Electric energy quality comprehensive treatment method and device based on multi-level collaborative optimization

Abstract

The invention discloses an electric energy quality comprehensive treatment method and device based on multilevel collaborative optimization, and belongs to the technical field of electric energy quality treatment of an electric power system. According to the method, voltage and current data of a power grid are collected in real time and multi-parameter monitoring is performed; processing the data by using an adaptive wavelet transform and singular value decomposition combined algorithm, extracting disturbance characteristics and identifying the type and position of a disturbance source; a multi-objective optimization function including harmonic suppression, voltage stabilization, three-phase balance and power factor improvement is constructed, a Pareto optimal solution set is solved, and an optimal solution is selected; and according to an optimization result, generating and executing a cooperative control strategy of each type of electric energy quality treatment equipment. Cooperative treatment of various electric energy quality problems and cooperative control of various treatment devices are realized, the accuracy, self-adaptability and economical efficiency of electric energy quality treatment are improved, and the reliability and stability of an electric power system are enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system power quality management, and specifically relates to a comprehensive power quality management method and device based on multi-level collaborative optimization. Background Art

[0002] With the widespread application of power electronics technology and nonlinear loads, power quality issues in power systems are becoming increasingly prominent. Power quality issues are mainly manifested in various forms such as harmonic distortion, voltage deviation, three-phase imbalance and power factor reduction. These problems seriously affect the safe and stable operation of power systems and the normal operation of electrical equipment.

[0003] Traditional governance methods often target single power quality issues, such as harmonics or reactive power compensation only, but lack comprehensive consideration of multiple power quality issues, resulting in limited governance effects. Existing governance equipment mostly operates with preset parameters, which makes it difficult to make adaptive adjustments based on the grid operation status and disturbance characteristics, and cannot achieve optimal control effects. Various types of power quality governance equipment mostly operate independently, lacking a unified coordination mechanism, which can easily lead to control conflicts or waste of resources. The accuracy of identifying the type and location of power quality disturbance sources is not high, making it difficult to implement targeted governance measures. With the expansion of the grid and the diversification of electricity demand, power quality issues are becoming more complex, diversified, and dynamic, and traditional single governance methods can no longer meet the needs of modern power systems for high-quality electricity.

[0004] Therefore, there is an urgent need to develop a comprehensive power quality management method that can comprehensively consider various power quality issues, realize coordinated control of multiple management equipment, and have adaptive optimization capabilities. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method and device for comprehensive power quality management based on multi-level collaborative optimization, so as to achieve the purpose of simultaneously managing multiple power quality problems such as harmonic distortion, voltage deviation, three-phase imbalance and power factor, improve the accuracy, adaptability and economy of power quality management, and enhance the safe and stable operation of the power system.

[0006] The specific technical solutions are as follows: In a first aspect, the present invention provides a method for comprehensive power quality management based on multi-level collaborative optimization, the method comprising the following steps: Step S1, real-time collection of grid voltage and current data, and multi-parameter monitoring of power quality, wherein the multi-parameters include harmonic distortion rate, voltage deviation, three-phase imbalance and power factor.

[0007] Step S2, using a combination algorithm of adaptive wavelet transform and singular value decomposition to process the collected data, extract disturbance features and identify the type and location of the disturbance source.

[0008] Step S3, construct a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solve the multi-objective optimization problem to generate a Pareto optimal solution set, and select the optimal solution based on the system status and user preference.

[0009] Step S4, generating a coordinated control strategy for various types of power quality management equipment based on the optimization results, calculating the equipment output parameters and executing them.

[0010] Adaptive wavelet transform adaptively selects the optimal scale parameter according to the local characteristics of the signal and translation parameters , so that the wavelet function The match with the disturbance signal is the greatest: ; Calculate the wavelet transform coefficients of voltage and current signals : ; is the power quality disturbance signal to be analyzed, is the wavelet function The conjugate function of .

[0011] Establish the disturbance event feature vector : ,in, is the disturbance amplitude characteristic, is the disturbance duration characteristic, is the frequency characteristic, is the energy distribution characteristic, It is the dynamic change characteristic of disturbance.

[0012] Singular value decomposition is to form a matrix of multi-parameter time series data of power quality , perform singular value decomposition: , where and are the left singular matrix and the right singular matrix respectively, is a diagonal matrix of singular values.

[0013] Extracting disturbance features and identifying the type and location of disturbance sources include: Combining the time series data of power quality multi-parameters into a matrix Perform singular value decomposition to obtain the singular value sequence ,in , and The matrices are The number of rows and columns; Construct disturbance pattern recognition indicators based on the distribution characteristics of singular value sequences: ;in, For the matrix The singular values; The improved support vector machine algorithm is used to identify the disturbance source type: Constructing training sample sets based on historical data ,in is the extracted feature vector , is the corresponding disturbance type label; Radial basis kernel function Perform feature mapping: ; In the formula, is the kernel function parameter; Construct a classification decision function: in, is the Lagrange multiplier, is the bias parameter; For the newly collected disturbance signal, the disturbance type identification result is output, and the disturbance types include: harmonic source, voltage deviation source, three-phase unbalance source and power factor abnormal source; The location of the disturbance source is determined using power network topology analysis and disturbance propagation characteristics: Build the network impedance matrix ; Calculate the disturbance propagation sensitivity matrix : ; In the formula, For Node The voltage complex number representation is: For Node The complex number representation of the injected current; Based on the collected voltage and current distortion data at the measurement point, the following optimization problem is solved to determine the location of the disturbance source: ,in, For measuring point The measured voltage, Assume that the disturbance source is located at the node The measured points calculated at The voltage, is the total number of measurement points.

