Intelligent chemical manufacturing visual scheduling system based on big data technology

Abstract

The invention discloses an intelligent chemical manufacturing visual scheduling system based on a big data technology, and the system comprises a data collection layer, relates to the technical field of chemical manufacturing, employs an analytic hierarchy process to combine with an entropy weight method, and takes equipment function importance, fault consequence severity, maintenance cost and difficulty, and rare and substitutability as criteria. The chemical production equipment is divided into key equipment and general equipment. The fault occurrence probability threshold value of the differentiated equipment fault prediction model is formulated for different levels of equipment, when the fault occurrence probability of key equipment or general equipment is larger than or equal to the corresponding threshold value, a scheduling adjustment mechanism is triggered, the estimated shutdown time length is obtained through calculation, the estimated shutdown time length is brought into the production scheduling model, and the production scheduling is completed. And the starting time, the processing sequence and the resource allocation scheme of each work order on each chemical production device are regenerated, so that the influence of the device failure on the production progress is reduced to the greatest extent on the premise of meeting the process requirements, and the efficient operation of the production line is maintained.

Description

Technical Field

[0001] The present invention relates to the field of chemical manufacturing technology, and in particular to an intelligent chemical manufacturing visualization scheduling system based on big data technology. Background Art

[0002] Intelligent chemistry is the product of the deep integration of the chemical industry and modern information technology. It is a chemical production model that uses advanced intelligent technology to improve the efficiency and safety of chemical production, management, and operations.

[0003] Traditional chemical production scheduling relies heavily on manual experience, and managers rely on subjective judgment to arrange work orders, allocate resources, and respond to emergencies such as equipment failures.

[0004] Publication No. CN113050567B discloses a dynamic scheduling method for an intelligent manufacturing system. An incremental extreme learning machine based on a decaying regularization term is used to construct a regression scheduling model for the intelligent manufacturing system. A decaying regularization term is introduced to realize adaptive acquisition of a regularization coefficient.

[0005] Publication No. CN111679637B discloses a flexible multi-task scheduling method in a manufacturing system, which takes the time limit and quality limit of the task as constraints, minimizes the total energy consumption and balances the workload as the goal, and establishes a dual-objective optimization model. The service allocation vector is first iteratively processed in stages, and then the subtask sequence vector is generated and iteratively processed. Finally, the service allocation vector and the subtask sequence vector are processed. The service allocation and subtask sequence are considered simultaneously in the search space that has been narrowed by the first two stages, so as to balance the total energy consumption and workload balance under the constraints of quality and completion time.

[0006] However, the above applications still have the following problems: CN113050567B focuses on the technical improvement of the model construction itself, and the method of CN111679637B focuses on general manufacturing goals such as energy consumption and workload balance. Both are responses to equipment failures in chemical production, and do not go deep into the risks of shutdowns caused by equipment failures in chemical production. It is impossible to ensure that the impact of equipment failures on production progress is minimized to the greatest extent while meeting process requirements, and to maintain efficient operation of the production line. Summary of the invention

[0007] In order to solve the technical problems existing in the background technology, the present invention proposes an intelligent chemical manufacturing visualization scheduling system based on big data technology.

[0008] The present invention proposes an intelligent chemical manufacturing visualization scheduling system based on big data technology, comprising: Data acquisition layer: Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, obtain equipment operating status, switch quantity information and process parameter setting values, and transmit data to the data storage layer through wired and wireless communication; Data storage layer: including big data storage platform and data warehouse; Data processing and analysis layer: Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, build a time series production prediction model based on long short-term memory networks, and build an equipment failure prediction model; Production planning and scheduling module, the production planning and scheduling module includes: Order management unit: used to receive customer orders, analyze order requirements to generate preliminary work orders, and calculate the urgency of each work order; Scheduling optimization unit: The predicted equipment capacity data is obtained based on the time series output prediction model based on the long short-term memory network. According to the urgency of each work order, the material supply cycle and process switching time data are obtained based on big data analysis. The production scheduling model is generated using an intelligent algorithm to determine the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment; Application layer: including visual scheduling console and mobile applications.

