Intelligent production scheduling system and method for coal industry scene

Abstract

The invention provides an intelligent production scheduling system and method for a coal industry scene, and the method comprises the steps: fusing multi-source production information (including equipment state information, coal quality index information and safety detection information) through a server, constructing a fault prediction model, and carrying out the fault prediction of the real-time production information; and generating a task list based on the fault prediction result, the real-time production information and the historical production information, matching a workflow and an optimal executor for each task in the task list, then sending the task to the PC terminal for execution, monitoring the execution progress of each task in real time, and urging the overdue task. And generating a corresponding report for each executed task based on the fused information, the fault prediction result, the real-time production information, the historical production information and the execution process information of each task, and sending the report to a display device for display. By adopting the method, the efficiency and the accuracy of production scheduling in a coal industry scene can be improved, the real-time performance of production scheduling can be ensured, and the statistics and the checking of the whole process link can be facilitated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial data processing, and in particular to a smart production scheduling system and method for a coal industry scene. BACKGROUND

[0002] In the coal industry scene, the traditional production scheduling method mainly involves screening from collected data, which requires a high degree of reliance on human experience, and dispatchers issue instructions based on their accumulated knowledge and judgment. This makes information transmission mainly rely on telephone, intercom and other means, which is prone to information distortion, omission and delay, and cannot guarantee timely coordination of each link; resource allocation also lacks precise data support, making it difficult to be scientific and reasonable, and situations such as equipment idling and material accumulation are common; in the face of failures, manual troubleshooting is time-consuming and slow to respond, resulting in prolonged production downtime. In the current wave of large-scale expansion of coal preparation plants and intelligentization, the drawbacks of the traditional production scheduling method have become increasingly prominent and have become difficult to meet the new development needs.

[0003] Because human intervention is involved in the important data screening link, information omission or errors are likely to occur in the data transmission process, leading to information distortion and delay, which affects the coordination efficiency; and the data is not analyzed accurately, which cannot dynamically optimize resource allocation, leading to resource waste; at the same time, when the equipment is abnormal, it cannot immediately make a real-time early warning and treatment plan, resulting in a very low safety factor in the production process; finally, the data generated in the entire process during production needs to be manually sorted, and multi-dimensional analysis reports cannot be automatically generated, which is not conducive to the statistics and viewing of the entire process link. SUMMARY

[0004] Therefore, the present application aims to provide a smart production scheduling system and method for a coal industry scene to alleviate the above-mentioned problems existing in the traditional production scheduling method in the coal industry scene.

[0005] In a first aspect, the embodiments of the present application provide a smart production scheduling system for a coal industry scene, applied to a server, comprising: an information fusion module configured to fuse production information from a plurality of different data sources, and construct a fault prediction model based on the fused information; wherein the production information comprises device state information, coal quality index information, and safety detection information; a task generation and distribution module configured to perform fault prediction on real-time production information through the fault prediction model, generate a task list based on the fault prediction result, the real-time production information, and historical production information, then match a corresponding workflow and an optimal executor for each task in the task list, and send each task matched with the workflow and the optimal executor to a PC terminal connected with the server for execution; wherein the workflow defines the order, conditions, and rules of execution of the corresponding task; a task monitoring and feedback module configured to monitor the execution progress of each task in real time, and determine whether there is an overdue task based on the execution progress of each task, and if there is an overdue task, urge each overdue task; a report automatic generation module configured to generate a corresponding report for each executed task based on the fused information, the fault prediction result, the real-time production information, the historical production information, and the execution process information of each task, and send the obtained report to a display device connected with the server for display.

[0006] In a second aspect, the embodiments of the present application also provide a smart production scheduling method for a coal industry scene, applied to the server of the first aspect, comprising: fusing production information from a plurality of different data sources, and constructing a fault prediction model based on the fused information; wherein the production information comprises device state information, coal quality index information, and safety detection information; performing fault prediction on real-time production information through the fault prediction model, generating a task list based on the fault prediction result, the real-time production information, and historical production information, then matching a corresponding workflow and an optimal executor for each task in the task list, and sending each task matched with the workflow and the optimal executor to a PC terminal connected with the server for execution; wherein the workflow defines the order, conditions, and rules of execution of the corresponding task; monitoring the execution progress of each task in real time, and determining whether there is an overdue task based on the execution progress of each task, and if there is an overdue task, urging each overdue task; generating a corresponding report for each executed task based on the fused information, the fault prediction result, the real-time production information, the historical production information, and the execution process information of each task, and sending the obtained report to a display device connected with the server for display.

[0007] The embodiment of the present application provides a kind of intelligent production scheduling system and method of coal industry scene, server fusion multi-source production information (including equipment state information, coal quality index information and safety detection information) constructs fault prediction model to carry out fault prediction to real-time production information, generates task list based on fault prediction result, real-time production information and historical production information, for each task in task list Match workflow and optimal executor are sent to PC end and executed, the execution progress of each task is monitored in real time and overdue task is urged, and based on the information after fusion, fault prediction result, real-time production information, historical production information and the execution process information of each task, the corresponding report of each task executed is generated and sent to display equipment for display.The above-mentioned technology can be used to construct a fault prediction model using multi-source production information to realize automatic fault prediction, automatically generate tasks in combination with real-time production information and historical production information, match workflow and optimal executor for the tasks, and then send the tasks to PC end for execution, so as to realize automatic production scheduling, improve the efficiency and accuracy of production scheduling in coal industry scene;The overdue task can be urged by monitoring the task execution progress in real time, which is beneficial to ensure the real-time performance of production scheduling;In combination with the related information in the production scheduling and task execution process, a report can be generated and sent to the display device for display, so as to realize automatic generation of visual report, which is beneficial to the statistics and viewing of the whole process link.

[0008] Other features and advantages of the present application will be set forth in the descriptions below, and in part will become apparent to those skilled in the art from the descriptions, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and achieved by the structure particularly pointed out in the specification, claims and drawings.

[0009] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings without creative labor based on these drawings.

[0011] Figure 1 The structure diagram of the intelligent production scheduling system of coal industry scene in the embodiment of the present application is shown in the figure. Figure 2 The work flow configuration interface diagram in the embodiment of the present application is shown in the figure. Figure 3A work flow diagram of the intelligent production scheduling system in the embodiment of the present application; Figure 4 A flowchart of the intelligent production scheduling method in the coal industry scene in the embodiment of the present application. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme of the present application will be described clearly and completely below in combination with embodiments. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0013] At present, the traditional production scheduling mode in the coal industry scene mainly has the following problems: (1) the data such as equipment operation parameters, material flow, safety detection is scattered, and lacks real-time integration and sharing; (2) abnormal events depend on manual investigation, the task allocation efficiency is low, and the fault processing period is long; (3) lacking data-driven intelligent analysis, resource allocation depends on subjective judgment, which is easy to cause equipment idling or material backlog; (4) the task execution process lacks transparent tracking, and it is difficult to realize closed-loop management.

[0014] Based on this, the intelligent production scheduling system and method of the coal industry scene provided by the embodiment of the present application can alleviate the above problems existing in the traditional production scheduling mode in the coal industry scene.