[0014] The multi-objective optimization function is: ; represents the harmonic distortion minimization objective, which is defined as: , For the Subharmonic voltage effective value, is the effective value of the fundamental voltage, is the highest harmonic order considered; represents the voltage deviation minimization objective, which is defined as: ; is the actual voltage effective value, is the rated voltage value; represents the three-phase unbalance minimization objective, which is defined as: ; is the negative sequence voltage component, is the positive sequence voltage component; represents the power factor maximization objective and is defined as: ; is the phase difference between voltage and current.

[0015] An improved multi-objective particle swarm algorithm is used to solve multi-objective optimization problems, including: The control parameters to be solved for harmonic suppression, voltage stability, three-phase balance and power factor improvement are encoded as particle position vectors; Initialize the particle swarm position and velocity; evaluate each particle in the multi-objective optimization function The fitness value of each component is calculated, and the individual optimal position and the global optimal position are updated based on the Pareto dominance relationship; Update the external archive based on the crowding distance sorting mechanism to maintain the diversity of solutions and update the particle positions and velocities: ; ; In the formula, and Respectively The particle in The velocity and position in the iteration, is the optimal position of an individual, is the global optimal position, is the inertia weight, and is the acceleration constant, and is a random number in the interval [0,1]; the particle swarm is updated according to the Pareto dominance relationship until the termination condition is met.

[0016] Methods for selecting the optimal solution based on system status and user preferences include: Constructing a decision matrix ,in Indicates The Pareto solution is The value of the objective function; Introducing user preference weight vector ,satisfy and ; Calculate the weighted normalized decision matrix ,in: ; Determine the ideal solution and negative ideal solution : ; ; Compute the distance of each Pareto solution to the ideal solution and the negative ideal solution: ; ; Calculate the relative closeness of each solution : ; Select the solution with the largest relative proximity as the optimal solution, which is the final optimization solution.

[0017] The power quality control equipment includes: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ; ; ; In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

[0018] The method further comprises the following steps: Step S5, establishing a governance effect evaluation indicator system, including technical indicators, economic indicators and environmental indicators; Step S6, calculating each evaluation index based on the real-time monitoring data and comparing it with the set target; Step S7, dynamically adjust the optimization target weights and control strategy parameters according to the evaluation results to form a closed-loop optimization mechanism.

[0019] In the second aspect, based on the same inventive concept, the present invention provides a comprehensive power quality management device based on multi-level collaborative optimization, which is used to execute the method of the first aspect. The device includes a monitoring perception module, a disturbance identification module, a decision optimization module and an execution control module.

[0020] The monitoring and sensing module is used to collect grid voltage and current data in real time, and perform multi-parameter monitoring of power quality, wherein the multi-parameters include harmonic distortion rate, voltage deviation, three-phase imbalance and power factor.

[0021] The disturbance identification module is used to process the collected data using a combination algorithm of adaptive wavelet transform and singular value decomposition, extract disturbance features and identify the type and location of the disturbance source.

[0022] The decision optimization module is used to construct a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solve the multi-objective optimization problem to generate a Pareto optimal solution set, and select the optimal solution based on system status and user preferences.

[0023] The execution and control module is used to generate collaborative control strategies for various types of power quality management equipment according to the optimization results, calculate the output parameters of the equipment and execute them.

[0024] The power quality control equipment includes: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ; ; ; In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

[0025] Compared with the prior art, the present invention has the following beneficial effects: The present invention constructs a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, so that the system can deal with multiple power quality problems at the same time; adopts a combined algorithm of adaptive wavelet transform and singular value decomposition to extract features of power quality disturbance signals, and combined with an improved support vector machine algorithm, can accurately identify the type and location of disturbance sources; solves the Pareto optimal solution set through a multi-objective particle swarm algorithm, and selects the optimal solution in combination with system status and user preferences, so that the governance strategy can be flexibly adjusted according to actual conditions; through comprehensive governance of power quality problems, grid disturbances are reduced, power supply quality is improved, the safe and stable operation capability of the power system is enhanced, and the service life of electrical equipment is extended. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a comprehensive power quality management method based on multi-level collaborative optimization of the present invention; Figure 2 It is a schematic diagram of the composition of a comprehensive power quality management device based on multi-level collaborative optimization of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described implementation mode is a part of the present invention, not all implementation modes. Based on the implementation modes of the present invention, all other implementation modes obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] Example 1 like Figure 1 As shown, it is a flow chart of a comprehensive power quality management method based on multi-level collaborative optimization of the present invention, and the method comprises the following steps: Step S1, real-time collection of grid voltage and current data, and multi-parameter monitoring of power quality, wherein the multi-parameters include harmonic distortion rate, voltage deviation, three-phase imbalance and power factor.

