Intelligent operation and maintenance system for traditional Chinese medicine preparation workshop

Abstract

The invention relates to an intelligent operation and maintenance system for a traditional Chinese medicine preparation workshop, and the system is characterized in that the system employs a cloud-side-end collaborative architecture, and comprises an equipment layer which comprises traditional Chinese medicine production equipment and an online sensor, and is used for executing the production operation and collecting the data of a production process; the edge layer is in communication connection with the equipment layer and comprises an edge computing gateway and a control system, the edge computing gateway is provided with multi-protocol conversion middleware and is used for carrying out collection, cleaning and protocol conversion on equipment data from different protocols, and the control system is used for executing basic automatic control; the platform layer is in communication connection with the edge layer and comprises a data middle platform, a model platform and a block chain node, the data middle platform is used for storing and managing production data from the edge layer, and the model platform is deployed with a process model and a self-adaptive control algorithm and is used for performing data analysis and optimization calculation; the block chain node is used for storing a hash value of key traceability data; and the application layer is in communication connection with the platform layer, comprises a self-adaptive optimization control application and a traceability query application, and is used for providing an optimization control interface and full-process traceability information query for a user. According to the scheme, the technical combination is advanced, the implementation difficulty is high, cooperation with a solution provider with industrial internet and intelligent manufacturing implementation experience is suggested, and compound talent cultivation which knows processes and data in an enterprise is paid attention to.

Description

Technical Field

[0001] This invention relates to the field of traditional Chinese medicine pharmaceutical technology, and in particular to an information management system for traditional Chinese medicine preparation workshops based on the Internet of Things, big data and artificial intelligence, which realizes digital monitoring and intelligent analysis of the entire production process. Background Technology

[0002] Traditional Chinese medicine (TCM) production suffers from complex processes, reliance on experience for parameters, and data silos across different stages. TCM production relies heavily on human experience, and process parameters (such as extraction temperature and drying time) are not precisely controlled.

[0003] Quality inspection is lagging behind, and traditional offline inspection cannot intervene in production in real time.

[0004] Material traceability is difficult, and batch information records are fragmented.

[0005] Energy and equipment utilization is low, and dynamic optimization is lacking.

[0006] Traditional DCS systems are not well-suited for the field of traditional Chinese medicine and lack the ability to integrate multi-source data (such as component analysis and environmental temperature and humidity).

[0007] Existing systems struggle to achieve functions such as automatic GMP compliance verification and quality traceability. To address these shortcomings, this invention proposes improvements. Summary of the Invention

[0008] The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops proposed in this invention solves the problems of isolated multiple systems, extensive process control, and difficulty in quality traceability in the existing technology of traditional Chinese medicine production, and provides a highly compatible and scalable intelligent production information platform.

[0009] The technical solution of this invention is implemented as follows: A smart operation and maintenance system for a traditional Chinese medicine preparation workshop, characterized in that the system adopts a cloud-edge-device collaborative architecture, comprising: The equipment layer includes traditional Chinese medicine production equipment and online sensors, which are used to perform production operations and collect production process data. The edge layer, which is communicatively connected to the device layer, includes an edge computing gateway and a control system. The edge computing gateway is configured with a multi-protocol conversion middleware for collecting, cleaning, and converting device data from different protocols. The control system is used to perform basic automation control. The platform layer, which communicates with the edge layer, includes a data platform, a model platform, and blockchain nodes. The data platform is used to store and manage production data from the edge layer. The model platform is deployed with process models and adaptive control algorithms for data analysis and optimization calculations. The blockchain nodes are used to store the hash values ​​of key traceability data. The application layer, which communicates with the platform layer, includes an adaptive optimization control application and a traceability query application, used to provide users with an optimized control interface and full-process traceability information query.

[0010] Preferably, the process model deployed on the model platform is a partial least squares (PLS) model, and the adaptive control algorithm is configured as follows: Real-time process data is acquired from the data platform at preset time intervals; To optimize the key quality attributes predicted by the PLS model to reach the target value, an optimization algorithm is used to back-calculate a set of optimal key process parameter settings. The optimal key process parameter settings are recommended to the operator in suggestion mode or directly issued to the control system of the edge layer in automatic mode to form closed-loop optimization control.

