Intelligent switching and adaptive execution method for multi-mode control of pump stations

Abstract

This invention relates to the field of industrial automation control technology and discloses an intelligent switching and adaptive execution method for multi-mode control of pumping stations. The method includes: a condition feature identification step that generates a multi-dimensional condition feature vector based on real-time operating parameters and predicted water flow trend indicators; an intelligent switching decision step that executes dual-path arbitration logic, prioritizing the response to recommended control mode commands or determining the target control mode through rule matching; a parameter dynamic loading and seamless transition step that performs output weighted interpolation calculations during mode switching to generate a final frequency control command for smooth switching; a parameter online self-tuning step that iteratively corrects PID control parameters using a gradient descent algorithm during steady-state operation; and an anomaly handling and safety degradation step that is executed in parallel as high-priority tasks, forcibly locking the safe operating mode in case of a fault. This invention achieves bumpless switching and adaptive parameter optimization for pumping station control.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, specifically to an intelligent switching and adaptive execution method for multi-mode control of pump stations. Background Technology

[0002] As core nodes of municipal water supply networks, urban water supply pumping stations' operational stability and energy efficiency directly impact water supply security and operating costs. With the development of industrial internet technology, optimized scheduling strategies based on upper-level computers (such as SCADA systems or cloud platforms) are increasingly being applied to pumping station control to achieve energy conservation and consumption reduction. However, in practical engineering applications, existing pumping station control systems still face the following technical challenges:

[0003] Existing pump station control architectures typically employ a unidirectional model where the upper layer formulates strategies and the lower layer executes instructions. The lower-level control units (PLCs) often only possess simple constant pressure or constant current logic, lacking the ability to autonomously assess complex operating conditions. When the communication network of the upper-level optimization system fails or scheduling instructions are delayed, the lower-level control units can usually only maintain the instructions from the previous moment or revert to simple hard-wired logic. This approach cannot utilize locally detected flow trends or pressure distribution data for autonomous adjustment, making it difficult to cope with special operating conditions such as peak water usage periods and plateaus. It easily leads to fluctuations in water supply pressure or energy waste, and lacks intelligent adaptive scheduling capabilities in the event of communication anomalies.

[0004] Pumping stations need to switch between various modes, such as constant pressure control, constant flow control, and variable load control, to meet water demand at different times. Current control schemes typically override these modes by directly calling the corresponding PID parameter sets. However, due to differences in algorithm structure, gain parameters, and reference values ​​between different control modes, direct switching often leads to a step jump in the controller's output signal. This jump causes a drastic change in the inverter's operating frequency, resulting in hydraulic hammer effects in the pipeline network. This not only affects water supply stability but also accelerates the mechanical wear of pump units and pipeline valves.

[0005] The physical characteristics of pumping station systems drift over time; for example, impeller wear leads to decreased efficiency, and pipe scaling alters the resistance coefficient. Existing systems often use fixed-parameter PID controllers, which cannot detect and compensate for these steady-state errors online, resulting in decreased control accuracy after long-term operation. Furthermore, conventional fault monitoring logic is typically nested within the main control program and executed sequentially, resulting in long scan cycles and susceptibility to main program logic blockages. When sensor disconnections or severe communication failures occur, without an independent high-priority degradation mechanism, the system often fails to promptly lock into a safe operating state, posing a risk of equipment overload or water outages. Summary of the Invention

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent switching and adaptive execution method for multi-mode control of pumping stations. This method operates in a system comprised of a field sensing unit, a core control unit, an execution drive unit, and an upper-level optimization unit. The method includes the following steps: Step S1, System Initialization Step: After the system is powered on, the core control unit executes the initialization program; the core control unit includes an internal data register, a communication interface, a pre-stored parameter storage area, a current active data area, and a loop interrupt organization block. The core control unit resets the internal data register and the communication interface, loads the default running parameters indexed as safe mode in the pre-stored parameter storage area into the current active data area, and starts the loop interrupt organization block; Step S2, Real-time Operating Condition Feature Identification Step: In each scanning cycle triggered by the cyclic interruption tissue block, the core control unit includes an analog input interface and an operating condition feature identification module, and reads the real-time pressure signal, flow signal and liquid level signal collected by the field sensing unit through the analog input interface; The operating condition feature recognition module calculates the flow rate change trend rate, flow tracking deviation, average pressure value, and pressure distribution uniformity based on the current sampled value and a historical sampling sequence of preset length. The core control unit also includes a communication data mapping area, and the upper-level optimization unit writes the predicted water flow trend flag into the communication data mapping area. The operating condition feature recognition module combines the predicted water volume trend indicator read from the communication data mapping area to generate a multi-dimensional operating condition feature vector. Step S3, Intelligent Switching Decision Step: The core control unit includes an intelligent decision arbitration module, and the upper-level optimization unit generates a recommended control mode instruction and writes it into the communication data mapping area; The intelligent decision-making arbitration module periodically reads the multi-dimensional operating condition feature vector and the recommended control mode instruction from the communication data mapping area, executes priority-based dual-path arbitration logic, and determines the target control mode. Step S4, Parameter Dynamic Loading and Seamless Transition Step: The core control unit includes a parameter dynamic loading module, a smooth transition control module, and a transition parameter storage area; The core control unit compares the target control mode with the current operating mode in real time. When the two are inconsistent, the parameter dynamic loading module extracts the initial control parameter set from the pre-stored parameter storage area according to the index address of the target control mode and writes it into the transition parameter storage area; within the preset transition time window, the smooth transition control module performs a weighted summation operation on the first output component calculated based on the current operating mode and the second output component calculated based on the transition parameter storage area, and generates a final frequency control command to be sent to the frequency converter in the execution drive unit; Step S5, Online parameter self-tuning step: When the target control mode is consistent with the current operating mode, the core control unit includes an online parameter self-tuning module. The online parameter self-tuning module calculates the steady-state deviation between the real-time feedback pressure value and the optimal pressure setpoint issued by the upper-level optimization unit, and uses the gradient descent algorithm to perform step-by-step correction on the currently operating PID control parameters. Step S6, Anomaly Handling and Security Degradation Step: The core control unit includes an anomaly monitoring and security degradation module, which polls the electrical connection status of the field sensing unit and the heartbeat messages of the industrial Ethernet communication network in real time. Once a sensor disconnection fault or communication timeout fault is detected, the control output of the intelligent decision arbitration module and the online parameter self-tuning module is immediately blocked, the target control mode is forcibly overwritten as the safe operation mode, and the final frequency control command sent to the inverter is locked to the preset safe frequency fixed value.

