A real-time fault warning method and device for a drug production line

Abstract

The present invention discloses a real-time fault warning method and device for a drug production line, which relates to the field of production line monitoring. The method includes: analyzing each production node in the drug production line to obtain multiple waste treatment nodes; obtaining waste characteristics corresponding to each node; deploying multiple sensor modules; performing data sensor monitoring to obtain multiple waste monitoring data sets; performing abnormal analysis to obtain multiple abnormal indicators corresponding to multiple waste treatment nodes; performing waste reaction impact analysis to output multiple reaction impact indicators; fitting according to multiple reaction impact indicators to construct a reaction impact relationship network, and performing fault warning on the drug production line. The method solves the technical problems of the existing real-time fault warning of drug production lines, that is, the fault warning is not timely, and it is difficult to ensure the stable operation of the production line and the product quality, and achieves the technical effect of timely discovering and warning of potential faults and effectively reducing the risks in the production process.

Description

Technical Field

[0001] The present application relates to the field of production line monitoring, and in particular to a real-time fault warning method and device for a drug production line. Background Art

[0002] In the modern pharmaceutical industry, the efficiency, stability and safety of drug production lines are key factors in ensuring the quality and output of drugs. With the development of automation and intelligent technology, drug production lines are becoming increasingly complex, involving multiple production links and a large number of production equipment. The operating status of these equipment directly affects the efficiency and product quality of the entire production line. Therefore, real-time monitoring and fault warning of the production line are particularly important. In the process of drug production, various wastes will be generated, including but not limited to residual raw materials, packaging materials, by-products in the production process, and waste liquids and waste gases that may contain harmful substances. If these wastes are not handled properly, they will not only pollute the environment, but also have adverse effects on the subsequent drug production process, such as contaminating raw materials, affecting production equipment, etc., thereby threatening the quality and safety of the final drug. At present, pharmaceutical companies generally use regular inspections, manual records, and monitoring methods based on fixed sensors to manage waste treatment links. Due to the limitations of monitoring methods, these methods cannot capture abnormal conditions in the waste treatment process in a timely manner. When a waste treatment node fails, it often has an adverse impact on other links of the production line, resulting in a lag in fault warning.

[0003] In the current relevant technologies, the real-time fault warning of drug production lines has the technical problem that the fault warning is not timely, making it difficult to ensure the stable operation of the production line and the product quality. Summary of the invention

[0004] The present application provides a real-time fault warning method and device for a drug production line, analyzes and determines waste treatment nodes, identifies key monitoring parameters according to waste types and treatment requirements, installs multiple types of sensors on waste treatment nodes, obtains real-time data through sensor modules, and performs abnormal analysis to identify potential faults, evaluates the impact of abnormalities on other parts of the production line, constructs a reaction influence relationship network, and performs fault warning and location based on the reaction influence relationship network. The present application realizes real-time and comprehensive monitoring of waste treatment nodes in drug production lines, achieves the technical effect of timely discovering and warning of potential faults, and effectively reducing risks in the production process.

[0005] The present application provides a real-time fault warning method for a drug production line, comprising:

[0006] Analyze each production node in the drug production line to obtain multiple waste treatment nodes; obtain waste characteristics corresponding to each node in the multiple waste treatment nodes; deploy multiple sensor modules according to the waste characteristics corresponding to each node; perform data sensing monitoring according to the multiple sensor modules to obtain multiple waste monitoring data sets; perform abnormality analysis on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes; perform waste reaction impact analysis according to the multiple abnormal indicators and output multiple reaction impact indicators; perform fitting according to the multiple reaction impact indicators to construct a reaction impact relationship network, and call the reaction impact relationship network to provide fault warning for the drug production line.

[0007] In a possible implementation, waste reaction impact analysis is performed according to the multiple abnormal indicators, multiple reaction impact indicators are output, and the following processing is performed:

[0008] Acquire the positions of the multiple waste treatment nodes; acquire multiple production nodes connected to the multiple waste treatment nodes based on the positions of the multiple waste treatment nodes, wherein the multiple waste treatment nodes correspond one-to-one to the multiple production nodes; output multiple connection components by identifying connection components between the multiple production nodes and the multiple waste treatment nodes; perform waste reaction impact analysis on the multiple connection components according to the multiple abnormal indicators, and output multiple reaction impact indicators.

[0009] In a possible implementation, waste reaction impact analysis is performed on the multiple connection components according to the multiple abnormal indicators, and the following processing is performed:

[0010] Determine the component material, connection air tightness, connection layer relationship and connection pipeline length of each connecting component; simulate each abnormal indicator based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connecting component as a feature vector, predict the reaction influence indicator corresponding to each connecting component, and output multiple reaction influence indicators corresponding to the multiple abnormal indicators.

[0011] In a possible implementation, the locations of the multiple waste processing nodes are obtained, and the following processing is performed:

[0012] Based on the positions of the multiple waste treatment nodes, determine whether the multiple waste treatment nodes include multi-branch waste treatment nodes, wherein the multi-branch waste treatment node refers to a waste treatment node composed of N branches, and each branch is connected to a different production node, and N is a positive integer greater than or equal to 2; if the multiple waste treatment nodes do not include multi-branch waste treatment nodes, a single-branch waste reaction impact analysis is performed according to the abnormal indicators of each waste treatment node.

