Distributed job node state prediction method and system based on software defined control

Abstract

The invention provides a distributed job node state prediction method and system based on software defined control, and belongs to the field of industrial equipment management. According to the method, state data of a device set of the same model is collected, heterogeneous state sorting is carried out, and operation heterogeneous devices are recognized; the model and state data of the heterogeneous equipment and the underground coal mine environment parameters are uploaded to an application layer; the application layer extracts equipment control parameters, calls a state prediction model based on a heterogeneous equipment model, and performs state prediction by fusing the environment parameters and the control parameters; and when the predicted data is consistent with the heterogeneous equipment state data, performing health identification. According to the method and the device, the technical problems of low distributed operation reliability and low operation and maintenance efficiency caused by incapability of performing accurate heterogeneous state identification and health assessment on the same type of equipment set in the distributed operation environment in the prior art are solved, and the purposes of accurately identifying the heterogeneous equipment state and realizing predictive health management are achieved; and the distributed operation reliability and the operation and maintenance efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of industrial equipment management, and particularly to a distributed job node status prediction method and system based on software-defined control. Background Art

[0002] With the continuous improvement of industrial automation and intelligence levels, the number and complexity of devices in a distributed job environment are increasing day by day. Especially in environments such as underground coal mines, the status monitoring and health management of sets of devices of the same model have become the key to ensuring job safety and efficiency.

[0003] Currently, traditional device status monitoring methods mainly rely on regular inspections and single-point monitoring, and cannot accurately identify heterogeneous status devices in a set of devices of the same model, resulting in the mixed treatment of faulty devices and normal devices; at the same time, there is a lack of an effective health assessment mechanism, and only after-the-fact maintenance can be carried out without early warning. These problems lead to frequent device failures, high maintenance costs, and an increased risk of job interruption in a distributed job environment, seriously restricting the reliability and operation and maintenance efficiency of distributed jobs. Summary of the Invention

[0004] Aiming at the technical problems in the prior art that it is impossible to accurately identify heterogeneous status and perform health assessment on a set of devices of the same model in a distributed job environment, resulting in low reliability and operation and maintenance efficiency of distributed jobs, the present invention provides a distributed job node status prediction method and system based on software-defined control to solve this problem.

[0005] The technical solutions of the present invention to solve the above technical problems are as follows: In a first aspect, the present invention provides a distributed job node status prediction method based on software-defined control, which is applied to an SDC system. The SDC system includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visual configuration interface and a data analysis module. The execution steps include: collecting device status data of a set of devices of the same model performing a first task through the hardware abstraction layer, performing heterogeneous status sorting to obtain job heterogeneous devices; when the number of the job heterogeneous devices is not equal to 0, uploading the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters to the application layer; extracting device control parameters of the job heterogeneous devices from the control logic layer through the application layer, invoking a device status prediction model of the data analysis module based on the heterogeneous device model, processing the underground coal mine environment parameters and the execution status of the device control parameters to perform status prediction, and obtaining device status prediction data; when the device status prediction data is consistent with the heterogeneous device status data, performing a health identification on the set of devices of the same model in the visual configuration interface.

[0006] Second aspect, the present invention provides a distributed job node status prediction system based on software-defined control, which is applied to an SDC system. The SDC system includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visualization configuration interface and a data analysis module. The distributed job node status prediction system includes: a data collection and sorting unit, configured to collect device status data of a set of devices of the same model executing a first task through the hardware abstraction layer, perform heterogeneous status sorting, and obtain job heterogeneous devices; a heterogeneous device upload unit, configured to upload the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters to the application layer when the number of the job heterogeneous devices is not equal to 0; a status prediction and analysis unit, configured to extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, retrieve the device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain device status prediction data; a health identification display unit, configured to perform a health identification on the set of devices of the same model in the visualization configuration interface when the device status prediction data is consistent with the heterogeneous device status data.

[0007] The beneficial effects of the present invention are as follows: First, by collecting the device status data of a set of devices of the same model executing a first task through the hardware abstraction layer, performing heterogeneous status sorting, and obtaining job heterogeneous devices, accurate screening and identification of devices with abnormal status in the set of devices of the same model are realized; when the number of the job heterogeneous devices is not equal to 0, the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters are uploaded to the application layer, thereby establishing an associated data basis for heterogeneous device information and environmental factors; by extracting the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, retrieving the device status prediction model of the data analysis module based on the heterogeneous device model, processing the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtaining device status prediction data, intelligent status prediction based on multi-dimensional parameter fusion is realized; when the device status prediction data is consistent with the heterogeneous device status data, a health identification is performed on the set of devices of the same model in the visualization configuration interface, thereby completing the verification of the prediction result and the visualization management of the device health status.

[0008] Through the above technical solution, the status prediction of distributed job nodes based on the software-defined control architecture is realized, effectively solving the problems of inaccurate heterogeneous status recognition and lack of health assessment, accurately identifying the status of heterogeneous devices and realizing predictive health management, and improving the reliability and operation and maintenance efficiency of distributed jobs. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic flowchart of the distributed job node status prediction method based on software-defined control provided by the present invention; Figure 2 Schematic structural diagram of the distributed job node status prediction system based on software-defined control provided by the present invention.

[0010] In the attached drawings, the components represented by each reference numeral are as follows: Data acquisition and sorting unit 11, heterogeneous device upload unit 12, status prediction and analysis unit 13, health indicator display unit 14. Detailed implementation manners

[0011] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.

[0012] In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the present invention, "a plurality of" means two or more, unless otherwise specifically defined.

[0013] In the description of the present invention, the term "for example" is used to mean "serving as an example, illustration, or explanation". Any embodiment described as "for example" in the present invention is not necessarily construed as being more preferred or more advantageous than other embodiments. In order to enable any person skilled in the art to implement and use the present invention, the following description is given. In the following description, details are set forth for the purpose of explanation. It should be understood that those skilled in the art can recognize that the present invention can be implemented without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid unnecessary details from obscuring the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

[0014] Embodiment 1, as Figure 1 shown, the embodiment of the present invention provides a distributed job node status prediction method based on software-defined control, which is applied to an SDC system. The SDC system includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visualization configuration interface and a data analysis module.

[0015] Specifically, the distributed job node status prediction method provided by the embodiments of this application is built based on a Software Defined Control (SDC) system. The SDC system adopts a hierarchical decoupled architecture design, including a hardware abstraction layer, a control logic layer, and an application layer, to achieve virtualized management of hardware resources and software implementation of control logic.

