CN112241602A - Electromagnetic transient simulation parameter optimization method based on particle swarm optimization - Google Patents
Electromagnetic transient simulation parameter optimization method based on particle swarm optimization Download PDFInfo
- Publication number
- CN112241602A CN112241602A CN202011125690.5A CN202011125690A CN112241602A CN 112241602 A CN112241602 A CN 112241602A CN 202011125690 A CN202011125690 A CN 202011125690A CN 112241602 A CN112241602 A CN 112241602A
- Authority
- CN
- China
- Prior art keywords
- particle
- parameters
- electromagnetic transient
- transient simulation
- parameter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000002245 particle Substances 0.000 title claims abstract description 140
- 238000004088 simulation Methods 0.000 title claims abstract description 110
- 238000005457 optimization Methods 0.000 title claims abstract description 75
- 230000001052 transient effect Effects 0.000 title claims abstract description 74
- 238000000034 method Methods 0.000 title claims abstract description 58
- 230000035945 sensitivity Effects 0.000 claims abstract description 38
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 30
- 238000012163 sequencing technique Methods 0.000 claims abstract description 6
- 230000003044 adaptive effect Effects 0.000 claims description 43
- 238000004364 calculation method Methods 0.000 claims description 14
- 230000000694 effects Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 8
- 238000012795 verification Methods 0.000 claims description 6
- 238000010206 sensitivity analysis Methods 0.000 claims description 5
- 230000009191 jumping Effects 0.000 claims description 3
- 238000012937 correction Methods 0.000 abstract description 7
- 238000012360 testing method Methods 0.000 description 6
- 238000004804 winding Methods 0.000 description 4
- 238000013016 damping Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000013433 optimization analysis Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000002904 solvent Substances 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/25—Design optimisation, verification or simulation using particle-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides an electromagnetic transient simulation parameter optimization method based on a particle swarm optimization algorithm, and belongs to the technical field of power system simulation models. The method comprises the following steps: acquiring an initial electromagnetic transient simulation function; according to the initial electromagnetic transient simulation function, analyzing the track sensitivity of all the parameters; quantifying the sensitivity of the track, sequencing the sensitivity of the track, and selecting a plurality of parameters with higher sensitivity of the track as parameters to be optimized; and based on a PSO algorithm, operating an electromagnetic transient simulation model to obtain optimized electromagnetic transient simulation parameters. The method adopts machine correction simulation model parameters to replace the traditional artificial experience correction, and improves the accuracy of the parameters. And parameters to be optimized are analyzed and screened through the track sensitivity, so that the optimization efficiency is improved. The PSO algorithm is adopted to solve the problem of dimensionality disaster when the number of parameters is large, and the optimization process is simplified.
Description
Technical Field
The invention belongs to the technical field of power system simulation models, and particularly relates to an electromagnetic transient simulation parameter optimization method based on particle swarm optimization.
Background
With the increasing expansion of the power grid scale and the construction of direct current transmission projects, the dynamic behavior of the power system is increasingly complex. At present, a power grid containing large-scale new energy and complex alternating-current and direct-current power transmission equipment is large in scale, strong in randomness and numerous in parameters, and has great significance for accident inversion and production practice guidance by establishing a detailed and reliable simulation model. However, in the face of a complex alternating current-direct current network with strong nonlinear characteristics, the accuracy of establishing a traditional electromechanical transient model is low, and the use requirement cannot be met, and the accuracy of a substituted electromechanical-electromagnetic hybrid simulation method (namely, the electromagnetic transient modeling is adopted for the direct current and new energy network, and the electromechanical transient modeling is still maintained for the traditional alternating current part) is increased to a certain extent, but the problem of interface handover error of the hybrid simulation is gradually exposed along with the improvement of the coupling degree of an alternating current-direct current system. The other alternative method is full electromagnetic transient simulation, and although the method has the defects of small integration step length, huge calculated amount, low simulation efficiency and the like, the accuracy of the result is ensured, and the dynamic process of the alternating current system can be accurately depicted, so that the method is favored.
At present, electromagnetic transient simulation is widely applied, but people pay attention to the accuracy problem of simulation parameters. The vast majority of modeling processes are summarized as follows: for most of the parameters such as line parameters, generator parameters and the like can be calculated by inquiring data of the nameplate and related materials, and part of measurable parameters can be obtained by experimental measurement (such as no-load test and short-circuit test for determining transformer parameters). Other parameters that are not easily measured and do not have relevant production data as a reference are often measured using empirical data. And then, the parameters are used for establishing an electromagnetic transient simulation model on a related platform through simple verification and put into use to perform simulation work such as accident inversion. If the difference between the result and the actual situation is too large in the verification process, the experience parameters are modified as appropriate, and the result is manually adjusted by methods such as trial and error so as not to violate the actual situation. This type of adjustment is common among experienced operators and often performs poorly in small scale systems.
