Research on dynamic reactive power optimization al

  • Detail

Research on dynamic reactive power optimization algorithm of high and medium voltage distribution

1 introduction reactive power optimization of distribution is an important measure to reduce losses and save energy. Its control means mainly include the switching of capacitors and the regulation of on load voltage regulating transformer taps. In the actual operation of power system, because the load is always changing, the static optimization calculation for a single time section can not meet the actual operation needs. The system optimization of load form change at each point is considered as dynamic optimization. In order to get an operable reactive power optimization scheme, it is necessary to have a feasible dynamic optimization algorithm. The zigzag experimental machine of Jinan new era Gold Testing Instrument Co., Ltd. is suitable for mechanical experiments in iron and steel plants, quality inspection stations, scientific research, colleges and universities, etc. dynamic reactive power optimization is a very complex space-time distribution nonlinear optimization problem. On the one hand, the single moment optimization is a complex nonlinear integer optimization; On the other hand, the dynamic change of load must be considered in the optimization of a period of time. For such a problem, it is very difficult to find the global optimal solution. The usual approach is to make a compromise between the computational efficiency and the global optimization, that is, to obtain a better optimization result on the basis of the simplified model

in the existing literature on dynamic reactive power optimization, the simplified models used in the algorithm can be roughly divided into two types: ① simplification of state solution space [1,2]; ② Simplification of dynamic load model. The first method is to use a certain strategy to reduce the search space of the solution. In reference [1], the state variable at stage n is defined as the total switching times of capacitors from stage 0 to stage n, which effectively reduces the search dimension. However, the capacitor control variable is simplified to 0/1 variable, and when the number of capacitors or the maximum allowable number of actions in 2025 increases, the solution scale of the whole problem will become large. Literature [2] adopts the idea of heuristic search, and reduces the search space by saving the best state of a limited number of objective function values in each stage. The amount of memory and computation required by this algorithm increases with the increase of control variables, so it can not be applied to the actual large-scale distribution system. The second method is to transform the very complex dynamic optimization problem of space-time distribution into several simple static optimization problems of space-time distribution by handling the dynamic load [3]. "Graphene also helps to reduce the weight and size of the battery, so that the results of static optimization automatically meet the dynamic optimization constraints. Literature [3] first segmented and equivalent the optimal loss curve, and then divided the system/node load curve segment accordingly, so as to equalize the load curve of the whole day, so that the load segmentation automatically meets the requirements of dynamic optimization constraints. However, this algorithm requires the control equipment to act at the same time, that is, all the equipment perform switching or do not act in a certain period of time

the existing dynamic reactive power optimization algorithms are all aimed at a single substation or medium voltage distribution. This paper presents a dynamic reactive power optimization algorithm for high and medium voltage distribution, which has good results in practicality and optimality

2 dynamic reactive power optimization model

2.1 dynamic load model

dynamic reactive power optimization is calculated on the premise that the system load curve and bus load distribution (provided by the load forecasting module) of the next day are known. Generally, the daily system load forecast and bus load forecast of power distribution will give the system active load data in the next 24 hours and the proportion of each bus active load in the system load, and calculate the bus reactive load according to the statistical value of load power factor

the actual load of power system changes continuously, but the continuous load curve can not be used for optimization. The usual treatment method is static in different periods, that is, the continuously changing load curve is simplified into a ladder like distribution curve, and it is considered that the load remains unchanged in each period. The more segments, the closer the final solution to the actual optimal solution. This paper selects the next 24 hours as the research object, and divides it into 24 periods in hours

2.2 restriction of action times

due to the limitation of manufacturing technology and level of China's new materials replacing plastic products, stone to plastic equipment, taps and switches have a service life of equipment characterized by allowable operation times. In the actual system operation, in order to extend the service life of the equipment, the operation regulations have a clear limit on the number of operations of the tap and switch within a certain time limit (such as one day). Therefore, the frequently changing optimization scheme has no operability, and each control variable must meet the constraint that the number of actions does not exceed a certain upper limit

it is this constraint that greatly increases the difficulty of dynamic reactive power optimization. If the constraint of action times is not considered, the optimal operation scheme of power at each time is only related to the current state, so the optimization of each time period is independent. The dynamic optimization can be transformed into the static optimization of each time period according to the load segmentation, and the time complexity of the dynamic optimization is completely eliminated. The restriction of the number of actions destroys the independence of each period, making dynamic reactive power optimization a global optimization that must be considered from the time as a whole. How to deal with this constraint is the key to dynamic optimization

2.3 optimization model

due to the radial characteristics of the distribution network structure, the optimization problem can be decomposed into sub optimization problems for each independent sub problem. The following discussion is aimed at the optimization of each sub, and it is assumed that there are n buses in the sub, m switchable capacitor banks, and 1 transformer with on load voltage regulating tap

in this paper, the objective function is to minimize the sum of system losses (energy loss) throughout the day, and capacitor switching and on load voltage regulating tap adjustment are selected as control means. Considering the voltage constraint and integer constraint of switching amount, it is assumed that the maximum allowable switching times of capacitor bank in a day for the actual system is, and the maximum allowable adjustment times of transformer tap is Research on dynamic reactive power optimization algorithm for the hottest high and medium voltage distribution network