Multi-machine running-in test distributed control system and method based on digital twinning

Abstract

The invention relates to the technical field of digital twinning, in particular to a multi-machine running-in test distributed control system and method based on digital twinning. The method comprises the steps that a sensing execution unit collects equipment operation state data based on a multi-source sensor module and carries out preliminary data processing and detection; the digital twinning unit receives multi-dimensional sensing data of the sensing execution unit based on a dynamic twinning modeling module, establishes an equipment state model, and performs simulation verification and parameter optimization on a control strategy through a Q-learning strategy to obtain optimized control parameters; the intelligent control unit constructs an optimal decision-making model of multi-machine cooperative control based on a cooperative decision-making module, adjusts parameters of an execution mechanism in real time through a PID control algorithm and feed-forward compensation, and generates a self-adaptive control instruction; and the fusion analysis unit constructs a multi-dimensional equipment operation knowledge base based on the knowledge graph module, and generates global optimization parameters through a multi-target genetic algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twins, and more specifically, to a distributed control system and method for multi-machine running-in test based on digital twins. Background Art

[0002] The distributed control system and method for multi-machine running-in test based on digital twins is an intelligent test system that integrates advanced simulation modeling, real-time data interaction, and collaborative control technologies. By constructing high-fidelity digital twins of physical devices, this system enables the synchronous operation, interactive verification, and performance evaluation of multiple devices in a virtual environment, thereby effectively improving the efficiency and reliability of equipment integration and commissioning. Its core lies in using digital twin technology to simulate the operating states and interaction relationships of real devices in a virtual space, and through a distributed collaborative control architecture between edge computing nodes and cloud platforms, achieving unified scheduling and dynamic optimization of multi-machine systems. This method not only supports communication, coordination, and fault prediction between devices but also can detect potential problems in advance during the test phase, reducing on-site commissioning time and costs, and is widely used in the integrated testing and operation and maintenance management of complex systems such as intelligent manufacturing, smart grids, and rail transit.

[0003] In the existing distributed control system and method for multi-machine running-in test based on digital twins, insufficient state monitoring is caused by incomplete data perception and processing lag; inaccurate modeling of the impact of equipment wear results in virtual-real synchronization errors and weak long-term prediction capabilities; control strategies lack adaptability and coordination, making it difficult to cope with complex and changing working environments; and there is insufficient system optimization ability and a lack of a global perspective. Therefore, a distributed control system and method for multi-machine running-in test based on digital twins are provided. Summary of the Invention

[0004] The purpose of the present invention is to provide a distributed control system and method for multi-machine running-in test based on digital twins to solve the problems in the existing distributed control system and method for multi-machine running-in test based on digital twins, including insufficient state monitoring caused by incomplete data perception and processing lag; inaccurate modeling of the impact of equipment wear resulting in virtual-real synchronization errors and weak long-term prediction capabilities; control strategies lacking adaptability and coordination, making it difficult to cope with complex and changing working environments; and insufficient system optimization ability and a lack of a global perspective as mentioned in the above background art.

[0005] To achieve the above object, on the one hand, the present invention aims to provide a distributed control system for multi-machine running-in test based on digital twins, including: a sensing and execution unit, which collects device operation state data based on a multi-source sensor module and performs preliminary data processing and detection.

[0006] Digital twin unit, which receives multi-dimensional perception data of the perception and execution unit based on the dynamic twin modeling module, establishes a device state model considering the impact of equipment wear, and then verifies and optimizes the control strategy through the Q-learning strategy to obtain optimized control parameters;

[0007] Intelligent control unit, which constructs an optimal decision-making model for multi-machine cooperative control based on the collaborative decision-making module, and adjusts the actuator parameters in real time through the PID control algorithm and feedforward compensation to generate adaptive control instructions;

[0008] Fusion analysis unit, which constructs a multi-dimensional device operation knowledge base based on the knowledge graph module and generates global optimization parameters through the multi-objective genetic algorithm.