[0009] Preferably, in the data acquisition layer, the sensors include temperature, pressure, flow, liquid level, vibration and component analysis sensors.

[0010] Preferably, in the data storage layer, the big data storage platform adopts the Hadoop distributed file system; The data warehouse cleans, transforms, and integrates the collected raw data based on Hive or Snowflake, and organizes the data by subject domain.

[0011] Preferably, in the order management unit, the urgency of each work order is calculated as follows: The urgency level of each work order is calculated by fuzzy comprehensive evaluation method, and the urgency levels include very urgent, urgent, general urgent, not too urgent, and not urgent.

[0012] Preferably, the factory layout diagram of the visual scheduling console of the application layer displays the equipment operation status in real time, and different statuses are marked with different colors; The production progress Gantt chart of the application layer shows the comparison between the planned and actual execution time of each work order in different processes.

[0013] Preferably, in the scheduling optimization unit, the intelligent algorithm adopted by the production scheduling model includes a genetic algorithm or a simulated annealing algorithm.

[0014] Preferably, in the scheduling optimization unit, the material supply cycle and process switching time data obtained based on big data analysis are: Use exponential smoothing method to analyze and predict material supply data to obtain material supply cycle; Based on the historical process switching big data, the process switching time is estimated using the regression analysis model.

[0015] Preferably, in the data processing and analysis layer, the data obtained by the data storage layer is analyzed: The hierarchical analysis method is adopted, with the division of key equipment and general equipment as the target layer, and the importance of equipment function, severity of failure consequences, maintenance cost and difficulty, rarity and substitutability as the criterion layer. The comprehensive score of each equipment is obtained by the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the threshold is considered key equipment, and equipment with a score lower than the threshold is considered general equipment.

[0016] Preferably, in the data processing and analysis layer, an equipment failure prediction model is constructed based on multivariate time series analysis and deep neural network algorithm. By analyzing the trend of equipment operating parameters and vibration spectrum characteristics, the equipment failure prediction model is used to predict the probability and time of equipment failure, providing a basis for preventive maintenance.

[0017] Preferably, in the equipment failure prediction model, a time threshold is set, and the failure probability thresholds of the three equipment failure prediction models are set to P1, P2 and P3 respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued and a basis is provided for preventive maintenance; When the equipment failure prediction model predicts that the probability of failure of a key device within a time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of failure of a general device within a time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered.

[0018] Preferably, the scheduling adjustment mechanism is as follows: By collecting historical maintenance data; Historical maintenance data includes fault type, maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time; For a certain fault type, let the downtime be T, the arrival time of the maintenance personnel be t1, the preparation time of maintenance tools and spare parts be t2, and the maintenance time be t3, then T=t1+t2+t3; In the historical maintenance data obtained, multiple sets of maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time data for a certain fault type; Calculate the normal distribution of the maintenance personnel's arrival time t1, is the average arrival time, is the standard deviation, and the time range of the maintenance personnel's arrival time t1 is ( ); Calculate the uniform distribution of the maintenance tool and spare parts preparation time t2, a and b are the lower and upper limits of the preparation time, and assume that the time range of the maintenance tool and spare parts preparation time t2 is (a, b); Calculate the average maintenance time of maintenance time t3, assuming that the average maintenance time of maintenance time t3 is m hours; In the time range of t1 ( ), use the random number generator to generate random numbers as the sampling value of t1, and repeat multiple times; In the time range of t2 (a, b), a random number generator is used to generate a random number as the sampling value of t2, and this is repeated multiple times; t3 = m; Then, substitute t1, t2, and t3 obtained from each sampling into the downtime formula T=t1+t2+t3, and then calculate the average value of T as the estimated downtime of a certain fault type; After obtaining the estimated downtime of a certain fault type, the estimated downtime is brought into the production scheduling model to regenerate the optimal start time, processing sequence and resource allocation plan for each work order in each chemical production equipment.

