Intelligent layered distributed control method and system for coaxial shaft engine

Abstract

The application discloses a kind of coaxial shafting engine intelligent layered distributed control method and system, to solve the problem of slow control response, poor fuel adaptability and safety linkage lag caused by high-speed shafting strong coupling.The method comprises: performing system self-checking calibration, driving shafting to establish self-sustaining operation, loading to steady state and performing speed fuel double closed loop, double flow path thermal management, multi-fuel self-adaptation and efficiency optimization, and triggering hardware level protection when abnormal.The system comprises: multi-core lockstep MCU decision layer, FPGA fast execution layer and sensor executor network;MCU is responsible for logical decision and optimization, FPGA is responsible for hard real-time control and physical level emergency stop.The application separates decision and execution logic through layered architecture, can realize hard real-time stable control under high-speed working condition and multi-fuel variable working condition disturbance-free switching, while ensuring material safety, significantly improves system cycle efficiency and operation and maintenance intelligent level.

Description

Technical Field

[0001] This invention belongs to the field of engine control technology, specifically, it relates to an intelligent hierarchical distributed control method and system for coaxial shaft engine. Background Technology

[0002] High-speed gas turbine propulsion systems have broad application value in energy conversion fields such as vehicle range extenders, distributed power generation, and micro gas turbines. These systems achieve closed-loop energy conversion through complex thermodynamic cycles and mechanical transmissions. Their control performance directly determines the overall power output characteristics and operational safety of the machine, making it a core research direction in modern power engineering.

[0003] Among these technologies, coaxial engine control is crucial for ensuring stable system operation. It aims to achieve dynamic balance under high-speed conditions through the coordinated control of coaxially coupled components such as the gas motor, turbine, and compressor. This technological direction focuses on the real-time performance of speed regulation, the stability of multi-fuel combustion, and the optimization of overall system thermal efficiency, forming the technological foundation for achieving high-performance power output.

[0004] Existing technologies suffer from several shortcomings in handling high-speed coaxial shaft control. First, traditional control architectures struggle to cope with strong torque coupling and inertial disturbances between shafts, resulting in poor power balance at extremely high speeds. Furthermore, software-level response delays cannot meet the demands of hard real-time control under transient conditions, easily leading to system instability, oscillations, or even stalling. Second, existing systems lack the ability to adaptively adjust to multiple fuel components, making it difficult to automatically optimize efficiency while ensuring combustion stability. The lack of multi-level safety interlocks and data-driven health management mechanisms results in delayed protection responses to dangerous conditions such as overheating and vibration, hindering the scientific prediction of remaining equipment lifespan. Additionally, traditional control logic exhibits mutual exclusion in coordinating thermal management safety and system cycle efficiency, making it difficult to achieve an optimal balance between waste heat recovery and material protection through dynamic diversion. These issues severely limit the reliability and intelligence of coaxial shaft engines in complex application scenarios. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent hierarchical distributed control method and system for coaxial shaft system engines, which mainly solves the problems of slow control response, poor fuel adaptability and delayed safety linkage caused by strong coupling in high-speed shaft systems.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for intelligent hierarchical distributed control of a coaxial shaft engine includes the following steps: S1, perform hardware logic architecture configuration verification, establish signal compensation model, and verify the mechanical position accuracy and response hysteresis of the actuator by driving the fuel valve, air distribution valve and vent valve to perform full-stroke reciprocating motion; S2 monitors the shaft system engine speed and load feedback in real time and adjusts the frequency of the drive motor in a closed loop. After establishing a stable airflow field in the combustion chamber, it opens the ignition excitation unit and fuel valve. After determining that the system has entered the self-sustaining operation state, it cuts off the external drive power supply. S3, increase the fuel valve opening to drive the shaft engine to accelerate, adjust the turbine guide vanes or bypass valve to make the system output power tend to the rated target value, and run the speed control inner loop to suppress transient disturbances caused by load fluctuations; S4 utilizes dual closed-loop control of speed and fuel to maintain power balance. Based on thermodynamic parameters, it establishes an extreme value optimization model for shaft engine efficiency. Using the output of the extreme value optimization model, it applies a disturbance signal to adjust the split ratio and combustion chamber pressure rise ratio, driving the system state to converge towards the highest efficiency range. S5 monitors the vibration amplitude, shaft trajectory, combustion status, combustion temperature and differential pressure parameters of the shaft engine in real time. When abnormal operating conditions are detected, it triggers the protection program to achieve fuel cut-off, rotor support protection and rapid shutdown. When weak combustion and flameout are detected, it performs auxiliary ignition and combustion repair.

[0007] Furthermore, in S1, the hardware logic architecture configuration verification is achieved by starting the boot program through the multi-core lockstep MCU decision layer and performing read / write tests on internal registers and external memory, and by loading a preset bitstream file using the FPGA fast execution layer.

