consistent transportation environment characteristics is screened out from the equipment history database.

[0112] Among them, the data volume of the small sample first component health service data is ≤30 groups, which can not only ensure the targetedness of data statistics, but also avoid calculation delays caused by excessive sample size to ensure response speed.

[0113] Furthermore, we traverse these small sample first component healthy service data and perform a box plot analysis on each state attribute of each data set. That is, we calculate the quartile of each state attribute to determine the distribution range of the state attribute in the healthy state.

[0114] On this basis, the intersection statistics of several groups of distribution intervals with the same state attribute are performed, and the overlapping part of each distribution interval is taken as the unified distribution interval of the state attribute. The unified distribution interval is the state standard threshold of the state attribute corresponding to the initial element in the first cluster. These state standard thresholds are added one by one to the second state standard threshold to obtain the complete second state standard threshold.

[0115] Through this step, the statistical results of small sample healthy service data were used to form a second state standard threshold that adapts to the service stage and environmental characteristics of the equipment. It works synergistically with the first state standard threshold, laying an important foundation for improving the comprehensiveness and accuracy of the protection threshold of mining belt conveyors.

[0116] Step S130 in the method provided in the embodiment of the present application includes: Extracting the initial component in the first cluster from the first component, extracting the service time of the initial component in the first cluster from the service time of the first component, and extracting the transportation environment information of the initial component in the first cluster from the transportation environment information; Using the service life of the initial components in the first cluster and the transportation environment information of the initial components in the first cluster as constraints, several small sample first component health service data are retrieved; Traversing the healthy service data of the plurality of small samples of the first components, performing box plot analysis of the same attributes, and obtaining a plurality of groups of state attribute distribution intervals; For the groups of state attribute distribution intervals, intersection statistics of intervals with the same attribute are performed to obtain the initial component state standard threshold in the first cluster, and add it to the second state standard threshold.

[0117] In the embodiment of the present application, in order to make the protection threshold of the mining belt conveyor more in line with the actual health status of the equipment in different service stages and environments, it is necessary to combine the service time and transportation environment information of the initial components in the first cluster, and obtain the second state standard threshold through statistics of small sample healthy service data, which complements the first state standard threshold to further improve the reliability and adaptability of the threshold.

[0118] Specifically, the initial components within the first cluster are first extracted from the first component. The service life of the initial components within the cluster (e.g., 1 month, 6 months, 1 year, etc.) is also extracted from the service life of the first component. Corresponding transportation environment information (e.g., underground humidity, dust concentration, ambient temperature, and roadway support conditions) is also extracted from the transportation environment information. This information together forms the constraints for retrieving the healthy service data for the first component, ensuring that the selected data is targeted.

[0119] For example, the initial element in the first cluster is a driving roller group, which has been in service for 8 months and the transportation environment information is underground humidity 85%, dust concentration 3mg / m 3 , the ambient temperature is 25℃, and this is used as a constraint condition for data retrieval.

[0120] Furthermore, based on the service life and transportation environment information of the initial components in the first cluster, a small sample of healthy service data of the first components is retrieved from the equipment history database. The number of small samples is usually ≤30 groups.

[0121] Among them, the healthy service data of the first component of these small samples must meet the requirements of being similar to the service time of the components in the current cluster (such as the deviation is within ±1 month), consistent with the transportation environment characteristics (such as parameters such as humidity and dust concentration are in the same range), and all are status record data when the components are operating without faults, including monitoring values ​​of status attributes such as temperature, vibration, and pressure.

[0122] For example, 25 groups of healthy service data that meet the conditions are retrieved, all of which are from the same type of drive roller group that has been in service for 7-9 months, with a humidity of 80%-90%, and a dust concentration of 2-4 mg / m 3 Trouble-free operation record in the environment.

[0123] Furthermore, the 25 groups of small sample first component healthy service data were traversed, and the same-state attribute box plot analysis was performed on each state attribute of each group of data.

[0124] Specifically, by calculating the quartiles of each state attribute (lower quartile X1, median X2, upper quartile X3), the distribution range of the state attribute in the healthy state (i.e., the box range from X1 to X3) is determined.

[0125] For example, the temperature attribute is analyzed. The temperature values ​​of the healthy service data of the first element of 25 groups of small samples are concentrated in the range of 32-48°C, where X1 is 35°C and X3 is 45°C. The state attribute distribution range of the temperature attribute is 35-45°C. The X1 of the vibration attribute is 0.3mm / s and X3 is 0.7mm / s, and its state attribute distribution range is 0.3-0.7mm / s. The distribution range of all state attributes can be obtained by analogy.

