Water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system
By dynamically correcting the parameters of the pump group model and quantifying energy consumption and disturbance costs, the problems of decreased accuracy and narrow optimization objectives of the pump group model were solved, thus achieving stable and efficient operation of the pump group.
Patent Information
- Application Number
- CN202511511768.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-22
- Publication Date
- 2025-11-28
AI Technical Summary
The existing digital twin models of water pump groups suffer from decreased accuracy due to wear and other factors during long-term operation, leading to the failure of optimization decisions. Furthermore, the narrow optimization objectives result in frequent state switching, causing disturbances that affect system stability and equipment lifespan.
The model parameters are dynamically corrected by an online identification and self-healing model construction module, and the steady-state energy consumption and instantaneous switching cost are quantified by a composite target generation module. The optimal state point is found by a global optimal operating state solution module, and the safe control trajectory is planned by a disturbance-free switching path planning module.
It achieves continuous accuracy of model precision and stability of optimization decisions, balancing energy saving and stability, obtaining sustainable energy-saving effects and improving system operational reliability.
Smart Images
Figure CN121028566A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of automation control technology, in particular to a water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system. BACKGROUND
[0002] Water pumps are common fluid conveying equipment in industrial and municipal fields, and their energy consumption accounts for a considerable proportion of the total operating cost. To reduce energy consumption, existing technologies have adopted energy-saving control systems based on digital twin models and optimization algorithms. These systems establish mathematical models of water pump groups and use optimization algorithms to find the combination of operating parameters with the lowest energy consumption.
[0003] However, the existing technology has a fundamental technical defect in achieving long-term effective energy-saving control: the digital twin model on which its decisions depend is static. In the long-term operation process, the physical equipment will experience performance degradation due to factors such as wear and tear, pipe wall fouling, etc., resulting in a gradually increasing deviation between the digital twin model and the physical entity. This continuous decline in model accuracy fundamentally undermines the effectiveness of any optimization decisions made based on the model.
[0004] Further, the objective function of its optimization algorithm is usually limited to minimizing steady-state operating energy consumption. This narrow optimization objective will drive a system, even with an accurate model, to generate control instructions that require frequent and drastic state switching in pursuit of theoretical energy efficiency optimization. This dynamic process of state switching itself generates significant transient disturbances such as pressure shocks, current spikes, and equipment wear in the pipe network. These disturbance costs, which are not quantified and considered in the optimization by existing technologies, offset the theoretical energy-saving benefits in practice and have a negative impact on the system's operational stability and equipment life. SUMMARY
[0005] The present application provides a water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system to solve the technical problems of unsustainable energy-saving effect and poor operational stability caused by the inability to sequentially address the long-term decline in model accuracy and the narrow optimization objective in existing technologies.
[0006] In view of the above problems, the present application provides a water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system, which comprises: an online identification and self-healing model construction module configured to receive real-time operating data streams of a physical pump group and synchronous prediction data streams of a current digital twin model, and output a high-fidelity digital twin model dynamically corrected for key physical degradation parameters and a model health index representing the current model's credibility; a composite target generating module, which is operatively connected to the online identification and self-healing model construction module and is configured to receive the high-fidelity digital twin model, the model health index, and external working condition requirements, and output a composite optimization target function including a steady-state energy consumption item and a transient switching cost item; a globally optimal operating state solving module, which is operatively connected to the composite target generating module and the online identification and self-healing model construction module and is configured to receive the composite optimization target function and the high-fidelity digital twin model, and output a target operating state point that is optimal under the composite optimization target; a disturbance-free switching path planning module, which is operatively connected to the globally optimal operating state solving module and is configured to receive the optimal target operating state point and the current actual operating state of the physical pump group, and output a set of control trajectory instructions to a control execution unit of the water pump group.
[0007] The technical scheme provided in the application has at least the following technical effects or advantages: First, the online identification and self-healing model construction module solves the prerequisite problem of ensuring the long-term accuracy of decision-making. By using the residual error between model prediction and actual data, the module can identify and correct the physical parameters of the model in reverse, thereby suppressing the decline in model accuracy caused by equipment performance degradation, and providing a sustainable and reliable technical foundation for subsequent optimization decisions.
