Industrial park cloud side cooperative control system based on digital twinning

Abstract

The invention relates to the technical field of industrial park control, and discloses an industrial park cloud edge cooperative control system based on digital twinning, which comprises a cloud engine and a plurality of edge control nodes, a cloud engine constructs a historical steady state snapshot library offline, and an edge control node retrieves historical steady states locally according to the library and real-time working conditions; and furthermore, the edge control node takes the deviation dynamic state of the current state and the health characteristic value of the controlled equipment as feedback, and performs dynamic modulation and compensatory correction on the retrieved historical steady state so as to generate a final control instruction. According to the method, the system is converted into distributed local reproduction based on sharing history, system regulation and control oscillation caused by instruction time delay is avoided, meanwhile, the system is provided with the elastic adaptive capacity when dealing with unknown impact, self-sensing and closed-loop compensation of the physical state of equipment are achieved, and it is ensured that the control effect is accurately achieved.

Description

Technical Field

[0001] The present invention relates to a cloud-edge collaborative control system for an industrial park based on digital twins, and belongs to the technical field of industrial park control. Background Art

[0002] Currently, in real industrial park scenarios where power generation capacity is changing rapidly, the ideal low-latency, high-bandwidth communication assumptions between the cloud, edge, and end that the above architecture relies on are difficult to hold true in reality, exposing an architectural control mismatch problem. Specifically, when the cloud completes minute-level optimization calculations and issues instructions based on past aggregated data, the physical working conditions of the park may have changed again, causing the cloud instructions to become outdated, thereby exacerbating the chain reaction of system regulation and converting the theoretical energy-saving benefits into physical losses and operation and maintenance costs caused by frequent start-stop and fluctuations of equipment.

[0003] To address this challenge, the industry has tried to compensate by increasing edge computing power or optimizing communication protocols. However, these improvements still operate within the original control framework based on communication latency, and fail to change the situation where a tightly coupled physical system with millisecond-level response is constrained by minute-level decision cycles.

[0004] Specifically, the existing technology has the following deficiencies: 1. The decision-making basis of the control system lags behind the real-time state of the physical world in time, resulting in a mismatch risk between the control behavior and the control target; 2. In the absence of real-time global information, the local response behaviors of each edge node are separated from each other, making it difficult to form effective coordination and may even conflict with each other; 3. In order to avoid system instability, engineering practice requires sacrificing system operating efficiency in exchange for basic operating stability, resulting in the system being unable to operate within a more efficient range. Therefore, how to construct a control system that can allow all distributed edge nodes in the entire region to quickly converge to a verified operating state based only on locally available information without relying on high-cost real-time communication and centralized online optimization calculations, and on the basis of coordination, dynamically adjust the response intensity according to the disturbance to adapt to the impact, and compensate the control according to the health status of the equipment, has become the technical problem to be solved by the present invention. Summary of the Invention

[0005] The present invention provides an industrial park cloud-edge collaborative control system based on digital twins. Its main purpose is to solve the problem that the existing control architecture cannot achieve instantaneous communication between distributed nodes due to communication delays and computing lags, and lacks the ability to cope with unknown disturbances and equipment aging.

[0006] To achieve the above objectives, the present invention provides an industrial park cloud-edge collaborative control system based on digital twins, including: A cloud-based engine is configured to analyze the industrial park's historical operating data offline to extract and construct a steady-state snapshot library containing multiple steady-state snapshots, each of which consists of a working condition fingerprint vector representing a macro-working condition and an associated device state vector that records the target state of each key controllable device under that working condition; Multiple edge control nodes are set up in the industrial park. Each edge control node has a pre-stored steady-state snapshot library and is configured to: collect current environmental parameters in real time to form the current working condition fingerprint vector, and calculate the minimum deviation distance between the current working condition fingerprint vector and all working condition fingerprint vectors in the steady-state snapshot library. ; Furthermore, a spectrum analysis is performed on the load current waveform collected when a controlled electromechanical device is executing control to extract the health characteristic value; furthermore, based on the minimum deviation distance Search the steady-state snapshot library to determine the target state adjustment of the benchmark, and compare the target state adjustment of the benchmark with a minimum deviation distance. Acceleration over time The dynamic resilience adjustment coefficient determined by the edge control node and a compensation coefficient determined by the health eigenvalue are calculated to generate the final control instruction to control the device, so that all edge control nodes can achieve distributed self-consistent collaborative control based on the shared steady-state snapshot library.