[0011] Preferably, the adaptive control algorithm further includes a model update module for periodically fine-tuning or retraining the PLS model using new production data.

[0012] Preferably, the multi-protocol conversion middleware includes: Multiple protocol driver plugins are available to support communication with devices or control systems that use different protocols. A unified data model is used to map the collected raw data into standardized data points; The data routing and forwarding module is used to transmit standardized data to the data middleware of the platform layer via MQTT or OPC UA protocol; The edge disaster recovery module is used to cache data locally when the network is interrupted and to resume transmission from where it left off after the network is restored.

[0013] Preferably, the blockchain node adopts a consortium blockchain architecture and runs smart contracts; the smart contracts define logic for data uploading, traceability information querying, and data verification, and are used for: Receive traceability data hash values, centered on product batch numbers, from the data platform and record them on the blockchain; In response to a traceability query request, return an index of full-chain traceability information corresponding to the specified product batch number; Provides a data verification interface to verify whether the detailed traceability data stored off-chain is consistent with the hash value recorded on-chain.

[0014] Preferably, the original data corresponding to the hash value of the traceability data on the blockchain includes at least one of the following: material information, key process parameters, environmental data, quality inspection results, and operation information.

[0015] Preferably, the online sensor includes a near-infrared spectroscopy (NIR) sensor for detecting material concentration.

[0016] Preferably, the edge layer and the platform layer, as well as the platform layer and the application layer, interact with each other via the OPC UA protocol or a RESTful API.

[0017] Preferably, the optimization algorithm is one of gradient descent, genetic algorithm, or model predictive control algorithm.

[0018] An adaptive optimization control method for a traditional Chinese medicine preparation workshop, characterized in that the method operates on the model platform of the system as described in any one of claims 1-9, and includes the following steps: Acquire real-time production data from the data platform, including key process parameters and online sensor data; The real-time production data is input into a pre-trained PLS process model to predict the key quality attributes under the current state. To optimize the prediction of key quality attributes to the target value, optimization calculations are performed to obtain a set of optimized key process parameter settings. The optimized key process parameter settings are output to the adaptive optimization control application at the application layer for operator confirmation in suggestion mode or direct execution in automatic mode.

[0019] In summary, the intelligent operation and maintenance system for traditional Chinese medicine preparation workshops disclosed in this invention has the following beneficial effects: 1. Quality Improvement and Stabilization: Reduce batch-to-batch variations and increase product yield through real-time prediction and optimization.

[0020] 2. Efficiency and cost optimization: Achieving better process parameters may shorten the production cycle and reduce energy and material consumption.

[0021] 3. Compliance and Trustworthy Traceability: Meets the requirements of the National Medical Products Administration for drug traceability, and blockchain technology enhances the credibility of data, empowering the brand.

[0022] 4. Digital Transformation: Break down information silos, form data assets, and lay a solid foundation for future AI applications and intelligent decision-making.

[0023] This solution features an advanced technology combination, but it is also quite difficult to implement. It is recommended to collaborate with solution providers who have experience in implementing industrial internet and smart manufacturing, and to focus on cultivating compound talents within the enterprise who understand both processes and data. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the overall framework of the intelligent operation and maintenance system for the traditional Chinese medicine preparation workshop in this embodiment.

[0026] Figure 2 The flowchart for adaptive optimization control.

[0027] Figure 3 This is a schematic diagram of the blockchain traceability data process. Detailed Implementation

[0028] The following will refer to the appendices in the embodiments of the present invention. Figure 1-3 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Example

[0029] like Figures 1 to 3 As shown in the accompanying drawings, the present invention will now be described in further detail.

[0030] This embodiment presents a smart operation and maintenance system for a traditional Chinese medicine (TCM) preparation workshop. The system adopts a cloud-edge-device collaborative architecture, comprising: an equipment layer, including TCM production equipment and online sensors, for performing production operations and collecting production process data; an edge layer, communicatively connected to the equipment layer, including an edge computing gateway and a control system. The edge computing gateway is configured with multi-protocol conversion middleware for collecting, cleaning, and converting device data from different protocols. The control system is used to perform basic automated control; a platform layer, communicatively connected to the edge layer, including a data middleware, a model platform, and blockchain nodes. The data middleware stores and manages production data from the edge layer. The model platform deploys process models and adaptive control algorithms for data analysis and optimization calculations. The blockchain nodes store hash values ​​of key traceability data; and an application layer, communicatively connected to the platform layer, including an adaptive optimization control application and a traceability query application, for providing users with an optimized control interface and full-process traceability information query.

