Multi-stage embedded control equipment state sensing and energy cascade scheduling system

Abstract

The invention provides a multi-stage embedded control equipment state sensing and energy cascade scheduling system. Comprising a master control decision center module, a distributed edge embedded node module, an equipment full-dimension state sensing module, an energy dynamic optimization scheduling module, a fault prediction and self-healing control module, a cross-protocol communication interconnection module and a man-machine cooperative command module. According to the invention, through constructing a three-layer time domain control chain of edge node nanosecond-level signal processing, cloud second-level optimization scheduling and equipment hour-level strategy presetting, seamless cooperation of turbine bearing pedestal micro-vibration monitoring and a power grid peak regulation strategy is realized, and real-time wavelet noise reduction preprocessing of embedded nodes is combined with cloud LSTM life prediction. A taboo search algorithm is driven to dynamically reconstruct a power supply scheme, and the pain point of control response lag in a high-fluctuation scene is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment control, and in particular to a multi-level embedded control equipment state perception and energy cascade scheduling system. Background Art

[0002] In a boiler production environment, the current industrial equipment status monitoring and energy dispatching system suffers from three major structural flaws: First, millisecond-level anomalies, such as sudden changes in drum water level and steam pressure fluctuations, must be transmitted through multiple levels of the DCS system, resulting in safety valve actuation delays exceeding 300ms. For example, in a false water level condition, while the embedded node can quickly shut down the feedwater pump, the cloud cannot immediately adjust the coal consumption, exacerbating the risk of drum pressure fluctuations. Second, the superheater tube wall overtemperature prediction system operates independently, and the pipe creep warning it outputs cannot automatically adjust the combustion intensity. When a local temperature exceeds the limit, the blower continues to operate at the original load until carbonization of the tube wall triggers a shutdown. Third, the boiler room coexists with Modbus RTU (legacy instrumentation), PROFIBUS (smart valves), and EtherCAT (new sensor) protocols. The energy center needs to customize an OPC gateway to integrate data, delaying the generation of load dispatch instructions by more than 15 minutes, resulting in a mismatch between steam production and heat demand.

[0003] Therefore, to address the above problems, a multi-level embedded control device status perception and energy cascade scheduling system is being developed. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, the present invention provides a multi-level embedded control device state perception and energy cascade scheduling system.

[0005] The technical solution of the present invention is: a multi-level embedded control device state perception and energy cascade scheduling system, including:

[0006] Main control decision-making center module: As the core processing unit of the system, it integrates a multi-threaded real-time operating system, coordinates the data flow and instruction distribution of all modules, and implements dynamic resource allocation strategies;

[0007] Distributed edge embedded node module: An embedded controller group deployed at the physical device layer, directly connected to field devices via an industrial bus, performing millisecond-level device status acquisition and issuing underlying control instructions;

[0008] Full-dimensional device status perception module: Builds a three-layer perception system, namely the physical layer sensor network, edge layer real-time preprocessing, and cloud-based deep modeling, integrating multi-source heterogeneous data such as temperature, vibration, current, and electromagnetic fields;

[0009] Energy dynamic optimization scheduling module: implements a three-order time scale scheduling strategy, namely nanosecond emergency response, second-level operation optimization, and hour-level strategy planning, and dynamically adjusts energy allocation based on the status of related equipment;

[0010] Fault prediction and self-healing control module: Generates a fault probability matrix based on the equipment degradation model and real-time state deviation, triggering the self-healing control sequence in the plan library;

[0011] Cross-protocol communication interconnection module: Adopts a layered communication architecture, namely the field layer TSN network, the edge layer OPC UA gateway, and the cloud-side MQTT agent, to achieve lossless transparent transmission of control instructions and status data;

[0012] Human-machine collaborative command module: provides a three-dimensional visual operation and maintenance interface, supporting operator intervention in scheduling strategy modification and emergency takeover control;

[0013] The system operation includes a three-stage closed loop: the edge node collects device status data in real time to the main control center, integrates and analyzes and generates a scheduling strategy to the embedded node to execute energy control instructions. At the same time, the fault prediction module continuously verifies the system stability, and the human-machine module dynamically adjusts the control parameter threshold.

[0014] As a preferred embodiment of the present invention, the device full-dimensional state perception module is characterized by executing the following refined process:

[0015] Physical sensing layer deployment: triaxial vibration sensors are installed in the axial, radial, and tangential directions of the turbine bearing seat. A temperature sensor array is deployed on the surface of the superheater tube panel. The drum water level gauge uses a differential pressure transmitter and radar dual-mode redundant detection. High-frequency current transformers are coupled to power equipment cables. Thin-film temperature array sensors are mounted on the surface of heating components. The sensor network adopts a star-tree hybrid topology to ensure signal integrity.

[0016] Edge preprocessing process: The embedded node performs five steps on the original signal: adaptive wavelet noise reduction to eliminate electromagnetic interference; FFT spectrum analysis to extract characteristic frequency band energy; peak-to-peak value and kurtosis index calculation of time domain signal; spatiotemporal alignment of multi-sensor data to compensate for transmission delay; and generation of standardized device health indicator vectors.

