Power distribution system power quality improvement method based on X-LMS adaptive algorithm

Abstract

The invention relates to a power distribution system electric energy quality improving method based on an X-LMS self-adaptive algorithm. The method comprises the following specific steps that the three-phase voltage and the load current of a grid-connected point of a power distribution network and the direct-current side voltage of a distributed static synchronous compensator are collected in real time; based on the three-phase voltage, generating a phase unit voltage of each phase and an orthogonal unit voltage orthogonal to the phase unit voltage; an X-LMS algorithm is adopted to extract a fundamental wave active weight and a fundamental wave reactive weight of a load current, wherein the X-LMS algorithm realizes rapid convergence and high steady-state precision of the weights by introducing a crossover error product term; an active compensation component and a reactive adjustment component are calculated through a direct-current voltage control loop and an alternating-current voltage control loop respectively, the direct-current voltage control loop generates an active compensation weight through a PI controller based on a direct-current side voltage error, and the alternating-current voltage control loop generates a reactive adjustment weight through the PI controller based on an alternating-current voltage amplitude error.

Description

Technical Field

[0001] This invention relates to the field of power system automation technology, specifically to a method for improving power quality in distribution systems based on the X-LMS adaptive algorithm. Background Technology

[0002] As the penetration rate of nonlinear loads in distribution networks continues to increase, the power quality problems they cause pose a serious challenge to the power grid. Existing solutions often improve power quality by introducing passive or active devices in series or parallel with the power grid.

[0003] Passive devices typically refer to passive filters and parallel capacitor banks. Passive filters are composed of passive components such as resistors, inductors, and capacitors connected in series and parallel. By adjusting relevant parameters, they can absorb harmonics of specific frequencies. Parallel capacitor banks are devices used in power systems for reactive power compensation, absorbing inductive reactive power from the grid, improving the power factor of the circuit, and enhancing power quality. Both are simple in structure, low in cost, highly reliable, and easy to maintain. However, both passive filters and parallel capacitor banks are prone to resonance with the grid, which can adversely affect the grid. Passive filters can only compensate for harmonics of specific orders, and parallel capacitor banks are difficult to use in situations with large load fluctuations.

[0004] Common active devices include APFs and distributed static synchronizing compensators (PSCs). APFs are primarily used to improve circuit power factor and mitigate harmonics, and their function is relatively simple. Distributed static synchronizing compensators can improve power quality by controlling the converter to output compensation current. Existing research on distributed static synchronizing compensators often employs both hardware and software methods to enhance power quality.

[0005] On the hardware side, improvements to the distributed static synchronous compensator (SSC) circuit topology, such as increasing the DC-side capacitor and adding filters at the grid connection point, are often made to suppress harmonics and improve the power factor. On the software side, appropriate control algorithms are designed to improve power quality. Existing research has proposed various neural network algorithms for LMS, such as the KHLMS algorithm, the greedy LMS algorithm, and the sparse LMS algorithm. These control algorithms enable the SSC to inject optimal reactive power at the grid connection point, and can also reduce the requirements for the SSC's rated capacity. However, existing LMS control methods still have limitations such as insufficient convergence speed and poor performance. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention proposes a method for improving power quality in power distribution systems based on the X-LMS adaptive algorithm.

[0007] The technical solution of the present invention is as follows: On the one hand, this invention proposes a method for improving power quality in distribution systems based on the X-LMS adaptive algorithm, the specific steps of which are as follows: Real-time acquisition of three-phase voltage, load current, and DC-side voltage of distributed static synchronous compensators at the grid connection point of the distribution network; based on the three-phase voltage, generation of phase unit voltage and orthogonal unit voltage for each phase; The X-LMS algorithm is used to extract the fundamental active and reactive weights of the load current. The X-LMS algorithm achieves fast convergence and high steady-state accuracy of the weights by introducing a cross error product term. The active power compensation component and reactive power regulation component are calculated by DC voltage control loop and AC voltage control loop respectively. The DC voltage control loop generates active power compensation weight based on DC voltage error through PI controller, and the AC voltage control loop generates reactive power regulation weight based on AC voltage amplitude error through PI controller. The total active power weight is obtained by superimposing the fundamental active power weight and the active power compensation weight, and the total reactive power weight is obtained by superimposing the fundamental reactive power weight and the reactive power regulation weight. The reference compensation current of each phase is generated by combining the phase unit voltage of each phase and the orthogonal unit voltage thereto. The switching transistors of the distributed static synchronous compensator are controlled by a hysteresis current controller to output the compensation current.