[0015] In order to facilitate the understanding of the present embodiment, first, a coal industry scene intelligent production scheduling system disclosed in the embodiment of the present application is introduced in detail, which can be applied to a server, as shown in Figure 1 The intelligent production scheduling system can include the following modules: The information fusion module 102 is configured to fuse production information from multiple different data sources, and construct a fault prediction model based on the fused information.

[0016] Among them, the production information can include equipment state information, coal quality index information, safety detection information, etc., which is not limited.

[0017] Specifically, multi-source data of industrial Internet of Things (IoT) devices, digital twin models, and AI large models can be integrated, including device status, coal quality indicators, safety detection data, etc. Device status can include power-on, shutdown, running, failure, etc. Coal quality indicators can include raw coal ash content, clean coal ash content, TDS content, and gangue rate (determined by the coal quality properties of different plants). Safety detection can include video monitoring device status, regional intrusion, illegal personnel intrusion, safety helmet recognition, fire detection, post patrol supervision, key process equipment detection, etc.

[0018] The fault prediction model can be trained using machine learning or deep learning models, such as using support vector machine (SVM), random forest, or neural network algorithms for training, and then evaluating the model to analyze its accuracy and reliability. The performance of the model can be evaluated through cross-validation, confusion matrix, etc. Based on the performance evaluation results of the model, the model parameters are adjusted to optimize the model performance.

[0019] The task generation and distribution module 104 is used to perform fault prediction on real-time production information through the fault prediction model, and generate a task list based on the fault prediction results, real-time production information, and historical production information. Then, for each task in the task list, a corresponding workflow and optimal executor are matched, and each task matched with a workflow and an optimal executor is sent to a PC connected to the server for execution.

[0020] The workflow defines the order, conditions, and rules of the corresponding task execution.

[0021] The task monitoring and feedback module 106 is used to monitor the execution progress of each task in real time, and determine whether there are overdue tasks based on the execution progress of each task. If there are overdue tasks, each overdue task is reminded.

[0022] The report automatic generation module 108 is used to generate a corresponding report for each executed task based on the fused information, fault prediction results, real-time production information, historical production information, and execution process information of each task, and send the obtained report to a display device connected to the server for display.

[0023] The intelligent production scheduling system for a coal industry scene provided by the embodiment of the present application can utilize multi-source production information to construct a fault prediction model to realize automatic fault prediction, automatically generate a task in combination with real-time production information and historical production information, match a workflow and an optimal executor for the task, and then send the task to a PC end for execution, so as to realize automatic production scheduling and improve the efficiency and accuracy of production scheduling in a coal industry scene. The overdue task can be urged by monitoring the task execution progress in real time, which is beneficial to ensuring the real-time performance of production scheduling. In addition, a report can be generated in combination with relevant information in the production scheduling and task execution process and sent to a display device for display, so as to realize automatic generation of a visual report and facilitate the statistics and viewing of the entire process link.

[0024] As a possible implementation, the fusion of the production information from the multiple different data sources can include first preprocessing, first feature extraction and analysis mining of the production information. Based on this, referring to FIG. 1, the information fusion module 102 can also be configured to perform first preprocessing on the production information, perform first feature extraction and analysis mining on the first preprocessed information, and then construct a fault prediction model based on the obtained first features and analysis mining results. Figure 1

[0025] The first preprocessing can include noise removal, missing value filling, standardization processing and the like, and the analysis mining result can include association rules between device faults and coal quality fluctuations.

[0026] For ease of understanding, the production information fusion mode is described below by way of example with actual application as an example. The production information fusion can be performed according to the following operation mode: First, collect device fault data and coal quality fluctuation data. The device fault data can be obtained from the running log, sensor monitoring data and maintenance record of the device, and the coal quality fluctuation data can be obtained through a coal quality detection device. Ensuring the accuracy and integrity of the collected data is crucial, and the accuracy of the collected data can be improved by regularly calibrating the sensor and checking the data transmission line.

[0027] Then, the collected data is preprocessed to ensure the consistency of the data, unify the data format and standardize the data acquisition process. The preprocessing mainly includes heterogeneous data standardization and time series alignment.

[0028] Heterogeneous data standardization: uniform format conversion (for example, JSON standardization) is performed on heterogeneous data from IoT devices (such as vibration sensors, temperature sensors, etc.), video monitoring, manual inspection records and the like.

[0029] ​Time series alignment: use the dynamic time warping (DTW) algorithm to align data streams of different sampling frequencies (for example, align device status data with video streams according to timestamps). Specifically, device status data can include sensor data (vibration, temperature, current, etc.), device start-stop status, fault signals, etc. Device status data can be aligned with video streams, manual inspection records, etc. to accurately attribute abnormal events; video monitoring data can include safety helmet recognition data, area intrusion recognition data, dangerous behavior detection data, etc. Video monitoring data can be aligned with device status data, safety detection data according to timestamps to ensure the spatio-temporal correlation of behavior and device anomalies; production execution data can include start-stop operation logs, material flow, coal quality test results (such as ash content, gangue content, etc.), etc. Production execution data can be aligned with device status data, task execution records according to timestamps to optimize production scheduling decisions (such as when the ash content of clean coal fluctuates, device parameters can be traced back according to timestamps); environmental monitoring data can include real-time monitoring values of under-warehouse medium concentration, barrel liquid level, dust, noise, etc. Real-time monitoring values can be aligned with device running status to support safety warning linkage (such as when the high liquid level of the concentrated medium barrel is detected, a barrel overflow warning is given and the device is stopped).

[0030] Then the first key features are extracted from the preprocessed device fault data and coal quality fluctuation data. For example, the running parameters, status information and event records of the device can be extracted from the device operation log as the first key features, and various indicators of coal quality (such as ash content, moisture content, volatile matter) can be extracted from the coal quality fluctuation data as the first key features.

[0031] Then the first key features are analyzed and mined. Specifically, the Apriori algorithm can be used to mine the association rules between device faults and coal quality fluctuations (for example, when the vibration anomaly threshold is greater than X, the probability of coal quality ash deviation increases by 60%). The strength of the association rules can be measured by support and confidence. Support represents the frequency of the association rule in the data set, and confidence represents the probability of the occurrence of the association rule under given conditions.

[0032] Then the training set and the test set are constructed based on the extracted first key features and the mined association rules, and the long short-term memory (LSTM) network is trained using the training set and the performance of the LSTM network is tested using the test set, so as to use the trained model as a fault prediction model to realize the construction of the fault prediction model. The constructed fault prediction model can be applied to actual production to monitor the relationship between device faults and coal quality fluctuations through the fault prediction model, discover potential problems in time and take corresponding measures; the practicality and effectiveness of the fault prediction model can be verified through actual cases.

[0033] As a possible implementation, the real-time production information can include structured data and unstructured data, the structured data can be device sensor data (such as vibration, temperature, current, etc.), production indicators (such as coal ash content, processing capacity, etc.), task execution records, etc., and the unstructured data can be video streams (such as safety behavior identification data, etc.), voice records (such as intercom instruction data, etc.), text reports (such as manual inspection logs, etc.), etc., and the fault prediction result can include the fault probability of the device corresponding to the real-time production information. Based on this, referring to Figure 1 As shown, the task generation and distribution module 104 can also be configured to: perform second preprocessing on the structured data and the unstructured data, and perform second feature extraction on the preprocessed information, and then input the extracted second features into the fault prediction model to perform fault prediction, and output the fault probability of the device corresponding to the real-time production information through the fault prediction model; if the fault probability of the target device is greater than the preset threshold, the task corresponding to the target device is generated based on the fault probability of the target device and the real-time production information and the historical production information corresponding to the target device.