[0019] An intelligent chemical manufacturing visualization scheduling method based on big data technology includes the following steps: S1. Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, and obtain equipment operation status, switch quantity information and process parameter setting values, and transmit the data to the data storage layer through wired and wireless communication, and the data storage layer performs data storage; S2. Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, and build a time series production prediction model based on long short-term memory networks to predict equipment production capacity data. The model is trained based on historical production data and related factors. S3. Build an equipment failure prediction model based on multivariate time series analysis and deep neural network algorithm. By analyzing the trend of equipment operating parameters and vibration spectrum characteristics, predict the probability and time of equipment failure, and provide a basis for preventive maintenance. S4. Adopting the hierarchical analysis method, with the target layer being key equipment and general equipment, and the criterion layer being the importance of equipment functions, the severity of failure consequences, the cost and difficulty of repair, and the rarity and substitutability, the comprehensive score of each equipment is obtained through the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the score threshold is key equipment, and equipment with a score lower than the score threshold is general equipment; S5. Receive customer orders, analyze order requirements and generate preliminary work orders; use fuzzy comprehensive evaluation method to calculate the urgency level of each work order, which includes very urgent, urgent, general urgent, not too urgent, and not urgent. When using fuzzy comprehensive evaluation method to calculate, the order delivery deadline, customer importance, and process dependency are comprehensively considered; S6. Based on the predicted equipment capacity data and the urgency level of each work order obtained by the time series production prediction model based on the long short-term memory network, the material supply data is analyzed and predicted using the exponential smoothing method to obtain the material supply cycle; based on the historical process switching big data, the process switching time is estimated using the regression analysis model, and the production scheduling model is generated using the genetic algorithm or simulated annealing algorithm to determine the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment; S7, setting a time threshold, and the failure probability thresholds of the three equipment failure prediction models are P1, P2 and P3 respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued to provide a basis for preventive maintenance; When the equipment failure prediction model predicts that the probability of failure of a key device within a time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of failure of a general device within a time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered.

[0020] In the present invention, the proposed intelligent chemical manufacturing visualization scheduling system based on big data technology has the following beneficial technical effects: 1. Adopting the hierarchical analysis method combined with the entropy weight method, chemical production equipment is divided into key equipment and general equipment based on the importance of equipment functions, severity of failure consequences, maintenance costs and difficulty, rarity and substitutability. Formulate the failure probability threshold of the differentiated equipment failure prediction model for different levels of equipment. When the failure probability of key equipment or general equipment is greater than or equal to the corresponding threshold, the scheduling adjustment mechanism is triggered, and the estimated downtime is calculated. The estimated downtime is brought into the production scheduling model, and the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment are regenerated to ensure that the impact of equipment failure on production progress is minimized to the greatest extent possible while meeting the process requirements, and the efficient operation of the production line is maintained.

[0021] 2. Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, build a time series production prediction model based on long short-term memory networks, deeply mine historical data patterns, and accurately predict equipment production capacity. Compared with traditional prediction methods, it can more keenly capture production trends, plan production in advance, and avoid insufficient or excessive production capacity. At the same time, based on multivariate time series analysis and deep neural network algorithms, build an equipment failure prediction model. By analyzing the trends of equipment operating parameters and vibration spectrum characteristics, it can accurately predict the probability and time of failure, far exceeding the early warning capabilities of traditional equipment maintenance methods, winning valuable time for preventive maintenance, reducing equipment failure rates, and reducing production stoppage losses caused by sudden failures.

[0022] 3. After receiving the customer order, the order management unit quickly analyzes the demand to generate a preliminary work order, and uses the fuzzy comprehensive evaluation method to comprehensively consider factors such as order delivery deadlines, customer importance, and process dependencies, accurately calculate the urgency level of each work order, and provide key decision-making basis for scheduling. The scheduling optimization unit generates a production scheduling model based on production forecast data, work order urgency, material supply cycle obtained by exponential smoothing method, and process switching time obtained based on historical data regression analysis, using intelligent algorithms such as genetic algorithms or simulated annealing algorithms. Compared with traditional empirical scheduling, this intelligent scheduling can significantly optimize the start time, processing sequence, and resource allocation plan of work orders in various chemical production equipment, effectively improve equipment utilization, shorten order delivery cycle, and improve customer satisfaction.