[0008] In a preferred embodiment, the step of generating the phase unit voltage of each phase and the orthogonal unit voltage based on the three-phase voltage specifically includes: The three-phase voltages are normalized to obtain the phase unit voltages that are in phase with the grid voltage. By performing Clark transformation and 90-degree phase shift on the phase unit voltage, and then inversely transforming it to the abc coordinate system, an orthogonal unit voltage is generated.

[0009] In a preferred embodiment, when the X-LMS algorithm extracts the fundamental active power weight and fundamental reactive power weight of the load current, the weight update formula is as follows:

[0010] In the formula, These are the fundamental active power weights for phases A, B, and C, respectively. These are the fundamental reactive power weights for phases A, B, and C, respectively. , , These are the prior active power errors for phases A, B, and C, respectively. , , These are the phase unit voltages for phases A, B, and C, respectively. Step size factor; , , These are the load currents for phases A, B, and C, respectively. , , These represent the active power auxiliary errors for phases A, B, and C, respectively; n is the sampling time index for discrete time. , , These are the orthogonal unit voltages of phases A, B, and C, respectively. , , These are the prior reactive power errors for phases A, B, and C, respectively. , , These are the reactive power auxiliary errors for phases A, B, and C, respectively.

[0011] In a preferred embodiment, the specific calculation process for the step of generating active power compensation weights based on the DC-side voltage error using a PI controller in the DC voltage control loop is as follows:

[0012] In the formula, v dce For DC voltage error signal, v dc(ref) The DC-side reference voltage, v dc This is the actual voltage on the DC side.

[0013] As a preferred embodiment, the specific calculation process for the step of generating reactive power adjustment weights based on AC voltage amplitude error using a PI controller in the AC voltage control loop is as follows:

[0014] In the formula, v te The error signal for the AC voltage amplitude, v t(ref) For AC reference voltage amplitude, v t This represents the actual AC voltage amplitude.

[0015] In a preferred embodiment, the step of generating the reference compensation current for each phase by combining the phase unit voltage of each phase with the orthogonal unit voltage therein specifically includes: Calculate the average value of the three-phase fundamental active power weight and the average value of the fundamental reactive power weight; The total active power weight is obtained by superimposing the average value of the fundamental active power weight and the active power compensation weight. The total reactive power weight is obtained by superimposing the average value of the fundamental reactive power weight and the reactive power adjustment weight. The reference compensation current for each phase is obtained by adding the product of the total active power weight and the phase unit voltage, and the product of the total reactive power weight and the orthogonal unit voltage.

[0016] In a preferred embodiment, the distributed static synchronous compensator is a six-bridge-arm three-phase two-level converter based on IGBT, with a large capacitor connected in parallel on the DC side to maintain voltage stability, eliminating the need for an external DC power supply.

[0017] In a preferred embodiment, the hysteresis current controller compares the reference compensation current with the actual compensation current to limit the error within a preset hysteresis band and directly generates a PWM signal to control the switching transistor.

[0018] On the other hand, the present invention proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a power quality improvement method for a power distribution system based on the X-LMS adaptive algorithm as described in any embodiment of the present invention.

[0019] On the other hand, the present invention proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a power quality improvement method for a power distribution system based on the X-LMS adaptive algorithm as described in any embodiment of the present invention.

[0020] The present invention has the following beneficial effects: 1. This invention introduces a crossover error term, resulting in faster dynamic response and convergence speed. Traditional LMS algorithms rely on a single error signal to update weights, leading to slow convergence and response. This invention innovatively introduces a crossover error term, enabling faster weight adjustment during load surges and minimizing the adverse effects of load switching on the power system.

[0021] 2. This invention utilizes an adaptive neural network to extract the active and reactive components of the fundamental current, eliminating the need for additional filters and exhibiting strong anti-interference capabilities. Traditional control strategies require low-pass filters to extract the fundamental current component, inevitably introducing amplitude and phase system errors. This invention, through the adaptive learning capability of neural networks, directly separates the fundamental active and reactive components from the load current. This algorithm eliminates the need for additional filters, avoiding the delay and error problems introduced by filters, and also improving robustness.