[0016] Among them, the hardware abstraction layer is responsible for uniformly abstracting and encapsulating the underlying physical devices, shielding the differences of different hardware devices, and providing a standardized device access interface for the upper layer. Through the hardware abstraction layer, unified management and control of various heterogeneous devices in the distributed job node can be achieved, including but not limited to downhole operation devices such as sensors, actuators, and controllers. The control logic layer runs the core control algorithms and business logics, and realizes intelligent control of the job node in a software-based manner. The control logic layer is decoupled from the hardware abstraction layer and the application layer, enabling the control logic to be developed, deployed, and maintained independently of the specific hardware platform, improving flexibility and scalability. The application layer, as the top layer, undertakes human-computer interaction and data analysis functions, including a visual configuration interface and a data analysis module. The visual configuration interface provides intuitive device status monitoring and configuration functions for operators, supporting real-time display of device operation status, alarm information, and health assessment results; the data analysis module integrates machine learning algorithms and data mining technologies to realize intelligent prediction of device status and anomaly detection, providing a basis for preventive maintenance.

[0017] Through the three-layer architecture of the SDC system, the SDC system can achieve comprehensive monitoring, intelligent analysis, and accurate prediction of the distributed job node, laying a foundation for improving industrial production efficiency and device reliability.

[0018] The distributed job node status prediction method includes: S1. Collect device status data of a set of devices of the same model performing the first task through the hardware abstraction layer, perform heterogeneous status sorting, and obtain job heterogeneous devices.

[0019] Specifically, first, perform unified data collection operations on a set of devices of the same model performing the first task through the hardware abstraction layer to obtain device status data. The set of devices of the same model refers to multiple devices deployed in a distributed job environment with the same model specifications and functional characteristics, and these devices cooperate to perform a predefined first task. The device status data includes but is not limited to multi-dimensional information such as device operation parameters, performance indicators, and working status identifiers, and is realized through the standardized interface of the hardware abstraction layer for real-time collection and unified formatting processing.

[0020] After obtaining the device status data, perform heterogeneous status sorting operation, which aims to identify and screen out devices with abnormal or deviated operating status from a set of devices of the same model. Heterogeneous status sorting refers to the comparison and analysis of the status data of devices of the same model through data analysis to identify device status patterns that are significantly different from the normal operating mode. After the heterogeneous status sorting process, heterogeneous devices for operation are obtained, that is, devices in the set of devices of the same model that are identified as having abnormal operating status or potential failure risks. These heterogeneous devices for operation will be the key monitoring objects for subsequent status prediction and health assessment, providing a list of target devices for achieving precise preventive maintenance.

[0021] By performing heterogeneous status sorting, abnormal devices in a distributed operation environment can be automatically identified, avoiding the cumbersome process of manually checking the status of each device in traditional methods and improving the efficiency and accuracy of anomaly detection.

[0022] S2. When the number of the heterogeneous devices for operation is not equal to 0, upload the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters to the application layer.

[0023] Specifically, first, count the number of the obtained heterogeneous devices for operation and perform a conditional judgment operation. When the number of the heterogeneous devices for operation is not equal to 0, it indicates that there are devices with abnormal operating status in the set of devices of the same model, and further status prediction and analysis processing are required. This conditional judgment ensures the reasonable allocation of resources and avoids unnecessary prediction calculations in the case of no abnormal devices.

[0024] When the number of the heterogeneous devices for operation is not equal to 0, a data upload operation will be performed, specifically including the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters. Among them, the heterogeneous device model is used to identify the specific model specifications of the heterogeneous devices for operation, providing an index basis for subsequent invocation of the corresponding device status prediction model; the heterogeneous device model includes identification information such as the manufacturer, product series, and technical parameters of the device. The heterogeneous device status data is the real-time operation data collected in step S1 and identified as the heterogeneous devices for operation; the heterogeneous device status data contains multi-dimensional operation parameters of the device, such as physical measurement values of current, voltage, temperature, vibration, pressure, etc., and status identification information such as the working mode, operation duration, and fault code of the device. The underground coal mine environment parameters reflect the external environmental conditions where the heterogeneous devices for operation are located, and the underground coal mine environment parameters include, but are not limited to, environmental factors affecting the operation performance of the device such as environmental temperature, humidity, gas concentration, dust concentration, atmospheric pressure, and ventilation conditions.

[0025] The above three types of data are uniformly uploaded to the application layer, providing a complete data foundation for subsequent intelligent analysis and status prediction. By uploading heterogeneous device models, heterogeneous device status data, and underground coal mine environmental parameters to the application layer, automatic screening of abnormal device information and accurate transmission of effective data are achieved, laying a data foundation for improving the accuracy and pertinence of status prediction.

[0026] S3. Extract the device control parameters of the heterogeneous devices for the operation from the control logic layer through the application layer, retrieve the device status prediction model of the data analysis module based on the heterogeneous device models, and process the underground coal mine environmental parameters and the execution status of the device control parameters to perform status prediction and obtain device status prediction data.

[0027] Specifically, first, the application layer extracts the device control parameters of the heterogeneous devices for the operation through the data interface with the control logic layer. The device control parameters refer to the control instructions and configuration parameters issued by the control logic layer to the heterogeneous devices for the operation, including but not limited to the set values of the working modes of the devices, the control quantities of the running speeds, the adjustment parameters of the output powers, the start-stop control signals, etc.

[0028] Second, based on the uploaded heterogeneous device models, retrieve the corresponding device status prediction models from the data analysis module. The data analysis module pre-stores dedicated prediction models trained for different device models, and each model has been trained and optimized with a large amount of historical data and has the ability to accurately predict the status of specific model devices. Through model matching, the most suitable device status prediction model for the current heterogeneous devices for the operation can be automatically selected to ensure the pertinence and accuracy of the prediction results.

[0029] Then, the retrieved device status prediction models receive the underground coal mine environmental parameters and the execution status of the device control parameters to perform status prediction. Among them, the underground coal mine environmental parameters reflect the external working environment where the heterogeneous devices for the operation are located; the device control parameters reflect the internal control status of the heterogeneous devices for the operation. The device status prediction models comprehensively analyze and perform pattern recognition on the above-mentioned underground coal mine environmental parameters and the device control parameters through deep learning, machine learning, or other artificial intelligence algorithms, and establish a complex mapping relationship between the environmental conditions, control parameters, and device status. Finally, through the calculation and processing of the device status prediction models, device status prediction data is obtained. The device status prediction data refers to the device status data that the heterogeneous devices for the operation should have under specific underground coal mine environmental parameters and device control parameters, providing a basis for subsequent device health assessment and maintenance decision-making.