[0023] 4. The visual scheduling console and mobile applications at the application layer create a convenient and efficient management and control platform for managers. The factory layout diagram uses different colors to mark the equipment operation status in real time, and the production progress Gantt chart intuitively presents the comparison of the work order execution progress, allowing managers to know the production site at any time and anywhere and quickly find problems. When the equipment failure prediction model issues an early warning or triggers the scheduling adjustment mechanism, the downtime of the faulty equipment is estimated based on historical maintenance data, and timely feedback is given to the production scheduling model for re-optimization, ensuring that production can be adjusted agilely in the face of emergencies such as equipment failures, maintaining stable operation, and minimizing losses.

[0024] 5. The big data storage platform uses the Hadoop distributed file system, which can easily cope with the storage needs of massive chemical data and ensure high data availability; the data warehouse cleans, converts, integrates and organizes the original data by subject domain based on Hive or Snowflake, making the data clear and easy to query and call, effectively shortening the data processing cycle.

[0025] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a principle block diagram of the system of the present invention; Figure 2 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0027] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0028] like Figure 1 An intelligent chemical manufacturing visualization scheduling system based on big data technology is shown, comprising: Data acquisition layer: Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, obtain equipment operating status, switch quantity information and process parameter setting values, and transmit data to the data storage layer through wired and wireless communication; In the data acquisition layer, sensors include temperature, pressure, flow, liquid level, vibration and component analysis sensors.

[0029] Data storage layer: including big data storage platform and data warehouse; In the data storage layer, the big data storage platform uses the Hadoop distributed file system; the Hadoop distributed file system is a distributed file system.

[0030] The data warehouse cleans, transforms, and integrates the collected raw data based on Hive or Snowflake, and organizes the data by subject domain; Hive is the data warehouse infrastructure, which provides data storage, query, and analysis functions, and Snowflake is a cloud-based data warehouse platform that provides efficient data storage, processing, and analysis functions. Subject domain is a high-level classification method for data in the data warehouse.

[0031] The big data storage platform uses the Hadoop distributed file system, which can easily cope with the storage needs of massive chemical data and ensure high data availability; the data warehouse cleans, converts, integrates and organizes the original data by subject domain based on Hive or Snowflake, making the data clear and easy to query and call, effectively shortening the data processing cycle.

[0032] Data processing and analysis layer: Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, build a time series production prediction model based on long short-term memory networks, and build an equipment failure prediction model; Spark is a fast and general distributed computing framework for large-scale data processing; Flink is an open source distributed stream processing and batch processing framework; In an optional embodiment, in the data processing and analysis layer, the data obtained by the data storage layer is analyzed: The hierarchical analysis method is adopted, with the division of key equipment and general equipment as the target layer, and the importance of equipment function, severity of failure consequences, maintenance cost and difficulty, rarity and substitutability as the criterion layer. The comprehensive score of each equipment is obtained by the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the threshold is considered key equipment, and equipment with a score lower than the threshold is considered general equipment.

[0033] In an optional embodiment, in the data processing and analysis layer, an equipment failure prediction model is constructed based on multivariate time series analysis and deep neural network algorithm. By analyzing the trend of equipment operating parameters and vibration spectrum characteristics, the equipment failure prediction model is used to predict the probability and time of equipment failure, providing a basis for preventive maintenance; Through multivariate time series analysis and deep neural network algorithms, the probability and time of chemical equipment failure can be predicted more accurately, and maintenance measures can be taken in advance to ensure the safe and stable operation of chemical production.

[0034] Using Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, we build a time series production prediction model based on long short-term memory networks, deeply mine historical data patterns, and accurately predict equipment capacity. Compared with traditional prediction methods, it can more keenly capture production trends, plan production in advance, and avoid insufficient or excessive capacity. At the same time, we build an equipment failure prediction model based on multivariate time series analysis and deep neural network algorithms. By analyzing the trends of equipment operating parameters and vibration spectrum characteristics, we can accurately predict the probability and time of failure, which far exceeds the early warning capabilities of traditional equipment maintenance methods, wins valuable time for preventive maintenance, reduces equipment failure rates, and reduces production stoppage losses caused by sudden failures.