[0034] In actual application, the operation mode of predicting the fault of the real-time production information through the fault prediction model can include: first, collecting real-time production information data, which is divided into structured data and unstructured data; next, further preprocessing the collected data, including aligning according to the timestamp, and specifically using the DTW algorithm to align heterogeneous data streams (such as 1Hz sensor data and 25fps video frames); then, extracting features from the preprocessed data, for numerical data, extracting time series features through sliding window statistics (such as mean, variance, peak, etc.), for images and videos, using ResNet model to extract visual features of safety violations and device states, and for text data, extracting entities (such as fault types, processing measures, etc.) and semantic relationships through BERT model; then, inputting the extracted features into the fault prediction model (i.e. pre-trained LSTM network) to predict the device fault probability (such as vibration overrun probability in the next 30 minutes, etc.). The fault prediction process is a multi-modal prediction process combining visual data, sensor data and text data, which weights the input features through attention mechanism, and finally outputs the prediction result, which can include abnormal event types (such as device fault, safety violation, coal quality deviation, etc.), fault probability, and fault impact range (such as single device or whole production line).

[0035] The task generation trigger rule can be: predicting the device failure probability by using a pre-trained LSTM network as a failure prediction model, and triggering task generation when the device failure probability is greater than a pre-set threshold value, and the threshold value can be set according to different device indexes of different factories (for example, the threshold value can be set to 60% by default, and the value of the threshold value can be adjusted according to actual conditions).

[0036] As a possible implementation, the real-time production information can include downtime information and yield loss information of the target device, and the historical production information can include historical work orders of the target device. Based on this, referring to FIG. 1, the task generation and distribution module 104 can be further configured to: perform third preprocessing on the historical work orders, and extract the failure type, failure description and solution of each target device from the third preprocessed information, and then assign a corresponding weight to each obtained solution by using a preset attention mechanism, and determine the target solution with the maximum weight for each target device; determine the task level of the task corresponding to each target device based on the downtime information and yield loss information of each target device, the preset first weight information and the preset first mapping relationship; and configure the scene type and the associated department of the task corresponding to each target device. Figure 1

[0037] Referring to FIG. 1, the information fusion module 102 can be further configured to: perform root cause analysis on the third preprocessed information, and generate the failure cause of each target device based on the root cause analysis result; wherein the failure cause is described in natural language. Figure 1

[0038] The task level can be divided into a first level representing a high-risk task, a second level representing a medium-risk task, and a third level representing a low-risk task. Based on this, referring to FIG. 1, the task generation and distribution module 104 can be further configured to: generate a corresponding task for each target device based on the target solution of each target device and the task level, the scene type, the associated department and the failure cause of the task corresponding to each target device; wherein each task includes a corresponding target solution, a task level, a scene type, an associated department and a failure cause. Figure 1

[0039] ​​​In practical application, the above task generation mode can be: based on the abnormal events (such as equipment failure, coal quality fluctuation, etc.) predicted by the large model (including the fault prediction model) to automatically generate tasks in combination with real-time production information and historical production information. The generated task contains fault reason (obtained by root cause analysis), treatment suggestion (i.e. treatment measures, i.e. solution), task level (divided into red / yellow / blue three levels), associated department (used for task organization to match workflow), scene type, etc. Task attributes. In the process of generating tasks, task attributes will be automatically filled in. The solution contained in the generated task is the high-frequency solution (i.e. target solution) extracted from the historical work order by using the attention mechanism. The task level can be dynamically calculated according to the impact range (such as equipment downtime, yield loss).

[0040] Specifically, the task generation process can be performed according to the following operation mode: First, the historical work order is cleaned and arranged, including removing redundant information, formatting data, ensuring data consistency and availability. Then, the second key features (including fault type, fault description, solution) are extracted from the historical work order. Then, the extracted second key features are used as the input of the attention mechanism, and the attention mechanism is used to weight the solutions extracted from the historical work order, so that the solutions more related to the current fault obtain higher weights. The higher the weight, the higher the frequency of the solution. The high-frequency solution can be extracted from the weighted solution as a treatment suggestion provided to the user. In the task attribute details of the task generation stage, specific treatment measures (i.e. high-frequency solutions) will be given.

[0041] Then, the task level is dynamically calculated according to the impact range. The task level can be divided into red level, yellow level and blue level. Among them, the red level represents a larger impact range (which may cause long-time equipment downtime or large yield loss), the yellow level represents a medium impact range (which may cause equipment downtime or yield loss, but the impact range is smaller than that of the red level), and the blue level represents a smaller impact range (which has limited impact on equipment downtime and yield loss). The task level is mainly calculated according to the two factors of equipment downtime and yield loss. The equipment downtime can specifically refer to the time that the task processing is not timely and may cause the equipment to stop. The longer the equipment downtime, the higher the task level. The yield loss can specifically refer to the yield loss that the task processing is not timely and may cause. The larger the yield loss, the higher the task level.

[0042] When calculating the task level, certain weight coefficients can be set, and the actual situation of the equipment downtime and yield loss is quantified to obtain the scores of the two factors (i.e., the equipment downtime score and the yield loss score), and then the scores of the two factors are weighted and summed to obtain the task level score. Finally, the task level score is mapped to the task level.

[0043] The task level score can be calculated using the following formula:

[0044] wherein, and are weight coefficients, which can be adjusted according to the actual situation.

[0045] According to the calculation result of the task level score and the pre-set score range of the red level, yellow level and blue level (i.e., there is a one-to-one mapping relationship between the three task levels and the three score ranges), the task level score can be mapped to the corresponding task level (for example, if the task level score falls within the score range of the red level, the task level score can be mapped to the red level).

[0046] After calculating the task level, the scene type and the associated department are configured for the task.

[0047] After data cleaning and sorting of historical work orders, root cause analysis is performed. The root cause analysis method can be: using Random Forest to classify historical task data (contained in historical work orders) to identify the root cause of high-frequency problems (for example, identifying that 80% of equipment failures are caused by "insufficient lubrication"). When performing root cause analysis, the Random Forest in the Python-based machine learning library sklearn can be called to generate a JSON result, and the JSON result can be converted into a natural language description (such as "possible insufficient lubrication, suggest checking the oil circuit") through a large language model.

[0048] After calculating the task level and configuring the scene type and the associated department and analyzing the failure cause, a task with empty task attribute values can be established, and the task level, scene type, associated department, and failure cause are filled into the values of the corresponding task attributes, thereby obtaining a task containing information such as task level, scene type, associated department, and failure cause, and the task generation stage is completed.

[0049] As a possible implementation, see Figure 1As shown, the task generation and distribution module 104 can also be used to match the task level, scene type and associated department of each task with multiple workflows in a preset workflow set to match the workflow corresponding to each task; and match the corresponding optimal performer for each task based on the real-time post status of the associated department of each task; wherein the real-time post status includes the working status of each executable task personnel corresponding to the associated department.