[0009] As a further improvement of this technical solution, the multi-source sensor module includes a heterogeneous sensing cluster and an edge intelligent processing module;

[0010] Among them, the heterogeneous sensing cluster is used to integrate multi-type industrial sensors such as vibration, temperature, and pressure to achieve full-dimensional perception of the device operation state;

[0011] The edge intelligent processing module realizes real-time data noise reduction and anomaly detection based on the sliding window filtering and spatio-temporal alignment algorithms.

[0012] As a further improvement of this technical solution, the specific steps for the edge intelligent processing module to realize real-time data noise reduction and anomaly detection based on the sliding window filtering and spatio-temporal alignment algorithms are as follows:

[0013] Select the window length according to the sensor type , and use recursive calculation for mean filtering ;

[0014] Compensate for network delay based on the Timestamp interpolation ;

[0015] Calculate the alarm threshold for dynamic threshold anomaly detection in real time through the sliding window statistics.

[0016] As a further improvement of this technical solution, the digital twin unit includes a dynamic twin modeling module and a virtual commissioning module;

[0017] Among them, the dynamic twin modeling module receives multi-dimensional perception data of the perception and execution unit, establishes a device state model considering the impact of equipment wear, and realizes virtual-real synchronization through Kalman filtering;

[0018] The virtual commissioning module is updated and optimized through the Q-learning strategy to obtain optimized control parameters, and the optimized control parameters are fed back to the dynamic twin modeling module in real time to dynamically correct the wear term in the state equation of the Kalman filter, realizing the collaborative update of model parameters and strategies.

[0019] As a further improvement of this technical solution, the specific steps for the dynamic twin modeling module to establish the equipment state model and achieve virtual-real synchronization through Kalman filtering are as follows:

[0020] According to multi-dimensional perception data and wear state , and considering the dust concentration to accelerate the wear degree, the dust concentration is introduced , and an equipment state model is established by expanding the state equation:

[0021] ;

[0022] In the formula, represents the wear state; represents the dynamic model of the system; represents the control input variable; represents the wear rate coefficient of the control input variable; represents the multi-dimensional perception data expected value; represents the deviation value; represents the wear rate coefficient of the deviation value; represents the wear rate coefficient of the dust concentration; represents the dust concentration; represents the process noise; represents the equipment state change rate;

[0023] Perform state prediction and state update .

[0024] As a further improvement of this technical solution, the intelligent control unit includes a collaborative decision-making module and a dynamic compensation module;

[0025] Among them, the collaborative decision-making module realizes multi-machine collaborative task allocation and resource scheduling through the policy function;

[0026] The dynamic compensation module performs non-linear disturbance suppression and error correction based on real-time sensing data and model prediction.

[0027] As a further improvement of this technical solution, the specific steps for the dynamic compensation module to use the PID control algorithm and feed-forward compensation to output the control instruction vector are as follows:

[0028] The updated current device state is obtained through the digital twin model and Kalman filter and compared with the expected state to calculate the state deviation ;

[0029] According to the state deviation the PID control algorithm is adopted and a feedforward compensation term is introduced to calculate the output control instruction vector .

[0030] As a further improvement of this technical solution, the sensing and execution unit further includes an execution module. The specific steps for the execution module to convert the control instruction generated by the intelligent control unit into an execution action are as follows:

[0031] Receive the control instruction vector output by the intelligent control unit ;

[0032] Through the conversion coefficient Convert the control instruction vector into an execution action .

[0033] As a further improvement of this technical solution, the fusion analysis unit includes a knowledge graph module and an optimization feedback module;

[0034] Among them, the knowledge graph module realizes the intelligent identification and root cause analysis of fault modes by constructing a multi-dimensional association network of device state - fault - wear;

[0035] The optimization feedback module uses a multi-objective genetic algorithm to dynamically generate global optimization parameters and adjusts the decision-making priorities of the digital twin model and control strategy in real time.