[0031] Preferably, the process model deployed on the model platform is a partial least squares (PLS) model, and the adaptive control algorithm is configured as follows: Real-time process data is acquired from the data platform at preset time intervals; To optimize the key quality attributes predicted by the PLS model to reach the target value, an optimization algorithm is used to back-calculate a set of optimal key process parameter settings. The optimal key process parameter settings are recommended to the operator in suggestion mode or directly issued to the control system of the edge layer in automatic mode to form closed-loop optimization control.

[0032] Preferably, the adaptive control algorithm further includes a model update module for periodically fine-tuning or retraining the PLS model using new production data.

[0033] Preferably, the multi-protocol conversion middleware includes: Multiple protocol driver plugins are available to support communication with devices or control systems that use different protocols. A unified data model is used to map the collected raw data into standardized data points; The data routing and forwarding module is used to transmit standardized data to the data middleware of the platform layer via MQTT or OPC UA protocol; The edge disaster recovery module is used to cache data locally when the network is interrupted and to resume transmission from where it left off after the network is restored.

[0034] Preferably, the blockchain node adopts a consortium blockchain architecture and runs smart contracts; the smart contracts define logic for data uploading, traceability information querying, and data verification, and are used for: Receive traceability data hash values, centered on product batch numbers, from the data platform and record them on the blockchain; In response to a traceability query request, return an index of full-chain traceability information corresponding to the specified product batch number; Provides a data verification interface to verify whether the detailed traceability data stored off-chain is consistent with the hash value recorded on-chain.

[0035] Preferably, the original data corresponding to the hash value of the traceability data on the blockchain includes at least one of the following: material information, key process parameters, environmental data, quality inspection results, and operation information.

[0036] Preferably, the online sensor includes a near-infrared spectroscopy (NIR) sensor for detecting material concentration.

[0037] Preferably, the edge layer and the platform layer, as well as the platform layer and the application layer, interact with each other via the OPC UA protocol or a RESTful API.

[0038] Preferably, the optimization algorithm is one of gradient descent, genetic algorithm, or model predictive control algorithm.

[0039] An adaptive optimization control method for a traditional Chinese medicine preparation workshop, characterized in that the method runs on a model platform within the system and includes the following steps: Acquire real-time production data from the data platform, including key process parameters and online sensor data; The real-time production data is input into a pre-trained PLS process model to predict the key quality attributes under the current state. To optimize the prediction of key quality attributes to the target value, optimization calculations are performed to obtain a set of optimized key process parameter settings. The optimized key process parameter settings are output to the adaptive optimization control application at the application layer for operator confirmation in suggestion mode or direct execution in automatic mode.

[0040] I. Overall Architecture Design The solution adopts a collaborative architecture of "cloud-edge-device" to ensure the system's flexibility, real-time performance, and security.

[0041] 1. Equipment Layer (End): This includes production equipment such as reaction vessels, extraction tanks, concentrators, dryers, and packaging machines, as well as sensors for temperature, pressure, pH, and concentration (near-infrared spectroscopy, NIR). These are the sources of data and the terminals for execution and control.

[0042] 2. Edge layer (edge): DCS / PLC system: responsible for real-time, high-reliability basic automation control of the production process (such as PID control and interlocking alarms).

[0043] Edge computing gateway: Deploys multi-protocol conversion middleware, responsible for collecting device data with different protocols, performing preliminary cleaning, caching and protocol conversion, and uploading it to the platform layer in a unified manner.

[0044] 3. Platform Layer (Cloud / Private Server): Data middle platform / industrial internet platform: Receives and stores all production data from the edge layer to build a unified data warehouse.

[0045] Digital Twin and Model Platform: Deploy and run PLS process models and adaptive control algorithms, and perform data analysis and optimization instruction calculations.

