A backup power automatic throw-in adaptive load shedding scheme based on multi-agent heuristic algorithm

Abstract

This invention presents an adaptive load shedding scheme for automatic transfer switch (ATS) based on a multi-agent heuristic algorithm. It aims to address the problems of low matching accuracy, susceptibility to over-switching, and inability to dynamically adjust traditional ATS overload shedding strategies under conditions of high renewable energy penetration. The scheme includes: establishing a dynamic database to collect and calculate pre-fault and real-time power; sorting the pre-fault and real-time load power using an insertion method to form an ordered array; targeting the power deficit, using a multi-agent heuristic algorithm to quickly find the load closest to the deficit in the ordered array and disconnecting it. After the ATS trips the main power supply, it first disconnects the matching load based on the pre-fault operating conditions. If overload persists after reclosing, it dynamically disconnects again based on the real-time operating conditions. This invention can accurately match power deficits, reduce overload risk and outage range, and improve power supply reliability, making it particularly suitable for complex power grids with a high proportion of renewable energy.

Description

Technical Field

[0001] This invention relates to the field of power systems and their automation technology, and in particular to an adaptive load shedding scheme for automatic backup switching based on a multi-agent heuristic algorithm. Background Technology

[0002] With the accelerated global energy transition, the penetration rate of new energy power generation, represented by wind power and photovoltaics, in the power system continues to increase. While bringing clean and low-carbon benefits, the inherent intermittency, randomness, and volatility of new energy sources also pose unprecedented challenges to the safe and stable operation of the power grid. Traditional power grids have relatively simple structures and fixed power flow directions. Automatic transfer switches (ATS) are key automation measures to ensure power supply reliability. Their operation logic is clear and reliable, and they are widely used in distribution networks and substations.

[0003] However, in complex power grids with a high proportion of renewable energy integration, the traditional automatic transfer switch (ATS) overload shedding control strategy has revealed significant limitations. First, fluctuations in renewable energy output can lead to uncertainties in the magnitude and direction of power flow on backup power lines, posing a risk of malfunction for traditional ATS under abnormal conditions of no voltage and no current. Furthermore, traditional ATS cannot dynamically shed loads based on load changes, potentially resulting in prolonged heavy-load operation of lines or main transformers after ATS activation, a situation increasingly prevalent due to load fluctuations. Additionally, traditional load shedding strategies often employ fixed cycles or priorities for load shedding, lacking a precise consideration of the matching degree between power deficit and load size. This often leads to a total load shedding amount far exceeding the actual deficit, causing unnecessary expansion of the outage area.

[0004] New energy sources such as wind power and photovoltaics are significantly affected by external factors such as weather, resulting in considerable fluctuations in their load-carrying capacity. Therefore, this paper proposes an adaptive load reduction scheme for automatic transfer switch (ATS) based on a multi-agent heuristic algorithm. This scheme, after the ATS disconnects the main power supply, firstly, based on the total load before the fault and the power deficit between the backup power supply and new energy sources, quickly and effectively matches the target load to be disconnected, minimizing overload after reconnection. After the target load is disconnected and the ATS reconnects to the backup power supply, if overload still occurs due to fluctuations in new energy sources, the scheme dynamically identifies and disconnects the target load in real time based on the load and the power deficit between the power supply and new energy sources. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an adaptive load reduction scheme for automatic transfer switch based on a multi-agent heuristic algorithm. Through the multi-agent heuristic algorithm, the load closest to the power deficit is automatically identified and disconnected, thereby reducing the risk of overload after the automatic transfer switch is closed, while minimizing the total amount of load disconnected and improving power supply reliability.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution: This invention provides a backup self-starting adaptive load reduction scheme based on a multi-agent heuristic algorithm, comprising the following steps: a) Establish a dynamic database: collect electrical quantities of each power source, busbar and load bay in the power system, calculate and store the power before the fault and the real-time power; b) Load power insertion sorting: The load power before the fault and the real-time load power are sorted from smallest to largest by insertion method to form an ordered load array; c) Multi-agent heuristic algorithm: Taking the power deficit as the target replacement object, the multi-agent heuristic algorithm is used to find the load closest to the power deficit in the ordered load array; d) Disconnect the target load: Disconnect the matching target load according to the operating conditions before the fault. If the load is still overloaded after the circuit is closed, disconnect the matching target load again according to the real-time operating conditions.

