Multi-time Scale Coordinated Optimization Scheduling Considering Flexible Resources

With the growth of new energy installations such as wind power and photovoltaic, the contradiction between supply and demand balance in the power system is becoming increasingly prominent, mainly reflected in the abandonment of new energy, deep peak shaving of units, and Primary Frequency Regulation of units. This article proposes a multi-time scale optimization scheduling strategy that considers multi-link flexible resources. It aims to promote the consumption of clean energy, enhance the flexibility of the power system, and promote the coordinated operation of different time scales. Firstly, the operational characteristics of the power system at different time scales were analyzed, and the current situation and problems of deep peak shaving and Primary Frequency Regulation were analyzed. Secondly, an optimization model was established with the goal of minimizing system operating costs and new energy waste electricity, which includes three stages: weekly optimization, daily optimization, and real-time optimization. Finally, an IEEE-30 node network structure was used for example verification, and the results showed that the proposed scheduling mode could effectively improve system flexibility.


Introduction
Recently, the randomness and volatility of the power system have significantly increased, and the scheduling mode of the power system is facing significant challenges [1][2].The randomness and volatility of new energy output reflect different characteristics at different time scales, which poses a huge challenge to the power balance of the power system.It is prone to problems such as new energy abandonment, increased cost of deep peak shaving of units, and unqualified Primary Frequency Regulation indicators of units [3][4].It is urgent to study the connection and scheduling modes at different time scales.
A certain amount of research has been conducted domestically and internationally on the optimization and scheduling of flexible resources and different time scales in the power system.The research mainly focuses on the optimization of flexible resources, the construction of optimization models, and the optimization of solving algorithms.In terms of flexible resource optimization, Ma et al. [5] studied the long-term peak-shaving problem.They pointed out that deep peak shaving of units can promote the consumption of new energy, but it will increase the operating costs of units and environmental pollution costs.Liao et al. [6] focused on studying the problem of short-term Primary Frequency Regulation (The abbreviation PFR is used in the following text to represent), pointing out that fluctuations in the output of new energy sources can lead to changes in system frequency which is an effective stabilizing method for PFR of units.Liu et al. [7] provided a brief explanation of the issue of linkage, reducing operational costs by coordinating and optimizing scheduling at three-time scales: daily, intraday, and real-time.In terms of optimizing the model, researchers [7][8][9] have established the objective function with the lowest system operating cost.Wu et al. [8] have cited the penalty of new energy waste to promote new energy consumption in the objective function.Shu et al. [9] have added network line constraints to the constraint conditions, thereby promoting the coordinated optimization of the power supply and grid.In terms of the solution algorithm, Zhou et al. [10] combed the solution algorithm, compared, and analyzed the accurate algorithm and approximation algorithm in many aspects.Shu et al. [11] adopted the accurate solution algorithm of 48-hour rolling iteration, but its calculation time is still long, which is difficult to meet the actual engineering needs.
Overall, current research mainly focuses on analyzing a single type of resource, lacking a careful consideration of multi-link flexible resources and a consideration for deep peak shaving and primary frequency modulation.In addition, the optimization scheduling mode proposed for a single time scale has a significant deviation from actual production, making it difficult to reveal the operational patterns of the system in different cycles.
This article first analyzes the operational characteristics of the power system at different time scales, especially for long-term unit peak shaving and short-term PFR.Then a multi-time scale optimization model was established.The objective function was set to minimize system costs and new energy waste electricity, and the constraints included factors such as deep peak shaving and PFR.Finally, an IEEE30 node network structure was used for example analysis, and research conclusions and policy recommendations were proposed.

Multiple time scale flexibility requirements
Considering the operating characteristics of units, load forecasting accuracy, the flexibility of power systems varies at different time scales.As shown in Figure 1  The operational flexibility of the annual time scale is reflected in the adequacy of capacity, which needs to be considered during the planning of the grid.The flexibility of long-time scales, such as month and week, is reflected in the flexibility of unit peak shaving.The flexibility of mid-time scales, such as pre-day and intra-day, is reflected in the deviation of new energy and load forecasting.The real-time short-term flexibility is reflected in the frequency modulation performance of the unit and energy storage.