[0008] Furthermore, in S4, the shaft engine efficiency extreme value optimization model includes power balance scheduling, dual-flow path thermal management regulation, multi-fuel characteristic adaptive correction, and extreme value optimization of system cycle efficiency. The dual-flow path thermal management regulation includes high-pressure path control and merging path control. The high-pressure path control monitors the outlet pressure and inlet flow of the volumetric compressor and calculates the real-time pressure ratio through the FPGA fast execution layer. When the operating point approaches the surge line, it drives the high-speed return valve to perform anti-surge protection. The merging path control distributes the flow between the cooling branch and the heat exchanger branch through a three-way proportional regulating valve. The multi-core lockstep MCU decision layer performs weighted calculations based on combustion chamber wall temperature data and turbine inlet temperature data, dynamically allocating the flow ratio through the cooling branch and the heat exchanger branch. When the combustion chamber wall temperature approaches the material's allowable limit, the flow ratio of the cooling branch is increased; when the temperature is within a safe range, the airflow is guided to the heat exchanger branch.

[0009] Furthermore, the multi-fuel characteristic adaptive correction supports multiple fuels including compressed natural gas, gasoline, hydrogen, and methanol. A fuel composition sensor installed on the fuel supply pipeline identifies the fuel type and component ratio in real time. The multi-core lockstep MCU decision layer automatically switches the characteristic parameter map preset in memory based on the identification results. During fuel switching, a smooth switching algorithm completes the transition from one fuel characteristic map to another, and the injection pressure, ignition timing, and fuel valve flow characteristic curves are corrected in real time based on combustion chamber pressure pulsation data, exhaust oxygen content, and temperature. To ensure smooth power delivery during multi-fuel switching, a transition compensation stage based on virtual torque is added to the algorithm. This stage calculates the torque difference between different fuels at the same flow rate and instructs the FPGA fast execution layer to offset the torque difference by instantaneously adjusting the injection pressure.

[0010] Furthermore, the FPGA fast execution layer integrates an electromagnetic bearing control module, which monitors the rotor's shaft trajectory in real time through a displacement sensor and uses active damping and notch filtering algorithms to suppress rotor vibration when it exceeds the critical speed. The active damping algorithm generates electromagnetic force compensation commands by calculating the first and second derivatives of the rotor displacement, and the digital notch filter is used to lock and filter out vibration components near the speed frequency. When the shaft system passes through the critical speed region, the FPGA fast execution layer dynamically adjusts the stiffness and damping parameters. When shaft system instability, system power failure, or vibration amplitude exceeding the preset safety limit is detected, the rotor is forcibly triggered to drop to the auxiliary bearing, and a fuel cut-off command is issued simultaneously.

[0011] Furthermore, the protection program in S5 is triggered by the hardware emergency stop logic built into the FPGA fast execution layer. This logic is implemented by hardware combinational circuits and is not affected by software interrupt nesting. When the triggering condition is met, the system completes the interlock protection action within a preset time threshold. The interlock protection action includes: closing the fuel shut-off valve to cut off the energy input, opening the vent valve to release the system pressure, driving the speed increaser to trip to isolate the load, and opening the bypass valve to change the airflow path.

[0012] Furthermore, the present invention also includes health monitoring and predictive maintenance steps: the FPGA fast execution layer extracts shaft vibration spectrum features, combustion chamber pressure pulsation frequency, and pressure ratio fluctuation features, and uploads the processed feature vector to the multi-core lockstep MCU decision layer; the multi-core lockstep MCU decision layer runs a lightweight neural network model to complete fault mode diagnosis. The input layer of the neural network model receives speed, temperature, pressure, vibration features, and fuel flow data, the hidden layer performs feature fusion on the data through a nonlinear activation function, and the output layer provides a health index score; the multi-core lockstep MCU decision layer predicts the remaining service life based on the system's cumulative runtime, thermal cycle count, and the performance degradation slope of key components.

[0013] Furthermore, the method supports automatic switching between multiple modes, including start-up mode, idle mode, loading mode, steady-state mode, unloading mode, normal shutdown mode, and emergency shutdown mode. The transition logic between modes is controlled by a state machine within the multi-core lockstep MCU decision layer, which determines the switching conditions based on speed, temperature, pressure, and external control commands. The sensor network layer adopts a redundant configuration, with key sensors having multi-channel backups. The multi-core lockstep MCU decision layer compares multi-channel signals in real time, switching to the backup channel and recording the fault code when a single-point signal fails.

[0014] Furthermore, the multi-core lockstep MCU decision layer sends the processed, de-identified operating data to the cloud optimization platform through the cloud optimization interface to obtain the control parameters and characteristic maps optimized by swarm intelligence analysis, and then transmits them back to the multi-core lockstep MCU decision layer via remote firmware upgrade to achieve online updates of the control strategy; in the high-speed flash memory mounted on the multi-core lockstep MCU decision layer, the system records the operating parameters cyclically at preset time intervals, and locks the data when a shutdown event occurs for subsequent fault backtracking analysis.