[0007] The establishment of the dynamic database includes: The automatic transfer switch collects and monitors the current, voltage of bus I, voltage of bus II, and secondary current of power supply 1, power supply 2, new energy bus I, and new energy bus II, and converts them into primary values ​​according to the transformation ratio; and performs power calculations according to different operating conditions to obtain the pre-fault power and real-time power of power supply 1, power supply 2, new energy bus I, new energy bus II, and each load interval. The operating conditions include at least split-line operation and parallel operation, and the calculation method for power deficit differs under different operating conditions.

[0008] The load power interpolation sorting method includes: The load power before the fault is sorted from smallest to largest by the insertion method to form the load array before the fault. The real-time load power is sorted from smallest to largest using the insertion method to form a real-time load array.

[0009] The multi-agent heuristic algorithm includes: The multi-agent heuristic algorithm includes a target replacement object ΔP and a target search array P[n], where P[n] is an array sorted in ascending order, containing n numbers: P[0], P[1], P[2], ..., P[n-1]. The multi-agent heuristic algorithm is used in array P[n] to find the number closest to the target replacement object ΔP, and then the found target replacement object is subtracted from ΔP. The specific algorithm steps are shown below.

[0010] Step 1: min=0, max=n-1, mid=(min+max) / 2; Step 2: Is ΔP less than P[max]? If it is less, proceed to step 3; otherwise, proceed to step 15. Step 3: Is ΔP greater than P[min]? If it is, proceed to step 4; otherwise, proceed to step 9. Step 4: min = mid, mid = (min + max) / 2; Step 5: Is mid greater than or equal to max-1? If yes, proceed to step 6; otherwise, continue to step 2. Step 6: Is ΔP greater than P[mid]? If so, proceed to step 7; otherwise, proceed to step 8. Step 7: ΔP = ΔP - P[max] and end; Step 8: ΔP = ΔP - P[mid] and end; Step 9: Is ΔP less than P[mid]? If so, proceed to step 10; otherwise, proceed to step 14. Step 10: Check if mid is equal to min+1. If it is, proceed to step 11; otherwise, proceed to step 2. Step 11: Is ΔP greater than P[mid]? If so, proceed to step 12; otherwise, proceed to step 13. Step 12: ΔP = ΔP - P[max] and end; Step 13: ΔP = ΔP - P[mid] and end; Step 14: ΔP = ΔP - P[max] and end; Step 15: ΔP = ΔP - P[max], and proceed to step 16; Step 16: max = max - 1, mid = (min + max) / 2, then proceed to step 17; Step 17: Determine if max equals min. If it does, proceed to step 18; otherwise, proceed to step 2. Step 18: ΔP = ΔP - P[max] and end.

[0011] The target load to be removed includes: After the automatic transfer switch disconnects the main power supply, the power deficit is calculated based on the power of each power source and load before the fault. Then, a multi-agent heuristic algorithm is used to find the load with the closest power deficit among the sorted loads before the fault and disconnect it. After the automatic transfer switch reconnects the backup power supply, it is determined whether the load is still overloaded based on the real-time power. If it is still overloaded, a multi-agent heuristic algorithm is used to find the load with the closest power deficit among the sorted real-time loads and disconnect it. The calculation of the power deficit is determined based on the operating conditions: when operating in separate sections, the power deficit is equal to the sum of the total power of the two bus sections minus the overload power setting value of the standby incoming line; when operating in parallel sections, the power deficit is equal to the total power of the main power supply bus minus the overload power setting value of the standby incoming line.

[0012] The advantages of this invention are as follows: by dynamically collecting electrical quantities and flexibly calculating power deficits based on operating conditions, it effectively addresses the uncertainty of power flow caused by fluctuations in renewable energy output; by employing a multi-agent heuristic algorithm to quickly and accurately match the load closest to the power deficit in the sorted load array, it avoids the "over-cutting" phenomenon caused by traditional fixed-cycle disconnection, minimizing the scope of power outages; and by having a dual guarantee mechanism of pre-fault prediction and real-time dynamic adjustment after closing, it significantly reduces the risk of long-term heavy loads on lines or main transformers after automatic transfer switching, and improves the power supply reliability and automation adaptability under complex power grids. Attached Figure Description

[0013] Figure 1 This is the main connection diagram of the automatic switching system of the present invention.

[0014] Figure 2 This is a flowchart of the multi-agent heuristic algorithm of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use.