[0013] In a possible implementation, the following processing is performed:

[0014] If the multiple waste treatment nodes include multi-branch waste treatment nodes, obtain abnormal indicators of the multi-branch waste treatment nodes; perform waste reaction impact analysis on N branches respectively according to the abnormal indicators of the multi-branch waste treatment nodes, and output N reaction impact indicators; when any reaction impact indicator among the N reaction impact indicators is greater than a preset impact indicator, obtain a warning signal.

[0015] In a possible implementation, anomaly analysis is performed on the multiple waste monitoring data sets, multiple anomaly indicators corresponding to the multiple waste processing nodes are obtained, and the following processing is performed:

[0016] Set historical normal waste monitoring samples and historical local abnormal waste monitoring samples for the multiple waste treatment nodes; define an initialization abnormality analysis model according to the abnormal layout factor algorithm; train the initialization abnormality analysis model according to the historical normal waste monitoring samples, the historical local abnormal waste monitoring samples and identification information of the abnormality degree to obtain an abnormality analysis model; call the abnormality analysis model to perform abnormality analysis on the multiple waste monitoring data sets, and obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes respectively.

[0017] In a possible implementation, each abnormal indicator is simulated based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, and the following processing is performed:

[0018] A resistance network layer is established; a protection evaluation is performed using the component material, connection air tightness, connection layer relationship, and connection pipeline length of each connection component as feature vectors, and the protection index of the connection component is output; the protection index of each connection component and each abnormal index are predicted using the resistance network layer, and the reaction influence index corresponding to each connection component is output.

[0019] The present application also provides a real-time fault warning device for a drug production line, comprising:

[0020] a waste treatment node acquisition module, the waste treatment node acquisition module is used to analyze each production node in the drug production line and obtain multiple waste treatment nodes; a node waste feature acquisition module, the node waste feature acquisition module is used to obtain the waste features corresponding to each of the multiple waste treatment nodes; a sensor layout module, the sensor layout module is used to layout multiple sensor modules according to the waste features corresponding to each node; a data sensor monitoring module, the data sensor monitoring module is used to perform data sensor monitoring according to the multiple sensor modules and obtain multiple waste monitoring data sets; an abnormality analysis module, the abnormality analysis module is used to perform abnormality analysis on the multiple waste monitoring data sets and obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes; a waste reaction impact analysis module, the waste reaction impact analysis module is used to perform waste reaction impact analysis according to the multiple abnormal indicators and output multiple reaction impact indicators; a fault warning module, the fault warning module is used to fit according to the multiple reaction impact indicators, build a reaction impact relationship network, and call the reaction impact relationship network to perform fault warning on the drug production line.

[0021] A real-time fault warning method and device for a drug production line proposed in the present application first analyzes each production node in the drug production line to obtain multiple waste treatment nodes, then obtains the waste characteristics corresponding to each of the multiple waste treatment nodes, and then arranges multiple sensor modules according to the waste characteristics corresponding to each node, and then performs data sensing monitoring based on the multiple sensor modules to obtain multiple waste monitoring data sets, and then performs abnormal analysis on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes, and then performs waste reaction impact analysis based on the multiple abnormal indicators, outputs multiple reaction impact indicators, and finally fits according to the multiple reaction impact indicators to construct a reaction impact relationship network, and calls the reaction impact relationship network to perform fault warning for the drug production line, thereby achieving the technical effect of timely discovering and warning of potential faults and effectively reducing the risks in the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the device according to the embodiment of the present application. It should be understood that the previous or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0023] Figure 1A flow chart of a real-time fault warning method for a drug production line provided in an embodiment of the present application.

[0024] Figure 2 A schematic structural diagram of a real-time fault warning device for a drug production line provided in an embodiment of the present application.

[0025] Explanation of the reference numerals: waste treatment node acquisition module 10, node waste feature acquisition module 20, sensor deployment module 30, data sensor monitoring module 40, abnormality analysis module 50, waste reaction impact analysis module 60, fault warning module 70. DETAILED DESCRIPTION

[0026] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0027] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0028] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions. For example, a process, method, device, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0029] The present application embodiment provides a real-time fault warning method for a drug production line, such as Figure 1 As shown, the method includes:

[0030] Step S100, analyze each production node in the drug production line to obtain multiple waste treatment nodes. Specifically, collect basic information such as the overall layout diagram, equipment list, and production flow chart of the drug production line, conduct on-site inspection of the drug production line, and determine the actual location, function, and mutual relationship of each production node. According to the production process and function, each link in the drug production line is divided into different production nodes, wherein each production node has a specific function or task, such as raw material feeding, reactor, filtration, drying, packaging, etc. On the basis of classification, further identify nodes directly related to waste treatment, which will generate waste or need to treat waste during the production process, and perform functional analysis on each identified node to determine whether it has a waste treatment function or whether it has a direct impact on waste treatment. Nodes responsible for waste collection, storage, treatment, and discharge are selected as waste treatment nodes, such as waste collection stations, sewage treatment stations, and waste gas treatment devices. These nodes are crucial to ensuring the cleanliness of the production environment, resource utilization efficiency, and environmental compliance. Compile a list of identified waste treatment nodes, including key information such as node name, location, treatment type (such as solid waste, liquid waste, gaseous waste, etc.), and treatment capacity.