[0035] In an optional embodiment, in the equipment failure prediction model, a time threshold is set, and the failure probability thresholds of the three equipment failure prediction models are set to P1, P2 and P3 respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued and a basis is provided for preventive maintenance; When the equipment failure prediction model predicts that the probability of a key device failing within a time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of a general device failing within a time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered; In an optional embodiment, 50%<P1<P2, the value range of P1 is [70%-80%], the value range of P2 is [85%-90%], and the value range of P3 is [50%-60%].

[0036] The scheduling adjustment mechanism is as follows: By collecting historical maintenance data; Historical maintenance data includes fault type, maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time; For a certain fault type, let the downtime be T, the arrival time of the maintenance personnel be t1, the preparation time of maintenance tools and spare parts be t2, and the maintenance time be t3, then T=t1+t2+t3; In the historical maintenance data obtained, multiple sets of maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time data for a certain fault type; Calculate the normal distribution of the maintenance personnel's arrival time t1, is the average arrival time, is the standard deviation, and the time range of the maintenance personnel's arrival time t1 is ( ); Calculate the uniform distribution of the maintenance tool and spare parts preparation time t2, a and b are the lower and upper limits of the preparation time, and assume that the time range of the maintenance tool and spare parts preparation time t2 is (a, b); Calculate the average maintenance time of maintenance time t3, assuming that the average maintenance time of maintenance time t3 is m hours; In the time range of t1 ( ), use the random number generator to generate random numbers as the sampling value of t1, and repeat multiple times; In the time range of t2 (a, b), a random number generator is used to generate a random number as the sampling value of t2, and this is repeated multiple times; t3 = m; Then, substitute t1, t2, and t3 obtained from each sampling into the downtime formula T=t1+t2+t3, and then calculate the average value of T as the estimated downtime of a certain fault type; After obtaining the estimated downtime of a certain fault type, the estimated downtime is brought into the production scheduling model to regenerate the optimal start time, processing sequence and resource allocation plan for each work order in each chemical production equipment.

[0037] Ensure that while meeting process requirements, the impact of equipment failure on production progress is minimized and the efficient operation of the production line is maintained.

[0038] The analytic hierarchy process combined with the entropy weight method is used to divide chemical production equipment into key equipment and general equipment based on the importance of equipment functions, severity of failure consequences, maintenance costs and difficulty, rarity and substitutability. Differentiated equipment failure prediction models are formulated with failure probability thresholds for different levels of equipment. When the failure probability of key equipment or general equipment is greater than or equal to the corresponding threshold, the scheduling adjustment mechanism is triggered, and the estimated downtime is calculated. The estimated downtime is brought into the production scheduling model, and the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment are regenerated to ensure that the impact of equipment failure on production progress is minimized to the greatest extent possible while meeting process requirements, and to maintain efficient operation of the production line.

[0039] Production planning and scheduling module, the production planning and scheduling module includes: Order management unit: used to receive customer orders, analyze order requirements to generate preliminary work orders, and calculate the urgency of each work order; In an optional embodiment, in the order management unit, the urgency of each work order is calculated as follows: The urgency level of each work order is calculated by fuzzy comprehensive evaluation method, and the urgency levels include very urgent, urgent, general urgent, not too urgent, and not urgent.

[0040] The urgency level of each work order is calculated through the fuzzy comprehensive evaluation method. The urgency of each work order on the current production line can be determined by the order delivery deadline, customer importance, and process dependency. The order delivery deadline is measured by the number of days, hours, or other time units remaining to the delivery date. Customer importance is classified and assigned based on the customer's past order volume, degree of cooperation, and credit rating. The process dependency is determined based on the completion of the work order's predecessor process and the urgency of the subsequent process to the overall production process.

[0041] Scheduling optimization unit: The predicted equipment capacity data is obtained based on the time series output prediction model based on the long short-term memory network. According to the urgency of each work order, the material supply cycle and process switching time data are obtained based on big data analysis. The production scheduling model is generated using an intelligent algorithm to determine the optimal start time, processing sequence and resource allocation plan for each work order in each chemical production equipment; In an optional embodiment, in the scheduling optimization unit, the intelligent algorithm adopted by the production scheduling model includes a genetic algorithm or a simulated annealing algorithm.