[0030] By obtaining the device status prediction data, intelligent status prediction based on multi-source information fusion is achieved, fully considering the comprehensive influence of the internal control status and external environmental conditions of the devices, and improving the accuracy and practicality of status prediction.

[0031] S4. When the device status prediction data is consistent with the heterogeneous device status data, perform a health identification on the set of devices of the same model in the visual configuration interface.

[0032] First, perform a consistency comparison operation between the device status prediction data and the heterogeneous device status data. Among them, the device status prediction data is the predicted value of the device status calculated by the device status prediction model in step S3, and the heterogeneous device status data is the actual operating status of the job heterogeneous devices collected in step S1. Through data comparison, perform deviation analysis on the two types of data to determine whether the prediction result and the actual status meet the preset consistency standard. Among them, the consistency determination is evaluated based on a preset tolerance range or similarity threshold. When the deviation between the device status prediction data and the heterogeneous device status data is within the acceptable range, it is considered that the two reach a consistent state. The consistency between the device status prediction data and the heterogeneous device status data verifies that the job heterogeneous devices identified in step S1 do have abnormal status.

[0033] When the consistency condition is met, perform a health identification on the entire set of devices of the same model in the visual configuration interface. Among them, the health identification refers to graphically displaying the overall health status of the set of devices of the same model in the visual configuration interface, such as using visual elements such as color coding, status icons, or numerical indicators. For example, use green to represent the healthy state, yellow to represent the warning state, red to represent the abnormal state, etc.

[0034] It should be noted that although only some devices are identified as job heterogeneous devices, when verifying the consistency between the device status prediction data and the heterogeneous device status data, perform a health identification on the entire set of devices of the same model. This is because the verification of the accuracy of the device status prediction model indicates the ability to perform reliable status monitoring on this type of device, so that a comprehensive assessment of the health status of the entire set of devices of the same model can be given.

[0035] By identifying the set of devices of the same model in the visual configuration interface, an intelligent health assessment based on prediction verification is realized, which not only verifies the accuracy of anomaly detection and status prediction, but also provides an intuitive display of the device health status for operators, effectively supporting device management and maintenance decisions.

[0036] Furthermore, the training sample devices of the device status prediction model are less than or equal to the first service life; extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, and process the underground coal mine environment parameters and the device control parameter execution status prediction to obtain device status prediction data, including: S31. Obtain the second service life of the job heterogeneous device; S32. Through the device sample index model of the data analysis module, set the underground coal mine environment attributes and device control attributes as constant attributes, retrieve the first device status record sample that meets the heterogeneous device model and the first service life, and retrieve the second device status record sample that meets the heterogeneous device model and the second service life; S33. Perform mode analysis of the same attributes on the device status disturbance amount record data of the first device status record sample and the second device status record sample to obtain the device status disturbance amplitude; S34. Extract the device control parameters of the operation heterogeneous device from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, and process the underground coal mine environment parameters and the device control parameters to perform status prediction to obtain the initial device status prediction data; S35. Based on the device status disturbance amplitude, correct the initial device status prediction data to obtain the device status prediction data.

[0037] In a preferred embodiment, the service life of the training sample devices of the device status prediction model is less than or equal to the first service life. The first service life is a preset reference service time, usually corresponding to the designed service life or the optimal performance period of the device. By restricting the service life range of the training sample devices, the device status prediction model is ensured to have good prediction ability and stability.

[0038] First, obtain the second service life of the current operation heterogeneous device. The second service life refers to the actual running time of the operation heterogeneous device from the time of commissioning to the current moment, reflecting the aging degree and performance attenuation status of the operation heterogeneous device.

[0039] Then, perform a historical data retrieval operation through the device sample index model preset in the data analysis module. To ensure the comparability of the retrieval results, set the underground coal mine environment attributes and device control attributes as constant attributes, that is, fix the influence of these external factors and focus on analyzing the influence of the service life on the device status. Based on this constraint condition, retrieve the first device status record sample that meets the heterogeneous device model and the first service life, and the second device status record sample that meets the heterogeneous device model and the second service life respectively. These two sets of sample data provide a comparison benchmark for the subsequent disturbance analysis.

[0040] Subsequently, perform the same-attribute mode analysis on the equipment state perturbation amount record data of the first equipment state record sample and the second equipment state record sample to quantify the influence of the service life. Among them, the equipment state perturbation amount refers to the deviation amount of the equipment state record data corresponding to the second service life relative to the equipment state record data corresponding to the first service life. Specifically, for any one state parameter (such as temperature, vibration amplitude, current, pressure value, etc.) under the same equipment model, calculate the difference between the value of this parameter in the second equipment state record sample and the value of the corresponding parameter in the first equipment state record sample. This difference is the equipment state perturbation amount of this parameter. For example, if the average operating temperature of a certain model of equipment under the first service life (such as 3 years) is 75°C, and the average operating temperature of the same model of equipment under the second service life (such as 5 years) is 78°C, then the equipment state perturbation amount of the temperature parameter is +3°C. The positive and negative values of this perturbation amount respectively represent the upward or downward trend of the equipment state parameter with the increase of the service life. The same-attribute mode analysis refers to performing statistical analysis on the equipment state parameters of the same attribute type respectively. Classify the equipment state parameters according to their physical attributes, such as temperature type parameters, vibration type parameters, electrical type parameters, etc., and then perform mode statistics on the perturbation amount data set of each type of parameter. Specifically, first, for the first equipment state record sample and the second equipment state record sample, calculate the perturbation amount of the corresponding state parameter of each equipment to form a perturbation amount data set; then, for each type of state parameter, count the occurrence frequency of the equipment state perturbation amount value, and identify the perturbation value with the highest occurrence frequency as the mode perturbation value of this attribute, which is used as the corresponding equipment state perturbation amplitude. For example, for temperature type parameters, collect the temperature perturbation amount data of 100 same-model equipment from the first equipment state record sample and the second equipment state record sample. Among them, +2°C appears 35 times, +3°C appears 28 times, +1°C appears 20 times, and the occurrence frequencies of other values are relatively low. Then +2°C is determined as the mode perturbation value of the temperature parameter and is the equipment state perturbation amplitude corresponding to the temperature parameter.