[0046] Blockchain node: Deploys a blockchain traceability module to receive key traceability data from the data platform and upload it to the blockchain.

[0047] 4. Application Layer: Adaptive Optimization Control App: Displays optimization suggestions to operators and can authorize algorithms to automatically send optimized setpoints (SP) to the DCS.

[0048] Traceability Inquiry App: Provides a full-process traceability information inquiry interface for internal quality inspectors, regulatory agencies, and consumers.

[0049] Production monitoring dashboard: Provides comprehensive monitoring of production status.

[0050] II. Phased Implementation Plan Module 1: Adaptive Control Algorithm for Traditional Chinese Medicine Processes (DCS + PLS) Objective: To achieve online prediction of critical quality attributes (CQA) and dynamic closed-loop optimization of critical process parameters (CPP).

[0051] step: 1. Data preparation and historical analysis: Collect historical production data, including CPP (such as extraction temperature, time, solvent flow rate) and CQA (such as active ingredient content, extract yield, and other endpoint indicators).

[0052] By using mechanistic analysis and data mining (such as variable projection importance, VIP), the core CPPs affecting CQA are identified.

[0053] 2. Offline PLS model development: An offline PLS model is trained using historical data. This model takes real-time process parameters (CPP) and online sensor data (such as NIR spectra) as input to predict the critical quality attributes (CQA) of the final product.

[0054] The model is rigorously validated and tested to ensure that its prediction accuracy meets engineering requirements.

[0055] 3. Adaptive control algorithm design: Integration: The trained PLS model is embedded into the optimization algorithm. This algorithm periodically (e.g., every 5 minutes) retrieves the latest real-time data from the data platform.

[0056] Optimization: The algorithm aims to "make the predicted CQA reach the target value" or "be optimal within the constraints". It uses optimization algorithms (such as gradient descent, genetic algorithm, model predictive control MPC) to back-calculate a set of optimal CPP settings.

[0057] Decision-making and execution: Recommended mode: The optimized settings are recommended to the operator, who then confirms them manually before issuing them to the DCS.

[0058] Automatic mode (advanced): After gaining sufficient trust, the algorithm can automatically write the optimized setpoints into the DCS through a standard interface (such as OPC UA) to form a closed-loop optimized control.

[0059] Model self-updating: Design a model update strategy to periodically fine-tune or retrain the PLS model with new production data to adapt to process drift and raw material changes.

[0060] Module 2: Multi-protocol conversion middleware Objective: To achieve unified access and data standardization for heterogeneous devices.

[0061] step: 1. Equipment Protocol Survey: Inventory all equipment, instruments, and control systems in the plant, including their brands, models, and supported protocols (such as Modbus RTU / TCP, Siemens S7, OPC DA / UA, HTTP / RESTful API, MQTT, etc.).

[0062] 2. Middleware selection and development: Option A (Recommended): Use mature industrial IoT platform edge components (such as Huawei IoTA, ThingsBoard, BeeNet, etc.), which usually have rich protocol drivers built-in.

[0063] Option B (Self-developed): Custom development based on open-source frameworks (such as Node-RED, Eclipse Kura) or using high-performance languages ​​(such as Go).

[0064] 3. Core functionality implementation: Protocol driver plug-in: Develop independent driver plug-ins for each protocol to achieve "hot-plugging".

[0065] Data mapping and modeling: Define a unified PointList for each data point (such as "temperature of reactor No. 1") to establish a mapping relationship from the original address to the unified data model.

[0066] Data routing and forwarding: The processed standardized data is transmitted stably and efficiently to the cloud data platform via standard protocols such as MQTT (preferred) or OPC UA.

[0067] Edge disaster recovery: It has the ability to resume transmission after network interruption, caches data locally when the network is interrupted, and retransmits it after recovery.

[0068] 4. Deployment and Testing: Deploy middleware on the edge computing gateway (industrial-grade PC or dedicated gateway device), and conduct interoperability testing on each device to ensure the accuracy and stability of data collection.

[0069] Module 3: Blockchain-based Traceability Module Objective: To ensure that core production data is tamper-proof and to provide reliable traceability services.