When the scale of the power grid is large, firstly, the parameters of the equipment have dispersion, namely, for the same equipment, the related parameters are not completely the same due to different equipment use time, different abrasion degrees and the like. And a large amount of manpower and material resources are consumed for parameters which need to be determined by methods such as no-load tests, short-circuit tests and the like, and it is not practical to correct parameters of various devices of the whole system respectively. Even if the parameter estimation is carried out by using the node information data measured by the PMU, the complicated calculation process is also needed. In addition, parameters which are difficult to measure possibly exist in a large power grid are more, and if empirical data are adopted, the accuracy cannot be guaranteed. And the manual adjustment of the empirical data can improve the simulation model to a certain extent, but when the parameters to be adjusted are excessive, the adjustment complexity can be involved in dimension disaster. The manual adjustment can not find the optimal parameters at all, and only can approximately find the parameter combination with the consistent trend, so that the model loses the accuracy.
Disclosure of Invention
In view of the above, the invention provides an electromagnetic transient simulation parameter optimization method based on particle swarm optimization, so as to solve the technical problems of the prior art that the simulation parameter acquisition process is complicated, and the accuracy is poor depending on subjective experience.
The technical scheme adopted by the invention for solving the technical problems is as follows:
an electromagnetic transient simulation parameter optimization method based on particle swarm optimization comprises the following steps:
a. acquiring an initial electromagnetic transient simulation function;
b. according to the initial electromagnetic transient simulation function, analyzing the track sensitivity of all the parameters;
c. quantifying the sensitivity of the track, sequencing the sensitivity of the track, and selecting a plurality of parameters with higher sensitivity of the track as parameters to be optimized;
d. and based on a PSO algorithm, operating an electromagnetic transient simulation model to obtain optimized electromagnetic transient simulation parameters.
Preferably, in step d, the step of operating the electromagnetic transient simulation model based on the PSO algorithm and obtaining the optimized electromagnetic transient simulation parameters includes the following steps:
d1. connecting the electromagnetic transient simulation model with a PSO main program;
d2. setting PSO parameters: setting a C1 value, a C2 value, the number of particles, upper and lower speed limits, upper and lower particle position limits, the maximum iteration times and convergence conditions; meanwhile, an adaptive function is set;
d3. the PSO main program is operated, and the method comprises the following steps:
d31. initializing the state of each particle, including randomly selecting the position and the speed of the particle within a set range;
d32. calculating the corresponding adaptive function value of each particle position, at the moment, bringing the parameter values corresponding to the particle positions into a simulation model and running simulation, and calculating the adaptive function value by collecting the state of a simulation result; updating the pbest parameter for each particle according to the adaptive function value, and recording the particle function value corresponding to the pbest of each particle;
d33. finding out the particle with the minimum adaptive function in all the particles of the generation, comparing the particle with the function value corresponding to the gbest, if the value is less than the minimum adaptive function, updating the gbest to be the position of the particle, and recording the adaptive function value corresponding to the gbest;
d34. for each particle, updating the particle velocity;
d35. updating the positions of the particles according to the updated particle speed of each particle;
d36. detecting whether a termination condition is met: and if the termination condition is met, jumping out the loop termination program and outputting the gbest and the corresponding position, otherwise, returning to the step d32.
d4. Outputting corresponding particle parameters according to the obtained gbest, wherein the particle parameters are optimized electromagnetic transient simulation parameters; and (4) bringing the particle parameters into an electromagnetic transient simulation model, running simulation, observing whether the observation result is the same as the adaptive function of the gbest, and finishing verification if the observation result is the same as the adaptive function of the gbest.
Preferably, in step d34, in the "updating the particle velocity for each particle", the updating manner of the particle velocity is as follows:
Vik+1＝w×Vik+c1×rand1×(pbest-Xik)+c2×rand2×(gbest-Xik)
wherein, Vik+1Representing the updated particle velocity, VikRepresenting the particle velocity before update, w is the inertial weight, c1，c2Representing the weighting factors, rand1 and rand2 are random numbers in the interval of (0,1), XikRepresenting the position of the particle at the moment;
in this case, in step d35, the update method of the position of the particle in the "update of the position of the particle according to the particle velocity after update of each particle" is as follows:
Xik+1＝Xik+Vik+1
wherein, Xik+1Representing the updated particle position.
Preferably, after the position of the particle is updated, checking whether the particle is out of limit; if the particle exceeds the set limit, the position of the particle is set as the limit value, and the particle velocity is updated.
Preferably, in step d36, the termination condition includes reaching a maximum number of iterations or reaching a desired optimization effect.
Preferably, in step d, the step of operating the electromagnetic transient simulation model based on the PSO algorithm and obtaining the optimized electromagnetic transient simulation parameters further includes the following steps:
d5. and optimizing a PSO algorithm according to the optimized electromagnetic transient simulation parameters, and performing secondary local PSO.
Preferably, in step d5, in the step "optimizing the PSO algorithm according to the optimized electromagnetic transient simulation parameters, and performing secondary local PSO", the PSO algorithm is optimized by:
improving the PSO algorithm; or
Improving the selection of the adaptive function value and the weight of each state; or
And step-by-step optimization is carried out, and dimensionality is reduced.