[0041] The equipment state perturbation amplitude obtained through the same-attribute mode analysis represents the typical change amplitude that the equipment state parameter is most likely to appear under the influence of the service life difference, filters out the abnormal fluctuations and accidental factors of individual equipment, and reflects the general law and common characteristics of the change of this model of equipment with the service life. The equipment state perturbation amplitude contains a set of perturbation amplitudes of multiple parameters, such as {temperature perturbation amplitude: +2°C, vibration perturbation amplitude: +0.5mm / s, current perturbation amplitude: -1.2A,...}. Each perturbation amplitude quantifies the typical influence degree of the service life difference on the corresponding equipment state parameter, providing a reliable correction benchmark for subsequent prediction correction.

[0042] Subsequently, according to the prediction process of S3, the device control parameters of the job heterogeneous devices are extracted from the control logic layer through the application layer. Based on the heterogeneous device model, the device status prediction model of the data analysis module is retrieved, and the underground coal mine environment parameters and the execution status of the device control parameters are processed to perform status prediction, obtaining the initial device status prediction data. This initial device status prediction data is the prediction result within the service life range of the training samples and does not yet consider the influence of the actual service life of the current device. Then, based on the obtained device status disturbance amplitude, the initial device status prediction data is refined and corrected to obtain the device status prediction data. In the correction process, the device status disturbance amplitude is used as a correction factor and applied to the initial device status prediction data. Specifically, the corresponding device status disturbance amplitude is directly applied to the initial prediction data for addition and subtraction operations, thereby obtaining the device status prediction data dynamically corrected by the service life.

[0043] Through the above steps, the adaptive status prediction based on the service life disturbance is realized. Compared with the prediction in step S3, it fully considers the influence of the equipment aging law on the status change, improves the prediction accuracy by introducing the service life correction mechanism, and provides a more accurate and reliable status prediction service for the equipment in different service stages.

[0044] Furthermore, the device status data includes a number of device status data; the device status data of the same-type device set executing the first task is collected through the hardware abstraction layer, and heterogeneous status sorting is performed to obtain the job heterogeneous devices, including: S11. Perform similarity analysis on the number of device status data to obtain multiple device status similarities; S12. Based on the multiple device status similarities, traverse the number of device status data for local density evaluation to obtain a number of device status distribution densities; S13. Use the distribution density mean value to traverse and divide the number of device status distribution densities to obtain a number of device status heterogeneous parameters; S14. Based on the number of device status heterogeneous parameters, extract the device numbers of the devices whose device status heterogeneous parameters are greater than or equal to the predefined heterogeneous parameter threshold from the same-type device set and add them to the job heterogeneous devices.

[0045] In a preferred implementation manner, a refined heterogeneous device identification method based on similarity analysis and density evaluation is proposed, and through multi-level data analysis, accurate identification and screening of abnormal devices in the same-type device set are realized.

[0046] The device status data includes a number of device status data, that is, the status information of each device is composed of status parameters in multiple dimensions, such as physical quantity parameters like temperature, vibration, current, pressure, etc., and status identification parameters like operating mode, workload, etc. The device status data of each device constitutes a multi-dimensional device status vector, providing a data basis for subsequent similarity calculation and heterogeneous analysis.

[0047] First, perform similarity analysis on a number of device status data to obtain multiple device status similarities. Among them, similarity analysis refers to calculating the similarity degree of status data between any two devices in a set of devices of the same model. For example, using Euclidean distance, cosine similarity or other distance measurement methods, pairwise comparison and calculation are performed on the device status vectors of any two devices to generate a similarity matrix between devices. Multiple device status similarities reflect the mutual relationship of the internal status distribution in the set of devices of the same model, providing a similarity benchmark for subsequent density evaluation.

[0048] Then, based on multiple device status similarities, traverse a number of device status data for local density evaluation to obtain a number of device status distribution densities. Among them, local density evaluation refers to taking each device as the center and statistically analyzing the distribution density of similar devices in its neighboring area. Specifically, taking the status data of the current device as the reference point, according to a preset similarity threshold or the number of neighbors, determine the local neighborhood range of this device, and then calculate the device status distribution density of the devices within this neighborhood range. The higher the device status distribution density, the more devices with similar status exist around this device, that is, the status mode of this device conforms more to the group characteristics; the lower the device status distribution density, it indicates that the status of this device is relatively isolated or abnormal.

[0049] Subsequently, use the average of the distribution densities to traverse and calculate the ratios of a number of device status distribution densities to obtain a number of device status heterogeneous parameters. First, calculate the arithmetic mean of all device status distribution densities to obtain the average of the distribution densities, which is used as the reference standard for judging the normality of the device status. Then, calculate the ratio of the device status distribution density of each device to this average of the distribution densities to obtain the device status heterogeneous parameter of this device. This device status heterogeneous parameter quantitatively reflects the degree of deviation of a single device from the group average level. The smaller the device status heterogeneous parameter, the more the status of this device deviates from the normal group mode, and the higher the degree of heterogeneity.

[0050] After that, based on several heterogeneous device status parameters, heterogeneous devices that meet the screening criteria are extracted from the set of devices of the same model. Specifically, the heterogeneous device status parameters of each device are compared with the predefined heterogeneous parameter thresholds, and the device tag numbers of the devices whose heterogeneous device status parameters are greater than or equal to the thresholds are extracted. Among them, the device tag number is the unique identifier of the device in the set of devices of the same model, which is used to accurately locate and manage specific devices. The device tag numbers that meet the conditions are added to the job heterogeneous devices and used as the key monitoring objects for subsequent status prediction and health assessment.

[0051] Through the above steps, automatic heterogeneous device identification is achieved. Compared with the simple threshold judgment method, it can more accurately identify abnormal individuals in the group, improving the accuracy and reliability of heterogeneous status sorting.

[0052] Furthermore, based on the multiple device status similarities, the several device status data are traversed for local density evaluation to obtain several device status distribution densities, including: S121. Extract the first device status data from the several device status data; S122. Extract multiple selected device status similarities based on the first device status data from the multiple device status similarities; S123. From the multiple selected device status similarities, select neighbor selected device status similarities that are 0.5 times the number of the set of devices of the same model in descending order, and perform mean calculation to obtain the first device status distribution density; S124. Add the first device status distribution density to the several device status distribution densities.