[0126] On this basis, the intersection statistics of several groups of distribution intervals with the same state attribute are taken, and the overlapping part of each distribution interval is taken as the unified distribution interval of the state attribute, that is, the state standard threshold of the initial element in the first cluster.

[0127] For example, if the temperature attribute of 20 of the 25 small sample first component healthy service data sets has a distribution range of 36-44°C, then 36-44°C is determined as the final state standard threshold for the temperature attribute. The intersection range of the vibration attribute is 0.35-0.65 mm / s, which is used as the state standard threshold for the vibration attribute. These thresholds are added to the second state standard threshold to form a complete second state standard threshold.

[0128] The use of a small sample of healthy service data for the first component (≤30 groups) reduces the computational complexity of the statistical process and enables rapid output of results, meeting the response requirements of real-time protection. Furthermore, this data, correlated with the service stage and environment of the initial components within the first cluster, can reflect the health status boundaries of mining belt conveyors under specific conditions, overcoming the limitation of relying solely on real-time parameters for the first health threshold.

[0129] S140: Take the intersection of the first state standard threshold and the second state standard threshold to obtain a modified state standard threshold, and perform comprehensive protection management.

[0130] In the embodiment of the present application, in order to achieve the most accurate and reliable protection of the mining belt conveyor, it is necessary to combine the advantages of the first state standard threshold and the second state standard threshold, and obtain the corrected state standard threshold by taking the intersection of the two, which is used as the final judgment basis for equipment protection to ensure that the protection of the equipment is adapted to both real-time working conditions and long-term health status characteristics of the equipment.

[0131] Specifically, the first state standard threshold is a threshold dynamically output by the first component standard threshold configurator based on real-time operation control monitoring parameters and transmission load monitoring parameters, and can quickly respond to changes in the current operating state of the equipment.

[0132] Meanwhile, the second health threshold is derived from a small sample of healthy service data, combining the first component's service life and transportation environment information. This threshold reflects the health boundary of the device under specific conditions and circumstances. The revised health threshold is formed by taking the intersection of the two thresholds, retaining the more restrictive portion of the two thresholds.

[0133] For example, if the temperature threshold in the first state standard threshold is 50-60°C, and the temperature threshold in the second state standard threshold is 48-55°C, the intersection of the two is 50-55°C, and this intersection is used as the corrected state standard threshold for temperature. This range not only meets the temperature limit under real-time high-load operation, but also meets the health status requirements of the equipment under the current service time (such as 8 months) and environment (high humidity), which is more stringent and accurate.

[0134] In addition, if the vibration threshold in the first state standard threshold is 0.5-1.2 mm / s, and the vibration threshold in the second state standard threshold is 0.4-1.0 mm / s, the intersection of the two is 0.5-1.0 mm / s. This intersection is also used as the corrected state standard threshold for vibration to ensure that the equipment vibration is within a safer range.

[0135] On the contrary, if the intersection of the first state standard threshold and the second state standard threshold is empty, it means that there is a conflict between the two threshold judgments. At this time, it is necessary to update the healthy service data of the small sample first component (such as expanding the search range, adjusting the service time or environmental information constraints), and recalculate the second state standard threshold until the intersection of the two is not empty.

[0136] In addition, if the intersection of the first state standard threshold and the second state standard threshold is still empty after multiple updates (such as 3-5 times), it is determined that the mining belt conveyor may have an abnormal state or data anomaly, and an early warning will be directly issued to remind the operation and maintenance personnel to conduct manual inspection to avoid protection failure due to threshold conflicts.

[0137] Furthermore, after obtaining the correction status standard threshold, comprehensive protection management will be performed based on the threshold.

[0138] Specifically, the status monitoring data of the mining belt conveyor (such as temperature, vibration, speed, etc.) are compared with the corrected status standard threshold in real time. When the monitoring data exceeds the corrected status standard threshold range, the corresponding protection measures are immediately activated, such as reducing the equipment operating speed, shutting down for maintenance, and activating fire prevention and extinguishing devices.

[0139] At the same time, the comparison results of the status monitoring data with the corrected status standard threshold, the protection action type and the execution time are recorded to form a complete protection log, providing data support for the subsequent maintenance and threshold optimization of the mining belt conveyor.

[0140] Through this step, the advantages of dynamic real-time thresholds and historical health thresholds are integrated, making the corrected status standard thresholds both dynamically adaptable and long-term reliable, significantly improving the accuracy and safety of the comprehensive protection of mining belt conveyors, and effectively reducing the risk of equipment failure and the probability of production interruption.