[0036] On the other hand, the present invention provides a distributed control method for multi-machine running-in test based on digital twin, which is used for the distributed control system for multi-machine running-in test based on digital twin described in any one of the above, and includes the following steps:

[0037] S10.1. Collect device operation state data through multi-source sensors. After real-time filtering, alignment and anomaly detection by the edge intelligent processing module, convert the control instruction into a physical execution action to achieve comprehensive perception and rapid response to the device state;

[0038] S10.2. Based on the dynamic modeling module, fuse the wear factor to establish a high-precision device state model, and realize virtual-real synchronization through Kalman filtering. Combine the Q-learning strategy of the virtual commissioning module to complete the simulation optimization of the control strategy and parameter adaptive adjustment;

[0039] S10.3. Construct an optimal multi-machine control strategy using the collaborative decision-making module, and generate an adaptive control instruction through the dynamic compensation module using PID and feedforward control;

[0040] S10.4. Construct a multi-dimensional association network of equipment status - fault - wear through the knowledge graph module to achieve intelligent diagnosis, and generate global optimization parameters by combining the multi-objective genetic algorithm of the optimization feedback module, dynamically adjust the model weights and control priorities, and form a cross-unit collaborative optimization closed loop.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] 1. In the distributed control system and method for multi-machine running-in test based on digital twin, by collecting the equipment operation status data and performing preliminary processing, and at the same time converting the control instruction into a physical execution action, the real-time perception and control closed loop of the system are realized; an equipment status model integrating wear factors is constructed based on multi-dimensional perception data, and the simulation optimization and dynamic update of the control strategy are realized through virtual commissioning.

[0043] 2. In the distributed control system and method for multi-machine running-in test based on digital twin, the optimal control instruction is generated through collaborative decision-making and adaptive control algorithms to achieve the efficient collaboration and stable operation of the multi-machine system. An intelligent analysis and feedback mechanism is constructed using the knowledge graph and multi-objective optimization algorithm to improve the global optimization ability and fault diagnosis level of the system. Description of the Drawings

[0044] Figure 1 is the overall flow block diagram of the present invention;

[0045] The meanings of each label in the figure are as follows:

[0046] 1. Perception and execution unit; 11. Multi-source sensor module; 111. Heterogeneous sensing cluster; 112. Edge intelligent processing module; 12. Execution module; 2. Digital twin unit; 21. Dynamic twin modeling module; 22. Virtual commissioning module; 3. Intelligent control unit; 31. Collaborative decision-making module; 32. Dynamic compensation module; 4. Fusion analysis unit; 41. Knowledge graph module; 42. Optimization feedback module. Specific Embodiments

[0047] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0048] Embodiment 1: Please refer to Figure 1As shown, a distributed control system for multi-machine running-in test based on digital twin is provided, including a sensing and execution unit 1. The sensing and execution unit 1 collects device operation status data based on a multi-source sensor module 11 and performs preliminary data processing and detection.

[0049] In this example, the multi-source sensor module 11 includes a heterogeneous sensing cluster 111 and an edge intelligent processing module 112.

[0050] Among them, the heterogeneous sensing cluster 111 is used to integrate multi-type industrial sensors such as vibration, temperature, and pressure to achieve full-dimensional perception of the device operation status.

[0051] Specifically, vibration sensors, temperature sensors, and pressure sensors are deployed to generate vibration raw data , temperature raw data , pressure raw data ;

[0052] Output multi-dimensional perception data :

[0053] ;

[0054] In the formula, represents the transpose operation; represents the wear variable characteristic signal;

[0055] The edge intelligent processing module 112 realizes real-time data noise reduction and anomaly detection based on the sliding window filtering and spatio-temporal alignment algorithms.

[0056] In this example, the specific steps for the edge intelligent processing module 112 to realize real-time data noise reduction and anomaly detection based on the sliding window filtering and spatio-temporal alignment algorithms are as follows:

[0057] Select the window length according to the sensor type , and adopt recursive calculation for mean filtering ;

[0058] Specifically, the mean filtering is:

[0059] ;

[0060] In the formula, represents the current time point; represents the th time point of the multi-dimensional perception data;

[0061] Compensate for network delay based on the Timestamp interpolation ;

[0062] Specifically, the network delay is:

[0063] ;

[0064] In the formula, represents the average network delay time; represents the multi-dimensional perception data received at the actual arrival time; represents the sampling time;

[0065] Among them, the average network delay time is:

[0066] ;

[0067] In the formula, represents the number of transmissions; represents the th actual delay time of transmission; represents the transmission number index;

[0068] The alarm threshold is calculated in real time through the sliding window statistic for dynamic threshold anomaly detection.