[0017] Cloud-based deep modeling process: After the main control center receives edge pre-processed data: establish a digital twin model of the equipment to simulate the stress distribution under load mutation conditions; apply long short-term memory networks to analyze historical HHI-Vector sequences and predict the remaining life curves of key components; calculate the coupling coefficients of associated equipment states based on the equipment topology relationship diagram to identify systemic risks; output equipment status level labels, namely normal, decay, critical, fault, and confidence assessment reports, and finally generate an equipment status portrait that is transmitted to the scheduling module.

[0018] As a preferred embodiment of the present invention, the energy dynamic optimization scheduling module is characterized by achieving three-level energy efficiency optimization:

[0019] Level 1: Millisecond-level transient response scheduling. When the fault prediction module issues an overload warning, the power switching sequence is immediately initiated: cut off the power supply to non-critical loads → start the heat storage tank to instantly release heat energy to compensate for the steam shortage, start the backup water supply pump in parallel → parallel redundant power supply to initiate seamless switching → record the voltage fluctuation spectrum of each node during the transient process;

[0020] Level 2: Minute-level operation optimization and scheduling, periodically executing energy cost minimization strategies: collecting real-time energy efficiency mapping tables for devices; building a multi-objective optimization model: using device priority coefficients as constraints and total system energy consumption as the objective function; using a tabu search algorithm to solve the optimal power allocation solution, for example: reducing the power of standby devices to a maintenance threshold, shifting the load curve of high-efficiency devices; and adjusting the motor speed curve through group control of inverters using the OPC UA protocol.

[0021] Level 3: Strategic energy planning and scheduling, pre-allocating resources based on production plans: analyzing the production task list for the next 24 hours; combining the peak and valley period of electricity prices with the photovoltaic output curve; dynamically planning the charging and discharging time window of energy storage equipment: charging to 80% capacity during valley periods, and releasing energy storage during peak periods to reduce grid power purchases; generating equipment start-up and shutdown plan matrices to control the warm-up start-up time of large units. The scheduling results are synchronized to the edge node through the communication module for execution, and the actual energy consumption is compared with the predicted deviation in real time.

[0022] As a preferred embodiment of the present invention, the fault prediction and self-healing control module is characterized by performing fault lifecycle management:

[0023] Real-time fault diagnosis mechanism: Continuously receives device state vectors from the state perception module and determines anomalies using a dual verification model: (1) Physical rule model: Verifies whether the harmonic components in the vibration spectrum exceed the material fatigue threshold; (2) Data-driven model: Convolutional neural network identifies abnormal hot spot patterns in the temperature field distribution. The diagnostic results are linked to the equipment failure case library in the knowledge graph;

[0024] Predictive maintenance triggering process: When the equipment degradation level increases: extract MTBF data of similar equipment from historical maintenance records; simulate the degradation trajectory of key components based on the Wiener process; reversely calculate the remaining safe operating time and generate maintenance time window recommendations; and link the scheduling module to limit the maximum load rate of the equipment;

[0025] Self-healing control sequence execution: After the fault is confirmed, a four-level response is initiated: in the first level, the local embedded node executes the preset protection action; in the second level, the scheduling module isolates the energy supply of the faulty equipment; in the third level, redundant equipment is activated and the load is redistributed; in the fourth level, the maintenance work order is uploaded to the human-machine module to generate a three-dimensional disassembly guidance map; the control process implements dual-channel monitoring: the main control center verifies the integrity of the instruction, and the edge node feedbacks the execution status code.

[0026] As a preferred embodiment of the present invention, the cross-protocol communication interconnection module is characterized by layered transmission protocol conversion:

[0027] Edge layer communication architecture: The field device layer uses the CANopen bus to transmit raw sensor data. The protocol conversion chip embedded in the embedded node converts CAN frames into EtherCAT frames. The time-sensitive network switch is used to achieve microsecond-level jitter control of control instructions. The transmission path implements a dual-ring redundant topology.

[0028] Wide-area communication pipeline: Triple channels are deployed between edge nodes and the cloud control platform: optical fiber dedicated lines transmit key control instructions; 4G / 5G wireless network backup transmits status monitoring data; and LoraWAN network transmits low-power device status. Data packets are embedded with time stamps to ensure timing consistency.

[0029] Cloud data governance: After receiving data, the main control center executes: OPC UA Pub / Sub agent filters duplicate messages; MQTT Broker distributes data according to QoS level: QoS0 level transmits historical records, QoS1 level transmits scheduling instructions, and QoS2 level transmits fault alarms; the communication anomaly detector monitors the link status: when the packet loss rate is >5% for three consecutive cycles, the transmission path is automatically switched. All cross-domain communications use AES-256-GCM end-to-end encryption and additional device digital signatures.