[0053] In a preferred implementation, first, the first device status data is extracted from the several device status data. The first device status data refers to the device status information that is currently selected as the density calculation target during the traversal process. According to the preset traversal order, each device in the set of devices of the same model is selected one by one as the target of density evaluation to ensure comprehensive density analysis of all devices.

[0054] Then, multiple selected device status similarities based on the first device status data are extracted from the multiple device status similarities. For example, in a set of devices of the same model including Q devices of the same model, the first device forms Q - 1 device status similarities with the remaining Q - 1 devices, that is, multiple selected device status similarities. The multiple selected device status similarities are the similarity set between the first device and all other devices, reflecting the relative position and similarity distribution of the first device in the entire device group.

[0055] Subsequently, the similarity degrees of multiple selected device states are sorted from largest to smallest by value, and the similarity degrees of the neighbor selected device states corresponding to the first 0.5 times the number of devices of the same model in the set are filtered, and the mean value is calculated to obtain the first device state distribution density. For example, according to the similarity degrees of multiple selected device states, among the Q - 1 remaining devices, they are arranged in descending order of the similarity degrees of the selected device states, and the devices corresponding to the first 0.5×Q similarity degree values are selected as the neighbor devices of the first device, obtaining 0.5×Q similarity degrees of neighbor selected device states; the arithmetic mean of the 0.5×Q similarity degrees of neighbor selected device states is calculated, and the obtained mean value is the first device state distribution density, providing a density benchmark for subsequent heterogeneous parameter calculation.

[0056] After that, the first device state distribution density is added to several device state distribution densities. By repeatedly executing step S121 to step S123, the density calculation is sequentially performed on each device in the device set of the same model, obtaining all device state distribution densities, providing a comprehensive density distribution data basis for subsequent heterogeneous parameter calculation.

[0057] Through the above steps, the accurate calculation of the local density based on neighbor selection is realized. By limiting the neighborhood range and statistical calculation, the density distribution characteristics of each device in the group are effectively quantified, providing reliable data support for accurately identifying heterogeneous devices.

[0058] Furthermore, the embodiment of the present application further includes: S51. When the number of the job heterogeneous devices is equal to 0, in the visualization configuration interface, a health label is given to the device set of the same model; S52. When the device state prediction data is inconsistent with the heterogeneous device state data, in the visualization configuration interface, an execution anomaly label is given to the job heterogeneous devices of the device set of the same model, and a health label is given to the non - job heterogeneous devices of the device set of the same model.

[0059] In a feasible implementation manner, when the number of job heterogeneous devices is equal to 0, it indicates that during the heterogeneous state sorting process in step S1, all devices in the device set of the same model are not identified as devices with abnormal states, that is, the operating states of the entire device set are within the normal range. In this case, a health label is given to the entire device set of the same model in the visualization configuration interface. This health label visually shows the overall health state of the device set to the operator through graphical interface elements (such as green status indicators, normal operation icons, etc.), indicating that there is no need for special state prediction and abnormal monitoring currently, and the device set of the same model can operate normally.

[0060] When the device status prediction data is inconsistent with the heterogeneous device status data, it indicates that there is a significant deviation between the prediction result of the device status prediction model in step S3 and the actually collected heterogeneous device status, and the prediction verification fails. In this case, first, the heterogeneous devices in the same model device set are marked as abnormal on the visual configuration interface. Among them, the abnormal mark highlights the abnormal status of these devices through eye-catching visual elements (such as red warning icons, abnormal status prompts, etc.), reminding the operator to pay key attention and handle them in a timely manner. Since the prediction verification fails, these devices are considered to have unpredictable abnormal risks and require manual intervention or further inspection. At the same time, the non-operating heterogeneous devices in the same model device set are marked as healthy. Among them, the non-operating heterogeneous devices refer to normal devices that are not identified as abnormal in the heterogeneous status sorting. Since the failure of the prediction verification does not affect the determination of the normal status of these devices, they can still be marked as healthy to ensure that the operator can accurately distinguish the normal devices and abnormal devices in the device set.

[0061] Through the above steps, intelligent device status management based on heterogeneous detection results and prediction verification results is realized, providing accurate status identification and classification processing for device monitoring in different scenarios, and effectively supporting device operation and maintenance decision-making.

[0062] Furthermore, the device control parameters of the job heterogeneous devices are extracted from the control logic layer through the application layer, and the device status prediction model of the data analysis module is retrieved based on the heterogeneous device model. The underground coal mine environment parameters and the execution status of the device control parameters are processed to perform status prediction, and device status prediction data is obtained, including: S361. Constrained by the preset device model, collect multiple groups of training sample data with the service life less than or equal to the first service life. Among them, any one of the multiple groups of training sample data includes device control record data, deployment environment record parameters, and a label indicating the device status data; S362. Using the label indicating the device status data as the supervision, and using the device control record data and the deployment environment record parameters as the input, train the first device status predictor; S363. Statistically analyze the first residual data set of the first device status predictor; S364. When the number of residuals in the first residual data set is greater than or equal to the quantity threshold, using the first residual data set as the supervision, and using the device control record data and the deployment environment record parameters as the input, train the second device status predictor; S365. Until the number of residuals in the Qth residual data set is less than the quantity threshold, sum the outputs of the first device status predictor, the second device status predictor, and up to the Qth device status predictor to obtain the device status prediction model, which is stored in association with the preset device model and embedded in the data analysis module.

[0063] In a preferred embodiment, first, with a preset device model as a constraint condition, multiple groups of training sample data with a service life less than or equal to the first service life are collected. The multiple groups of training sample data constitute the basic data set for training the device state prediction model. Each group of training sample data includes device control record data, deployment environment record parameters, and a label identifying the device state data. Among them, the device control record data records the control parameter setting information of the device during historical operation; the deployment environment record parameters record the external environmental condition information where the device is located; the label identifying the device state data is used as the target output of supervised learning and identifies the actual state performance of the device under specific control and environmental conditions. By restricting the service life range of the training samples, the consistency and representativeness of the training data are ensured.

[0064] Then, using the label identifying the device state data as the supervision signal and the device control record data and deployment environment record parameters as input features, the first device state predictor is trained. The first device state predictor uses machine learning algorithms (such as neural networks, support vector machines, or decision trees, etc.) to establish a mapping relationship between the input features and the device state through the supervised learning method. The first device state predictor, as the basic prediction model, can handle most common device state prediction tasks. Subsequently, the first residual data set of the first device state predictor is statistically analyzed. The first residual data set contains the prediction error information of the first device state predictor on the training data, that is, the set of differences between the predicted values and the true labels, which reflects the data patterns and prediction blind spots that the first device state predictor has not learned, providing the target direction for subsequent model improvement.