[0016] Example 1: This embodiment describes the implementation process of a self-adaptive load reduction scheme based on a multi-agent heuristic algorithm, such as... Figure 1 As shown, it includes: The analog quantities to be acquired include: the current of power supply 1, power supply 2, new energy source of bus I, new energy source of bus II, voltage of bus I, voltage of bus II, and secondary current of the load interval; Power calculation includes: power supply 1, power supply 2, new energy source of bus I, new energy source of bus II, and the power before the fault and the real-time power of each load interval.

[0017] Furthermore, when the backup automatic transfer switch is running in separate phases, the total power supply on bus I and bus II is as shown in formulas (1) and (2).

[0018] (1) (2) in This refers to the total power supply on bus I. This refers to the power of incoming line number 1. This refers to the number of new energy sources connected to Motherboard I. The output power of each new energy source connected to the main grid; This refers to the total power supply on bus II. Power is applied to No. 2. The number of new energy sources connected to Motherway II. The output power of each new energy source connected to the II motherboard.

[0019] Furthermore, if the fault occurs on incoming line 1, the power deficit after the automatic transfer switch is successfully activated is as shown in Formula 3.

[0020] (3) Among them This is the overload power setting value for incoming line No. 2.

[0021] If the fault occurs at station 2, the power deficit after the automatic transfer switch is successfully activated will be as shown in formula 4.

[0022] (4) in This is the overload power setting value for incoming line 1.

[0023] Furthermore, the power of each load interval is sorted from smallest to largest using the interpolation method, thereby forming... , , ... There are a total of n loads.

[0024] Furthermore, using a multi-agent heuristic algorithm, we can find the loads that are related to the power deficit among these n loads. The closest load is removed. The specific algorithm steps are as follows.

[0025] Step 1: min=0, max=n-1, mid=(min+max) / 2; Step 2: Is it less than P[max]? If it is, proceed to step 3; otherwise, proceed to step 15. Step 3: If the value is greater than P[min], proceed to step 4; otherwise, proceed to step 9. Step 4: min = mid, mid = (min + max) / 2; Step 5: Is mid greater than or equal to max-1? If yes, proceed to step 6; otherwise, continue to step 2. Step 6: If the value is greater than P[mid], proceed to step 7; otherwise, proceed to step 8. Step 7: = -P[max] and end; Step 8: = -P[mid] and end; Step 9: If the value is less than P[mid], proceed to step 10; otherwise, proceed to step 14. Step 10: Check if mid is equal to min+1. If it is, proceed to step 11; otherwise, proceed to step 2. Step 11: If it is greater than P[mid], then execute step 12; otherwise, execute step 13. Step 12: = -P[max] and end; Step 13: = -P[mid] and end; Step 14: = -P[max] and end; Step 15: = -P[max], and execute step 16; Step 16: max = max - 1, mid = (min + max) / 2, then proceed to step 17; Step 17: Determine if max equals min. If it does, proceed to step 18; otherwise, proceed to step 2. Step 18: = -P[max] and end.

[0026] Furthermore, the target load removal includes: After the automatic transfer switch disconnects the main power supply, the power deficit is calculated based on the power of each power source and load before the fault. Then, a multi-agent heuristic algorithm is used to find the load with the closest power deficit among the sorted pre-fault load power and disconnect it. After the automatic transfer switch reconnects the backup power supply, it is determined whether the load is still overloaded based on the real-time power. If it is still overloaded, the multi-agent heuristic algorithm is used to find the load with the closest real-time power deficit among the sorted real-time load power and disconnect it.

[0027] Example 2: This embodiment describes the implementation process of an adaptive load reduction scheme based on a multi-agent heuristic algorithm for automatic backup switching. The main difference between Embodiment 2 and Embodiment 1 is that the automatic backup switching operation mode is different, resulting in different calculation formulas for power deficit. The following focuses on the power deficit calculation method of Embodiment 2.

[0028] When the automatic transfer switch is in parallel operation, the No. 1 incoming line is the main power supply, and the total power supply on Bus I is as shown in Formula 5.

[0029] (5) If the fault occurs on incoming line 1, the power deficit after the automatic transfer switch is successfully activated will be as shown in Formula 6.

[0030] (6) If incoming line 2 is the main power supply, then the total power supply on bus II is as shown in Formula 7.

[0031] (7) If the fault occurs on incoming line 2, the power deficit after the automatic transfer switch is successfully activated will be as shown in Formula 8.

[0032] (8) Furthermore, using a multi-agent heuristic algorithm, the system identifies the power deficit from the load power sorted in ascending order. The closest load was removed.