[0042] In an optional embodiment, in the scheduling optimization unit, the material supply cycle and process switching time data obtained based on big data analysis are: Use exponential smoothing method to analyze and predict material supply data to obtain material supply cycle; Based on historical process switching big data, the process switching time is estimated using a regression analysis model; After receiving the customer order, the order management unit quickly analyzes the demand to generate a preliminary work order, and uses the fuzzy comprehensive evaluation method to comprehensively consider factors such as order delivery deadlines, customer importance, and process dependencies, accurately calculate the urgency level of each work order, and provide key decision-making basis for scheduling. The scheduling optimization unit generates a production scheduling model based on production forecast data, work order urgency, material supply cycle obtained by exponential smoothing method, and process switching time obtained based on historical data regression analysis, using intelligent algorithms such as genetic algorithms or simulated annealing algorithms. Compared with traditional empirical scheduling, this intelligent scheduling can significantly optimize the start time, processing sequence, and resource allocation plan of work orders in various chemical production equipment, effectively improve equipment utilization, shorten order delivery cycle, and improve customer satisfaction.

[0043] Application layer: including visual scheduling console, which provides intuitive and interactive visual interface of factory layout diagram, production progress Gantt chart and key indicator dashboard in the form of web application; It also includes mobile applications to facilitate mobile office and on-site scheduling for on-site managers and operators.

[0044] In an optional embodiment, the factory layout diagram of the visual scheduling console of the application layer displays the equipment operation status in real time, and different statuses are marked with different colors.

[0045] The different states include normal, fault and standby.

[0046] The visual scheduling console and mobile applications at the application layer create a convenient and efficient management and control platform for managers. The factory layout diagram uses different colors to identify the equipment operation status in real time, and the production progress Gantt chart intuitively presents the comparison of the work order execution progress, allowing managers to know the production site at any time and anywhere and quickly find problems. When the equipment failure prediction model issues an early warning or triggers the scheduling adjustment mechanism, the downtime of the faulty equipment is estimated based on historical maintenance data, and timely feedback is given to the production scheduling model for re-optimization, ensuring that production can be adjusted agilely in the face of emergencies such as equipment failures, maintaining stable operation, and minimizing losses.

[0047] In an optional embodiment, the production progress Gantt chart of the application layer presents the comparison between the planned and actual execution time of each work order in different processes.

[0048] like Figure 2The method for visual scheduling of intelligent chemical manufacturing based on big data technology includes the following steps: S1. Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, and obtain equipment operation status, switch quantity information and process parameter setting values, and transmit the data to the data storage layer through wired and wireless communication, and the data storage layer performs data storage; S2. Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, and build a time series production prediction model based on long short-term memory networks to predict equipment production capacity data. The model is trained based on historical production data and related factors. S3. Build an equipment failure prediction model based on multivariate time series analysis and deep neural network algorithm. By analyzing the trend of equipment operating parameters and vibration spectrum characteristics, predict the probability and time of equipment failure, and provide a basis for preventive maintenance. S4. Adopting the hierarchical analysis method, with the target layer being key equipment and general equipment, and the criterion layer being the importance of equipment functions, the severity of failure consequences, the cost and difficulty of repair, and the rarity and substitutability, the comprehensive score of each equipment is obtained through the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the score threshold is key equipment, and equipment with a score lower than the score threshold is general equipment; S5. Receive customer orders, analyze order requirements and generate preliminary work orders; use fuzzy comprehensive evaluation method to calculate the urgency level of each work order, which includes very urgent, urgent, general urgent, not too urgent, and not urgent. When using fuzzy comprehensive evaluation method to calculate, the order delivery deadline, customer importance, and process dependency are comprehensively considered; S6. Based on the predicted equipment capacity data and the urgency level of each work order obtained by the time series production prediction model based on the long short-term memory network, the material supply data is analyzed and predicted using the exponential smoothing method to obtain the material supply cycle; based on the historical process switching big data, the process switching time is estimated using the regression analysis model, and the production scheduling model is generated using the genetic algorithm or simulated annealing algorithm to determine the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment; S7, setting a time threshold, and the failure probability thresholds of the three equipment failure prediction models are P1, P2 and P3 respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued to provide a basis for preventive maintenance; When the equipment failure prediction model predicts that the probability of failure of a key device within a time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of failure of a general device within a time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered.