[0065] When the number of residuals in the first residual data set is greater than or equal to the preset number threshold, it indicates that there is still room for improvement in the prediction performance of the first device state predictor, and the second round of predictor training is started. Using the first residual data set as the new supervision signal and still using the device control record data and deployment environment record parameters as the input, the second device state predictor is trained. The second device state predictor specifically learns the prediction residual pattern of the first device state predictor to correct the prediction bias of the first device state predictor.

[0066] The system repeatedly executes step S363 and step S364 until the number of residuals in the Q - th residual data set is less than the quantity threshold, indicating that the prediction accuracy has reached the preset requirement. At this time, perform an addition operation on the outputs of the first device state predictor, the second device state predictor up to the Q - th device state predictor to obtain the final device state prediction model. This device state prediction model integrates the prediction capabilities of multiple sub - predictors and improves the prediction performance through summation fusion. The final device state prediction model is associated with the preset device model and stored, and is embedded in the data analysis module to provide model support for subsequent online prediction.

[0067] Through the above steps, progressive model training based on residual learning is achieved. Compared with the single - predictor method, this integrated training method can effectively improve the accuracy and generalization ability of the prediction model, providing a more reliable technical guarantee for equipment state prediction in complex industrial environments.

[0068] Furthermore, the process for determining the labels of the device state data includes: S3611: Receive the rated range of device control attributes from the control logic layer, and receive the constraint range of environmental attributes from the visual configuration interface, and randomly configure the device control record data and environmental record parameters; S3612: With the preset device model, the device control record data, and the environmental record parameters as constraints, collect an operating sample set with a service life less than or equal to the first service life, where the operating sample set includes a device state data set; S3613: Extract the centroid device state data from the device state data set to obtain the labels identifying the device state data.

[0069] In a preferred embodiment, first, receive the rated range of device control attributes from the control logic layer and receive the constraint range of environmental attributes from the visual configuration interface. Based on these constraint ranges, randomly configure the device control record data and environmental record parameters. Among them, the rated range of device control attributes defines the normal working range of each device control parameter, such as the minimum and maximum values of control quantities such as voltage, current, and rotational speed; the constraint range of environmental attributes defines the typical change range of the device deployment environment, such as the range of environmental parameters such as temperature, humidity, and pressure. Use the random sampling method within the rated range of device control attributes and the constraint range of environmental attributes to generate multiple different combinations of device control record data and environmental record parameters, ensuring that the training samples can cover various possible working states of the device.

[0070] Subsequently, with the preset device model, device control record data, and environmental record parameters as constraints, an operation sample set with a service life less than or equal to the first service life is collected. This operation sample set refers to the actual device operation records retrieved from the historical operation database under specific constraints. This collection process ensures the consistency and comparability of the sample data: the device model constraint guarantees the hardware consistency of the samples; the control and environmental parameter constraints guarantee the operating condition consistency of the samples; the service life constraint guarantees the time consistency of the samples. Among them, the operation sample set includes a device state data set, that is, the actual operation state records of the device under the constraint conditions, including multi-dimensional state parameters such as temperature, vibration, and current.

[0071] Next, centroid device state data extraction is performed on the device state data set to obtain labels identifying the device state data. This centroid device state data extraction refers to extracting representative central state values from the device state data under multiple similar operating conditions through statistical methods. Specifically, clustering analysis or centroid calculation is performed on all the device state data in the operation sample set to extract the centroid values (such as mean, median, or cluster center, etc.) of each state parameter as the typical device state performance under this combination of operating conditions. The label identifying the device state data is the standardized state data obtained through centroid extraction, which has good representativeness and stability and can effectively guide the training process of the device state prediction model.

[0072] Through the above steps, the automated generation and standardization processing of training labels are realized, solving the technical problem of large-scale training sample annotation, providing a reliable data basis for the efficient training of the device state prediction model, and improving the automation degree and label quality of model training.

[0073] Furthermore, for the device state perturbation amount record data of the first device state record sample and the second device state record sample, co-attribute mode analysis is performed to obtain the device state perturbation amplitude, including: S331. Extract the positive perturbation amount record data and the negative perturbation amount record data from the device state perturbation amount record data, where the positive perturbation amount refers to the second device state record being greater than the first device state record value, and the negative perturbation amount refers to the second device state record being less than the first device state record value; S332. Perform co-attribute mode analysis on the positive perturbation amount record data to obtain the device positive perturbation amplitude; S333. Perform co-attribute mode analysis on the negative perturbation amount record data to obtain the device negative perturbation amplitude; S334. Add the device positive perturbation amplitude and the device negative perturbation amplitude to the device state perturbation amplitude.

[0074] In a preferred embodiment, by distinguishing positive and negative perturbations, a two-way quantitative analysis of the changing trend of the device state is achieved.

[0075] First, from the device state perturbation amount record data, positive perturbation amount record data and negative perturbation amount record data are extracted. Among them, the positive perturbation amount refers to the situation where the second device state record is greater than the first device state record, reflecting the change pattern in which the device state parameter shows an upward trend with the increase of the service life. The negative perturbation amount refers to the situation where the second device state record is less than the first device state record, reflecting the change pattern in which the device state parameter shows a downward trend with the increase of the service life. Through this directional classification, the change directions of different device state parameters can be accurately identified, laying a foundation for subsequent separate analysis.

[0076] Then, perform a same-attribute mode analysis on the positive perturbation amount record data to obtain the device positive perturbation amplitude. This analysis process statistically counts the occurrence frequencies of the perturbation amount values for all state parameters showing a positive change trend according to the parameter type, identifies the positive perturbation value with the highest occurrence frequency, and obtains the device positive perturbation amplitude. The device positive perturbation amplitude reflects the typical amplitude of the upward change of the device state parameter, providing a quantitative basis for the positive adjustment in the prediction and correction.

[0077] At the same time, perform a same-attribute mode analysis on the negative perturbation amount record data to obtain the device negative perturbation amplitude. This analysis process statistically counts the occurrence frequencies of the perturbation amount values for all state parameters showing a negative change trend according to the parameter type, identifies the negative perturbation value with the highest occurrence frequency, and obtains the device negative perturbation amplitude. The device negative perturbation amplitude reflects the typical amplitude of the downward change of the device state parameter, providing a quantitative basis for the negative adjustment in the prediction and correction.