Preferably, in the step b, in the "analyzing the track sensitivity of all parameters according to the initial electromagnetic transient simulation function", the track sensitivity is analyzed according to the following formula:
wherein x represents a state variable of the system, y represents an output variable of the system, and α represents a relevant parameter.
Preferably, a parameter perturbation method is adopted to solve the trajectory sensitivity analysis, discretization is carried out on the parameters, linearization processing is carried out only near a preset value, and the variation trend of the adaptive function along with the parameters is calculated, wherein the calculation formula is as follows:
wherein x represents a state variable of the system, y represents an output variable of the system, and α represents a relevant parameter.
Preferably, in the step c, the method for quantifying the sensitivity of the track comprises the following steps of:
according to the norm of the difference phasor between the measured data and the simulation data; or
According to the distortion ratio of the waveform in the fault state.
According to the technical scheme, the invention provides an electromagnetic transient simulation parameter optimization method based on particle swarm optimization, which has the beneficial effects that: due to the fact that calculation of electromagnetic transient simulation is long in time consumption, the optimization problem is solved by a traditional method and is inevitably involved in dimension disaster, and calculation efficiency cannot be guaranteed when the number of parameters is large. And the introduction of the particle swarm optimization algorithm can better solve the problem. The method can completely replace the parameter estimation method which takes expert experience as the leading factor in the past, save a large amount of manpower and material resources, and ensure the accuracy and reliability of simulation. In addition, the algorithm has certain flexibility, if the effect does not meet the requirement, certain improvement space is provided, the PSO parameters and the optimization route can be modified as appropriate according to the collected particle running track information, and even the optimization range is reduced according to the estimated value, so that the efficiency is improved or the optimization effect is improved.
Drawings
FIG. 1 is a PSO optimization model parameter flow diagram.
Detailed Description
The technical scheme and the technical effect of the invention are further elaborated in the following by combining the drawings of the invention.
Referring to fig. 1, in a specific embodiment, a method for optimizing electromagnetic transient simulation parameters based on a particle swarm optimization algorithm aims to correct parameters of a conventional electromagnetic transient simulation, and estimate unknown parameters in a circuit through PSO and a large amount of measured data to guide accident inversion and production practice. The algorithm takes accuracy and calculation efficiency into consideration, the estimated parameters meet the error requirement, and certain model training efficiency is ensured.
An electromagnetic transient simulation parameter optimization method based on particle swarm optimization comprises the following steps:
a. acquiring an initial electromagnetic transient simulation function;
b. according to the initial electromagnetic transient simulation function, analyzing the track sensitivity of all the parameters;
c. quantifying the sensitivity of the track, sequencing the sensitivity of the track, and selecting a plurality of parameters with higher sensitivity of the track as parameters to be optimized;
d. and based on a PSO algorithm, operating an electromagnetic transient simulation model to obtain optimized electromagnetic transient simulation parameters.
According to the electromagnetic transient simulation parameter optimization method based on particle swarm optimization, the traditional manual experience correction is replaced by the machine correction simulation model parameters, and the accuracy of the parameters is improved. And parameters to be optimized are analyzed and screened through the track sensitivity, so that the optimization efficiency is improved. The PSO algorithm is adopted to solve the problem of dimensionality disaster when the number of parameters is large, and the optimization process is simplified.
In particular, the amount of the solvent to be used,
b. according to the initial electromagnetic transient simulation function, performing track sensitivity analysis on all parameters:
the parameters of the power system are numerous, and for a large system, it is difficult and unnecessary to correct all the parameters, generally speaking, the processing mode is to sort the parameters according to the influence on the result, and screen out the parameters with higher track sensitivity according to the calculation efficiency and the application scene for subsequent optimization analysis.
The trajectory sensitivity calculation formula is as follows:
where x represents the state variable of the system, y represents the output variable of the system, and α represents the relevant parameter.
The trajectory sensitivity is generally calculated by numerical integration, and the formula (1) is an analytic method based on a mathematical model, namely a model analytic expression at the position needs to be known, an electromagnetic transient model is considered as a black box model at the position, and the analytic expression cannot be obtained, so that a parameter perturbation method is adopted for solving, parameter summation is discretized, linearization processing is only carried out near a preset value, and the variation trend of an adaptive function along with the parameters is calculated, wherein the calculation formula is as follows:
of course, this approximate calculation method is effective in the vicinity of the reference value of the parameter. In the actual use process, the parameter is required to be within a reasonable variation range, and as long as the variation range is not large, the parameter track sensitivity near the reference value can be used as the track sensitivity in the whole interval for processing. The state variable x here can be selected to be measurable at any position in the circuit. Since parameters in the fault inversion process are to be examined, the waveform should be considered in segments.
c. Quantifying the sensitivity of the track, sequencing the sensitivity of the track, and selecting a plurality of parameters with higher sensitivity of the track as the parameters to be optimized:
and (4) analyzing the track sensitivity of all parameters to be optimized, and quantizing the state in the process. Different quantification modes can be adopted for different scenes, such as norm of difference phasor between measured data and simulation data (1-norm, 2-norm and infinite norm can be used according to conditions), or distortion proportion of waveforms in a fault state and the like. When a fault exists, the fault existence time can be considered independently and is split into two states: steady state and transient state. And each part is given a different weight depending on its degree of importance, the final fitness function being a weighted average of all considered states.