[0008] On the basis of the above technical effects, the composite target generating module solves the deep-seated problem of narrow optimization targets. By quantifying the transient disturbance generated during state switching as a calculable switching cost item and dynamically weighting it with the steady-state energy consumption item, the module constructs a comprehensive optimization target, so that the subsequent optimization decisions can take into account both energy saving and stability.
[0009] Finally, through the collaborative work of the above modules, comprehensive energy-saving and stability control is achieved. The application can optimize a comprehensive optimization target under the premise of ensuring the accuracy of decision-making, and plan a safe execution path, thereby achieving sustainable energy-saving effect and improving the reliability of system operation. BRIEF DESCRIPTION OF DRAWINGS
[0010] Figure 1 The structure block diagram of the water pump group operating efficiency dynamic modeling and energy-saving control parameter self-optimization system provided in the embodiments of the application. DETAILED DESCRIPTION
[0011] The above technical solutions will be described in detail below in combination with the accompanying drawings and specific embodiments, so that the above technical solutions can be better understood. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments for explaining the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application. In addition, it should be noted that, for the convenience of description, only the parts related to the present application are shown in the drawings, not all.
[0012] Please refer to Figure 1 , the water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system, comprising: An online identification and self-healing model construction module is configured to receive real-time operation data streams of a physical pump group and synchronous prediction data streams of a current digital twin model, and output a high-fidelity digital twin model dynamically corrected with key physical attenuation parameters and a model health index representing the credibility of the current model; A composite target generation module is operatively connected to the online identification and self-healing model construction module and is configured to receive the high-fidelity digital twin model, the model health index, and external operating condition requirements, and output a composite optimization target function containing steady-state energy consumption items and instantaneous switching cost items; A global optimal operating state solving module is operatively connected to the composite target generation module and the online identification and self-healing model construction module and is configured to receive the composite optimization target function and the high-fidelity digital twin model, and output an optimal target operating state point under the composite optimization target; A disturbance-free switching path planning module is operatively connected to the global optimal operating state solving module and is configured to receive the optimal target operating state point and the current actual operating state of the physical pump group, and output a set of control trajectory instructions to the control execution unit of the water pump group.
[0013] The specific implementation of the water pump group operation efficiency dynamic modeling and energy-saving control parameter self-optimization system disclosed by the present application is as follows.
[0014] The full-process working mode of the system starts from continuous online identification of the physical pump group and self-healing correction of the model. The technical goal of this stage is to build and continuously maintain a high-fidelity digital twin model that can accurately reflect the real state of the physical world.
[0015] To achieve this goal, a data acquisition interface module embedded in the system collects real-time operating data streams in parallel from multiple physical measurement points of the water pump group at a preset sampling frequency of no less than 10 Hz through the industrial field bus protocol. The real-time operating data stream at least includes: the current operating frequency and output current value fed back by the frequency converter of each water pump, and the real-time pressure and flow values fed back by the pressure sensor and flow sensor installed on the key nodes of the pipe network. The data acquisition interface module will also apply a high-precision time stamp to the multi-channel data collected for each frame to ensure the time synchronization of subsequent data processing.
[0016] After obtaining the real-time operating data stream with a synchronized time stamp, the online identification and self-healing model construction module immediately drives the current digital twin model to perform synchronous prediction. The specific implementation path of this process is: the module takes the control input type data in the real-time operating data stream, i.e., the current operating frequency of each water pump, as the input variable of the digital twin model. The digital twin model is a mathematical model that couples the performance curve equation describing the relationship between the water pump head and flow and the fluid dynamics equation describing the pressure loss of the pipe network.
[0017] Specifically, the mathematical model is composed of the following nonlinear equation set: The "performance curve equation describing the relationship between the water pump head and flow" for the i-th water pump in the pump group has the mathematical expression: wherein, H i is the outlet head of the i-th water pump, Q i is the instantaneous flow of the i-th water pump, f i is the normalized operating frequency of the i-th water pump (i.e., the ratio of the actual operating frequency to the rated frequency), C i is the inherent performance coefficient of the i-th water pump obtained through factory testing or on-site calibration.