[0141] The embodiments of the present application achieve the following technical effects through the above specific implementation methods: The present application proposes a comprehensive protection method for a mining belt conveyor. First, the first component of the mining belt conveyor is extracted, and its initial service data is collected. The operating control parameters and conveying load parameters are used as input and the preset state standard threshold is used as supervision to train the first component standard threshold configurator. Through component clustering and model training, the foundation for dynamic threshold generation is laid. Secondly, the real-time monitoring parameters are processed by the first component standard threshold configurator to output the first state standard threshold, so that the threshold can be adjusted in real time with the working conditions, overcoming the defect that the fixed threshold is difficult to adapt to different operating states. At the same time, based on the current service time and transportation environment as constraints, a small sample of healthy service data is retrieved and the second state standard threshold is obtained by statistics. Combined with the equipment service time and environmental characteristics, the threshold is made to fit the long-term health status of the equipment more closely, making up for the limitations of the real-time threshold. Finally, the intersection of the first state standard threshold and the second state standard threshold is taken as the corrected state standard threshold, and comprehensive protection management is performed accordingly to integrate the advantages of dynamic and historical thresholds, ensure more accurate and reliable protection, and effectively reduce the impact of control errors on production.

[0142] The method provided in the embodiment of the present application solves the problems of insufficient accuracy and control error caused by the inability of fixed thresholds to adapt to different service states and operating conditions in traditional mining belt conveyor protection through the technical solution of "component clustering-dynamic threshold generation-health threshold supplementation-corrected threshold protection", improves the reliability and flexibility of equipment protection, and provides strong technical support for ensuring the safe and stable operation of mining belt conveyors.

[0143] Example 2, as shown in the attached Figure 1 and attached Figure 2As shown, the present application provides a comprehensive protection system for a mining belt conveyor and a comprehensive protection method for a mining belt conveyor.

[0144] As attached Figure 1 As shown, based on the inventive concept of the comprehensive protection method for a mining belt conveyor provided in Example 1, the present application also provides a comprehensive protection system for a mining belt conveyor, specifically comprising: A configurator training module is used to extract the first component of the mining belt conveyor, collect initial service data of the first component, use operation control parameters and conveying load parameters as input, and use a preset state standard threshold as supervision to train the first component standard threshold configurator, where initial service indicates a service time of less than or equal to 3 months and no failures, and the preset state is a continuous variable; a first threshold output module, configured to process the operation control monitoring parameter and the transport load monitoring parameter through a first component standard threshold configurator, and output a first state standard threshold; A second threshold statistics module is used to retrieve a small sample of healthy service data of the first component based on the service time of the first component and the transportation environment information, perform statistics, and obtain a second state standard threshold; The modified threshold comprehensive protection module is used to take the intersection of the first state standard threshold and the second state standard threshold, obtain the modified state standard threshold, and perform comprehensive protection management.

[0145] In one embodiment, the configurator training module is further configured to: Obtaining initial service data of a mining belt conveyor, wherein the initial service data of the mining belt conveyor includes state record data of the mining belt conveyor; According to the parts list, the mining belt conveyor is disassembled to obtain the first initial component to the Nth initial component, where N represents the total number of parts; Based on the mining belt conveyor state record data, clustering the first initial element to the Nth initial element in the same state to obtain the first cluster initial element to the Qth cluster initial element, where Q represents the total number of clusters; The first element is obtained by combining any cluster of elements from the first cluster of initial elements to the Qth cluster of initial elements.

[0146] obtaining an initial element in a first cluster of the first element, and collecting an initial service data set of the initial elements in the first cluster, wherein any piece of initial service data of the initial elements in the first cluster in the initial service data set includes an operation record control parameter, a transport record load parameter, and a status record data; Traversing the initial component initial service data set in the first cluster, dividing the initial component initial service data in the first cluster having the same operation record control parameters and the same transmission record load parameters into the same group, and obtaining a plurality of groups of initial component initial service data in the first cluster; Traversing the plurality of groups of state record data of the initial service data of the initial components in the plurality of first clusters to perform a same-state box plot analysis to obtain a plurality of state identification intervals, wherein the state identification intervals are upper and lower limit constraint intervals of the state record data box; The plurality of groups of initial component initial service data in the first cluster and the plurality of state identification intervals are associated and stored, and added to the first component initial service data.