[0050] In actual application process, scene configuration (such as configuration of the association relationship of scene serial number, scene name, scene type, device number, task level, etc.), workflow design (such as configuration of the association relationship of workflow serial number, workflow name, scene type, virtual department, task level, current state, etc.) and organization configuration (such as configuration of the binding relationship of organization serial number, organization name, post name, etc.) can be performed in advance in background configuration process, and then the operation mode for task matching workflow and optimal performer mainly can include dynamic workflow matching and virtual organization mapping. Figure 2 The workflow configuration interface is shown, and the user can perform related configuration operations (such as query, new, edit, preview, delete, copy process, reset, etc.) of the workflow on the workflow configuration interface.

[0051] Dynamic workflow matching: according to the scene type, task level, and associated department (such as virtual department), the generated task is matched with the preset workflow, and through virtual organization mapping to the entity post or robot, so as to realize task assignment or broadcast task pushing.

[0052] Virtual organization mapping: through the skill group tag (such as “mechanical and electrical maintenance group”) and real-time post status (such as online / busy), the optimal performer for the generated task is matched, for example, the optimal performer can be a person with a specified skill group tag and in an online state.

[0053] Virtual organization mapping and workflow matching are two key mechanisms in the intelligent command system (i.e., intelligent production scheduling system), which cooperate with each other to achieve efficient allocation and execution of tasks. Virtual organization mapping combines multiple skill group tags (such as “mechanical and electrical maintenance group”) with real-time post status (online / busy) to dynamically identify the optimal performer currently available. This process takes into account the current working status of the performer, ensuring that the task can be assigned to the most suitable person (find the most suitable person through organization configuration relationship). Workflow defines the order, conditions and rules of task execution, and is the basis for task automation. Once the virtual organization mapping determines the optimal performer, the system will assign the task to the optimal performer according to the preset workflow and start the corresponding task processing process.

[0054] As a possible implementation, the above-mentioned server can also be connected with a mobile terminal; based on this, referring toFigure 1 As shown, the task generation and distribution module 104 can also be configured to: for each task, if the task level of the task is the first level, push the task to all mobile terminals and all display devices in the target area, and if the task level of the task is the second level or the third level, push the task to the mobile terminal corresponding to the department associated with the task.

[0055] In actual application, the push logic can be pre-configured, and the push logic can have a broadcast push mode and a directional push mode.

[0056] The broadcast push mode: push high-risk tasks (i.e., tasks with a red level) to all terminals (such as mobile phones, large screens, etc.) in the relevant area.

[0057] The directional push mode: ordinary tasks (i.e., tasks with a yellow level or a blue level) are matched to corresponding terminals through skill groups (such as matching the "lubrication task" to the mobile terminal of the personnel corresponding to the "mechanical and electrical maintenance group").

[0058] As a possible implementation, see Figure 1 As shown, the task monitoring and feedback module 106 can also be configured to: monitor the status of each task in real time, and synchronize the status of each task to the PC terminal, the mobile terminal, and the display device.

[0059] In actual application, the PC terminal, the large screen (i.e., the display device at this time), and the smart terminal (i.e., the mobile terminal at this time, such as a smart phone, a smart bracelet, a smart helmet, etc.) can be connected to the WebSocket server (i.e., the server at this time), and the task status can be synchronized through the WebSocket server, the PC terminal, the large screen, and the smart terminal, and the cross-department collaboration can be realized through voice, video, and intercom functions, and multi-terminal collaboration can be realized.

[0060] The WebSocket server generates tasks, monitors task status, and sends task status change notifications to the PC terminal. The task status synchronization process is a multi-terminal real-time communication process based on the WebSocket protocol between the WebSocket server, the PC terminal, and the smart terminal, and the event of triggering the WebSocket server to broadcast the task status change notification is triggered when the task status changes. The PC terminal is connected to the WebSocket server to receive task status change notifications and send control instructions. The smart terminal is connected to the same WebSocket server through a mobile network / Wi-Fi to realize the same two-way communication capability as the PC terminal (including receiving task status change notifications and sending control instructions). The large screen is used as a data display terminal to accept the control of the PC terminal or the smart terminal through the control instructions forwarded by the WebSocket server.

[0061] As a possible implementation, see Figure 1 As shown, the task generation and distribution module 104 can also be used to: determine the task priority score of the task corresponding to each target device based on the urgency information, impact range information and resource consumption information of each target device and the preset second weight information; for each task, if the task priority score of the task is greater than the preset score threshold, the task is pushed to a predetermined target mobile terminal and / or target display device.

[0062] In actual application, the following priority scoring models can be pre-built:

[0063] in, 、 、 The urgency level is a dynamic weight that can be adjusted based on the production stage. Urgency is a preset value based on the different environments of different factories and can be adjusted based on the current production status, equipment health, safety monitoring data, and the actual conditions of dispatchers and relevant personnel. Impact is a value evaluated for different task types by analyzing historical work orders and task processing records. The impact range value of a task can be dynamically adjusted based on current production data. Different types of tasks generally consume different resources, and the system can preset resource consumption based on task type.

[0064] Continuing with the previous example, a certain scoring threshold can be set in advance and a smart terminal and large screen can be specified for a task. When the system uses the priority scoring model to calculate the task priority score of the task, it will determine whether the task priority score of the task is greater than the scoring threshold. If the task priority score of the task is greater than the scoring threshold, the task will be pushed to the pre-specified smart terminal and large screen.

[0065] As a possible implementation, each task may have corresponding time information, which may include the occurrence time, reporting time, planned completion time, and actual completion time of the corresponding task; based on this, see Figure 1 As shown, the task monitoring and feedback module 106 can also be used to: for each task, if the execution progress of the task is not updated after reaching the first target time point, the task is determined to be an overdue task; wherein the time interval between the first target time point and the planned completion time of the corresponding task is less than the preset duration.

[0066] See also Figure 1 As shown, the task monitoring and feedback module 106 can also be used to: for each overdue task, if the execution progress of the overdue task is not updated after reaching the second target time point, add a reminder mark to the overdue task, and send the overdue task with the reminder mark to the display device for display.

[0067] Continuing from the previous example, the full life cycle tracking function of a task can be achieved through collaboration among the WebSocket server, PC, large screen, and smart terminal. This function mainly includes: real-time monitoring of task progress (such as not started, in progress, overdue, completed, overdue completion, etc.), supporting automatic reminders for overdue tasks, task feedback, and task adjustments.

[0068] Automatic reminder for overdue tasks: The progress of tasks will be updated in real time as the relevant stakeholders matched to different branches of the workflow complete the tasks. Each task has an occurrence time (the time when it occurs), a reporting time (the time when it is reported after it occurs), a planned completion time (the time closest to the actual completion time of the matching task based on historical data), and an actual completion time (the actual completion time of the task). If the task progress is not updated within 70% of the time interval between the occurrence time and the planned time of the task (70% can be adjusted according to the actual on-site situation), the system will mark the task as overdue (that is, the task is an overdue task). The system will transfer the task by selecting whether the task is an internal task or an external task when submitting the task (whether the task is an internal task or an external task has been determined in the task generation stage). If it is an internal task, it will be directly transferred to the direct leader. If it is an external task, it will be transferred according to the configuration process of the workflow. The backend service scheduled task monitors overdue tasks in real time. When the progress of an overdue task has not been updated within 80% of the time interval between the occurrence time and the planned time, the system will regularly urge the overdue task, mark the overdue tasks that need to be urged, and display them on the big screen first.