[0018] The "fluid dynamics equation describing the pressure loss of the pipe network" has the mathematical expression: wherein, H tot is the total head required by the pipe network system to meet the demand, H stat is the static head (or static water head) of the pipe network, Q tot is the total flow of the pipe network, i.e., the sum of the flows of all operating water pumps. The "global resistance coefficient value" (i.e., C tot) in the "fluid dynamics equation" is a comprehensive physical quantity that comprehensively reflects the characteristics of the along-path resistance and local resistance of all pipe lengths, pipe diameters, bends, valves, etc. in the pipe network, and its initial value can be set through fluid mechanics theory calculation or empirical data.
[0019] When the system is running stably, the total head provided by the pump group is balanced with the total head required by the pipeline network. That is, for each pump operating in parallel, its outlet head is equal to and equal to the total head of the pipeline network system. ).
[0020] The "Online Identification and Self-Healing Model Construction Module" obtains predicted values by solving the aforementioned system of equations. Since this system of equations is nonlinear, the "Iterative Numerical Solution Method" specifically employs the Newton-Raphson method. The specific steps are as follows: First, an initial flow rate value is set based on the current operating frequency. Then, a Jacobian matrix is constructed, and the flow rate solution is continuously corrected through iterative calculations until the residuals of each equation are less than a preset convergence threshold, thereby obtaining the uniquely determined predicted flow rate and predicted pressure values at that operating frequency. The predicted flow rate and predicted pressure values, together with the collected timestamps, constitute the synchronous prediction data stream. The output of this step, namely the real-time operating data stream and the synchronous prediction data stream, will serve as the technical input for the next step, namely model residual calculation and parameter back-identification.
[0021] After successfully generating the synchronous prediction data stream, the online identification and self-healing model building module then performs inverse parameter identification based on model residuals. Its technical goal is to quantify the deviation between the model and reality and diagnose the physical causes of this deviation. The specific calculation steps are as follows: First, the module calculates a set of model residual signals by subtracting the corresponding items of the synchronous prediction data stream and the real-time running data stream at the same timestamp. Next, the module calls a preset "model residual-physical attenuation" knowledge base. The knowledge base is a set of decision rules that take the statistical characteristics of the residual signals as input and output the correction direction and magnitude of specific physical attenuation parameters. The query mechanism is implemented by the module continuously calculating multiple statistical characteristics of the model residual signals within a sliding time window, including but not limited to mean, variance, and skewness. Subsequently, the module uses these statistical characteristics as a combination condition and matches them with multiple "IF-THEN" rules stored in the knowledge base. For example, a rule in the knowledge base can be defined as: "IF the mean flow residual is consistently less than -5% and the mean pressure residual is consistently greater than +3% in the past hour, THEN the diagnostic event is 'increased pipe friction resistance', and the instruction 'increase the global resistance coefficient in the pipe network model by 0.1%' is output." When the statistical characteristics calculated by this module meet the IF condition of a certain rule, a match is completed, and specific instructions for correcting key physical attenuation parameters are obtained.
[0022] After identifying the physical decay events causing the deviation and obtaining correction instructions through the "model residual-physical decay" knowledge base, the online identification and self-healing model building module ultimately performs dynamic correction and health assessment of the model parameters. The specific implementation path is as follows: based on the received correction instructions, this module directly accesses and modifies the corresponding parameter values in the high-fidelity digital twin model. For example, for the instruction to "increase the global resistance coefficient in the pipeline model by 0.1%", this module performs a multiplicative update on the global resistance coefficient value in the fluid dynamics equation. In this way, the digital twin model completes the dynamic correction of its parameters, thus enabling it to output more realistic predictions in the next calculation cycle. Simultaneously, to quantify the overall reliability of the current model, this module also performs parallel calculations of the model health index. A specific calculation method is that within the same sliding time window, the module first calculates the root mean square error (RMSE) of the flow residual signal, and then divides this RMSE value by the average value of the actual flow measurements within that time window, thereby obtaining a normalized RMSE value.
[0023] To resolve the intuitive contradiction between the normalized root mean square error (the smaller the better) and the model health metric (the larger the expected value, the better), and to normalize its value range to between 0 and 1, this invention defines the model health metric as follows: The normalized root mean square error (denoted as ) is expressed by the following mathematical formula. The calculation yielded: in, This is the final output model health index, with values ranging from 0 to 1, where 1 represents an extremely healthy model. exp() is the natural exponential function; c is an adjustable model health sensitivity coefficient used to adjust the degree of penalty of error on health indicators. Its empirical value is greater than 0. In this embodiment, c=5 can be taken.