After selecting proper states and evaluation criteria, solving the track sensitivity at the reference value for each parameter by using an equation (2), sequencing the parameters according to the weighted state variables, and selecting a plurality of parameters with the maximum track sensitivity in a reasonable quantity as optimization parameters. If the state is more, certain waveforms influenced by only a few parameters can be considered preferentially, and the complexity of the problem can be reduced by preferentially processing a simpler optimization problem. If a certain waveform is only related to a single parameter, the accurate value of the parameter can be found directly by discretizing the interval and then traversing the whole interval, so that the dimensionality of the problem can be directly reduced.
If it is desired to consider waveform accuracy for only certain states, the parameter trajectory sensitivities for those states can be analyzed separately and only those states can be analyzed as the fitness function. The selection of the adaptive function and the optimization parameter is directly related to the specific problem and the requirement, the solution can be designed according to different situations, the problem is solved through one-time PSO (particle swarm optimization) without being carved, the original problem can be decomposed into a plurality of sub-problems to be solved, and the half-effort is achieved.
d. And based on a PSO algorithm, operating an electromagnetic transient simulation model to obtain optimized electromagnetic transient simulation parameters.
The parameter correction of the electromagnetic transient simulation model of the power system by using the PSO is specifically divided into the following steps:
d1 connecting simulation software and PSO, sorting optimization examples, and ensuring that simulation program can be run through the bottom end;
d2 sets PSO parameters including, but not limited to, C1 value, C2 value, number of particles, upper and lower speed limits, upper and lower particle position limits, maximum iteration number, convergence condition, etc. And meanwhile, an adaptive function is set, and the quantization standard of each state is selected according to the situation and weighted average is carried out.
d3 running the PSO main program, the concrete implementation process is as follows:
d31 initializes the state of each particle, including randomly selecting the particle position and velocity within a set range. The particle position represents the value of each parameter, and the particle range represents the variation range of each parameter.
d32 calculating the adaptive function value corresponding to each particle position, and taking the parameter value corresponding to each particle position into the simulation model and running the simulation, and calculating the adaptive function value by collecting the state of the simulation result. Updating pbest parameter for each particle according to the adaptive function value (pbest represents the optimal state position of the particle history, when the adaptive function is smaller than the current optimal state of the particle history, the corresponding position of the optimal adaptive function value is updated, the function value corresponding to the pbest of each particle is set to infinity during initialization), and then recording the particle function value corresponding to the pbest of each particle
d33 finding the minimum particle of the adaptive function of all the particles, comparing with the function value corresponding to the gbest, if it is less than this value, updating the gbest to the position of the particle (gbest represents the historical optimum state of all the particles, when the adaptive function is less than the current gbest, the optimal value will be updated, when the adaptive function value corresponding to the gbest is set to infinity at initialization), and recording the adaptive function value corresponding to the gbest.
d34, updating the speed parameter of each particle according to the updating mode shown in formula (3):
Vik+1＝w×Vik+c1×rand1×(pbest-Xik)+c2×rand2×(gbest-Xik)
(3)
wherein Vik+1Representing the updated particle velocity, VikRepresenting the particle velocity before update, w is the inertial weight, c1，c2Representing the weighting factors, rand1 and rand2 are random numbers in the interval of (0,1), XikRepresenting the position of the particle at this time. The basic meaning of this equation is that the traveling direction of the particle at the next time is affected by the velocity of the particle of the previous generation, the distance between the particle and the optimum position of all the particles, and the distance between the particle and the optimum position of the particle. The velocity of the final particle is updated to the sum of the three phasors multiplied by the weight and it is necessary to check that the velocity cannot be exceeded for each dimension (if the velocity is exceeded then this velocity is set as the threshold).
d35 updates the positions of the particles by the updated velocity of each particle calculated as described above, and the updating method is shown in equation (4):
Xik+1＝Xik+Vik+1
(4)
wherein Xik+1Representing the updated particle position. The same check is carried out after the updating is finishedAnd whether the particles are out of limit is caused. If the particle exceeds the set limit, the particle position is set as the limit value and the velocity is updated (where the velocity processing mode has multiple processing modes, such as setting to zero or keeping the original value and reversing)
d36 detects whether the termination condition is met: and (4) reaching the maximum iteration number or achieving the expected optimization effect, if any one of the iteration numbers is met, jumping out of the loop termination program and outputting the gbest and the corresponding position, and otherwise, returning to the step d32.
d4 outputting the corresponding particle parameters according to the obtained gbest, wherein the parameter combination is the optimized parameter combination corrected by the algorithm. And (4) bringing the parameter combination into an electromagnetic transient simulation model, running simulation, and finishing verification if an observation result is the same as the adaptive function of the gbest.
d5 optimizing PSO algorithm according to the optimization result (difference between parameter value and simulation data and measured data under the parameter), improving PSO parameter or adaptive function weight, and performing secondary local PSO.