[0069] As a possible implementation, see Figure 1 As shown, the automatic report generation module 108 can also be used to: generate corresponding field information for each completed task based on the fused information, fault prediction results, real-time production information, historical production information and the execution process information of each task; use a preset report template engine to match the corresponding target report template for each completed task from the preset report template set, and render the field information and target report template corresponding to the same task to generate a report.

[0070] In actual application, the automatic generation of reports is mainly based on historical data obtained from the data of the entire life cycle of the task. The data of the life cycle can include Figure 1 The data of the information fusion module 102, the task generation and distribution module 104 and the task monitoring and feedback module 106 are shown.

[0071] In the report generation phase, the task completion rate, task average processing time (e.g., "electromechanical maintenance task average time 2.3 hours"), and high-frequency problem statistics (e.g., "bearing failure accounts for 35% of equipment problems") can be analyzed from task status, task execution duration, and task overdue records; device status data can be obtained from device operation and failure records to analyze device idle rate, failure rate (e.g., "crusher monthly average failure 3 times"), and maintenance recommendations; required data can be read from safety detection data and production index data to analyze violation events (e.g., "not wearing safety helmet accounts for 60% of violation events"), environmental abnormal trend (e.g., "concentrated medium bucket value statistics"), and coal quality fluctuation history comparison (e.g., "monthly deviation of clean coal ash ±0.5%").

[0072] In the report generation phase, real-time data can be used to analyze real-time task progress (e.g., "40% of tasks in progress"), real-time task completion rate (the ratio between the number of tasks completed in the current shift and the total number of tasks), and overdue task warning (e.g., "3 tasks remaining time <30 minutes"); instantaneous failure alarm (e.g., "temperature of belt conveyor rises 5°C") and real-time capacity (e.g., "current hour processing capacity 80 tons") can be analyzed from real-time data stream of IoT sensors and digital twin platform (i.e., real-time device status data); current safety violation times (e.g., "2 incidents of not wearing safety helmet today") and environmental real-time risk (e.g., "concentrated medium bucket value approaching limit") can be analyzed from instant detection results of edge computing devices (e.g., smart helmets, cameras, etc.) (i.e., safety detection data); instantaneous coal quality data (e.g., "current clean coal ash 8.5%") and bucket position dynamic prediction data (e.g., "expected full storage in 5 minutes") can be analyzed from real-time pushed coal quality index test results, sensor detection results, etc.

[0073] In the report generation phase, based on historical data and real-time analysis, production daily reports, safety reports, etc. can be automatically generated, supporting key indicators (e.g., task completion rate, high-frequency problem statistics) to be displayed by shift / day / week dimension. The generated report content can include data visualization charts, root cause analysis and optimization suggestions, etc., supporting natural language query and custom template export. The data visualization chart generation method can be: using ECharts library to automatically generate heat maps (representing device status distribution), line charts (representing task response trend), etc.

[0074] In the report generation phase, report template engine (e.g., Jinja2 template engine) can be used to dynamically generate reports, supporting custom combination of fields (e.g., "device failure rate", "task completion rate", etc.) to generate report content filling into report templates. The fields that need to be configured by the report template engine mainly include core indicator fields, analysis type fields, etc., which can be obtained from the database or real-time analysis results.

[0075] The core indicator field mainly includes production indicator data, equipment state data, task management data, safety detection data, and coal quality data. The production indicator data can include planned production, actual production, completion rate, etc., and the data is derived from production execution records, etc. The equipment state data can include failure rate, MTBF (Mean Time Between Failure), score, etc., and the data is derived from device sensor logs, fault history table, etc. The task management data can include total number of tasks, overdue rate, average processing time, etc., and the data is derived from tasks and workflows monitored by the task monitoring and feedback module 106, etc. The safety detection data can include violation times, bucket violation times, rectification rate, etc., and the data is derived from video analysis, environmental sensor detection, etc. The coal quality data can include ash deviation value, gangue rate, washing efficiency, etc., and the data is derived from laboratory data, production prediction results, etc.

[0076] The analysis type field mainly includes root cause analysis results (i.e., root cause analysis results), optimization suggestions, and trend comparison data. The root cause analysis results can include high-frequency fault reasons and associated equipment, which are clustered from historical data by a random forest model. The optimization suggestions can include recommended maintenance periods and parameter adjustment schemes, which can be obtained based on rule-based reasoning of a large language model. The trend comparison data can include monthly rankings and week-on-week changes, which can be calculated based on time series data aggregation.

[0077] In addition to configuring the fields, the report template engine also needs to define template parameters to adapt to different scenarios. The template parameters mainly include report parameters, visualization parameters, and output control parameters.

[0078] The report parameters mainly include template configuration options for time dimensions (such as daily reports, weekly reports, and monthly reports), data aggregation options for shift statistics (such as shift reports), report types (such as production reports, safety reports, and mechanical and electrical reports), chapter order customization, overview, problem analysis, and suggestion configuration parameters.

[0079] The visualization parameters mainly include chart types (such as bar charts and line charts), specified field display forms, and rendering parameters. For example, the rendering parameters can include color configuration parameters, with red representing failure, green representing safety, and yellow representing warning. The colors can be automatically rendered according to the set threshold (such as marking relevant task information red when the failure rate is >10%).

[0080] The output control parameters mainly include report export format and permission filtering configuration parameters. The report export format can be PDF, Excel, HTML, etc., which determines the file export format of the generated report. When performing permission filtering configuration operations, the report can be set to display only department data, and the visible fields of the report for different roles can be configured to limit the visible fields of the report by role.

[0081] The technical implementation process of dynamically generating a report using a report template engine mainly includes data preparation, template matching, dynamic rendering, and field language query in sequence.

[0082] Taking the use of a Jinja2 engine as an example, the detailed process of dynamically generating a report can be as follows: real-time data is aggregated through Flink stream processing, and historical data is obtained from a time series database; a corresponding Jinja2 template is loaded according to the report type selected by a user (such as a “daily safety report”); fields are bound to parameters (such as fault_rate=12.5 and time_dimension=“daily report”); the Jinja2 engine combines text, charts (through ECharts), and tables to generate HTML / PDF; the user is supported to edit the Jinja2 template again (such as adjusting the order of chapters); the user’s question content (such as “which devices had the most faults last month?”) is converted into a database query language and mapped to a template field to generate a field language, and then the relevant query result is obtained by querying related data through the field language.

[0083] For ease of understanding, the working process of the intelligent production scheduling system is exemplarily described below by taking a specific application as an example. Referring to FIG. 1, Figure 3 The working process of the intelligent production scheduling system mainly can include the following steps: Step 1, information fusion.

[0084] Data generation: data is generated from multiple different data sources such as industrial Internet of Things (IoT) devices, digital twin models, and AI large models. The generated data is derived from device data (such as sensor vibration logs, temperature logs, and PLC control signals), production data (such as coal quality test results and yield records), safety data (such as video analysis), and text data (such as work order records and inspection reports).