[0007] Preferably, in step S2, the core control unit includes a data register area, and the operating condition feature recognition module is configured with a first-in-first-out cyclic sampling buffer of depth N in the data register area; The operating condition feature recognition module extracts discrete time series data from the first-in-first-out cyclic sampling buffer, and uses the least squares linear regression principle to calculate the covariance and variance of the time variable and the flow variable through discrete point accumulation. The ratio of the covariance to the variance is determined as the slope of the fitted line, which is the flow change trend rate. The multidimensional operating condition feature vector includes the flow tracking deviation value, the flow change trend rate value, the average pressure value, the pressure distribution uniformity value, and the predicted water flow trend indicator.

[0008] Preferably, in step S3, the core control unit includes a memory, which stores a preset working condition mapping rule lookup table, and the intelligent decision arbitration module reads the working condition mapping rule lookup table; The intelligent decision-making arbitration module executes priority-based dual-path arbitration logic as follows: First path: The intelligent decision arbitration module detects the instruction validity bit in the recommended control mode instruction; if the instruction validity bit is logically true, the recommended control mode instruction is directly parsed and updated to the target control mode; Second path: If the instruction valid bit is logical false or the communication status word indicates a timeout, the intelligent decision arbitration module traverses the working condition mapping rule lookup table, performs logical operations on the values ​​of each component of the multi-dimensional working condition feature vector with the preset comparison threshold in the table, retrieves the matching control mode, and assigns it to the target control mode.

[0009] Preferably, in step S3, the intelligent decision arbitration module is configured with a delay judgment timer, which only outputs a trigger signal to update the target control mode when the logical condition in the working condition mapping rule lookup table is always true for multiple consecutive scanning cycles and the duration exceeds the preset time window of the delay judgment timer. The intelligent decision arbitration module performs feedforward prediction based on the flow change trend rate and the predicted water volume trend indicator: when the predicted water volume trend indicator indicates a peak platform and the absolute value of the flow change trend rate is less than the preset steady-state threshold, it matches the steady-state optimization rule in the operating condition mapping rule lookup table, sets the target control mode to the variable load adaptive control mode, and sets the self-tuning permission flag.

[0010] Preferably, in step S4, the smooth transition control module calculates the real-time weighting coefficient based on the current count value of the transition timer, and uses the real-time weighting coefficient to generate the final frequency control command; The real-time weight coefficient is calculated by: calculating the ratio of the current count value of the transition timer to the preset transition period, and the value of the ratio is in the range of zero to one; The final frequency control command is generated by: calculating the difference between the real-time weight coefficient and the value of the difference, and multiplying the difference by the first output component calculated based on the original control mode algorithm before the switch to obtain the first weighting term. The second weighting term is obtained by multiplying the real-time weighting coefficient by the second output component calculated based on the target control mode algorithm after switching. The first weighted term and the second weighted term are added together to obtain the final frequency control command; When the current count value of the transition timer reaches the preset transition period, the parameter dynamic loading module performs a data block copy operation to completely overwrite the initial control parameter set in the transition parameter storage area to the current running parameter area.

[0011] Preferably, in step S5, the online parameter self-tuning module uses the gradient descent algorithm to perform step-by-step correction of the currently running PID control parameters in the following manner: The change in the controller output is calculated, and the change in the controller output is defined as the difference between the operating frequency value that the inverter is executing at the current iteration time and the operating frequency value that the inverter is executing at the previous self-tuning iteration time. The sign function value of the change in the controller output is determined. When the change is positive, the sign function value is positive one; when the change is negative, the sign function value is negative one; and when the change is zero, the sign function value is zero. The updated proportional gain coefficient is calculated by subtracting the product of the preset learning rate constant, the steady-state deviation, and the sign function value from the proportional gain coefficient currently in use during the control cycle. When the change in the controller output is zero, the sign function value returns zero, and the control parameters are not updated.

[0012] Preferably, in step S5, before performing parameter update calculation, the online parameter self-tuning module first determines whether the absolute value of the steady-state deviation is greater than the preset adjustment dead zone; If the result is false, then the current iteration is terminated; After generating the updated proportional gain coefficient, the online parameter self-tuning module executes the safety limiting logic: comparing the updated proportional gain coefficient with the safety upper limit and safety lower limit values ​​set in the pre-stored parameter storage area; If the updated proportional gain coefficient is greater than the safety upper limit value, then the safety upper limit value is assigned to the updated proportional gain coefficient. If the updated proportional gain coefficient is less than the safety lower limit, then the safety lower limit is assigned to the updated proportional gain coefficient.

[0013] Preferably, in step S6, the anomaly monitoring and security degradation module is configured as an independent high-priority cyclic interrupt task in the control program of the core control unit; The anomaly monitoring and security degradation module executes the highest priority control takeover logic as follows: when any bit of the sensor fault flag or communication fault flag is set, the update operation of the intelligent decision arbitration module and the online parameter self-tuning module on the output register is stopped by resetting the enable terminal of the PID function block or executing the program jump instruction. The safe operation mode is configured as a constant liquid level-fixed frequency joint control strategy. Under this strategy, the anomaly monitoring and safety degradation module directly reads the hard-wired signal status of the liquid level sensor. When the liquid level signal indicates a normal water level, the frequency converter includes a frequency setpoint register, and the anomaly monitoring and safety degradation module directly writes a preset safe frequency fixed value into the frequency setpoint register.

