Compact PLC controller function module

Abstract

The invention provides a compact PLC (Programmable Logic Controller) function module. A functional module belongs to the technical field of PLC controllers and comprises a data acquisition unit, an edge computing unit, a central processing unit, a data storage unit, a machine learning accelerator, a communication unit, a security encryption unit and a self-adaptive control logic unit. The output end of the data acquisition unit is connected with the input end of the edge calculation unit; the edge computing unit is bidirectionally connected with the central processing unit and the machine learning accelerator; and the central processing unit is bidirectionally connected with the data storage unit, the communication unit and the adaptive control logic unit. By adopting a high-density integrated circuit and a system-level packaging technology, the PLC controller realizes tight integration of multifunctional modules, reduces the volume, improves the space utilization rate, and is suitable for various harsh industrial environments.

Description

Technical Field

[0001] The invention provides a compact PLC controller function module, belonging to the technical field of PLC controllers. Background Art

[0002] In the field of industrial automation, with the rapid development of the Internet of Things (IoT), big data and artificial intelligence (AI) technologies, traditional programmable logic controllers (PLCs) are facing unprecedented challenges and opportunities. Early PLCs were mainly used to perform simple logic control tasks, such as switch logic, timing and counting. Their designs tended to be hardware-based and modular. Although they were reliable and stable, they were unable to handle complex working condition analysis, large-scale data processing and intelligent decision-making. With the transformation of manufacturing to intelligent manufacturing, higher requirements are placed on the real-time, flexibility, intelligence and safety of production systems, and the limitations of traditional PLCs are becoming increasingly prominent.

[0003] In recent years, control systems have developed towards a more integrated and intelligent direction. As the core component of the new generation of automated control, the compact PLC controller aims to achieve full-chain intelligent management from data acquisition to decision execution through highly integrated functional modules. However, to meet this demand, multiple technical obstacles need to be overcome, including how to efficiently integrate diversified functional modules in a limited space, how to achieve efficient processing and analysis of real-time data, how to perform remote monitoring and management while ensuring information security, and how to automatically adjust control strategies in a dynamically changing production environment to achieve optimal system performance. Summary of the invention

[0004] The present invention provides a compact PLC controller function module to solve the problems raised in the above background technology:

[0005] The present invention proposes a compact PLC controller functional module, which includes: a data acquisition unit, an edge computing unit, a central processing unit, a data storage unit, a machine learning accelerator, a communication unit, a security encryption unit and an adaptive control logic unit; the output end of the data acquisition unit is connected to the input end of the edge computing unit; the edge computing unit is bidirectionally connected to the central processing unit and the machine learning accelerator; the central processing unit is bidirectionally connected to the data storage unit, the communication unit and the adaptive control logic unit; the security encryption unit is bidirectionally connected to the data acquisition unit, the edge computing unit, the central processing unit, the data storage unit, the machine learning accelerator, the communication unit and the adaptive control logic unit.

[0006] Furthermore, the PLC controller functional module adopts high-density integrated circuits and is packaged through system-level packaging technology.

[0007] Furthermore, the working method of the functional module includes:

[0008] The data acquisition unit receives raw data from the sensor through the I / O interface, preprocesses the raw data, and transmits the preprocessed real-time data to the edge computing unit;

[0009] The edge computing unit processes the received real-time data and performs basic logical judgment and real-time control. At the same time, it identifies key data that needs further analysis and forwards it to the machine learning accelerator through an efficient data channel.

[0010] The machine learning accelerator uses pre-trained models to conduct in-depth analysis of the filtered data and feeds the analysis results back to the edge computing unit to further optimize the control strategy in combination with real-time data;

[0011] The central processing unit synthesizes the preliminary decisions from the edge computing unit, the intelligent analysis results from the machine learning accelerator, and the remote instructions, formulates the final control strategy, and stores the strategy data in the data storage unit;

[0012] The security encryption unit monitors the entire data transmission process and manages access rights to internal data;

[0013] The adaptive control logic unit dynamically adjusts the control logic according to the strategy of the central processing unit. The control instructions are output through the edge computing unit or directly by the central processing unit to drive the actuator to act and perform control;

[0014] The communication unit communicates with the remote monitoring center, uploading system status, alarm information and performance data, and receiving remote control instructions or parameter updates.

[0015] Furthermore, the edge computing method of the edge computing unit includes:

[0016] The edge computing unit receives pre-processed data from the data acquisition unit and dynamically allocates the priority of data processing according to the importance and real-time requirements of the data through an adaptive data stratification algorithm;

[0017] Through feature extraction algorithm, key features are extracted from the preprocessed data;

[0018] Build a lightweight rule engine to execute basic control instructions according to preset logic. Adopt an event-driven architecture. When a specific event is detected, the corresponding processing flow is immediately triggered.

[0019] Based on the online performance evaluation model, the control instruction execution effect is monitored in real time, the control parameters are automatically fine-tuned, and the machine learning algorithm is used to predict future data traffic and computing needs. Through the dynamic load balancing strategy, the computing resource allocation is dynamically adjusted according to the load of each processing unit in the edge computing unit;

[0020] Based on data characteristics and model requirements, intelligently select data subsets and push them to the machine learning accelerator. Through the model online fusion mechanism, the analysis results of the machine learning accelerator are integrated with the real-time control logic of the edge computing unit to form an accurate control strategy.

[0021] Based on the feedback from the machine learning accelerator, the local decision logic is adjusted to conduct online learning and self-optimization of the control strategy.

[0022] In the transmission stage before and after data processing, it works closely with the security encryption unit and adopts dynamic key management to encrypt data.

[0023] Furthermore, the online performance evaluation model is used to monitor the execution effect of control instructions in real time, automatically fine-tune control parameters, use machine learning algorithms to predict future data traffic and computing needs, and dynamically adjust the computing resource allocation according to the load conditions of each processing unit in the edge computing unit through a dynamic load balancing strategy, including:

[0024] According to the characteristics of the control task, a multi-dimensional performance evaluation index system is dynamically constructed. Through the built-in intelligent monitoring system, based on time series analysis and pattern recognition, the changing trend of indicators can be tracked in real time;

[0025] A method combining genetic algorithm and gradient descent method is used to conduct local and detailed search and global optimal exploration of control parameters, and fine-tune the control parameters;

[0026] Based on the reinforcement learning algorithm, the control strategy template is automatically generated or updated according to the current system status and historical optimization experience. The long short-term memory network is used to combine the output of multiple prediction models, and the data flow and computing requirements of the edge computing unit in the future are predicted through integrated learning technology.

[0027] Physical computing resources are virtualized to form a resource pool, and virtual resource blocks are dynamically allocated to each task or service. An adaptive load balancing algorithm that integrates task priority, resource consumption, and inter-task dependencies is used to introduce a resource reservation mechanism, and a preemption mechanism is designed to schedule resources.