[0007] Preferably, the cloud engine is configured to extract steady-state snapshots by performing K-means clustering analysis on historical operating data, wherein each cluster center corresponds to a working condition fingerprint vector, and the mean of the device state in the data set associated with the cluster center is determined as the device state vector.

[0008] Preferably, the operating condition fingerprint vector includes at least two of the following physical quantities: outdoor temperature, outdoor humidity, solar radiation intensity, time mark, and electricity price mark.

[0009] Preferably, when adjusting the device state according to the final control instruction, the edge control node further includes a smooth transition module, which is configured to make the device state gradually reach the target state along a trajectory defined by a ramp function within a certain time period.

[0010] Preferably, the edge control node is further configured to: when the calculated minimum deviation distance When the value of is continuously greater than a safety threshold stored in the edge control node, the controlled device is switched to a conservative operation mode that does not rely on the steady-state snapshot library, and an alarm message is sent to the cloud engine.

[0011] Preferably, the dynamic toughness adjustment coefficient is generated by calculating the minimum deviation distance in real time. Acceleration over time The absolute value of ; and, based on acceleration The absolute value of , real-time generation of dynamic toughness adjustment coefficient , the dynamic toughness adjustment coefficient The calculation method is: ,in, is a positive real number stored in the edge control node for adjusting the system response sensitivity.

[0012] Preferably, the edge control node collects the load current waveform in the following manner: reuse the power monitoring module in the industrial park, and activate high-frequency sampling of the load current waveform of the electromechanical equipment only when the edge control node executes the control instruction to change the operating state of the electromechanical equipment and during the stable period determined thereafter.

[0013] Preferably, the edge control node extracts the health characteristic value by performing a fast Fourier transform on the collected load current waveform to obtain its spectrum data, and extracting the energy value or energy proportion within one or more characteristic harmonic frequency bands predetermined by the cloud engine from the spectrum data, and using the energy value or energy proportion as the health characteristic value.

[0014] Preferably, the compensation coefficient is generated as follows: the cloud engine is also configured to build a device health baseline library, which stores the baseline health feature values ​​corresponding to healthy devices when performing different actions; the edge control node compares the health feature values ​​extracted in real time with the corresponding baseline health feature values ​​in the device health baseline library to determine the compensation coefficient.

[0015] Preferably, the edge control node also includes a decision arbitration module, which: stores a rule set containing multiple arbitration rules based on physical common sense or safety regulations; and is configured to use the arbitration rule set to perform a rationality check on the final control instruction before issuing the final control instruction to the device, and only allow execution if the instruction passes the check. If it fails the check, the instruction is rejected and the system is forced to enter a safe state.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The system control method of the present invention avoids the instruction delay problem caused by traditional centralized optimization calculations. The core task of its cloud engine is no longer online solution, but offline construction of a historical steady-state snapshot library covering various working conditions. The edge control nodes distributed in various locations independently query and approach the same historically verified collaborative operation state based on the shared steady-state snapshot library and the same environmental data collected in real time. As a result, the coordination of the entire system no longer depends on real-time cloud communication and instruction issuance that are susceptible to delay interference, but is realized in a distributed architecture, thereby avoiding control shocks and equipment losses caused by instructions lagging behind changes in working conditions.

[0017] 2. The present invention also provides a system resilience adjustment mechanism for dealing with unknown disturbances. When the system encounters a severe impact that has never occurred in history, the edge control node not only pays attention to the deviation distance between the current state and the historical steady state, but also pays attention to the acceleration of this deviation distance over time, and uses this acceleration as a direct basis for judging the severity of the external disturbance. Then, based on the numerical value of the acceleration, the control adjustment amount of its own output is subjected to real-time nonlinear dynamic convergence. This means that the system will automatically slow down the amplitude of its adjustment action at the moment it senses a severe impact, and absorb the impact energy in a flexible manner, rather than rigidly confronting it with a fixed response logic. This enables the system to switch from a rigid safety mode to a flexible adaptation with self-protection capabilities when facing unknown risks.

