Adaptive coordinated frequency control method and system for port multi-source load

Abstract

This invention discloses an adaptive coordinated frequency regulation control method and system for multi-source loads in ports, relating to the field of microgrid control technology. The method includes the following steps: acquiring high-resolution load and renewable energy data for the port; dividing the load into continuous controllable loads and discrete controllable loads; dynamically adjusting the virtual inertia and damping coefficient based on the frequency change rate and deviation; constructing a hybrid prediction model containing continuous and integer decision variables, and generating the total frequency regulation power command using mixed-integer linear quadratic programming; and allocating the command to a hybrid energy storage system composed of supercapacitors and lithium-ion batteries through adaptive spectrum separation. This invention solves the frequency oscillation defects of traditional methods when dealing with nonlinear operating conditions in ports, extends battery life, and improves the system's disturbance rejection capability and transient frequency stability.

Description

Technical Field

[0001] This invention relates to the field of microgrid control technology, specifically to an adaptive coordinated frequency regulation control method and system for multi-source loads in ports. Background Technology

[0002] The increasing penetration of renewable energy sources such as photovoltaics and offshore wind power in port microgrids has led to a sharp decline in system inertia. Simultaneously, heterogeneous loads such as cranes, cold chain equipment, and shore power exhibit frequent abrupt changes: crane lifting power can jump from zero to several megawatts within a few seconds; cold chain equipment exhibits thermal hysteresis; and shore power connection / disconnection is highly random. These factors easily trigger frequency instability, potentially leading to widespread power outages in severe cases.

[0003] In existing technologies, load frequency control for microgrids often employs traditional proportional-integral-derivative (PID) control or conventional droop control. However, traditional PID or droop control lacks adaptability and struggles to cope with nonlinear and sudden operating conditions in ports. Specifically, it suffers from the following blind spots: it cannot characterize the start-stop jumps of discrete equipment, simplifying them into static curves and leading to control deviations; its inertia and damping coefficients are fixed, failing to provide flexible compensation during impact transients; single energy storage devices frequently withstand high-power pulses, accelerating battery life degradation; and offline parameter tuning struggles to adapt to real-time environmental changes such as tides and weather.

[0004] Therefore, an adaptive and coordinated frequency regulation control method for multi-source loads in ports is needed, which can characterize the dynamic characteristics of heterogeneous load mixture, adaptively adjust virtual inertia and damping, coordinate mixed energy storage distribution, and have online parameter optimization capabilities. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing an adaptive coordinated frequency modulation control method and system for multi-source loads in ports.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An adaptive coordinated frequency regulation control method for multi-source loads in ports includes the following steps: Step S1: Obtain high-resolution power parameters of the port's multi-source load and real-time power parameters of the renewable energy supply side; Step S2: Perform feature extraction and cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. Step S3: Monitor the system frequency deviation and frequency change rate of the microgrid in real time. Based on the frequency deviation and frequency change rate, adaptively and dynamically adjust the virtual inertia parameters and damping coefficients on the control side of the microgrid inverter to construct an adaptive virtual synchronous generator control model. Step S4: Based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, construct the objective function and operating constraints of the hybrid model predictive control; Step S5: Solve the objective function of the hybrid model predictive control to generate a total frequency regulation active power reference command for smoothing frequency fluctuations; Step S6: Based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, the total frequency modulation active power reference command is decoupled into high-frequency power command and low-frequency power command through an adaptive low-pass filter, and then distributed to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

[0007] Furthermore, step S2 specifically includes the following steps: Step S2.1: Extract two features from the power time series of each load: the power mutation pulse exponent and the load dynamic inertia coefficient, and construct a two-dimensional feature vector; Step S2.2: Perform cluster analysis on the two-dimensional feature vector using an adaptive density peak clustering algorithm. Specifically, this includes: calculating the local density and relative distance of each feature point; constructing decision values ​​and arranging them in descending order; automatically determining cluster centers through decision value difference; assigning non-cluster center points to cluster centers with higher density and the closest distance; and finally outputting continuous controllable load clusters and discrete controllable load clusters. Step S2.3: Define the refrigerated container load with thermodynamic isothermal hysteresis characteristics as a continuous controllable load, and model its power regulation within its temperature control domain as a continuous variable; define the bridge crane equipment with trapezoidal power curve characteristics and operating state in the form of start-stop jump and the shore power demand for berthing as discrete controllable loads; Step S2.4: Using a hybrid logic dynamic modeling method, the nonlinear switching dynamics of the discrete controllable load is transformed into linear inequality constraints containing logical integer variables, and a unified state-space equation for the heterogeneous load is established.