[0031] Step S200, obtaining waste characteristics corresponding to each of the multiple waste processing nodes. Specifically, collecting detailed information about each waste processing node, including waste type, composition, generation amount, treatment method, etc., formulating classification standards based on the properties (such as physical properties, chemical properties, biological properties), hazards, etc. of the waste, and identifying and classifying the waste generated by each waste processing node based on the classification standards.

[0032] For each type of waste, identify its key characteristics, such as physical characteristics (density, viscosity, particle size, color, odor, etc.), chemical characteristics (pH value, toxicity, flammability, reactivity, etc.), biological characteristics (biodegradability, pathogen content, etc.), etc. Establish a special database or data table for each waste treatment node to store the characteristic data of the waste, enter the historical data or the waste characteristic data obtained by field measurement into the corresponding database, conduct regular sampling and testing of the waste, and update its characteristic data.

[0033] Step S300, multiple sensor modules are arranged according to the waste characteristics corresponding to each node. Specifically, according to the waste characteristics of the waste treatment node (such as chemical properties, physical state, temperature, pressure, etc.), the parameters that need to be monitored are determined, and based on the results of the analysis, a suitable sensor module is selected. The sensor module is a device that can sense and convert specific physical quantities (such as temperature, pressure, chemical composition, etc.) into measurable signals for real-time monitoring of the state of the waste. For example, for waste treatment nodes that need to monitor temperature, a temperature sensor is selected; for those that need to monitor chemical composition, a corresponding chemical sensor is selected. According to the specific requirements of the waste treatment node and the performance of the sensor module, the parameters of the sensor module, such as measurement range, accuracy, sampling frequency, etc., are set, and the sensor module is interfaced with the data acquisition system or control system to ensure that the data can be transmitted accurately and in real time. Conduct on-site surveys of waste treatment nodes to determine the best locations for sensor modules, which can accurately reflect the status of the waste and are easy to install and maintain. Fix the sensor modules in the corresponding positions according to the best locations and installation specifications. After installation, debug the sensor modules and verify their accuracy and reliability by comparing the data output by the sensor modules with the actual waste status.

[0034] Step S400, data sensing monitoring is performed according to the multiple sensor modules to obtain multiple waste monitoring data sets. Specifically, the sensor module is started to collect various data in the waste treatment process in real time, such as temperature, pressure, chemical content, flow rate, etc. The sensor module converts the collected raw data into digital signals, and transmits the converted data to the central control system or data processing center by wired or wireless means. A special database is established in the data processing center to store the data received from each sensor module, remove noise, outliers or duplicates in the raw data, ensure the accuracy and consistency of the data, integrate the pre-processed data from different sensor modules, and store them in the corresponding database table according to time sequence and node classification to form a complete waste monitoring data set.

[0035] Step S500, performing anomaly analysis on the multiple waste monitoring data sets, and obtaining multiple anomaly indicators corresponding to the multiple waste processing nodes. Specifically, according to the characteristics of the waste processing nodes and the requirements of the drug production line, key monitoring indicators that are crucial to fault warning are identified, and at the same time, representative features are extracted from the original data by means of statistical methods, machine learning algorithms, etc. According to the data characteristics and business needs, an anomaly detection algorithm is selected, such as a statistical method (such as Z-score, IQR, etc.), a distance-based method (such as K-means, DBSCAN, etc.) or a density-based method (such as LOF, Isolation Forest, etc.). The preprocessed data and the anomaly detection algorithm are applied to the waste monitoring data set, and the algorithm outputs the anomaly score or label of each waste processing node, and identifies the anomaly point or abnormal time period. Specific anomaly indicators are extracted from the anomaly detection results, such as the time when the anomaly occurs, the degree of the anomaly (such as the anomaly score), and the type of the anomaly (such as temperature anomaly, pressure anomaly, etc.), and the anomaly indicators of each waste processing node are summarized and sorted to form multiple anomaly indicator sets.