[0014] Preferably, a multi-mode adaptive scheduling system for pumping stations based on water volume prediction and dynamic optimization is provided, comprising: The field sensing unit is used to collect the physical operating parameters of the pumping station system, including pressure sensor group, flow sensor, liquid level sensor and power parameter acquisition instrument; The core control unit, which employs a programmable logic controller, is connected to the field sensing unit. The core control unit includes a central processing unit, a memory, an internal data register, a communication interface, a pre-stored parameter storage area, a currently active data area, a loop interrupt organization block, an analog input interface, a communication data mapping area, a transition parameter storage area, and control program logic; The control program logic is constructed as a data preprocessing module, a working condition feature identification module, an intelligent decision arbitration module, a parameter dynamic loading module, a smooth transition control module, an online parameter self-tuning module, and an anomaly monitoring and safety degradation module. The execution drive unit includes a frequency converter and a water pump unit, wherein the control input terminal of the frequency converter is connected to the core control unit; The upper-level optimization unit is connected to the core control unit via an industrial Ethernet communication network; The core control unit is configured to execute the above-described intelligent switching and adaptive execution method for multi-mode control of a pumping station.

[0015] Preferably, a pump station control device includes a processor, a memory, and a communication bus; The memory stores computer programs that can be executed by the processor; The communication bus is used to enable communication between the processor and the memory; When the processor executes the computer program, it implements the above-mentioned intelligent switching and adaptive execution method for multi-mode control of pumping stations.

[0016] This invention provides an intelligent switching and adaptive execution method for multi-mode control of pumping stations. It has the following beneficial effects: (1) This invention implements priority-based dual-path arbitration logic through an intelligent decision-making arbitration module, thereby achieving the coordination between upper-level optimization strategies and lower-level autonomous control. When the recommended control mode instruction is valid, the system prioritizes responding to the scheduling of the upper-level optimization unit. When the instruction is missing or communication times out, the second path is activated. The system traverses the working condition mapping rule lookup table and combines it with the predicted water volume trend flag to determine the target control mode. This not only utilizes external prediction information to cope with complex working conditions such as peak platforms, but also ensures that the lower-level controller can independently maintain system operation based on multi-dimensional working condition feature vectors in abnormal scenarios such as communication interruption.

[0017] (2) This invention utilizes a parameter dynamic loading module and a smooth transition control module to achieve a smooth transition without disturbance during multi-mode switching. When the target control mode changes, the system constructs an independent parameter environment through the transition parameter storage area and calculates the real-time weight coefficient within the preset transition time window. It performs a weighted summation operation on the first output component based on the current mode and the second output component based on the target mode, eliminating the step disturbance caused by the control logic jump, and ensuring that the final frequency control command sent to the frequency converter always maintains a continuous linear change, thus avoiding pipeline pressure shock.

[0018] (3) This invention improves the control accuracy and safety of the system through the online parameter self-tuning module and the anomaly monitoring and safety degradation module. When the online parameter self-tuning module is running in steady state, it uses the gradient descent algorithm to iteratively update the PID parameters based on the steady-state deviation, and automatically compensates for the characteristic drift caused by equipment aging. At the same time, the anomaly monitoring and safety degradation module, as an independent high-priority cyclic interrupt task, can immediately bypass the algorithm logic when it detects a sensor disconnection fault or communication timeout fault, forcibly overwrite the target control mode to the safe operation mode and write the preset safe frequency fixed value to the frequency setpoint register, so as to ensure that the pump station can still operate safely under fault conditions. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1This invention provides an intelligent switching and adaptive execution system for multi-mode control of pumping stations, comprising a field sensing unit, a core control unit, an execution drive unit, and an upper-level optimization unit, specifically: The field sensing unit is used to collect the physical operating parameters of the pumping station system. The field sensing unit includes a pressure sensor group set at different monitoring points in the pumping station outlet pipe network, a flow sensor set at the main outlet pipe, a liquid level sensor set at the water storage tank, and a power parameter acquisition instrument connected to the main circuit of the motor. The signal output terminals of the above sensors are connected to the analog input module and communication interface of the core control unit through shielded cables.

[0022] The core control unit adopts a programmable logic controller (PLC). The core control unit includes a central processing unit (CPU), memory, and input / output interfaces. The memory of the core control unit stores control program logic that can be executed by the CPU. When the control program logic is executed, it is logically constructed into multiple functional modules, including a data preprocessing module, a working condition feature recognition module, an intelligent decision arbitration module, a parameter dynamic loading module, a smooth transition control module, an online parameter self-tuning module, and an anomaly monitoring and safety degradation module.

[0023] The data preprocessing module is used to filter and perform range conversion processing on the raw signals collected by the field sensing unit. The working condition feature recognition module reads the data output by the data preprocessing module and generates a multi-dimensional working condition feature vector based on the time series sliding window algorithm. The multi-dimensional working condition feature vector is configured as structured data stored in the data register of the core control unit, which includes the flow tracking deviation value, the flow change trend rate value, the average pressure value, and the pressure distribution uniformity value.

[0024] The upper-level optimization unit includes an industrial control computer, which is connected to the core control unit via an industrial Ethernet communication network. The upper-level optimization unit writes the calculated recommended control mode instructions and predicted water volume trend flags into the communication data mapping area specified by the core control unit.

[0025] The intelligent decision arbitration module reads the multi-dimensional operating condition feature vector and the recommended control mode instruction. The intelligent decision arbitration module stores a preset logical judgment rule table. The intelligent decision arbitration module is configured to perform logical comparison operations. When the recommended control mode instruction exists, it responds to the instruction first. When the recommended control mode instruction is missing, it determines the target control mode based on the matching result of the multi-dimensional operating condition feature vector and the logical judgment rule table.