[0008] Furthermore, step S3 specifically includes the following steps: Step S3.1: Set the foundation inertia coefficient and foundation damping coefficient; Step S3.2: Calculate the absolute value of the frequency change rate at the current moment. When the absolute value exceeds the set system safety dead zone threshold, add a virtual inertia compensation component that is positively correlated with the frequency change rate to the system to release or absorb transient kinetic energy. Step S3.3: Calculate the absolute value of the frequency deviation at the current moment, and add a virtual damping compensation component that is positively correlated with the frequency deviation to the system to suppress the frequency drop depth in the initial stage of the sudden increase in the crane load, and accelerate frequency stabilization in the frequency recovery stage to suppress the secondary oscillation phenomenon.

[0009] Furthermore, step S4 specifically includes the following steps: Step S4.1: Define the prediction time domain and the control time domain, and take the system frequency deviation, regional tie line power deviation, the adjustable capacity of continuous loads and the switching state of discrete loads as multidimensional state variables. Step S4.2: Construct a quadratic objective function that includes a frequency deviation penalty term, a control increment cost term, a regional tie-line power over-limit penalty term, and a discrete load switching frequency penalty term; Step S4.3: Introduce physical limitations of equipment, operational safety boundaries, and special requirements for port operations as operational constraints. These constraints specifically include global power balance constraints of the microgrid, ramp rate limits for conventional generator sets and peak-shaving units, state-of-charge safety domain constraints for energy storage systems, upper and lower limits for continuous load regulation, and minimum continuous operation and downtime constraints for discrete loads.

[0010] Furthermore, step S5 specifically includes the following steps: Step S5.1: Transform the quadratic objective function and multivariate constraints into a mixed-integer linear quadratic programming mathematical problem; Step S5.2: Using a hybrid solution algorithm combining the branch and bound method and the effective set method, rolling optimization calculations are performed within the given discrete control period to extract the first set of elements of the control sequence as the reference command for the total frequency regulation active power at the current moment. Step S5.3: Use one of the particle swarm optimization algorithm, gray wolf optimization algorithm or differential evolution algorithm to perform online adaptive optimization of the weight matrix in the hybrid model predictive control, and dynamically update the penalty weight according to the port load fluctuation law under different seasons and meteorological conditions to cope with the nonlinear perturbation of port microgrid system parameters and strong external interference.

[0011] Furthermore, step S6 specifically includes the following steps: Step S6.1: Obtain the charge-discharge cycle life loss state model of the lithium-ion battery and the real-time state of charge feedback value of the supercapacitor. Step S6.2: Construct a dynamic adjustment mechanism for the filter time constant, and extract the low-frequency steady-state power deviation by smoothing the low-pass filter; Step S6.3: Distribute the filtered low-frequency commands to the lithium-ion battery cells and strip the unfiltered high-frequency transient power fluctuations and distribute them to the supercapacitors. Step S6.4: When the state of charge of the supercapacitor deviates from the lower limit of the safe operating range, the low-pass filter time constant is dynamically reduced and a compensation power term proportional to the state of charge deviation is introduced. Part of the high-frequency power and compensation power are transferred to the lithium-ion battery, thereby ensuring that the supercapacitor continuously provides high-frequency modulation capability and avoiding leakage current.

[0012] An adaptive cooperative frequency modulation control system for multi-source loads in ports, used to implement any one of the adaptive cooperative frequency modulation control methods for multi-source loads in ports, including: The multi-source data acquisition module is used to acquire high-resolution power parameters of multi-source loads at the port and real-time power parameters of the renewable energy supply side. The heterogeneous load sensing and classification module is used to extract features and perform cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. The virtual inertia adaptive module is used to monitor the system frequency deviation and frequency change rate of the microgrid in real time, and adaptively and dynamically adjust the virtual inertia parameters and damping coefficient on the control side of the microgrid inverter. The hybrid predictive optimization module is used to construct and solve the objective function and operating constraints of the hybrid model predictive control based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, and generate the total frequency regulation active power reference command. The frequency domain decoupling and power allocation module is used to decouple the total frequency modulation active power reference command into high-frequency power command and low-frequency power command through an adaptive low-pass filter, based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, and allocate them to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

[0013] Furthermore, the multi-source data acquisition module collects the transient stator current of the bridge crane motor in real time through a distributed Internet of Things sensor network, obtains the thermal cycle status of the refrigerated container through infrared thermal imaging and bus communication, and performs nanosecond-level timestamp synchronization. The overall data sampling frequency of the system is not less than 1kHz.