[0049] An intelligent chemical manufacturing visualization scheduling method based on big data technology also includes the following steps: In S1, the sensor may also include a gas concentration sensor for monitoring the concentration of harmful gases in the chemical production environment to ensure production safety.

[0050] In S1, the data storage layer also has a data redundancy backup function, which regularly performs redundant backup of stored data to prevent data loss.

[0051] In S2, when training the time series yield prediction model based on the long short-term memory network, seasonal factors and market demand fluctuation factors are also introduced as auxiliary variables to improve the accuracy of yield prediction.

[0052] In S5, the fuzzy comprehensive evaluation method can also combine the product profit contribution rate when calculating the urgency level of each work order, and give a higher urgency weight to the work order with a high profit contribution rate.

[0053] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0054] In the embodiments provided by the present invention, it should be understood that the disclosed system or method can be implemented in other ways. For example, the above-described embodiments of the invention are only illustrative, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation.

[0055] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0056] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0057] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic features of the present invention.

[0058] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent chemical manufacturing visualization scheduling system based on big data technology, characterized in that: include: Data acquisition layer: Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, obtain equipment operating status, switch quantity information and process parameter setting values, and transmit data to the data storage layer through wired and wireless communication; Data storage layer: including big data storage platform and data warehouse; Data processing and analysis layer: Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, build a time series production prediction model based on long short-term memory networks, and build an equipment failure prediction model; Production planning and scheduling module, the production planning and scheduling module includes: Order management unit: used to receive customer orders, analyze order requirements to generate preliminary work orders, and calculate the urgency of each work order; Scheduling optimization unit: The predicted equipment capacity data is obtained based on the time series output prediction model based on the long short-term memory network. According to the urgency of each work order, the material supply cycle and process switching time data are obtained based on big data analysis. The production scheduling model is generated using an intelligent algorithm to determine the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment; Application layer: including visual scheduling console and mobile applications.

2. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 1 is characterized in that: In the data storage layer, the big data storage platform uses the Hadoop distributed file system; The data warehouse cleans, transforms, and integrates the collected raw data based on Hive or Snowflake, and organizes the data by subject domain.

3. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 1 is characterized in that: In the order management unit, the urgency of each work order is calculated as follows: The urgency level of each work order is calculated by fuzzy comprehensive evaluation method, and the urgency levels include very urgent, urgent, general urgent, not too urgent, and not urgent.

4. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 1 is characterized in that: In the scheduling optimization unit, the intelligent algorithms used in the production scheduling model include genetic algorithms or simulated annealing algorithms.

5. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 4 is characterized in that: In the scheduling optimization unit, the material supply cycle and process switching time data obtained based on big data analysis are as follows: Use exponential smoothing method to analyze and predict material supply data to obtain material supply cycle; Based on the historical process switching big data, the process switching time is estimated using the regression analysis model.

6. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 1 is characterized in that: In the data processing and analysis layer, the data obtained from the data storage layer is analyzed: The hierarchical analysis method is adopted, with the division of key equipment and general equipment as the target layer, and the importance of equipment function, severity of failure consequences, maintenance cost and difficulty, rarity and substitutability as the criterion layer. The comprehensive score of each equipment is obtained by the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the threshold is considered key equipment, and equipment with a score lower than the threshold is considered general equipment.

7. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 6 is characterized in that: In the data processing and analysis layer, an equipment failure prediction model is built based on multivariate time series analysis and deep neural network algorithm. By analyzing the trends of equipment operating parameters and vibration spectrum characteristics, the equipment failure prediction model is used to predict the probability and time of equipment failure, providing a basis for preventive maintenance.

8. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 7 is characterized in that: In the equipment failure prediction model, a time threshold is set, and the failure probability thresholds of the three equipment failure prediction models are set to P1, P2, and P3, respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued and a basis is provided for preventive maintenance; When the equipment failure prediction model predicts that the probability of failure of a key device within a time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of failure of a general device within a time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered.