[0026] The parameter dynamic loading module is connected to the intelligent decision arbitration module. The memory of the core control unit is divided into a pre-stored parameter storage area (DataBlock). This area stores multiple sets of PID gain parameters and frequency reference values ​​corresponding to constant pressure control, constant flow control, constant liquid level control and variable load adaptive control. The parameter dynamic loading module is used to address and read the corresponding initial control parameter set from the pre-stored parameter storage area according to the index value of the target control mode.

[0027] The smooth transition control module is used to perform output weighted interpolation calculations during control mode switching. Based on a preset linear or nonlinear time function, the smooth transition control module performs weighted synthesis of the output value of the current mode and the calculated output value of the target mode to generate the final frequency control command.

[0028] The drive unit includes a frequency converter and a water pump unit. The control input terminal of the frequency converter is connected to the analog output module or fieldbus interface of the core control unit to receive the final frequency control command, thereby adjusting the operating frequency of the water pump unit.

[0029] The online parameter self-tuning module is used to calculate the deviation between the optimized setpoint issued by the upper-level optimization unit and the real-time feedback value when the system is in steady state, and to use the gradient descent algorithm to perform step-by-step correction of the currently running PID control parameters.

[0030] The anomaly monitoring and safety degradation module is configured as an independent high-priority interrupt routine. When a sensor disconnection or communication timeout fault signal is detected, the anomaly monitoring and safety degradation module immediately blocks the output of the intelligent decision arbitration module and locks the target control mode to the safe operation mode.

[0031] Please see Figure 2 Based on the above, this invention provides an intelligent switching and adaptive execution method for multi-mode control of pumping stations, specifically including the following steps: Step S1: System initialization step. After the system is powered on, the core control unit executes the initialization program to reset the internal data registers and communication interfaces. At the same time, the default running parameters indexed as "safe mode" in the pre-stored parameter storage area are loaded into the currently active data area. The core control unit configures and starts the CyclicInterruptOB block and sets the interrupt scan period to a preset time constant (preferably 100ms).

[0032] Step S2, Real-time Operating Condition Feature Identification Step: In each scanning cycle triggered by the cyclic interruption organization block, the data preprocessing module reads the real-time pressure signal, flow signal, and liquid level signal collected by the field sensing unit through the analog input interface, and performs moving average filtering on them. The operating condition feature identification module receives the processed data, calculates the flow rate change trend rate based on the current sampled value and the historical sampling sequence of preset length through a linear fitting algorithm, and calculates the flow tracking deviation, average pressure value, and pressure distribution uniformity. Combined with the predicted water flow trend indicator read from the communication mapping area of ​​the upper optimization unit, a structured multi-dimensional operating condition feature vector is generated.

[0033] Step S3: Intelligent switching decision-making step. The intelligent decision arbitration module periodically reads the multi-dimensional operating condition feature vector and the recommended control mode instruction. The intelligent decision arbitration module executes a priority-based dual-path arbitration logic: The first path detects the status word in the recommended control mode instruction. If the status word indicates that the instruction is valid, the recommended control mode instruction is directly parsed and updated to the target control mode. If the status word indicates that the instruction is invalid or times out, the second path is activated. The multi-dimensional operating condition feature vector is traversed and compared with the preset operating condition mapping rules in the memory to retrieve the matching control mode and assign it to the target control mode.

[0034] Step S4, Parameter Dynamic Loading and Seamless Transition: The core control unit compares the target control mode with the current operating mode in real time. When they are inconsistent, the parameter dynamic loading module retrieves the corresponding proportional-integral-derivative (PID) gain parameter and frequency reference value from the pre-stored parameter storage area based on the index address of the target control mode, and writes them into the transition parameter storage area. The smooth transition control module responds to the mode inconsistency signal by starting the transition timer and calculates the weighting coefficient that increases linearly with time within the preset transition time window. During this period, the smooth transition control module performs a weighted summation operation on the first output component calculated based on the current operating mode and the second output component calculated based on the transition parameter storage area, generates the final frequency control command, and sends it to the execution drive unit until the weighting coefficient reaches the preset threshold. The system then completely switches to the target control mode and releases the computing resources occupied by the original mode.

[0035] Step S5, Online Parameter Self-Tuning Step: When the target control mode is consistent with the current operating mode, the system maintains steady-state operation. The online parameter self-tuning module is equipped with a self-tuning timer. The online parameter self-tuning module calculates the steady-state deviation between the real-time feedback pressure value and the optimal pressure setpoint issued by the upper-level optimization unit. If the steady-state deviation exceeds the allowable dead zone, the online parameter self-tuning module uses the gradient descent algorithm to calculate the correction amount of the proportional-integral-derivative parameters, and iteratively updates the control parameters of the current operation with a limited small step size, so that the steady-state deviation converges to zero.

[0036] Step S6, the anomaly handling and safety degradation step, is executed in parallel with steps S2 to S5 as an independent high-priority task or hardware interrupt task. The anomaly monitoring and safety degradation module polls the electrical connection status of the field sensing unit and the heartbeat messages of the industrial Ethernet communication network in real time. Once a sensor disconnection fault or communication timeout fault is detected, the module immediately blocks the control output of the intelligent decision arbitration module and the online parameter self-tuning module, forces the target control mode to be overwritten as the safe operation mode, and locks the final frequency control command sent to the frequency converter to the preset safe frequency fixed value.

[0037] In step S2, the operating condition feature recognition module configures a depth of [missing information] in the data register area of ​​the core control unit. A first-in-first-out (FIFO) circular sampling buffer, which is used to store the most recent preset time window. (set up The historical operational data sequence within ).

[0038] During each scan cycle of the cyclic interrupted tissue block, the data preprocessing module synchronously reads the instantaneous flow rate values ​​collected by the flow sensor. and the data collected by the pressure sensor group Pressure monitoring values ​​at each point The data preprocessing module performs a moving average filtering process on the original signal, pushes the filtered current time data into the head pointer position of the circular sampling buffer, and removes the oldest data from the tail pointer position to maintain the real-time data sequence.