[0014] Furthermore, the frequency domain decoupling and power distribution module is equipped with an anti-aliasing filter and a droop control interface, which are used to provide continuous underlying voltage and frequency support for the port microgrid in off-grid island operation mode and under extreme weather disturbances.

[0015] Furthermore, the hybrid prediction optimization module is deployed in the centralized energy management system or edge computing node of the port microgrid, and interacts with the underlying distributed energy storage inverter main controller in a closed loop through industrial Ethernet. A communication delay compensation mechanism is configured to address communication delays. The communication delay compensation mechanism uses a predefined time observer and extended Kalman filter for system state prediction and advance correction control.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention breaks through the limitation of traditional load frequency control, which can only handle continuous state variables, by using physical information-driven heterogeneous load classification and hybrid logic dynamic modeling. It can directly transform the physical actions of discrete equipment such as bridge cranes into power grid trajectory optimization.

[0017] 2. This invention introduces an adaptive virtual inertia mechanism, which dynamically adjusts the virtual inertia and damping coefficient according to the frequency change rate and frequency deviation, greatly enhancing the transient shock resistance of low-inertia microgrids when encountering sudden heavy load impacts, and effectively suppressing frequency drop depth and secondary oscillations.

[0018] 3. This invention utilizes frequency domain decoupling and the synergistic cooperation of a hybrid energy storage system to guide high-frequency destructive pulses to the supercapacitor and distribute low-frequency steady-state power to the lithium-ion battery, thereby significantly extending the entire life cycle of the lithium battery and reducing the replacement cost of the energy storage system.

[0019] 4. This invention achieves coordinated optimization of multiple objectives such as frequency deviation, tie line power overrun, and discrete load switching by using hybrid model predictive control and hybrid integer linear quadratic programming. Combined with swarm intelligence algorithm for online adaptive optimization, it significantly improves the transient and steady-state frequency quality of the system and effectively reduces the levelized cost of electricity (LCOE) of port energy management. Attached Figure Description

[0020] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart illustrating an embodiment of the present invention; Figure 2 This is a schematic diagram of the port multi-energy hybrid microgrid topology and heterogeneous load classification according to an embodiment of the present invention; Figure 3 This is a diagram of the hybrid model predictive control and adaptive virtual inertia two-layer control framework of an embodiment of the present invention; Figure 4 This is a schematic diagram of the system structure according to an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown, the adaptive cooperative frequency regulation control method for multi-source loads in a port includes the following steps: Step S1: Obtain high-resolution power parameters of the port's multi-source load and real-time power parameters of the renewable energy supply side; Step S2: Perform feature extraction and cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. Step S3: Monitor the system frequency deviation and frequency change rate of the microgrid in real time. Based on the frequency deviation and frequency change rate, adaptively and dynamically adjust the virtual inertia parameters and damping coefficients on the control side of the microgrid inverter to construct an adaptive virtual synchronous generator control model. Step S4: Based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, construct the objective function and operating constraints of the hybrid model predictive control; Step S5: Solve the objective function of the hybrid model predictive control to generate a total frequency regulation active power reference command for smoothing frequency fluctuations; Step S6: Based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, the total frequency modulation active power reference command is decoupled into high-frequency power command and low-frequency power command through an adaptive low-pass filter, and then distributed to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

[0023] The following data are collected synchronously through a distributed Internet of Things sensor network at a sampling frequency of not less than 1 kHz: transient stator current of the bridge crane motor, thermal cycle status of refrigerated containers, power of shore power system, real-time output of photovoltaic and wind power, and frequency of microgrid common coupling point.