[0085] Data aggregation: the data generated from different data sources is integrated, and data preprocessing is performed (such as removing noise, filling missing values, standardizing heterogeneous data, and aligning timestamps).

[0086] Analysis and mining: the preprocessed multi-source data is subjected to entity extraction, relationship establishment, large model prediction of anomalies, root cause analysis, and the like to realize analysis and mining. For example, in entity extraction, entities in the form of “device A_vibration value_anomaly” are extracted from numerical data, and entities in the form of “fault type: bearing wear” and “handler: Zhang San” are extracted from text data. For another example, in relationship establishment, a rule engine is used to establish a relationship between “vibration value greater than threshold” and “possible fault: bearing wear”, and historical statistical analysis is used to obtain a relationship between temperature and fault, such as “80% of bearing wear is accompanied by temperature > 80℃”.

[0087] When performing large model prediction of anomalies, the LSTM model pre-trained can be used to predict anomalies for time series data. The LSTM model can predict future states of equipment (such as trends in vibration values, etc.), types of abnormal events, probabilities of failure, and ranges of failure impact, etc. For example, the large model can predict a safety violation event based on the video recognition result "person not wearing safety helmet" in the multi-modal fusion data and the detected equipment state "running". The large model can also predict a process abnormal event based on the coal quality ash deviation and washing parameters. For another example, the large model can perform root cause analysis based on historical work order text data and generate a natural language description.

[0088] Large model generation task: The large model can generate executable tasks based on the abnormal prediction results and real-time data. For example, the large model can automatically calculate the corresponding content (such as treatment measures, task levels, and failure causes) and fill in the corresponding task attributes.

[0089] When performing information fusion, edge computing can also be used to implement local processing of related information, reducing dependence on the cloud.

[0090] Step 2, information analysis and accurate push, distribute tasks to the correct performers.

[0091] Background configuration: Virtual organization mapping and workflow presetting can be performed through background configuration. When performing virtual organization mapping, skill group tags can be configured through the background, and personnel / equipment can be tagged (such as "mechanical and electrical maintenance group" and "inspection robot"). When performing workflow presetting, the order, conditions, and rules of task execution in the workflow can be defined through the background, and the scene type, virtual department, task level, current state, etc. of the workflow can be configured.

[0092] Generate task list: Configure other task attributes (such as scene type, associated department, etc.) for the task, and generate the task list after configuration.

[0093] Match workflow and performer for task: Match the workflow for the tasks in the task list according to the scene type, task level, and associated department, and match the performer for the task according to the real-time state of the personnel, so as to distribute the task to the correct personnel for execution. For example, the task generated for fault A needs to be performed by an electrician and a bench worker, and the current idle degree of the electrician is 70%, so the idle designated electrician can be matched as the performer of the task. Manual adjustment of the allocation of the performer can also be supported, such as the dispatcher dragging the task to a specific personnel.

[0094] Task distribution: tasks can be pushed to corresponding terminals (such as mobile terminals, large screens, etc.) based on pre-configured push logic (such as broadcast push and targeted push), realizing the matching of task organization interaction modes so that relevant personnel can perform tasks. The push mode of task distribution can be broadcast push or targeted push based on post skill group, and the specific push mode can be selected flexibly according to the actual situation of the task. Manual adjustment of the priority or execution path of the task can also be supported.

[0095] Step 3, real-time monitoring of task status.

[0096] Multi-terminal collaboration: WebSocket protocol synchronizes task status in real time among multiple terminals and inputs, outputs and displays relevant information. For example, detailed operation steps are displayed on the PC terminal, and fault points are marked by AR and safety specifications are announced by voice on the smart helmet terminal.

[0097] Task update: when the task status changes, the corresponding information of the task (such as associated departments, fault reasons, assigned executors, etc.) is updated so that the updated task can be distributed to the correct personnel for execution.

[0098] Step 4, report generation.

[0099] When the terminal has completed all tasks in the task list, a report will be generated based on the historical process of the task. For example, daily reports are generated for tasks completed on the same day, and monthly reports are generated by summarizing the task data completed in a month and the daily reports of each day. For example, a daily report can be generated at 8 pm every day (the generation time of the daily report can be changed and configured according to the scheduling requirements of different factories), and a monthly report can be generated at 8 am on the last day of each month. When the report is completed, the server will push the report to the large screen for display.

[0100] The generated report can be used to generate a report list, and the generated report can be used as the basic data for data summarization and analysis in the above information fusion data step to ensure the real-time nature of the task-related information.

[0101] The report generation method can be a template-based report generation method based on a pre-set report template, or a custom report generation method based on a custom field combination, to meet the needs of different management levels.

[0102] In summary, the working process of the above-mentioned intelligent production scheduling system can mainly include the following aspects: 1) Data collection: collect data from IoT devices, digital twin platforms, AI large models and other different data sources to build a multi-modal data pool.

[0103] 2) Intelligent analysis: large model predicts anomalies, generates task instances, and triggers workflow matching.

[0104] 3) Task distribution: dynamically assigned to posts / robots, multi-end task state synchronization.

[0105] 4) Execution monitoring: real-time feedback on task execution progress, automatic reminders for overdue tasks, and task closure.

[0106] 5) Report generation: data aggregation, root cause analysis results, and visual report output.

[0107] The above-mentioned intelligent production scheduling system improves the following technical effects through information fusion, task generation, distribution, flow, monitoring, and report automation: By multi-modal perception and using DTW algorithm for time alignment and Apriori algorithm for association rule mining in multi-modal data fusion, the problem of heterogeneous data integration is solved, "prevention" is achieved, and the accuracy of dangerous behavior recognition is improved; through automatic task generation and dynamic workflow matching, optimal allocation of tasks-resources is achieved, task allocation time is shortened, abnormal processing efficiency is improved, device idle rate is reduced, and material backlog is reduced; through automatic root cause analysis, problem positioning time is shortened; by configuring the report engine combined with Jinja2 template and random forest root cause analysis results for report automation, multi-dimensional data visualization and root cause tracing are supported, and the timeliness of report generation is improved; through real-time task monitoring based on WebSocket protocol and edge computing, and automatic reminders for overdue tasks, low-latency and high-precision safety warnings are ensured.

[0108] In practical application, by using the above-mentioned intelligent production scheduling system, massive data generated in the production process (such as device running parameters, material flow, quality indicators, etc.) can be deeply mined and analyzed with the help of AI large models, intelligent interaction terminals, etc., providing accurate basis for decision-making; through industrial Internet of Things and digital twinning, device interconnection is realized, device state information is collected in real time, and managers have a clear understanding of the running status of all plant devices; the large model of the coal preparation plant can perform intelligent analysis and reasoning based on real-time data, providing scientific basis for scheduling decisions and reasonable overall scheduling; through terminals such as mobile phones, intercoms, voice broadcasts, smart bracelets, smart helmets, and regional large screens, multi-end collaboration is realized, relevant information such as coal quality, mechanical and electrical, safety, and production can be pushed in time, and system warnings, fault abnormalities, safety violations, and other problems and cause analysis, treatment suggestions are pushed to relevant intelligent agents and specific personnel, assisting the scheduling and management departments to speed up task flow, improving production management efficiency, and changing the production control mode from people-centered to intelligent command + intelligent decision-making-centered.