[0147] Taking the operation control parameters and the transport load parameters as input and the preset state standard threshold as supervision, the fully connected neural network is trained to obtain the first-level standard threshold configurator; Performing verification on the primary standard threshold configurator to obtain a first state attribute verification residual vector up to an M-th state attribute verification residual vector, where M represents the total number of state attributes; When any residual vector modulus value of the first state attribute verification residual vector until the M-th state attribute verification residual vector is greater than or equal to the residual modulus threshold, updating the first-level standard threshold configurator network topology based on the first state attribute verification residual vector until the M-th state attribute verification residual vector, and performing cyclic training; Otherwise, the primary standard threshold configurator is set to the first component standard threshold configurator.

[0148] In one embodiment, the second threshold statistics module is further configured to: Extracting the initial component in the first cluster from the first component, extracting the service time of the initial component in the first cluster from the service time of the first component, and extracting the transportation environment information of the initial component in the first cluster from the transportation environment information; Using the service life of the initial components in the first cluster and the transportation environment information of the initial components in the first cluster as constraints, several small sample first component health service data are retrieved; Traversing the healthy service data of the plurality of small samples of the first components, performing box plot analysis of the same attributes, and obtaining a plurality of groups of state attribute distribution intervals; For the groups of state attribute distribution intervals, intersection statistics of intervals with the same attribute are performed to obtain the initial component state standard threshold in the first cluster, and add it to the second state standard threshold.

[0149] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0150] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0151] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A comprehensive protection method for a mining belt conveyor, characterized in that: include: Extract the first element of the mining belt conveyor, collect the initial service data of the first element, use the operation control parameters and conveying load parameters as input, and use the preset state standard threshold as supervision to train the first element standard threshold configurator; Processing the operation control monitoring parameter and the transport load monitoring parameter through the first component standard threshold value configurator to output a first state standard threshold value; Using the service life of the first component and the transportation environment information as constraints, a small sample of the first component's healthy service data is retrieved, statistics are performed, and the second status standard threshold is obtained; Take the intersection of the first state standard threshold and the second state standard threshold to obtain a modified state standard threshold and perform comprehensive protection management.

2. The method according to claim 1, wherein Extract the first element of the mining belt conveyor, including: Obtaining initial service data of a mining belt conveyor, wherein the initial service data of the mining belt conveyor includes state record data of the mining belt conveyor; According to the parts list, the mining belt conveyor is disassembled to obtain the first initial component to the Nth initial component, where N represents the total number of parts; Based on the mining belt conveyor state record data, clustering the first initial element to the Nth initial element in the same state to obtain the first cluster initial element to the Qth cluster initial element, where Q represents the total number of clusters; The first element is obtained by combining any cluster of elements from the first cluster of initial elements to the Qth cluster of initial elements.

3. The method according to claim 2, wherein Based on the mining belt conveyor state record data, clustering the first initial element to the Nth initial element in the same state to obtain the first cluster initial element to the Qth cluster initial element includes: Extracting first mining belt conveyor state record data from the mining belt conveyor state record data, wherein the first mining belt conveyor state record data includes first initial component state record data and second initial component state record data; When the first initial element state record data and the second initial element state record data are consistent, the same state frequency of the first initial element and the second initial element is increased by one, otherwise the same state frequency remains unchanged, wherein the initial value of the same state frequency is equal to 0; Until the Yth mining belt conveyor state record data is extracted from the mining belt conveyor state record data, the same state frequency is updated, and the same state frequency is output, Y ≥ 30, where Y represents the total number of mining belt conveyor state record data; When the ratio of the same-state frequency to Y is greater than or equal to a ratio threshold, clustering the first initial component and the second initial component into one category; When all initial elements are clustered, the first cluster of initial elements is output until the Qth cluster of initial elements.

4. The method according to claim 3, wherein When the first initial component state record data and the second initial component state record data are consistent, it includes: Configure the state attribute consistency tolerance deviation for the state attribute set; When the state attribute deviations of the first initial component state record data and the second initial component state record data both meet the corresponding state attribute consistency fault tolerance deviation, the first initial component state record data and the second initial component state record data are considered to be consistent; Otherwise, it is considered that the first initial component state record data and the second initial component state record data are inconsistent.