[0036] In a possible implementation, the plurality of waste monitoring data sets are subjected to abnormal analysis, and a plurality of abnormal indicators corresponding to the plurality of waste processing nodes are obtained. Step S500 further includes step S510, setting historical normal waste monitoring samples and historical local abnormal waste monitoring samples for the plurality of waste processing nodes. Specifically, from the previous production records, the monitoring data of each waste processing node under normal operation is collected as the historical normal waste monitoring samples. Similarly, from the production records, the monitoring data recorded as abnormal or suspected abnormal due to some reason (such as equipment failure, operating error, raw material abnormality, etc.) is screened out as the historical local abnormal waste monitoring samples. The collected historical normal samples and historical local abnormal samples are sorted to ensure that each sample contains complete monitoring data and relevant timestamps, node identifiers and other information. Step S520, according to the abnormal layout factor algorithm, the initialization abnormal analysis model is defined. Specifically, according to the characteristics of the waste monitoring data and the needs of abnormal detection, the abnormal layout factor algorithm is selected as the basis of the abnormal analysis model, and according to the algorithm, the parameters and structure of the abnormal analysis model are initialized to prepare for subsequent training. Among them, the abnormal layout factor algorithm is an algorithm for anomaly detection, which identifies abnormal points or abnormal patterns by analyzing the distribution characteristics of data in space or time. Step S530, train the initialized abnormal analysis model according to the historical normal waste monitoring samples, the historical local abnormal waste monitoring samples and the identification information of the abnormal degree to obtain the abnormal analysis model. Specifically, the collected historical samples are preprocessed, including data cleaning, normalization, feature extraction and other steps to improve the efficiency and accuracy of model training. The preprocessed historical normal samples and local abnormal samples are input into the initialized abnormal analysis model, and the identification information of the abnormal degree (information used to identify the abnormal degree of each abnormal sample, such as abnormal score, abnormal type, etc., which plays a role of supervised learning in the model training process) is used to guide the training process of the model. By iteratively optimizing the parameters of the model, the model can accurately distinguish between normal data and abnormal data. Step S540, call the abnormal analysis model to perform abnormal analysis on the multiple waste monitoring data sets, and obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes. Specifically, multiple waste monitoring data sets acquired in real time are input into a trained anomaly analysis model. The model automatically performs anomaly detection based on the input monitoring data, identifies abnormal points or abnormal patterns, and converts the detected abnormal results into specific abnormal indicators, such as the time when the abnormality occurred, the degree of the abnormality, the type of abnormality, etc., and outputs them to subsequent processing procedures.This implementation method collects historical normal samples and local abnormal samples, and uses these samples to train the anomaly analysis model, so that the model can more accurately identify abnormal situations in waste treatment nodes, achieving the technical effect of improving the accuracy and efficiency of anomaly detection.

[0037] Step S600, performing waste reaction impact analysis according to the multiple abnormal indicators, and outputting multiple reaction impact indicators. Specifically, the reaction impact refers to the negative impact of the abnormality of the waste treatment node on the upstream and downstream nodes or system components of the drug production line. Specifically, the abnormal indicators corresponding to each waste treatment node are analyzed to determine the specific meaning and cause of each abnormal indicator. According to the process flow of the waste treatment node and the layout of the drug production line, the upstream and downstream production nodes or system components that may be affected by each abnormal indicator are identified, that is, the specific path of the abnormality of the waste treatment node affecting the upstream and downstream nodes or system components, wherein the abnormality of the waste treatment node will affect the upstream and downstream nodes through material flow, energy flow or information flow. Using domain knowledge, historical data and expert experience, each identified impact path is evaluated for the reaction impact, and the specific impact degree and impact type that the abnormality may cause to the upstream and downstream nodes are determined, and the impact type includes reduced production efficiency, product quality problems, equipment damage, etc. The evaluated reaction impact is converted into a quantifiable indicator, that is, a reaction impact indicator, such as an impact degree score, an impact range estimate, etc., which is used to describe the specific impact of the abnormality on the production line.

[0038] In a possible implementation, waste reaction impact analysis is performed according to the multiple abnormal indicators, and multiple reaction impact indicators are output. Step S600 further includes step S610, obtaining the positions of the multiple waste treatment nodes. Specifically, using tools such as production line layout diagrams or geographic information systems (GIS), determine the specific coordinates or relative positions of each waste treatment node (waste treatment equipment or area) in the drug production line. Step S620, using the positions of the multiple waste treatment nodes, obtain multiple production nodes connected to the multiple waste treatment nodes, wherein the multiple waste treatment nodes correspond one-to-one to the multiple production nodes. Specifically, according to the positions of the waste treatment nodes, trace their upstream and downstream relationships in the drug production line, use production flow charts and material flow diagrams and other materials to identify the production nodes directly connected to each waste treatment node (equipment, process or area responsible for specific production tasks in the drug production line), confirm the one-to-one correspondence between these production nodes and waste treatment nodes, and record them. Among them, the connection refers to the direct association between the production node and the waste treatment node due to material flow, information flow or energy flow. Step S630, by identifying the connection components between the multiple production nodes and the multiple waste treatment nodes, multiple connection components are output. Specifically, the connection parts between each production node and the waste treatment node are analyzed, key connection components such as pipelines, conveyor belts, valves, sensors, etc. are identified, and the types, locations, functions, etc. of these connection components are recorded. Among them, the connection components refer to physical components that connect the production node and the waste treatment node and undertake the functions of material transmission, energy transfer or information exchange. Step S640, according to the multiple abnormal indicators, the waste reaction impact analysis is performed on the multiple connection components, and multiple reaction impact indicators are output. Specifically, for the abnormal indicators of each waste treatment node, the specific impact that it may have on the connected production nodes and connection components is analyzed, and the damage or potential risk caused to the connection components is evaluated in combination with the nature (such as temperature abnormality, pressure abnormality, flow abnormality, etc.) and severity of the abnormal indicators. These impacts are quantified using risk assessment models or expert experience, and corresponding reaction impact indicators (such as risk level, impact range, loss degree, etc.) are output. This implementation method obtains the location of the waste treatment node and identifies the production nodes and connecting components connected to it, so as to more accurately evaluate the potential impact of abnormalities in the waste treatment node on the entire production line, achieving the technical effect of improving the accuracy of waste reaction impact analysis.