[0028] Furthermore, the central processing method of the central processing unit includes:

[0029] The central processing unit receives and integrates multi-source data from the edge computing unit, machine learning accelerator, communication unit, and adaptive control logic unit, uses deep learning algorithms to perform pattern recognition and semantic understanding on multi-source data, automatically classifies data types, and analyzes hidden associations;

[0030] Based on current working conditions, historical data and external environmental factors, a situational awareness model is constructed to dynamically adjust the control strategy generation logic; based on genetic algorithms, control strategies are generated; simulation software is used to pre-simulate the effects of new strategies in a virtual environment to evaluate potential risks and benefits;

[0031] Dynamically schedule computing resources and I / O operations and optimize the order of task execution based on policy execution requirements and system resource status; use machine learning algorithms to predict future resource requirements and automatically adjust the allocation of data storage units, computing resources, and communication bandwidth;

[0032] Work closely with the security encryption unit to implement a multi-level security strategy; establish a secure connection with the remote monitoring center through the communication unit to perform remote configuration updates, command issuance and data synchronization, and conduct online learning based on feedback data and remotely updated algorithm models.

[0033] Furthermore, based on the current working conditions, historical data and external environmental factors, a situational awareness model is constructed to dynamically adjust the generation logic of the control strategy; based on the genetic algorithm, the control strategy is generated; the effect of the new strategy is pre-simulated in a virtual environment by simulation software to evaluate the potential risks and benefits, including:

[0034] Using time series analysis and data fusion technology, the data of edge computing units, machine learning accelerators, communication units, and adaptive control logic units are integrated into a unified platform;

[0035] Use deep neural networks to automatically extract key features of working conditions;

[0036] Construct a situational awareness model based on fuzzy logic, use genetic algorithms to optimize strategy parameters, and continuously iterate to generate better control strategies by simulating natural selection and genetic mechanisms;

[0037] Build a physics-based high-fidelity simulation environment, pre-execute candidate control strategies in a virtual environment, and use Monte Carlo simulation to evaluate the long-term effects of the strategies;

[0038] Through the closed-loop control system, the control strategy is adjusted online, and the online learning algorithm is integrated to enable the control strategy to learn from each execution and gradually optimize the adjustment logic.

[0039] Furthermore, the communication method of the communication unit includes:

[0040] When the communication unit starts, it performs self-diagnosis. After the self-diagnosis is completed, it dynamically loads the appropriate communication protocol stack according to the preset or remote configuration;

[0041] Use TLS / SSL protocol to establish a secure encryption channel with the remote monitoring center, and implement an identity authentication mechanism that combines device certificates with dynamic tokens, using pre-shared security keys and real-time generated one-time passwords for identity authentication;

[0042] Before uploading data, data is screened and prioritized according to data type and importance, and Huffman coding and Run-Length Encoding are used to compress data. Timestamps, data source identifiers, and integrity check codes are added to each data packet.

[0043] Based on network status monitoring and historical communication quality analysis, the optimal path is dynamically selected to send data, and the backup link is automatically switched if a network failure occurs; the real-time communication protocol is integrated, and when multiple communication units work simultaneously, a distributed load balancing algorithm is used to allocate communication tasks;

[0044] For the received remote control commands, confirmation signals are sent before and after execution, including command reception confirmation and execution result feedback;

[0045] Continuously monitor the status of the communication link and identify abnormal communication patterns based on AI algorithms. If a communication anomaly is detected, immediately trigger the preset recovery process and report the fault details to the monitoring center.

[0046] Furthermore, the adaptive logic control method of the adaptive control logic unit includes:

[0047] Obtain real-time data through the central processing unit, apply online identification algorithms, and update the built-in system model based on real-time data;

[0048] Combining the analysis results of the edge computing unit and the machine learning accelerator, time series analysis is used to predict the future behavior trend of the system;

[0049] Based on the prediction results and system goals, the genetic algorithm is used to find the optimal control strategy. Through the dynamic rule base, the control rule weights are automatically adjusted according to the current system status, historical control effects and newly received remote instructions;

[0050] Integrate the global decision-making of the central processing unit, the fast response decision-making of edge computing, and the optimization strategy of adaptive control logic, and use fuzzy logic, expert system, and deep reinforcement learning methods for comprehensive decision-making;

[0051] According to the decision results, the output path of the control instruction is selected through the edge computing unit or directly through the central processing unit to accurately control the actuator;

[0052] Through the feedback loop, the actual effect data after control execution is collected and compared with the expected target. Based on the comparison and evaluation results, the control logic is retrained using the machine learning accelerator to continuously adjust and optimize the control algorithm parameters to form a closed-loop self-learning mechanism.

[0053] The integrated anomaly detection algorithm monitors abnormal behaviors in the control process in real time. When an anomaly or failure is detected, the preset fault-tolerant strategy is activated.

[0054] A computer program product proposed by the present invention includes a computer program / instruction, and when the computer program / instruction is executed by a processor, the compact PLC controller functional module as described above is implemented.

[0055] Beneficial effects of the present invention: By adopting high-density integrated circuits and system-level packaging technology, the PLC controller realizes the close integration of multi-functional modules, reduces the volume, improves space utilization, and is suitable for various harsh industrial environments; the adaptive data processing and intelligent decision-making functions of the edge computing unit can respond to on-site needs in real time, reduce the burden of the central processing unit, reduce latency, and improve the real-time performance and response speed of the control system; the integration of the machine learning accelerator not only improves the depth and breadth of data analysis, but also enables the system to have advanced functions such as predictive maintenance and production optimization, enhances the system's autonomous decision-making ability, and improves production efficiency and maintenance level; the close collaboration between the adaptive control logic unit and the central processing unit realizes the online optimization and self-learning of the control strategy, so that the system can dynamically adapt to changes in working conditions and the external environment and maintain the optimal operating state. It reduces human intervention and improves the level of automation; the full-chain data protection measures and strict access control of the security encryption unit ensure the safe transmission of data and instructions, effectively prevent external attacks and internal misoperations, and improve the stability and reliability of the system; through dynamic resource scheduling, load balancing and predictive resource allocation, the system can efficiently utilize computing, storage and communication resources, avoid resource bottlenecks, ensure stable operation under high load, and reduce operating costs; the communication unit supports multiple protocols and has powerful data compression and transmission optimization mechanisms, ensuring efficient, secure and reliable data communication, and maintaining good communication quality even in unstable network conditions; the integrated anomaly detection, fault recovery strategy and self-learning mechanism enable the system to respond quickly and self-repair when encountering problems, reducing the risk of downtime and ensuring production continuity and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1This is a schematic diagram of the functional module structure of a compact PLC controller described in the present invention. DETAILED DESCRIPTION

[0057] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0058] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0060] One embodiment of the present invention provides a compact PLC controller functional module, which includes: a data acquisition unit, an edge computing unit, a central processing unit, a data storage unit, a machine learning accelerator, a communication unit, a security encryption unit and an adaptive control logic unit; the output end of the data acquisition unit is connected to the input end of the edge computing unit; the edge computing unit is bidirectionally connected to the central processing unit and the machine learning accelerator; the central processing unit is bidirectionally connected to the data storage unit, the communication unit and the adaptive control logic unit; the security encryption unit is bidirectionally connected to the data acquisition unit, the edge computing unit, the central processing unit, the data storage unit, the machine learning accelerator, the communication unit and the adaptive control logic unit.

