Network construction type bidirectional low-harmonic energy storage converter based on virtual synchronous control and intelligent load identification method thereof

Abstract

The invention provides a network construction type bidirectional low-harmonic energy storage converter based on virtual synchronous control and an intelligent load identification method thereof, and belongs to the field of power electronics. The converter comprises a power module and a control module, wherein the power module is used for bidirectional conversion and output of electric energy; the control module comprises a virtual synchronous control module and a harmonic suppression control module, and the virtual synchronous control module adopts a virtual synchronous generator technology to realize active / reactive power bidirectional adjustment during electric energy bidirectional conversion of the power module and network construction level voltage and frequency support during electric energy output; and the harmonic suppression module is provided with a parallel QPR controller group (comprising a first QPR controller, a second QPR controller and a third QPR controller) which are used for processing third-time, fifth-time and seventh-time characteristic harmonics respectively and suppressing the characteristic harmonics of the electric energy output side of the power module. According to the invention, the dual modules are arranged in the control module, so that the cooperation of a network construction function and low harmonic output is realized, and the electric energy quality and the system operation stability are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of power electronics technology, specifically relating to a grid-type bidirectional low harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method. Background Technology

[0002] In the development of modern society, the stability and efficiency of power supply are crucial to all sectors. Especially in the construction industry, the reliability of temporary power systems directly impacts project progress and construction safety, making its importance self-evident.

[0003] Traditional construction site power supply solutions often rely on oil-immersed transformers connected to the mains power grid or diesel generator sets as emergency power. However, oil-immersed transformers are bulky and heavy, making transportation and installation difficult in mountainous or remote construction sites. While diesel generators can operate independently, they suffer from significant noise pollution, high fuel costs, and slow start-up response. Furthermore, when the main power grid is interrupted due to faults or maintenance, traditional solutions struggle to provide seamless, uninterrupted power supply, posing a serious risk to critical construction processes such as continuous concrete pouring. In addition, while energy storage technology has advanced and lithium iron phosphate batteries offer numerous advantages, efficiently converting their DC power into the AC power required by construction equipment and maintaining stable power quality under complex load conditions still presents many technical challenges.

[0004] In practical engineering applications, existing grid-type energy storage converters have insufficient harmonic suppression capabilities. Harmonics generated by nonlinear loads not only reduce power quality but may also cause system resonance. Summary of the Invention

[0005] In view of this, the present invention provides a grid-type bidirectional low harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method, aiming to solve the above-mentioned shortcomings of grid-type energy storage converters in practical engineering applications.

[0006] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0007] In a first aspect, the present invention provides a grid-type bidirectional low harmonic energy storage converter based on virtual synchronous control, comprising: a power module and a control module; the control module is signal-connected to the power module;

[0008] The power module is configured to achieve bidirectional power conversion and power output;

[0009] The control module includes a virtual synchronization control module and a harmonic suppression control module;

[0010] The virtual synchronous control module is configured to use virtual synchronous generator technology to achieve bidirectional regulation of active and reactive power during the bidirectional power conversion process of the power module, as well as grid-level voltage and frequency support during the power output process.

[0011] The harmonic suppression module is equipped with a parallel QPR controller group; the parallel QPR controller group includes at least a first QPR controller, a second QPR controller and a third QPR controller;

[0012] The harmonic suppression module is configured to suppress characteristic harmonics on the power output side of the power module through a parallel QPR controller group; wherein the first QPR controller, the second QPR controller and the third QPR controller handle the 3rd, 5th and 7th characteristic harmonics, respectively.

[0013] Furthermore, the transfer function of the parallel QPR controller group is:

[0014]

[0015] In the formula, G total (s) is the transfer function of the total controller; G fundamental (s) is the fundamental wave controller transfer function; G QPRn (s) is the QPR controller transfer function for the nth harmonic;

[0016] The closed-loop transfer function of the grid-type bidirectional low-harmonic energy storage converter is:

[0017]

[0018] In the formula, H cl (s) is the closed-loop transfer function of the converter; G inv (s) is the inverter transfer function, G fb (s) is the transfer function of the feedback loop.

[0019] Furthermore, the parameter configuration for the parallel QPR controller group is as follows:

[0020] First harmonic QPR controller: Gain K r3 =12, cutoff frequency ω c3 =4 rad / s, resonant angular frequency =942.5 rad / s;

[0021] Second harmonic QPR controller: Gain K r5 =15, cutoff frequency ω c5 =5 rad / s, resonant angular frequency =1570.8 rad / s;

[0022] Third harmonic QPR controller: Gain K r7=18, cutoff frequency ω c7 =6 rad / s, resonant angular frequency =2199.1 rad / s.

[0023] Furthermore, the virtual synchronization control module is configured to simulate the electromechanical transient characteristics of a synchronous generator through the rotor motion equation and to achieve autonomous reactive power adjustment through virtual excitation control.

[0024] The rotor motion equation is:

[0025]

[0026] In the formula, J is the virtual moment of inertia; ω is the virtual rotor angular velocity; T m T represents mechanical torque. e ω is the electromagnetic torque; D is the damping coefficient; ω0 is the rated angular velocity;

[0027] The control law for virtual excitation control is:

[0028]

[0029] In the formula, E is the amplitude of the virtual internal potential; E0 is the no-load potential; k q Q is the reactive power-voltage droop factor. ref V is the reactive power reference value; Q is the actual output reactive power; ref V is the voltage reference value; k is the actual output voltage; v This is the voltage-regulated integral gain.

[0030] Furthermore, the virtual synchronization control module also includes a virtual impedance control unit, the voltage equation of which in the dq coordinate system is:

[0031]

[0032] In the formula, v d and v q These represent the d-axis and q-axis output voltage components of the converter in the dq rotating coordinate system, respectively; e d and e q These represent the d-axis and q-axis potential components of the virtual synchronous generator in the dq coordinate system, respectively; R v For virtual resistance; L v For virtual inductance; i d and i q These represent the d-axis output current component and the q-axis output current component of the converter in the dq rotating coordinate system, respectively.

[0033] Furthermore, it also includes an energy storage capacity optimization configuration module, which is configured to achieve the optimal configuration of energy storage capacity based on a multi-objective optimization model and an improved particle swarm optimization algorithm;

[0034] The objective function of the multi-objective optimization model is:

[0035]

[0036] In the formula, F obj To optimize the comprehensive evaluation index for multiple objectives; C total The total cost over the entire system lifecycle; LPSP represents the probability of power loss; T response The system's dynamic response time is the indicator; α1, α2, and α3 are weighting coefficients.

[0037] The formula for calculating the total cost of the system's entire lifecycle is:

[0038]

[0039] In the formula, C initial N represents the initial investment cost; C represents the total lifespan (in years); om (t) represents the operating and maintenance cost in year t; r is the discount rate; C replacement Cost of equipment replacement.

[0040] Furthermore, it also includes a deep reinforcement learning energy management module, which is configured to optimize energy storage charging and discharging strategies based on a deep Q-network algorithm;

[0041] The optimization objective function for energy management is:

[0042]

[0043] In the formula, J represents the optimization objective of energy management; T represents the total number of optimization periods; C grid (t) represents the grid electricity price at time t; P grid (t) represents the power drawn from the grid at time t; Δt is the duration of each scheduling period; λ is the weighting coefficient; L battery (t) represents the loss cost of the energy storage battery at time t;

[0044] The objective function is transformed into a Markov decision process based on the deep Q-network algorithm. The state space of the deep Q-network algorithm is:

[0045]

[0046] In the formula, S(t) is the state space at time t; SOC(t) is the state space at time t; P load (t) represents the state of charge of the battery at time t; t hourt represents the current hour; day This is the current date.

[0047] Furthermore, the power module includes at least one power conversion module, which adopts a three-level NPC topology. When multiple power conversion modules are connected in parallel, a current sharing control strategy is employed. The control equation for the current sharing control strategy is as follows:

[0048]

[0049] In the formula, P i and Q i These are the actual output active power and actual output reactive power of the i-th power conversion module, respectively. and These are the rated active power and rated reactive power of the i-th power conversion module, respectively; and These are the active power droop coefficient and reactive power droop coefficient of the i-th power conversion module, respectively; f i and V i These are the local output frequency and output voltage, respectively; f0 and V0 are the rated frequency and rated voltage, respectively.