The optimization result of the algorithm has two possibilities, if the error requirement is met, the cycle is ended, and the parameters are verified, so that the set of parameters can meet the use requirement without subsequent optimization. However, if the maximum number of iterations is reached and the error requirement is not met, the algorithm needs to be improved. The specific improvement method comprises the following ideas:
firstly, the PSO algorithm is improved, and the improvement can be selected according to actual conditions. The motion track of the particles can be output in the optimization process, if the particles easily touch the boundary, the speed of the particles can be set to be reverse when the particles touch the wall, and the upper limit of the speed is reduced. If the result is still in the fall-off period and does not converge completely, an attempt may be made to increase the number of particles or the number of iterations. If the particles are clustered in a very short time and fall into local optima, the upper limit of the particle velocity can be reduced or the weights of pbest and gbest can be reduced. And selecting different optimization measures according to different conditions, and then carrying out secondary PSO to observe the improvement effect.
And secondly, the selection of the adaptive function value and the weight of each state are improved. Since the states in the circuit can be selected manually, it is contemplated to change the state selected into the fitness function. Meanwhile, if the difference between a certain state and the actual condition is found to be particularly large in the process of verifying the result, the weight of the state can be increased in the process of secondary optimization, and therefore the optimization effect of the corresponding parameters is guaranteed.
And thirdly, if the optimization effect is still unsatisfactory after the adjustment of the first step and the second step, the step-by-step optimization can be considered so as to reduce the dimensionality of the problem. Not every state is related to all parameters in the circuit for a state in the circuit and a parameter to be optimized, so the optimization problem can be broken down into simpler small problems: some of the parameters are optimized with several associated states, which are preferably independent of or less sensitive to other parameters, to ensure that the values of other parameters are not correct and do not affect the verification of the problem. PSO converges more easily at lower dimensions.
The technical concept and technical effects of the present invention are further described below by a specific embodiment.
In the embodiment, a single-machine infinite non-excitation speed regulation system is adopted, wherein the simulation time is 20s, the fault is introduced in the 3 rd s, and the fault is cleared in the 3.1 s. Fault type three is short-circuited to ground. The parameter to be optimized is d-axis synchronous reactance X of the generatordD-axis transient reactance Xd', d-axis sub-transient reactance Xd", q-axis synchronous reactance XqQ-axis transient reactance Xq', q-axis sub-transient reactance Xq", d-axis exciting winding f stator open-circuit time constant Td0' D-axis damping winding D stator open-circuit time constant Td0", q-axis damping g-winding stator open-circuit time constant Tq0Stator open-circuit time constant T of' Q-axis damping Q windingq0", rotor inertia time constant Tj. The measurable states include motor speed, bus voltage, generator output active power, power angle, bus current and load power.
The problem contains 11 parameters to be optimized, the states are numerous, and due to the introduction of faults, the states at the moment of the fault can be independently used as an evaluation criterion. Therefore, the state is changed into eleven states, namely the motor rotating speed, the fault voltage, the voltage after the fault, the generator power during the fault, the generator power after the fault, the steady-state power angle, the transient power angle, the fault current, the current after the fault, the load power during the fault and the load power after the fault.
As the complete running data and accurate values of the simulation parameters exist in the calculation example, all the parameters are considered to be unknown, and the accurate values of the simulation parameters are only used as final evaluation criteria to be compared with an optimization result. The preset parameter ranges are up and down 20% of the accurate value. The precise values of each simulation parameter are as follows:
TABLE 1 exact values of simulation parameters in the examples
Firstly, calculating the track sensitivity of each parameter, wherein the calculation formula is (1) and (2) for each state, and the selected adaptive function is the difference phasor 2-norm of the state phasor calculated by simulation and the actually measured state phasor:
TABLE 2 trajectory sensitivity of the simulation parameters in the example
Analysis data shows that the steady-state power angle is only related to Xq, so that the parameters can be optimized independently, and the parameters can be calibrated by adopting an equidistant segmentation method and then a trial-and-error method. For simplicity, a one-dimensional PSO was performed here, and the value of this parameter was directly confirmed to be 1.8899.
Confirmation of XqThen, for the remaining ten parameters, the problem of smaller dimension is solved first in consideration of the calculation efficiency. Note that rotor speed only includes XqThe five variables are related, if the state is taken as the adaptive function, the solution containing T can be solved separatelyj，Xd，Xd’,Td0' optimization of four parameters (other parameter values are chosen independently of the state)
Therefore, the 2-norm of the difference phasor between the simulation result and the rotor speed of the actual data is used as an adaptive function to carry out the second PSO optimization. The optimization result is as follows:
TABLE 3 corrected parameter results obtained after second PSO
| Xd | Xd’ | Td0’ | Tj |
| 1.8989 | 0.2755 | 4.9974 | 45.6145 |
It can be seen that the error between the back-calculated parameters and the actual parameters after simulation is small, and the simulation effect can be considered to be good.