[0078] Subsequently, add the device positive perturbation amplitude and the device negative perturbation amplitude into the device state perturbation amplitude. The device state perturbation amplitude contains the positive and negative two-way perturbation information of each state parameter, forming a complete perturbation feature description. For example, the perturbation amplitude of a certain device can be expressed as {temperature perturbation amplitude: +2°C, vibration perturbation amplitude: +0.5 mm / s, current perturbation amplitude: -1.2 A,...}, comprehensively reflecting the two-way change law of each device state parameter with the service life.

[0079] Through the above steps, a classification and statistical analysis based on the perturbation direction is achieved. Compared with the simple overall mode analysis, it can more accurately quantify the changing trend of the device state, provide more accurate and comprehensive correction parameters for subsequent prediction and correction, and improve the accuracy and applicability of the service life correction.

[0080] Example 2, as Figure 2As shown, based on the same inventive concept as the distributed job node status prediction method based on software-defined control provided in Embodiment 1, an embodiment of the present invention further provides a distributed job node status prediction system. This system is applied to an SDC system, which includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visual configuration interface and a data analysis module. The distributed job node status prediction includes: A data acquisition and sorting unit 11, configured to collect device status data of a set of devices of the same model executing a first task through the hardware abstraction layer, perform heterogeneous status sorting, and obtain job heterogeneous devices; A heterogeneous device upload unit 12, configured to upload the heterogeneous device model, heterogeneous device status data, and underground coal mine environment parameters to the application layer when the number of the job heterogeneous devices is not equal to 0; A status prediction and analysis unit 13, configured to extract device control parameters of the job heterogeneous devices from the control logic layer through the application layer, retrieve a device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain device status prediction data; A health identification display unit 14, configured to, when the device status prediction data is consistent with the heterogeneous device status data, perform health identification on the set of devices of the same model in the visual configuration interface.

[0081] Furthermore, the training sample devices of the device status prediction model are less than or equal to the first service life; the status prediction and analysis unit 13 includes the following execution steps: Obtain the second service life of the job heterogeneous devices; Through the device sample index model of the data analysis module, set the underground coal mine environment attributes and device control attributes as constant attributes, retrieve a first device status record sample that meets the heterogeneous device model and the first service life, and retrieve a second device status record sample that meets the heterogeneous device model and the second service life; Perform co-attribute mode analysis on the device status perturbation amount record data of the first device status record sample and the second device status record sample to obtain the device status perturbation amplitude; Extract device control parameters of the job heterogeneous devices from the control logic layer through the application layer, retrieve a device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain initial device status prediction data; Based on the device status perturbation amplitude, correct the initial device status prediction data to obtain the device status prediction data.

[0082] Further, the device status data includes several device status data; the data acquisition and sorting unit 11 includes the following execution steps: Perform a similarity analysis on the several device status data to obtain multiple device status similarities; Based on the multiple device status similarities, traverse the several device status data for local density evaluation to obtain several device status distribution densities; Use the distribution density mean value to traverse and calculate the ratio of the several device status distribution densities to obtain several device status heterogeneity parameters; Based on the several device status heterogeneity parameters, extract the device item numbers from the same model device set whose device status heterogeneity parameters are greater than or equal to the predefined heterogeneity parameter threshold, and add them to the job heterogeneous devices.

[0083] Further, the data acquisition and sorting unit 11 further includes the following execution steps: Extract the first device status data from the several device status data; Extract multiple selected device status similarities based on the first device status data from the multiple device status similarities; Sort the multiple selected device status similarities from largest to smallest, screen 0.5 times the number of the same model device set of neighbor selected device status similarities, perform mean calculation to obtain the first device status distribution density; Add the first device status distribution density to the several device status distribution densities.

[0084] Further, the embodiment of the present application further includes a device status identification unit, and the execution steps of the device status identification unit include: When the number of the job heterogeneous devices is equal to 0, perform a health identification on the same model device set in the visualization configuration interface; When the device status prediction data is inconsistent with the heterogeneous device status data, perform an execution anomaly identification on the job heterogeneous devices of the same model device set and perform a health identification on the non-job heterogeneous devices of the same model device set in the visualization configuration interface.

[0085] Further, the status prediction and analysis unit 13 includes the following execution steps: Constrained by a preset device model, collect multiple groups of training sample data whose service life is less than or equal to the first service life. Among them, any one of the multiple groups of training sample data includes device control record data, deployment environment record parameters, and a label indicating device status data; Using the label indicating device status data as supervision, and using the device control record data and the deployment environment record parameters as inputs, train a first device status predictor; Statistically analyze the first residual data set of the first device status predictor; When the number of residuals in the first residual data set is greater than or equal to the quantity threshold, use the first residual data set as supervision, and use the device control record data and the deployment environment record parameters as inputs to train a second device status predictor; Until the number of residuals in the Qth residual data set is less than the quantity threshold, sum the outputs of the first device status predictor, the second device status predictor, up to the Qth device status predictor to obtain the device status prediction model, which is stored in association with the preset device model and embedded in the data analysis module.

[0086] Furthermore, the process for determining the label of the device status data includes: Receive the rated range of device control attributes from the control logic layer, and receive the constraint range of environmental attributes from the visual configuration interface, and randomly configure the device control record data and the environmental record parameters; Taking the preset device model, the device control record data, and the environmental record parameters as constraints, collect an operation sample set with a service life less than or equal to the first service life, where the operation sample set includes a device status data set; Extract the centroid device status data from the device status data set to obtain the label identifying the device status data.

[0087] Furthermore, the status prediction analysis unit 13 also includes the following execution steps: Extract the positive disturbance amount record data and the negative disturbance amount record data from the device status disturbance amount record data, where the positive disturbance amount refers to the second device status record being greater than the first device status record value, and the negative disturbance amount refers to the second device status record being less than the first device status record value; Perform the same-attribute mode analysis on the positive disturbance amount record data to obtain the device positive disturbance amplitude; Perform the same-attribute mode analysis on the negative disturbance amount record data to obtain the device negative disturbance amplitude; Add the device positive disturbance amplitude and the device negative disturbance amplitude to the device status disturbance amplitude.

[0088] It should be noted that in the above embodiments, the descriptions of the various embodiments have their own emphases. For parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0089] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0090] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0091] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0093] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they know the basic creative concept.

[0094] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention also intends to include these changes and modifications.