[0050] Furthermore, the power module also includes an energy storage module, which uses a lithium iron phosphate battery pack. The DC / DC converter control strategy for the energy storage module is as follows:

[0051]

[0052] In the formula, P is the reference current of the DC / DC converter corresponding to the k-th energy storage module; bat,total For; V dc C is the DC bus voltage. k C represents the rated capacity of the k-th energy storage module. total k represents the total capacity of the entire energy storage system. soc This is the equilibrium coefficient based on SOC.

[0053] Secondly, the present invention provides an intelligent load identification method for a grid-type bidirectional low-harmonic energy storage converter based on virtual synchronization control, comprising the following steps:

[0054] Collect multidimensional operating data of the load and extract the power characteristics, current characteristics, frequency domain characteristics and time domain characteristics of the multidimensional operating data;

[0055] Power characteristics, current characteristics, frequency domain characteristics, and time domain characteristics are input into a pre-built load classifier to identify the load type and obtain the load identification result.

[0056] Adjust power supply parameters based on load identification results.

[0057] In summary, this invention provides a grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method. The converter includes a power module and a control module; the control module is signal-connected to the power module; the power module is configured to realize bidirectional energy conversion and energy output; the control module includes a virtual synchronous control module and a harmonic suppression control module; the virtual synchronous control module is configured to use virtual synchronous generator technology to realize bidirectional regulation of active and reactive power during the bidirectional energy conversion process of the power module, as well as grid-level voltage and frequency support during the energy output process; the harmonic suppression module is provided with a parallel QPR controller group; the parallel QPR controller group includes at least a first QPR controller, a second QPR controller, and a third QPR controller; the harmonic suppression module is configured to suppress characteristic harmonics on the energy output side of the power module through the parallel QPR controller group; wherein, the first QPR controller, the second QPR controller, and the third QPR controller process the 3rd, 5th, and 7th characteristic harmonics, respectively. This invention solves the problems of insufficient harmonic suppression capability, easy system resonance, and poor power quality of existing grid-type energy storage converters by setting a virtual synchronization control module and a harmonic suppression control module with parallel QPR controller groups in the control module of the grid-type bidirectional energy storage converter. It realizes the synergy between grid-type function and low harmonic output, and effectively improves power quality and system operation stability. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 The overall architecture diagram of a network-type bidirectional low harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method for coordinated control provided in the embodiments of the present invention is shown.

[0060] Figure 2 A flowchart illustrating the coordinated control of a grid-type bidirectional low-harmonic energy storage converter and its intelligent load identification method based on virtual synchronous control, provided in an embodiment of the present invention.

[0061] Figure 3 The figure shows the simulation verification results of the harmonic suppression performance of the multi-QPR controller provided by this invention.

[0062] Figure 4 The diagram shows the verification results of the dynamic response of VSG network control provided by this invention.

[0063] Figure 5 The simulation verification results of the energy storage capacity optimization configuration (improved PSO algorithm) provided by the present invention are shown in the figure. Detailed Implementation

[0064] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0065] The following section first introduces some of the technical terms used in the prior art involved in this invention:

[0066] (1) Grid-Forming Converter: A power electronic converter that can actively establish voltage and frequency references. It has inertial and damping characteristics similar to synchronous generators and can independently support the operation of the power grid without relying on the external power grid to provide voltage and frequency references.

[0067] (2) Virtual Synchronous Generator (VSG): By simulating the rotor motion equation and excitation characteristics of a traditional synchronous generator through a control algorithm, the power electronic converter has rotational inertia and damping coefficient, enabling it to participate in system frequency regulation and provide voltage support.

[0068] (3) Total Harmonic Distortion (THD): describes the ratio of harmonic components in a signal to the fundamental frequency. It is an important indicator for evaluating power quality. The calculation formula is the ratio of the square root of the sum of the squares of the effective values ​​of each harmonic to the effective value of the fundamental frequency.

[0069] (4) Non-Intrusive Load Monitoring (NILM): A technology that identifies and monitors the status of each electrical device by analyzing the overall current or voltage waveform characteristics, without the need to install a dedicated sensor at each load end.

[0070] (5) Quasi-Proportional Resonant Controller (QPR): An improved controller that introduces a bandwidth parameter on the basis of the traditional proportional resonant controller, which improves the robustness of the system while maintaining high gain for a specific frequency signal.

[0071] (6) Active Neutral Point Clamped (ANPC): A three-level inverter topology that uses fully controlled power devices to achieve neutral point clamping. Compared with the traditional diode clamping type, it can achieve a more balanced loss distribution and more flexible control.

[0072] Secondly, the technical background of the present invention will be further introduced.

[0073] In modern construction, the reliability of temporary power systems directly impacts project progress and construction safety. Traditional construction site power supply solutions primarily rely on oil-immersed transformers connected to the mains or diesel generator sets to provide emergency power, but these solutions have significant limitations. Oil-immersed transformers are bulky and heavy, posing numerous difficulties for transportation and installation in mountainous or remote construction sites; while diesel generators have independent operating capabilities, they are constrained by noise pollution, high fuel costs, and slow start-up response. More importantly, when the main power grid is interrupted due to faults or maintenance, traditional solutions cannot provide seamless, uninterrupted power supply, posing a significant risk to critical construction processes such as continuous concrete pouring operations.

[0074] Advances in energy storage technology have made the transformation and upgrading of construction site power supply systems possible. Lithium iron phosphate batteries, with their high safety due to their olivine crystal structure, cycle life exceeding 2000 cycles, and stable performance over a wide temperature range of -10℃ to 45℃, have become an ideal choice for outdoor energy storage applications. However, efficiently converting the DC power from energy storage batteries into the 220V / 380V AC power required by construction equipment, and maintaining stable power quality under complex load conditions, still faces multiple technical challenges.

[0075] The electrical environment at construction sites presents distinct characteristics. Typical equipment such as mixers, cranes, and welding machines exhibit significantly different load characteristics: the inrush current generated when a mixer starts can reach 3 to 7 times its rated value, and the load fluctuates periodically with changes in material properties; the power demand of cranes changes non-linearly during lifting heavy objects, resulting in severe transient power fluctuations; and rectifier-type equipment such as welding machines generate substantial harmonic pollution. These characteristics place stringent demands on the dynamic response speed, overload capacity, and harmonic suppression capabilities of the power supply system. Traditional grid-connected energy storage converters rely on phase-locked loops to track grid voltage, lacking active grid-connection capabilities in islanded operation mode, and are prone to voltage drops or even system instability when facing sudden load switching.

[0076] The concept of grid-based control has opened up a new path for solving this problem. Unlike grid-following control, which passively follows the power grid, grid-based converters actively establish voltage and frequency references and simulate the mechanical inertia and electromagnetic damping characteristics of synchronous machines through a virtual synchronous generator control strategy. When the load changes abruptly, the virtual inertia can smooth the frequency fluctuation rate, while the virtual damping suppresses the frequency deviation amplitude. This mechanism provides an essential guarantee for the stable operation of the system. However, in practical engineering applications, grid-based energy storage converters still face three core technical bottlenecks: First, insufficient harmonic suppression capability. Harmonics generated by nonlinear loads not only reduce power quality but may also cause system resonance. Second, low load identification accuracy leads to energy management efficiency losses, making it impossible to optimize charging and discharging strategies according to load characteristics. Third, the contradiction between lightweight equipment and performance indicators. Construction sites have an urgent need for equipment portability, but high power density designs are often accompanied by challenges in thermal management and reliability.

[0077] In the field of harmonic suppression technology, existing research mainly focuses on two directions: filter optimization and modulation strategy improvement. Passive LC filters, while inexpensive, are bulky and have a fixed resonant frequency, making them difficult to handle dynamically changing harmonic spectra. Active power filters can achieve dynamic compensation but increase system complexity and cost. Multilevel inverter topologies reduce voltage steps by increasing the number of output levels, thereby reducing high-frequency harmonic content. Among three-level topologies, the diode-neutral-point-clamped (NPC) type is widely used due to its simple structure, but its inherent midpoint potential imbalance and uneven distribution of internal and external transistor losses limit performance improvement. The active-neutral-point-clamped (ANPC) topology overcomes the shortcomings of the NPC by introducing additional fully controlled devices, but the increased control complexity also brings new challenges.