[0070] step: 1. Blockchain Selection: Choose a consortium blockchain (such as FISCO BCOS, Hyperledger Fabric). Consortium blockchains offer high performance and strong controllability, making them suitable for enterprise-level applications. Enterprises, regulatory agencies, and other stakeholders act as consensus nodes on the chain.

[0071] 2. Source tracing data model design: Identify the key traceability units on the blockchain (usually centered around the "batch number").

[0072] Design the on-chain data structure, including: Material information: raw material batch, place of origin, and inspection report hash.

[0073] Process parameters: CPP data hash of key processes (in order to save on-chain space, the hash values ​​of a large amount of raw data are usually put on the chain, while the raw data is stored in an off-chain database).

[0074] Environmental data: workshop temperature, humidity, cleanliness, etc.

[0075] Quality inspection: Hash of quality inspection results reports for intermediate and finished products.

[0076] Operation information: Operator, timestamp, device ID.

[0077] 3. Smart Contract Development: Write smart contracts to define the logic for data uploading, querying, and verification.

[0078] Main functions: `addTraceData` (add trace data), `getBatchTrace` (query full-chain information by batch number), `verifyData` (verify whether off-chain data is consistent with on-chain hash).

[0079] 4. System Integration: After receiving a message that a batch of production has been completed, the data platform automatically triggers a smart contract to upload the core data hash value of that batch to the blockchain.

[0080] Develop a traceability query application. The front-end interface inputs the product batch number, the application obtains the hash index from the blockchain, and then retrieves detailed data from the off-chain database, allowing users to verify the integrity of the data.

[0081] III. Implementation Roadmap (It is recommended to divide it into three phases) Phase 1: Infrastructure Development and Data Integration (3-6 months) 1. Deploy an edge computing gateway, develop and implement multi-protocol conversion middleware to achieve unified collection and uploading of data from all devices.

[0082] 2. Build a data middle platform and complete the construction of the data warehouse.

[0083] 3. Complete the construction of the blockchain consortium chain and the initial development of smart contracts.

[0084] Phase Two: Model Development and Pilot Application (6-8 months) 1. Select 1-2 core processes (such as extraction and concentration) as pilot projects.

[0085] 2. Develop an offline PLS prediction model for this process based on historical data.

[0086] 3. Develop an initial version of the adaptive control algorithm, run it in the recommended mode first, and verify its effectiveness in collaboration with senior process engineers.

[0087] 4. Complete the integration of the blockchain traceability module with the pilot process to achieve automatic data upload to the blockchain.

[0088] Phase 3: Comprehensive Promotion and Optimization (6 months+) 1. Extend the adaptive control algorithm to other key processes.

[0089] 2. After verifying the reliability of the algorithm, try the closed-loop automatic optimization mode in the pilot process.

[0090] 3. Improve the traceability query app and grant consumers some query permissions (such as checking traceability by scanning a code).

[0091] 4. Establish a mechanism for continuous learning and updating of the model.

[0092] This embodiment takes the "extraction workshop" of a Chinese medicine pharmaceutical company as the implementation object and focuses on the core equipment, the "multifunctional extraction tank".

[0093] 1. System Composition and Deployment Equipment layer (end): Production equipment: 1 multi-functional extraction tank.

[0094] Sensors: Temperature sensors, pressure sensors, and an online near-infrared spectroscopy (NIR) analyzer installed on the extraction tank for real-time detection of the concentration of target components in the extract.

[0095] Edge layer (edge): Control system: A Siemens S7-1500 series PLC is responsible for basic PID control and interlocking alarms of the actuators for heating, stirring, liquid inlet and liquid outlet of the extraction tank.

[0096] Edge computing gateway: A Huawei Atlas 500 smart station is used, on which a multi-protocol conversion middleware based on Node-RED is deployed.

[0097] This middleware is configured with a Siemens S7 protocol driver to collect data such as temperature, pressure, and steam valve opening from the PLC.

[0098] It also includes a Modbus TCP driver for acquiring real-time spectral data and calculating concentration values ​​from an online NIR analyzer.

[0099] The middleware maps all data into JSON format and sends data packets to the cloud platform layer every 10 seconds via the MQTT protocol.