[0061] The working principle of the above technical solution is as follows: the data acquisition unit receives raw data (such as temperature, pressure, position information, etc.) from the sensor through the I / O interface, and preprocesses the raw data, including preliminary filtering and standardization; and transmits the preprocessed real-time data to the edge computing unit; the edge computing unit processes the received real-time data and performs basic logical judgment and real-time control, such as emergency stop, basic alarm triggering, etc. At the same time, it identifies key data that need further analysis and forwards it to the machine learning accelerator through an efficient data channel; the machine learning accelerator uses a pre-trained model to perform in-depth analysis on the screened data, such as predictive maintenance prediction and production efficiency optimization suggestions, and feeds back the analysis results to the edge computing unit, and combines the real-time data to perform control strategies. Further optimization; the central processing unit integrates the preliminary decisions from the edge computing unit, the intelligent analysis results of the machine learning accelerator, and the remote instructions, and formulates the final control strategy, and stores the strategy data in the data storage unit; the security encryption unit monitors the data transmission process throughout, ensures the encryption and decryption of all communication content, protects the system from external attacks, and manages the access rights to internal data; the adaptive control logic unit dynamically adjusts the control logic according to the strategy of the central processing unit, and the control instructions are output through the edge computing unit or directly by the central processing unit to drive the actuator (such as motor, valve) to operate and control; the communication unit communicates with the remote monitoring center, uploads system status, alarm information and performance data, and receives remote control instructions or parameter updates.

[0062] The effects of the above technical solution are: directly acquiring data from sensors through the data acquisition unit, and performing preliminary processing and real-time control in the edge computing unit, which reduces the time delay of data transmission to the cloud and improves the control response speed. The use of machine learning accelerators further enhances data analysis capabilities, making the prediction and decision-making process more efficient; using machine learning models to conduct in-depth analysis of data can not only achieve predictive maintenance and detect potential faults in advance, but also optimize production efficiency based on production data and propose improvement measures, thereby reducing maintenance costs and improving overall operational efficiency; the adaptive control logic unit can dynamically adjust the control logic according to the strategy of the central processing unit, so that the system can quickly adapt to changes in production conditions or external environmental interference, and improve the accuracy and adaptability of control; the presence of a secure encryption unit ensures the security of data during transmission and storage, prevents data leakage or malicious tampering, protects the system from external attacks, and manages internal data access rights. , which improves the overall security level of the system; the communication unit allows efficient communication with the remote monitoring center, which can not only upload system status and alarm information in real time, but also accept remote commands or parameter updates, which is convenient for remote centralized management and decision-making, reduces on-site maintenance requirements, and improves management efficiency; the entire system ensures the effective use of computing resources, avoids resource bottlenecks, reduces operating costs, and improves system stability and reliability through intelligent management such as dynamic load balancing and resource allocation adjustment strategies; the machine learning model and adaptive control logic design in the system support online learning and self-optimization, which means that the system can learn from each control execution and continuously evolve the control strategy to better match actual production needs and achieve continuous improvement.

[0063] In one embodiment of the present invention, the PLC controller functional module is packaged using a high-density integrated circuit (HDI) and a system-in-package (SiP) technology.

[0064] The working principle and effect of the above technical solution are as follows: HDI technology greatly reduces the distance between components by increasing the number of circuit board layers and using miniaturized lines, while SiP technology further integrates multiple independent chips and passive components into a single package, greatly reducing the size of the entire controller. This is especially important for space-constrained application scenarios, allowing the device to be designed to be more compact and easy to install in various industrial environments; the increase in integration reduces the distance and time of signal transmission, reduces signal delay, and improves data transmission speed and system responsiveness. In addition, the close connection between components in the system-level package helps to improve overall operating efficiency and reduce energy consumption; SiP technology reduces the probability of failure caused by welding point failure by reducing the number of external connection points and components inside the package, and enhances the stability and durability of the system. At the same time, integrated packaging also improves the system's anti-interference ability, especially in industrial environments with strong electromagnetic interference; highly integrated modules reduce external wiring and assembly steps, simplify system design and manufacturing processes, and reduce production costs and complexity. At the same time, unified packaging standards are also conducive to standardized production and improved production efficiency; the modular design of system-level packaging makes it easier to upgrade specific functions (such as computing power and communication protocols). It only requires replacing or adding the corresponding SiP modules without having to redesign the entire system, reducing the cost and difficulty of maintenance and upgrades; through SiP technology, functional modules originally distributed in different locations (such as processors, storage, and communication modules) can be integrated together to achieve multi-functional integration, thereby improving the comprehensive processing capabilities and application scope of the PLC controller.

[0065] In one embodiment of the present invention, the edge computing method of the edge computing unit includes:

[0066] The edge computing unit receives pre-processed data from the data acquisition unit and dynamically allocates the priority of data processing according to the importance and real-time requirements of the data through an adaptive data stratification algorithm;

[0067] Through the feature extraction algorithm, key features are extracted from the preprocessed data.

[0068] Build a lightweight rule engine to execute basic control instructions according to preset logic, adopt an event-driven architecture, and immediately trigger the corresponding processing flow when a specific event (such as data anomaly, threshold exceeded) is detected;

[0069] Based on the online performance evaluation model, the control instruction execution effect is monitored in real time, the control parameters are automatically fine-tuned, and the machine learning algorithm is used to predict future data traffic and computing needs. Through the dynamic load balancing strategy, the computing resource allocation is dynamically adjusted according to the load of each processing unit in the edge computing unit;

[0070] Based on data characteristics and model requirements, intelligently select data subsets and push them to the machine learning accelerator. Through the model online fusion mechanism, the analysis results of the machine learning accelerator are integrated with the real-time control logic of the edge computing unit to form an accurate control strategy.

[0071] Based on the feedback from the machine learning accelerator, the local decision logic is adjusted to conduct online learning and self-optimization of the control strategy.

[0072] In the transmission stage before and after data processing, it works closely with the security encryption unit and adopts dynamic key management to encrypt data.