[0015] The present invention also provides an intelligent hierarchical distributed control system for a coaxial shaft engine, used to execute the above method, the system comprising: The multi-core lockstep MCU decision layer uses multiple symmetrical processor cores to execute the same instruction stream and monitors the consistency of execution results in real time through a hardware comparator. It is used to handle non-real-time operation mode management, system-level power allocation strategy, overall thermal efficiency optimization algorithm, and parameter scheduling logic during multi-fuel switching. The FPGA fast execution layer is connected to the multi-core lockstep MCU decision layer through a high-speed serial communication link. Its internal logic is constructed by hardware gate circuits to drive the high-speed input / output interface, pulse width modulation module and electromagnetic bearing actuator, and undertakes the inner loop logic of speed control, active vibration control of electromagnetic bearing and hardware emergency stop protection. The sensor network layer is equipped with an electromagnetic speed sensor, a thermocouple temperature sensor, a piezoelectric pressure sensor, a fuel composition analyzer, and an electromagnetic bearing displacement sensor, which are used to sense the full-dimensional operating status of the system and transmit the signals to the FPGA fast execution layer. The actuator network layer includes a fuel valve driver, an airflow distribution valve, a venting valve stepper motor control module, an ignition excitation unit, an electromagnetic bearing power amplifier, and an external drive motor controller, which are used to receive control commands and drive the corresponding mechanical components to perform actions.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention adopts a hierarchical distributed control architecture of multi-core lockstep MCU decision layer and FPGA fast execution layer, decoupling and separating non-real-time global decision logic and hard real-time control logic. It can not only rely on the multi-core computing power of MCU to realize the stable operation of complex algorithms such as global efficiency optimization and multi-mode scheduling, but also realize microsecond-level control response through the parallel hardware logic of FPGA. It effectively eliminates the transient disturbance caused by the strong coupling characteristics of high-speed shaft system, and significantly improves the operating stability and control reliability of the system under high speed conditions.

[0017] (2) This invention constructs a multi-fuel characteristic adaptive adjustment mechanism and a full-level safety interlocking system, which can automatically match the corresponding control parameter spectrum based on the fuel component identification results, and achieve seamless switching of different fuel types by combining the torque transition compensation algorithm. At the same time, it realizes microsecond-level safety response under abnormal operating conditions through the pure hardware emergency shutdown logic built into the FPGA. Combined with the health prediction model based on feature extraction and lightweight neural network, it can realize early warning of faults and scientific assessment of remaining service life, greatly improving the fuel applicability range and operational safety of the system.

[0018] (3) The present invention designs a dual-flow-path dynamic thermal management adjustment mechanism, which can dynamically adjust the flow distribution ratio of the cooling branch and the heat exchanger branch based on real-time monitoring data of combustion chamber wall temperature and turbine inlet temperature. When the combustion chamber wall temperature is in a safe range, the waste heat recovery efficiency is maximized, and when the wall temperature is close to the allowable limit of the material, the safety of the components is prioritized. This effectively solves the mutual exclusion problem between thermal safety protection and cycle efficiency improvement in traditional control logic, and realizes the synergistic optimization of system operation safety and energy conversion efficiency.

[0019] (4) The present invention is equipped with a redundant sensor network and a black box recording mechanism for operation data, which can achieve seamless switching when a single sensor fails, ensuring the continuity of the control process. At the same time, it relies on the cloud optimization interface to realize the desensitized uploading of operation data and the online iteration of control strategies. It can continuously optimize the control effect by combining group operation data and adapt to the usage needs of more complex application scenarios. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the core principle framework of the hierarchical collaborative control between the MCU unit decision layer and the FPGA execution layer in this invention; Figure 3 This is a flowchart of the method control in this invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments. The embodiments of the present invention include, but are not limited to, the following embodiments.

[0022] like Figure 1 , Figure 2 As shown, this invention discloses an intelligent hierarchical distributed control system for a coaxial shaft engine, whose hardware architecture adopts a hierarchical and decoupled distributed design. The system mainly consists of a multi-core lockstep MCU decision layer, an FPGA fast execution layer, a sensor network layer, an actuator network layer, and a cloud optimization platform. The multi-core lockstep MCU decision layer, as the core brain of the entire system, integrates at least two symmetrical processor cores. These two cores run the same instruction sequence at the hardware level and monitor the instruction execution results in real time through a built-in comparator. If the calculation results of the two cores deviate, the system will immediately trigger an exception interrupt handler, thereby ensuring the absolute reliability of the control commands during the generation stage. This decision layer communicates bidirectionally with the external cloud optimization platform via a high-speed Ethernet interface, enabling it to receive control mapping tables and proportional-integral-derivative control parameter sets trained with large amounts of data from the cloud in real time. Simultaneously, the decision layer also establishes a physical connection with the FPGA fast execution layer through a high-speed parallel bus or serial peripheral interface, disseminating complex global management strategies to the execution layer.