[0100] Platform layer (cloud / private server): Data Platform: Deployed based on the open-source platform ThingsBoard, it creates an "extraction workshop" data model to receive and persistently store all real-time data from the edge gateway.

[0101] Model platform: RESTful API services are developed using the Python Flask framework, and the following core algorithms are deployed: PLS model: A pre-trained partial least squares model. Its input variables (X) include: extraction temperature, extraction time, and real-time NIR concentration; its output variables (Y) are the predicted endpoint effective ingredient content and extract yield.

[0102] Adaptive control algorithm: Model predictive control (MPC) algorithm is adopted. This algorithm retrieves the latest data from the data platform every 5 minutes, calls the PLS model for rolling optimization, and calculates the optimal temperature setpoint for a period of time in the future with the goal of "stabilizing the predicted endpoint effective ingredient content at the target value (e.g., 5.0 mg / g)".

[0103] Blockchain nodes: Deployed traceability smart contracts based on the FISCO BCOS consortium blockchain. The enterprise itself and its superior regulatory agency serve as consensus nodes.

[0104] Application layer: Adaptive Optimization Control App: A web application interface that displays current process parameters, NIR concentration, PLS model prediction results, and recommended temperature setpoints calculated by the MPC algorithm. Operators can choose to "adopt the suggestion" or "ignore" on this interface.

[0105] Traceability Query App: A web application accessible via PC or mobile phone that provides the function of querying traceability information by product batch number.

[0106] 2. Workflow Step 1: Data Acquisition and Processing The edge gateway collects data from both Siemens PLCs and NIR analyzers simultaneously via a multi-protocol conversion middleware.

[0107] After being cleaned and standardized, the data is uploaded to the data platform via the MQTT protocol.

[0108] The data platform stores and manages data and provides data access interfaces for the model platform.

[0109] Step 2: Adaptive optimization control (running in "recommendation mode") The MPC algorithm is triggered periodically (every 5 minutes) to retrieve real-time data for the current batch of extraction jobs from the data platform.

[0110] The MPC algorithm calls the PLS model deployed on the model platform to predict the final component content if the current operating conditions continue.

[0111] If the predicted result deviates from the target value, the MPC algorithm will perform optimization calculations. For example, it might calculate that "raising the temperature setpoint from 98°C to 100°C within the next 15 minutes" would make the predicted content closest to the target value.

[0112] The optimized setpoint (100℃) is sent to the adaptive optimization control APP and displayed prominently on the interface to the operator, along with the reason for the optimization (such as "to improve the predictive content").

[0113] Once an experienced operator confirms the suggestion is reasonable, they click the "Accept" button. This instruction, transmitted through the platform and edge layers, is ultimately sent to the Siemens PLC as a write command (via the OPC UA protocol) to modify the setpoint (SP) of the temperature control loop.

[0114] The PLC automatically adjusts the opening of the steam valve based on the new set values ​​to achieve precise temperature control.

[0115] Step 3: Blockchain Traceability When this batch of extraction operations is completed, the data platform generates a "critical traceability data package," which includes: Raw material batch ID and place of origin.

[0116] The core process parameters extracted in this batch (such as average temperature, final NIR concentration, and MPC optimization records) are as follows.

[0117] Operator ID, timestamp.

[0118] Batch ID link for subsequent processes (such as concentration).

[0119] The data platform calculates the hash value (such as SHA-256) of the data packet and calls the smart contract method addTraceData on the blockchain to bind the hash value with the product batch number "EX20231026001" and put it on the chain.

[0120] When a consumer scans the QR code on the product packaging, the traceability query app is triggered. It uses the `getBatchTrace` method of the smart contract to query all on-chain hash indices corresponding to batch number "EX20231026001," then retrieves the corresponding detailed raw data from the off-chain database of the data platform, combining it into a complete and visualized traceability report presented to the consumer. A "Verify" button is provided at the bottom of the report, using the `verifyData` method to prove that the data has not been tampered with.

[0121] 3. Implementation Results Through the implementation of this embodiment, the traditional Chinese medicine enterprise has achieved the following benefits: Quality improvement: The batch-to-batch variation (RSD) of the final active ingredient content in the extraction process was reduced from ±15% to within ±5%, and the yield of superior products increased by about 8%.