[0078] Virtual synchronous generator (VSG) technology has been successfully applied in the field of renewable energy grid connection, but its parameter tuning methods are mainly designed for grid-connected operating conditions. In the scenario of isolated power supply on construction sites, how to maintain sufficient stability margin while ensuring dynamic response speed, and how to adaptively adjust virtual inertia and damping coefficient according to load characteristics, are issues that directly affect the practicality of grid-based energy storage systems. In addition, traditional VSG control often neglects the coordinated design of harmonic suppression components, resulting in the amplification of high-frequency harmonic components while suppressing low-frequency power oscillations. This contradiction is particularly prominent when the proportion of rectifier-type loads is high on construction sites.

[0079] Load characteristic identification is fundamental to the intelligent scheduling of energy storage systems. Traditional methods rely on modeling the electrical parameters of the load, but the diverse types of construction equipment and their complex operating conditions make it difficult to establish a universal mathematical model. In recent years, non-intrusive load monitoring technology based on machine learning has shown promising promise, classifying loads by extracting time-frequency domain features of current and voltage waveforms. However, existing research primarily focuses on residential power consumption scenarios, and the accuracy in identifying high-power inductive loads and impulsive loads specific to construction sites still needs improvement. Furthermore, the real-time performance and computational complexity of load identification algorithms directly impact their deployability in resource-constrained embedded controllers.

[0080] Through in-depth analysis of existing technical solutions, the following main defects can be summarized, which are precisely the technical obstacles that this invention aims to overcome:

[0081] (1) Insufficient harmonic suppression capability, power quality is difficult to meet the needs of sensitive equipment.

[0082] Existing grid-connected energy storage converters often prioritize power balance and frequency stability in their control strategy design, resulting in relatively simplistic handling of harmonic suppression. Traditional PI controllers, inherent in AC steady-state errors, exhibit insufficient suppression capability against characteristic harmonics generated by nonlinear loads such as rectifiers and frequency converters. While some studies have introduced proportional resonant (PR) controllers to enhance regulation at specific frequencies, the inherent contradiction between the theoretical assumption of infinite gain at the resonant frequency and the bandwidth limitations of actual systems makes them prone to control failure during grid frequency fluctuations or rapid load changes. Furthermore, traditional methods mostly employ single harmonic suppression strategies, failing to differentiate between the generation mechanisms and propagation characteristics of different harmonics. This can lead to the amplification of other harmonic components while suppressing certain harmonics. This deficiency is particularly pronounced in construction scenarios, where the harmonic spectrum generated by equipment such as welding machines often covers multiple characteristic frequencies, including the 3rd, 5th, and 7th harmonics. Optimizing the fundamental frequency alone is insufficient to achieve full-band harmonic suppression.

[0083] (2) Virtual synchronous control parameter tuning lacks adaptive capability, making it difficult to balance dynamic performance and stability.

[0084] The core of virtual synchronous generator (VSG) control lies in simulating the inertia and damping characteristics of a synchronous machine through software algorithms. However, the setting of these virtual parameters directly affects the dynamic response and stability margin of the system. Existing research mostly adopts a fixed-parameter VSG control strategy, where the virtual inertia and damping coefficients are determined offline based on a linearized small-signal model before the system is put into operation. This method is acceptable in grid-connected scenarios with relatively stable loads, but it faces severe challenges in isolated power supply modes on construction sites. When the mixer starts, the instantaneous power jumps from zero to tens of kilowatts. If the virtual inertia is set too large, although it can effectively suppress the rate of frequency change, it will lead to excessively long settling time and slow frequency recovery. If the virtual inertia is too small, the system response is rapid, but the frequency overshoot is large and may even trigger frequency protection. To complicate matters further, different types of loads have conflicting requirements for the optimal VSG parameters: inductive loads require large damping to suppress oscillations, while capacitive loads require small damping to ensure response speed. Traditional fixed-parameter schemes cannot be adjusted in real time according to load characteristics and can only adopt a compromise strategy during the design phase, which inevitably leads to poor system performance under certain operating conditions.

[0085] (3) The load identification method relies on prior knowledge and has weak generalization ability for complex working conditions at the construction site.

[0086] Existing load identification technologies are mainly divided into two categories: physical model-based methods and feature matching-based methods. The former requires establishing accurate mathematical models for various loads, including equivalent circuit parameters and power factor characteristics. However, construction equipment comes in many models and operates under complex conditions, making it difficult to establish a universal model. For example, the torque characteristics and power fluctuation patterns of the same mixer will change significantly when mixing concrete with different proportions. This condition-dependent nature leads to low accuracy for identification methods based on fixed models. The latter extracts feature quantities from current and voltage waveforms, such as RMS values, power factors, and higher harmonic content, for pattern matching. However, feature extraction algorithms are often optimized for specific application scenarios, and their effectiveness is greatly reduced when migrated to new scenarios. More importantly, both methods fall into the category of offline identification, requiring the load to run for a period of time before identification can be completed. They cannot provide a judgment result the instant the equipment starts up, and this latency constitutes a significant weakness in energy storage scheduling that requires rapid response. Furthermore, the load decoupling problem when multiple devices operate simultaneously has not been effectively solved. When cranes and mixers operate in parallel, the system can only monitor the overall current waveform, making it difficult to separate the contributions of each device.

[0087] To address the aforementioned deficiencies in the existing technology, embodiments of the present invention provide a grid-type bidirectional low-harmonic energy storage converter based on virtual synchronization control. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronization control includes: a power module and a control module; the control module is signal-connected to the power module.

[0088] The power module is configured to achieve bidirectional power conversion and power output;

[0089] The control module includes a virtual synchronization control module and a harmonic suppression control module;

[0090] The virtual synchronous control module is configured to use virtual synchronous generator technology to achieve bidirectional regulation of active and reactive power during the bidirectional power conversion process of the power module, as well as grid-level voltage and frequency support during the power output process.

[0091] The harmonic suppression module is equipped with a parallel QPR controller group; the parallel QPR controller group includes at least a first QPR controller, a second QPR controller and a third QPR controller;

[0092] The harmonic suppression module is configured to suppress characteristic harmonics on the power output side of the power module through a parallel QPR controller group; wherein the first QPR controller, the second QPR controller and the third QPR controller handle the 3rd, 5th and 7th characteristic harmonics, respectively.

[0093] In this embodiment, the power module is the main execution unit of the entire converter, undertaking the basic functions of bidirectional power conversion and power output. Through the orderly switching of power electronic switching devices, it realizes the bidirectional flow of DC and AC power between the energy storage unit and the grid or load: it can convert AC power into DC power for storage in the energy storage unit during grid off-peak hours or when there is a demand for energy storage charging, and it can also invert DC power into AC power for output to the grid or load during grid peak hours or when there is a demand for load power supply. The control module, as the decision-making and regulation unit, consists of a virtual synchronization control module and a harmonic suppression control module. The virtual synchronization control module is based on virtual synchronous generator technology. By simulating the electromechanical transient characteristics and excitation regulation mechanism of a synchronous generator, it achieves bidirectional regulation of active and reactive power during the bidirectional power conversion process, ensuring the controllability of energy flow. On the other hand, it provides grid-level voltage and frequency support during power output, enabling the converter to have the ability to operate independently in a grid and meet the technical requirements of grid-type equipment. The harmonic suppression control module is specifically configured with a parallel QPR (quasi-proportional resonance) controller group, which contains at least three independent QPR controllers, corresponding to the 3rd, 5th and 7th characteristic harmonics respectively. By designing differentiated parameters for the resonant frequency, gain and bandwidth of each controller, it can be used to handle the main characteristic harmonic frequencies generated by nonlinear loads.

[0094] During operation, the virtual synchronization control module first generates active and reactive power commands, as well as voltage and frequency reference signals for the power module based on the system operating status (such as grid frequency, voltage, load power demand, and energy storage SOC status). This ensures that the converter maintains frequency and voltage stability of the grid or microgrid during bidirectional charging and discharging. Simultaneously, the harmonic suppression control module collects the AC current and voltage signals from the output side of the power module in real time. It extracts the 3rd, 5th, and 7th characteristic harmonic components through a harmonic detection algorithm. Then, it activates the corresponding parallel QPR controller group. Each QPR controller, through its quasi-resonant characteristics, forms a high-gain regulation at the target harmonic frequency to compensate for and suppress the current deviation caused by the corresponding harmonic, thereby offsetting the harmonic distortion effect caused by nonlinear loads. Ultimately, the AC power output by the power module meets both the stability requirements of grid operation and has high-quality characteristics with low harmonic distortion rate, achieving a synergistic realization of grid capability and power quality optimization.