[0018] 3. The present invention integrates control and equipment health status perception without increasing additional hardware costs. When executing control instructions, the edge control node reuses the existing power monitoring module on site to collect the load current waveform of the controlled electromechanical equipment and perform spectrum analysis on it to extract health characteristic values ​​that can characterize its physical health status. This health characteristic value is not only used for fault alarm, but also serves as a key compensation coefficient and directly participates in the generation operation of the final control instruction, thereby performing feedforward compensation for the response deviation caused by equipment wear and aging. Its ultimate goal is to ensure that the physical effect of the equipment output is consistent with the system expectations, rather than just the precise issuance of electrical instructions. This extends the control loop from the instruction level to the effect level, solving the gap between the ideal control model and the real physical world caused by equipment degradation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the operation flow of the control system of the present invention; Figure 2 Schematic diagram for comparing the effects of the compensation mechanism of the present invention; Figure 3 Schematic diagram of the architecture deployment of the control system of the present invention. DETAILED DESCRIPTION

[0020] In order to make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Of course, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; it should be understood that the following embodiments are intended to explain the present invention rather than to limit the scope of protection of the present invention.

[0021] The present invention provides an industrial park cloud-edge collaborative control system based on digital twins, which is mainly composed of a cloud engine and multiple edge control nodes arranged in the industrial park; the core operation mode of the system is: the cloud engine is configured to analyze historical data offline to build a shared steady-state snapshot library, and multiple edge control nodes, without relying on real-time communication with the cloud, query the library locally and combine with real-time working conditions to achieve a distributed self-consistent collaborative control; in physical systems with multi-variable and tightly coupled characteristics such as industrial parks, the control architecture that relies on the cloud to perform minute-level online optimization calculations and then issue instructions often causes control instructions to lag behind instantaneous changes in physical working conditions due to communication and computing delays; to meet this challenge, the responsibility of the cloud engine of the present invention is configured to build a historical steady-state snapshot library offline, and the cloud engine collects historical operating data of the industrial park and executes historical operating data. For example, in the K-means cluster analysis, each cluster center that is finally formed is determined as a working condition fingerprint vector that characterizes a specific working condition. At the same time, the mean value of the device state in the data set associated with the cluster center is determined as the device state vector associated with the working condition fingerprint vector. In this way, each steady-state snapshot consists of a working condition fingerprint vector and a device state vector, and the collection of all these steady-state snapshots constitutes a lightweight steady-state snapshot library. The working condition fingerprint vector can specifically include at least two of the outdoor temperature, outdoor humidity, solar radiation intensity, time stamp, and electricity price stamp, while the device state vector records the target state of each key controllable device under the working condition. By transforming the core task of the cloud from online solution to offline construction of a historical knowledge base, the basis of control decision-making is transformed from online calculation relying on high-latency communication to local reproduction of verified global collaborative states.

[0022] Accordingly, in the absence of real-time global information, the local response behaviors of the various edge control nodes in the park are easily separated from each other. In view of this, each edge control node is configured to pre-store the same steady-state snapshot library built by the cloud engine. When the system is running, each edge control node collects the current environmental parameters in real time to form a current working condition fingerprint vector with the same dimension as the historical working condition fingerprint vector in the library. Then, the edge control node retrieves the vector with the minimum deviation distance from the current working condition fingerprint vector in the locally stored steady-state snapshot library through vector distance calculation, such as Euclidean distance calculation. The historical operating condition fingerprint vector is retrieved, and the device state vector associated with the retrieved historical operating condition fingerprint vector is read as the target state for controlling the connected device; since all edge control nodes share the same steady-state snapshot library and query based on the same real-time collected environmental data, they can converge to the same global operating state that has been historically verified to be coordinated at the same time without communicating with each other, thus forming distributed self-consistent collaborative control.