[0024] The final output of this step, a dynamically corrected high-fidelity digital twin model and a quantified model health index, will serve as the technical input for the subsequent dynamic generation stage of composite optimization objectives.
[0025] After obtaining an accurate and continuously self-healing high-fidelity digital twin model, the system then enters the dynamic generation stage of composite optimization objectives. The technical goal of this stage is to construct an optimization function that can simultaneously achieve long-term energy saving and instantaneous stability.
[0026] To this end, the composite target generation module within the system first performs predictive calculations of steady-state energy consumption. Specifically, this module calls the high-fidelity digital twin model output from the previous stage and, through simulation calculations, predicts the steady-state operating power values corresponding to the stable operation of the pump group under multiple potential target operating states. The implementation path of this prediction process is as follows: for each potential target operating state generated by the subsequent optimization algorithm (this state defines the start-up, shutdown, and specific speed of each pump), this module uses these speed values as input to the high-fidelity digital twin model and solves the fluid dynamics equations within the model to obtain the stable flow rate and head under that state. Subsequently, based on the pump efficiency curve model, the module calculates the mechanical power required to drive the pump at that flow rate and head, and further combines it with the motor efficiency model to finally calculate the corresponding total input electrical power. This total input electrical power is defined as the steady-state operating power of that target operating state. By performing this calculation for all potential target operating states, the module obtains a set of steady-state operating power values. These steady-state operating power values will serve as the first basic component for subsequently constructing the composite optimization objective function.
[0027] Meanwhile, to quantify the instantaneous disturbances during the transition from the current state to different target states, the composite target generation module performs parallel quantitative calculations of the instantaneous switching cost term. This calculation process first requires determining a basic switching intensity. The method for calculating this basic switching intensity is to quantify the operational differences between each potential target operating state and the current operating state. The specific calculation steps are as follows: First, identify the sets of pumps whose operating state changes from "stopped" to "operating" during the switching process, and the sets of pumps whose operating state changes from "operating" to "stopped," and count the total number of pumps in these two sets. Second, calculate the sum of the absolute values of the changes in operating frequency of all pumps during this switching process. Finally, using a preset weighted summation function, sum the previously counted total number of pumps started and stopped and the absolute values of their frequency changes.
[0028] Meanwhile, to address the issue that the physical meaning, unit, and order of magnitude of the "total number of pumps started and stopped" (unit: pumps) and the "absolute value of frequency change" (unit: Hz) are different, resulting in unclear physical meaning when directly weighted summing, a normalization process is used before weighted summation to convert both into dimensionless relative values. The specific calculation steps are as follows: Step 1: Calculate the normalized value of the number of start-stop operations. The total number of water pumps whose status changed during the switching process (from "Running" to "Stopped" or from "Stopped" to "Running"). Divide by the total number of pumps in the pump group : .
[0029] Step 2: Calculate the normalized value of the frequency variation. The sum of the absolute values of the changes in the operating frequency of all water pumps ( Divide by a reference frequency variation (e.g., the total number of pumps in the pump group multiplied by the rated frequency of a single pump). ), to reflect its relative range of change: .
[0030] Step 3: Sum the two dimensionless normalized values with weights to obtain the basic switching intensity. .
[0031] in, The preset weighting coefficients, This is used to adjust the relative importance of start-stop operations and frequency regulation in disturbance assessment.
[0032] After determining the basic switching intensity, this module also needs to calculate a network vulnerability index. This index characterizes the sensitivity of the current network operating conditions to disturbances. The calculation method involves using a high-fidelity digital twin model to perform a rapid virtual disturbance response simulation. Specifically, in the digital twin model, the module applies a preset, standardized virtual control disturbance to a currently operating main pump, such as a 1% instantaneous frequency step increase lasting 0.5 seconds. Subsequently, the module analyzes and records the response curves of the pressure values at key network nodes output by the model simulation, and extracts the maximum fluctuation amplitude of the pressure response curve within a predetermined time window. This maximum pressure fluctuation amplitude obtained through virtual disturbance simulation is quantified and defined as the current network vulnerability index. A higher index value indicates that the current network state is more sensitive to control disturbances. These two calculated basic switching intensity and network vulnerability index will jointly serve as inputs for the next step of generating the final instantaneous switching cost.