[0029] In this embodiment, the grid voltage and current data acquisition adopts high-precision digital sampling technology, and the sampling frequency is set to 20kHz to ensure that harmonic components up to 10kHz can be accurately captured. The monitoring equipment is deployed at key nodes of the distribution network, such as substation busbars, important branch line branch points, and large nonlinear load access points. In multi-parameter monitoring, the harmonic distortion rate is calculated to the 50th harmonic according to the IEEE-519 standard; the voltage deviation is evaluated based on the national standard GB / T 12325-2008; the three-phase imbalance is calculated using the symmetrical component method; and the power factor is obtained by the ratio of active power to apparent power. For example, the distribution network of an industrial park is monitored using this system. Within a week, the total harmonic distortion rate (THD) was detected to be an average of 4.8%, with a maximum of 7.2%; the voltage deviation fluctuated within the range of ±5.8%; the three-phase imbalance was an average of 2.3%; and the power factor was an average of 0.86.

[0030] Step S2, using a combination algorithm of adaptive wavelet transform and singular value decomposition to process the collected data, extract disturbance features and identify the type and location of the disturbance source.

[0031] This system uses db4 wavelet as the mother wavelet, because it has good time-frequency localization characteristics and is particularly suitable for analyzing transient disturbances in power systems. The scale parameter a is set to a value range of [2 −3 ,2 5 ], and the translation parameter b is adaptively determined according to the signal window length. For example, for a typical voltage sag event, the adaptive wavelet transform is used to determine the voltage sag at a=2 −1 The characteristics of the start and end times of the temporary sag are clearly captured at the scale. The calculated wavelet coefficients show an obvious peak at the point where the temporary sag occurs, and the amplitude is 3.5 times the normal value, providing a reliable basis for accurately identifying the disturbance type and the time of occurrence.

[0032] Step S3, construct a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solve the multi-objective optimization problem to generate a Pareto optimal solution set, and select the optimal solution based on the system status and user preference.

[0033] Step S4, generating a coordinated control strategy for various types of power quality management equipment based on the optimization results, calculating the equipment output parameters and executing them.

[0034] Adaptive wavelet transform adaptively selects the optimal scale parameter according to the local characteristics of the signal and translation parameters , so that the wavelet function The match with the disturbance signal is the greatest: ; Calculate the wavelet transform coefficients of voltage and current signals : ; is the power quality disturbance signal to be analyzed, is the wavelet function The conjugate function of .

[0035] Establish the disturbance event feature vector : ,in, is the disturbance amplitude characteristic, is the disturbance duration characteristic, is the frequency characteristic, is the energy distribution characteristic, It is the dynamic change characteristic of disturbance.

[0036] The specific calculation method of each component of the disturbance characteristic vector is as follows: Ea represents the disturbance amplitude characteristic, which is quantified by the maximum relative deviation between the disturbance signal and the reference signal; Et is the disturbance duration characteristic, which is determined by detecting the time difference between the start and end of the disturbance; Ef is the frequency characteristic, which is characterized by the energy distribution of wavelet decomposition coefficients in each frequency band; Ee is the energy distribution characteristic, which calculates the ratio of signal energy during the disturbance period to that during normal operation; Ed is the dynamic change characteristic of the disturbance, which is characterized by the maximum and average values ​​of the first-order derivative of the signal during the disturbance process. Taking the harmonic disturbance in a factory as an example, the extracted characteristic vector is E=[0.23, 0.85, 0.67, 0.31, 0.12], where a larger Et value indicates that this is a continuous disturbance, and a higher Ef value indicates that the disturbance is mainly concentrated in a specific frequency band.

[0037] Singular value decomposition is to form a matrix of multi-parameter time series data of power quality , perform singular value decomposition: , where and are the left singular matrix and the right singular matrix respectively, is a diagonal matrix of singular values.

[0038] U is an m×m orthogonal matrix, known as a left singular matrix, and its column vectors are called left singular vectors, which form a set of standard orthogonal bases in the row space of the matrix X. In power quality analysis, the column vectors of the U matrix can be interpreted as the basic mode or characteristic mode of power quality disturbance, and each mode represents an independent form of disturbance. For example, the first column of the U matrix may correspond to the basic mode of harmonic disturbance, the second column may correspond to the basic mode of voltage fluctuation, and so on.

[0039] V is an n×n orthogonal matrix, called a right singular matrix, and its column vectors are called right singular vectors, which constitute a set of standard orthogonal bases in the column space of the matrix X. In power quality data analysis, the column vectors of the V matrix can be understood as the distribution pattern of the disturbance in the time or frequency domain. Each column of the V matrix represents the distribution weight of the corresponding disturbance pattern at different time points or frequency points, which helps to analyze the time characteristics or spectrum characteristics of the disturbance.

[0040] Σ is an m×n diagonal matrix with the elements σ on the main diagonal 1 ≥σ 2 ≥...≥σ r ≥0 is the singular value of the matrix X, where r=min(m,n) is the rank of the matrix X; the singular value reflects the energy or importance of the corresponding disturbance mode. The larger the singular value, the more significant the contribution of the mode in the original data; in power quality analysis, the first few larger singular values ​​usually occupy the vast majority of the total energy, which shows that although the power quality disturbance is complex, it can actually be represented by a combination of a small number of basic modes. This is the theoretical basis for the singular value decomposition to effectively identify disturbance characteristics.