[0030] As a preferred embodiment of the present invention, the human-machine collaborative command module is characterized by providing a deep interaction function:

[0031] 3D Visualization Engine: Renders 3D equipment models in real time and overlays physical parameters: thermal maps of motor winding temperature fields, animations showing real-time flow rates in pipelines, and dynamic current flow indications for distribution networks. Key parameters are displayed using spectral color mapping, with blue indicating normal, yellow indicating warning, and red indicating fault.

[0032] Collaborative decision-making mechanism: Operators can perform four types of interventions: drag the timeline to replay historical operating conditions; manually delineate energy-saving areas on the topology map, and the system automatically generates optimization plans for these areas; modify equipment operation priority coefficients to trigger scheduling policy recalculation; insert emergency maintenance plans and simultaneously freeze related equipment scheduling instructions;

[0033] Alarm management strategy: When the fault prediction module outputs a high-risk alarm: an audible and visual alarm sequence is initiated and pushed to the mobile terminal; a 3D disassembly view of the faulty equipment is automatically displayed in a pop-up window; historical processing cases are pushed through the maintenance knowledge base; operator response time is recorded and emergency response efficiency is evaluated; an audit log is generated for all manual operations, and key control instructions require secondary password confirmation.

[0034] As a preferred embodiment of the present invention, the distributed edge embedded node module is characterized by hardware-level control logic:

[0035] Onboard control program: The MCU equipped with a real-time operating system executes four tasks: μs-level interrupt service routine to collect sensor raw signals; ms-level task to process data preprocessing algorithm; 10ms-level task to verify the validity of the main control instruction; scheduled task to upload local cache data;

[0036] Safety protection mechanism: The node has three built-in protection layers: electrical isolation, with optocoupler relays isolating the 220VAC high-voltage control circuit; logic monitoring, with a watchdog chip monitoring program operation status; and boundary protection, with preset device parameter safety intervals. When the limit is exceeded, the power supply is immediately cut off and the main control instructions are ignored.

[0037] Dynamic configuration function: Receives policy scripts issued by the master control through the communication module to achieve online update of device sampling frequency; resets control parameter thresholds; switches to the backup control algorithm library, and initiates the neighbor takeover protocol when a node is abnormal: adjacent nodes temporarily take over the devices within the jurisdiction of the faulty node through daisy-chain communication.

[0038] As a preferred embodiment of the present invention, the collaborative operation process of the system includes:

[0039] Step 1: The embedded node self-checks sensor connectivity, then the main control center loads the device topology map. The energy module then initializes the grid / energy storage / load status matrix, and finally the human-machine module creates a visualization scene.

[0040] Step 2: The edge node uploads a snapshot of the device status every 50ms. The master control center integrates the data and updates the digital twin. The fault prediction module scans for abnormal points. The energy scheduler allocates power according to the preset strategy, and control instructions are issued through an encrypted channel.

[0041] Step 3: When the communication delay exceeds 10ms, the edge node activates the local cache strategy: comparing the current device status with the preset safety template. If the vibration amplitude exceeds 120% of the template value, load reduction is performed and the communication redundant link is reconnected.

[0042] Step 4: Perform an energy efficiency audit at the end of each day's production: compare the actual dispatch curve with the theoretical optimization model, revise the equipment energy efficiency coefficient, and update the load priority rule base. An audit trail is generated for all operational events, and key control steps are recorded on the blockchain for evidence.

[0043] The present invention has the following advantages: 1. The present invention realizes seamless coordination of micro-vibration monitoring of turbine bearing seats and peak-shaving strategy of power grid by constructing a three-layer time domain control chain of nanosecond-level signal processing of edge nodes, second-level optimization scheduling on the cloud, and hour-level strategy preset of equipment. The real-time wavelet noise reduction preprocessing of embedded nodes is combined with LSTM life prediction on the cloud to drive the taboo search algorithm to dynamically reconstruct the power supply plan, thereby solving the pain point of delayed control response in high-fluctuation scenarios.

[0044] 2. This invention establishes a closed-loop linkage mechanism between fault prediction and energy scheduling. When the vibration spectrum exceeds the material fatigue threshold, the system automatically triggers triple protection: load reduction to preserve equipment life, flywheel energy storage to compensate for power mutations, and optocoupler relays to physically isolate the fault source. The four-level self-healing control chain makes the equipment's emergency stop response speed dozens of times faster than traditional solutions, avoiding secondary disasters caused by overload. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a structural schematic diagram of the present invention.

[0046] Figure 2 This is a detailed flowchart executed by the full-dimensional state perception module of the device of the present invention.

[0047] Figure 3 This is a schematic diagram of the three-level optimization of the energy dynamic optimization scheduling module of the present invention.