[0022] 3. This invention is computationally efficient and easy to implement digitally. Existing methods, such as dq coordinate transformation, require complex matrix operations. Although this invention introduces cross terms, the overall algorithm structure still maintains the simplicity of the LMS algorithm, making it suitable for real-time operation in embedded controllers such as DSPs, and enabling high-performance control with a relatively low sampling rate. Attached Figure Description

[0023] Figure 1 This is a topology diagram of a distributed static synchronization compensator. Figure 2The control block diagram for generating switching signals using the X-LMS algorithm. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0026] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0027] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0028] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0029] Example 1: See Figure 1 A method for improving power quality in a distribution system based on the X-LMS adaptive algorithm, the specific steps of which are as follows: Real-time acquisition of three-phase voltage, load current, and DC-side voltage of distributed static synchronous compensators at the grid connection point of the distribution network; based on the three-phase voltage, generation of phase unit voltage and orthogonal unit voltage for each phase; In this embodiment, there are two main types of distributed static synchronization compensators: series type and parallel type. The topology used in this invention is a parallel type distributed static synchronization compensator. Figure 1The topology of a distributed static synchronous compensator (SSC) is illustrated. This invention employs a six-arm, three-phase, two-level topology based on IGBTs. By controlling the switching states of the IGBTs, the constant DC voltage is converted into a pulsed AC voltage, thereby generating the required compensation current. A large capacitor is connected in parallel on the DC side, and its voltage is maintained stable through a specific control algorithm, eliminating the need for an external DC power supply. As an energy storage element, it enables the exchange of active and reactive energy at the grid connection point. A filter inductor is connected at the grid connection point to filter out high-frequency harmonic currents generated by IGBT switching.

[0030] The X-LMS algorithm is used to extract the fundamental active and reactive weights of the load current. The X-LMS algorithm achieves fast convergence and high steady-state accuracy of the weights by introducing a cross error product term. The active power compensation component and reactive power regulation component are calculated by DC voltage control loop and AC voltage control loop respectively. The DC voltage control loop generates active power compensation weight based on DC voltage error through PI controller, and the AC voltage control loop generates reactive power regulation weight based on AC voltage amplitude error through PI controller. The total active power weight is obtained by superimposing the fundamental active power weight and the active power compensation weight, and the total reactive power weight is obtained by superimposing the fundamental reactive power weight and the reactive power regulation weight. The reference compensation current of each phase is generated by combining the phase unit voltage of each phase and the orthogonal unit voltage thereto. The switching transistors of the distributed static synchronous compensator are controlled by a hysteresis current controller to output the compensation current.

[0031] In this embodiment, Figure 2This paper presents a control block diagram for generating switching signals using the X-LMS algorithm. The entire control system can be divided into three main parts: signal acquisition, reference current calculation, and PWM signal generation. The signal acquisition stage primarily acquires the three-phase grid voltage, load current, and DC-side capacitor voltage of the distributed static synchronous compensator (DSC) as the main processing objects for the algorithm, generating phase unit voltages and quadrature unit voltages. Reference current calculation is the core of the control system. The algorithm iteratively adjusts weights to track and extract the fundamental active and reactive power weights of the load current in real time. Simultaneously, it compares the DC reference voltage with the actual DC voltage and sends the error to the PI controller to obtain the active power loss compensation weight, which compensates for the switching and resistive losses of the DSC, stabilizing VDC at the reference value. It also compares the AC reference voltage amplitude with the actual AC voltage amplitude and sends the error to the PI controller to obtain the reactive power regulation weight, injecting reactive current to stabilize the grid connection point voltage during load fluctuations. The active power weight and reactive power weight are then superimposed to obtain the total active power weight and total reactive power weight, which are used to synthesize the reference current. The PWM signal generation module compares the reference current with the actual current and sends the error to the hysteresis comparator to generate the trigger signal for each phase IGBT, producing a compensation current and achieving comprehensive management of the power grid's power quality.

[0032] Based on the above fundamental principles, the specific application of the improvements to the LMS algorithm in this scheme will now be explained in detail: The traditional LMS algorithm requires extracting the fundamental active and reactive components of the load voltage to generate the phase unit voltage (hereinafter referred to as phase unit voltage) and the unit voltage orthogonal to the phase unit voltage (hereinafter referred to as orthogonal unit voltage). The phase unit voltage is obtained by normalizing the three-phase voltages a, b, and c, and is in phase with the grid voltage; the orthogonal unit voltage has the same amplitude as the phase unit voltage, and each phase lags the phase unit voltage by 90°.