[0041] Singular value decomposition is a mature matrix decomposition method, and its basic mathematical form has been widely understood and applied in related fields. Therefore, the three matrices are not elaborated in detail here.

[0042] When constructing the matrix X, the voltage and current waveforms at different measurement points are arranged in time order to form the rows of the matrix, and the sampling values ​​at different times constitute the columns of the matrix. For example, for data containing 10 measurement points and 1000 sampling values ​​collected at each point, a 10×1000 matrix is ​​formed. After singular value decomposition, it is usually found that the first few singular values ​​account for more than 90% of the total energy, indicating that the power quality disturbance has obvious low-rank characteristics. In a substation monitoring example, the data containing harmonic disturbances are decomposed, and the sum of the first three singular values ​​accounts for 92.7% of the total sum of all singular values, of which the first singular value accounts for 78.3%. This distribution feature is an important indicator for identifying harmonic sources.

[0043] Extracting disturbance features and identifying the type and location of disturbance sources include: Combining the time series data of power quality multi-parameters into a matrix Perform singular value decomposition to obtain the singular value sequence ,in , and The matrices are The number of rows and columns.

[0044] Construct disturbance pattern recognition indicators based on the distribution characteristics of singular value sequences: ;in, For the matrix The singular values; The improved support vector machine algorithm adopted introduces a kernel parameter adaptive optimization mechanism, and the optimal kernel function parameter γ is determined by cross-validation. In the training phase, a database containing 500 groups of disturbance samples of different types is used, covering typical disturbance types such as harmonic sources, voltage fluctuation sources, three-phase unbalanced sources, and power factor anomaly sources. After training, the SVM classifier achieves a recognition accuracy of 95.2% on an independent test set. For example, for the disturbance signal monitored by a substation in an industrial area, the feature vector E=[0.32, 0.14, 0.78, 0.45, 0.21], the system identifies it as a harmonic source caused by the inverter. Further on-site inspections confirm that the newly added multiple inverters in the area are indeed the main source of disturbance.

[0045] The improved support vector machine algorithm is used to identify the disturbance source type: Constructing training sample sets based on historical data ,in is the extracted feature vector , is the corresponding perturbation type label.

[0046] Radial basis kernel function Perform feature mapping: ; In the formula, is the kernel function parameter.

[0047] Construct a classification decision function: in, is the Lagrange multiplier, is the bias parameter.

[0048] For the newly collected disturbance signal, the disturbance type identification result is output, and the disturbance types include: harmonic source, voltage deviation source, three-phase unbalance source and power factor abnormality source.

[0049] The location of the disturbance source is determined using power network topology analysis and disturbance propagation characteristics: Build the network impedance matrix ; The construction of the network impedance matrix Z is based on the actual parameters of the power system, including line impedance, transformer parameters and load characteristics. For medium-sized distribution networks, 50-100 key nodes usually need to be considered. The disturbance propagation sensitivity matrix calculation uses the sensitivity of node voltage to injected current to reflect the propagation law of disturbances in the network. The practical application of the location identification algorithm is as follows: A harmonic problem occurred in an industrial park. The system collected voltage harmonic data at 12 measurement points. By solving the optimization problem, it was identified that the vicinity of node 5 was the most likely harmonic source location. The optimization objective function value was 0.026, which was much lower than other candidate locations. On-site inspection found that the node was indeed connected to a large-capacity rectifier, which was highly consistent with the identification results.

[0050] Calculate the disturbance propagation sensitivity matrix : ; In the formula, For Node The voltage complex number representation is: For Node The complex number representation of the injected current.

[0051] Based on the collected voltage and current distortion data at the measurement point, the following optimization problem is solved to determine the location of the disturbance source: ,in, For measuring point The measured voltage, Assume that the disturbance source is located at the node The measured points calculated at The voltage, is the total number of measurement points.

[0052] The multi-objective optimization function is: ; represents the harmonic distortion minimization objective, which is defined as: , For the Subharmonic voltage effective value, is the effective value of the fundamental voltage, is the highest harmonic order considered; represents the voltage deviation minimization objective, which is defined as: ; is the actual voltage effective value, is the rated voltage value; represents the three-phase unbalance minimization objective, which is defined as: ; is the negative sequence voltage component, is the positive sequence voltage component; represents the power factor maximization objective and is defined as: ; is the phase difference between voltage and current.

[0053] An improved multi-objective particle swarm algorithm is used to solve multi-objective optimization problems, including: The control parameters to be solved for harmonic suppression, voltage stability, three-phase balance and power factor improvement are encoded as particle position vectors.

[0054] Initialize the particle swarm position and velocity; evaluate each particle in the multi-objective optimization function The fitness value of each component is calculated, and the individual optimal position and the global optimal position are updated based on the Pareto dominance relationship.