[0048] Figure 4 It is a flow chart of the operating steps of the present invention. DETAILED DESCRIPTION

[0049] Reference herein to an embodiment means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present invention. The appearance of such a phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] This system coordinates six core functions through the main control decision-making center module: the distributed edge embedded node module collects multi-source data of equipment temperature / vibration / current with millisecond accuracy, and performs triple processing (physical layer signal noise reduction → edge layer feature extraction → cloud-based LSTM life prediction) through the equipment full-dimensional state perception module to generate a device health status vector and transmit it to the fault prediction and self-healing control module; this module calculates the remaining life based on the Wiener process degradation model, triggers a four-level self-healing mechanism (local emergency stop → energy isolation → load transfer → maintenance guidance) and links the energy dynamic optimization scheduling module to implement three-level energy efficiency regulation (nanosecond-level flywheel energy storage to compensate for power gaps → second-level taboo search algorithm to optimize load distribution → hour-level optical storage and charging coordination strategy); the entire process relies on cross-protocol communication and interconnection modules to build a layered transmission channel (TSN network guarantees microsecond-level instruction delivery → optical fiber / 5G dual-channel redundancy → OPC UA-MQTT protocol conversion), and finally the human-machine collaborative command module dynamically renders the equipment status heat map through the 3D visualization engine, supporting operators to modify scheduling thresholds in real time, replay historical working conditions, and intervene in emergency control, forming a "perception-prediction-scheduling-execution-feedback" closed loop.

[0051] When the system starts, the burner edge node uploads the drum water level pulse signal (1kHz sampling rate) in real time through the TSN network → the communication module converts the superheater Modbus temperature data into a time-standardized PROFINET frame → the state perception module completes the drum water level wavelet noise reduction and steam pressure spectrum analysis at the edge node, while the cloud digital twin simulates the tube screen stress distribution under the condition of sudden change in coal calorific value → the main control center integrates the drum oscillation coefficient and the tube wall creep prediction curve. When the water level change rate is detected to be greater than 10mm / s or the local temperature is greater than 450℃ → the fault prediction module calculates the remaining safety The system generates a load reduction maintenance window based on the time (e.g., the 3-hour lifespan of an overheated pipe section) and immediately limits the airflow to the faulty area to 75%. Using a tabu search algorithm, the system redistributes the steam load (reducing the output of adjacent boilers by 5% and releasing energy from the heat storage tank to compensate). If the prediction module issues a red pipe burst alarm, the self-healing control module, directly connected to the edge node, triggers a three-level response: 0.5s fast closing of the main steam valve (hard cutoff of the optocoupler relay), emergency water replenishment to prevent dry boil, and a 300% flow rate override for the desuperheating water. The synchronized human-machine interface displays a 3D thermal map of the superheater burst point and a database of similar boiler maintenance cases, allowing operators to manually correct the combustion curve. The system dynamically verifies the water level fluctuation band (±15mm) and air-fuel ratio every 50ms. If the TSN packet loss rate exceeds 1%, it automatically switches to the 5G channel, achieving full-time coordinated control from nanosecond-level sensing (water level fluctuations) to hourly-level strategies (heat storage tank scheduling).

[0052] Specifically, such as Figure 1As shown in the figure, the multi-level embedded control device state perception and energy cascade scheduling system includes a main control decision-making center module, a distributed edge embedded node module, a full-dimensional device state perception module, an energy dynamic optimization scheduling module, a fault prediction and self-healing control module, a cross-protocol communication interconnection module and a human-machine collaborative command module. The main control decision-making center module serves as the core processing unit of the system, integrates a multi-threaded real-time operating system, coordinates the data flow and instruction distribution of all modules, and executes dynamic resource allocation strategies. The distributed edge embedded node module is deployed in the embedded controller group of the physical device layer, directly connected to the field equipment through the industrial bus, and performs millisecond-level device state acquisition and After underlying control commands are issued, the full-dimensional device status perception module constructs a three-layer perception system: a physical-layer sensor network, real-time preprocessing at the edge layer, and deep modeling at the cloud. This system integrates heterogeneous data from multiple sources, including temperature, vibration, current, and electromagnetic fields. The dynamic energy optimization and scheduling module implements a three-order timescale scheduling strategy: nanosecond-level emergency response, second-level operational optimization, and hour-level strategy planning. It dynamically adjusts energy allocation based on the status of associated devices. The fault prediction and self-healing control module generates a fault probability matrix based on device degradation models and real-time status deviations, triggering self-healing control sequences within the pre-plan library. The cross-protocol communication and interconnection module utilizes a layered communication architecture: a field-layer TSN network, an edge-layer OPC UA gateway, and a cloud-based MQTT proxy. This enables lossless transmission of control commands and status data. The human-machine collaborative command module provides a three-dimensional visual operation and maintenance interface, supporting operator intervention, scheduling strategy modification, and emergency takeover control. System operation consists of a three-stage closed loop: edge nodes collect real-time device status data and transmit it to the main control center. This data is integrated and analyzed to generate a scheduling strategy, which is then sent to embedded nodes to execute energy control commands. Simultaneously, the fault prediction module continuously verifies system stability, and the human-machine module dynamically adjusts control parameter thresholds.