[0069] Specifically, the calculation of the sliding window statistic is:

[0070] ;

[0071] ;

[0072] In the formula, represents the mean value within the current window; represents the standard deviation within the current window;

[0073] The alarm threshold is:

[0074] ;

[0075] In the formula, represents the multi-dimensional perception data of a data point at a certain moment;

[0076] The specific steps for the execution module 12 to convert the control instruction generated by the intelligent control unit 3 into an execution action are:

[0077] Receive the control instruction vector output by the intelligent control unit 3 ;

[0078] Through the conversion coefficient Convert the control instruction vector into the execution action .

[0079] Specifically, the execution action is as follows:

[0080] ;

[0081] Furthermore, in the distributed control system for multi-machine running-in test based on digital twin, the perception and execution unit 1 collects and preliminarily processes the device operation status data (such as vibration, temperature, pressure, and wear variable characteristic signals) through the multi-source sensor module 11, performs real-time noise reduction and anomaly detection using the edge intelligent processing module 112, and converts the control instructions generated by the intelligent control unit 3 into actual execution actions through the execution module 12. This process not only ensures the comprehensive, accurate perception and timely response to the device status, but also realizes the dynamic synchronization between the virtual model and the physical entity through digital twin technology, thereby supporting the predictive maintenance, fault diagnosis, and performance optimization of the system, and greatly improving the efficiency and reliability of multi-machine collaborative work. Specifically, this mechanism enables the system to quickly adjust the operation strategy according to the real-time monitoring data, prevent potential faults, extend the service life of the device, and ensure the stability and efficiency of the production process.

[0082] The distributed control system for multi-machine running-in test based on digital twin further includes a digital twin unit 2. The digital twin unit 2 receives the multi-dimensional perception data of the perception and execution unit 1 based on the dynamic twin modeling module 21, establishes a device status model considering the influence of device wear, and then performs simulation verification and parameter optimization on the control strategy through the Q-learning strategy to obtain the optimized control parameters;

[0083] In this example, the digital twin unit 2 includes a dynamic twin modeling module 21 and a virtual commissioning module 22;

[0084] Among them, the dynamic twin modeling module 21 receives the multi-dimensional perception data of the perception and execution unit 1, establishes a device status model considering the influence of device wear, and realizes virtual-real synchronization through Kalman filtering;

[0085] In this example, the specific steps to establish a device status model and realize virtual-real synchronization through Kalman filtering are as follows:

[0086] According to the multi-dimensional perception data and the wear status , and considering the dust concentration to accelerate the wear degree, the dust concentration is introduced , and a device status model is established through the extended state equation:

[0087] ;

[0088] In the formula, represents the wear status; represents the dynamic model of the system; represents the control input variable; represents the wear rate coefficient of the control input variable; represents the expected value of multi-dimensional perception data; represents the deviation value; represents the wear rate coefficient of the deviation value; represents the wear rate coefficient of the dust concentration; represents the dust concentration; represents the process noise; represents the rate of change of the equipment state;

[0089] Perform state prediction through Kalman filtering and state update .

[0090] Specifically, the dynamic model of the system is based on the current system state and the control input applied to the system , calculates the rate of change of the system state, that is, the predicted value of the system state at the next moment;

[0091] Among them, the wear state will be affected by the dust concentration, which has an accelerating effect on wear;

[0092] Then the wear state The defined evolution equation after being affected by the dust concentration is:

[0093] ;

[0094] In the formula, represents the instantaneous change rate of the wear state;

[0095] Among them, the wear rate coefficient of the control input variable is:

[0096] ;

[0097] In the formula, represents the wear increment within the time interval ; represents the time interval;

[0098] The wear rate coefficient of the deviation value is:

[0099] ;

[0100] The wear rate coefficient of the dust concentration is:

[0101] ;

[0102] State prediction through Kalman filtering And state update 。

[0103] State prediction is:

[0104] ;

[0105] In the formula, represents the state transition matrix; represents the prediction at time represents the control input matrix; represents the control input at time

[0106] State update is:

[0107] ;

[0108] In the formula, represents the Kalman gain; represents the measurement vector; represents the observation matrix;

[0109] The virtual debugging module 22 is updated and optimized through the Q-learning strategy to obtain optimized control parameters, and the optimized control parameters are fed back to the dynamic twin modeling module 21 in real time to dynamically correct the wear term in the state equation of the Kalman filter, realizing the collaborative update of model parameters and strategies.