Finally, the remaining six parameters are optimized, and PSO is carried out for three times, the adaptive function can select or carry out weighted average on all states except the two used states, and the selection mode of the adaptive function is directly related to the convergence rate and the optimization effect, so that the weights of some states can be increased or reduced through repeated tests, and a solution with good performance is obtained.
After a plurality of tests, the following calculation formula is used as the adaptive function:
where result _ norm1-11 represents the 2-norm of the difference phasors of the eleven state simulation data and measured data in the above table, respectively.
After PSO optimization, the output result is as follows:
TABLE 4 comparison of simulation results with accurate values
| Parameter name | Xd | Xd’ | Xd” | Xq | Xq’ | Xq” |
| Precise value | 1.8899 | 0.2717 | 0.1952 | 1.8899 | 0.2717 | 0.1952 |
| Simulation value | 1.8989 | 0.2755 | 0.1923 | 1.8899 | 0.2709 | 0.1981 |
| Parameter name | Td0’ | Td0” | Tq0’ | Tq0” | Tj | |
| Precise value | 5 | 0.035 | 0.5 | 0.035 | 45.762 | |
| Simulation value | 4.9974 | 0.036 | 0.4969 | 0.036 | 45.6145 |
It can be seen that the error between the parameter value after simulation correction and the parameter accurate value does not exceed 3%, and the parameter correction function of the algorithm is proved.
While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.
Claims (10)
1. An electromagnetic transient simulation parameter optimization method based on particle swarm optimization is characterized by comprising the following steps:
a. acquiring an initial electromagnetic transient simulation function;
b. according to the initial electromagnetic transient simulation function, analyzing the track sensitivity of all the parameters;
c. quantifying the sensitivity of the track, sequencing the sensitivity of the track, and selecting a plurality of parameters with higher sensitivity of the track as parameters to be optimized;
d. and based on a PSO algorithm, operating an electromagnetic transient simulation model to obtain optimized electromagnetic transient simulation parameters.
2. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method according to claim 1, wherein in the step d, the step of operating the electromagnetic transient simulation model based on the PSO algorithm to obtain the optimized electromagnetic transient simulation parameters comprises the following steps:
d1. connecting the electromagnetic transient simulation model with a PSO main program;
d2. setting PSO parameters: setting a C1 value, a C2 value, the number of particles, upper and lower speed limits, upper and lower particle position limits, the maximum iteration times and convergence conditions; meanwhile, an adaptive function is set;
d3. the PSO main program is operated, and the method comprises the following steps:
d31. initializing the state of each particle, including randomly selecting the position and the speed of the particle within a set range;
d32. calculating the corresponding adaptive function value of each particle position, at the moment, bringing the parameter values corresponding to the particle positions into a simulation model and running simulation, and calculating the adaptive function value by collecting the state of a simulation result; updating the pbest parameter for each particle according to the adaptive function value, and recording the particle function value corresponding to the pbest of each particle;
d33. finding out the particle with the minimum adaptive function in all the particles of the generation, comparing the particle with the function value corresponding to the gbest, if the value is less than the minimum adaptive function, updating the gbest to be the position of the particle, and recording the adaptive function value corresponding to the gbest;
d34. for each particle, updating the particle velocity;
d35. updating the positions of the particles according to the updated particle speed of each particle;
d36. detecting whether a termination condition is met: and if the termination condition is met, jumping out the loop termination program and outputting the gbest and the corresponding position, otherwise, returning to the step d32.
d4. Outputting corresponding particle parameters according to the obtained gbest, wherein the particle parameters are optimized electromagnetic transient simulation parameters; and (4) bringing the particle parameters into an electromagnetic transient simulation model, running simulation, observing whether the observation result is the same as the adaptive function of the gbest, and finishing verification if the observation result is the same as the adaptive function of the gbest.
3. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method according to claim 2, wherein in the step d34, the particle velocity is updated in the following manner in the "updating the particle velocity for each particle":
Vik+1＝w×Vik+c1×rand1×(pbest-Xik)+c2×rand2×(gbest-Xik)
wherein, Vik+1Representing the updated particle velocity, VikRepresenting the particle velocity before update, w is the inertial weight, c1，c2Representing the weighting factors, rand1 and rand2 are random numbers in the interval of (0,1), XikRepresenting the position of the particle at the moment;
in this case, in step d35, the update method of the position of the particle in the "update of the position of the particle according to the particle velocity after update of each particle" is as follows:
Xik+1＝Xik+Vik+1
wherein, Xik+1Representing the updated particle position.
4. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method of claim 3, wherein after the positions of the particles are updated, whether the particle out-of-limit condition occurs is checked; if the particle exceeds the set limit, the position of the particle is set as the limit value, and the particle velocity is updated.
5. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method of claim 1, wherein in step d36, the termination condition comprises reaching a maximum number of iterations or reaching a desired optimization effect.
6. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method according to claim 2, wherein in step d, the step of operating the electromagnetic transient simulation model based on the PSO algorithm and obtaining the optimized electromagnetic transient simulation parameters further comprises the steps of:
d5. and optimizing a PSO algorithm according to the optimized electromagnetic transient simulation parameters, and performing secondary local PSO.
7. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method according to claim 6, wherein in the step d5, in the step "optimizing the PSO algorithm according to the optimized electromagnetic transient simulation parameters and performing the secondary local PSO", the PSO algorithm is optimized by:
improving the PSO algorithm; or
Improving the selection of the adaptive function value and the weight of each state; or
And step-by-step optimization is carried out, and dimensionality is reduced.
8. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method according to claim 1, wherein in the step b, in the step of performing trajectory sensitivity analysis on all parameters according to the initial electromagnetic transient simulation function, the trajectory sensitivity analysis is performed according to the following formula:
wherein x represents a state variable of the system, y represents an output variable of the system, and α represents a relevant parameter.
9. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method of claim 8, wherein a parameter perturbation method is adopted to solve trajectory sensitivity analysis, discretization is carried out on a parameter sum, linearization processing is carried out only near a preset value, and a change trend of an adaptive function along with the parameter is calculated, wherein the calculation formula is as follows:
wherein x represents a state variable of the system, y represents an output variable of the system, and α represents a relevant parameter.
10. The particle swarm optimization-based electromagnetic transient simulation parameter optimization method of claim 1, wherein in the step c, the track sensitivities are quantized and sorted, and a plurality of parameters with higher track sensitivities are selected as the parameters to be optimized, and the method for quantizing the track sensitivities comprises the following steps:
according to the norm of the difference phasor between the adopted measured data and the simulation data; or
According to the distortion ratio of the waveform in the fault state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125690.5A CN112241602A (en) | 2020-10-20 | 2020-10-20 | Electromagnetic transient simulation parameter optimization method based on particle swarm optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011125690.5A CN112241602A (en) | 2020-10-20 | 2020-10-20 | Electromagnetic transient simulation parameter optimization method based on particle swarm optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112241602A true CN112241602A (en) | 2021-01-19 |
Family
ID=74169229
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011125690.5A Pending CN112241602A (en) | 2020-10-20 | 2020-10-20 | Electromagnetic transient simulation parameter optimization method based on particle swarm optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112241602A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406434A (en) * | 2021-05-14 | 2021-09-17 | 杭州电子科技大学 | SVG dynamic parameter segmentation optimization identification method based on parameter fault characteristics |
| CN113468792A (en) * | 2021-07-22 | 2021-10-01 | 国网宁夏电力有限公司电力科学研究院 | Parameter correction method and device of electromagnetic transient simulation model and electronic equipment |
| CN113704990A (en) * | 2021-08-23 | 2021-11-26 | 贵州电网有限责任公司 | Single-parameter online correction method for electromagnetic transient model of direct-current measurement data |
| CN114236401A (en) * | 2021-12-20 | 2022-03-25 | 上海正泰电源***有限公司 | Battery state estimation method based on adaptive particle swarm optimization |
| CN114325387A (en) * | 2022-01-17 | 2022-04-12 | 湖南科技大学 | Method for monitoring state of induction motor based on particle swarm inversion sliding-mode observer |
| CN115130359A (en) * | 2022-05-27 | 2022-09-30 | 南方电网科学研究院有限责任公司 | Digital twin parameter correction method for improving convergence effect |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101369002A (en) * | 2008-10-08 | 2009-02-18 | 东北电力大学 | Method for amending simulation parameter of generator by measured track and trace sensitivity |
| CN102520353A (en) * | 2011-12-09 | 2012-06-27 | 清华大学 | Synchronous generator model parameter multi-step identification method |
| US20150188415A1 (en) * | 2013-12-30 | 2015-07-02 | King Abdulaziz City For Science And Technology | Photovoltaic systems with maximum power point tracking controller |
| CN106169747A (en) * | 2016-07-20 | 2016-11-30 | 河海大学 | A kind of double fed induction generators parameter identification method |
-
2020
- 2020-10-20 CN CN202011125690.5A patent/CN112241602A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101369002A (en) * | 2008-10-08 | 2009-02-18 | 东北电力大学 | Method for amending simulation parameter of generator by measured track and trace sensitivity |
| CN102520353A (en) * | 2011-12-09 | 2012-06-27 | 清华大学 | Synchronous generator model parameter multi-step identification method |