[0073] The working principle of the above technical solution is as follows: the edge computing unit first receives the pre-processed data from the data acquisition unit. Then, the adaptive data stratification algorithm is used to dynamically adjust the priority of data processing according to the importance and real-time requirements of the data. This mechanism ensures that critical data can be processed immediately, taking precedence over other non-urgent or secondary data; the data is reduced in dimension through feature extraction algorithms (such as autoencoders, PCA) to extract key features that affect control decisions. This step reduces the amount of calculation, improves the efficiency of data processing, and enables the system to respond faster; the constructed lightweight rule engine performs basic control operations such as emergency stop or alarm triggering according to preset logic. The engine adopts an event-driven architecture. Once a predefined event (such as data anomaly or exceeding the set threshold) is detected, the corresponding processing flow is immediately triggered to achieve a rapid response; the edge computing unit monitors the execution effect of the control instruction in real time, and automatically adjusts the control parameters according to the online performance evaluation model to optimize system performance and resource utilization. At the same time, machine learning is used to predict future data traffic and computing needs, combined with dynamic load balancing strategies, and computing resources are dynamically allocated according to the real-time load of each processing unit to ensure smooth system operation and avoid resource bottlenecks; intelligently identify data characteristics and model requirements, and only select highly correlated data subsets to be transmitted to the machine learning accelerator. Through the model online fusion mechanism, the advanced analysis results of machine learning are combined with the real-time control logic of edge computing to generate a more accurate control strategy. This process promotes the intelligence and accuracy of the control strategy; according to the feedback from the machine learning accelerator, the edge computing unit can adjust its local decision logic to achieve online learning and self-optimization of the control strategy, that is, the system can continuously adjust and improve the control strategy based on new data and experience; during the entire data processing and transmission process, the edge computing unit works closely with the security encryption unit, and uses dynamic key management technology to encrypt the data to ensure the security of the data and prevent the data from being illegally accessed or tampered with during transmission.

[0074] The effects of the above technical solutions are as follows: by dynamically allocating data processing priorities, the immediate processing of key data is ensured, and the real-time response capability of the system is improved, which is crucial for industrial automation, autonomous driving and other fields that require rapid response; by reducing the computing burden through feature extraction algorithms, and intelligently allocating computing resources through dynamic load balancing strategies, resource utilization is optimized to the maximum extent, operating costs are reduced, and efficient and stable operation of the system is maintained; the event-driven architecture enables the system to respond quickly to specific events, and the automatic fine-tuning mechanism and machine learning prediction based on the online performance evaluation model enable the system to self-optimize according to the current state and future trends, enhancing the adaptability and robustness of the system; intelligently selecting data subsets for machine learning analysis, and integrating the analysis results with the real-time control logic to form a more accurate control strategy. This model online fusion mechanism makes control decisions more accurate and improves the performance and reliability of the overall system; dynamic key management is used for encryption during the data transmission stage, which effectively prevents data leakage and unauthorized access, ensures data privacy and security, and complies with strict data protection regulations; through the feedback results of the machine learning accelerator, the local decision logic is continuously adjusted to achieve online learning and self-optimization of the control strategy, which means that the system can continue to evolve and continuously improve its performance and decision-making quality.

[0075] In one embodiment of the present invention, the online performance evaluation model is used to monitor the execution effect of control instructions in real time, automatically fine-tune control parameters, use machine learning algorithms to predict future data traffic and computing requirements, and dynamically adjust the allocation of computing resources according to the load conditions of each processing unit in the edge computing unit through a dynamic load balancing strategy, including:

[0076] According to the characteristics of the control task, a multi-dimensional performance evaluation index system is dynamically constructed. Through the built-in intelligent monitoring system, based on time series analysis and pattern recognition, the changing trend of indicators can be tracked in real time;

[0077] A method combining genetic algorithm and gradient descent method is used to conduct local and detailed search and global optimal exploration of control parameters, and fine-tune the control parameters;

[0078] Based on the reinforcement learning algorithm, the control strategy template is automatically generated or updated according to the current system status and historical optimization experience. The long short-term memory network (LSTM) is used to combine the outputs of multiple prediction models and predict the data flow and computing requirements of the edge computing unit in the future through integrated learning technology.

[0079] Physical computing resources are virtualized to form a resource pool, and virtual resource blocks are dynamically allocated to each task or service. An adaptive load balancing algorithm that integrates task priority, resource consumption, and inter-task dependencies is used to introduce a resource reservation mechanism to ensure resource supply for critical tasks, and a preemption mechanism is designed to schedule resources.

[0080] The working principle of the above technical solution is as follows: according to the characteristics of the control task, a comprehensive performance evaluation index system is dynamically established. This system covers multiple dimensions such as response time, resource utilization, control accuracy and energy consumption. The built-in intelligent monitoring system uses time series analysis and pattern recognition technology to continuously track the changes of these indicators over time, aiming to capture any signs of performance fluctuations or degradation in a timely manner; once performance problems or optimization space are detected, the system starts the parameter optimization process. A combination strategy of genetic algorithm and gradient descent method is adopted here. The genetic algorithm is responsible for searching the possible optimal solution space globally and conducting large-scale exploration, while the gradient descent method focuses on local areas and performs detailed parameter fine-tuning. The combination of the two realizes both fast and accurate control parameter optimization; using reinforcement learning algorithms, especially the model combined with long short-term memory network (LSTM), the system learns how to optimally adjust the control strategy based on the current state and historical data. LSTM is good at capturing long-term dependencies in time series data, and combined with ensemble learning technology, it integrates information from multiple prediction models to more accurately predict future data traffic and computing needs. This not only takes into account the periodicity and trend of the data, but also can handle random factors, making the prediction closer to the actual situation; based on the prediction results, the system virtualizes and manages physical computing resources to form a resource pool. Through the adaptive load balancing algorithm, each task or service dynamically obtains a virtual resource block based on its priority, resource consumption requirements, and dependencies with other tasks. In order to ensure the execution of critical tasks, the system has designed a resource reservation mechanism and introduced a preemption mechanism, which can reallocate resources when resources are tight, give priority to meeting the needs of critical tasks, and ensure the overall stability and efficiency of the system.

[0081] The effect of the above technical solution is: by dynamically constructing a multi-dimensional performance evaluation index system, the system can instantly identify and respond to various performance changes to ensure the efficient execution of control tasks. The optimized control parameters improve the task response time while maintaining high control accuracy, so that the edge computing unit can more accurately meet the application requirements; combining genetic algorithms and gradient descent methods to fine-tune parameters, and fine-tuning resource usage through intelligent monitoring systems, helps to maximize resource utilization and reduce unnecessary energy consumption, which is in line with the trend of green computing; using reinforcement learning and LSTM networks, the system can accurately predict future data traffic and computing needs, and prepare resources in advance. This not only reduces the risk of service interruption caused by improper resource allocation, but also improves the overall service quality and user experience; by virtualizing physical resources and implementing dynamic resource allocation, the system can quickly respond to load changes and reasonably allocate resources for tasks of different priorities. The adaptive load balancing algorithm and preemption mechanism ensure that critical tasks can still obtain necessary resources under resource constraints, ensuring the stable operation of the system; this technical solution reduces the need for manual intervention and reduces operation and maintenance costs through automation and intelligent means. At the same time, by predicting future demand and dynamically adjusting resources, failures caused by insufficient resources can be effectively avoided, thereby enhancing business continuity and stability. The edge computing platform, which has efficient resource management and control strategy optimization capabilities, provides developers and enterprises with strong infrastructure support, lowers the threshold for new service deployment, and promotes technological innovation and rapid iteration of services, which is conducive to the continuous expansion and diversified development of the business.