[0110] Specifically, the Q-learning strategy is:

[0111] ;

[0112] In the formula, represents the current system state; represents the learning rate; represents the action; represents the immediate reward; represents the next state; represents the next action; represents the discount factor, ;

[0113] Among them, the current system state is:

[0114] ;

[0115] Specifically, the current system state including the updated device state estimation value and the cumulative wear state of the device ;

[0116] wherein, the immediate reward is:

[0117] ;

[0118] In the formula, represents the weight coefficient of the wear state in the immediate reward; represents the weight coefficient of the deviation value in the immediate reward; represents the weight coefficient of the control input variable in the immediate reward; represents the weight coefficient of the dust concentration in the immediate reward;

[0119] Specifically, the larger, the more significant the negative impact of wear on the reward; the importance of controlling the state tracking accuracy; weighing the control efficiency and the actuator life; reflecting the influence weight of dust on the system health;

[0120] Specifically, the wear term in the state equation of the dynamic correction Kalman filter is modified to realize the collaborative update of the model parameters and the strategy as follows:

[0121] Modify the state equation of the dynamic twin modeling module:

[0122] ;

[0123] In the formula, , , represents the time-varying wear rate coefficient dynamically optimized by the virtual commissioning module 22;

[0124] Then, the time-varying wear rate coefficient specifically dynamically optimized by the virtual commissioning module 22 is:

[0125] ;

[0126] In the formula, represents the time-varying wear rate coefficient dynamically optimized, reflecting the real-time influence weights of the control input, state deviation, and dust concentration on wear; represents the learning rate; represents the adjustment amount of the wear rate coefficient; represents the dynamic optimization index variable;

[0127] Furthermore, in the distributed control system for multi-machine running-in test based on digital twin, the digital twin unit 2 receives multi-dimensional perception data from the perception and execution unit 1 through the dynamic twin modeling module 21, establishes an accurate equipment state model in combination with the equipment wear condition, and realizes real-time synchronization between the virtual model and the physical entity by using the Kalman filtering technology; at the same time, the virtual commissioning module 22 uses the Q-learning strategy to simulate and verify the control strategy and optimize the parameters to obtain the optimal control parameters. This mechanism can not only accurately predict the state changes of the equipment, timely detect potential faults and evaluate the wear degree, but also optimize the control logic through multiple iterations in the virtual environment to ensure the implementation of the most effective control instructions in actual operation, thereby significantly improving the reliability and efficiency of the system, extending the service life of the equipment, and reducing the maintenance cost. This enables the entire system to operate efficiently and stably in a complex and changeable working environment, realizing intelligent management of the entire process from monitoring, analysis to decision-making support.

[0128] The distributed control system for multi-machine running-in test based on digital twin further includes an intelligent control unit 3. The intelligent control unit 3 constructs an optimal decision-making model for multi-machine collaborative control based on the collaborative decision-making module 31, and adjusts the actuator parameters in real time through the PID control algorithm and feed-forward compensation to generate adaptive control instructions;

[0129] In this example, the intelligent control unit 3 includes a collaborative decision-making module 31 and a dynamic compensation module 32;

[0130] Among them, the collaborative decision-making module 31 realizes multi-machine collaborative task allocation and resource scheduling through the policy function;

[0131] Specifically, the policy function is:

[0132] ;

[0133] In the formula, represents the policy function; represents the time range;

[0134] The dynamic compensation module 32 uses the PID control algorithm and feed-forward compensation to output a control instruction vector.