[0024] Step S2 specifically includes the following steps: Step S2.1: Extract two features from the power time series of each load: the power mutation pulse exponent and the load dynamic inertia coefficient, and construct a two-dimensional feature vector; Step S2.2: Perform cluster analysis on the two-dimensional feature vector using an adaptive density peak clustering algorithm. Specifically, this includes: calculating the local density and relative distance of each feature point; constructing decision values ​​and arranging them in descending order; automatically determining cluster centers through decision value difference; assigning non-cluster center points to cluster centers with higher density and the closest distance; and finally outputting continuous controllable load clusters and discrete controllable load clusters. Step S2.3: Define the refrigerated container load with thermodynamic isothermal hysteresis characteristics as a continuous controllable load, and model its power regulation within its temperature control domain as a continuous variable; define the bridge crane equipment with trapezoidal power curve characteristics and operating state in the form of start-stop jump and the shore power demand for berthing as discrete controllable loads; Step S2.4: Using a hybrid logic dynamic modeling method, the nonlinear switching dynamics of the discrete controllable load is transformed into linear inequality constraints containing logical integer variables, and a unified state-space equation for the heterogeneous load is established.

[0025] The specific formula for the power surge pulse exponent is as follows: in, The power surge pulse index represents the average drasticness of power jumps in the load per unit time. Indicates the number of sampling points. and This represents the nth and (n+1)th sampling times. express Instantaneous power at a given moment express Instantaneous power at a given moment Indicates the sampling interval. This represents the average power within the sampling window. It represents small positive numbers and is used to prevent the denominator from being zero; The specific formula for the load dynamic inertia coefficient is as follows: in, This represents the load dynamic inertia coefficient, used to characterize the smoothness of load power changes. Represents the power time series of the load. The first derivative of power, The second derivative of power, Indicates the sampling time window; Each load's two features form a feature vector, and the feature vectors of all loads form a feature point set. For each feature point, a local density is defined, and the specific formula is as follows: in, This represents the local density at point i. Let represent the Euclidean distance between the i-th point and the j-th point. and Represents the adaptive bandwidth parameters for the i-th and j-th points, in order to For example, the specific formula is: in, This represents the point m closest to the i-th point. ; For each feature point, a relative distance is defined using the following formula: Among them, for the point with the highest density, its relative distance It equals the maximum distance to all points; for other points, it is the relative distance. It is equal to the distance to the nearest point with a higher density; Calculate the decision value for each point, which is the product of relative distance and local density. Sort the decision values ​​from largest to smallest, and define the difference between the decision values ​​of adjacent points: in, This represents the difference between the decision values ​​of adjacent points. This represents the i-th largest decision value after sorting. The goal is to find the first decision value that satisfies this condition. The value of i, where, The sensitivity coefficient is set to 0.1, and the first i points are the cluster centers. Since the port load is physically divided into two categories, this method will automatically identify two cluster centers. If more than two are identified, the two with the largest decision values ​​are selected. If only one is identified, the feature space is divided according to the midline. For points that are not cluster centers, they are assigned to the category of the nearest cluster center with a higher density, ultimately outputting two clusters: the continuous controllable load cluster CCL and the discrete controllable load cluster DCL. Among them, the continuous controllable load defines loads with thermodynamic isothermal hysteresis characteristics, such as refrigerated containers, as CCLs, and models the power regulation within their temperature control domain as continuous variables; the discrete controllable load defines bridge cranes with trapezoidal power curve characteristics and operating states of start-stop switching, as well as shore power demand, as DCLs. The hybrid logic dynamic modeling method (MLD) is adopted, introducing Boolean variables to transform nonlinear switching dynamics into linear inequality constraints containing logical integer variables, and establishing a unified state-space equation for heterogeneous loads.

[0026] Step S3 specifically includes the following steps: Step S3.1: Set the foundation inertia coefficient and foundation damping coefficient; Step S3.2: Calculate the absolute value of the frequency change rate at the current moment. When the absolute value exceeds the set system safety dead zone threshold, add a virtual inertia compensation component that is positively correlated with the frequency change rate to the system to release or absorb transient kinetic energy. Step S3.3: Calculate the absolute value of the frequency deviation at the current moment, and add a virtual damping compensation component that is positively correlated with the frequency deviation to the system to suppress the frequency drop depth in the initial stage of the sudden increase in the crane load, and accelerate frequency stabilization in the frequency recovery stage to suppress the secondary oscillation phenomenon.