[0082] In one embodiment of the present invention, the central processing method of the central processing unit includes:

[0083] The central processing unit receives and integrates multi-source data from the edge computing unit, machine learning accelerator, communication unit, and adaptive control logic unit, including real-time control signals, analysis results, remote commands, and system status information. It uses deep learning algorithms to perform pattern recognition and semantic understanding on multi-source data, automatically classify data types, and resolve hidden associations.

[0084] Based on current working conditions, historical data and external environmental factors, a situational awareness model is constructed to dynamically adjust the control strategy generation logic; based on genetic algorithms, the optimal control strategy is generated; the effects of new strategies are pre-simulated in a virtual environment through simulation software to evaluate potential risks and benefits;

[0085] Dynamically schedule computing resources and I / O operations and optimize the order of task execution based on policy execution requirements and system resource status; use machine learning algorithms to predict future resource requirements and automatically adjust the allocation of data storage units, computing resources, and communication bandwidth;

[0086] Work closely with the security encryption unit to implement a multi-level security strategy, including real-time security auditing, intrusion detection and prevention, to ensure the secure transmission of data and instructions, establish a secure connection with the remote monitoring center through the communication unit, perform remote configuration updates, issue instructions and synchronize data, and conduct online learning based on feedback data and remotely updated algorithm models.

[0087] The working principle of the above technical solution is as follows: the central processing unit (CPU) first receives data from different components, such as real-time control signals from edge computing units, analysis results from machine learning accelerators, remote instructions from communication units, and system status information from adaptive control logic units. Deep learning algorithms are used to perform pattern recognition and semantic understanding on these multi-source data, automatically classify data and analyze the associations therein, laying a solid data foundation for subsequent intelligent decision-making. This process improves the intelligence level of data processing and ensures the accuracy of decision-making basis; based on the current actual working conditions, historical data and changes in the external environment, the CPU builds a situational awareness model to dynamically adjust the generation logic of the control strategy. Genetic algorithms are used to explore the control strategy space and find the optimal solution. In order to reduce the risk of implementing new strategies, simulation software is used to preview new strategies in a virtual environment to evaluate their possible impact and ensure the effectiveness and safety of the strategies; according to the specific requirements of the control strategy and the actual situation of the current system resources, the CPU flexibly schedules computing resources and I / O operations, optimizes the task execution order, reduces task waiting time, and speeds up system response. Use machine learning algorithms to predict future resource needs, automatically adjust the allocation of data storage, computing resources and communication bandwidth, achieve optimal resource configuration, and improve system efficiency; collaborate with the security encryption unit to implement comprehensive security protection measures, covering real-time security audits, intrusion detection and defense, to ensure the security of all data and instructions during transmission. Maintain a connection with the remote monitoring center through a secure communication link to achieve remote configuration updates, command issuance and data synchronization. Using feedback data obtained from remote locations and algorithm model updates, the CPU supports online learning, continuously optimizes internal control logic and decision-making algorithms, and promotes system self-evolution to better adapt to changing environments and needs.

[0088] The effect of the above technical solution is: through deep learning algorithms to perform pattern recognition and semantic understanding of multi-source data, the system can automatically classify data and discover hidden associations, thereby providing a more accurate and comprehensive information basis for decision-making. This not only improves the scientific nature of decision-making, but also speeds up decision-making, enabling the system to respond quickly to various working conditions; the situational awareness model and the control strategy generation mechanism based on genetic algorithms allow the system to dynamically adjust the strategy according to the current working conditions, historical data and external environment, ensuring that the control logic always matches the actual production conditions, improving the adaptability and flexibility of the system; dynamic resource scheduling and task optimization reduce the waiting time of computing resources and I / O operations, and improve the overall system response speed. Through machine learning, future resource needs are predicted and resource allocation is automatically adjusted, ensuring efficient use of resources, reducing operating costs and improving processing efficiency. Close cooperation with the security encryption unit and the implementation of multi-level security strategies provide strong security protection for the transmission of data and instructions, effectively preventing data leakage and malicious attacks, and ensuring the stable operation of the system and information security. Through the online learning mechanism, the system can optimize the control logic and decision-making algorithm in real time based on feedback data and remotely updated algorithm models, promoting the continuous evolution of the system and improving long-term performance and stability. The secure remote connection function supports remote configuration updates, command issuance and data synchronization, greatly simplifying the system maintenance and upgrade process, reducing the need for manual intervention, and improving maintenance efficiency and system manageability.

[0089] In one embodiment of the present invention, a situational awareness model is constructed based on current working conditions, historical data and external environmental factors to dynamically adjust the generation logic of the control strategy; a control strategy is generated based on a genetic algorithm; and the effect of the new strategy is pre-simulated in a virtual environment by simulation software to evaluate potential risks and benefits, including:

[0090] Using time series analysis and data fusion technology, the data of edge computing units, machine learning accelerators, communication units, and adaptive control logic units are integrated into a unified platform;

[0091] Use deep neural networks (such as convolutional neural networks (CNN) and recurrent neural networks (RNN)) to automatically extract key features of working conditions;

[0092] Construct a situational awareness model based on fuzzy logic, use genetic algorithms to optimize strategy parameters, and continuously iterate to generate better control strategies by simulating natural selection and genetic mechanisms;

[0093] Build a physics-based high-fidelity simulation environment, pre-execute candidate control strategies in a virtual environment, and use Monte Carlo simulation to evaluate the long-term effects of the strategies, including expected benefits, stability, resource consumption, and potential risks, to ensure the feasibility and reliability of the strategies.

[0094] Through the closed-loop control system, the control strategy is adjusted online, and the online learning algorithm is integrated to enable the control strategy to learn from each execution and gradually optimize the adjustment logic.

[0095] The working principle of the above technical solution is as follows: First, the system integrates data from different sources (such as real-time data from edge computing units, analysis results from machine learning accelerators, command information from communication units, and status reports from adaptive control logic units) into a unified platform through time series analysis and data fusion technology. Deep neural networks (CNN and RNN, etc.) are used to conduct in-depth analysis of the integrated data and automatically extract key features of the working conditions, such as the working status of the equipment, environmental change trends, resource usage patterns, etc. These features become the basis for building a situational awareness model, helping the system understand current and past working conditions and providing detailed basis for the formulation of control strategies. The constructed fuzzy logic situational awareness model has the ability to adjust dynamically, and can automatically adjust model parameters according to changes in real-time working conditions and historical data to ensure the adaptability of the control strategy. Genetic algorithms are further involved to optimize strategy parameters by simulating natural selection and genetic mechanisms, and iteratively generate more efficient and adaptable control strategies. A high-fidelity simulation environment built on the basis of physical principles is used to pre-execute and test these candidate control strategies. Through methods such as Monte Carlo simulation, a comprehensive assessment of the long-term impact of the strategy is conducted, including expected benefits, system stability, resource consumption, and possible risks, to ensure the feasibility and reliability of the selected strategy. Finally, through a closed-loop control system, the system can combine simulation feedback and actual runtime data to adjust the control strategy in real time. The integrated online learning algorithm allows the system to continuously learn and optimize during execution, allowing the control strategy to evolve by itself and adapt to changing and increasingly complex working conditions, ensuring continuous improvement and optimization of system performance.