[0135] In this example, the specific steps for the dynamic compensation module 32 to use the PID control algorithm and feed-forward compensation to output a control instruction vector are as follows:

[0136] Obtain the updated of the current equipment state through the digital twin model and Kalman filtering, and compare it with the expected state to calculate the state deviation

[0137] Specifically, the state deviation is:

[0138] ;

[0139] According to the state deviation adopt the PID control algorithm and introduce a feedforward compensation term calculate the output control command vector .

[0140] Specifically, the output control command vector is:

[0141] ;

[0142] In the formula, represents the proportional gain parameter; represents the integral gain parameter; represents the derivative gain parameter; represents the state deviation at time represents the th system state deviation at the time step; represents the time step index variable;

[0143] Among them, the feedforward compensation term is:

[0144] ;

[0145] In the formula, represents the reference control input; represents the inverse function of the system dynamics model , that is, according to the current system state estimate and the reference control input , calculate the required ideal control quantity; represents differentiating the inverse dynamics output to generate a feedforward compensation quantity to offset the system dynamic delay;

[0146] Furthermore, in the distributed control system for multi-machine running-in test based on digital twin, the intelligent control unit 3 constructs an optimal decision-making model for multi-machine collaborative control through the collaborative decision-making module 31 to achieve task allocation and resource scheduling, and uses the dynamic compensation module 32 to output an adaptive control instruction vector by combining PID control with feedforward compensation according to the real-time state deviation. This design not only ensures the efficient cooperation and optimal allocation of resources among multiple devices, but also can precisely adjust according to the actual operating state of each device to minimize errors, improve the system response speed and stability. Specifically, through the comprehensive monitoring and intelligent regulation of the multi-machine system, this unit realizes the rapid adaptation to complex working conditions and optimized operation, thus significantly improving the overall production efficiency and equipment reliability, while reducing the maintenance cost and failure risk. This mechanism enables the system to maintain the best performance in a dynamically changing working environment.

[0147] The distributed control system for multi-machine running-in test based on digital twin further includes a fusion analysis unit 4. The fusion analysis unit 4 constructs a multi-dimensional equipment operation knowledge base based on the knowledge graph module 41 and generates global optimization parameters through a multi-objective genetic algorithm.

[0148] In this example, the fusion analysis unit 4 includes a knowledge graph module 41 and an optimization feedback module 42;

[0149] Among them, the knowledge graph module 41 realizes the intelligent identification and root cause analysis of fault modes by constructing a multi-dimensional association network of equipment status - fault - wear;

[0150] Specifically, constructing a multi-dimensional association network of equipment status - fault - wear means establishing a multi-level and dynamic association model covering the whole process of "normal state - abnormal state - early fault - performance degradation - severe wear" by integrating equipment operation status data (such as vibration, temperature, pressure, etc.), fault type information, and wear evolution characteristics. This network not only describes the state evolution path of the equipment under different working conditions, but also reveals the causal relationship and mutual influence mechanism between fault occurrence and wear accumulation, thus providing a theoretical support and data-driven analysis basis for predictive maintenance, fault diagnosis, and life assessment.

[0151] The optimization feedback module 42 dynamically generates global optimization parameters using a multi-objective genetic algorithm and adjusts the decision-making priorities of the digital twin model and control strategy in real time.

[0152] Specifically, the dynamic generation of global optimization parameters using a multi-objective genetic algorithm is as follows:

[0153] ;

[0154] In the formula, represents the global optimization parameter; denote as the weight coefficient; denote the maximum wear threshold; denote the weight coefficient after wear; denote the weight coefficient of the device state; denote the reference state vector;

[0155] Furthermore, in the distributed control system for multi-machine running-in test based on digital twin, the fusion analysis unit 4 constructs a multi-dimensional association network of device state - fault - wear through the knowledge graph module 41 to achieve intelligent identification and root cause analysis, and uses the optimization feedback module 42 to dynamically generate global optimization parameters by using the multi-objective genetic algorithm, and adjusts the decision-making priorities of the digital twin model and the control strategy in real time. This mechanism not only supports the comprehensive monitoring and in-depth analysis of the device operation state, fault mode and wear trend, but also can automatically optimize the control parameters according to multiple factors such as system performance, wear degree and state deviation, so as to form a cross-unit collaborative optimization link. This makes the whole system have stronger adaptability and robustness, can continuously optimize the operation efficiency, extend the device life and reduce the maintenance cost in a complex and changeable working environment, and ensures the efficient and stable operation of the system. Through this comprehensive intelligent control method, the full-process intelligent management from data collection to decision execution is realized.