[0033] Assuming the grid voltage is a symmetrical three-phase sinusoidal wave, v a v b v c The voltages of phases a, b, and c of the power grid are represented as follows:

[0034] Among them, V m Let be the voltage amplitude. Then the sum of the squares of the three-phase voltages is:

[0035] Let the normalization parameter v t for:

[0036] Therefore, the phase unit voltage u pa upb u pc The normalized formula is as follows:

[0037] To generate an orthogonal unit voltage, a Clark transformation is performed on the phase unit voltage to convert the three-phase u pa u pb u pc Projected onto the stationary αβ coordinate system:

[0038] will u pα u pβ Multiplying the column vectors by the rotation matrix on the left yields a new vector that is 90° behind the whole. u qα u qβ :

[0039] Finally, u qα u qβ By inversely transforming to the abc coordinate system, the required orthogonal unit voltage u is obtained. qa u qb u qc :

[0040] Simplifying the above formulas, we have:

[0041] The goal of the LMS algorithm is to minimize the following mean square error to make the actual current close to the reference current (taking the fundamental active power weight of phase A as an example):

[0042] Where J(n) is the cost function, e(n) is the error between the reference current and the actual current of phase a, and ω pa (n) represents the fundamental active power weight of phase a. The core of the LMS algorithm lies in adjusting the weights using gradient descent. The cost function is calculated with respect to the weights ω. pa The gradient is:

[0043] The updated weights are obtained by subtracting the product of the step size and the gradient from the weights:

[0044] Where μ is the step size factor, controlling the convergence speed and stability. Similarly, the fundamental active power weights of phases B and C can be obtained, and the fundamental reactive power weights of the three phases can also be obtained by the above method, which can be simplified to:

[0045] The LMS algorithm, with its simple computation and ease of implementation, has been widely used in engineering practice. Its core advantages lie in its low computational complexity, real-time online processing, and easy convergence stability. However, it has some limitations. The LMS algorithm has a relatively slow convergence speed and poor dynamic response performance, especially in scenarios with high correlation of input signals or poor conditions.

[0046] To address the aforementioned shortcomings and solve the convergence speed problem of the traditional LMS algorithm, this invention selects relevant signals and employs a multiplication mechanism to achieve an adaptive variable step size. The principle of the X-LMS algorithm is illustrated using the fundamental active power weight of phase A as an example. Two error quantities, e, are now introduced. pa and E pa ,have:

[0047] Among them, i la e is the load current; pa Similar to the definition in the traditional LMS algorithm, i represents the prior estimation error, which is the difference between the actual load current and the active current calculated by weighting at the previous moment; lpA E represents the average active current of the load. pa To assist in error calculation, the degree of deviation between the load active current and the average active current was measured. lpa Let be the reference active current of the load, which can be expressed by the following formula:

[0048] The iterative formula for the fundamental active power weight of the X-LMS algorithm is given below:

[0049] In the traditional LMS algorithm, the weight change Δω ∝ e(n), meaning the weight change is linearly related to the error. In the X-LMS algorithm, the weight change Δω ∝ e(n)·E(n)·i l (n). The introduced multiplier i lThis algorithm allows the system to increase the step size of weight updates under heavy load for rapid compensation, and automatically decrease the step size under light load, avoiding overshoot and oscillation. When the system is stable, the auxiliary error and prior estimation error are very small, and their product is also minimal. The cross term e(n)·E(n) keeps the step size small, resulting in minimal jitter in the reference signal and reducing THD. When the load changes abruptly, the product of the auxiliary error and prior estimation error becomes large, allowing for a larger update step size and a fast system response. In summary, this algorithm achieves an adaptive variable step size effect. If the input signal contains noise, the error term in the traditional LMS algorithm will also include this noise, causing fluctuations in the weights. The XLMS algorithm performs correlation detection by calculating the product of e(n) and E(n). Only signal components that appear in both e(n) and E(n), i.e., the fundamental signal components, can drive weight updates. Therefore, this algorithm can function normally without an additional low-pass filter.

[0050] Similarly, through the above analysis, we can obtain the fundamental active power weights of phases B and C and the fundamental reactive power weights of the three phases, which can be summarized as follows:

[0051] The expressions for errors e and E are as follows:

[0052] Three-phase fundamental active power weight ω p and the three-phase fundamental reactive power weight ω q The average values ​​are as follows:

[0053] The core of a distributed static synchronous compensator (DSC) is a voltage source converter. Without an external DC power supply, its normal operation depends on the stability of the DC-side capacitor voltage. The grid voltage is converted into DC voltage through the switching on and off of the switching transistors. To maintain DC voltage stability, the DSC must absorb a certain amount of active power from the grid to compensate for switching losses. The formula for calculating the DC voltage error is:

[0054] Among them, v dce For DC voltage error signal, v dc(ref) The DC-side reference voltage, v dc This is the actual DC-side voltage. The error is eliminated to zero using a PI controller. The output of the PI controller is the active current weight that the distributed static synchronous compensator needs to absorb from the grid.