[0023] The FPGA fast execution layer is responsible for high-frequency real-time computing and hardware-level security in the system. Internally, it utilizes a hardware description language to construct multiple independent parallel processing modules, including a speed closed-loop control module, an electromagnetic bearing active vibration control module, a high-speed signal sampling and processing module, and a hardware emergency stop logic circuit. Due to the parallel hardware structure of the FPGA, its response time to the control loop can reach the microsecond level, which is crucial for power systems like coaxial shaft engines with extremely high rotor speeds and extremely low rotational inertia. The input of the fast execution layer connects to the sensor network layer, while its output directly drives the actuator network layer.

[0024] The sensor network layer is distributed across various key components of the shaft system engine. Electromagnetic speed sensors are installed on the shaft system to capture the instantaneous angular velocity of the rotor. Multiple sets of displacement sensors are evenly spaced around the bearing housings of the electromagnetic bearings to acquire two-dimensional displacement signals of the rotor in the radial and axial directions, as well as the shaft center trajectory. Thermocouples and pressure transmitters are distributed along the thermodynamic circulation path, such as at the combustion chamber inlet and outlet, and turbine inlet, to monitor the thermodynamic state of the airflow. Furthermore, infrared spectral fuel composition sensors are specifically installed on the fuel supply pipeline to identify the molecular structure of the fuel within the pipeline in real time. The analog or digital signals generated by these sensors are all collected at the analog / digital conversion front-end of the fast execution layer.

[0025] The actuator network layer contains all the key actuators controlling engine operation. The fuel valve, installed on the main fuel supply line, precisely controls the mass flow rate of fuel entering the combustion chamber by adjusting its valve core opening. A three-way proportional control valve, located downstream of the compressor assembly, connects its outlet to the cooling passage of the combustion chamber and the heat exchange path of the regenerator, responsible for distributing compressed air flow. The ignition exciter is connected to the ignition electrode in the combustion chamber, providing ignition energy during startup. The power amplifier of the electromagnetic bearing receives commands from the rapid actuator layer, providing controlled current to the bearing coil to generate levitation force. An external drive motor is connected to the shaft system via a coupling, providing initial rotational kinetic energy to the system during startup. A vent valve, installed at the compressor outlet, releases high-pressure air during shutdown or surge conditions.

[0026] like Figure 3 As shown, based on the above system, the control method of the present invention is as follows: During the system power-on initialization boot phase, the multi-core lockstep MCU decision layer first performs a bit-by-bit comparison of the startup code execution streams of the two symmetrical cores through an internal bus monitor to ensure the hardware integrity of the computing environment. Simultaneously, the FPGA fast execution layer loads a preset bitstream file and performs a full address space read / write traversal test on the external high-speed static random access memory. After the hardware self-test passes, the system enters the actuator calibration phase. The decision layer instructs the fast execution layer to drive the fuel valve, airflow distribution valve, and vent valve to perform a full-stroke reciprocating motion. During this process, the sensor network layer collects the mechanical position feedback data of the actuators in real time, and the fast execution layer calculates the time difference between the control command issuance time and the feedback signal arrival time based on the sampling clock, thereby determining the response time constant of each actuator. By comparing the deviation between the command value and the actual feedback value, the system generates an error compensation table covering the entire range and stores it in non-volatile memory. This table is used to correct the control commands in real time during subsequent operation, eliminating control steady-state errors caused by mechanical wear or manufacturing tolerances.

[0027] Subsequently, the rapid execution layer acquires the rotational speed signal of the shaft engine in real time through the sensor network layer and outputs a high-frequency pulse width modulation signal to drive the inverter of the external drive motor. As the external drive motor drives the shaft to rotate, the rotational speed steadily increases. When the rotational speed reaches the preset ignition speed range and a stable airflow field is established in the combustion chamber by the compressor, the decision layer, according to the preset timing logic, instructs the rapid execution layer to sequentially trigger the ignition excitation unit and the fuel valve. At this time, the rapid execution layer monitors the change gradient of the combustion chamber exhaust temperature at a kilohertz sampling frequency. Once it is determined that the exhaust temperature rise rate exceeds the preset threshold, it is considered that the combustion chamber has been successfully ignited and the system has entered a self-sustaining operation state. At this time, the rapid execution layer immediately cuts off the power output of the external drive motor and instructs the tripping mechanism to physically disengage the drive motor from the main shaft system.