[0156] Embodiment 2:

[0157] The difference between Embodiment 2 and Embodiment 1 of the present invention is that this embodiment introduces the static mechanical property acquisition and analysis method used in the distributed control system for multi-machine running-in test based on digital twin.

[0158] The distributed control method for multi-machine running-in test based on digital twin, which is used for the distributed control system for multi-machine running-in test based on digital twin in any one of the above, includes the following steps:

[0159] S10.1. Collect device operation state data through multi-source sensors. After real-time filtering, alignment and anomaly detection by the edge intelligent processing module, convert the control instructions into physical execution actions to achieve the comprehensive perception and rapid response to the device state;

[0160] S10.2. Based on the dynamic modeling module, fuse the wear factor to establish a high-precision device state model, and realize the virtual-real synchronization through Kalman filtering. Combine the Q-learning strategy of the virtual commissioning module to complete the simulation optimization and parameter adaptive adjustment of the control strategy;

[0161] S10.3. Use the collaborative decision-making module to construct an optimal control strategy for multiple machines, and generate an adaptive control instruction by using PID and feedforward control through the dynamic compensation module;

[0162] S10.4. Construct a multi-dimensional association network of equipment status - fault - wear through the knowledge graph module to achieve intelligent diagnosis, generate global optimization parameters by combining the multi-objective genetic algorithm of the optimization feedback module, dynamically adjust the model weights and control priorities, and form a cross-unit collaborative optimization closed loop.

[0163] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A distributed control system for multi-machine running-in test based on digital twin, characterized in that: Including: A sensing and execution unit (1), which collects device operation status data based on a multi-source sensor module (11) and performs preliminary data processing and detection; A digital twin unit (2), which receives multi-dimensional sensing data from the sensing and execution unit (1) based on a dynamic twin modeling module (21), establishes a device state model considering the impact of device wear, and then simulates and verifies the control strategy and optimizes the parameters through a Q-learning strategy to obtain optimized control parameters; An intelligent control unit (3), which constructs an optimal decision-making model for multi-machine cooperative control based on a collaborative decision-making module (31), and adjusts the actuator parameters in real time through a PID control algorithm and feedforward compensation to generate an adaptive control instruction; A fusion analysis unit (4), which constructs a multi-dimensional device operation knowledge base based on a knowledge graph module (41) and generates global optimization parameters through a multi-objective genetic algorithm.

2. The distributed control system for multi-machine running-in test based on digital twin according to claim 1, characterized in that: The multi-source sensor module (11) includes a heterogeneous sensing cluster (111) and an edge intelligent processing module (112); Among them, the heterogeneous sensing cluster (111) is used to integrate multi-type industrial-level sensors such as vibration, temperature, and pressure to achieve full-dimensional perception of the device operation status; The edge intelligent processing module (112) realizes real-time data noise reduction and anomaly detection based on a sliding window filtering and spatio-temporal alignment algorithm.

3. The distributed control system for multi-machine running-in test based on digital twin according to claim 2, wherein: The specific steps for the edge intelligent processing module (112) to realize real-time data noise reduction and anomaly detection based on a sliding window filtering and spatio-temporal alignment algorithm are as follows: Select the window length according to the sensor type , and calculate the mean filtering using a recursive formula ; Timestamp Interpolation Compensation for Network Latency ; Calculate the alarm threshold for dynamic threshold anomaly detection in real time through the sliding window statistic.