| US20150188415A1 (en) * | 2013-12-30 | 2015-07-02 | King Abdulaziz City For Science And Technology | Photovoltaic systems with maximum power point tracking controller |
| CN106169747A (en) * | 2016-07-20 | 2016-11-30 | 河海大学 | A kind of double fed induction generators parameter identification method |
Non-Patent Citations (5)
| Title |
|---|
| JIAN ZHANG等: "Parameters Identification of Equivalent Model of Permanent Magnet Synchronous Generator (PMSG) Wind Farm Based on Analysis of Trajectory Sensitivity", 《ENERGIES》 * |
| YANKAN SONG等: "Efficient GPU-Based Electromagnetic Transient Simulation for Power Systems With Thread-Oriented Transformation and Automatic Code Generation", 《IEEE ACCESS》 * |
| 张爽等: "基于电压源换流器的多端柔性直流输电***拟交流最优潮流", 《电力***保护与控制》 * |
| 文静等: "基于轨迹灵敏度的模型参数校正方法研究", 《黑龙江电力》 * |
| 潘学萍等: "基于短路电流辨识双馈感应发电机的模型参数", 《电力自动化设备》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406434A (en) * | 2021-05-14 | 2021-09-17 | 杭州电子科技大学 | SVG dynamic parameter segmentation optimization identification method based on parameter fault characteristics |
| CN113406434B (en) * | 2021-05-14 | 2022-05-31 | 杭州电子科技大学 | SVG dynamic parameter segmentation optimization identification method based on parameter fault characteristics |
| CN113468792A (en) * | 2021-07-22 | 2021-10-01 | 国网宁夏电力有限公司电力科学研究院 | Parameter correction method and device of electromagnetic transient simulation model and electronic equipment |
| CN113704990A (en) * | 2021-08-23 | 2021-11-26 | 贵州电网有限责任公司 | Single-parameter online correction method for electromagnetic transient model of direct-current measurement data |
| CN113704990B (en) * | 2021-08-23 | 2023-11-10 | 贵州电网有限责任公司 | Electromagnetic transient model single-parameter online correction method for direct-current measurement data |
| CN114236401A (en) * | 2021-12-20 | 2022-03-25 | 上海正泰电源***有限公司 | Battery state estimation method based on adaptive particle swarm optimization |
| CN114236401B (en) * | 2021-12-20 | 2023-11-28 | 上海正泰电源***有限公司 | Battery state estimation method based on self-adaptive particle swarm algorithm |
| CN114325387A (en) * | 2022-01-17 | 2022-04-12 | 湖南科技大学 | Method for monitoring state of induction motor based on particle swarm inversion sliding-mode observer |
| CN114325387B (en) * | 2022-01-17 | 2023-10-03 | 湖南科技大学 | Method for monitoring state of induction motor of sliding mode observer based on particle swarm inversion |
| CN115130359A (en) * | 2022-05-27 | 2022-09-30 | 南方电网科学研究院有限责任公司 | Digital twin parameter correction method for improving convergence effect |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112241602A (en) | Electromagnetic transient simulation parameter optimization method based on particle swarm optimization | |
| CN109598435B (en) | Power distribution network cable running state evaluation method and system | |
| CN105512799B (en) | Power system transient stability evaluation method based on mass online historical data | |
| CN110417011B (en) | Online dynamic security assessment method based on mutual information and iterative random forest | |
| CN110119570B (en) | Actually measured data driven wind farm model parameter checking method | |
| US11900031B2 (en) | Systems and methods of composite load modeling for electric power systems | |
| CN108155648A (en) | Method for estimating state based on the infinite Extended Kalman filter of adaptive H | |
| CN107358542B (en) | Method for constructing excitation system performance evaluation model | |
| CN111523778A (en) | Power grid operation safety assessment method based on particle swarm algorithm and gradient lifting tree | |
| CN112054552A (en) | Identification method and device for control parameters of photovoltaic inverter | |
| CN107565553A (en) | A kind of power distribution network robust dynamic state estimator method based on UKF | |
| CN112200694B (en) | Dominant instability mode identification model construction and application method based on graph neural network | |
| CN104716646B (en) | A kind of node Coupling Degrees method based on Injection Current | |
| CN113805064A (en) | Lithium ion battery pack health state prediction method based on deep learning | |
| CN110738232A (en) | grid voltage out-of-limit cause diagnosis method based on data mining technology | |
| CN108090615B (en) | Minimum frequency prediction method after power system fault based on cross entropy integrated learning | |
| CN112907075A (en) | Method for identifying parameters of comprehensive load model of power system | |
| CN115293257A (en) | Detection method and system for abnormal electricity utilization user | |
| CN112395814A (en) | Optimization method for parameter identification of load model of power system | |
| CN114583767B (en) | Data-driven wind power plant frequency modulation response characteristic modeling method and system | |
| CN103793556A (en) | Transformer substation load modeling method based on outlet load synthesis | |
| CN109921462B (en) | New energy consumption capability assessment method and system based on LSTM | |
| CN109799454A (en) | Generator stator insulation residual breakdown strength prediction technique based on particle group optimizing | |
| CN117131341A (en) | Electric meter operation error estimation method, and out-of-tolerance electric meter judgment method and system | |
| CN116227320A (en) | Double-fed fan control parameter identification method based on LSTM-IPSO |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210119 |
|
| RJ01 | Rejection of invention patent application after publication |