[0055] The particle swarm size is set to 50, and the maximum number of iterations is 100. The inertia weight ω adopts a linear decreasing strategy from 0.9 to 0.4, and the acceleration constant and c2 are both set to 2.0.

[0056] In order to avoid falling into the local optimum, mutation operation is introduced, and the mutation probability is set to 0.05. In the actual optimization process, the system initially randomly generates 50 particles, each of which represents a possible combination of control parameters. After iterative optimization, 15 Pareto non-inferior solutions are finally obtained to form the Pareto frontier. For example, in a certain distribution system, the control scheme corresponding to one Pareto solution makes the harmonic distortion rate 2.1%, the voltage deviation 1.8%, the three-phase unbalance degree 1.3%, and the power factor 0.95; the other Pareto solution makes the harmonic distortion rate 1.8%, the voltage deviation 2.2%, the three-phase unbalance degree 1.4%, and the power factor 0.92. The two schemes have their own advantages.

[0057] Update the external archive based on the crowding distance sorting mechanism to maintain the diversity of solutions and update the particle positions and velocities: ; ; In the formula, and Respectively The particle in The velocity and position in the iteration, is the optimal position of an individual, is the global optimal position, is the inertia weight, and is the acceleration constant, and is a random number in the interval [0,1]; the particle swarm is updated according to the Pareto dominance relationship until the termination condition is met.

[0058] Reaching the preset maximum number of iterations is the most basic termination condition. The system will set a sufficiently large upper limit for the number of iterations (such as 100 or 200 times) to ensure that the algorithm has sufficient search time. When the iteration counter reaches this preset value, regardless of the optimization effect, the algorithm will stop running and output the current Pareto optimal solution set.

[0059] In addition, the system calculates the change in the Pareto front in multiple consecutive iterations (for example, 10 consecutive times). When the change is less than a preset threshold (for example, 0.001), the algorithm is considered to have converged and the iteration can be terminated. This change can be measured by calculating the difference in the hypervolume or coverage index of the Pareto solution set between the previous and next two iterations.

[0060] The system monitors the update of the global optimal position. If the global optimal position is not significantly improved (the improvement is less than the preset threshold, such as 0.0005) in multiple consecutive iterations (for example, 20 times), the algorithm is considered to have reached the optimal state or fallen into a local optimal state, and the algorithm is terminated at this time. Considering the needs of real-time control, the system also sets a calculation time limit. When the algorithm running time exceeds the preset maximum allowed time (for example, for real-time control scenarios, it may be set to 200 milliseconds), the algorithm will be forced to terminate and output the current optimal solution to ensure that the control instructions can be generated and executed in time.

[0061] Methods for selecting the optimal solution based on system status and user preferences include: Constructing a decision matrix ,in Indicates The Pareto solution is The value of the objective function; Introducing user preference weight vector ,satisfy and ; Weight coefficient Adjust according to the specific application scenario. The four weights correspond to the multi-objective optimization function The importance of the four optimization objectives: the weight of the harmonic distortion rate minimization objective, the weight of the voltage deviation minimization objective, the weight of the three-phase unbalance minimization objective and the weight of the power factor maximization objective.

[0062] For example, in the power quality management of precision manufacturing enterprises, due to the high voltage stability requirements of the equipment, w2=0.4 is set, which is greater than other weights; in the metallurgical industry, due to the prominent harmonic problem, w1=0.5 is set. In the actual application of an electronics manufacturing plant, the optimal control scheme calculated by this method reduces the harmonic distortion rate from 7.6% to 2.3%, the voltage deviation from ±6.8% to ±2.1%, the three-phase imbalance from 3.5% to 1.2%, and the power factor from 0.83 to 0.96.

[0063] Calculate the weighted normalized decision matrix ,in: ; Determine the ideal solution and negative ideal solution : ; ; Compute the distance of each Pareto solution to the ideal solution and the negative ideal solution: ; ; Calculate the relative closeness of each solution : ; Select the solution with the largest relative proximity as the optimal solution, which is the final optimization solution.

[0064] The power quality control equipment includes: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ; ; ; In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

[0065] A comprehensive power quality management method based on multi-level collaborative optimization also includes the following steps: Step S5: Establish a governance effect evaluation index system, including technical indicators, economic indicators and environmental indicators. Establish a dynamic evaluation and optimization feedback mechanism for the comprehensive governance effect of power quality, including: Construct a multi-dimensional evaluation indicator system: ;in, It is a set of technical indicators, including the improvement of harmonic distortion rate, voltage deviation, three-phase imbalance and power factor, etc. It is a collection of economic indicators, including equipment investment cost, operation and maintenance cost, and energy-saving benefits; It is a collection of environmental indicators, including carbon emission reduction, resource utilization efficiency, etc. Based on the grid voltage and current data collected in real time in step S1, the current value of each evaluation index is calculated: In the formula, , and are the calculation functions of technical indicators, economic indicators and environmental indicators respectively, V and I are the voltage and current measurement values, and C is the cost parameter set; Compare with the target value and calculate the performance deviation: in, is the target value of each indicator; Step S6, calculating each evaluation index based on the real-time monitoring data and comparing it with the set target; According to performance deviation Dynamically adjust the weight vector of the multi-objective optimization function in step S3: ; It is an adaptive learning rate, which is determined by the size and change trend of the performance deviation; According to the adjusted weight vector , re-execute the optimal solution selection process in step S3 to generate a new control strategy; Construct an adaptive adjustment mechanism for optimization strategy parameters: ;in, is the control strategy parameter vector, is the parameter update step size, Performance evaluation function About parameters The gradient of Step S7, dynamically adjust the optimization target weights and control strategy parameters according to the evaluation results to form a closed-loop optimization mechanism.