[0095] To address the differences in the generation mechanisms and propagation characteristics of the 3rd, 5th, and 7th harmonics, this embodiment employs a differentiated design using parallel QPR controller groups. Each controller focuses on suppressing a specific harmonic, avoiding the problem that traditional single-control strategies might amplify other harmonics while suppressing a particular harmonic, thus achieving corresponding and balanced suppression of multiple frequency harmonics. Secondly, compared to the steady-state error defects of traditional PI controllers and the control failure problem of ideal PR controllers during frequency fluctuations, the quasi-resonant characteristics of the QPR controller are achieved through a rationally designed bandwidth (introducing the cutoff frequency parameter ω). c This allows for the formation of a high-gain band with controllable width near the resonant frequency ω0, ensuring both high-gain suppression of the target harmonic frequency and adaptability to small fluctuations in the grid frequency. It resolves the contradiction between the theoretical assumptions of the ideal PR controller and the bandwidth limitations of the actual system (improving the traditional single-controller wideband suppression mode into a multi-controller narrowband precise suppression mode, avoiding mutual interference between different harmonic suppressions), and enhances the robustness of the harmonic suppression strategy.

[0096] Please see Figure 1 , Figure 1This paper presents the overall technical structure of a grid-based bidirectional low-harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method to achieve coordinated control. This technical solution effectively solves key technical challenges in power quality, system stability, and intelligent management in construction site power supply through an innovative multi-level collaborative control architecture. The entire solution comprises five core technical modules: full-band harmonic suppression technology using parallel multi-quasi-proportional resonant controllers, a grid-based control strategy based on virtual synchronous generators, an intelligent energy storage capacity optimization configuration method, a deep reinforcement learning energy management system, and a modular and scalable system architecture. These modules work collaboratively to form a complete technical solution. The following section combines... Figure 1 Some other embodiments of the present invention will be described.

[0097] In a further embodiment of the present invention, a parallel QPR controller group design is proposed. First, the transfer function design of a single QPR controller is as follows:

[0098] (1)

[0099] Where: K r ω is the resonant gain coefficient, which determines the magnitude of the gain at the resonant frequency; c The cutoff frequency controls the bandwidth of the resonant peak; This is the resonant angular frequency, i.e., the angular frequency of the target harmonic. ; s is the complex frequency variable in the Laplace transform.

[0100] To address the main harmonic components at the construction site, a parallel QPR controller group is designed, and its comprehensive transfer function is:

[0101] (2)

[0102] In the formula, G total (s) is the transfer function of the total controller; G fundamental (s) is the fundamental wave controller transfer function; G QPRn (s) is the QPR controller transfer function for the nth harmonic;

[0103] The optimized closed-loop transfer function of the grid-type bidirectional low-harmonic energy storage converter is as follows:

[0104] (3)

[0105] In the formula, H cl (s) is the closed-loop transfer function of the converter; G inv (s) is the inverter transfer function, G fb(s) represents the transfer function of the feedback loop. By precisely designing the parameters of each loop, this invention can achieve efficient suppression of multiple harmonics while ensuring system stability. Experimental results show that, under rated operating conditions, the total harmonic distortion (THD) of the output voltage can be controlled within 2%, and the content of each single harmonic is less than 0.8%, which is far superior to the 5% limit required by the IEEE 519-2014 standard.

[0106] In a further embodiment of the present invention, the parameters of each QPR controller are independently optimized according to the actual operating conditions. The parameter configuration of the parallel QPR controller group is as follows:

[0107] First harmonic QPR controller: Gain K r3 =12, cutoff frequency ω c3 =4 rad / s, resonant angular frequency =942.5 rad / s;

[0108] Second harmonic QPR controller: Gain K r5 =15, cutoff frequency ω c5 =5 rad / s, resonant angular frequency =1570.8 rad / s;

[0109] Third harmonic QPR controller: Gain K r7 =18, cutoff frequency ω c7 =6 rad / s, resonant angular frequency =2199.1 rad / s.

[0110] This parameter configuration can achieve a harmonic suppression gain of over 40dB in the frequency range of 250Hz±0.8Hz, effectively reducing the 5th harmonic content to below 0.5%.

[0111] In one embodiment of the present invention, the virtual synchronization control module is configured to simulate the electromechanical transient characteristics of a synchronous generator through the rotor motion equation, and to achieve autonomous adjustment of reactive power through virtual excitation control.

[0112] The core of VSG control lies in simulating the electromechanical transient characteristics of a synchronous generator, and its rotor motion equation is:

[0113] (4)

[0114] In the formula, J is the virtual moment of inertia, which determines the system's response speed to power disturbances; ω is the virtual rotor angular velocity; T m T represents mechanical torque, corresponding to active power command. e ω0 is the electromagnetic torque, reflecting the actual output power; D is the damping coefficient, providing frequency regulation capability; ω0 is the rated angular velocity.

[0115] Electromagnetic torque T e The calculation is based on the power balance relationship:

[0116] (5)

[0117] Where: e d e q i represents the internal potential components in the dq coordinate system; d i q Let be the current component in the dq coordinate system.

[0118] To achieve autonomous adjustment of reactive power, this embodiment incorporates a virtual excitation control circuit with the following control law:

[0119] (6)

[0120] In the formula, E is the amplitude of the virtual internal potential; E0 is the no-load potential; k q Q is the reactive power-voltage droop factor. ref V is the reactive power reference value; Q is the actual output reactive power; ref V is the voltage reference value; k is the actual output voltage; v This is the voltage-regulated integral gain.

[0121] The power systems at construction sites are often in a weak grid or isolated grid state, making it difficult for traditional grid-connected inverters to provide sufficient voltage and frequency support. This embodiment innovatively applies Virtual Synchronous Generator (VSG) technology to energy storage grid-connected power systems, enabling the inverter to have external characteristics similar to a synchronous generator, and to actively establish and maintain grid voltage and frequency.

[0122] In a further embodiment of the present invention, in order to improve the transient stability of the system, a virtual impedance control strategy is introduced, and the voltage equation in the dq coordinate system is:

[0123] (7)

[0124] In the formula, v d and v q These represent the d-axis and q-axis output voltage components of the converter in the dq rotating coordinate system, respectively; e d and e q These represent the d-axis and q-axis potential components of the virtual synchronous generator in the dq coordinate system, respectively; R v For virtual resistance; L v For virtual inductance; i d and i q These represent the d-axis output current component and the q-axis output current component of the converter in the dq rotating coordinate system, respectively.

[0125] The virtual impedance parameters are optimized based on system stability constraints. Small-signal stability analysis shows that the system exhibits good damping characteristics when the virtual inductance Lv satisfies the following conditions:

[0126] (8)

[0127] in:

[0128] V is the effective value of the system voltage; For steady-state work angle; These are steady-state active and reactive power, respectively.

[0129] By rationally designing the VSG control parameters, this embodiment enables the energy storage system to possess excellent grid-connection capabilities. Experimental verification shows that under the condition of a 50% load change, the system frequency deviation is controlled within ±0.2Hz, the voltage deviation does not exceed ±3%, and the recovery time is less than 200ms, fully meeting the power supply requirements of sensitive loads.

[0130] In one embodiment of the present invention, an energy storage capacity optimization configuration module is further included, which is configured to achieve the optimal configuration of energy storage capacity based on a multi-objective optimization model and an improved particle swarm optimization algorithm.

[0131] The objective function of the multi-objective optimization model is:

[0132] (9)

[0133] In the formula, F obj To optimize the comprehensive evaluation index for multiple objectives; C total The total cost over the entire system lifecycle includes initial investment costs and operation and maintenance costs; LPSP (Loss of Power Supply Probability) represents the probability of power supply loss; T response The system's dynamic response time is the key performance indicator; α1, α2, and α3 are weighting coefficients, satisfying the following conditions: ;

[0134] The formula for calculating the total cost of the system's entire lifecycle is:

[0135] (10)

[0136] In the formula, C initial This is the initial investment cost, and C initial =P rated ×c p +E rated ×c e P rated The rated power (kW) of the energy storage system; cp Cost per unit power (RMB / kW); E rated The rated capacity of the energy storage system (kWh); c e Unit capacity cost (RMB / kWh); N is the total lifespan (years); C om (t) represents the operating and maintenance cost in year t; r is the discount rate; C replacement Cost of replacing equipment (battery).