[0027] The specific formulas for the virtual inertia and virtual damping are as follows: in, Represents virtual inertia. Indicates the fundamental inertia coefficient. This represents the adaptive gain coefficient for inertia. Indicates the rate of change of frequency. Indicates the system's safe dead zone threshold. Indicates virtual damping. Indicates the foundation damping coefficient. This represents the damping adaptive gain coefficient. Indicates frequency deviation; When the load on the bridge crane suddenly increases, the rate of frequency change surges, and the system adds a virtual inertia compensation component to absorb the impact kinetic energy; during the frequency recovery phase, the absolute value of the frequency deviation is large, and the system adds a damping compensation component to accelerate frequency stabilization and suppress secondary oscillations.

[0028] Step S4 specifically includes the following steps: Step S4.1: Define the prediction time domain and the control time domain, and take the system frequency deviation, regional tie line power deviation, the adjustable capacity of continuous loads and the switching state of discrete loads as multidimensional state variables. Step S4.2: Construct a quadratic objective function that includes a frequency deviation penalty term, a control increment cost term, a regional tie-line power over-limit penalty term, and a discrete load switching frequency penalty term; Step S4.3: Introduce physical limitations of equipment, operational safety boundaries, and special requirements for port operations as operational constraints. These constraints specifically include global power balance constraints of the microgrid, ramp rate limits for conventional generator sets and peak-shaving units, state-of-charge safety domain constraints for energy storage systems, upper and lower limits for continuous load regulation, and minimum continuous operation and downtime constraints for discrete loads.

[0029] The specific formula for the quadratic objective function is as follows: in, Let k represent the quadratic objective function at the current time k. This indicates the prediction time domain, i.e., the number of steps in the future prediction. This represents the control time domain, i.e., the number of decision variable steps in the optimization calculation. This represents the frequency deviation predicted at time k+p from the predicted frequency at time k. This represents the regional tie-line power deviation predicted at time k+p from time k. This represents the frequency deviation penalty weight, used to adjust the importance of frequency deviation in the objective function. This represents the tie-line power deviation penalty weight, used to suppress tie-line power exceeding limits. This represents the conventional unit power control increment predicted at time k+c from time k. This represents the control increment cost weight, used to suppress drastic changes in control actions. This represents the conventional unit power command predicted at time k+c from time k. This represents the control cost weight, used to penalize excessively large control commands. This represents the total number of Discrete Controllable Loads (DCLs). This represents the switching penalty weight of the d-th DCL, used to suppress frequent start-stop operations. This represents the switching state of the d-th DCL at time k+c, predicted at time k. This indicates the switch state at the previous moment, used to calculate the switching action. This represents the penalty coefficient for slack variables. This represents a slack variable, used to soften constraints; Operational constraints include: global power balance constraints of the microgrid; ramp rate limits for conventional generators and peak-shaving units; state-of-charge safety domain constraints for energy storage systems; upper and lower limits for continuous load regulation; and minimum continuous operation and downtime constraints for discrete loads.

[0030] Step S5 specifically includes the following steps: Step S5.1: Transform the quadratic objective function and multivariate constraints into a mixed-integer linear quadratic programming mathematical problem; Step S5.2: Using a hybrid solution algorithm combining the branch and bound method and the effective set method, rolling optimization calculations are performed within the given discrete control period to extract the first set of elements of the control sequence as the reference command for the total frequency regulation active power at the current moment. Step S5.3: Use one of the particle swarm optimization algorithm, gray wolf optimization algorithm or differential evolution algorithm to perform online adaptive optimization of the weight matrix in the hybrid model predictive control, and dynamically update the penalty weight according to the port load fluctuation law under different seasons and meteorological conditions to cope with the nonlinear perturbation of port microgrid system parameters and strong external interference.

[0031] The above quadratic objective function and multivariate constraints are transformed into a mixed-integer linear quadratic programming problem. A hybrid solution algorithm combining the branch and bound method and the effective set method is adopted to perform rolling optimization calculations within a given discrete control period. The first set of elements of the control sequence is extracted as the reference command for the total frequency regulation active power at the current moment.

[0032] The weight matrix in the predictive control of the hybrid model is adaptively optimized online using one of the particle swarm optimization algorithm (PSO), gray wolf optimization algorithm (GWO), or differential evolution algorithm (DE). The penalty weight is dynamically updated according to the port load fluctuation pattern under different seasons and meteorological conditions to cope with the nonlinear perturbation of port microgrid system parameters and strong external interference.