[0028] The decision-making layer drives the engine speed to continue increasing by gradually increasing the step opening command of the fuel valve. To achieve dynamic balance, the fast execution layer synchronously adjusts the guide vane angle of the turbine assembly or the flow cross-sectional area of ​​the exhaust bypass valve, so that the system's output torque gradually approaches the target value. During this process, due to the extremely small moment of inertia of the high-speed rotor, any slight fluctuation in load will cause severe oscillations in speed. To address this, a speed control inner loop operates within the fast execution layer. This loop is directly implemented in the hardware logic. It obtains the speed change rate by performing high-order differential analysis on the speed signal and adjusts the fuel injection pressure in real time, thereby counteracting the transient inertial disturbances caused by load fluctuations within milliseconds and ensuring the stability of the shaft system operation.

[0029] Next, the system adopts a dual closed-loop control structure for both engine speed and fuel flow. The multi-core lockstep MCU decision layer constructs an optimization model for shaft engine efficiency based on real-time collected thermodynamic parameters such as pressure, temperature, and flow rate. This model comprehensively considers power balance scheduling, dual-flow path thermal management regulation, and adaptive correction of multi-fuel characteristics. In the dual-flow path thermal management regulation, the system is divided into high-pressure path control and combined path control. The fast execution layer monitors the outlet pressure and inlet flow rate of the volumetric compressor in real time and calculates the real-time pressure ratio. When the operating point approaches the surge boundary line, the fast execution layer directly drives the high-speed return valve to open to release excess air. In the combined path control, the three-way proportional control valve located downstream of the compressor assembly plays a crucial role. The decision layer acquires thermocouple data installed on the outer wall of the combustion chamber and temperature sensor data at the turbine inlet, and uses a weighted average algorithm to calculate the heat load index. When the combustion chamber wall temperature approaches the allowable temperature limit of the material, the decision-making layer instructs the three-way proportional control valve to increase the deflection angle to the cooling branch, guiding more cold air to the outer wall of the combustion chamber for convective heat exchange; when the wall temperature is within the safe range, the flow distribution ratio to the heat exchanger branch is increased, so that the air is fully preheated by the regenerator before entering the combustion chamber, thereby utilizing the exhaust waste heat to improve the overall heat-power conversion efficiency of the system.

[0030] To address the issue of variable fuel composition, this embodiment introduces an infrared spectral fuel composition sensor in the adaptive fuel characteristic correction. This sensor identifies the molecular structure ratio of the fuel in the pipeline in real time and transmits the identification code to the decision layer via a bus. The decision layer retrieves and loads the corresponding fuel characteristic parameter map from memory based on the identification code. This map defines in detail the calorific value constant, viscosity characteristics, and ignition delay parameters for different components. During fuel switching, the system runs a smooth switching algorithm, linearly weighting and fusing the old and new parameter maps within a preset time step to prevent power shocks caused by sudden changes in fuel calorific value. Simultaneously, a pressure pulsation sensor on the combustion chamber monitors combustion stability, and the fast execution layer corrects the injection pressure, ignition timing, and fuel valve flow characteristic curves in real time based on combustion chamber pressure pulsation data, exhaust oxygen content, and temperature. Furthermore, a virtual torque-based transition compensation module is connected in series in the control loop. This module can estimate the theoretical torque difference between different fuels at the same mass flow rate, and the fast execution layer compensates for this torque fluctuation in real time by fine-tuning the instantaneous pulse width of the fuel valve.

[0031] Finally, the fast execution layer utilizes the built-in electromagnetic bearing control module to acquire the rotor's two-dimensional displacement signal through displacement sensors deployed circumferentially around the bearing housing. An internally running active damping algorithm suppresses the rotor's radial vibration; the expression for the damping algorithm is: In the formula, This represents the real-time radial displacement signal. , where c represents the output of the damping negative feedback control, and 'c' represents the active damping coefficient. To eliminate in-frequency vibrations at specific speeds, the fast execution layer is also equipped with a digital notch filter. This filter can track the speed frequency in real time and accurately filter out vibration components with the same frequency as the speed in the control signal. When the speed crosses the critical speed range of the shaft system, the system automatically increases the stiffness parameter of the electromagnetic bearing. If the deviation of the shaft center trajectory detected by the displacement sensor exceeds 80% of the preset physical clearance, or if a serious power failure occurs in the system, the hardware emergency stop logic circuit inside the fast execution layer will be triggered immediately. This logic circuit is implemented using pure hardware combinational logic gates and does not rely on the instruction scheduling of the software operating system. Once triggered, the system will directly issue interlock protection commands within microseconds: physically cut off the power to the fuel valve so that it closes under the action of spring force, fully open the compressor vent valve, drive the speed increaser tripping mechanism on the output shaft to physically isolate the load, and switch the three-way proportional regulating valve to the fully cooled position. At the same time, the power amplifier of the electromagnetic bearing is cut off, and the rotor falls smoothly onto the auxiliary bearing with a self-lubricating coating.