[0137] The probability of power loss (LPSP) is calculated based on the probability distribution of load demand and energy storage status:

[0138] (11)

[0139] Where: P def (t) represents the power deficit at time t; P load (t) represents the load demand at time t; T represents the total number of statistical periods.

[0140] To solve this multi-objective optimization problem, this embodiment employs an improved Particle Swarm Optimization (PSO) algorithm, introducing adaptive inertia weights and mutation operations to enhance the algorithm's global search capability and convergence speed. The particle position update formula is as follows:

[0141] (12)

[0142] Where: v i (k) represents the velocity of the i-th particle in the k-th iteration; x i (k) represents the position of the i-th particle in the k-th iteration (i.e., the energy storage capacity configuration scheme); w(k) = w max -(w max -w min )·k / K max For adaptive inertia weights; For learning factors; p is a random number in the interval [0,1]. besti The historical best position of the i-th particle; g best This is the globally optimal position.

[0143] The rational configuration of energy storage capacity is crucial to ensuring the economical and efficient operation of uninterruptible power supply systems at construction sites. This embodiment proposes an energy storage capacity configuration method based on multi-objective optimization, comprehensively considering multiple constraints such as system reliability, economy, and dynamic response performance. Through this optimization strategy, this embodiment can achieve optimal energy storage capacity configuration while meeting the uninterrupted power supply needs of construction sites. Taking a typical construction site (average daily load 50kW, peak load 80kW) as an example, the optimization results show that the optimal configuration of the energy storage system is 60kW power and 240kWh capacity. Compared with empirical configuration methods, this can reduce the initial investment cost by 23% while improving power supply reliability to over 99.95%.

[0144] In one embodiment of the present invention, a deep reinforcement learning energy management module is further included, which is configured to optimize the energy storage charging and discharging strategy based on a deep Q-network algorithm.

[0145] The optimization objective function for energy management is:

[0146] (13)

[0147] In the formula, J represents the optimization objective of energy management; T represents the total number of optimization periods; C grid (t) represents the grid electricity price at time t; P grid (t) represents the power drawn from the grid at time t; Δt is the duration of each scheduling period; λ is the weighting coefficient; L battery (t) represents the loss cost of the energy storage battery at time t.

[0148] The constraints include:

[0149] 1) Power balance constraint: P load (t)=P grid (t)+P bat (t)-P loss (t)

[0150] 2) Energy storage power constraint: -P bat,max ≤P bat (t)≤P bat,max

[0151] 3) Energy storage capacity constraints: SOC min ≤SOC(t)≤SOC max

[0152] 4) Power grid constraint: 0 ≤ P grid (t)≤P grid,max

[0153] Where: C grid (t) represents the grid electricity price at time t; P grid(t) represents the power purchased from the grid; P bat (t) represents the energy storage charging and discharging power (discharging is positive); P loss (t) represents the system power loss; λ is the battery life loss weighting coefficient; L battery (t) is the battery life loss function; SOC(t) is the state of charge at time t.

[0154] The battery life loss function uses the rainflow counting method and the equivalent cycle number model:

[0155] (14)

[0156] Where: n is the number of charge / discharge cycles; DOD i N represents the depth of discharge in the i-th cycle; life (DOD) = a·DOD (-b) This represents the relationship between cycle life and depth of discharge.

[0157] To achieve online optimization, this embodiment employs the Deep Q-Network (DQN) algorithm to model the energy management problem as a Markov decision process. The state space is defined as follows:

[0158] (15)

[0159] In the formula, S(t) is the state space at time t; SOC(t) is the state space at time t; P load (t) represents the state of charge of the battery at time t; t hour t represents the current hour; day This is the current date.

[0160] The action space is discretized as follows:

[0161] (16)

[0162] The reward function is designed as follows:

[0163] (17)

[0164] Where: P penalty (t) represents the penalty for violating the constraint.

[0165] The DQN network employs a 3-layer fully connected structure with 128, 64, and 32 neurons in the hidden layers, using ReLU as the activation function. Network parameter updates utilize empirical replay and target network techniques to improve training stability. The loss function during training is:

[0166] (18)

[0167] Where: θ represents the network parameters; γ represents the target network parameters; γ is the discount factor.

[0168] Construction site operations exhibit distinct time-varying characteristics, with significant variations in load demand across different periods. This embodiment develops an energy management system based on deep reinforcement learning, capable of achieving optimal charging and discharging strategies based on load forecasting, electricity price information, and energy storage status. Through a combination of offline training and online fine-tuning, this energy management system can adapt to the load characteristics of different construction sites. Practical applications demonstrate that, compared to traditional rule-based control strategies, intelligent optimization scheduling can reduce electricity costs by 15%–25% and extend battery life by over 20%.

[0169] Considering the diverse scale of construction sites, this invention also adopts a modular design concept, developing a flexibly expandable system architecture. The entire system consists of a Power Conversion Module (PCM), an Energy Storage Module (ESM), a Control Module (CM), and a Human-Machine Interface Module (HMI), with each module connected via standardized interfaces. In one embodiment of this invention, the power module includes at least one power conversion module, which adopts a three-level NPC topology, with a single module rated power of 30kW and an efficiency of 98.5%. Multiple PCMs can operate in parallel, reducing output ripple through carrier phase shifting technology. Parallel current sharing control employs an improved droop control strategy.

[0170] (19)

[0171] In the formula, P i and Q i These are the actual output active power and actual output reactive power of the i-th power conversion module, respectively. and These are the rated active power and rated reactive power of the i-th power conversion module, respectively; and These are the active power droop coefficient and reactive power droop coefficient of the i-th power conversion module, respectively; f i and V i These are the local output frequency and output voltage, respectively; f0 and V0 are the rated frequency and rated voltage, respectively.

[0172] To improve the power distribution accuracy of the parallel system, this embodiment introduces a secondary adjustment stage:

[0173] (20)

[0174] Where: Pavg Q avg k is the average power of all modules. pf k qv This is the integral gain.

[0175] In one embodiment of the present invention, the energy storage module uses lithium iron phosphate battery packs, with each battery pack having a capacity of 50kWh, and is equipped with an independent Battery Management System (BMS). Multiple battery packs can be connected in parallel to the DC bus via a DC / DC converter to achieve flexible capacity configuration. The DC / DC converter adopts an interleaved parallel Boost topology to reduce input current ripple, and its control strategy is as follows:

[0176] (twenty one)

[0177] in: P is the reference current of the DC / DC converter corresponding to the k-th energy storage module; bat,total For; V dc C is the DC bus voltage. k C represents the rated capacity of the k-th energy storage module. total k represents the total capacity of the entire energy storage system. soc Equalization coefficient based on SOC

[0178] System scalability is achieved through plug-and-play technology. When a new module is added, the control system automatically identifies and assigns an address, updating the system topology information. Communication uses a CAN bus, supporting the interconnection of up to 32 modules. Practical experience has shown that this modular architecture can cover a power range from 10kW to 1MW, meeting the diverse needs from small renovation sites to large construction sites.

[0179] To further enhance the level of intelligence, this invention integrates machine learning-based intelligent load identification technology, which can automatically identify the type of connected load and optimize the power supply strategy. Therefore, embodiments of this invention also provide an intelligent load identification method for a grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control, comprising the following steps:

[0180] S1: Collect multi-dimensional operating data of the load and extract the power characteristics, current characteristics, frequency domain characteristics and time domain characteristics of the multi-dimensional operating data.

[0181] It should be noted that power characteristics may include active power, reactive power, and apparent power; current characteristics may include RMS current, current waveform factor, and current peak factor; frequency domain characteristics may include harmonic content, spectral centroid, and spectral spread; and time domain characteristics may include starting current surge and load fluctuation characteristics.

[0182] S2: Input the power characteristics, current characteristics, frequency domain characteristics and time domain characteristics into the pre-built load classifier to identify the load type and obtain the load identification result.

[0183] It should be noted that the Support Vector Machine (SVM) algorithm is used to implement load classification. The decision function is:

[0184] (twenty two)

[0185] Where: K(x) i Let K(x) be the kernel function, using a radial basis function (RBF) kernel: K(x) i ,x)=exp(-γ||x i -x||²).

[0186] S3: Adjust power supply parameters based on load identification results.