5. The method according to claim 2, wherein Collect initial service data of the first component, including: obtaining an initial element in a first cluster of the first element, and collecting an initial service data set of the initial elements in the first cluster, wherein any piece of initial service data of the initial elements in the first cluster in the initial service data set includes an operation record control parameter, a transport record load parameter, and a status record data; Traversing the initial component initial service data set in the first cluster, dividing the initial component initial service data in the first cluster having the same operation record control parameters and the same transmission record load parameters into the same group, and obtaining a plurality of groups of initial component initial service data in the first cluster; Traversing the plurality of groups of state record data of the initial service data of the initial components in the plurality of first clusters to perform a same-state box plot analysis to obtain a plurality of state identification intervals, wherein the state identification intervals are upper and lower limit constraint intervals of the state record data box; The plurality of groups of initial component initial service data in the first cluster and the plurality of state identification intervals are associated and stored, and added to the first component initial service data.

6. The method according to claim 1, wherein Extract the first element of the mining belt conveyor, collect the initial service data of the first element, use the operation control parameters and conveying load parameters as input, and use the preset state standard threshold as supervision to train the first element standard threshold configurator, including: Taking the operation control parameters and the transport load parameters as input and the preset state standard threshold as supervision, the fully connected neural network is trained to obtain the first-level standard threshold configurator; Performing verification on the primary standard threshold configurator to obtain a first state attribute verification residual vector up to an M-th state attribute verification residual vector, where M represents the total number of state attributes; When any residual vector modulus value of the first state attribute verification residual vector until the M-th state attribute verification residual vector is greater than or equal to the residual modulus threshold, updating the first-level standard threshold configurator network topology based on the first state attribute verification residual vector until the M-th state attribute verification residual vector, and performing cyclic training; Otherwise, the primary standard threshold configurator is set to the first component standard threshold configurator.

7. The method according to claim 6, wherein When the modulus of any residual vector of the first state attribute verification residual vector until the M-th state attribute verification residual vector is greater than or equal to the residual modulus threshold, based on the first state attribute verification residual vector until the M-th state attribute verification residual vector, updating the first-level standard threshold configurator network topology, and performing loop training, including: When the modulus of the first state attribute verification residual vector is greater than or equal to the residual modulus threshold, the operation control parameter and the transport load parameter are used as input, and the first state attribute verification residual vector is used as supervision to train a fully connected neural network to obtain a first state attribute output corrector, which is integrated into the first state attribute output node of the primary standard threshold configurator. The integration rule is to add the outputs of the first state attribute output corrector and the first state attribute output node and use the sum as the final output of the first state attribute output node; Until the modulus value of the Mth state attribute verification residual vector is greater than or equal to the residual modulus threshold, the operation control parameter and the transport load parameter are used as input, and the Mth state attribute verification residual vector is used as supervision to train a fully connected neural network to obtain an Mth state attribute output corrector, which is integrated into the Mth state attribute output node of the primary standard threshold configurator, wherein the integration rule is to add the outputs of the Mth state attribute output corrector and the Mth state attribute output node as the final output of the Mth state attribute output node; Obtain the first-level standard threshold configurator network topology and perform cyclic training.

8. The method according to claim 1, wherein Using the service life of the first component and the transportation environment information as constraints, retrieve a small sample of the first component's healthy service data, perform statistics, and obtain the second status standard threshold, including: Extracting the initial component in the first cluster from the first component, extracting the service time of the initial component in the first cluster from the service time of the first component, and extracting the transportation environment information of the initial component in the first cluster from the transportation environment information; Using the service life of the initial components in the first cluster and the transportation environment information of the initial components in the first cluster as constraints, several small sample first component health service data are retrieved; Traversing the healthy service data of the plurality of small samples of the first components, performing box plot analysis of the same attributes, and obtaining a plurality of groups of state attribute distribution intervals; For the groups of state attribute distribution intervals, intersection statistics of intervals with the same attribute are performed to obtain the initial component state standard threshold in the first cluster, and add it to the second state standard threshold.

9. Comprehensive protection system for mining belt conveyor, characterized by: The system is used to implement the comprehensive protection method for a mining belt conveyor according to any one of claims 1 to 8, and the system comprises: a configurator training module, for extracting the first element of the mining belt conveyor, collecting the initial service data of the first element, using the operation control parameters and the conveying load parameters as input, and using the preset state standard threshold as supervision, to train the first element standard threshold configurator; a first threshold output module, configured to process the operation control monitoring parameter and the transport load monitoring parameter through a first component standard threshold configurator, and output a first state standard threshold; A second threshold statistics module is used to retrieve a small sample of healthy service data of the first component based on the service time of the first component and the transportation environment information, perform statistics, and obtain a second state standard threshold; The modified threshold comprehensive protection module is used to take the intersection of the first state standard threshold and the second state standard threshold, obtain the modified state standard threshold, and perform comprehensive protection management.

Images