[0053] Device full-dimensional status perception module, such as Figure 2 As shown, the following refined processes are performed in the specific process:

[0054] Physical sensing layer deployment: triaxial vibration sensors are installed in the axial, radial, and tangential directions of the turbine bearing seat. A temperature sensor array is deployed on the surface of the superheater tube panel. The drum water level gauge uses a differential pressure transmitter and radar dual-mode redundant detection. High-frequency current transformers are coupled to power equipment cables. Thin-film temperature array sensors are mounted on the surface of heating components. The sensor network adopts a star-tree hybrid topology to ensure signal integrity.

[0055] Edge preprocessing process: The embedded node performs five steps on the original signal: adaptive wavelet noise reduction to eliminate electromagnetic interference; FFT spectrum analysis to extract characteristic frequency band energy; peak-to-peak value and kurtosis index calculation of time domain signal; spatiotemporal alignment of multi-sensor data to compensate for transmission delay; and generation of standardized device health indicator vectors.

[0056] Cloud-based deep modeling process: After the main control center receives edge pre-processed data: establish a digital twin model of the equipment to simulate the stress distribution under load mutation conditions; apply long short-term memory networks to analyze historical HHI-Vector sequences and predict the remaining life curves of key components; calculate the coupling coefficients of associated equipment states based on the equipment topology relationship diagram to identify systemic risks; output equipment status level labels, namely normal, decay, critical, fault, and confidence assessment reports, and finally generate an equipment status portrait that is transmitted to the scheduling module.

[0057] Energy dynamic optimization scheduling module, such as Figure 3 As shown, it is used to achieve the following three-level energy efficiency optimization:

[0058] Level 1: Millisecond-level transient response scheduling. When the fault prediction module issues an overload warning, the power switching sequence is immediately initiated: cut off the power supply to non-critical loads → start the heat storage tank to instantly release heat energy to compensate for the steam shortage, start the backup water supply pump in parallel → parallel redundant power supply to initiate seamless switching → record the voltage fluctuation spectrum of each node during the transient process;

[0059] Level 2: Minute-level operation optimization and scheduling, periodically executing energy cost minimization strategies: collecting real-time energy efficiency mapping tables for devices; building a multi-objective optimization model: using device priority coefficients as constraints and total system energy consumption as the objective function; using a tabu search algorithm to solve the optimal power allocation solution, for example: reducing the power of standby devices to a maintenance threshold, shifting the load curve of high-efficiency devices; and adjusting the motor speed curve through group control of inverters using the OPC UA protocol.

[0060] Level 3: Strategic energy planning and scheduling. This pre-allocates resources based on production plans. The system analyzes the production task list for the next 24 hours, combines the peak and valley electricity price period with the predicted photovoltaic output curve, and dynamically plans the charging and discharging time windows for energy storage equipment: charging to 80% capacity during valley electricity price periods and releasing energy storage during peak electricity price periods to reduce grid power purchases. A matrix of equipment start and stop schedules is generated to control the warm-up start-up time of large units. Scheduling results are synchronized to edge nodes via a communication module for execution, and actual energy consumption is compared against predicted deviations in real time.

[0061] The fault prediction and self-healing control module performs fault lifecycle management:

[0062] Real-time fault diagnosis mechanism: Continuously receives device state vectors from the state perception module and determines anomalies using a dual verification model: (1) Physical rule model: Verifies whether the harmonic components in the vibration spectrum exceed the material fatigue threshold; (2) Data-driven model: Convolutional neural network identifies abnormal hot spot patterns in the temperature field distribution. The diagnostic results are linked to the equipment failure case library in the knowledge graph;

[0063] Predictive maintenance triggering process: When the equipment degradation level increases: extract MTBF data of similar equipment from historical maintenance records; simulate the degradation trajectory of key components based on the Wiener process; reversely calculate the remaining safe operating time and generate maintenance time window recommendations; and link the scheduling module to limit the maximum load rate of the equipment;

[0064] Self-healing control sequence execution: After the fault is confirmed, a four-level response is initiated: in the first level, the local embedded node executes the preset protection action; in the second level, the scheduling module isolates the energy supply of the faulty equipment; in the third level, redundant equipment is activated and the load is redistributed; in the fourth level, the maintenance work order is uploaded to the human-machine module to generate a three-dimensional disassembly guidance map; the control process implements dual-channel monitoring: the main control center verifies the integrity of the instruction, and the edge node feedbacks the execution status code.

[0065] The layered transmission protocol conversion of the cross-protocol communication interconnection module is as follows:

[0066] Edge layer communication architecture: The field device layer uses the CANopen bus to transmit raw sensor data. The protocol conversion chip embedded in the embedded node converts CAN frames into EtherCAT frames. The time-sensitive network switch is used to achieve microsecond-level jitter control of control instructions. The transmission path implements a dual-ring redundant topology.