Long-time scale deep peak shaving
As shown in Figure 2 below, basic peak shaving refers to the normal operating state of the unit from the minimum output to the rated output, mainly including fuel cost.The state of deep peak shaving without oil injection is between the stable limit without oil injection and the minimum output, and the cost includes fuel and additional losses.The deep peak shaving of oil injection refers to the stage from the stable limit of unit oil injection to the stable limit of no oil injection, and the cost includes fuel cost, unit loss, i.e., oil injection cost, etc.In actual scheduling, the cost and constraints of deep peak shaving were not considered, resulting in cost waste and profit loss for the unit.

Short time scale primary frequency modulation
The principle of PFR in the power system is shown in Figure 3.
f 50 The PFR of the unit belongs to the short-term active power balance, which is essentially the process of adjusting the output of the unit to complete the frequency or load response within a relatively short time scale.To cope with system frequency fluctuations caused by renewable energy, the PFR capability of thermal power units is crucial for maintaining grid frequency stability.The speed inequality rate is the main indicator to measure the PFR ability of a unit, usually expressed as a percentage of the difference between the corresponding no-load and full-load speed and the rated speed ratio, which is the slope of the static characteristic curve of the steam turbine control system in the following figure.The speed difference rate of thermal power units should be 3-6%, and the corresponding reserve capacity for PFR is require

Weekly optimization model
The objective function of the model is shown in Equation ( 1): in Equation ( 1), F is the equivalent total operating cost of the system, N is the number of units, T represents scheduling periods, and , i t P is the output of the units i during the period t ;   The unit climbing constraints are shown in Equations (6-7): Rotate Alternate Constraint is shown in Equation ( 9): , ,max , Line flow constraint is shown in Equation ( 10 is the maximum output of the unit i , and  is the rotation reserve coefficient., l t P is the active power flowing through the branch l , ,max l P is the capacity of the branch l , N is the number of units, L N is the number of loads, G is the distribution factor matrix of DC power flow transfer.

Example parameters
The calculation example adopts the modified IEEE-30 node network structure in [5], which has a total of 30 nodes and 41 branches.Six units are located at nodes 1, 2, 5, 8, 13, and 15, respectively.The parameters of the units are obtained from typical data in reference literature, and the deep peak shaving data of the units is obtained from Ma et al.'s work [5].The speed difference rate of the unit's PFR is set at 5%; The new energy output data comes from the actual data of a new energy base.The load is based on the typical weekly and daily load curves of a certain location.

Optimal results
By substituting the example parameters into the aforementioned three models, the results can be obtained as shown below, limited to displaying the output of units.Figure 5 shows the results of the weekly optimization model, it can be seen that as the load and new energy fluctuate within a week, the output of units increases and decreases in an orderly manner.There is a difference in the peak valley difference between working and nonworking days, and the unit has effectively adjusted the peak.The weekly optimization operation has a power consumption rate of 90% for wind turbines and 88% for photovoltaics.The policy recommendations mainly include the following points.Firstly, PFR is an effective means to improve frequency fluctuations in the power grid and improve power quality.We need to consider PFR capacity.Secondly, the additional costs brought by deep peak shaving of the unit are needed to be calculated and reflected in the unit scheduling or compensation.Finally, we should pay attention to the operational characteristics and requirements of the power system at different time scales and comprehensively consider various types of flexible resources.

Figure 1 .
Figure 1.Operating characteristics of the power system at different time scales.

Figure 3 .
Figure 3. Principle of unit primary frequency regulation.

Figure 7 .
Figure 7. Unit output results of real-time optimization model.5. Conclusions This article studies the operating characteristics of different time scales, such as deep peak shaving and PFR.It proposes a multi-time scale collaborative optimization scheduling model that considers multiple flexible resources.The operating characteristics of power systems at different time scales are different.Through the collaborative optimization scheduling model proposed in this article, the new energy consumption rate can be effectively improved, and the peak valley difference of the load curve can be suppressed.It can alleviate the pressure of PFR and deep peak shaving of the unit.The policy recommendations mainly include the following points.Firstly, PFR is an effective means to improve frequency fluctuations in the power grid and improve power quality.We need to consider PFR capacity.Secondly, the additional costs brought by deep peak shaving of the unit are needed to be calculated and reflected in the unit scheduling or compensation.Finally, we should pay attention to the operational characteristics and requirements of the power system at different time scales and comprehensively consider various types of flexible resources.
, oil C represents the cost of deep peak shaving and oil injection; wu C is the punish for the discharge of oil pollutants; sun C is the loss cost of the unit.oil S is the oil price (yuan/ton), is the basic operating level of the unit i at the time t of the day