Claims

1. A method for predicting the state of distributed job nodes based on software-defined control, characterized in that Applied to the SDC system, the SDC system includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visualization configuration interface and a data analysis module. The execution steps include: Collect the device status data of the same-type device set performing the first task through the hardware abstraction layer, perform heterogeneous status sorting, and obtain job heterogeneous devices; When the number of the job heterogeneous devices is not equal to 0, upload the heterogeneous device model, the heterogeneous device status data, and the underground coal mine environment parameters to the application layer; Extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain device status prediction data; When the device status prediction data is consistent with the heterogeneous device status data, perform a health identification on the same-type device set in the visualization configuration interface.

2. The method according to claim 1, characterized in that The training sample devices of the device status prediction model are less than or equal to the first service life. Extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain device status prediction data, including: Obtain the second service life of the job heterogeneous devices; Through the device sample index model of the data analysis module, set the underground coal mine environment attributes and device control attributes as constant attributes, retrieve the first device status record sample that meets the heterogeneous device model and the first service life, and retrieve the second device status record sample that meets the heterogeneous device model and the second service life; Perform the same-attribute mode analysis on the device status perturbation amount record data of the first device status record sample and the second device status record sample, and obtain the device status perturbation amplitude; Extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the device control parameters to perform status prediction, and obtain the initial device status prediction data; Based on the device status perturbation amplitude, correct the initial device status prediction data to obtain the device status prediction data.

3. The method according to claim 1, characterized in that, The device status data includes a number of device status data. Collect the device status data of the same-type device set performing the first task through the hardware abstraction layer, perform heterogeneous status sorting, and obtain job heterogeneous devices, including: Perform a similarity analysis on the number of device status data to obtain a number of device status similarities; Based on the number of device status similarities, traverse the number of device status data for local density evaluation to obtain a number of device status distribution densities; Use the distribution density mean value to traverse and divide the number of device status distribution densities to obtain a number of device status heterogeneous parameters; Based on the several heterogeneous device status parameters, extract the device tag numbers of the devices in the same model device set whose heterogeneous device status parameters are greater than or equal to the predefined heterogeneous parameter threshold, and add them to the job heterogeneous devices.

4. The method according to claim 3, wherein Based on the multiple device status similarities, traverse the several device status data for local density evaluation to obtain several device status distribution densities, including: Extract the first device status data from the several device status data; Extract multiple selected device status similarities based on the first device status data from the multiple device status similarities; From the multiple selected device status similarities in descending order, screen out 0.5 times the number of the same model device set of neighbor selected device status similarities, perform mean calculation to obtain the first device status distribution density; Add the first device status distribution density to the several device status distribution densities.

5. The method according to claim 1, wherein It further includes: When the number of the job heterogeneous devices is 0, perform health identification on the same model device set in the visualization configuration interface; When the device status prediction data is inconsistent with the heterogeneous device status data, perform an execution exception identification on the job heterogeneous devices of the same model device set in the visualization configuration interface, and perform health identification on the non-job heterogeneous devices of the same model device set.

6. The method according to claim 2, wherein Extract the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, call the device status prediction model of the data analysis module based on the heterogeneous device model, process the underground coal mine environment parameters and the execution status prediction of the device control parameters to obtain device status prediction data, including: Taking the preset device model as a constraint, collect multiple groups of training sample data whose service life is less than or equal to the first service life. Any one of the multiple groups of training sample data includes device control record data, deployment environment record parameters, and a label identifying the device status data; Taking the label identifying the device status data as a supervision, and taking the device control record data and the deployment environment record parameters as inputs, train the first device status predictor; Statistically analyze the first residual data set of the first device status predictor; When the number of residuals in the first residual data set is greater than or equal to the quantity threshold, taking the first residual data set as a supervision, and taking the device control record data and the deployment environment record parameters as inputs, train the second device status predictor; Until the number of residuals in the Qth residual data set is less than the quantity threshold, sum up the outputs of the first device status predictor, the second device status predictor until the Qth device status predictor to obtain the device status prediction model, associate and store it with the preset device model, and embed it in the data analysis module.

7. The method according to claim 6, wherein The determination process of the label identifying the device status data includes: Receive the rated interval of the device control attribute from the control logic layer, and receive the environmental attribute constraint interval from the visualization configuration interface, and randomly configure the device control record data and the environmental record parameters; Collect an operation sample set with a service life less than or equal to the first service life, constrained by the preset device model, the device control record data, and the environmental record parameters, where the operation sample set includes a device status data set. Extract the centroid device status data from the device status data set to obtain a label identifying the device status data.

8. The method according to claim 2, wherein Perform a same-attribute mode analysis on the device status perturbation amount record data of the first device status record sample and the second device status record sample to obtain the device status perturbation amplitude, including: Extract the positive perturbation amount record data and the negative perturbation amount record data from the device status perturbation amount record data, where the positive perturbation amount refers to the second device status record being greater than the first device status record value, and the negative perturbation amount refers to the second device status record being less than the first device status record value; Perform a same-attribute mode analysis on the positive perturbation amount record data to obtain the device positive perturbation amplitude; Perform a same-attribute mode analysis on the negative perturbation amount record data to obtain the device negative perturbation amplitude; Add the device positive perturbation amplitude and the device negative perturbation amplitude to the device status perturbation amplitude.

9. A distributed job node status prediction system based on software-defined control, characterized in that, For implementing the method according to any one of claims 1 to 8, applied to an SDC system, the SDC system includes a hardware abstraction layer, a control logic layer, and an application layer. The application layer includes a visualization configuration interface and a data analysis module. The execution steps include: A data acquisition and sorting unit for collecting device status data of a set of devices of the same model performing the first task through the hardware abstraction layer, performing heterogeneous state sorting to obtain job heterogeneous devices; A heterogeneous device upload unit for uploading the heterogeneous device model, the heterogeneous device status data, and the underground coal mine environmental parameters to the application layer when the number of the job heterogeneous devices is not equal to 0; A status prediction and analysis unit for extracting the device control parameters of the job heterogeneous devices from the control logic layer through the application layer, invoking the device status prediction model of the data analysis module based on the heterogeneous device model, and processing the underground coal mine environmental parameters and the device control parameters to perform status prediction to obtain device status prediction data; A health identification display unit for performing health identification on the set of devices of the same model in the visualization configuration interface when the device status prediction data is consistent with the heterogeneous device status data.

Images