[0067] Wide-area communication pipeline: Triple channels are deployed between edge nodes and the cloud control platform: optical fiber dedicated lines transmit key control instructions; 4G / 5G wireless network backup transmits status monitoring data; and LoraWAN network transmits low-power device status. Data packets are embedded with time stamps to ensure timing consistency.

[0068] Cloud data governance: After receiving data, the main control center executes: OPC UA Pub / Sub agent filters duplicate messages; MQTT Broker distributes data according to QoS level: QoS0 level transmits historical records, QoS1 level transmits scheduling instructions, and QoS2 level transmits fault alarms; the communication anomaly detector monitors the link status: when the packet loss rate is >5% for three consecutive cycles, the transmission path is automatically switched. All cross-domain communications use AES-256-GCM end-to-end encryption and additional device digital signatures.

[0069] The human-machine collaborative command module provides deep interactive functions:

[0070] 3D Visualization Engine: Renders 3D equipment models in real time and overlays physical parameters: thermal maps of motor winding temperature fields, animations showing real-time flow rates in pipelines, and dynamic current flow indications for distribution networks. Key parameters are displayed using spectral color mapping, with blue indicating normal, yellow indicating warning, and red indicating fault.

[0071] Collaborative decision-making mechanism: Operators can perform four types of interventions: drag the timeline to replay historical operating conditions; manually delineate energy-saving areas on the topology map, and the system automatically generates optimization plans for these areas; modify equipment operation priority coefficients to trigger scheduling policy recalculation; insert emergency maintenance plans and simultaneously freeze related equipment scheduling instructions;

[0072] Alarm management strategy: When the fault prediction module outputs a high-risk alarm: an audible and visual alarm sequence is initiated and pushed to the mobile terminal; a 3D disassembly view of the faulty equipment is automatically displayed in a pop-up window; historical processing cases are pushed through the maintenance knowledge base; operator response time is recorded and emergency response efficiency is evaluated; an audit log is generated for all manual operations, and key control instructions require secondary password confirmation.

[0073] The hardware-level control logic of the distributed edge embedded node module is:

[0074] Onboard control program: The MCU equipped with a real-time operating system executes four tasks: μs-level interrupt service routine to collect sensor raw signals; ms-level task to process data preprocessing algorithm; 10ms-level task to verify the validity of the main control instruction; scheduled task to upload local cache data;

[0075] Safety protection mechanism: The node has three built-in protection layers: electrical isolation, with optocoupler relays isolating the 220VAC high-voltage control circuit; logic monitoring, with a watchdog chip monitoring program operation status; and boundary protection, with preset device parameter safety intervals. When the limit is exceeded, the power supply is immediately cut off and the main control instructions are ignored.

[0076] Dynamic configuration function: Receives policy scripts issued by the master control through the communication module to achieve online update of device sampling frequency; resets control parameter thresholds; switches to the backup control algorithm library, and initiates the neighbor takeover protocol when a node is abnormal: adjacent nodes temporarily take over the devices within the jurisdiction of the faulty node through daisy-chain communication.

[0077] like Figure 4 As shown in the figure, the collaborative operation process steps of the system are as follows:

[0078] Step 1: The embedded node self-checks sensor connectivity, then the main control center loads the device topology map. The energy module then initializes the grid / energy storage / load status matrix, and finally the human-machine module creates a visualization scene.

[0079] Step 2: The edge node uploads a snapshot of the device status every 50ms. The master control center integrates the data and updates the digital twin. The fault prediction module scans for abnormal points. The energy scheduler allocates power according to the preset strategy, and control instructions are issued through an encrypted channel.

[0080] Step 3: When the communication delay exceeds 10ms, the edge node activates the local cache strategy: comparing the current device status with the preset safety template. If the vibration amplitude exceeds 120% of the template value, load reduction is performed and the communication redundant link is reconnected.

[0081] Step 4: Perform an energy efficiency audit at the end of each day's production: compare the actual dispatch curve with the theoretical optimization model, revise the equipment energy efficiency coefficient, and update the load priority rule base. An audit trail is generated for all operational events, and key control steps are recorded on the blockchain for evidence.

[0082] The above embodiments are provided to persons familiar with the art for implementing or using the present invention. Personnel familiar with the art may make various modifications or changes to the above embodiments without departing from the spirit of the present invention. Therefore, the scope of protection of the present invention is not limited to the above embodiments, but should be the maximum scope of the innovative features mentioned in the claims.