[0109] Based on the above-mentioned intelligent production scheduling system for the coal industry scenario, the embodiment of the present invention also provides an intelligent production scheduling method for the coal industry scenario. The intelligent production scheduling method can be applied to the above-mentioned server. Figure 4 As shown, the smart production scheduling method may include the following steps: Step S402: Fusion the production information from multiple different data sources, and build a fault prediction model based on the fused information; wherein the production information includes equipment status information, coal quality index information, and safety detection information.

[0110] Step S404: perform fault prediction on the real-time production information through the fault prediction model, and generate a task list based on the fault prediction result, the real-time production information and the historical production information. Then, match the corresponding workflow and the optimal executor for each task in the task list, and send each task matched with the workflow and the optimal executor to the PC connected to the server for execution; wherein, the workflow defines the order, conditions and rules for the execution of the corresponding tasks.

[0111] Step S406: monitor the execution progress of each task in real time, and determine whether there is an overdue task based on the execution progress of each task. If there is an overdue task, urge the completion of each overdue task.

[0112] Step S408: Based on the fused information, the fault prediction results, the real-time production information, the historical production information and the execution process information of each task, a corresponding report is generated for each completed task, and the obtained report is sent to a display device connected to the server for display.

[0113] By adopting the above-mentioned intelligent production scheduling method for the coal industry scenario, a fault prediction model can be constructed using multi-source production information to realize automatic fault prediction. Tasks can be automatically generated by combining real-time production information and historical production information, and workflows and optimal executors can be matched for the tasks. The tasks can then be sent to the PC for execution, thereby realizing automated production scheduling and improving the efficiency and accuracy of production scheduling in the coal industry scenario. Overdue tasks can be urged to proceed by real-time monitoring of the progress of task execution, which is conducive to ensuring the real-time nature of production scheduling. Reports can also be generated by combining relevant information in the production scheduling and task execution process and sent to display devices for display, thereby realizing automatic generation of visual reports, which is conducive to statistics and viewing of the entire process.

[0114] The fusion can include first preprocessing, first feature extraction and analysis mining of the production information; based on this, the step S402 (i.e. fusing the production information from multiple different data sources, and constructing a fault prediction model based on the fused information) can include: first preprocessing the production information, and first feature extraction and analysis mining of the first preprocessed information, and then constructing the fault prediction model based on the obtained first features and analysis mining results; wherein the analysis mining results include the association rules of device faults and coal quality fluctuations.

[0115] The real-time production information can include structured data and unstructured data, and the fault prediction result can include the fault probability of the device corresponding to the real-time production information; based on this, the fault prediction of the real-time production information by the fault prediction model in the step S404 can include: second preprocessing the structured data and the unstructured data, and second feature extraction of the second preprocessed information, then inputting the extracted second features into the fault prediction model for fault prediction, and outputting the fault probability of the device corresponding to the real-time production information by the fault prediction model.

[0116] Before generating the task list based on the fault prediction result, the real-time production information and the historical production information in the step S404, the intelligent production scheduling method in the coal industry scenario can further include: if there is a target device whose fault probability is greater than a preset threshold, generating a task corresponding to the target device based on the fault probability of the target device and the real-time production information and the historical production information corresponding to the target device.

[0117] The real-time production information can include downtime information and yield loss information of the target device, and the historical production information can include historical work orders of the target device; based on this, the generation of the task list based on the fault prediction result, the real-time production information and the historical production information in the step S404 can include: third preprocessing the historical work orders, and extracting the fault type, fault description and solution of each target device from the third preprocessed information, then assigning a corresponding weight to each obtained solution by using a preset attention mechanism, and determining the target solution with the maximum weight of each target device; determining the task level of the task corresponding to each target device based on the downtime information and yield loss information of each target device, a preset first weight information and a preset first mapping relationship; configuring the scene type and the associated department of the task corresponding to each target device.

[0118] After the third preprocessing of the historical work order, the intelligent production scheduling method for the coal industry scene can further include: performing root cause analysis on the third preprocessed information, and generating a failure cause of each target device based on the root cause analysis result; wherein the failure cause is described in natural language.

[0119] The generation of the task list based on the failure prediction result, the real-time production information and the historical production information in the step S404 can further include: generating a corresponding task for each target device based on the target solution of each target device and the task level, the scene type, the associated department and the failure cause of the task corresponding to each target device; wherein each task contains the corresponding target solution, task level, scene type, associated department and failure cause.

[0120] The matching of the corresponding workflow and the optimal performer for each task in the task list in the step S404 can include: matching the task level, scene type and associated department of each task with a plurality of workflows in a preset workflow set to match the workflow corresponding to each task; matching the corresponding optimal performer for each task based on the real-time post state of the associated department; wherein the real-time post state contains the working state of each executable task personnel corresponding to the associated department.

[0121] Each task has corresponding time information, and the time information includes the occurrence time, the reporting time, the planned completion time and the actual completion time of the corresponding task; based on this, the determination of whether there is an overdue task based on the execution progress of each task in the step S406 can include: for each task, if the execution progress of the task is not updated after reaching a first target time point, the task is determined as an overdue task; wherein the time interval between the first target time point and the planned completion time of the corresponding task is less than a preset time length.

[0122] The urging of each overdue task in the step S406 can include: for each overdue task, if the execution progress of the overdue task is not updated after reaching a second target time point, an urging identifier is added to the overdue task, and the overdue task with the urging identifier is sent to the display device for display.

[0123] The step S408 of generating the report for each task based on the fused information, the fault prediction result, the real-time production information, the historical production information and the execution process information of each task can include: generating field information for each task based on the fused information, the fault prediction result, the real-time production information, the historical production information and the execution process information of each task; and matching a target report template for each task from a preset report template set by using a preset report template engine, and rendering the field information corresponding to the same task and the target report template to generate the report.

[0124] The task level can be divided into a first level representing a high-risk task, a second level representing a medium-risk task and a third level representing a low-risk task, and the server is further connected with a mobile terminal; based on this, the intelligent production scheduling method for the coal industry scene can further include: for each task, if the task level of the task is the first level, the task is pushed to all mobile terminals and all display devices of the target area, and if the task level of the task is the second level or the third level, the task is pushed to the mobile terminal corresponding to the associated department of the task.

[0125] The intelligent production scheduling method for the coal industry scene can further include: monitoring the state of each task in real time, and synchronizing the state of each task to the PC terminal, the mobile terminal and the display device.

[0126] The intelligent production scheduling method for the coal industry scene can further include: determining the task priority score of the task corresponding to each target device based on the urgency information, the influence range information and the resource consumption information of each target device and the preset second weight information; for each task, if the task priority score of the task is greater than a preset score threshold, the task is pushed to a predetermined target mobile terminal and / or target display device.

[0127] The intelligent production scheduling method for the coal industry scene provided by the embodiment of the application has the same implementation principle, technical effects and the intelligent production scheduling system for the coal industry scene as described above. For brevity, the part of the intelligent production scheduling method for the coal industry scene not mentioned in the embodiment can refer to the corresponding content in the intelligent production scheduling system for the coal industry scene.