[0039] In a possible implementation, the waste reaction impact analysis is performed on the multiple connection components according to the multiple abnormal indicators, and step S640 further includes step S641, determining the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component. Specifically, collect and record the manufacturing material information (such as metal, plastic, rubber, etc.) of each connection component, because different materials have different characteristics such as tolerance and corrosion to waste; evaluate the sealing performance between the connection components, including the use of gaskets, sealants, etc., and the tightness of the connection joints to determine whether there is a risk of leakage; analyze the hierarchical structure between the connection components to determine the multi-layer structure that the waste passes through during the transmission process, such as pipes, valves, filters, etc.; measure and record the pipeline length between the connection components, because the pipeline length will affect the flow rate, pressure loss and deposition of the waste. Step S642, simulate each abnormal indicator based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, predict the reaction impact indicator corresponding to each connection component, and output multiple reaction impact indicators corresponding to the multiple abnormal indicators. Specifically, according to the detailed attributes of each connection component collected in step S641, combined with the physical and chemical properties of the waste (such as density, viscosity, corrosiveness, toxicity, etc.), a simulation model of the flow of waste in the connection component is constructed. Multiple abnormal indicators (such as temperature abnormality, pressure abnormality, flow abnormality, etc.) are used as input parameters of the simulation model to simulate the flow behavior of the waste under these abnormal conditions. The simulation model is run to simulate the flow of waste in the connection component, including flow velocity, flow direction, pressure distribution, temperature distribution, etc. According to the simulation results, the possible reaction effects of each connection component under the action of abnormal indicators are predicted, such as increased corrosion, increased leakage risk, pipeline blockage, etc. The predicted reaction effects are quantified into specific indicators (such as corrosion rate, leakage probability, blockage degree, etc.), and these indicators are output for subsequent analysis and decision-making. This implementation method provides accurate basic data for reaction simulation analysis by collecting the characteristics of each connection component, thereby achieving the technical effect of improving the accuracy of reaction impact analysis.

[0040] In a possible implementation, the positions of the multiple waste treatment nodes are obtained, and step S610 further includes step S611, where the positions of the multiple waste treatment nodes are used to determine whether the multiple waste treatment nodes include multi-branch waste treatment nodes, wherein the multi-branch waste treatment node refers to a waste treatment node composed of N branches, and each branch is connected to a different production node, and N is a positive integer greater than or equal to 2. Specifically, the location information of all waste treatment nodes is collected, including the physical location of the waste treatment nodes, the connection relationship between each other, etc. These nodes are traversed, and for each waste treatment node, it is checked whether it is connected to multiple different production nodes, that is, whether there are multiple branches. If a waste treatment node is connected to N (N≥2) different production nodes, and each branch is independently connected to a different production node, then the node is marked as a multi-branch waste treatment node, and all identified multi-branch waste treatment nodes are recorded. Step S612, if the multiple waste treatment nodes do not include multi-branch waste treatment nodes, perform single-branch waste reaction impact analysis based on the abnormal indicators of each waste treatment node. Specifically, if no multi-branch waste treatment node is found in step S611, that is, all waste treatment nodes are single-branch nodes, then single-branch waste reaction impact analysis is performed. That is, for each single-branch waste treatment node, according to the production nodes connected to it and its own abnormal indicators (such as processing efficiency, failure rate, etc.), analyze the possible reaction effects (such as reduced production efficiency, product quality problems, environmental pollution, etc.) of the node when treating waste on the entire production system, including evaluating the flow path of the waste, changes in processing efficiency, and the impact on subsequent production links. According to the analysis results, output the reaction impact assessment report of each single-branch waste treatment node. Since multi-branch nodes are connected to multiple different production nodes, their anomalies will have a wider and more complex impact on the system, so more detailed analysis is required; for single-branch waste treatment nodes, due to their relatively simple structure and limited impact range, a relatively simplified analysis method can be adopted. This implementation method can select the analysis method more specifically by first determining whether there are multi-branch waste treatment nodes, thereby achieving the technical effect of improving the targetedness and efficiency of the analysis.

[0041] In a possible implementation, step S610 further includes step S613, if the multiple waste treatment nodes include multi-branch waste treatment nodes, obtain the abnormal indicators of the multi-branch waste treatment nodes. Specifically, after identifying the multi-branch waste treatment nodes, collect the detailed abnormal indicator data of these nodes, including but not limited to the proportion of reduced processing efficiency, failure rate, pollutant emission exceeding the standard, etc. Step S614, according to the abnormal indicators of the multi-branch waste treatment nodes, respectively perform waste reaction impact analysis on N branches, and output N reaction impact indicators. Specifically, for each multi-branch waste treatment node, respectively perform impact analysis on the N branches connected to it, and analyze the flow of waste on each branch, processing efficiency, potential impact on subsequent production links, etc. According to the abnormal indicators of the branch and the overall performance of the waste treatment node, evaluate the waste reaction impact generated by each branch, and output the corresponding reaction impact indicators, which can be specific values ​​or ratings, used to quantify the degree of influence of the branch on the system. Step S615, when any of the N reaction impact indicators is greater than the preset impact indicator, obtain an early warning signal. Specifically, the N reaction impact indicators are compared with the preset impact indicator threshold. Among them, the preset impact indicator is the reaction impact indicator threshold set according to system requirements and actual conditions, which is used to determine whether it is necessary to issue an early warning signal. If the reaction impact indicator of any branch exceeds the preset threshold, an early warning signal is output immediately, and the early warning signal can be sent to relevant personnel through sound and light alarms, SMS notifications, email reminders, etc., so that timely countermeasures can be taken. This implementation method realizes the refined management of multi-branch waste treatment nodes by performing separate analysis and evaluation on each branch, which helps to timely discover and deal with potential problems, avoid the expansion of problems, and achieve the technical effect of refined management.