[0096] The effect of the above technical solution is: through the deep integration of current working conditions, historical data and external environmental factors, the constructed situational awareness model can provide a more accurate basis for decision-making. This comprehensive analysis based on real-time data and historical experience makes the generation logic of the control strategy closer to the actual situation and improves the accuracy of decision-making; the combination of fuzzy logic situational awareness model and genetic algorithm not only realizes the dynamic optimization of control strategy parameters, but also can quickly adjust the strategy according to environmental changes, ensuring the high adaptability and flexibility of the control system, especially in the face of complex and changeable working conditions, can effectively improve system performance; through the high-fidelity simulation environment to pre-execute the control strategy and use Monte Carlo simulation evaluation, it can fully identify and quantify potential risks and benefits before actual application, avoiding the possible losses caused by directly implementing new strategies on the physical system, and ensuring the stability and security of the system; resource consumption In-depth analysis and prediction of patterns help to identify inefficient links, achieve rational allocation and utilization of resources through optimized control strategies, reduce waste, and improve overall operational efficiency and economic benefits; the integration of closed-loop control systems and online learning algorithms enables control strategies to learn and self-adjust from each execution, accelerating the iterative optimization process of strategies, reducing human intervention, and improving the intelligence level of the system. In the long run, it can keep the system in an optimal state in a constantly changing environment; the application of time series analysis and data fusion technology breaks down information silos, promotes data collaboration between different devices and fields, improves the information processing capabilities and response speed of the entire system, and provides a basis for achieving global optimization.

[0097] In one embodiment of the present invention, the communication method of the communication unit includes:

[0098] When the communication unit starts, it performs self-diagnosis. After the self-diagnosis is completed, it dynamically loads the appropriate communication protocol stack according to the preset or remote configuration.

[0099] Use TLS / SSL protocol to establish a secure encryption channel with the remote monitoring center, and implement an identity authentication mechanism that combines device certificates with dynamic tokens, using pre-shared security keys and real-time generated one-time passwords for identity authentication;

[0100] Before uploading data, data is screened and prioritized according to data type and importance, and Huffman coding and Run-Length Encoding are used to compress data. Timestamps, data source identifiers, and integrity check codes are added to each data packet.

[0101] Based on network status monitoring and historical communication quality analysis, the optimal path is dynamically selected to send data, and the backup link is automatically switched if a network failure occurs; real-time communication protocols such as DDS (Data Distribution Service) are integrated, and when multiple communication units work simultaneously, a distributed load balancing algorithm is used to allocate communication tasks;

[0102] For the received remote control commands, confirmation signals are sent before and after execution, including command reception confirmation and execution result feedback;

[0103] Continuously monitor the status of communication links and identify abnormal communication patterns based on AI algorithms, such as increased packet loss rate and sudden delay changes. If communication abnormalities are detected, the preset recovery process will be triggered immediately, including automatic reconnection, link switching or fault isolation, and the fault details will be reported to the monitoring center.

[0104] The working principle of the above technical solution is as follows: after the communication unit is started, it first performs self-diagnosis to ensure that the hardware is normal, the software version is the latest, and the communication interface is available. Based on preset rules or configuration information received from the remote monitoring center, the most suitable communication protocol stack (such as MQTT, OPC UA, etc.) is dynamically loaded to adapt to different network environments and system architectures. This process ensures seamless connection with the remote monitoring system; in order to ensure the security of data transmission, the communication unit uses the TLS / SSL protocol to create an encrypted channel to achieve data confidentiality and integrity protection. Through the dual verification mechanism of device certificates and dynamic tokens, combined with pre-shared keys and one-time passwords, access rights are strictly controlled to prevent illegal intrusion; before uploading data, the communication unit will filter and prioritize the data according to its importance and type, remove redundant information, and use efficient coding technology (such as Huffman coding, Run-Length Encoding) to compress the data to reduce bandwidth usage and speed up transmission. In addition, each data packet will be attached with a timestamp, source identifier and checksum to ensure data integrity and traceability; based on network status monitoring and historical communication data analysis, the communication unit dynamically selects the best path to transmit data, and can automatically switch to the backup link when the main link fails. When multiple communication units work together, a distributed load balancing algorithm is used to reasonably allocate communication tasks, avoid single-point overload, and improve communication efficiency. For the received remote control commands, the communication unit adopts a confirmation receipt mechanism, that is, a confirmation signal is sent after the command is received and after execution, to ensure the accurate execution of the control command. Through the integrated AI algorithm, the communication unit can continuously monitor the link status, promptly identify communication anomalies such as packet loss and delay anomalies, and quickly trigger the recovery process, including automatic reconnection, link switching or fault isolation measures, and report the fault details to the monitoring center to achieve rapid fault response and recovery.

[0105] The effects of the above technical solutions are: through the self-diagnosis mechanism, hardware or software problems can be discovered and solved in advance to ensure that the communication unit is always in good working condition. Dynamic loading of suitable communication protocol stacks and automatic fault recovery processes (such as automatic reconnection and link switching) reduces the risk of communication interruption and improves the stability and reliability of the entire system; the use of TLS / SSL protocol and composite authentication mechanism (device certificate + dynamic token) builds a solid security line of defense, effectively preventing data from being eavesdropped or tampered with, ensuring the confidentiality and integrity of data transmission, and improving the security of the system; through intelligent screening, priority sorting and efficient coding compression of data, it greatly reduces unnecessary data transmission, reduces bandwidth consumption, and speeds up data transmission. This strategy not only saves network resources, but also improves data transmission efficiency, especially in bandwidth-constrained environments; dynamic path selection and distributed load balancing algorithms based on network status monitoring ensure that data can be transmitted through the optimal path, effectively avoiding network congestion and balancing the communication load, especially in scenarios where multiple communication units work together, significantly improving the overall communication efficiency and system response speed; by sending confirmation signals before and after execution, the accurate execution of remote control instructions is ensured, the controllability and operational accuracy of the system are improved, and it helps to achieve more refined remote management and control; the integrated AI algorithm can monitor the status of the communication link in real time, quickly identify abnormal patterns, and automatically trigger the recovery process, reducing the need for manual intervention, shortening the fault handling time, and further enhancing the system's robustness and maintenance efficiency.

[0106] In one embodiment of the present invention, the adaptive logic control method of the adaptive control logic unit includes:

[0107] The central processing unit obtains real-time data and uses online identification algorithms to update the built-in system model based on real-time data.