[0032] This embodiment also features health monitoring and predictive maintenance capabilities. The fast execution layer utilizes the Fast Fourier Transform algorithm to extract the time-domain signal of shaft vibration. Spectral characteristics to capture time-domain signals of combustion chamber pressure pulsations The abnormal harmonic components are detected, and these eigenvectors are uploaded to the decision layer. (Shaft vibration time-domain signal) Combustion chamber pressure pulsation time domain signal The spectral characteristics obtained by Fast Fourier Transform are as follows: In the formula, k represents the harmonic frequency order. , Characteristic vectors of each harmonic of vibration and pressure pulsation in the complex frequency domain; , This represents a discrete sampling sequence, where N represents the total number of samples. Abnormal harmonic components are extracted through threshold filtering: Φ={A(k err ),P(k err )},in, Indicates the abnormal harmonic order / abnormal characteristic frequency point. It represents the characteristic quantity of harmonic spectrum amplitude at abnormal characteristic frequencies in shaft vibration signals, characterizing fault features such as abnormal rotor vibration, frequency doubling disturbance, and resonance distortion. It represents the harmonic spectrum amplitude characteristic at the abnormal characteristic frequency in the combustion chamber pressure pulsation signal, characterizing abnormal combustion characteristics such as combustion oscillation, thermoacoustic instability, and pressure pulsation distortion. When weak combustion and flameout are detected, auxiliary ignition is performed to complete combustion.

[0033] The decision-making layer runs a lightweight deep belief network model, which generates a system health status score by nonlinearly mapping the coupled features of rotational speed, temperature, and pressure. Specifically, it integrates rotational speed, temperature, pressure, and spectral anomaly features to construct a global input feature vector: In the formula, Indicates the real-time rotational speed of the shaft system; Indicates the thermal load temperature of the combustion chamber / critical components; This indicates the system's operating baseline pressure; ~ This represents the characteristic components of abnormal harmonics in vibration and pressure pulsations. The deep belief network consists of multiple stacked restricted Boltzmann machines, achieving nonlinear mapping; the expression for a single-layer feature mapping (the i-th layer) is: The global health score is then output as follows: Integrate and simplify engineering expressions: In the formula, This indicates the system's health status score; , This represents the weight matrix and bias vector of the i-th layer of the network; represents the Sigmoid activation function; L represents the number of layers in the lightweight DBN network.

[0034] By combining the cumulative number of thermal cycles recorded in the memory with the slope of performance degradation of key components, the decision-making layer can calculate the predicted remaining life of the shaft engine. When the score falls below a preset threshold, the system issues a maintenance warning to the user via the communication interface. The expression for the performance degradation law of key components is as follows: In the formula, This indicates the real-time performance degradation index of a component. Indicates initial rated performance; Indicates the performance degradation slope (aging rate coefficient). This represents the cumulative number of thermal cycles. Remaining Life (RUL) calculation formula: In the formula, RUL represents the remaining service life of the shaft system engine critical components. This indicates the critical performance threshold for component failure.

[0035] In actual operation, this system supports automatic state switching between start-up mode, idle mode, loading mode, steady-state mode, load reduction mode, normal shutdown mode, and emergency shutdown mode. The transition between modes is managed by a finite state machine within the decision-making layer. The sensor network layer employs a dual-redundancy configuration, with all speed and temperature sensors having A / B dual channels. By comparing the deviations of the dual-channel signals in real time, the decision-making layer can seamlessly switch to the backup channel upon detecting distortion in the primary channel signal, ensuring the continuity of the control system. Furthermore, the decision-making layer is connected to a high-speed black box storage module, which cyclically records all-dimensional operating parameters within a preset time before shutdown at a high-frequency sampling rate. This data is not only used for local fault backtracking but is also periodically anonymized and sent to the cloud optimization platform for further optimization of the control mapping table, thereby achieving continuous evolution of the control strategy.

[0036] Regarding the control details of the electromagnetic bearing, the fast execution layer acquires real-time clearance data between the rotor and stator using high-precision displacement sensors. Due to the complex gyroscopic effect generated during high-speed rotation of the shaft system engine, the fast execution layer utilizes a cross-feedback control algorithm to introduce a portion of the X-axis displacement deviation signal into the Y-axis control loop to counteract the strongly coupled gyroscopic torque. When the system detects that the rotational speed is approaching the shaft system's critical bending speed, the decision layer instructs the fast execution layer to dynamically adjust the differential gain of the PID controller, artificially increasing the system's modal damping to ensure the rotor smoothly passes through the resonance zone without instability. For combustion chamber protection, the system not only monitors static temperature but also captures thermoacoustic instability using a dynamic pressure sensor installed at the burner head. Once a sustained increase in pressure pulsation amplitude is detected in a specific frequency band, the fast execution layer immediately intervenes in the opening of the three-way proportional control valve, altering the air distribution ratio entering the combustion chamber to disrupt the thermoacoustic coupling boundary conditions, thereby suppressing combustion oscillations.