[0033] After calculating the basic switching intensity and the network vulnerability index, this module then generates the final instantaneous switching cost item through a non-linear amplification method. The design idea behind this non-linear amplification mechanism is that when the network vulnerability index is high, even a small basic switching intensity should result in a sharply amplified switching cost. In a preferred embodiment of the invention, this non-linear amplification is achieved through an exponential function. The specific calculation process includes: substituting the basic switching intensity and the network vulnerability index into a preset exponential function used to non-linearly amplify the impact of the network vulnerability index to generate the final instantaneous switching cost item. This instantaneous switching cost item... It can be calculated using the following mathematical formula: in, The basic switching intensity calculated above, The aforementioned calculated pipeline vulnerability index uses β as a weighting factor that can be adjusted by the user to calibrate the degree of vulnerability impact. A specific numerical example is given: if the calculated... The calculated value is 20.5. Given a β value of 0.6 and a user-defined β value of 0.8, the calculated instantaneous switching cost is: =20.5*exp(0.6*0.8)=20.5*exp(0.48)≈33.15. As an alternative implementation, the nonlinear amplification here can also be achieved through a polynomial function, which can also achieve a nonlinear weighting effect on the pipeline vulnerability index.
[0034] After obtaining the steady-state energy consumption and instantaneous switching cost terms, the composite target generation module finally performs adaptive adjustment of the dynamic balance weights and constructs the final objective function. This process embodies the intelligent inter-module linkage in the architecture of this invention. The specific implementation steps are as follows: First, the module receives the previously calculated model health index from the online identification and self-healing model construction module in real time. Then, the module substitutes the received model health index into a preset mathematical function to calculate a dynamic balance weight factor. The design idea of this mathematical function is that when the model health is low, the weight used to penalize instantaneous switching costs should be automatically increased, making the system decision more conservative. In a preferred embodiment of this invention, this dynamic balance weight factor α can be calculated using the following mathematical formula: in, The received model health index has a value between 0 and 1, where 1 represents an extremely healthy model. The base weights represent the system's default decision-making tendency when the model is absolutely healthy; C is a coefficient used to adjust the system's risk aversion. A concrete numerical example is given: if the received... It is set to 0.8. If C is 0.15 and C is 2, then the calculated result of the dynamic balance weight factor is 0.15 + (1 - 0.8)^2 = 0.15 + 0.04 = 0.19. As an alternative implementation, the dynamic adjustment here can also be achieved through an S-shaped function to limit the value of the dynamic balance weight factor within a preset range.
[0035] After calculating the dynamic balance weighting factor, the composite objective generation module finally constructs the final composite optimization objective function. This is achieved by weighting and summing the previously calculated steady-state energy consumption and instantaneous switching cost terms using the calculated dynamic balance weighting factor. The result of this weighted sum is defined as the comprehensive cost of evaluating the operating state of a potential objective; a lower value indicates a better objective.
[0036] To address the issue that the units of the "steady-state energy consumption item" (i.e., "steady-state operating power value", in W) and the "instantaneous switching cost item" (calculated from the "basic switching intensity" and "pipeline vulnerability index", which is a dimensionless value) are inconsistent, and that direct weighted summation does not have a clear physical meaning, this invention introduces a cost conversion coefficient k when constructing the final composite optimization objective function.
[0037] The cost conversion factor k, measured in watts (W), is physically defined as "the equivalent power loss corresponding to a unit instantaneous switching cost." This factor is used to monetize or quantify the dimensionless switching cost into an equivalent cost with the same dimensions as the steady-state energy consumption. Its value can be determined through a comprehensive assessment based on factors such as equipment depreciation and grid stability penalties.