[0033] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A self-adaptive load reduction scheme for backup power supply based on a multi-agent heuristic algorithm, characterized in that, Includes the following steps: a) Establish a dynamic database: collect electrical quantities of each power source, busbar and load bay in the power system, calculate and store the power before the fault and the real-time power; b) Load power insertion sorting: The load power before the fault and the real-time load power are sorted from smallest to largest by insertion method to form an ordered load array; c) Multi-agent heuristic algorithm: Taking the power deficit as the target replacement object, the multi-agent heuristic algorithm is used to find the load closest to the power deficit in the ordered load array; d) Disconnect the target load: Disconnect the matching target load according to the operating conditions before the fault. If the load is still overloaded after the circuit is closed, disconnect the matching target load again according to the real-time operating conditions.

2. The adaptive load reduction scheme based on a multi-agent heuristic algorithm according to claim 1, characterized in that, The establishment of the dynamic database includes: The automatic transfer switch collects and monitors the current, voltage of bus I, voltage of bus II, and secondary current of power supply 1, power supply 2, new energy bus I, and new energy bus II, and converts them into primary values ​​according to the transformation ratio; and performs power calculations according to different operating conditions to obtain the pre-fault power and real-time power of power supply 1, power supply 2, new energy bus I, new energy bus II, and each load interval. The operating conditions include at least split-line operation and parallel operation, and the calculation method for power deficit differs under different operating conditions.

3. The adaptive load reduction scheme based on a multi-agent heuristic algorithm according to claim 1, characterized in that, The load power interpolation sorting method includes: The load power before the fault is sorted from smallest to largest by the insertion method to form the load array before the fault. The real-time load power is sorted from smallest to largest using the insertion method to form a real-time load array.

4. The self-adaptive load reduction scheme based on a multi-agent heuristic algorithm according to claim 1, characterized in that, The multi-agent heuristic algorithm includes: The multi-agent heuristic algorithm includes a target replacement object ΔP and a target search array P[n], where P[n] is an array sorted from smallest to largest, containing n numbers: P[0], P[1], P[2]...P[n-1]. The algorithm uses the multi-agent heuristic algorithm in array P[n] to find the number closest to the target replacement object ΔP, and then subtracts the found target replacement object from ΔP. The specific algorithm steps are as follows: Step 1: min=0, max=n-1, mid=(min+max) / 2; Step 2: Is ΔP less than P[max]? If it is less, proceed to step 3; otherwise, proceed to step 15. Step 3: Is ΔP greater than P[min]? If it is, proceed to step 4; otherwise, proceed to step 9. Step 4: min = mid, mid = (min + max) / 2; Step 5: Is mid greater than or equal to max-1? If yes, proceed to step 6; otherwise, continue to step 2. Step 6: Is ΔP greater than P[mid]? If so, proceed to step 7; otherwise, proceed to step 8. Step 7: ΔP = ΔP - P[max] and end; Step 8: ΔP = ΔP - P[mid] and end; Step 9: Is ΔP less than P[mid]? If so, proceed to step 10; otherwise, proceed to step 14. Step 10: Is mid less than min+1? If so, proceed to step 11; otherwise, proceed to step 2. Step 11: Is ΔP greater than P[mid]? If so, proceed to step 12; otherwise, proceed to step 13. Step 12: ΔP = ΔP - P[max] and end; Step 13: ΔP = ΔP - P[mid] and end; Step 14: ΔP = ΔP - P[max] and end; Step 15: ΔP = ΔP - P[max], and proceed to step 16; Step 16: max = max - 1, mid = (min + max) / 2, then proceed to step 17; Step 17: Determine if max equals min. If it does, proceed to step 18; otherwise, proceed to step 2. Step 18: ΔP = ΔP - P[max] and end.

5. The self-adaptive load reduction scheme based on a multi-agent heuristic algorithm according to claim 1, characterized in that, The target load to be removed includes: After the automatic transfer switch disconnects the main power supply, the power deficit is calculated based on the power of each power source and load before the fault. Then, a multi-agent heuristic algorithm is used to find the load with the closest power deficit among the sorted loads before the fault and disconnect it. After the automatic transfer switch reconnects the backup power supply, it is determined whether the load is still overloaded based on the real-time power. If it is still overloaded, a multi-agent heuristic algorithm is used to find the load with the closest power deficit among the sorted real-time loads and disconnect it. The calculation of the power deficit is determined based on the operating conditions: when operating in separate sections, the power deficit is equal to the sum of the total power of the two bus sections minus the overload power setting value of the standby incoming line; when operating in parallel sections, the power deficit is equal to the total power of the main power supply bus minus the overload power setting value of the standby incoming line.

Images