[0066] Build a case library ,in is the system status description, To take control measures, For governance effectiveness; Implementing nearest neighbor reasoning based on similarity based on the case library: ;in, is the current system status, Calculate parameters for similarity; Combining historical experience and current conditions, optimize power quality management decisions and form a closed-loop self-optimizing adjustment mechanism.

[0067] In an actual application case of a large semiconductor manufacturing company, the company's power quality deteriorated significantly due to the expansion of its production line. The main manifestations were: a harmonic distortion rate of up to 8.4%, exceeding the national standard; a voltage fluctuation range of ±7.2%, causing frequent tripping of precision equipment; a three-phase imbalance of 3.9%, causing motor overheating; and a power factor drop to 0.79, resulting in increased power loss and additional electricity bills.

[0068] After adopting the multi-level collaborative optimization of power quality comprehensive management of the present invention, high-precision acquisition devices are first deployed in substations and key load points through the monitoring perception module to monitor the changes in power quality parameters in real time. The disturbance identification module uses adaptive wavelet transform and singular value decomposition algorithm to analyze the data, and accurately identifies that the disturbance source is located in the inverter group control area and the semiconductor etching equipment concentration area. Based on the company's high requirements for voltage stability, the decision optimization module sets a weight vector [0.25, 0.45, 0.15, 0.15] that focuses on voltage stability and generates an optimal control plan. The execution and control module coordinates the deployment of a 150kVA APF, two 100kvar SVCs and a 120kVA DVR to achieve collaborative control of multiple devices. Three months after the system was put into operation, the harmonic distortion rate dropped to 2.1%, the voltage fluctuation range narrowed to ±1.8%, the three-phase imbalance dropped to 1.1%, the power factor increased to 0.96, the number of equipment tripping times decreased from an average of 12 times per month to 0 times, the stable operation rate of the production line increased by 8.7%, and the annual electricity saving was about 420,000 kWh, and the annual electricity bill saving was about 357,000 yuan. The equipment investment is expected to recover its cost in 1.8 years. In addition, by reducing power loss, carbon emissions are reduced by about 328 tons each year, demonstrating significant environmental benefits.

[0069] This case fully verifies the comprehensive management capability and outstanding technical effect of the method of the present invention in solving complex power quality problems.

[0070] Example 2 like Figure 2As shown, it is a schematic diagram of the composition of a comprehensive power quality management device based on multi-level collaborative optimization of the present invention, and the device includes a monitoring perception module, a disturbance identification module, a decision optimization module and an execution control module.

[0071] The monitoring and sensing module is used to collect grid voltage and current data in real time, and to perform multi-parameter monitoring of power quality, including harmonic distortion rate, voltage deviation, three-phase imbalance and power factor; the monitoring and sensing module adopts a distributed structure, including several measurement terminal units and a central data integration unit. The measurement terminal unit consists of a high-precision voltage transformer, a current transformer, a digital signal processor DSP and a communication interface, which can monitor the voltage and current signals of 8 channels at the same time, with a sampling rate of up to 100kHz, ensuring that high-frequency disturbance components in the power system can be captured. The central data integration unit adopts an industrial-grade server, equipped with dual redundant power supplies, with high reliability and high throughput characteristics, and can receive and process monitoring data of up to 100 measurement points in real time. High-speed data transmission between units is achieved through the optical fiber ring network, and the communication rate reaches 1Gbps, ensuring the real-time and reliability of data transmission.

[0072] The disturbance identification module is used to process the collected data using a combination algorithm of adaptive wavelet transform and singular value decomposition, extract disturbance features and identify the type and location of the disturbance source.

[0073] The motion recognition module adopts a modular design, including a data preprocessing submodule, a feature extraction submodule and a classification recognition submodule. The data preprocessing submodule filters, denoises and normalizes the original collected data to improve the accuracy of subsequent analysis. The feature extraction submodule implements the adaptive wavelet transform and singular value decomposition algorithm of the present invention, and uses GPU accelerated calculation, which greatly improves the calculation efficiency. For typical power quality disturbance data, the feature extraction time is shortened to 20% of the traditional method. The classification recognition submodule has a built-in improved support vector machine algorithm. Through deep learning pre-training and transfer learning technology, the system has excellent disturbance recognition capabilities. Even in the face of new disturbance modes that have not been encountered before, it can achieve an identification accuracy of more than 85%.