[0039] The operating condition feature recognition module reads the data from the circular sampling buffer and performs the following operations to generate a multidimensional operating condition feature vector: First, calculate the flow tracking deviation. The operating condition feature recognition module reads the target flow value set by the system. Calculate its value and the filtered instantaneous flow rate. The difference, i.e. .

[0040] Second, calculate the rate of change in flow rate. The operating condition feature recognition module extracts discrete time series data from the cyclic sampling buffer. Using the principle of least squares linear regression, the covariance and variance of the time variable are calculated through discrete point accumulation. The ratio of these two variances is then used to determine the slope of the fitted straight line, which represents the trend rate of flow change. .

[0041] Third, calculate the average pressure value of the pipeline network. and pressure distribution uniformity The working condition feature recognition module 22 identifies the same time period. The arithmetic mean of the pressure monitoring values ​​at each point is obtained. Simultaneously, it retrieves results using bubble sort or comparison commands. The maximum value among the pressure monitoring values ​​at each point and minimum value The pressure distribution uniformity is obtained by calculating the difference between the two. .

[0042] Fourth, obtain external predicted feature components. The operating condition feature identification module accesses the input data area that establishes a communication mapping with the upper-level optimization unit and directly reads the predicted water volume trend flag. This flag is an enumerated integer data, corresponding to the predefined states of stable, upward trend, downward trend, and peak platform.

[0043] Finally, the working condition feature recognition module will calculate the results. , , , and read The data is written sequentially into the multi-dimensional working condition feature vector data structure according to the preset memory address offset, and then called by the intelligent decision arbitration module.

[0044] In step S3, the intelligent decision arbitration module is configured with a mode switching logic state machine and a preset working condition mapping rule lookup table in the data storage area of ​​the core control unit. The intelligent decision arbitration module periodically accesses the input buffer and reads the multi-dimensional working condition feature vector generated above and the recommended control mode instructions written by the upper-level optimization unit.

[0045] The recommended control mode instruction data structure includes a 16-bit mode index word (Mode_Index) and a boolean instruction validity bit (Valid_Bit). The intelligent decision arbitration module executes priority-based path arbitration logic.

[0046] In the first path, the intelligent decision arbitration module detects the instruction validity bit in the recommended control mode instruction. If the instruction validity bit is logically true (TRUE), the intelligent decision arbitration module masks the input of the multi-dimensional operating condition feature vector, directly extracts the mode index word and assigns it to the target control mode. This logical path ensures that the scheduling strategy of the upper-level optimization unit is directly mapped to the control target of the execution layer.

[0047] In the second path, when the instruction validity bit is logically false (FALSE) or the communication status word indicates a timeout, the intelligent decision arbitration module starts the rule matching logic. The intelligent decision arbitration module traverses the working condition mapping rule lookup table and performs logical operations on the values ​​of each component of the multi-dimensional working condition feature vector with the preset comparison threshold in the table.

[0048] Taking the switching from constant pressure control mode to constant flow control mode as an example, the operating condition mapping rule lookup table defines the following AND logic condition: The intelligent decision arbitration module compares the flow tracking deviation in the multi-dimensional operating condition feature vector. and average pressure value of pipeline network Determine whether the following conditions are met simultaneously: The absolute value is greater than the preset flow deviation threshold and " The absolute value of the deviation from the current pressure setting is less than the preset pressure deviation threshold.

[0049] The intelligent decision arbitration module is equipped with a delay judgment timer (TimerOnDelay). Only when the above logical condition is true for multiple consecutive scan cycles and the duration exceeds the preset time window of the delay judgment timer (set to 30 seconds), the intelligent decision arbitration module outputs a trigger signal to update the target control mode to "constant flow control mode". If any condition turns false within the judgment time window, the intelligent decision arbitration module immediately resets the delay judgment timer and maintains the current target control mode unchanged.

[0050] In addition, the intelligent decision arbitration module is based on the traffic change trend rate. And perform feedforward prediction based on the predicted water volume trend indicator. When the predicted water volume trend indicator indicates "peak plateau" and the flow rate change trend rate is... When the absolute value is less than the preset steady-state threshold, the intelligent decision arbitration module matches the steady-state optimization rule in the working condition mapping rule lookup table, sets the target control mode to "variable load adaptive control mode", and sets the self-tuning enable flag. At the end of each scan cycle, the intelligent decision arbitration module refreshes the determined target control mode to the global status register of the core control unit for the parameter dynamic loading module to call.

[0051] In step S4, specifically, when the core control unit determines that the target control mode is inconsistent with the current operating mode, the parameter dynamic loading module responds to the mode switching trigger signal and addresses the pre-stored parameter storage area according to the index address of the target control mode.

[0052] The parameter dynamic loading module reads the initial control parameter set corresponding to the target control mode from the pre-stored parameter storage area. The initial control parameter set includes the proportional gain parameter, integral time constant, derivative time constant, and feedforward frequency reference value. The parameter dynamic loading module writes the above initial control parameter set into the pre-allocated transition parameter storage area in the memory of the core control unit, thereby constructing a second parameter set independent of the current operating parameter area at the data level.

[0053] The smooth transition control module starts its internal transition timer, and the transition occurs within a preset transition period after the transition timer has counted. Within 10 seconds, the smooth transition control module executes two sets of control operations sequentially within the same scan cycle of the core control unit: the first set of operations calculates the original mode output component based on the data of the current operating mode and the current operating parameter area. The second set of calculations yields the new mode output components based on data from the target control mode and the transition parameter storage area. .

[0054] The smooth transition control module uses the current count value of the transition timer. Calculate real-time weighting coefficients The calculation formula is as follows: In the formula: express The real-time weighting coefficient at time t, its value range is: ; This indicates the current count value of the transition timer, in seconds. This indicates the preset transition period, which is the total time required to complete the mode switch.