[0187] For example, based on the load identification results, the power supply parameters are automatically adjusted as follows:

[0188] (1) Resistive load: constant voltage control mode is adopted, with voltage accuracy of ±1%;

[0189] (2) Inductive load: Enhance reactive power support and improve starting capability;

[0190] (3) Nonlinear load: Activate harmonic suppression function, THD < 2%;

[0191] (4) Sensitive load: Enable UPS mode, switching time <5ms.

[0192] Please see Figure 2 , Figure 2 This paper presents an algorithm flow for coordinated control of a network-type bidirectional low-harmonic energy storage converter based on virtual synchronous control and its intelligent load identification method. The flow presents a four-layer technical system: data acquisition, parallel processing, coordinated control, and mode execution. This architecture is initiated by the system initialization module, acquires operational status information through a multi-dimensional data acquisition module, performs specialized processing through five parallel algorithm modules, and finally achieves unified decision-making and mode selection at the system coordination control center, forming a complete closed-loop control system. The following section further describes the algorithm flow.

[0193] 1. Algorithm Flow Design:

[0194] The key technical features are reflected in its modular organization and parallel processing mechanism. The five core algorithm modules operate independently yet in coordination, avoiding the latency accumulation problems caused by traditional serial processing and significantly improving system response speed and processing efficiency. Each algorithm module employs a flow control method combining diamond-shaped decision nodes and rectangular processing nodes to ensure the logical consistency and reliability of algorithm execution.

[0195] 2. Technical Implementation Scheme of Data Acquisition Module

[0196] The multi-dimensional data acquisition module is located at the data source layer of the algorithm flow and is responsible for acquiring comprehensive information about the system operation in real time. This module uses four parallel acquisition units to complete voltage and current acquisition, power and frequency measurement, SOC temperature monitoring, and load characteristic analysis, forming a multi-dimensional data matrix.

[0197] The voltage and current acquisition unit employs high-precision AD conversion technology to achieve synchronous sampling of three-phase voltage and current at a sampling frequency of no less than 10kHz, ensuring complete capture of waveform details. The power and frequency measurement unit, based on the Fast Fourier Transform algorithm, realizes real-time calculation of active power, reactive power, and frequency, achieving a measurement accuracy of 0.1%. The SOC temperature monitoring unit communicates with the battery management system via a CAN bus to obtain the state of charge and temperature distribution information of each battery module. The load characteristic analysis unit uses a sliding window statistical method to extract the power characteristics, current waveform characteristics, and frequency domain characteristics of the load, providing a data foundation for subsequent intelligent identification.

[0198] 3. Cooperative processing mechanism of parallel algorithm modules:

[0199] The core technical feature of the algorithm flow lies in the parallel collaborative processing mechanism of five specialized algorithm modules. These five modules each undertake different technical functions, achieving efficient collaboration through time-division multiplexing and data sharing.

[0200] The intelligent load identification algorithm module employs a three-stage processing flow: feature extraction, classification, and type output. The feature extraction stage utilizes multidimensional signal analysis technology to extract load feature vectors from the time, frequency, and statistical domains. The SVM classification stage uses a radial basis function kernel to automatically identify the load type through a trained support vector machine model, achieving a classification accuracy of over 96%.

[0201] The multi-QPR harmonic suppression algorithm module achieves precise suppression of specific harmonics through a processing flow of harmonic detection and analysis, QPR parallel control, and harmonic suppression output. Harmonic detection and analysis employs a synchronous reference coordinate transformation method to quickly identify the amplitude and phase of each harmonic. QPR parallel control, through the coordinated operation of multiple quasi-proportional resonant controllers, independently suppresses the 3rd, 5th, and 7th harmonics, keeping the total harmonic distortion rate below 2%.

[0202] The VSG grid-building control algorithm module is based on the principle of a virtual synchronous generator. Through a processing sequence of rotor motion equation calculation, virtual excitation control, and grid-building signal output, it enables the energy storage system to establish a grid. The rotor motion equation calculation simulates the electromechanical transient characteristics of a synchronous generator, and the virtual excitation control achieves autonomous adjustment of reactive power, with a frequency stability accuracy of ±0.2Hz.

[0203] The energy storage capacity optimization algorithm module adopts an optimization process of establishing a multi-objective function, solving an improved PSO algorithm, and configuring the optimal capacity. It comprehensively considers economy, reliability, and response performance. By improving the particle swarm optimization algorithm to solve the optimal configuration scheme, the system investment cost is reduced by more than 25%.

[0204] The deep reinforcement learning energy management algorithm module achieves intelligent optimization of energy storage charging and discharging strategies through a learning process of state space definition, DQN network training, and optimal policy output. The state space includes multi-dimensional information such as SOC, load, electricity price, and time. The DQN network adopts a three-layer fully connected structure, and improves training stability through experience replay and target network techniques, achieving a response time of less than 100ms.

[0205] 4. Decision-making mechanism of the system coordination and control center:

[0206] The system coordination and control center, as the decision-making core of the algorithm flow, is responsible for integrating the output results of the five parallel algorithm modules and making unified decisions based on the system's operating status and external environmental conditions. This center adopts a multi-level decision-making architecture, first performing data fusion and consistency verification on the output information of each algorithm module, and then determining the operating mode based on preset decision rules and real-time constraints.

[0207] The operation mode determination node adopts a hierarchical decision tree structure, automatically selecting the most suitable operation mode based on multi-dimensional criteria such as grid connection status, load demand level, energy storage charge status, and system health status. The determination logic follows the principle of prioritizing safety while also considering economy, ensuring stable and reliable system operation under various conditions.

[0208] 5. Execution mechanism of multi-mode operation strategy:

[0209] The algorithm framework is designed with four basic operating modes to adapt to different application scenarios and operational requirements. Grid-connected operation mode is suitable for economically optimized operation under normal grid conditions, where the system acts as an active support unit for the grid, providing frequency regulation and voltage support services. Off-grid operation mode is suitable for independent power supply during grid faults or planned maintenance periods, where the system switches to grid-connected mode to provide uninterrupted power to critical loads.

[0210] The energy storage regulation mode is suitable for peak shaving and valley filling scenarios and renewable energy consumption scenarios. The system schedules charging and discharging based on electricity price signals and load forecasts to maximize economic benefits. The emergency backup mode is suitable for sudden failures or extreme operating conditions. The system activates protection mechanisms to ensure the normal operation of core functions and the safety of personnel and equipment.

[0211] Each operating mode is equipped with corresponding control parameters and execution strategies, and specific hardware actions are implemented through the execution module via control command output. The execution module adopts a real-time operating system architecture with a response time of less than 10ms, ensuring the timely and accurate execution of control commands.

[0212] 6. Closed-loop feedback control mechanism:

[0213] The algorithm framework implements closed-loop control through a feedback path represented by dashed lines, feeding back the results of control command execution to the multi-dimensional data acquisition module to form a complete control loop. This feedback mechanism can monitor the control effect in real time, promptly detect system deviations, and make dynamic adjustments to ensure that the system is always in an optimal operating state.

[0214] Feedback information includes key indicators such as actual output power, voltage frequency deviation, harmonic content changes, and load response. This information is filtered and standardized before being re-entered into the data acquisition module, providing an updated data foundation for the next round of algorithm processing and enabling the system to operate adaptively and optimally.

[0215] Through the systematic implementation of the above technical solutions, the algorithmic framework of this invention can effectively solve key technical problems such as power quality, system stability and intelligent management in power supply at construction sites, and provides a complete algorithmic solution for the engineering application of grid-type energy storage systems.

[0216] The technical solution of this invention, through the above detailed technical implementation steps, forms a complete technical solution, and its main innovations are reflected in:

[0217] For the first time, the parallel technology of multiple QPR controllers was applied to an energy storage converter, achieving precise suppression of specific subharmonics and solving the frequency coupling problem of traditional single controllers;

[0218] The innovative integration of VSG technology with energy storage systems enables them to have true grid-building capabilities, filling the technological gap in active support of energy storage systems in weak grid environments.

[0219] A full life-cycle cost optimization model considering battery life loss was established, realizing intelligent configuration of energy storage capacity and significantly improving system economy;

[0220] This is the first time that deep reinforcement learning technology has been applied to the energy management of construction site energy storage systems, achieving adaptive optimization scheduling and reducing operating costs by more than 20% compared to traditional methods.

[0221] A modular and scalable architecture was developed, supporting plug-and-play functionality and providing flexible solutions for construction sites of different sizes;

[0222] It integrates machine learning-based intelligent load identification technology, which enables automatic optimization of power supply strategies and significantly improves the system's intelligence level.