[0033] Step S6 specifically includes the following steps: Step S6.1: Obtain the charge-discharge cycle life loss state model of the lithium-ion battery and the real-time state of charge feedback value of the supercapacitor. Step S6.2: Construct a dynamic adjustment mechanism for the filter time constant, and extract the low-frequency steady-state power deviation by smoothing the low-pass filter; Step S6.3: Distribute the filtered low-frequency commands to the lithium-ion battery cells and strip the unfiltered high-frequency transient power fluctuations and distribute them to the supercapacitors. Step S6.4: When the state of charge of the supercapacitor deviates from the lower limit of the safe operating range, the low-pass filter time constant is dynamically reduced and a compensation power term proportional to the state of charge deviation is introduced. Part of the high-frequency power and compensation power are transferred to the lithium-ion battery, thereby ensuring that the supercapacitor continuously provides high-frequency modulation capability and avoiding leakage current.

[0034] like Figure 2 As shown, the port microgrid system uses the port microgrid bus as its core hub. The system includes: a power supply side, comprising a large power grid / gas turbine unit, offshore wind farm, and photovoltaic array, with energy input to the bus; a hybrid energy storage system, comprising supercapacitors and lithium-ion battery units, connected to the bus; and a multi-source heterogeneous load side, comprising container cranes, refrigerated containers / cold chain, and shore power for berthed ships. Among these, cranes and shore power are defined as discrete controllable loads (DCL), while refrigerated containers are defined as continuous controllable loads (CCL). The loads obtain power from the bus.

[0035] like Figure 3 As shown, the two-layer collaborative frequency regulation control logic includes: a sensing layer, which is a real-time state sensing module for the microgrid, acquiring frequency deviation and frequency change rate; a bottom-layer control layer, which is an adaptive virtual synchronization control based on transient indicators, including adaptive virtual inertia and adaptive damping coefficient, outputting transient kinetic energy compensation signals to the top layer; a top-layer control layer, which is a hybrid model predictive control layer, including a heterogeneous load hybrid logic dynamic model, state variable evolution, multi-objective cost function, and swarm intelligence online optimization and mixed integer linear quadratic programming solver, outputting a rolling optimization sequence; and an output layer, which is the total frequency regulation active power reference command.

[0036] In this embodiment, the total command output by the HMPC contains a massive amount of complex spectral components. A low-pass filter with dynamically adjustable cutoff frequency is used to smoothly distribute low-frequency, gentle commands to the lithium-ion battery, delaying its lifespan decline; unfiltered high-frequency transient power surges are stripped away and absorbed by the supercapacitor. When the supercapacitor's SOC falls below a safe threshold, the time constant is automatically reduced, high-frequency components are smoothly transferred back to the battery, and a trickle-current compensation component is introduced to maintain the supercapacitor's energy reserves to withstand the next impact.

[0037] like Figure 4 As shown, an adaptive cooperative frequency modulation control system for port multi-source loads is used to implement any one of the adaptive cooperative frequency modulation control methods for port multi-source loads, including: The multi-source data acquisition module is used to acquire high-resolution power parameters of multi-source loads at the port and real-time power parameters of the renewable energy supply side. The heterogeneous load sensing and classification module is used to extract features and perform cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. The virtual inertia adaptive module is used to monitor the system frequency deviation and frequency change rate of the microgrid in real time, and adaptively and dynamically adjust the virtual inertia parameters and damping coefficient on the control side of the microgrid inverter. The hybrid predictive optimization module is used to construct and solve the objective function and operating constraints of the hybrid model predictive control based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, and generate the total frequency regulation active power reference command. The frequency domain decoupling and power allocation module is used to decouple the total frequency modulation active power reference command into high-frequency power command and low-frequency power command through an adaptive low-pass filter, based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, and allocate them to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

[0038] The multi-source data acquisition module collects the transient stator current of the bridge crane motor in real time through a distributed Internet of Things sensor network, obtains the thermal cycle status of the refrigerated container through infrared thermal imaging and bus communication, and performs nanosecond-level timestamp synchronization. The overall data sampling frequency of the system is not less than 1kHz.