[0023] When the system encounters a severe shock that has never occurred in history, if the minimum deviation distance between the current working condition fingerprint and all historical fingerprints in the library is If the deviation distance is continuously greater than a preset safety threshold, the system will only switch to a conservative operation mode that does not make fine adjustments based on the real-time working conditions, and its adjustment ability will be limited. In order to make the system more adaptable when dealing with unknown disturbances, the edge control node is also configured to pay attention to the deviation distance when making control decisions. Dynamic characteristics that change over time. Specifically, the edge control node calculates the deviation distance in real time. Perform two consecutive difference operations on the time series to calculate the deviation acceleration The absolute value of , and then, based on the acceleration The absolute value of , a dynamic toughness adjustment coefficient is generated in real time , dynamic toughness adjustment coefficient The calculation method is ,in, is a preset positive real number, and its value can be determined by offline simulation after weighing the system response speed and overshoot suppression effect; It is used to dynamically modulate the retrieved target state adjustment. When the system is stable, Approaching zero, the coefficient Approaches 1, and when encountering a severe impact, Increase, coefficient The system automatically adjusts the amplitude of its regulating action to achieve a regulating mechanism in which the response intensity is adapted to the severity of the disturbance.

[0024] Traditional control models do not take into account the response deviation caused by physical wear and aging of equipment, which may lead to a gap between control instructions and physical effects. To compensate for this deviation, the edge control node is configured to reuse the existing power monitoring module in the industrial park when executing control instructions, and only activates high-frequency sampling of the load current waveform of the electromechanical equipment when the edge control node executes the control instruction to change the operating state of the electromechanical equipment and during the stable period determined thereafter. After collecting the load current waveform, the edge control node performs a fast Fourier transform on it to obtain its spectrum data, and extracts the energy value or energy proportion in one or more predetermined characteristic harmonic frequency bands from the spectrum data as a health feature value. Correspondingly, the cloud engine is also configured to build a device health baseline library, which stores the baseline health feature values ​​corresponding to healthy devices when performing different actions. The edge control node compares the real-time extracted health feature value with the corresponding baseline health feature value in the device health baseline library to determine a compensation coefficient. The compensation coefficient is directly The edge control node directly participates in the generation of the final control instruction, performing feedforward compensation for response deviations caused by device performance degradation, so that the physical effect of the device output approaches the system target. To further improve the system's operational stability and decision-making safety, when adjusting the device state based on the final control instruction, the edge control node also includes a smooth transition module. This module is configured to gradually bring the device state to the target state within a predetermined time period along a trajectory defined by, for example, a ramp function. Furthermore, the edge control node also includes a decision arbitration module. This module stores a rule set containing multiple arbitration rules based on physical common sense or safety regulations. Before issuing the final control instruction to the device, the module is configured to use the arbitration rule set to perform a rationality check on the final control instruction. For example, it checks whether cooling and heating instructions conflict. Execution is allowed only if the instruction passes the check. If not, the instruction is rejected and the system is forced into a safe state. This serves as a protection mechanism to prevent unreasonable decisions caused by upstream information contamination.

[0025] Example 1: In a specific industrial park application, photovoltaic power generation power drops instantaneously in the afternoon due to rapid cloud movement. Simultaneously, a key production line shuts down for unplanned maintenance, resulting in a sharp drop in the park's overall power load. The combination of these two events poses a direct challenge to the stability of the park's power grid and the economic operation of the energy system. When this condition occurs, all edge control nodes deployed within the park, including those controlling photovoltaic inverters, central air conditioning systems, and energy storage units, collect parameters representing changes in the external environment in real time and independently generate a local fingerprint vector corresponding to this condition. These edge control nodes calculate the deviation between the current fingerprint vector and all historical fingerprint vectors in a locally stored steady-state snapshot library, and retrieve the same historical steady-state snapshot associated with low solar irradiation intensity and low production load. This mechanism provides all distributed nodes with a unified global target operating state verified as coordinated by historical operating data, and transforms the control problem from how to online calculate an unknown optimal coordination strategy to how to approach a known optimal coordination history.

[0026] Furthermore, since the change of working conditions is sudden, the minimum deviation distance calculated by each edge control node is The time series shows an accelerating trend, which makes the distance The acceleration determined by the second derivative of The absolute value of Instantaneous increase, according to the dynamic toughness adjustment coefficient The generation method of , a larger The value will result in a coefficient The value of converges instantaneously to a value less than 1. At this time, although each edge control node has retrieved the target device state vector, the baseline target state adjustment amount generated based on it is multiplied by the dynamic resilience adjustment coefficient before being sent to the device. As a result, the actual execution amplitude of a series of adjustment actions such as the power reduction of the photovoltaic inverter and the power reduction of the air-conditioning system are instantaneously converged, and the system responds to the shock in a more moderate way, thus achieving a balance between the response speed when dealing with disturbances and maintaining system stability.