[0042] In a possible implementation, each abnormal index is simulated based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, and step S642 further includes step S6421, establishing a resistance network layer. Specifically, a resistance network model is selected, such as a deep neural network (DNN), a convolutional neural network (CNN) or a recurrent neural network (RNN), etc., the weights and biases of the resistance network layer are randomly initialized, and an activation function is selected for each neuron of the resistance network layer, such as ReLU (Rectified Linear Unit), Sigmoid or Tanh, etc., to increase the nonlinear processing capability of the network, and an optimizer (such as Adam, SGD, etc.) is configured for the resistance network layer to adjust the network weights to minimize the prediction error. Among them, the resistance network layer is a data layer in the neural network that is specifically used to process characteristics such as resistance and protection. Step S6422, the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component are used as feature vectors to perform a protection evaluation, and the protection index of the connection component is output. Specifically, the feature vectors of each connection component, such as component material, connection air tightness, connection layer relationship, and connection pipeline length, are standardized or normalized to ensure that features of different dimensions have the same weight in the model. Based on the preprocessed feature vectors, feature extraction methods (such as PCA, autoencoder, etc.) are used to further extract key information to simplify the model complexity and improve the prediction accuracy. The extracted features are input into the pre-trained evaluation model (which can be a linear model, a decision tree, or a machine learning model), and the protective index of each connection component is calculated. The protective index refers to a quantitative index that measures the strength of the protective role played by the connection component in the system, including the ability to resist external interference, maintain system stability, and prevent leakage. Step S6423, using the resistance network layer to predict the protective index of each connection component and each abnormal index, and output the reaction impact index corresponding to each connection component. Specifically, the protective index of each connection component is fused with the abnormal index (such as pressure abnormality, temperature abnormality, etc.) to form a complete input vector. The fused input vector is input into the resistance network layer, and the trained network parameters are used for forward propagation to calculate the reaction impact index corresponding to each connection component. According to the size and trend of the reaction impact index, the impact of each connection component on the overall performance of the system under abnormal conditions is analyzed to provide a basis for subsequent maintenance and optimization. This implementation method quantifies the protective role of the connection component in the system through protective evaluation, and uses the resistance network layer for prediction to accurately evaluate the reaction impact of each connection component on the overall performance of the system under abnormal conditions, thereby achieving the technical effect of providing more accurate guidance for system maintenance and optimization.

[0043] Step S700, according to the multiple reaction influence indicators, fit, build a reaction influence relationship network, call the reaction influence relationship network to perform fault warning on the drug production line. Specifically, the reaction influence relationship network is a model established based on graph theory or network analysis method, which is used to describe the reaction influence and mutual relationship of the abnormal waste treatment node on other parts of the drug production line. Specifically, multiple reaction influence indicators are integrated to ensure that all relevant waste treatment nodes and their abnormal states are taken into consideration, check the integrity and consistency of the data, and exclude errors or abnormal values. Each waste treatment node and its corresponding reaction influence indicator are used as nodes in the reaction influence relationship network, and connections are established between nodes according to the flow path of waste in the production line, the mutual dependence in the treatment process, and the potential causal relationship between abnormal indicators. According to historical data, expert knowledge and statistical analysis, weights are assigned to each connection to indicate the strength of the influence between different nodes. Using graph theory or network analysis methods, a reaction influence relationship network is constructed, which is used to intuitively display the potential impact of the abnormal waste treatment node on the entire production line. Based on historical data and simulation data, the reaction influence relationship network is trained and optimized to improve the accuracy and timeliness of fault warning. The abnormal indicators of the waste treatment nodes are monitored in real time and input into the trained reaction influence relationship network. Through the analysis and calculation of the relationship network, the potential risk points that may cause production line failures are predicted and identified. According to the prediction results, fault warning information is generated, including warning level, impact range, recommended measures, etc., and relevant personnel are notified in time for processing. The embodiment of the present application adopts the method of analyzing and determining the waste treatment nodes, identifying key monitoring parameters according to the waste type and treatment requirements, installing multiple types of sensors on the waste treatment nodes, obtaining real-time data through the sensor module, and performing abnormal analysis, identifying potential faults, evaluating the impact of abnormalities on other parts of the production line, constructing the reaction influence relationship network, and based on the reaction influence relationship network, performing fault warning and positioning and other technical means, realizing real-time and comprehensive monitoring of waste treatment nodes in the drug production line, achieving the technical effect of timely discovering and warning potential faults and effectively reducing the risks in the production process.

[0044] In the above, refer to Figure 1 A real-time fault warning method for a drug production line according to an embodiment of the present invention is described in detail. Figure 2 A real-time fault warning device for a drug production line according to an embodiment of the present invention is described.