Claims

1. A multi-level embedded control device state perception and energy cascade scheduling system, characterized by: It includes: Main control decision-making center module: as the core processing unit of the system, it integrates a multi-threaded real-time operating system, coordinates the data flow and instruction distribution of all modules, and executes dynamic resource allocation strategies; Distributed edge embedded node module: An embedded controller group deployed at the physical device layer, directly connected to field devices via an industrial bus, performing millisecond-level device status acquisition and issuing underlying control instructions; Full-dimensional device status perception module: Builds a three-layer perception system, namely the physical layer sensor network, edge layer real-time preprocessing, and cloud-based deep modeling, integrating multi-source heterogeneous data such as temperature, vibration, current, and electromagnetic fields; Energy dynamic optimization scheduling module: implements a three-order time scale scheduling strategy, namely nanosecond emergency response, second-level operation optimization, and hour-level strategy planning, and dynamically adjusts energy allocation based on the status of related equipment; Fault prediction and self-healing control module: Generates a fault probability matrix based on the equipment degradation model and real-time state deviation, triggering the self-healing control sequence in the plan library; Cross-protocol communication interconnection module: adopts a layered communication architecture, namely the field layer TSN network, the edge layer OPC UA gateway, and the cloud-side MQTT agent to achieve lossless transparent transmission of control instructions and status data; Human-machine collaborative command module: provides a three-dimensional visual operation and maintenance interface, supporting operator intervention in scheduling strategy modification and emergency takeover control; The system operation includes a three-stage closed loop: the edge node collects device status data in real time to the main control center, integrates and analyzes and generates a scheduling strategy to the embedded node to execute energy control instructions. At the same time, the fault prediction module continuously verifies the system stability, and the human-machine module dynamically adjusts the control parameter threshold.

2. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the device full-dimensional state perception module is characterized in that Perform the following refinement process: Physical sensing layer deployment: triaxial vibration sensors are installed in the axial, radial, and tangential directions of the turbine bearing seat. A temperature sensor array is deployed on the surface of the superheater tube panel. The drum water level gauge uses a differential pressure transmitter and radar dual-mode redundant detection. High-frequency current transformers are coupled to power equipment cables. Thin-film temperature array sensors are mounted on the surface of heating components. The sensor network adopts a star-tree hybrid topology to ensure signal integrity. Edge pre-processing process: The embedded node performs five steps of processing on the original signal: adaptive wavelet noise reduction to eliminate electromagnetic interference; FFT spectrum analysis to extract the energy of characteristic frequency bands; Calculate peak-to-peak and kurtosis indicators for time-domain signals; align multi-sensor data in time and space to compensate for transmission delays; and generate standardized device health indicator vectors. Cloud-based deep modeling process: After the main control center receives edge pre-processed data: establish a digital twin model of the equipment to simulate the stress distribution under load mutation conditions; apply long short-term memory networks to analyze historical HHI-Vector sequences and predict the remaining life curves of key components; calculate the coupling coefficients of associated equipment states based on the equipment topology relationship diagram to identify systemic risks; output equipment status level labels, namely normal, decay, critical, fault, and confidence assessment reports, and finally generate an equipment status portrait that is transmitted to the scheduling module.

3. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the energy dynamic optimization scheduling module is characterized in that Achieve three-level energy efficiency optimization: Level 1: Millisecond-level transient response scheduling. When the fault prediction module issues an overload warning, the power switching sequence is immediately initiated: cut off the power supply to non-critical loads → start the heat storage tank to instantly release heat energy to compensate for the steam shortage, start the backup water supply pump in parallel → parallel redundant power supply to initiate seamless switching → record the voltage fluctuation spectrum of each node during the transient process; Level 2: Minute-level operation optimization and scheduling, periodically executing energy cost minimization strategies: collecting real-time energy efficiency mapping tables for devices; building a multi-objective optimization model: using device priority coefficients as constraints and total system energy consumption as the objective function; using a tabu search algorithm to solve the optimal power allocation solution, for example: reducing the power of standby devices to a maintenance threshold, shifting the load curve of high-efficiency devices; and adjusting the motor speed curve through group control of inverters using the OPC UA protocol. Level 3: Strategic energy planning and scheduling. This pre-allocates resources based on production plans. The system analyzes the production task list for the next 24 hours, combines the peak and valley electricity price period with the predicted photovoltaic output curve, and dynamically plans the charging and discharging time windows for energy storage equipment: charging to 80% capacity during valley electricity price periods and releasing energy storage during peak electricity price periods to reduce grid power purchases. A matrix of equipment start and stop schedules is generated to control the warm-up start-up time of large units. Scheduling results are synchronized to edge nodes via a communication module for execution, and actual energy consumption is compared against predicted deviations in real time.

4. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the fault prediction and self-healing control module is characterized in that Perform incident lifecycle management: Real-time fault diagnosis mechanism: Continuously receives device state vectors from the state perception module and determines anomalies using a dual verification model: (1) Physical rule model: Verifies whether the harmonic components in the vibration spectrum exceed the material fatigue threshold; (2) Data-driven model: Convolutional neural network identifies abnormal hot spot patterns in the temperature field distribution. The diagnostic results are linked to the equipment failure case library in the knowledge graph; Predictive maintenance triggering process: When the equipment degradation level increases: extract MTBF data of similar equipment from historical maintenance records; simulate the degradation trajectory of key components based on the Wiener process; reversely calculate the remaining safe operating time and generate maintenance time window recommendations; The linkage scheduling module limits the maximum load rate of the device; the self-healing control sequence is executed: after the fault is confirmed, a four-level response is initiated: in the first level, the local embedded node executes the preset protection action; At the second level, the dispatch module isolates the energy supply of the faulty equipment; At the third level, redundant equipment is activated and the load is redistributed. At the fourth level, maintenance work orders are uploaded to the human-machine module to generate a three-dimensional disassembly guide map. The control process implements dual-path monitoring: the main control center verifies the integrity of the instructions, and the edge node feedbacks the execution status code.

5. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the cross-protocol communication interconnection module is characterized in that Layered transport protocol conversion: Edge layer communication architecture: The field device layer uses the CANopen bus to transmit raw sensor data. The protocol conversion chip embedded in the embedded node converts CAN frames into EtherCAT frames. The time-sensitive network switch is used to achieve microsecond-level jitter control of control instructions. The transmission path implements a dual-ring redundant topology. Wide-area communication pipeline: Triple channels are deployed between edge nodes and the cloud control platform: optical fiber dedicated lines transmit key control instructions; 4G / 5G wireless network backup transmits status monitoring data; and LoraWAN network transmits low-power device status. Data packets are embedded with time stamps to ensure timing consistency. Cloud data governance: After receiving data, the main control center executes: OPC UA Pub / Sub agent filters duplicate messages; MQTT Broker distributes data according to QoS level: QoS0 level transmission history, QoS1 level transmission scheduling instructions, QoS2 level transmission fault alarm; communication anomaly detector monitors link status: when the packet loss rate is >5% for three consecutive cycles, the transmission path is automatically switched. All cross-domain communications use AES-256-GCM end-to-end encryption and additional device digital signatures.

6. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the human-machine collaborative command module is characterized in that Provides deep interactive functions: 3D visualization engine: Real-time rendering of equipment 3D models and superimposition of physical parameters: rendering of motor winding temperature field thermal map, pipeline fluid animation display of real-time flow rate, and dynamic current flow direction indication of distribution network. Key parameters are displayed in a spectral color spectrum map, i.e. blue for normal, yellow for warning and red for fault; Collaborative decision-making mechanism: Operators can perform four types of interventions: drag the timeline to replay historical operating conditions; manually delineate energy-saving areas on the topology map, and the system automatically generates optimization plans for the areas; Modify the equipment operation priority coefficient to trigger the recalculation of the scheduling strategy; insert the emergency maintenance plan and freeze the relevant equipment scheduling instructions simultaneously; Alarm management strategy: When the fault prediction module outputs a high-risk alarm: an audible and visual alarm sequence is initiated and pushed to the mobile terminal; a pop-up window automatically displays a 3D disassembly view of the faulty device; Linked maintenance knowledge base pushes historical processing cases; records operator response time and evaluates emergency efficiency; generates audit logs for all manual operations; and key control instructions require secondary password confirmation.

7. Based on the multi-level embedded control device state perception and energy cascade scheduling system according to claim 1, the distributed edge embedded node module is characterized in that Hardware-level control logic: Onboard control program: The MCU equipped with a real-time operating system executes four-level tasks: μs-level interrupt service routines collect sensor raw signals; ms-level tasks process data preprocessing algorithms; 10ms-level tasks verify the validity of master control instructions; scheduled tasks upload local cache data; Safety protection mechanism: The node has three built-in protection layers: electrical isolation, with optocoupler relays isolating the 220VAC high-voltage control circuit; logic monitoring, with a watchdog chip monitoring program operation status; and boundary protection, with preset device parameter safety intervals. When the limit is exceeded, the power supply is immediately cut off and the main control instructions are ignored. Dynamic configuration function: Receives policy scripts issued by the master control through the communication module to achieve online update of device sampling frequency; resets control parameter thresholds; switches to the backup control algorithm library, and initiates the neighbor takeover protocol when a node is abnormal: adjacent nodes temporarily take over the devices within the jurisdiction of the faulty node through daisy-chain communication.

8. The device state perception and energy cascade scheduling system based on the multi-level embedded control according to claim 1 is characterized in that: The collaborative operation process of the system includes: Step 1: The embedded node self-checks sensor connectivity, then the main control center loads the device topology map. The energy module then initializes the grid / energy storage / load status matrix, and finally the human-machine module creates a visualization scene. Step 2: The edge node uploads a snapshot of the device status every 50ms. The master control center integrates the data and updates the digital twin. The fault prediction module scans for abnormal points. The energy scheduler allocates power according to the preset strategy, and control instructions are issued through an encrypted channel. Step 3: When the communication delay exceeds 10ms, the edge node activates the local cache strategy: comparing the current device status with the preset safety template. If the vibration amplitude exceeds 120% of the template value, load reduction is performed and the communication redundant link is reconnected. Step 4: Perform an energy efficiency audit at the end of each day's production: compare the actual dispatch curve with the theoretical optimization model, revise the equipment energy efficiency coefficient, and update the load priority rule base. An audit trail is generated for all operational events, and key control steps are recorded on the blockchain for evidence.

Images