[0027] At the same time, after the control command is issued, an edge control node that controls the central air-conditioning compressor collects the load current waveform of the compressor by reusing the power monitoring module, and extracts its health characteristic value through fast Fourier transform. After comparing this health characteristic value with the baseline value in the equipment health baseline library, the node finds that the compressor has a slight performance degradation due to long-term operation, and generates a compensation coefficient based on this. This compensation coefficient is combined with the dynamic resilience adjustment coefficient The target state adjustment amount acts together on the benchmark to generate the final control instruction. The existence of this mechanism makes the adjustment of the equipment state not only take into account the drastic changes in external working conditions, but also feedforward into account the physical health status of the equipment itself. The synergy of the two mechanisms adjusts the overall response behavior of the system in response to external shocks and the execution effect of specific instructions to the equipment. Ultimately, the overall operating state of the park smoothly transitions to a new equilibrium point, which matches the historically efficient operating state. This process avoids system oscillations and equipment losses that may be caused by excessive adjustment. The coordinated control and adaptive adjustment of the entire system are achieved through shared historical data and unified dynamic response rules without communication interaction among all edge nodes.

[0028] Example 2: In order to quantitatively verify the system response characteristics and stability of the technical solution of the present invention in response to sudden disturbances, a set of simulation test platforms based on hardware in the loop was built. The platform reproduced an industrial park power grid model including photovoltaic power generation units, variable industrial loads and energy storage units through a real-time simulator, and connected the physical edge control node hardware running the control logic of the present invention to the closed-loop system. At the same time, a control group using the traditional hierarchical optimization control method was set up, and a 60-second communication and computing delay was set between the cloud and the edge node of the control group; in the experiment, the dynamic resilience adjustment coefficient in the edge control node of the technical solution of the present invention was adjusted. Key parameters The value is set to balance the rapidity and stability of the system response. The specific value is determined by an offline calibration procedure. This procedure imposes a series of standardized step disturbances on the simulation model to minimize the key state variables of the system, such as the integral of the fluctuation of the power exchanged between the park and the large power grid. In this experiment, for the constructed industrial park model, The value of is determined to be 0.5.

[0029] After the test process was started, at the 100th second, a disturbance of a 50% step drop in solar irradiance simulating cloud cover and a 40% step drop in load simulating the offline of a key production line were simultaneously applied to the simulation system; in the control group, after the disturbance occurred, the exchange power between the park and the large power grid generated a reverse power shock because the local response lagged behind the load change, and a large overshoot occurred after the cloud optimization instruction was issued. The system power fluctuation continued for a long time after the disturbance before converging; in the experimental group running the technical solution of the present invention, when a disturbance occurred, all edge control nodes almost simultaneously obtained the coordinated target state based on the local query steady-state snapshot library, and calculated the dynamic resilience adjustment coefficient according to the severity of the disturbance, adjusted the control output, and the fluctuation of the system exchange power was suppressed to a smaller range and quickly converged to a new steady state.

[0030] Table 1: Comparison of key performance indicators between the experimental group and the control group after the disturbance

[0031] As shown in Table 1, the peak power overshoot and power fluctuation convergence time of the experimental group are significantly lower than those of the control group. The reason is that the control command of the control group has a minute-level delay, which causes the command to act on a changed system state, thereby exacerbating the system oscillation. The experimental group avoids this delay by locally querying the steady-state snapshot library and achieves near real-time response. At the same time, the deviation acceleration in the experimental group is The peak value is much larger than that of the control group, reflecting the system's rapid detection of disturbance changes and thus generating a dynamic resilience coefficient less than 1. , actively reduced the intensity of the control action, which was also confirmed by the data that the total variance of the key equipment adjustment action in the experimental group was much smaller than that in the control group.