[0045] A real-time fault warning device for a drug production line according to an embodiment of the present invention is used to solve the technical problem that the real-time fault warning of an existing drug production line is not timely, and it is difficult to ensure the stable operation of the production line and the quality of the product, so as to achieve the technical effect of timely discovering and warning potential faults and effectively reducing the risks in the production process. A real-time fault warning device for a drug production line includes: a waste treatment node acquisition module 10, a node waste feature acquisition module 20, a sensor layout module 30, a data sensor monitoring module 40, an abnormality analysis module 50, a waste reaction impact analysis module 60, and a fault warning module 70.

[0046] The waste treatment node acquisition module 10 is used to analyze each production node in the drug production line and obtain multiple waste treatment nodes; the node waste feature acquisition module 20 is used to obtain the waste features corresponding to each node in the multiple waste treatment nodes; the sensor layout module 30 is used to layout multiple sensor modules according to the waste features corresponding to each node; the data sensor monitoring module 40 is used to perform data sensor monitoring according to the multiple sensor modules and obtain multiple waste monitoring data sets; the abnormality analysis module 50 is used to perform abnormality analysis on the multiple waste monitoring data sets and obtain multiple abnormal indicators corresponding to the multiple waste treatment nodes respectively; the waste reaction impact analysis module 60 is used to perform waste reaction impact analysis according to the multiple abnormal indicators and output multiple reaction impact indicators; the fault warning module 70 is used to fit according to the multiple reaction impact indicators, construct a reaction impact relationship network, and call the reaction impact relationship network to perform fault warning on the drug production line.

[0047] The specific configuration of the waste reaction impact analysis module 60 will be described in detail below. As described above, the waste reaction impact analysis is performed according to the multiple abnormal indicators, and multiple reaction impact indicators are output. The waste reaction impact analysis module 60 may further include: a node position acquisition unit for acquiring the positions of the multiple waste treatment nodes; a production node acquisition unit for acquiring multiple production nodes connected to the multiple waste treatment nodes with the positions of the multiple waste treatment nodes, wherein the multiple waste treatment nodes correspond to the multiple production nodes one by one; a connection component identification unit for identifying the connection components between the multiple production nodes and the multiple waste treatment nodes, and outputting multiple connection components; a reaction impact indicator output unit for performing waste reaction impact analysis on the multiple connection components according to the multiple abnormal indicators, and outputting multiple reaction impact indicators.

[0048] Wherein, the waste reaction impact analysis is performed on the multiple connection components according to the multiple abnormal indicators, and the reaction impact indicator output unit may further include: a connection component information determination subunit is used to determine the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component; an abnormal indicator simulation subunit is used to simulate each abnormal indicator based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, predict the reaction impact indicator corresponding to each connection component, and output multiple reaction impact indicators corresponding to the multiple abnormal indicators.

[0049] Among them, the positions of the multiple waste treatment nodes are obtained, and the node position acquisition unit may further include: a judgment subunit is used to judge whether the multiple waste treatment nodes include multi-branch waste treatment nodes based on the positions of the multiple waste treatment nodes, wherein the multi-branch waste treatment node refers to a waste treatment node composed of N branches, and each branch is connected to a different production node, and N is a positive integer greater than or equal to 2; a single-branch waste reaction impact analysis subunit is used to perform a single-branch waste reaction impact analysis according to the abnormal indicators of each waste treatment node if the multiple waste treatment nodes do not include a multi-branch waste treatment node.

[0050] Among them, the node position acquisition unit may further include: a multi-branch waste treatment node abnormal indicator acquisition subunit is used to obtain the abnormal indicators of the multi-branch waste treatment nodes if the multiple waste treatment nodes include multi-branch waste treatment nodes; a branch waste reaction impact analysis subunit is used to perform waste reaction impact analysis on N branches respectively according to the abnormal indicators of the multi-branch waste treatment nodes, and output N reaction impact indicators; an early warning signal acquisition subunit is used to obtain an early warning signal when any reaction impact indicator among the N reaction impact indicators is greater than a preset impact indicator.

[0051] The specific configuration of the abnormality analysis module 50 will be described in detail below. As described above, the abnormality analysis is performed on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste processing nodes. The abnormality analysis module 50 may further include: a historical waste monitoring sample setting unit is used to set historical normal waste monitoring samples and historical local abnormal waste monitoring samples for the multiple waste processing nodes; an initialization abnormality analysis model definition unit is used to define an initialization abnormality analysis model according to an abnormal layout factor algorithm; a model training unit is used to train the initialization abnormality analysis model according to the historical normal waste monitoring samples, the historical local abnormal waste monitoring samples and the identification information of the abnormality degree to obtain an abnormality analysis model; an abnormality analysis unit is used to call the abnormality analysis model to perform abnormality analysis on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste processing nodes.

[0052] Among them, each abnormal indicator is simulated according to the component material, connection air tightness, connection layer relationship and connection pipeline length of each connecting component as the characteristic vector, and the abnormal indicator simulation subunit may further include: a resistance network layer establishment microunit for establishing a resistance network layer; a protection evaluation microunit for performing a protection evaluation with the component material, connection air tightness, connection layer relationship and connection pipeline length of each connecting component as the characteristic vector, and outputting the protection index of the connecting component; a prediction microunit is used to predict the protection index of each connecting component and each abnormal indicator using the resistance network layer, and output the reaction influence index corresponding to each connecting component.