[0223] Through systematic innovation and synergistic optimization of the above-mentioned technical solutions, this invention successfully solves key technical problems in construction site power supply, such as poor power quality, insufficient system stability, high operating costs, and low level of intelligence, providing a revolutionary technical solution for uninterrupted power supply at construction sites. Actual engineering verification has shown that this invention's technical solution has reached a leading level in terms of technological advancement, economic rationality, and application reliability.

[0224] In addition, the present invention also proposes the following alternatives:

[0225] (1) Regarding harmonic suppression:

[0226] a) Adaptive Notch Filter Solution: Besides using a quasi-proportional resonant controller, an adaptive notch filter can also be used to achieve harmonic suppression. The transfer function of the notch filter is:

[0227] (twenty three)

[0228] Where: ω n Here, Q is the notch filter frequency, and Q is the quality factor. By identifying harmonic frequencies online and adaptively adjusting the notch filter parameters, effective suppression of time-varying harmonics can be achieved.

[0229] b) Repetitive Control Scheme: For periodic harmonic disturbances, a repetitive controller can be used, and its discrete-domain transfer function is:

[0230] (twenty four)

[0231] Where N is the number of sampling points in one fundamental period, and Q(z) is the low-pass filter. Repetitive control can achieve zero steady-state error tracking for all integer harmonics.

[0232] (2) Regarding network control:

[0233] a) Sag control scheme: As a simplified alternative to VSG, traditional droop control can be used to achieve the network construction function:

[0234] (25)

[0235] Although it lacks inertia support, it is simple to implement and suitable for small-capacity systems.

[0236] b) Virtual Oscillator Control (VOC): Based on the principle of nonlinear oscillators, self-synchronization is achieved through the following equations:

[0237] (26)

[0238] The VOC scheme does not require a phase-locked loop and has better transient synchronization performance.

[0239] (3) Regarding energy management:

[0240] a) Model Predictive Control (MPC): Establishes a state-space model of the system and achieves optimal control by solving a finite-time optimization problem.

[0241] (27)

[0242] MPC can explicitly handle constraints and has higher prediction accuracy than reinforcement learning methods.

[0243] b) Fuzzy logic control: Establishing a fuzzy rule base based on expert experience, such as:

[0244] -IF SOC is Low AND Price is Low THEN Charge Fast;

[0245] -IF SOC is High AND Load is High THEN Discharge.

[0246] Fuzzy control is robust and does not depend on an accurate model.

[0247] (4) Regarding energy storage media:

[0248] a) Supercapacitor solution: For power applications, supercapacitors can be used to replace or supplement lithium batteries. Supercapacitors have high power density (>10kW / kg) and long cycle life (>1 million cycles), making them particularly suitable for frequent charge and discharge scenarios.

[0249] b) Flywheel energy storage solution: This solution uses a high-speed rotating flywheel to store kinetic energy, which is then converted into energy by a motor. Flywheel energy storage offers fast response times (milliseconds) and high efficiency (>95%), making it suitable for providing short-term high-power support.

[0250] c) Hydrogen fuel cell solution: For long-term energy storage needs, hydrogen fuel cell systems can be used. Hydrogen is produced by electrolyzing water to store energy, which is then used to generate electricity through fuel cells. This method has high energy density and is suitable for seasonal energy storage.

[0251] (5) Regarding topology:

[0252] a) Modular Multilevel Converter (MMC): Employing cascaded H-bridge or half-bridge sub-modules, it can achieve higher voltage levels and lower harmonic content.

[0253] b) Z-source inverter: Through a special impedance network, it realizes the step-up and step-down functions, improves DC voltage utilization, and simplifies the system structure.

[0254] (6) Regarding the communication architecture:

[0255] a) Wireless communication solution: Use ZigBee, LoRa or 5G technology to achieve communication between modules, avoid wiring difficulties and improve system flexibility.

[0256] b) Power Line Communication (PLC): Utilizes power lines to transmit control signals, eliminating the need for additional communication cables and reducing system costs.

[0257] The most suitable alternative can be selected based on the specific application scenario and technical requirements, or multiple solutions can be combined to further improve system performance.

[0258] The feasibility and superiority of this invention have been verified through simulation analysis. The specific verification process and results are as follows:

[0259] 1. Simulation verification stage:

[0260] A complete system simulation model was built using MATLAB, including the energy storage unit, power converter, control system, and load model. Simulation parameter settings: system capacity 100kW / 400kWh, DC bus voltage 700V, switching frequency 10kHz, output voltage 380V / 50Hz.

[0261] Main simulation results:

[0262] - During steady-state operation, the output voltage THD is less than 1.5%, and the efficiency reaches 97.8%.

[0263] - When the load increases by 50%, the maximum frequency deviation is 0.18Hz and the recovery time is 165ms;

[0264] - After a harmonic load (THD=25%) is connected, the output voltage THD remains below 2.1%;

[0265] - In the parallel expansion test, four 30kW modules were connected in parallel, and the power imbalance was less than 2%.

[0266] 2. Detailed analysis of the simulation verification results of the technical solution:

[0267] The technical solution of this invention has been fully demonstrated through three core simulation verifications, proving the significant advantages of the proposed grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control in key technical fields such as harmonic suppression, dynamic response, and capacity optimization. The simulation results fully verify the innovation, effectiveness, and practicality of the technical solution.

[0268] Simulation Verification 1: Simulation results of harmonic suppression performance of multi-QPR controller are as follows Figure 3 As shown.

[0269] Verification of Technological Innovations: The parallel multi-quasi-proportional resonant controller technology demonstrated excellent harmonic suppression performance in this simulation. Simulation results show that under typical nonlinear load conditions, the total harmonic distortion (THD) of the system output current significantly decreased from 38.91% before suppression to 9.19% after suppression, with a THD improvement of 29.73% and a harmonic suppression effect as high as 76.4%.

[0270] Frequency domain performance analysis: Spectrum analysis results show that the designed QPR controller achieves precise suppression of the 3rd, 5th, and 7th harmonic components. While maintaining the fundamental component's stability, the amplitudes of each harmonic are significantly attenuated, especially for the lower harmonics, which are of most concern in engineering applications. This result verifies the technical advantages of the parallel architecture of multiple QPR controllers in full-band harmonic suppression.

[0271] Engineering Application Value: Although the current THD level is 9.19%, slightly higher than the ideal target value of 5%, it is significantly better than the national standard requirement of 10%, and represents a qualitative breakthrough compared to traditional control methods. This performance level fully meets the power supply quality requirements of sensitive equipment on construction sites, providing reliable power quality assurance for uninterruptible power supply systems.

[0272] Simulation Verification 2: Verification Results of Dynamic Response of VSG Network Control are as follows Figure 4 As shown.

[0273] Grid connection capability verification: The virtual synchronous generator control strategy demonstrated excellent grid connection performance in the load step test. Under the condition of 100% load change from 40kW to 80kW, the system frequency deviation was controlled within ±0.000Hz, and the voltage deviation was also maintained within a very small range of ±0.00%, which fully proves the effectiveness of the VSG control algorithm.

[0274] Dynamic response characteristics: Simulation results show that the system can respond quickly to sudden load changes, and the power balance characteristic curve indicates that the energy storage converter can rapidly track load changes and achieve smooth power transition. The virtual excitation control maintains stable operation near the rated value, demonstrating the robustness and stability of the control system.

[0275] Technological advancements are evident in the fact that, compared to traditional grid-connected inverters, the grid-connected control strategy proposed in this invention achieves true "active support" capability. The system does not rely on an external grid reference and can independently maintain voltage and frequency stability, providing crucial technical support for microgrids and off-grid applications. This characteristic has significant engineering value in complex grid environments such as construction sites.

[0276] Simulation Verification 3: Energy Storage Capacity Optimization Configuration (Improved PSO Algorithm) Simulation verification results are as follows: Figure 5 As shown.

[0277] Performance verification of the optimization algorithm: The improved particle swarm optimization algorithm exhibits good convergence performance in energy storage capacity configuration. The optimization process shows that the objective function value converges rapidly from the initial 0.846 to the optimal value of 0.656. The algorithm basically reaches a stable state after 20 iterations, proving the efficiency and stability of the improved PSO algorithm.