[0038] Therefore, the "final composite optimization objective function" (i.e., comprehensive cost) The specific calculation formula is as follows: in, This refers to the "steady-state operating power value"; Here, k represents the "instantaneous switching cost item"; k is the aforementioned cost conversion coefficient; and α is the dynamic balancing weight factor. By introducing the cost conversion coefficient k, the units of the two sub-items in the weighted summation are unified, and their physical meanings are clear, thereby ensuring the effectiveness and rationality of the final optimization objective.
[0039] This final composite optimization objective function, which can dynamically adapt to changes in model health, will serve as the sole basis for the next stage of optimization decisions.
[0040] After constructing the composite optimization objective function, the system process then enters the optimal state solution and switching path planning stage. The technical objective of this stage is to find the optimal state among all possible operating states based on the aforementioned constructed optimization function, and to plan a safe execution path.
[0041] To this end, the global optimal operating state solution module within the system first executes the solution for the global optimal operating state point. This solution process is implemented through an intelligent optimization algorithm, such as a genetic algorithm. Its specific implementation path is as follows: First, the module encodes a potential target operating state (defining the start / stop and specific speed of each pump) into a digital gene string, i.e., a chromosome. Next, the module randomly generates an initial population containing a large number of such chromosomes. For each chromosome in the population, the module decodes it into a specific operating state and uses the newly constructed composite optimization objective function as the fitness function to evaluate the chromosome's fitness. During the evaluation process, the module calls a high-fidelity digital twin model as a simulation platform to calculate the steady-state energy consumption and instantaneous switching cost corresponding to the chromosome, and finally calculates its fitness value based on the composite optimization objective function. After completing the evaluation of the entire population, the module iteratively generates a new population with better fitness by simulating selection, crossover, and mutation operations in biological evolution. This iterative process will continue until a preset termination condition is met, such as reaching the maximum number of generations or fitness convergence. Finally, the module decodes the chromosome with the best fitness value in the final population, and the running state it represents is determined as the optimal target running state. Alternatively, this optimization process can also be implemented using other heuristic optimization algorithms such as particle swarm optimization.
[0042] After determining the optimal target operating state point, the final step in the system process is the generation of a non-disruptive switching path by the non-disruptive switching path planning module. The technical goal of this step is to transform a discrete state switching decision into a continuous, smooth, and physically safe control process. The specific implementation path is as follows: First, the module receives the current actual operating state and the optimal target operating state point solved in the previous step. Next, within a preset switching time window (e.g., 30 seconds), the module generates a set of time-coordinated, non-linear control trajectories. The generation method involves using the core constraint of "suppressing the maximum pressure fluctuation rate of the pipeline network during the switching process within a preset safety threshold," and employing an interpolation or spline function algorithm to calculate the control curve of the inverter frequency of each pump requiring speed adjustment over time. These control curves are designed to be non-linear, for example, changing relatively smoothly at the initial and final stages of the switching, while changing more rapidly in the middle stage, to minimize the hydraulic impact on the pipeline network.
[0043] Take a specific switching scenario as an example: Assume the current state is that pump A is running at 80% frequency, and pump B is stopped; the optimal target state is that pump A is stopped, and pump B is running at 75% frequency. The seamless switching path planning module will generate two cooperating control trajectories: Trajectory 1, within 0 to 30 seconds, smoothly reduces the output frequency of pump A from 80% to 0% according to a preset S-shaped downward curve; Trajectory 2, in coordination with pump A, within the same time window, smoothly increases the output frequency of pump B from 0% to 75% according to another S-shaped upward curve. This coordinated and smooth control method avoids sudden drops or rises in pipeline pressure caused by the sudden stop or start of a single pump.
[0044] After successfully generating the control trajectories for all relevant water pumps, the module finally executes the control commands and performs closed-loop control. This is achieved by discretizing each calculated continuous control trajectory into a series of timestamped, specific frequency setpoint commands. Subsequently, the module sends these commands sequentially to the corresponding water pump's frequency converter or the system's underlying programmable logic controller (PLC) via an industrial communication protocol, such as Modbus TCP / IP. Upon receiving these commands, the underlying control execution unit precisely drives the water pumps to complete the state transition according to the planned time and frequency path. Thus, the entire system completes a full closed-loop adaptive energy-saving control process, from state awareness and model self-healing to dynamic optimization and safe execution.