[0039] The frequency domain decoupling and power distribution module is equipped with an anti-aliasing filter and a droop control interface, which are used to provide continuous underlying voltage and frequency support for the port microgrid in off-grid island operation mode and under extreme weather disturbances.

[0040] The hybrid prediction optimization module is deployed in the centralized energy management system or edge computing node of the port microgrid. It interacts with the underlying distributed energy storage inverter main controller in a closed loop through industrial Ethernet. A communication delay compensation mechanism is configured to address communication delays. The communication delay compensation mechanism uses a predefined time observer and extended Kalman filter for system state prediction and advance correction control.

[0041] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0042] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

Claims

1. An adaptive cooperative frequency modulation control method for multi-source loads in ports, characterized in that, Includes the following steps: Step S1: Obtain high-resolution power parameters of the port's multi-source load and real-time power parameters of the renewable energy supply side. The power parameters include the transient stator current of the bridge crane motor, the thermal cycle status of the refrigerated container, the power of the shore power system, the real-time output of photovoltaic and wind power, and the frequency of the microgrid common coupling point. Step S2: Perform feature extraction and cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. Step S3: Monitor the system frequency deviation and frequency change rate of the microgrid in real time. Based on the frequency deviation and frequency change rate, adaptively and dynamically adjust the virtual inertia parameters and damping coefficients on the control side of the microgrid inverter to construct an adaptive virtual synchronous generator control model. Step S4: Based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, construct the objective function and operating constraints of the hybrid model predictive control; Step S5: Solve the objective function of the hybrid model predictive control to generate a total frequency regulation active power reference command for smoothing frequency fluctuations; Step S6: Based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, the total frequency modulation active power reference command is decoupled into high-frequency power command and low-frequency power command through an adaptive low-pass filter, and then distributed to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

2. The method according to claim 1, characterized in that, Step S2 specifically includes the following steps: Step S2.1: Extract two features from the power time series of each load: the power mutation pulse exponent and the load dynamic inertia coefficient, and construct a two-dimensional feature vector; Step S2.2: Perform cluster analysis on the two-dimensional feature vector using an adaptive density peak clustering algorithm. Specifically, this includes: calculating the local density and relative distance of each feature point; constructing decision values ​​and arranging them in descending order; automatically determining cluster centers through decision value difference; assigning non-cluster center points to cluster centers with higher density and the closest distance; and finally outputting continuous controllable load clusters and discrete controllable load clusters. Step S2.3: Define the refrigerated container load with thermodynamic isothermal hysteresis characteristics as a continuous controllable load, and model its power regulation within its temperature control domain as a continuous variable; define the bridge crane equipment with trapezoidal power curve characteristics and operating state in the form of start-stop jump and the shore power demand for berthing as discrete controllable loads; Step S2.4: Using a hybrid logic dynamic modeling method, the nonlinear switching dynamics of the discrete controllable load is transformed into linear inequality constraints containing logical integer variables, and a unified state-space equation for the heterogeneous load is established.

3. The method according to claim 2, characterized in that, Step S3 specifically includes the following steps: Step S3.1: Set the foundation inertia coefficient and foundation damping coefficient; Step S3.2: Calculate the absolute value of the frequency change rate at the current moment. When the absolute value exceeds the set system safety dead zone threshold, add a virtual inertia compensation component that is positively correlated with the frequency change rate to the system to release or absorb transient kinetic energy. Step S3.3: Calculate the absolute value of the frequency deviation at the current moment, and add a virtual damping compensation component that is positively correlated with the frequency deviation to the system to suppress the frequency drop depth in the initial stage of the sudden increase in the crane load, and accelerate frequency stabilization in the frequency recovery stage to suppress the secondary oscillation phenomenon.

4. The method according to claim 3, characterized in that, Step S4 specifically includes the following steps: Step S4.1: Define the prediction time domain and the control time domain, and take the system frequency deviation, regional tie line power deviation, the adjustable capacity of continuous loads and the switching state of discrete loads as multidimensional state variables. Step S4.2: Construct a quadratic objective function that includes a frequency deviation penalty term, a control increment cost term, a regional tie-line power over-limit penalty term, and a discrete load switching frequency penalty term; Step S4.3: Introduce physical limitations of equipment, operational safety boundaries, and special requirements for port operations as operational constraints. These constraints specifically include global power balance constraints of the microgrid, ramp rate limits for conventional generator sets and peak-shaving units, state-of-charge safety domain constraints for energy storage systems, upper and lower limits for continuous load regulation, and minimum continuous operation and downtime constraints for discrete loads.