[0122] Efficiency optimization: Through dynamic optimization, the average extraction time per batch was reduced by 10%, thus reducing energy consumption.

[0123] Traceability and credibility: An unalterable, end-to-end traceability system has been established, easily passing GMP audits and enhancing brand reputation.

[0124] This embodiment fully demonstrates the effectiveness and advancement of the system described in this invention in improving the quality, efficiency, and transparency of traditional Chinese medicine manufacturing.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart operation and maintenance system for a traditional Chinese medicine preparation workshop, characterized in that, The system adopts a cloud-edge-device collaborative architecture, including: The equipment layer includes traditional Chinese medicine production equipment and online sensors, which are used to perform production operations and collect production process data. The edge layer, which is communicatively connected to the device layer, includes an edge computing gateway and a control system. The edge computing gateway is configured with a multi-protocol conversion middleware for collecting, cleaning, and converting device data from different protocols. The control system is used to perform basic automation control. The platform layer, which communicates with the edge layer, includes a data platform, a model platform, and blockchain nodes. The data platform is used to store and manage production data from the edge layer. The model platform is deployed with process models and adaptive control algorithms for data analysis and optimization calculations. The blockchain nodes are used to store the hash values ​​of key traceability data. The application layer, which communicates with the platform layer, includes an adaptive optimization control application and a traceability query application, used to provide users with an optimized control interface and full-process traceability information query.

2. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 1, characterized in that, The process model deployed on the model platform is a partial least squares (PLS) model, and the adaptive control algorithm is configured as follows: Real-time process data is acquired from the data platform at preset time intervals; To optimize the key quality attributes predicted by the PLS model to reach the target value, an optimization algorithm is used to back-calculate a set of optimal key process parameter settings. The optimal key process parameter settings are recommended to the operator in suggestion mode or directly issued to the control system of the edge layer in automatic mode to form closed-loop optimization control.

3. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 2, characterized in that, The adaptive control algorithm further includes a model update module, which is used to periodically fine-tune or retrain the PLS model using new production data.

4. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 1, characterized in that, The multi-protocol conversion middleware includes: Multiple protocol driver plugins are available to support communication with devices or control systems that use different protocols. A unified data model is used to map the collected raw data into standardized data points; The data routing and forwarding module is used to transmit standardized data to the data middleware of the platform layer via MQTT or OPC UA protocol; The edge disaster recovery module is used to cache data locally when the network is interrupted and to resume transmission from where it left off after the network is restored.

5. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 1, characterized in that, The blockchain nodes adopt a consortium blockchain architecture and run smart contracts; the smart contracts define logic for data uploading, traceability information querying, and data verification, and are used for: Receive traceability data hash values, centered on product batch numbers, from the data platform and record them on the blockchain; In response to a traceability query request, return an index of full-chain traceability information corresponding to the specified product batch number; Provides a data verification interface to verify whether the detailed traceability data stored off-chain is consistent with the hash value recorded on-chain.

6. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 5, characterized in that, The original data corresponding to the hash value of the traceability data on the blockchain includes at least one of the following: material information, key process parameters, environmental data, quality inspection results, and operation information.

7. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 1, characterized in that, The online sensor includes a near-infrared spectroscopy (NIR) sensor for detecting material concentration.

8. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 1, characterized in that, The edge layer and the platform layer, as well as the platform layer and the application layer, interact with each other via the OPC UA protocol or RESTful API.

9. The intelligent operation and maintenance system for traditional Chinese medicine preparation workshops according to claim 2, characterized in that, The optimization algorithm is one of gradient descent, genetic algorithm, or model predictive control algorithm.

10. An adaptive optimization control method for a traditional Chinese medicine preparation workshop, characterized in that, The method operates on the model platform of the system as described in any one of claims 1-9, and includes the following steps: Acquire real-time production data from the data platform, including key process parameters and online sensor data; The real-time production data is input into a pre-trained PLS process model to predict the key quality attributes under the current state. To optimize the prediction of key quality attributes to the target value, optimization calculations are performed to obtain a set of optimized key process parameter settings. The optimized key process parameter settings are output to the adaptive optimization control application at the application layer for operator confirmation in suggestion mode or direct execution in automatic mode.

Images