[0037] In the coordinated operation of the fuel system and thermal management, the actuation accuracy of the three-way proportional control valve directly affects combustion efficiency. This valve is made of high-temperature resistant alloy and driven by a high-precision servo motor. The rapid execution layer uses closed-loop control based on encoder feedback from the servo motor to ensure the valve core's position repeatability reaches one-thousandth. During the system's transition from low to high load, the decision-making layer pre-calculates the required air split ratio based on a preset energy balance equation and instructs the three-way proportional control valve to actuate in advance. This feedforward control strategy effectively compensates for the thermal inertia of the thermal system, ensuring the combustion chamber wall temperature remains within the material's safe range while maximizing the preheating of the air entering the combustion chamber. This hierarchical distributed control architecture, combining rapid hardware-level response with advanced software-level decision-making, provides multi-dimensional safety assurance and performance optimization methods for coaxial shaft engines.

[0038] The above embodiments are merely one of the preferred embodiments of the present invention and should not be used to limit the scope of protection of the present invention. Any modifications or refinements made to the main design concept and spirit of the present invention that are not of substantial significance, but solve the same technical problem as the present invention, should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent hierarchical distributed control of a coaxial shaft engine, characterized in that, Includes the following steps: S1, perform hardware logic architecture configuration verification, establish signal compensation model, and verify the mechanical position accuracy and response hysteresis of the actuator by driving the fuel valve, air distribution valve and vent valve to perform full-stroke reciprocating motion; S2 monitors the shaft system engine speed and load feedback in real time and adjusts the frequency of the drive motor in a closed loop. After establishing a stable airflow field in the combustion chamber, it opens the ignition excitation unit and fuel valve. After determining that the system has entered the self-sustaining operation state, it cuts off the external drive power supply. S3, increase the fuel valve opening to drive the shaft engine to accelerate, adjust the turbine guide vanes or bypass valve to make the system output power tend to the rated target value, and run the speed control inner loop to suppress transient disturbances caused by load fluctuations; S4 utilizes dual closed-loop control of speed and fuel to maintain power balance. Based on thermodynamic parameters, it establishes an extreme value optimization model for shaft engine efficiency. Using the output of the extreme value optimization model, it applies a disturbance signal to adjust the split ratio and combustion chamber pressure rise ratio, driving the system state to converge towards the highest efficiency range. S5 monitors the vibration amplitude, shaft trajectory, combustion status, combustion temperature and differential pressure parameters of the shaft engine in real time. When abnormal operating conditions are detected, it triggers the protection program to achieve fuel cut-off, rotor support protection and rapid shutdown. When weak combustion and flameout are detected, it performs auxiliary ignition and combustion repair.

2. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 1, characterized in that, In S1, the hardware logic architecture configuration verification is achieved by starting the boot program through the multi-core lockstep MCU decision layer and performing read / write tests on internal registers and external memory, and by loading a preset bitstream file using the FPGA fast execution layer.

3. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 1, characterized in that, In S4, the shaft engine efficiency extreme value optimization model includes power balance scheduling, dual-flow path thermal management regulation, multi-fuel characteristic adaptive correction, and extreme value optimization of system cycle efficiency. The dual-flow path thermal management regulation includes high-pressure path control and merging path control. The high-pressure path control monitors the outlet pressure and inlet flow of the volumetric compressor and calculates the real-time pressure ratio through the FPGA fast execution layer. When the operating point approaches the surge line, it drives the high-speed return valve to perform anti-surge protection. The merging path control allocates flow between the cooling branch and the heat exchanger branch through a three-way proportional regulating valve. The multi-core lockstep MCU decision layer performs weighted calculations based on combustion chamber wall temperature data and turbine inlet temperature data, dynamically allocating the flow ratio through the cooling branch and the heat exchanger branch. When the combustion chamber wall temperature approaches the material's allowable limit, the flow allocation ratio of the cooling branch is increased; when the temperature is within the safe range, the airflow is guided to the heat exchanger branch.

4. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 3, characterized in that, The multi-fuel characteristic adaptive correction supports multiple fuels, including compressed natural gas, gasoline, hydrogen, and methanol. Fuel composition sensors installed on the fuel supply pipeline identify fuel type and component ratio in real time. The multi-core lockstep MCU decision layer automatically switches between preset characteristic parameter maps stored in memory based on the identification results. During fuel switching, a smooth switching algorithm completes the transition from one fuel characteristic map to another. Injection pressure, ignition timing, and fuel valve flow characteristic curves are corrected in real time based on combustion chamber pressure pulsation data, exhaust oxygen content, and temperature. To ensure smooth power delivery during multi-fuel switching, a transition compensation stage based on virtual torque is added to the algorithm. This stage calculates the torque difference between different fuels at the same flow rate and instructs the FPGA fast execution layer to offset the torque difference by instantaneously adjusting the injection pressure.

5. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 4, characterized in that, The FPGA fast execution layer integrates an electromagnetic bearing control module, which monitors the rotor's shaft trajectory in real time through a displacement sensor and uses active damping and notch filtering algorithms to suppress rotor vibration when it exceeds the critical speed. The active damping algorithm generates electromagnetic force compensation commands by calculating the first and second derivatives of the rotor displacement, and the digital notch filter is used to lock and filter out vibration components near the speed frequency. When the shaft system passes through the critical speed region, the FPGA fast execution layer dynamically adjusts the stiffness and damping parameters. When shaft system instability, system power failure, or vibration amplitude exceeding the preset safety limit is detected, the rotor is forcibly triggered to drop to the auxiliary bearing, and a fuel cut-off command is issued simultaneously.

6. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 5, characterized in that, The protection program in S5 is triggered by the hardware emergency stop logic built into the FPGA fast execution layer. This logic is implemented by hardware combinational circuits and is not affected by software interrupt nesting. When the triggering conditions are met, the system completes the interlocking protection action within a preset time threshold. The interlocking protection action includes: closing the fuel shut-off valve to cut off the energy input, opening the vent valve to release the system pressure, driving the speed increaser to trip to isolate the load, and opening the bypass valve to change the airflow path.

7. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 6, characterized in that, It also includes health monitoring and predictive maintenance steps: the FPGA fast execution layer extracts shaft vibration spectrum features, combustion chamber pressure pulsation frequency, and pressure ratio fluctuation features, and uploads the processed feature vector to the multi-core lockstep MCU decision layer; the multi-core lockstep MCU decision layer runs a lightweight neural network model to complete fault mode diagnosis. The input layer of the neural network model receives speed, temperature, pressure, vibration features, and fuel flow data, the hidden layer performs feature fusion on the data through a nonlinear activation function, and the output layer gives a health index score; the multi-core lockstep MCU decision layer predicts the remaining service life based on the system's cumulative runtime, thermal cycle count, and the performance degradation slope of key components.

8. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 1, characterized in that, The method supports automatic switching between multiple modes, including start-up mode, idle mode, loading mode, steady-state mode, unloading mode, normal shutdown mode, and emergency shutdown mode. The transition logic between modes is controlled by a state machine in the multi-core lockstep MCU decision layer, which determines the switching conditions based on speed, temperature, pressure, and external control commands. The sensor network layer adopts a redundant configuration, with key sensors having multi-channel backups. The multi-core lockstep MCU decision layer compares multi-channel signals in real time, and switches to the backup channel and records the fault code when a single signal fails.

9. The intelligent hierarchical distributed control method for a coaxial shaft engine according to claim 1, characterized in that, The multi-core lockstep MCU decision layer sends the processed, de-identified operating data to the cloud optimization platform through the cloud optimization interface, obtains the control parameters and characteristic maps optimized by swarm intelligence analysis, and transmits them back to the multi-core lockstep MCU decision layer via remote firmware upgrade to achieve online updates of the control strategy; in the high-speed flash memory mounted on the multi-core lockstep MCU decision layer, the system records the operating parameters cyclically at preset time intervals, and locks the data when a shutdown event occurs for subsequent fault backtracking analysis.

10. An intelligent hierarchical distributed control system for a coaxial shaft engine, used to execute the method according to any one of claims 1 to 9, characterized in that, The system includes: The multi-core lockstep MCU decision layer uses multiple symmetrical processor cores to execute the same instruction stream and monitors the consistency of execution results in real time through a hardware comparator. It is used to handle non-real-time operation mode management, system-level power allocation strategy, overall thermal efficiency optimization algorithm, and parameter scheduling logic during multi-fuel switching. The FPGA fast execution layer is connected to the multi-core lockstep MCU decision layer through a high-speed serial communication link. Its internal logic is constructed by hardware gate circuits to drive the high-speed input / output interface, pulse width modulation module and electromagnetic bearing actuator, and undertakes the inner loop logic of speed control, active vibration control of electromagnetic bearing and hardware emergency stop protection. The sensor network layer is equipped with an electromagnetic speed sensor, a thermocouple temperature sensor, a piezoelectric pressure sensor, a fuel composition analyzer, and an electromagnetic bearing displacement sensor, which are used to sense the full-dimensional operating status of the system and transmit the signals to the FPGA fast execution layer. The actuator network layer includes a fuel valve driver, an airflow distribution valve, a venting valve stepper motor control module, an ignition excitation unit, an electromagnetic bearing power amplifier, and an external drive motor controller, which are used to receive control commands and drive the corresponding mechanical components to perform actions.

Images