9. The intelligent chemical manufacturing visualization scheduling system based on big data technology according to claim 8 is characterized in that: The scheduling adjustment mechanism is as follows: By collecting historical maintenance data; Historical maintenance data includes fault type, maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time; For a certain fault type, let the downtime be T, the arrival time of the maintenance personnel be t1, the preparation time of maintenance tools and spare parts be t2, and the maintenance time be t3, then T=t1+t2+t3; In the historical maintenance data obtained, multiple sets of maintenance personnel arrival time, maintenance tool and spare parts preparation time, and maintenance time data for a certain fault type; Calculate the normal distribution of the maintenance personnel's arrival time t1, is the average arrival time, is the standard deviation, and the time range of the maintenance personnel's arrival time t1 is ( ); Calculate the uniform distribution of the maintenance tool and spare parts preparation time t2, a and b are the lower and upper limits of the preparation time, and assume that the time range of the maintenance tool and spare parts preparation time t2 is (a, b); Calculate the average maintenance time of maintenance time t3, assuming that the average maintenance time of maintenance time t3 is m hours; In the time range of t1 ( ), use the random number generator to generate random numbers as the sampling value of t1, and repeat multiple times; In the time range of t2 (a, b), a random number generator is used to generate a random number as the sampling value of t2, and this is repeated multiple times; t3 = m; Then, substitute t1, t2, and t3 obtained from each sampling into the downtime formula T=t1+t2+t3, and then calculate the average value of T as the estimated downtime of a certain fault type; After obtaining the estimated downtime of a certain fault type, the estimated downtime is brought into the production scheduling model to regenerate the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment.

10. The intelligent chemical manufacturing visualization scheduling method based on big data technology according to any one of claims 1 to 9 is characterized in that: The following steps are involved: S1. Install sensors on chemical production equipment, pipelines and storage tanks to collect physical and chemical parameters, and obtain equipment operation status, switch quantity information and process parameter setting values, and transmit the data to the data storage layer through wired and wireless communication, and the data storage layer performs data storage; S2. Use Spark and Flink distributed computing frameworks to process real-time streaming data and batch data, and build a time series production prediction model based on long short-term memory networks to predict equipment production capacity data. The model is trained based on historical production data and related factors. S3. Build an equipment failure prediction model based on multivariate time series analysis and deep neural network algorithm. By analyzing the trend of equipment operating parameters and vibration spectrum characteristics, predict the probability and time of equipment failure, and provide a basis for preventive maintenance. S4. Adopting the hierarchical analysis method, with the target layer being key equipment and general equipment, and the criterion layer being the importance of equipment functions, the severity of failure consequences, the cost and difficulty of repair, and the rarity and substitutability, the comprehensive score of each equipment is obtained through the entropy weight method, and a score threshold is set. Equipment with a score higher than or equal to the score threshold is key equipment, and equipment with a score lower than the score threshold is general equipment; S5. Receive customer orders, analyze order requirements and generate preliminary work orders; The fuzzy comprehensive evaluation method is used to calculate the urgency level of each work order, which includes very urgent, urgent, moderately urgent, not too urgent, and not urgent; S6. Based on the predicted equipment capacity data and the urgency level of each work order obtained by the time series production prediction model based on the long short-term memory network, the material supply data is analyzed and predicted using the exponential smoothing method to obtain the material supply cycle; based on the historical process switching big data, the process switching time is estimated using the regression analysis model, and the production scheduling model is generated using the genetic algorithm or simulated annealing algorithm to determine the start time, processing sequence and resource allocation plan of each work order in each chemical production equipment; S7, setting a time threshold, and the failure probability thresholds of the three equipment failure prediction models are P1, P2 and P3 respectively, and 0<P3<P1<P2; When the equipment failure prediction model predicts that the probability of failure of a key device or a general device within the time threshold is greater than or equal to P3, an early warning is issued to provide a basis for preventive maintenance; When the equipment failure prediction model predicts that the probability of failure of a key device within the time threshold is greater than or equal to P1, or when the equipment failure prediction model predicts that the probability of failure of a general device within the time threshold is greater than or equal to P2, the scheduling adjustment mechanism is triggered.

Images