[0074] The decision optimization module is used to construct a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solve the multi-objective optimization problem to generate a Pareto optimal solution set, and select the optimal solution based on the system state and user preference; a high-performance industrial computer is used as the hardware platform, equipped with a real-time operating system to ensure the stable operation of the optimization algorithm. The module has a built-in improved multi-objective particle swarm algorithm, and the solution efficiency is improved through parallel computing technology. In order to cope with the complex and changeable power grid conditions, the decision optimization module implements the intelligent scene recognition function, and can automatically adjust the optimization strategy according to the power grid operation status. For example, when the load fluctuates frequently, the system will automatically increase the weight of the voltage stability target; when a large number of nonlinear loads are detected, the harmonic suppression target is given priority. In addition, the module also has a user interaction interface, allowing professionals to manually adjust the optimization parameters and target weights according to actual needs.

[0075] The execution control module is used to generate collaborative control strategies for various types of power quality management equipment according to the optimization results, calculate the output parameters of the equipment and execute them; it consists of a main control unit and multiple power units, and adopts a hierarchical control architecture. The main control unit receives the control strategy generated by the decision optimization module, performs further real-time coordination and optimization, and generates control instructions for each power unit. The power unit includes an APF control unit, an SVC control unit, a DVR control unit, and an UPQC control unit, etc. Each unit uses a dedicated digital signal processor to achieve rapid response control. The execution control module adopts a redundant design, and the failure of any control unit will not affect the operation of the overall system. The control instructions are transmitted via industrial Ethernet, and the response time is less than 10ms, which meets the needs of real-time power quality management.

[0076] The power quality control equipment includes: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ; ; ; In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

[0077] The entire device adopts three levels of safety protection measures: hardware-level protection includes overcurrent protection, overvoltage protection and thermal protection; software-level protection includes abnormal condition detection and automatic protection strategy; system-level protection includes remote monitoring and fault self-diagnosis functions. When an abnormal situation is detected, the system can intelligently determine the type and severity of the fault and take corresponding protection measures, from alarm prompts for minor faults to emergency shutdown for serious faults, to ensure the safety of equipment and power grid.

[0078] In order to achieve long-term stable operation, a complete maintenance management system has also been designed. The system includes equipment health monitoring, predictive maintenance and remote diagnosis functions. Equipment health monitoring evaluates the health status of equipment by real-time parameter monitoring of key components such as power devices, filter capacitors and communication interfaces; predictive maintenance predicts possible failures based on big data analysis and historical equipment operation data, and arranges maintenance in advance; remote diagnosis allows professional technicians to access the system through a secure network connection to perform remote fault diagnosis and software upgrades, greatly reducing the need for on-site maintenance and system downtime.

[0079] In addition, the device has good scalability and interoperability. The system adopts modular design and standardized interfaces, and can flexibly configure hardware resources according to the application scale. By supporting standard communication protocols such as IEC 61850, the system can be seamlessly integrated into the existing power automation system and energy management system. In addition, the system has reserved an artificial intelligence expansion interface to support the introduction of advanced algorithms such as deep reinforcement learning in the future, so as to achieve continuous optimization and self-evolution of power quality management strategies.

[0080] All modules adopt industrial-grade design standards, and the overall system complies with the EMC standards and reliability requirements of power electronic equipment. It can work stably in an ambient temperature range of -20°C to 55°C, and the protection level reaches IP54, meeting the application requirements of various industries and public buildings. The system adopts redundant design and hot-swap technology to ensure the uninterrupted operation of key functions, with an annual availability rate of more than 99.9%, effectively ensuring the safe and stable operation of the power system.

[0081] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A comprehensive power quality management method based on multi-level collaborative optimization, characterized in that: The method comprises the following steps: Step S1, real-time collection of grid voltage and current data, and multi-parameter monitoring of power quality, wherein the multi-parameters include harmonic distortion rate, voltage deviation, three-phase imbalance and power factor; Step S2, using a combination algorithm of adaptive wavelet transform and singular value decomposition to process the collected data, extract disturbance features and identify the type and location of the disturbance source; Step S3, constructing a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solving the multi-objective optimization problem to generate a Pareto optimal solution set, and selecting the optimal solution based on the system state and user preference; Step S4, generating a coordinated control strategy for various types of power quality management equipment based on the optimization results, calculating the equipment output parameters and executing them.

2. According to claim 1, a comprehensive power quality management method based on multi-level collaborative optimization is characterized in that: Adaptive wavelet transform adaptively selects the optimal scale parameter according to the local characteristics of the signal and translation parameters , so that the wavelet function The match with the disturbance signal is the greatest: ; Calculate the wavelet transform coefficients of voltage and current signals : ; is the power quality disturbance signal to be analyzed, is the wavelet function The conjugate function of ; Establish the disturbance event feature vector : ,in, is the disturbance amplitude characteristic, is the disturbance duration characteristic, is the frequency characteristic, is the energy distribution characteristic, It is the dynamic change characteristic of disturbance.

3. The method for comprehensive power quality management based on multi-level collaborative optimization according to claim 2 is characterized in that: Singular value decomposition is to form a matrix of multi-parameter time series data of power quality , perform singular value decomposition: , where and are the left singular matrix and the right singular matrix respectively, is a diagonal matrix of singular values.