[0278] Capacity configuration optimization effect: Optimization results show that, for a typical construction site load curve (average daily load 47.1kW, peak load 80.0kW), the optimal configuration determined by the system is an 80.0kW power unit paired with a 50.0kWh energy storage capacity, with a capacity ratio of 0.62 hours. Compared with the traditional empirical configuration scheme (96.0kW / 384.0kWh), the optimized scheme achieves a cost saving of 78.7%, fully demonstrating the economic value of intelligent optimization algorithms.

[0279] Engineering practicality analysis: The optimized configuration scheme significantly reduces system investment costs while meeting load demands. By precisely matching load characteristics with energy storage capacity, it avoids the over-configuration problem in traditional designs, providing an important technical foundation for the commercialization of energy storage systems.

[0280] The comprehensive evaluation of the innovativeness and effectiveness of the technical solution is explained as follows:

[0281] Innovative Technology Integration: This patented technology successfully integrates three core technologies—multi-QPR harmonic suppression, VSG network control, and intelligent capacity optimization—forming a complete technical solution. The various technical modules work collaboratively to achieve high system reliability and economy while ensuring power quality.

[0282] Performance indicators met: Simulation results show that the system meets or exceeds design requirements in key indicators such as power quality, dynamic response, and economy. THD is controlled below 10%, frequency and voltage stability meet the requirements of sensitive loads, and cost optimization is significant, verifying the comprehensiveness and advancement of the technical solution.

[0283] Engineering Application Prospects: The technical solution addresses the specific application scenario of uninterrupted power supply at construction sites, solving the technical challenges of traditional energy storage systems in terms of power quality, grid construction capabilities, and capacity configuration. Simulation results fully demonstrate the feasibility and superiority of the technical solution in engineering applications, laying a solid foundation for promoting the industrial application of energy storage technology in the construction field.

[0284] In summary, through simulation verification of the three core technologies, the patented technical solution of this invention has achieved significant breakthroughs in key technical areas such as harmonic suppression, network control, and capacity optimization. The simulation results not only verify the effectiveness of each technology but, more importantly, demonstrate the overall innovation and engineering practical value of the technical solution. This technical solution provides a complete technical solution for constructing high-quality, highly reliable uninterruptible power supply systems for construction sites, possessing significant academic value and broad industrialization prospects.

[0285] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0286] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0287] In the embodiments disclosed in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0288] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control, characterized in that, include: A power module and a control module; the control module is signal-connected to the power module; The power module is configured to achieve bidirectional power conversion and power output; The control module includes a virtual synchronization control module and a harmonic suppression control module. The virtual synchronous control module is configured to use virtual synchronous generator technology to realize bidirectional regulation of active and reactive power during the bidirectional power conversion process of the power module, as well as grid-level voltage and frequency support during the power output process. The harmonic suppression module is equipped with a parallel QPR controller group; the parallel QPR controller group includes at least a first QPR controller, a second QPR controller and a third QPR controller; The harmonic suppression module is configured to suppress characteristic harmonics on the power output side of the power module through the parallel QPR controller group; wherein the first QPR controller, the second QPR controller and the third QPR controller process the 3rd, 5th and 7th characteristic harmonics, respectively.

2. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, The transfer function of the parallel QPR controller group is: In the formula, G total (s) is the transfer function of the total controller; G fundamental (s) is the fundamental wave controller transfer function; G QPRn (s) is the QPR controller transfer function for the nth harmonic; The closed-loop transfer function of the grid-type bidirectional low-harmonic energy storage converter is: In the formula, H cl (s) is the closed-loop transfer function of the converter; G inv (s) is the inverter transfer function, G fb (s) is the transfer function of the feedback loop.

3. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 2, characterized in that, The parameters of the parallel QPR controller group are configured as follows: First harmonic QPR controller: Gain K r3 =12, cutoff frequency ω c3 =4 rad / s, resonant angular frequency =942.5 rad / s; Second harmonic QPR controller: Gain K r5 =15, cutoff frequency ω c5 =5 rad / s, resonant angular frequency =1570.8 rad / s; Third harmonic QPR controller: Gain K r7 =18, cutoff frequency ω c7 =6 rad / s, resonant angular frequency =2199.1 rad / s.

4. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, The virtual synchronization control module is configured to simulate the electromechanical transient characteristics of a synchronous generator through rotor motion equations and to achieve autonomous reactive power adjustment through virtual excitation control. The rotor motion equation is: In the formula, J is the virtual moment of inertia; ω is the virtual rotor angular velocity; T m T represents mechanical torque. e ω is the electromagnetic torque; D is the damping coefficient; ω0 is the rated angular velocity; The control law for the virtual excitation control is: In the formula, E is the amplitude of the virtual internal potential; E0 is the no-load potential; k q Q is the reactive power-voltage droop factor. ref V is the reactive power reference value; Q is the actual output reactive power; ref V is the voltage reference value; k is the actual output voltage; v This is the voltage-regulated integral gain.

5. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 4, characterized in that, The virtual synchronization control module also includes a virtual impedance control unit, the voltage equation of which is given by the virtual impedance control unit in the dq coordinate system as follows: In the formula, v d and v q These represent the d-axis and q-axis output voltage components of the converter in the dq rotating coordinate system, respectively; e d and e q These represent the d-axis and q-axis potential components of the virtual synchronous generator in the dq coordinate system, respectively; R v For virtual resistance; L v For virtual inductance; i d and i q These represent the d-axis output current component and the q-axis output current component of the converter in the dq rotating coordinate system, respectively.

6. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, It also includes an energy storage capacity optimization configuration module, which is configured to achieve the optimal configuration of energy storage capacity based on a multi-objective optimization model and an improved particle swarm optimization algorithm; The objective function of the multi-objective optimization model is: In the formula, F obj To optimize the comprehensive evaluation index for multiple objectives; C total The total cost over the entire system lifecycle; LPSP represents the probability of power loss; T response The system's dynamic response time is the indicator; α1, α2, and α3 are weighting coefficients. The formula for calculating the total lifecycle cost of the system is as follows: In the formula, C initial N represents the initial investment cost; C represents the total lifespan (in years); om (t) represents the operating and maintenance cost in year t; r is the discount rate; C replacement Cost of equipment replacement.

7. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, It also includes a deep reinforcement learning energy management module, which is configured to optimize energy storage charging and discharging strategies based on a deep Q-network algorithm. The optimization objective function for energy management is: In the formula, J represents the optimization objective of energy management; T represents the total number of optimization periods; C grid (t) represents the grid electricity price at time t; P grid (t) represents the power drawn from the grid at time t; Δt is the duration of each scheduling period; λ is the weighting coefficient; L battery (t) represents the loss cost of the energy storage battery at time t; The optimization objective function is transformed into a Markov decision process based on the deep Q-network algorithm. The state space of the deep Q-network algorithm is as follows: In the formula, S(t) is the state space at time t; SOC(t) is the state space at time t; P load (t) represents the state of charge of the battery at time t; t hour t represents the current hour; day This is the current date.

8. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, The power module includes at least one power conversion module, which adopts a three-level NPC topology. When multiple power conversion modules are connected in parallel, a current sharing control strategy is employed. The control equation for the current sharing control strategy is: In the formula, P i and Q i These are the actual output active power and actual output reactive power of the i-th power conversion module, respectively. and These are the rated active power and rated reactive power of the i-th power conversion module, respectively; and These are the active power droop coefficient and reactive power droop coefficient of the i-th power conversion module, respectively; f i and V i These are the local output frequency and output voltage, respectively; f0 and V0 are the rated frequency and rated voltage, respectively.

9. The grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control according to claim 1, characterized in that, The power module also includes an energy storage module, which uses a lithium iron phosphate battery pack. The DC / DC converter control strategy for the energy storage module is as follows: In the formula, P is the reference current of the DC / DC converter corresponding to the k-th energy storage module; bat,total For; V dc C is the DC bus voltage. k C represents the rated capacity of the k-th energy storage module. total k represents the total capacity of the entire energy storage system. soc This is the equilibrium coefficient based on SOC.

10. A smart load identification method for a grid-type bidirectional low-harmonic energy storage converter based on virtual synchronous control, characterized in that, Includes the following steps: Collect multidimensional operating data of the load, and extract the power characteristics, current characteristics, frequency domain characteristics, and time domain characteristics of the multidimensional operating data; The power characteristics, current characteristics, frequency domain characteristics, and time domain characteristics are input into a pre-built load classifier to identify the load type and obtain the load identification result. Adjust the power supply parameters based on the load identification results.

Images