[0108] Combining the analysis results of the edge computing unit and the machine learning accelerator, time series analysis is used to predict the future behavior trend of the system;

[0109] Based on the prediction results and system goals (such as stability, efficiency, energy consumption, etc.), the genetic algorithm is used to find the optimal control strategy. Through the dynamic rule base, the control rule weights are automatically adjusted according to the current system status, historical control effects and newly received remote instructions;

[0110] Integrate the global decision-making of the central processing unit, the fast response decision-making of edge computing, and the optimization strategy of adaptive control logic, and use fuzzy logic, expert system, and deep reinforcement learning methods for comprehensive decision-making;

[0111] According to the decision results, the output path of the control instruction is selected through the edge computing unit or directly through the central processing unit to accurately control the actuator;

[0112] Through the feedback loop, the actual effect data after control execution, including control error, system response time, etc., is collected and compared with the expected target. Based on the comparison and evaluation results, the control logic is retrained using the machine learning accelerator, and the control algorithm parameters are continuously adjusted and optimized to form a closed-loop self-learning mechanism.

[0113] The integrated anomaly detection algorithm monitors abnormal behaviors in the control process in real time. When an anomaly or failure is detected, the preset fault-tolerant strategy is activated.

[0114] The working principle of the above technical solution is as follows: real-time data is collected from sensors and equipment through the central processing unit (CPU). These data cover process variables, equipment status and environmental parameters, etc., to form a comprehensive understanding of the current system status. Using online identification algorithms, these data are used to update and optimize the system model (whether linear, nonlinear or fuzzy model) to ensure that the model can accurately capture the real dynamic characteristics of the system and provide a basis for the formulation of subsequent control strategies; combining the fast data analysis capabilities of the edge computing unit with the efficient processing of the machine learning accelerator, time series analysis is used to predict the future state change trend of the system. Based on these prediction results and the preset system goals (such as improving stability, efficiency or reducing energy consumption), the genetic algorithm is used to search the control strategy space to find the control strategy that best suits the current situation. By dynamically adjusting the weights of the control rules, the online fine-tuning of the control strategy is achieved to cope with changes in the system status; integrating the global perspective decision of the central processing unit, the instant response capability of edge computing and the deep optimization strategy of the adaptive control logic, complex decision-making methods such as fuzzy logic, expert system and deep reinforcement learning are used to make multi-level and comprehensive judgments to determine the best control action. Subsequently, the control instructions are output to the actuator through the most appropriate path (edge ​​computing unit or directly by the CPU), such as dynamically adjusting the control parameters through PID self-tuning or sliding mode control, to achieve precise control of the system; through the closed-loop feedback system, the actual effect data after the control execution is collected, including key indicators such as control error and response time, and compared and evaluated with the predetermined goals. Based on the evaluation results, the control logic is retrained using the machine learning accelerator, and the control algorithm parameters are continuously optimized iteratively to form a control logic system that continuously self-learns and evolves; the integrated anomaly detection algorithm monitors the control process in real time. Once abnormal behaviors such as failure to execute control instructions and abnormal data fluctuations are found, the preset fault-tolerant strategy is immediately activated, such as switching to the backup control logic or executing the emergency shutdown procedure, to ensure the safe and stable operation of the system even in abnormal situations.

[0115] The effect of the above technical solution is: by updating the system model through real-time data collection and online identification algorithm, this method can quickly adapt to environmental changes and equipment status changes, ensuring that the control strategy always matches the actual dynamic characteristics of the system. The integrated anomaly detection and fault tolerance mechanism further enhances the system's ability to recover in the face of faults or abnormal situations, ensuring the stable operation of the system; using genetic algorithms to find the optimal control strategy, combining time series analysis to predict future trends, and dynamically adjusting the weights of control rules, the control strategy is more accurate and effective, thereby improving the overall performance of the system. This is not only reflected in the improvement of control accuracy, but also includes increased efficiency and reduced energy consumption; the application of edge computing units and machine learning accelerators shortens the time for data processing and decision-making, allowing the system to respond quickly to external changes, which is particularly important for scenarios that require instant control, such as industrial automation, autonomous driving, etc.; collecting execution effect data through feedback loops, and using machine learning to retrain the control logic, forming a closed-loop self-learning mechanism. This means that the system can continuously learn from operations, automatically adjust and optimize control algorithms, and the control logic will become more mature and complete over time; the integration of the global vision of central processing, the rapid response of edge computing, and advanced decision-making methods such as deep reinforcement learning make control decisions more comprehensive and accurate, able to handle complex and changing control scenarios and improve decision-making quality; anomaly detection algorithms and preset fault-tolerant strategies ensure that when control command failures or data anomalies occur, the system can quickly take measures, such as switching to backup control logic or executing emergency protection procedures, to effectively prevent the situation from deteriorating and ensure personnel safety and minimize property losses.

[0116] One embodiment of the present invention is a computer program product, including a computer program / instruction, characterized in that when the computer program / instruction is executed by a processor, it implements any of the compact PLC controller function modules described above.

[0117] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A compact PLC controller function module, characterized in that: The functional module includes: a data acquisition unit, an edge computing unit, a central processing unit, a data storage unit, a machine learning accelerator, a communication unit, a security encryption unit and an adaptive control logic unit; the output end of the data acquisition unit is connected to the input end of the edge computing unit; the edge computing unit is bidirectionally connected to the central processing unit and the machine learning accelerator; the central processing unit is bidirectionally connected to the data storage unit, the communication unit and the adaptive control logic unit; the security encryption unit is bidirectionally connected to the data acquisition unit, the edge computing unit, the central processing unit, the data storage unit, the machine learning accelerator, the communication unit and the adaptive control logic unit.

2. The compact PLC controller functional module according to claim 1, characterized in that: The PLC controller functional module adopts high-density integrated circuits and is packaged through system-level packaging technology.

3. The compact PLC controller functional module according to claim 1, characterized in that: The working method of the functional module includes: The data acquisition unit receives raw data from the sensor through the I / O interface, preprocesses the raw data, and transmits the preprocessed real-time data to the edge computing unit; The edge computing unit processes the received real-time data and performs basic logical judgment and real-time control. At the same time, it identifies key data that needs further analysis and forwards it to the machine learning accelerator through an efficient data channel. The machine learning accelerator uses pre-trained models to conduct in-depth analysis of the filtered data and feeds the analysis results back to the edge computing unit to further optimize the control strategy in combination with real-time data; The central processing unit synthesizes the preliminary decisions from the edge computing unit, the intelligent analysis results from the machine learning accelerator, and the remote instructions, formulates the final control strategy, and stores the strategy data in the data storage unit; The security encryption unit monitors the entire data transmission process and manages access rights to internal data; The adaptive control logic unit dynamically adjusts the control logic according to the strategy of the central processing unit. The control instructions are output through the edge computing unit or directly by the central processing unit to drive the actuator to act and perform control; The communication unit communicates with the remote monitoring center, uploading system status, alarm information and performance data, and receiving remote control instructions or parameter updates.