4. The method for comprehensive power quality management based on multi-level collaborative optimization according to claim 3 is characterized in that: Extracting disturbance features and identifying the type and location of disturbance sources include: Combining the time series data of power quality multi-parameters into a matrix Perform singular value decomposition to obtain the singular value sequence ,in , and The matrices are The number of rows and columns; Construct disturbance pattern recognition indicators based on the distribution characteristics of singular value sequences: ;in, For the matrix The singular values; The improved support vector machine algorithm is used to identify the disturbance source type: Constructing training sample sets based on historical data ,in is the extracted feature vector , is the corresponding disturbance type label; Radial basis kernel function Perform feature mapping: ; In the formula, is the kernel function parameter; Construct a classification decision function: in, is the Lagrange multiplier, is the bias parameter; For the newly collected disturbance signal, the disturbance type identification result is output, and the disturbance types include: harmonic source, voltage deviation source, three-phase unbalance source and power factor abnormal source; The location of the disturbance source is determined using power network topology analysis and disturbance propagation characteristics: Build the network impedance matrix ; Calculate the disturbance propagation sensitivity matrix : ; In the formula, For Node The voltage complex number representation is: For Node The complex number representation of the injected current; Based on the collected voltage and current distortion data at the measurement point, the following optimization problem is solved to determine the location of the disturbance source: ,in, For measuring point The measured voltage, Assume that the disturbance source is located at the node The measured points calculated at The voltage, is the total number of measurement points.

5. A comprehensive power quality management method based on multi-level collaborative optimization according to claim 1 or 4, characterized in that: The multi-objective optimization function is: ; represents the harmonic distortion minimization objective, which is defined as: , For the Subharmonic voltage effective value, is the effective value of the fundamental voltage, is the highest harmonic order considered; represents the voltage deviation minimization objective, which is defined as: ; is the actual voltage effective value, is the rated voltage value; represents the three-phase unbalance minimization objective, which is defined as: ; is the negative sequence voltage component, is the positive sequence voltage component; represents the power factor maximization objective and is defined as: ; is the phase difference between voltage and current.

6. The method for comprehensive power quality management based on multi-level collaborative optimization according to claim 5 is characterized in that: An improved multi-objective particle swarm algorithm is used to solve multi-objective optimization problems, including: The control parameters to be solved for harmonic suppression, voltage stability, three-phase balance and power factor improvement are encoded as particle position vectors; Initialize the particle swarm position and velocity; evaluate each particle in the multi-objective optimization function The fitness value of each component is calculated, and the individual optimal position and the global optimal position are updated based on the Pareto dominance relationship; Update the external archive based on the crowding distance sorting mechanism to maintain the diversity of solutions and update the particle positions and velocities: ; ； In the formula, and Respectively The particle in The velocity and position in the iteration, is the optimal position of an individual, is the global optimal position, is the inertia weight, and is the acceleration constant, and is a random number in the interval [0,1]; the particle swarm is updated according to the Pareto dominance relationship until the termination condition is met.

7. A comprehensive power quality management method based on multi-level collaborative optimization according to claim 6, characterized in that: Methods for selecting the optimal solution based on system status and user preferences include: Building a decision matrix ,in Indicates The Pareto solution is The value of the objective function; Introducing user preference weight vector ,satisfy and ; Calculate the weighted normalized decision matrix ,in: ; Determine the ideal solution and negative ideal solution : ； ； Compute the distance of each Pareto solution to the ideal solution and the negative ideal solution: ； ； Calculate the relative closeness of each solution : ; Select the solution with the largest relative proximity as the optimal solution, which is the final optimization solution.

8. The method for comprehensive power quality management based on multi-level collaborative optimization according to claim 7 is characterized in that: The power quality control equipment includes: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ； ； ； In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

9. A comprehensive power quality management device based on multi-level collaborative optimization, used to execute the method described in any one of claims 1 to 8, characterized in that: The device includes a monitoring perception module, a disturbance identification module, a decision optimization module and an execution control module; The monitoring and sensing module is used to collect grid voltage and current data in real time and perform multi-parameter monitoring on power quality, wherein the multi-parameters include harmonic distortion rate, voltage deviation, three-phase imbalance and power factor; The disturbance identification module is used to process the collected data using a combination algorithm of adaptive wavelet transform and singular value decomposition, extract disturbance features and identify the type and location of the disturbance source; The decision optimization module is used to construct a multi-objective optimization function including harmonic suppression, voltage stability, three-phase balance and power factor improvement, solve the multi-objective optimization problem to generate a Pareto optimal solution set, and select the optimal solution based on system status and user preferences; The execution and control module is used to generate collaborative control strategies for various types of power quality management equipment based on the optimization results, calculate the equipment output parameters and execute them.

10. The device for comprehensive power quality management based on multi-level collaborative optimization according to claim 9, characterized in that: The various types of power quality management equipment include: Active power filter APF, static VAR compensator SVC, dynamic voltage restorer DVR, unified power quality conditioner UPQC and hybrid power filter HPF; The collaborative control strategy generates device output parameters through the following formula: ； ； ； In the formula, is the output voltage of the active power filter, For the Subharmonic voltage components, is the compensated reactive power of the static VAR compensator, is the system active power, is the target power factor, is the current power factor, is the compensation voltage of the dynamic voltage restorer, is the reference voltage, is the actual voltage.

Images