[0032] Example 3: This example combines Figures 1 to 3 , describes the cloud-edge collaborative control system of the industrial park based on digital twins, such as Figure 1 As shown in the figure, the process starts with historical operating data and performs offline analysis and modeling through the cloud engine. Specifically, the cloud engine uses historical park temperature, load, equipment status and other data to perform offline analysis through algorithms such as K-means clustering, thereby constructing a steady-state snapshot library and an equipment health baseline library respectively. The steady-state snapshot library contains a series of snapshots consisting of working condition fingerprints and associated equipment state vectors, while the equipment health baseline library stores the characteristic values ​​of healthy equipment under different working conditions; in the real-time operation stage of the system, the edge control node collects real-time environmental parameters such as temperature, humidity, radiation intensity, etc. to generate a current working condition fingerprint vector with the same dimension as the historical library, and retrieves the historical steady-state snapshot based on this, and calculates the minimum deviation distance from the best matching snapshot in the library. At the same time, the system continues to calculate the deviation distance Acceleration over time , to generate a dynamic resilience adjustment coefficient for responding to impacts. On the other hand, when the system controls electromechanical equipment such as air conditioners, inverters, and energy storage units, it collects their load current waveforms and performs spectrum analysis to extract the current health characteristic value, and compares the characteristic value with the equipment health baseline library to generate a compensation coefficient. Finally, the final control instruction that integrates dynamic adjustment and health compensation is generated. After the decision arbitration module verifies the rationality and safety of the rule set, and after progressive execution through the smooth transition module using ramp functions and other methods, it is issued to the electromechanical equipment.

[0033] like Figure 2As shown in the figure, the horizontal axis represents the degree of degradation of the device performance, increasing from 0% to 30%, and the vertical axis represents the output efficiency of the device. The figure contains three curves, which represent the actual output when the target output is not compensated and the actual output when it is compensated. The target output curve is a horizontal line at 100% efficiency, representing the expected performance of the system, while the actual output curve without compensation is a dotted line that drops sharply as the device performance degrades, indicating that if no compensation is applied, the actual output efficiency of the device will be far lower than the target. In contrast, the actual output curve with compensation is a solid line that is always closely close to the target output within the entire device performance degradation range. It intuitively proves that the compensation coefficient mechanism of the present invention can effectively offset the performance degradation caused by physical aging and wear of the equipment, ensuring that the final physical execution effect is consistent with the expected target of the control system.

[0034] like Figure 3 As shown in the figure, the architecture diagram clearly shows the two core components of the system: CloudSide and EdgeSide. In the cloud, computing power is supported by a cloud server cluster, on which is deployed a cloud engine as the core logic unit, a historical database for storing historical data, and a device health baseline library for storing device health records. The core task of the cloud is offline analysis and modeling. On the edge, multiple edge control gateways are deployed, such as Edge Control Gateway #1 and Edge Control Gateway #2 shown in the figure. Each gateway is deployed with local edge control logic and a copy of the steady-state snapshot library issued by the cloud. The edge control gateway is directly connected to and controlled by on-site physical equipment such as central air conditioners and photovoltaic inverters through real-time control buses such as Modbus or OPCUA. This distributed architecture enables each edge node to make local decisions independently and quickly, realizing decoupling from real-time communication in the cloud.

[0035] Example 4: When the control system of the present invention is deployed in an industrial park for the first time or when a key electromechanical equipment is newly added to the park, in order to make the operating parameters of the system deterministic and based on field data, it is necessary to perform initialization and parameter calibration procedures; for the safety threshold used to judge unknown working conditions in the edge control node, the calibration process is to first use the cloud engine to process the historical operating data of at least one complete operating year of the park, and extract a set of working condition fingerprint vectors covering all time; secondly, for each working condition fingerprint vector in the set, calculate its minimum deviation distance from all other vectors in the set ; Finally, for all the calculated The value is statistically analyzed, and the value corresponding to the 99.9th percentile of the distribution is set as the safety threshold stored in the edge control node for triggering the conservative operation mode.

[0036] For the functional modules related to device health status perception in the edge control node, their initialization and calibration process takes a newly added variable-frequency water pump as an example. When the new equipment is installed and in its initial health state, the system enters a device baseline construction mode; in this mode, the edge control node controls the water pump to operate stably at a set of preset speed points covering its normal operating range. At each speed point, the edge control node uses a reused power monitoring module to collect the load current waveform of the water pump at high frequency and performs rapid transformation on the collected waveform data. By comparing the current spectra obtained at different speeds, it identifies those harmonic frequency bands where energy is concentrated and stable at a specific speed but changes regularly with speed changes, and determines these frequency bands as characteristic harmonic frequency bands used to characterize the health status of the water pump; then, the system uses the energy integrated average in these characteristic harmonic frequency bands when the water pump is stably operating at each benchmark speed point as the baseline health characteristic value, and stores it in the device health baseline library of the edge control node together with the corresponding speed information.