5. The method according to claim 4, characterized in that, Step S5 specifically includes the following steps: Step S5.1: Transform the quadratic objective function and multivariate constraints into a mixed-integer linear quadratic programming mathematical problem; Step S5.2: Using a hybrid solution algorithm combining the branch and bound method and the effective set method, rolling optimization calculations are performed within the given discrete control period to extract the first set of elements of the control sequence as the reference command for the total frequency regulation active power at the current moment. Step S5.3: Use one of the particle swarm optimization algorithm, gray wolf optimization algorithm or differential evolution algorithm to perform online adaptive optimization of the weight matrix in the hybrid model predictive control, and dynamically update the penalty weight according to the port load fluctuation law under different seasons and meteorological conditions to cope with the nonlinear perturbation of port microgrid system parameters and strong external interference.

6. The method according to claim 5, characterized in that, Step S6 specifically includes the following steps: Step S6.1: Obtain the charge-discharge cycle life loss state model of the lithium-ion battery and the real-time state of charge feedback value of the supercapacitor. Step S6.2: Construct a dynamic adjustment mechanism for the filter time constant, and extract the low-frequency steady-state power deviation by smoothing the low-pass filter; Step S6.3: Distribute the filtered low-frequency commands to the lithium-ion battery cells and strip the unfiltered high-frequency transient power fluctuations and distribute them to the supercapacitors. Step S6.4: When the state of charge of the supercapacitor deviates from the lower limit of the safe operating range, the low-pass filter time constant is dynamically reduced and a compensation power term proportional to the state of charge deviation is introduced. Part of the high-frequency power and compensation power are transferred to the lithium-ion battery, thereby ensuring that the supercapacitor continuously provides high-frequency modulation capability and avoiding leakage current.

7. An adaptive cooperative frequency modulation control system for multi-source loads in ports, used to implement the adaptive cooperative frequency modulation control method for multi-source loads in ports as described in any one of claims 1-6, characterized in that, include: The multi-source data acquisition module is used to acquire high-resolution power parameters of multi-source loads at the port and real-time power parameters of the renewable energy supply side. The heterogeneous load sensing and classification module is used to extract features and perform cluster analysis on the high-resolution power parameters, establish a heterogeneous load classification model oriented towards port operation characteristics, and divide the port load into continuous controllable load and discrete controllable load. The virtual inertia adaptive module is used to monitor the system frequency deviation and frequency change rate of the microgrid in real time, and adaptively and dynamically adjust the virtual inertia parameters and damping coefficient on the control side of the microgrid inverter. The hybrid predictive optimization module is used to construct and solve the objective function and operating constraints of the hybrid model predictive control based on the heterogeneous load classification model and the adaptive virtual synchronous generator control model, and generate the total frequency regulation active power reference command. The frequency domain decoupling and power allocation module is used to decouple the total frequency modulation active power reference command into high-frequency power command and low-frequency power command through an adaptive low-pass filter, based on the charging and discharging characteristics and frequency domain response capability of the hybrid energy storage system, and allocate them to the supercapacitor unit and the lithium-ion battery unit for coordinated output.

8. The system according to claim 7, characterized in that, The multi-source data acquisition module collects the transient stator current of the bridge crane motor in real time through a distributed Internet of Things sensor network, obtains the thermal cycle status of the refrigerated container through infrared thermal imaging and bus communication, and performs nanosecond-level timestamp synchronization. The overall data sampling frequency of the system is not less than 1kHz.

9. The system according to claim 8, characterized in that, The frequency domain decoupling and power distribution module is equipped with an anti-aliasing filter and a droop control interface, which are used to provide continuous underlying voltage and frequency support for the port microgrid in off-grid island operation mode and under extreme weather disturbances.

10. The system according to claim 9, characterized in that, The hybrid prediction optimization module is deployed in the centralized energy management system or edge computing node of the port microgrid. It interacts with the underlying distributed energy storage inverter main controller in a closed loop through industrial Ethernet. A communication delay compensation mechanism is configured to address communication delays. The communication delay compensation mechanism uses a predefined time observer and extended Kalman filter for system state prediction and advance correction control.

Images