[0053] A real-time fault warning device for a drug production line provided in an embodiment of the present invention can execute a real-time fault warning method for a drug production line provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to the execution method.

[0054] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0055] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A real-time fault warning method for a drug production line, characterized in that: The method comprises: Analyze each production node in the drug production line and obtain multiple waste treatment nodes; Obtaining waste characteristics corresponding to each of the multiple waste processing nodes; Arrange multiple sensor modules according to the waste characteristics corresponding to each node; Performing data sensing monitoring according to the multiple sensor modules to obtain multiple waste monitoring data sets; Performing anomaly analysis on the multiple waste monitoring data sets to obtain multiple anomaly indicators corresponding to the multiple waste processing nodes respectively; Perform waste reaction impact analysis according to the multiple abnormal indicators, and output multiple reaction impact indicators; Fitting is performed according to the plurality of reaction influence indicators to construct a reaction influence relationship network, and the reaction influence relationship network is called to perform fault warning on the drug production line; Wherein, waste reaction impact analysis is performed according to the multiple abnormal indicators, and multiple reaction impact indicators are output, including: Obtaining locations of the plurality of waste processing nodes; Acquire a plurality of production nodes connected to the plurality of waste treatment nodes based on the positions of the plurality of waste treatment nodes, wherein the plurality of waste treatment nodes correspond one-to-one to the plurality of production nodes; Outputting a plurality of connection components by identifying connection components between the plurality of production nodes and the plurality of waste treatment nodes; Performing waste reaction impact analysis on the plurality of connection components according to the plurality of abnormal indicators, and outputting a plurality of reaction impact indicators; Wherein, performing waste reaction impact analysis on the plurality of connection components according to the plurality of abnormal indicators includes: Determine the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component; Each abnormal index is simulated based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, the reaction influence index corresponding to each connection component is predicted, and multiple reaction influence indexes corresponding to the multiple abnormal indicators are output.

2. The method according to claim 1, characterized in that Acquiring the locations of the plurality of waste processing nodes, the method further comprises: Determining whether the plurality of waste treatment nodes include a multi-branch waste treatment node based on the positions of the plurality of waste treatment nodes, wherein the multi-branch waste treatment node refers to a waste treatment node consisting of N branches, and each branch is connected to a different production node, and N is a positive integer greater than or equal to 2; If the plurality of waste treatment nodes do not include a multi-branch waste treatment node, a single-branch waste reaction impact analysis is performed based on the abnormal index of each waste treatment node.

3. The method according to claim 2, characterized in that If the plurality of waste processing nodes include a multi-branch waste processing node, obtaining an abnormality indicator of the multi-branch waste processing node; According to the abnormal indicators of the multi-branch waste treatment node, respectively perform waste reaction impact analysis on N branches, and output N reaction impact indicators; When any of the N reaction impact indicators is greater than a preset impact indicator, an early warning signal is obtained.

4. The method according to claim 1, characterized in that Performing anomaly analysis on the multiple waste monitoring data sets to obtain multiple anomaly indicators corresponding to the multiple waste processing nodes, the method comprising: Setting historical normal waste monitoring samples and historical local abnormal waste monitoring samples for the multiple waste processing nodes; According to the abnormal layout factor algorithm, the initialization abnormal analysis model is defined; The initialized abnormality analysis model is trained according to the historical normal waste monitoring samples, the historical local abnormal waste monitoring samples and the identification information of the abnormality degree to obtain an abnormality analysis model; The abnormality analysis model is called to perform abnormality analysis on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste processing nodes respectively.

5. The method according to claim 1, characterized in that Each abnormal index is simulated based on the component material, connection air tightness, connection layer relationship and connection pipeline length of each connection component as a feature vector, and the method includes: Building a resistant network layer; The protectiveness of each connection component is evaluated using the component material, connection air tightness, connection layer relationship and connection pipeline length as feature vectors, and the protectiveness index of the connection component is output; The resistance network layer is used to predict the protective index and each abnormal index of each connection component, and the reaction impact index corresponding to each connection component is output.

6. A real-time fault warning device for a drug production line, characterized in that: The device is used to implement the real-time fault warning method for a drug production line according to any one of claims 1 to 5, and the device comprises: A waste treatment node acquisition module, which is used to analyze each production node in the drug production line and acquire multiple waste treatment nodes; A node waste feature acquisition module, the node waste feature acquisition module is used to acquire waste features corresponding to each node in the multiple waste processing nodes; A sensor deployment module, the sensor deployment module is used to deploy multiple sensor modules according to the waste characteristics corresponding to each node; A data sensing monitoring module, the data sensing monitoring module is used to perform data sensing monitoring according to the multiple sensing modules to obtain multiple waste monitoring data sets; An abnormality analysis module, the abnormality analysis module is used to perform abnormality analysis on the multiple waste monitoring data sets to obtain multiple abnormal indicators corresponding to the multiple waste processing nodes respectively; A waste reaction impact analysis module, the waste reaction impact analysis module is used to perform waste reaction impact analysis according to the multiple abnormal indicators and output multiple reaction impact indicators; A fault warning module is used to perform fitting according to the multiple reaction influence indicators, construct a reaction influence relationship network, and call the reaction influence relationship network to perform fault warning on the drug production line.

Images