4. The compact PLC controller functional module according to claim 1, characterized in that: The edge computing method of the edge computing unit includes: The edge computing unit receives pre-processed data from the data acquisition unit and dynamically allocates the priority of data processing according to the importance and real-time requirements of the data through an adaptive data stratification algorithm; Through feature extraction algorithm, key features are extracted from the preprocessed data; Build a lightweight rule engine to execute basic control instructions according to preset logic. Adopt an event-driven architecture. When a specific event is detected, the corresponding processing flow is immediately triggered. Based on the online performance evaluation model, the control instruction execution effect is monitored in real time, the control parameters are automatically fine-tuned, and the machine learning algorithm is used to predict future data traffic and computing needs. Through the dynamic load balancing strategy, the computing resource allocation is dynamically adjusted according to the load of each processing unit in the edge computing unit; Based on data characteristics and model requirements, intelligently select data subsets and push them to the machine learning accelerator. Through the model online fusion mechanism, the analysis results of the machine learning accelerator are integrated with the real-time control logic of the edge computing unit to form an accurate control strategy. Based on the feedback from the machine learning accelerator, the local decision logic is adjusted to conduct online learning and self-optimization of the control strategy. In the transmission stage before and after data processing, it works closely with the security encryption unit and adopts dynamic key management to encrypt data.

5. The compact PLC controller functional module according to claim 4, characterized in that: Based on the online performance evaluation model, the control instruction execution effect is monitored in real time, the control parameters are automatically fine-tuned, and the future data traffic and computing requirements are predicted using machine learning algorithms. Through the dynamic load balancing strategy, the computing resource allocation is dynamically adjusted according to the load of each processing unit in the edge computing unit, including: According to the characteristics of the control task, a multi-dimensional performance evaluation index system is dynamically constructed. Through the built-in intelligent monitoring system, based on time series analysis and pattern recognition, the changing trend of indicators can be tracked in real time; A method combining genetic algorithm and gradient descent method is used to conduct local and detailed search and global optimal exploration of control parameters, and fine-tune the control parameters; Based on the reinforcement learning algorithm, the control strategy template is automatically generated or updated according to the current system status and historical optimization experience. The long short-term memory network is used to combine the output of multiple prediction models, and the data flow and computing requirements of the edge computing unit in the future are predicted through integrated learning technology. Physical computing resources are virtualized to form a resource pool, and virtual resource blocks are dynamically allocated to each task or service. An adaptive load balancing algorithm that integrates task priority, resource consumption, and inter-task dependencies is used to introduce a resource reservation mechanism, and a preemption mechanism is designed to schedule resources.

6. The compact PLC controller functional module according to claim 1, characterized in that: The central processing method of the central processing unit comprises: The central processing unit receives and integrates multi-source data from the edge computing unit, machine learning accelerator, communication unit, and adaptive control logic unit, uses deep learning algorithms to perform pattern recognition and semantic understanding on multi-source data, automatically classifies data types, and analyzes hidden associations; Based on current working conditions, historical data and external environmental factors, a situational awareness model is constructed to dynamically adjust the control strategy generation logic; based on genetic algorithms, control strategies are generated; simulation software is used to pre-simulate the effects of new strategies in a virtual environment to evaluate potential risks and benefits; Dynamically schedule computing resources and I / O operations and optimize the order of task execution based on policy execution requirements and system resource status; use machine learning algorithms to predict future resource requirements and automatically adjust the allocation of data storage units, computing resources, and communication bandwidth; Work closely with the security encryption unit to implement a multi-level security strategy; establish a secure connection with the remote monitoring center through the communication unit to perform remote configuration updates, command issuance and data synchronization, and conduct online learning based on feedback data and remotely updated algorithm models.

7. The compact PLC controller functional module according to claim 6, characterized in that: Based on the current working conditions, historical data and external environmental factors, a situational awareness model is constructed to dynamically adjust the generation logic of the control strategy; Generate control strategies based on genetic algorithms; Use simulation software to pre-simulate the effects of new strategies in a virtual environment and evaluate potential risks and benefits, including: Using time series analysis and data fusion technology, the data of edge computing units, machine learning accelerators, communication units, and adaptive control logic units are integrated into a unified platform; Use deep neural networks to automatically extract key features of working conditions; Construct a situational awareness model based on fuzzy logic, use genetic algorithms to optimize strategy parameters, and continuously iterate to generate better control strategies by simulating natural selection and genetic mechanisms; Build a physics-based high-fidelity simulation environment, pre-execute candidate control strategies in a virtual environment, and use Monte Carlo simulation to evaluate the long-term effects of the strategies; Through the closed-loop control system, the control strategy is adjusted online, and the online learning algorithm is integrated to enable the control strategy to learn from each execution and gradually optimize the adjustment logic.

8. The compact PLC controller functional module according to claim 1, characterized in that: The communication method of the communication unit includes: When the communication unit starts, it performs self-diagnosis. After the self-diagnosis is completed, it dynamically loads the appropriate communication protocol stack according to the preset or remote configuration; Use TLS / SSL protocol to establish a secure encryption channel with the remote monitoring center, and implement an identity authentication mechanism that combines device certificates with dynamic tokens, using pre-shared security keys and real-time generated one-time passwords for identity authentication; Before uploading data, data is screened and prioritized according to data type and importance, and Huffman coding and Run-Length Encoding are used to compress data. Timestamps, data source identifiers, and integrity check codes are added to each data packet. Based on network status monitoring and historical communication quality analysis, the optimal path is dynamically selected to send data, and the backup link is automatically switched if a network failure occurs; the real-time communication protocol is integrated, and when multiple communication units work simultaneously, a distributed load balancing algorithm is used to allocate communication tasks; For the received remote control commands, confirmation signals are sent before and after execution, including command reception confirmation and execution result feedback; Continuously monitor the status of the communication link and identify abnormal communication patterns based on AI algorithms. If a communication anomaly is detected, immediately trigger the preset recovery process and report the fault details to the monitoring center.

9. The compact PLC controller functional module according to claim 1, characterized in that: The adaptive logic control method of the adaptive control logic unit comprises: Obtain real-time data through the central processing unit, apply online identification algorithms, and update the built-in system model based on real-time data; Combining the analysis results of the edge computing unit and the machine learning accelerator, time series analysis is used to predict the future behavior trend of the system; Based on the prediction results and system goals, the genetic algorithm is used to find the optimal control strategy. Through the dynamic rule base, the control rule weights are automatically adjusted according to the current system status, historical control effects and newly received remote instructions; Integrate the global decision-making of the central processing unit, the fast response decision-making of edge computing, and the optimization strategy of adaptive control logic, and use fuzzy logic, expert system, and deep reinforcement learning methods for comprehensive decision-making; According to the decision results, the output path of the control instruction is selected through the edge computing unit or directly through the central processing unit to accurately control the actuator; Through the feedback loop, the actual effect data after control execution is collected and compared with the expected target. Based on the comparison and evaluation results, the control logic is retrained using the machine learning accelerator to continuously adjust and optimize the control algorithm parameters to form a closed-loop self-learning mechanism. The integrated anomaly detection algorithm monitors abnormal behaviors in the control process in real time. When an anomaly or failure is detected, the preset fault-tolerant strategy is activated.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the compact PLC controller functional module as described in any one of claims 1-9 is implemented.

Images