[0055] The smooth transition control module utilizes real-time weighting coefficients Output components of the original mode With the new mode output components Perform a weighted summation operation to generate the final frequency control command. The calculation formula is expressed as follows:

[0056] In the formula: express The final frequency control command, i.e. the frequency setpoint of the inverter, is sent to the drive unit at all times. This represents the theoretical output value calculated based on the algorithm of the original control mode before the switch; This represents the theoretical output value calculated based on the algorithm of the switched target control mode; The real-time weighting coefficients obtained from the above calculations are denoted as .

[0057] The smooth transition control module will calculate the result. The input signal of the frequency converter is sent to the execution drive unit through the analog output port, so that the input signal of the frequency converter remains continuously linear during the mode switching, eliminating the step disturbance during the control mode switching process.

[0058] When the current count value of the transition timer Reaching the preset transition period At that time, that is Once the smooth transition control module determines that the transition process has ended, the parameter dynamic loading module performs a data block copy (BlockMove) operation, completely overwriting the initial control parameter set in the transition parameter storage area to the current operating parameter area. The core control unit then stops outputting components in the original mode. The computational logic locks the target control mode to a single master control logic.

[0059] In step S5, specifically, when the core control unit confirms that the target control mode is consistent with the current operating mode, and the flow rate change trend output by the operating condition characteristic identification module is... When the value falls below the preset steady-state threshold, the online parameter self-tuning module initiates adaptive iterative calculation. This module is equipped with a self-tuning timer independent of the PID control loop scan cycle, with a time interval... Set to 60 seconds.

[0060] The online parameter self-tuning module responds to the trigger signal of the self-tuning timer by reading the dynamic optimization setting value written by the upper-level optimization unit. and the real-time average pressure value fed back by the field sensing unit The online parameter self-tuning module calculates the absolute value of the steady-state deviation. The online parameter self-tuning module performs a dead-zone logic comparison: [This is followed by a seemingly unrelated sentence about judgment / determination.] Is it larger than the preset adjustment dead zone? (Set to 0.01MPa) If the judgment result is false, the online parameter self-tuning module terminates the current iteration and maintains the current control parameters unchanged; if the judgment result is true, the parameter update operation is performed.

[0061] The online parameter self-tuning module uses an extremum optimization algorithm based on the discrete gradient descent principle to optimize the proportional gain coefficient of the PID controller. and integral gain coefficient Corrections are made, and the following uses the proportional gain coefficient. The correction process will be described using an example.

[0062] The online parameter self-tuning module establishes performance index functions. The calculation formula is as follows:

[0063] In the formula: Indicates the first The performance index value of each self-tuning iteration cycle is used to quantify the quality of the control effect; Indicates the first The steady-state deviation of pressure during the next iteration is the difference between the real-time feedback value and the optimized set value. Indicates the first The average pressure value of the pipeline network collected by the field sensing unit and filtered during each iteration cycle; Indicates the first The dynamic optimization setting value issued by the upper-level optimization unit corresponding to each iteration cycle.

[0064] The online parameter self-tuning module calculates the correction amount for the proportional gain coefficient and generates the proportional gain coefficient for the next cycle. The iterative update formula is as follows:

[0065] In the formula: This indicates the updated proportional gain coefficient that will take effect in the next cycle. This indicates the proportional gain coefficient currently in use during the control cycle; This represents the preset learning rate constant, used to control the step size of parameter adjustment to prevent system overshoot; its value is 0.001. The above calculation represents the steady-state pressure deviation. The sign function is used to determine the adjustment direction of gradient descent. It returns +1 when the value in parentheses is positive, -1 when it is negative, and 0 when it is 0. It represents the change in the controller output, used to characterize the response trend of the control system to the input.

[0066] In the above formula Defined by the following formula:

[0067] In the formula: Indicates the current iteration time. The operating frequency value that the frequency converter is currently executing; Indicates the time of the previous self-tuning iteration. The operating frequency value executed by the frequency converter.

[0068] The online parameter self-tuning module generates... Then, the safety limiting logic is executed, and the online parameter self-tuning module compares... The safety upper limit value set in the pre-stored parameter storage area and safety lower limit ,like Then Assigned value ;like Then Assigned value ; when When it is 0, Returns 0, and the control parameters are not updated.

[0069] After verification, the online parameter self-tuning module will provide the final result. The numerical values ​​are overwritten into the background data block of the PID function block of the core control unit, so that the updated control parameters take effect in the next PID control scan cycle, thereby achieving adaptive dynamic matching of control parameters as the operating conditions drift.

[0070] In step S6, specifically, the anomaly monitoring and security degradation module is configured as an independent high-priority cyclic interrupt task (CyclicInterruptTask) in the control program of the core control unit. Its scan cycle is set to 20 milliseconds, which is shorter than the scan cycle of the main control program, so as to ensure priority capture of system faults.

[0071] The anomaly monitoring and safety degradation module executes analog signal range verification logic. It directly reads the original current signal value mapped from the field sensing unit to the input image area. The module compares the real-time current values ​​of the pressure sensor group and flow sensor with the preset open circuit threshold (3.8mA) and short circuit threshold (20.5mA). If the current value of any sensor is continuously lower than the open circuit threshold or higher than the short circuit threshold for a period of time exceeding the preset filtering confirmation time window (e.g., 2 seconds), the module determines that an electrical fault has occurred in the physical channel and sets the corresponding sensor fault flag bit (Fault_Flag_Sensor).

[0072] The anomaly monitoring and security degradation module executes the communication link heartbeat monitoring logic. The anomaly monitoring and security degradation module reads the rolling heartbeat counter value (Heartbeat_Counter) written by the upper-layer optimization unit through the industrial Ethernet communication network. The anomaly monitoring and security degradation module compares the heartbeat counter value of the current scan cycle with the cached value of the previous scan cycle. If the two values ​​remain the same and do not change within the preset communication timeout judgment time (e.g., 5 seconds), the anomaly monitoring and security degradation module determines that the communication link is interrupted and sets the communication fault flag bit (Fault_Flag_Comm).