[0037] In addition, the compensation coefficient used to compensate the control instruction is also generated in a deterministic way. When the water pump is in daily operation, the edge control node extracts its health feature value in real time. Query the corresponding baseline health characteristic value under the current operating state from the equipment health baseline library In this specific configuration, the health characteristic value is positively correlated with the equipment operating efficiency, and the compensation coefficient is defined as ; This coefficient is then used to adjust the control instructions issued to the water pump. If the real-time health characteristic value decreases, the compensation coefficient will be greater than 1, thereby increasing the output intensity of the instruction to compensate for the physical effect deviation caused by the decline in equipment performance.

[0038] Example 5: When the system of the present invention is deployed in a newly built industrial park that lacks long-term historical operating data, in order to achieve immediate activation and subsequent self-optimization of the system, the cloud engine first sends a universal steady-state snapshot library to all edge control nodes in the park. The universal library is based on the historical data of multiple stably operating industrial parks with similar industrial characteristics, and is integrated and constructed through clustering analysis. This provides an initial operating basis for the new park, enabling it to have basic collaborative control capabilities; on this basis, the cloud engine is configured to continuously collect the actual operating data of the new park itself, and when the accumulated data volume reaches a preset threshold, it automatically triggers a clustering analysis of the park's own data to generate a steady-state snapshot library based on the park's own data, and use it to replace the original universal library in the edge control node.

[0039] While carrying out the above deployment, to ensure that control behavior complies with the safety regulations of the specific site, the decision-making arbitration module in the edge control node needs to be configured. Each edge control node is pre-installed with a template library containing multiple basic physical and electrical safety arbitration rules at the factory. During the on-site commissioning phase, engineering technicians can use the configuration interface to select and activate applicable rules from this template library based on the specific equipment models installed in the new park, the process constraints, and safety management regulations. They can also add additional arbitration rules for the site's unique operating conditions. This supplementary capability allows the specific operating boundary conditions of specific equipment, such as large-capacity battery energy storage systems, such as the rule prohibiting charging when the grid voltage is abnormal, to be included in the decision-making arbitration module's verification range. After this configuration, the decision-making arbitration module has the deterministic instruction verification capability that conforms to the safety boundaries of the specific site.

[0040] Example 6: In executing the procedure of constructing a steady-state snapshot library through K-means cluster analysis, the cloud engine first performs data preprocessing and model hyperparameter optimization steps; in the data preprocessing stage, the cloud engine verifies the integrity and consistency of the collected historical operation data, and removes invalid data segments where the sensor readings exceed their physical range or the values ​​remain unchanged for a long time; in the hyperparameter optimization stage, in order to determine the number of clusters Value, the cloud engine for a preset The K-means clustering algorithm is executed multiple times, and the silhouette coefficient is calculated for each clustering result. Finally, the corresponding value that makes the global silhouette coefficient reach the maximum value is selected. value, used as the number of clusters for constructing the final steady-state snapshot library.

[0041] Similarly, the edge control node also performs the aforementioned validity check on the sensor data before collecting the current environmental parameters in real time to form the current working condition fingerprint vector. When an abnormality is detected in a key sensor data, the node will temporarily use a moving average based on the valid data of the sensor in the past short period of time as a substitute to maintain the continuity of the control function. In addition, by comparing the health feature value and baseline health characteristics To generate the compensation coefficient The generation method is further configured as follows: in the equipment baseline construction mode, by injecting simulated efficiency decay at multiple operating points, the nonlinear relationship between the health characteristic value and the required compensation strength of the equipment during the whole cycle from health to decay is calibrated, and the model parameters of the relationship are stored in the equipment health baseline library for real-time calculation. When called.