[0045] This detailed embodiment elaborates on the internal operating logic of each major functional module, aiming to provide a detailed basis and explanation for those skilled in the art to understand and implement it. It should be emphasized that the above description constitutes a specific, preferred embodiment, but the concept of the present invention is not limited thereto. Any equivalent transformations, modifications, or improvements based on the core spirit of the present invention, without departing from the technical principles and scope disclosed in this specification, should be considered to fall within the scope of protection claimed by the present invention, as long as they achieve the same or similar technical effects.
Claims
1. A system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters, characterized in that, The system includes: The online identification and self-healing model building module is configured to receive the real-time operation data stream of the physical pump group and the synchronous prediction data stream of the current digital twin model, and output a high-fidelity digital twin model with dynamically corrected key physical attenuation parameters and a model health index that characterizes the credibility of the current model. The composite objective generation module is connected to the online identification and self-healing model construction module and is configured to: receive the high-fidelity digital twin model, the model health index and external operating condition requirements, and output a composite optimization objective function that includes steady-state energy consumption and instantaneous switching cost. The global optimal running state solution module is connected to the composite target generation module and the online identification and self-healing model construction module, and is configured to: receive the composite optimization objective function and the high-fidelity digital twin model, and output an optimal target running state point under the composite optimization objective. The non-disruptive switching path planning module is connected to the global optimal operating state solution module and is configured to: receive the optimal target operating state point and the current actual operating state of the physical pump group, and output a set of control trajectory instructions to the control execution unit of the pump group accordingly.
2. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 1, characterized in that, The online identification and self-healing model construction module is configured to dynamically correct the key physical attenuation parameters by using the model residual signal between the real-time running data stream and the synchronous prediction data stream, and based on a preset "model residual-physical attenuation" knowledge base, to identify the correction values of one or more key physical attenuation parameters in reverse.
3. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 1 or 2, characterized in that, The composite target generation module is configured to calculate the instantaneous switching cost item based on the basic switching intensity and the pipeline vulnerability index derived from the high-fidelity digital twin model; wherein the contribution of the pipeline vulnerability index to the instantaneous switching cost item is non-linearly amplified as the pipeline vulnerability index increases.
4. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 1 or 2, characterized in that, The composite target generation module is further configured to: use a dynamic balance weighting factor to weight the steady-state energy consumption item and the instantaneous switching cost item; and the value of the dynamic balance weighting factor is configured to be dynamically and adaptively adjusted according to the model health index from the online identification and self-healing model construction module.
5. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 3, characterized in that, The nonlinear amplification is achieved through an exponential function.
6. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 5, characterized in that, The steps for the composite target generation module to calculate the instantaneous switching cost item include: a) Determine one or more operations that switch from the current operating state to the target operating state, and calculate the basic switching intensity accordingly; b) Using the high-fidelity digital twin model, simulate the response of the current pipeline network under disturbance, and calculate the pipeline network vulnerability index accordingly; c) Substitute the basic switching intensity and the pipeline vulnerability index into the exponential function to generate the instantaneous switching cost item; Furthermore, the instantaneous switching cost item Calculated using the following mathematical formula: Among them, the For the basic switching intensity, the The pipeline network vulnerability index is defined as β, where β is an adjustable vulnerability impact weighting factor.
7. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 4, characterized in that, The steps for the composite target generation module to dynamically and adaptively adjust the dynamic balance weight factor include: a) Receive the model health index from the online identification and self-healing model construction module; b) Substitute the model health index into a mathematical function that adjusts the weights according to changes in the model health index to calculate the value of the dynamic balance weight factor. Furthermore, the dynamic balance weighting factor α is calculated using the following mathematical formula: in, The health index of the model, is the base weight value, and c is the risk aversion coefficient.
8. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 1, characterized in that, The disturbance-free switching path planning module is configured to generate control trajectory instructions that can suppress the pipeline pressure fluctuation rate of the water pump group within a preset threshold during the state switching process.
9. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 3, characterized in that, The nonlinear amplification is achieved through a polynomial function.
10. The system for dynamic modeling of pump group operating efficiency and self-optimization of energy-saving control parameters as described in claim 4, characterized in that, The value of the dynamic balance weight factor is adjusted according to the model health index through an S-shaped function, so as to limit the value of the dynamic balance weight factor within a preset range.