[0055] Where, ω cp For the active compensation component, k pa kia For PI parameters.

[0056] In a power distribution network, voltage fluctuations are primarily related to the flow of reactive power. The voltage at the grid connection point is easily affected by load fluctuations. When load changes cause voltage rises or falls, distributed static synchronizing compensators (SSCs) can absorb or inject reactive power to stabilize the voltage. The formula for calculating the AC voltage error signal is:

[0057] Among them, v te The error signal for the AC voltage amplitude, v t(ref) For AC reference voltage amplitude, v t This represents the actual AC voltage amplitude. The output of the PI controller is the reactive current weight that the distributed static synchronous compensator needs to inject or absorb.

[0058] Where, ω cq For reactive power regulation component, k pr k ir For PI parameters.

[0059] Total active weight ω sp Total reactive power weight ω sq These are the average value of the three-phase fundamental active power weights superimposed with the active power compensation component, and the average value of the three-phase fundamental reactive power weights superimposed with the reactive power regulation component, respectively:

[0060] Therefore, the formulas for calculating the instantaneous active current reference values ​​i*aa, i*ab, i*ac and the instantaneous reactive current reference values ​​i*ra, i*rb, i*rc for each phase are as follows:

[0061] The combined reference currents i*sa, i*sb, and i*sc for each phase are the sum of the instantaneous active current and instantaneous reactive current for each phase. .

[0062] In a preferred embodiment of this invention, the step of generating the phase unit voltage of each phase and the orthogonal unit voltage based on the three-phase voltage specifically comprises: The three-phase voltages are normalized to obtain the phase unit voltages that are in phase with the grid voltage. By performing Clark transformation and 90-degree phase shift on the phase unit voltage, and then inversely transforming it to the abc coordinate system, an orthogonal unit voltage is generated.

[0063] In a preferred embodiment of this invention, the X-LMS algorithm uses the following weight update formula when extracting the fundamental active power weight and fundamental reactive power weight of the load current:

[0064] In the formula, These are the fundamental active power weights for phases A, B, and C, respectively. These are the fundamental reactive power weights for phases A, B, and C, respectively. , , These are the prior active power errors for phases A, B, and C, respectively. , , These are the phase unit voltages for phases A, B, and C, respectively. Step size factor; , , These are the load currents for phases A, B, and C, respectively. , , These represent the active power auxiliary errors for phases A, B, and C, respectively; n is the sampling time index for discrete time. , , These are the orthogonal unit voltages of phases A, B, and C, respectively. , , These are the prior reactive power errors for phases A, B, and C, respectively. , , These are the reactive power auxiliary errors for phases A, B, and C, respectively.

[0065] In a preferred embodiment of this invention, the specific calculation process for the step of generating active power compensation weights based on the DC-side voltage error using a PI controller in the DC voltage control loop is as follows:

[0066] In the formula, v dce For DC voltage error signal, v dc(ref) The DC-side reference voltage, v dc This is the actual voltage on the DC side.

[0067] As a preferred embodiment of this invention, the specific calculation process for the step of generating reactive power adjustment weights based on AC voltage amplitude error using a PI controller in the AC voltage control loop is as follows:

[0068] In the formula, v teThe error signal for the AC voltage amplitude, v t(ref) For AC reference voltage amplitude, v t This represents the actual AC voltage amplitude.

[0069] In a preferred embodiment of this invention, the step of generating the reference compensation current for each phase by combining the phase unit voltage of each phase with the orthogonal unit voltage therein specifically includes: Calculate the average value of the three-phase fundamental active power weight and the average value of the fundamental reactive power weight; The total active power weight is obtained by superimposing the average value of the fundamental active power weight and the active power compensation weight. The total reactive power weight is obtained by superimposing the average value of the fundamental reactive power weight and the reactive power adjustment weight. The reference compensation current for each phase is obtained by adding the product of the total active power weight and the phase unit voltage, and the product of the total reactive power weight and the orthogonal unit voltage.

[0070] In a preferred embodiment of this invention, the distributed static synchronous compensator is a six-bridge-arm three-phase two-level converter based on IGBT, with a large capacitor connected in parallel on the DC side to maintain voltage stability, and no external DC power supply is required.