[0042] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. The cloud-edge collaborative control system for industrial parks based on digital twins is characterized by: include: A cloud-based engine is configured to analyze the industrial park's historical operating data offline to extract and construct a steady-state snapshot library containing multiple steady-state snapshots, each of which consists of a working condition fingerprint vector representing a macro-working condition and an associated device state vector that records the target state of each key controllable device under that working condition; Multiple edge control nodes are set up in the industrial park. Each edge control node pre-stores a steady-state snapshot library and is configured to: collect current environmental parameters in real time to form a current working condition fingerprint vector, and calculate the minimum deviation distance between the current working condition fingerprint vector and all working condition fingerprint vectors in the steady-state snapshot library; In addition, a spectral analysis is performed on the load current waveform collected by a controlled electromechanical device when executing control to extract the health characteristic value; then, based on the minimum deviation distance, a steady-state snapshot library is retrieved to determine the benchmark target state adjustment amount, and the benchmark target state adjustment amount is respectively calculated with a dynamic toughness adjustment coefficient determined by the acceleration of the minimum deviation distance over time and a compensation coefficient determined by the health characteristic value to generate the final control instruction to control the device, so that all edge control nodes realize distributed self-consistent collaborative control based on the shared steady-state snapshot library.

2. The digital twin-based industrial park cloud-edge collaborative control system according to claim 1, characterized in that: The cloud engine is configured to extract steady-state snapshots by performing K-means clustering analysis on historical operating data, where each cluster center corresponds to a working condition fingerprint vector, and the mean of the device state in the data set associated with the cluster center is determined as the device state vector.

3. The cloud-edge collaborative control system for industrial parks based on digital twins according to claim 1 is characterized in that: The working condition fingerprint vector includes at least two of the following physical quantities: outdoor temperature, outdoor humidity, solar radiation intensity, time mark, and electricity price mark.

4. The digital twin-based industrial park cloud-edge collaborative control system according to claim 1, characterized in that: When adjusting the device state according to the final control instruction, the edge control node further includes a smooth transition module, which is configured to make the device state gradually reach the target state along a trajectory defined by a ramp function within a certain time period.

5. The cloud-edge collaborative control system for industrial parks based on digital twins according to claim 1, characterized in that: The edge control node is further configured to: when the calculated minimum deviation distance When the value of is continuously greater than a safety threshold stored in the edge control node, the controlled device is switched to a conservative operation mode that does not rely on the steady-state snapshot library, and an alarm message is sent to the cloud engine.

6. The digital twin-based industrial park cloud-edge collaborative control system according to claim 1, characterized in that: The dynamic toughness adjustment coefficient is generated by calculating the minimum deviation distance in real time. Acceleration over time The absolute value of ; And, based on the acceleration The absolute value of , real-time generation of dynamic toughness adjustment coefficient , the dynamic toughness adjustment coefficient The calculation method is: ,in, is a positive real number stored in the edge control node for adjusting the system response sensitivity.

7. The cloud-edge collaborative control system for industrial parks based on digital twins according to claim 1, characterized in that: The edge control node collects the load current waveform by reusing the power monitoring module in the industrial park and activating high-frequency sampling of the load current waveform of the electromechanical equipment only when the edge control node executes a control instruction to change the operating state of the electromechanical equipment and during the stable period determined thereafter.

8. The digital twin-based industrial park cloud-edge collaborative control system according to claim 7, characterized in that: The edge control node extracts the health characteristic value by performing a fast Fourier transform on the collected load current waveform to obtain its spectrum data, and extracting the energy value or energy proportion within one or more characteristic harmonic frequency bands predetermined by the cloud engine from the spectrum data, and using the energy value or energy proportion as the health characteristic value.

9. The cloud-edge collaborative control system for industrial parks based on digital twins according to claim 1, characterized in that: The compensation coefficient is generated as follows: the cloud engine is further configured to build a device health baseline library, which stores baseline health feature values ​​corresponding to healthy devices when performing different actions; The edge control node compares the health feature values ​​extracted in real time with the corresponding baseline health feature values ​​in the device health baseline library.

10. The cloud-edge collaborative control system for industrial parks based on digital twins as claimed in claim 1 is characterized in that: The edge control node also includes a decision arbitration module, which stores a rule set containing multiple arbitration rules based on physical common sense or safety regulations; and is configured to use the arbitration rule set to perform a rationality check on the final control instruction before issuing it to the device, and only allow execution if the instruction passes the check. If it fails the check, the instruction is rejected and the system is forced to enter a safe state.

Images