4. The distributed control system for multi-machine running-in test based on digital twin according to claim 1, wherein: The digital twin unit (2) includes a dynamic twin modeling module (21) and a virtual commissioning module (22); Among them, the dynamic twin modeling module (21) receives multi-dimensional sensing data from the sensing and execution unit (1), establishes a device state model considering the impact of device wear, and realizes virtual-real synchronization through Kalman filtering; The virtual commissioning module (22) updates and optimizes through a Q-learning strategy to obtain optimized control parameters, and feeds the optimized control parameters back to the dynamic twin modeling module (21) in real time to dynamically correct the wear term in the state equation of Kalman filtering, realizing the collaborative update of model parameters and strategies.

5. The distributed control system for multi-machine running-in test based on digital twin according to claim 4, characterized in that: The specific steps for the dynamic twin modeling module (21) to establish a device state model and realize virtual-real synchronization through Kalman filtering are as follows: Based on multi-dimensional perception data and wear state and taking into account the dust concentration to accelerate the degree of wear, the dust concentration is introduced , and a device state model is established by expanding the state equation: ； Wherein, represents the wear state; represents the dynamic model of the system; represents the control input variable; represents the wear rate coefficient of the control input variable; represents the multi-dimensional perception data expected value; represents the deviation value; represents the wear rate coefficient of the deviation value; represents the wear rate coefficient of the dust concentration; represents the dust concentration; represents the process noise; represents the equipment state change rate; State prediction by Kalman filtering and state update .

6. The distributed control system for multi-machine running-in test based on digital twin according to claim 1, characterized in that: The intelligent control unit (3) includes a collaborative decision-making module (31) and a dynamic compensation module (32); Among them, the collaborative decision-making module (31) realizes multi-machine cooperative task allocation and resource scheduling through a policy function; The dynamic compensation module (32) outputs a control instruction vector using a PID control algorithm and feedforward compensation.

7. The distributed control system for multi-machine running-in test based on digital twin according to claim 6, characterized in that: The specific steps for the dynamic compensation module (32) to output a control instruction vector using a PID control algorithm and feedforward compensation are as follows: The updated current device state is obtained through the digital twin model and Kalman filter and compared with the expected state to calculate the state deviation ; According to the state deviation Adopt the PID control algorithm and introduce a feedforward compensation term Calculate the output control instruction vector .

8. The distributed control system for multi-machine running-in test based on digital twin according to claim 7, characterized in that: The perception and execution unit (1) further includes an execution module (12). The specific steps for the execution module (12) to convert the control instructions generated by the intelligent control unit (3) into execution actions are as follows: Receive the control instruction vector output by the intelligent control unit (3) ; By means of a conversion coefficient the control instruction vector is converted into an execution action .

9. The distributed control system for multi-machine running-in test based on digital twin according to claim 1, wherein: The fusion analysis unit (4) includes a knowledge graph module (41) and an optimization feedback module (42); Among them, the knowledge graph module (41) realizes the intelligent identification and root cause analysis of fault modes by constructing a multi-dimensional association network of equipment status - fault - wear; The optimization feedback module (42) dynamically generates global optimization parameters using a multi-objective genetic algorithm and adjusts the decision priorities of the digital twin model and control strategy in real time.

10. A distributed control method for multi-machine running-in test based on digital twin, which is used for the distributed control system for multi-machine running-in test based on digital twin as described in any one of claims 1-9, characterized in that: It includes the following steps: S10.

1. Collect the operation status data of the equipment through multi-source sensors. After real-time filtering, alignment, and anomaly detection by the edge intelligent processing module, convert the control instructions into physical execution actions to achieve comprehensive perception and rapid response to the equipment status; S10.

2. Based on the dynamic modeling module, fuse the wear factor to establish a high-precision equipment status model, and achieve virtual-real synchronization through Kalman filtering. Combine the Q-learning strategy of the virtual commissioning module to complete the simulation optimization of the control strategy and parameter adaptive adjustment; S10.

3. Use the collaborative decision-making module to construct an optimal control strategy for multiple machines, and generate adaptive control instructions through the dynamic compensation module using PID and feedforward control; S10.

4. Construct a multi-dimensional association network of equipment status - fault - wear through the knowledge graph module to achieve intelligent diagnosis. Combine the multi-objective genetic algorithm of the optimization feedback module to generate global optimization parameters, dynamically adjust the model weights and control priorities, and form a cross-unit collaborative optimization closed loop.

Images