[0071] In a preferred embodiment of this invention, the hysteresis current controller compares the reference compensation current with the actual compensation current to limit the error within a preset hysteresis band and directly generates a PWM signal to control the switching transistor.

[0072] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for improving power quality in a distribution system based on the X-LMS adaptive algorithm, characterized in that, The specific steps are as follows: Real-time acquisition of three-phase voltage, load current, and DC-side voltage of distributed static synchronous compensators at the grid connection point of the distribution network; based on the three-phase voltage, generation of phase unit voltage and orthogonal unit voltage for each phase; The X-LMS algorithm is used to extract the fundamental active and reactive weights of the load current. The X-LMS algorithm achieves fast convergence and high steady-state accuracy of the weights by introducing a cross error product term. The active power compensation component and reactive power regulation component are calculated by DC voltage control loop and AC voltage control loop respectively. The DC voltage control loop generates active power compensation weight based on DC voltage error through PI controller, and the AC voltage control loop generates reactive power regulation weight based on AC voltage amplitude error through PI controller. The total active power weight is obtained by superimposing the fundamental active power weight and the active power compensation weight, and the total reactive power weight is obtained by superimposing the fundamental reactive power weight and the reactive power regulation weight. The reference compensation current of each phase is generated by combining the phase unit voltage of each phase and the orthogonal unit voltage thereto. The switching transistors of the distributed static synchronous compensator are controlled by a hysteresis current controller to output the compensation current.

2. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, The specific steps for generating the phase unit voltage and the orthogonal unit voltage of each phase based on the three-phase voltage are as follows: The three-phase voltages are normalized to obtain the phase unit voltages that are in phase with the grid voltage. By performing Clark transformation and 90-degree phase shift on the phase unit voltage, and then inversely transforming it to the abc coordinate system, an orthogonal unit voltage is generated.

3. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, When the X-LMS algorithm extracts the fundamental active power weight and fundamental reactive power weight of the load current, the weight update formula is as follows: In the formula, These are the fundamental active power weights for phases A, B, and C, respectively. These are the fundamental reactive power weights for phases A, B, and C, respectively. , , These are the prior active power errors for phases A, B, and C, respectively. , , These are the phase unit voltages for phases A, B, and C, respectively. Step size factor; , , These are the load currents for phases A, B, and C, respectively. , , These represent the active power auxiliary errors for phases A, B, and C, respectively; n is the sampling time index for discrete time. , , These are the orthogonal unit voltages of phases A, B, and C, respectively. , , These are the prior reactive power errors for phases A, B, and C, respectively. , , These are the reactive power auxiliary errors for phases A, B, and C, respectively.

4. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, The specific calculation process for the step of generating active power compensation weights based on the DC-side voltage error using a PI controller in the DC voltage control loop is as follows: In the formula, v dce This is the DC voltage error signal, v dc(ref) The DC-side reference voltage, v dc This is the actual voltage on the DC side.

5. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, The specific calculation process for the step of generating reactive power regulation weights based on the AC voltage amplitude error using a PI controller in the AC voltage control loop is as follows: In the formula, v te The error signal for the AC voltage amplitude, v t(ref) For AC reference voltage amplitude, v t This represents the actual AC voltage amplitude.

6. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, The specific steps for generating the reference compensation current for each phase by combining the phase unit voltage of each phase with the orthogonal unit voltage therein are as follows: Calculate the average value of the three-phase fundamental active power weight and the average value of the fundamental reactive power weight; The total active power weight is obtained by superimposing the average value of the fundamental active power weight and the active power compensation weight. The total reactive power weight is obtained by superimposing the average value of the fundamental reactive power weight and the reactive power adjustment weight. The reference compensation current for each phase is obtained by adding the product of the total active power weight and the phase unit voltage, and the product of the total reactive power weight and the orthogonal unit voltage.

7. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 6, characterized in that, The distributed static synchronous compensator is a six-bridge-arm three-phase two-level converter based on IGBTs, with a large capacitor connected in parallel on the DC side to maintain voltage stability, and no external DC power supply is required.

8. The method for improving power quality in a distribution system based on the X-LMS adaptive algorithm according to claim 1, characterized in that, The hysteresis current controller compares the reference compensation current with the actual compensation current, limits the error to a preset hysteresis band, and directly generates a PWM signal to control the switching transistor.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements a power quality improvement method for a power distribution system based on the X-LMS adaptive algorithm as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements a power quality improvement method for a power distribution system based on the X-LMS adaptive algorithm as described in any one of claims 1 to 8.

Images