[0128] Unless otherwise specifically stated, the relative steps, numerical expressions and numerical values of the components and steps set forth in these embodiments do not limit the scope of the application.

[0129] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a nonvolatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0130] In the description of the present application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second", "third" are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0131] Finally, it should be noted that: the above-described embodiments are only specific embodiments of the present application, which are used to illustrate the technical solutions of the present application, and are not limited thereto, the protection scope of the present application is not limited thereto, although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art within the technical range disclosed by the present application can modify or easily think of changes to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to part of the technical features; and these modifications, changes or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A smart production scheduling system for coal industry scenarios, characterized by: Applied to the server, including: An information fusion module is used to fuse production information from multiple different data sources and build a fault prediction model based on the fused information; wherein the production information includes equipment status information, coal quality index information and safety detection information; A task generation and distribution module is configured to perform fault prediction on real-time production information using the fault prediction model, generate a task list based on the fault prediction results, the real-time production information, and historical production information, match a corresponding workflow and optimal executor for each task in the task list, and send each task with a matching workflow and optimal executor to a PC connected to the server for execution; wherein the workflow defines the order, conditions, and rules for executing the corresponding tasks; The task monitoring and feedback module is used to monitor the execution progress of each task in real time and determine whether there are overdue tasks based on the execution progress of each task. If there are overdue tasks, each overdue task will be urged to be completed; The automatic report generation module is used to generate a corresponding report for each completed task based on the fused information, the fault prediction results, the real-time production information, the historical production information and the execution process information of each task, and send the obtained report to the display device connected to the server for display.

2. The intelligent production scheduling system for coal industry scenarios according to claim 1 is characterized in that: The fusion includes a first preprocessing, a first feature extraction and an analysis and mining of the production information; the information fusion module is also used to: perform a first preprocessing on the production information, and perform a first feature extraction and an analysis and mining on the first preprocessed information, and then construct the fault prediction model based on the obtained first feature and the analysis and mining results; wherein the analysis and mining results include association rules between equipment failures and coal quality fluctuations.

3. The intelligent production scheduling system for coal industry scenarios according to claim 2 is characterized in that: The real-time production information includes structured data and unstructured data, and the fault prediction result includes the failure probability of the equipment corresponding to the real-time production information; the task generation and distribution module is also used to: perform a second preprocessing on the structured data and the unstructured data, and perform a second feature extraction on the second preprocessed information, and then input the extracted second feature into the fault prediction model for fault prediction, and obtain the failure probability of the equipment corresponding to the real-time production information through the output of the fault prediction model; if there is a target equipment whose failure probability is greater than a preset threshold, the task corresponding to the target equipment is generated based on the failure probability of the target equipment and the real-time production information and historical production information corresponding to the target equipment.

4. The intelligent production scheduling system for coal industry scenarios according to claim 3 is characterized in that: The real-time production information includes downtime information and output loss information of the target equipment, and the historical production information includes historical work orders of the target equipment. The task generation and distribution module is further configured to: perform a third preprocessing on the historical work orders, extract the fault type, fault description, and solution of each target equipment from the third preprocessed information, then use a preset attention mechanism to assign a corresponding weight to each obtained solution, and determine the target solution with the largest weight for each target equipment; Determining a task level of a task corresponding to each target device based on downtime information and output loss information of each target device, preset first weight information, and a preset first mapping relationship; Configure the scenario type and associated department of each target device's corresponding task; The information fusion module is further configured to: perform a root cause analysis on the third pre-processed information, and generate a fault cause of each target device based on the root cause analysis result; wherein the fault cause is described in natural language; The task generation and distribution module is also used to generate corresponding tasks for each target device based on the target solution of each target device and the task level, scenario type, associated department and fault cause of the task corresponding to each target device; wherein each task contains the corresponding target solution, task level, scenario type, associated department and fault cause.

5. The intelligent production scheduling system for coal industry scenarios according to claim 4 is characterized in that: The task generation and distribution module is also used to: match the task level, scenario type and associated department of each task with multiple workflows in a preset workflow set to match the workflow corresponding to each task; based on the real-time job status of the associated department of each task, match the corresponding optimal executor for each task; wherein the real-time job status includes the work status of each task-executable person corresponding to the corresponding associated department.

6. The intelligent production scheduling system for coal industry scenarios according to claim 1 is characterized in that: Each task has corresponding time information, including the occurrence time, reporting time, planned completion time, and actual completion time of the corresponding task. The task monitoring and feedback module is further configured to: for each task, if the execution progress of the task is not updated after reaching a first target time point, determine the task as an overdue task; wherein the time interval between the first target time point and the planned completion time of the corresponding task is less than a preset time length; The task monitoring and feedback module is also used to: for each overdue task, if the execution progress of the overdue task is not updated after reaching the second target time point, add a reminder mark to the overdue task, and send the overdue task with the reminder mark to the display device for display.

7. The intelligent production scheduling system for coal industry scenarios according to claim 6 is characterized in that: The automatic report generation module is also used to: generate corresponding field information for each completed task based on the fused information, the fault prediction results, the real-time production information, the historical production information and the execution process information of each task; use a preset report template engine to match the corresponding target report template for each completed task from the preset report template set, and render the field information and target report template corresponding to the same task to generate a report.

8. The intelligent production scheduling system for coal industry scenarios according to claim 5 is characterized in that: The task levels are divided into a first level representing high-risk tasks, a second level representing medium-risk tasks, and a third level representing low-risk tasks. The server is also connected to a mobile terminal. The task generation and distribution module is further used to: for each task, if the task level of the task is the first level, push the task to all mobile terminals and all display devices in the target area; if the task level of the task is the second level or the third level, push the task to the mobile terminal corresponding to the department associated with the task; The task monitoring and feedback module is further used to monitor the status of each task in real time and synchronize the status of each task to the PC, the mobile terminal and the display device.

9. The intelligent production scheduling system for coal industry scenarios according to claim 8 is characterized in that: The task generation and distribution module is further configured to: determine a task priority score for the task corresponding to each target device based on the urgency information, impact range information, and resource consumption information of each target device and preset second weight information; For each task, if the task priority score of the task is greater than a preset score threshold, the task is pushed to a predetermined target mobile terminal and / or target display device.

10. A smart production scheduling method for coal industry scenarios, characterized in that: The server according to any one of claims 1 to 9 comprises: Fusion of production information from multiple data sources, including equipment status information, coal quality index information, and safety detection information, and construction of a fault prediction model based on the fused information. Fault prediction is performed on real-time production information using the fault prediction model, and a task list is generated based on the fault prediction result, the real-time production information, and historical production information. A corresponding workflow and optimal executor are then matched for each task in the task list, and each task matched with a workflow and optimal executor is sent to a PC connected to the server for execution; wherein the workflow defines the order, conditions, and rules for the execution of the corresponding tasks; Monitor the progress of each task in real time and determine whether there are overdue tasks based on the progress of each task. If there are overdue tasks, urge them to be completed. Based on the fused information, the fault prediction results, the real-time production information, the historical production information and the execution process information of each task, a corresponding report is generated for each completed task, and the obtained report is sent to a display device connected to the server for display.

Images