[0073] The anomaly monitoring and safety degradation module executes the highest priority control takeover logic. When any bit of the sensor fault flag or communication fault flag is set (i.e., logically true), the anomaly monitoring and safety degradation module triggers the safety interlock mechanism. The anomaly monitoring and safety degradation module bypasses the operation logic of the intelligent decision arbitration module and the online parameter self-tuning module by resetting the enable pin of the PID function block or executing a program jump instruction, thus stopping their update operation on the output register.

[0074] The anomaly monitoring and safety degradation module forcibly overwrites the target control mode with a preset safe operation mode code. The safe operation mode is configured as a "constant liquid level-fixed frequency joint control strategy". Under this strategy, the anomaly monitoring and safety degradation module directly reads the hard-wired signal status of the liquid level sensor and writes a preset safe frequency fixed value (40Hz) directly to the frequency setpoint register of the frequency converter when the liquid level signal indicates a normal water level. This safe frequency fixed value is a value calibrated based on the characteristics of the pump station pipeline network that can meet the minimum water supply requirements and will not cause motor overload.

[0075] The anomaly monitoring and safety degradation module simultaneously sets the alarm output coil, drives the local audible and visual alarm, and sends a fault status word through the backup communication port. This degradation control state will continue until the system detects that all fault flag bits have been reset (including automatic recovery or manual confirmation reset). Only then can the anomaly monitoring and safety degradation module release the interlock and restore the aforementioned automatic scheduling process.

[0076] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent switching and adaptive execution of multi-mode control in pumping stations, characterized in that, This method operates in a system consisting of a field sensing unit, a core control unit, an execution drive unit, and an upper-level optimization unit. The method includes the following steps: Step S1, System Initialization Step: After the system is powered on, the core control unit executes the initialization program; the core control unit includes an internal data register, a communication interface, a pre-stored parameter storage area, a current active data area, and a loop interrupt organization block. The core control unit resets the internal data register and the communication interface, loads the default running parameters indexed as safe mode in the pre-stored parameter storage area into the current active data area, and starts the loop interrupt organization block; Step S2, Real-time Operating Condition Feature Identification Step: In each scanning cycle triggered by the cyclic interruption tissue block, the core control unit includes an analog input interface and an operating condition feature identification module, and reads the real-time pressure signal, flow signal and liquid level signal collected by the field sensing unit through the analog input interface; The operating condition feature recognition module calculates the flow rate change trend rate, flow tracking deviation, average pressure value, and pressure distribution uniformity based on the current sampled value and a historical sampling sequence of preset length. The core control unit also includes a communication data mapping area, and the upper-level optimization unit writes the predicted water flow trend flag into the communication data mapping area. The operating condition feature recognition module combines the predicted water volume trend indicator read from the communication data mapping area to generate a multi-dimensional operating condition feature vector. Step S3, Intelligent Switching Decision Step: The core control unit includes an intelligent decision arbitration module, and the upper-level optimization unit generates a recommended control mode instruction and writes it into the communication data mapping area; The intelligent decision-making arbitration module periodically reads the multi-dimensional operating condition feature vector and the recommended control mode instruction from the communication data mapping area, executes priority-based dual-path arbitration logic, and determines the target control mode. Step S4, Parameter Dynamic Loading and Seamless Transition Step: The core control unit includes a parameter dynamic loading module, a smooth transition control module, and a transition parameter storage area; The core control unit compares the target control mode with the current operating mode in real time. When the two are inconsistent, the parameter dynamic loading module extracts the initial control parameter set from the pre-stored parameter storage area according to the index address of the target control mode and writes it into the transition parameter storage area; within the preset transition time window, the smooth transition control module performs a weighted summation operation on the first output component calculated based on the current operating mode and the second output component calculated based on the transition parameter storage area, and generates a final frequency control command to be sent to the frequency converter in the execution drive unit; Step S5, Online parameter self-tuning step: When the target control mode is consistent with the current operating mode, the core control unit includes an online parameter self-tuning module. The online parameter self-tuning module calculates the steady-state deviation between the real-time feedback pressure value and the optimal pressure setpoint issued by the upper-level optimization unit, and uses the gradient descent algorithm to perform step-by-step correction on the currently operating PID control parameters. Step S6, Anomaly Handling and Security Degradation Step: The core control unit includes an anomaly monitoring and security degradation module, which polls the electrical connection status of the field sensing unit and the heartbeat messages of the industrial Ethernet communication network in real time. Once a sensor disconnection fault or communication timeout fault is detected, the control output of the intelligent decision arbitration module and the online parameter self-tuning module is immediately blocked, the target control mode is forcibly overwritten as the safe operation mode, and the final frequency control command sent to the inverter is locked to the preset safe frequency fixed value.

2. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 1, characterized in that: In step S2, the core control unit includes a data register area, and the operating condition feature recognition module is configured with a first-in-first-out cyclic sampling buffer of depth N in the data register area. The operating condition feature recognition module extracts discrete time series data from the first-in-first-out cyclic sampling buffer, and uses the least squares linear regression principle to calculate the covariance and variance of the time variable and the flow variable through discrete point accumulation. The ratio of the covariance to the variance is determined as the slope of the fitted line, which is the flow change trend rate. The multidimensional operating condition feature vector includes the flow tracking deviation value, the flow change trend rate value, the average pressure value, the pressure distribution uniformity value, and the predicted water flow trend indicator.

3. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 1, characterized in that: In step S3, the core control unit includes a memory, which stores a preset working condition mapping rule lookup table, and the intelligent decision arbitration module reads the working condition mapping rule lookup table. The intelligent decision-making arbitration module executes priority-based dual-path arbitration logic as follows: First path: The intelligent decision arbitration module detects the instruction validity bit in the recommended control mode instruction; if the instruction validity bit is logically true, the recommended control mode instruction is directly parsed and updated to the target control mode; Second path: If the instruction valid bit is logical false or the communication status word indicates a timeout, the intelligent decision arbitration module traverses the working condition mapping rule lookup table, performs logical operations on the values ​​of each component of the multi-dimensional working condition feature vector with the preset comparison threshold in the table, retrieves the matching control mode, and assigns it to the target control mode.

4. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 3, characterized in that: In step S3, the intelligent decision arbitration module is configured with a delay judgment timer. The trigger signal is output to update the target control mode only when the logical condition in the working condition mapping rule lookup table is always true for multiple consecutive scanning cycles and the duration exceeds the preset time window of the delay judgment timer. The intelligent decision arbitration module performs feedforward prediction based on the flow change trend rate and the predicted water volume trend indicator: when the predicted water volume trend indicator indicates a peak platform and the absolute value of the flow change trend rate is less than the preset steady-state threshold, it matches the steady-state optimization rule in the operating condition mapping rule lookup table, sets the target control mode to the variable load adaptive control mode, and sets the self-tuning permission flag.

5. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 1, characterized in that: In step S4, the smooth transition control module calculates the real-time weighting coefficient based on the current count value of the transition timer, and uses the real-time weighting coefficient to generate the final frequency control command. The real-time weight coefficient is calculated by: calculating the ratio of the current count value of the transition timer to the preset transition period, and the value of the ratio is in the range of zero to one; The final frequency control command is generated by: calculating the difference between the real-time weight coefficient and the value of the difference, and multiplying the difference by the first output component calculated based on the original control mode algorithm before the switch to obtain the first weighting term. The second weighting term is obtained by multiplying the real-time weighting coefficient by the second output component calculated based on the target control mode algorithm after switching. The first weighted term and the second weighted term are added together to obtain the final frequency control command; When the current count value of the transition timer reaches the preset transition period, the parameter dynamic loading module performs a data block copy operation to completely overwrite the initial control parameter set in the transition parameter storage area to the current running parameter area.

6. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 1, characterized in that: In step S5, the online parameter self-tuning module uses the gradient descent algorithm to perform step-by-step correction of the currently running PID control parameters in the following specific manner: The change in the controller output is calculated, and the change in the controller output is defined as the difference between the operating frequency value that the inverter is executing at the current iteration time and the operating frequency value that the inverter is executing at the previous self-tuning iteration time. The sign function value of the change in the controller output is determined. When the change is positive, the sign function value is positive one; when the change is negative, the sign function value is negative one; and when the change is zero, the sign function value is zero. The updated proportional gain coefficient is calculated by subtracting the product of the preset learning rate constant, the steady-state deviation, and the sign function value from the proportional gain coefficient currently in use during the control cycle. When the change in the controller output is zero, the sign function value returns zero, and the control parameters are not updated.

7. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 6, characterized in that: In step S5, before performing parameter update calculation, the online parameter self-tuning module first determines whether the absolute value of the steady-state deviation is greater than the preset adjustment dead zone. If the result is false, then the current iteration is terminated; After generating the updated proportional gain coefficient, the online parameter self-tuning module executes the safety limiting logic: comparing the updated proportional gain coefficient with the safety upper limit and safety lower limit values ​​set in the pre-stored parameter storage area; If the updated proportional gain coefficient is greater than the safety upper limit value, then the safety upper limit value is assigned to the updated proportional gain coefficient. If the updated proportional gain coefficient is less than the safety lower limit, then the safety lower limit is assigned to the updated proportional gain coefficient.

8. The intelligent switching and adaptive execution method for multi-mode control of a pumping station according to claim 1, characterized in that: In step S6, the anomaly monitoring and security degradation module is configured as an independent high-priority cyclic interrupt task in the control program of the core control unit; The anomaly monitoring and security degradation module executes the highest priority control takeover logic as follows: when any bit of the sensor fault flag or communication fault flag is set, the update operation of the intelligent decision arbitration module and the online parameter self-tuning module on the output register is stopped by resetting the enable terminal of the PID function block or executing the program jump instruction. The safe operation mode is configured as a constant liquid level-fixed frequency joint control strategy. Under this strategy, the anomaly monitoring and safety degradation module directly reads the hard-wired signal status of the liquid level sensor. When the liquid level signal indicates a normal water level, the frequency converter includes a frequency setpoint register, and the anomaly monitoring and safety degradation module directly writes a preset safe frequency fixed value into the frequency setpoint register.

9. A multi-mode adaptive scheduling system for pumping stations based on water volume prediction and dynamic optimization, characterized in that, The system includes: The field sensing unit is used to collect the physical operating parameters of the pumping station system, including pressure sensor group, flow sensor, liquid level sensor and power parameter acquisition instrument; The core control unit, which employs a programmable logic controller, is connected to the field sensing unit. The core control unit includes a central processing unit, a memory, an internal data register, a communication interface, a pre-stored parameter storage area, a currently active data area, a loop interrupt organization block, an analog input interface, a communication data mapping area, a transition parameter storage area, and control program logic; The control program logic is constructed as a data preprocessing module, a working condition feature identification module, an intelligent decision arbitration module, a parameter dynamic loading module, a smooth transition control module, an online parameter self-tuning module, and an anomaly monitoring and safety degradation module. The execution drive unit includes a frequency converter and a water pump unit, wherein the control input terminal of the frequency converter is connected to the core control unit; The upper-level optimization unit is connected to the core control unit via an industrial Ethernet communication network; The core control unit is configured to execute an intelligent switching and adaptive execution method for multi-mode control of a pumping station as described in any one of claims 1 to 8.

10. A pump station control device, characterized in that, Includes processor, memory, and communication bus; The memory stores computer programs that can be executed by the processor; The communication bus is used to enable communication between the processor and the memory; When the processor executes the computer program, it implements an intelligent switching and adaptive execution method for multi-mode control of a pumping station as described in any one of claims 1 to 8.

Images