A method for unit agc performance ranking and key unit screening based on unit data mining

Abstract

The present application relates to a kind of based on unit data mining unit AGC performance ranking and key unit discrimination method. First, the instruction generation process of unit participating AGC regulation is carried out mechanism analysis, from grid frequency, tie-line power input algorithm, the power that each unit should participate in regulation is analyzed and generated;Second, the evaluation model of unit response characteristic and the difference of instruction exists is established and the reason is carried out analysis;Third, unit AGC performance evaluation method is optimized, and combined with unit AGC actual response situation, unit AGC response characteristic evaluation and importance sorting method are proposed, the unit in front of sorting shows that it has good regulation performance, the unit in front of sorting can obtain better CPS index for system when participating in AGC regulation. Therefore, when the CPS index of system is monitored to deteriorate, the unit in front of sorting can be considered to be called first, to improve the CPS performance of system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electric power, and particularly relates to a method for sorting unit AGC performance and identifying key units based on unit data mining. BACKGROUND

[0002] In recent years, with the increasing urgent demand for clean energy worldwide, power grid companies are actively responding to national policies while vigorously promoting large-scale grid connection of new energy, especially offshore wind power. This transformation has brought profound changes to the power supply structure, with a significant increase in the proportion of new energy, and has also brought challenges to active power balance of the power system. With large-scale renewable energy connected to the power grid, the problem of insufficient frequency modulation capability of the new type of power system is prominent, and the regional power grid control performance standard (CPS) cannot meet the expected standard. In this case, the AGC performance of the unit automatic generation control (AGC) performance, especially the thermal power unit as the "ballast" of the power system, directly affects the safe and stable operation of the power grid. Among various installed capacities of thermal power units, affected by factors such as unit maintenance, operation life, and fuel quality, the response of various types of units to the AGC regulation demand of the power grid is not the same, and even the same unit under different operating conditions also has a large difference. The previous evaluation of the AGC performance of the unit is mainly carried out through field tests and joint debugging tests with dispatching, which belongs to static evaluation, and in the actual operation process of the power system, with the change of the operating condition of the unit, the evaluation result also has a certain deviation, which is not conducive to the comprehensive and dynamic evaluation of the AGC performance of the unit. For the dispatching and operating personnel, although the AGC regulation system of the power system will calculate the required regulation power within a certain time period according to the frequency fluctuation of the power grid and the power deviation of the tie line, and then issue a dispatching instruction to the relevant unit according to a certain dispatching strategy, but the dispatching instruction is fixed and cannot be flexibly adjusted according to the real-time operating performance of the unit, resulting in that the dispatching strategy of the master station of the AGC regulation system of the power system cannot adapt to the flexible operation demand of the power system under different environments. SUMMARY

[0003] The application aims to provide a method for sorting unit AGC performance and identifying key units based on unit data mining, which can flexibly adjust the regulation strategy according to the real-time operating performance of the unit, so that the dispatching strategy of the master station of the AGC regulation system of the power system can adapt to the flexible operation demand of the power system under different environments.

[0004] To achieve the above object, the technical scheme of the present application is: a method for sorting and identifying key units based on unit data mining, which comprises the following steps: first, analyzing the instruction generation process of the unit participating in AGC regulation, and calculating the power of each unit participating in regulation from the grid frequency and tie-line power input; second, establishing an evaluation model for the difference between the unit response characteristics and the instruction; third, optimizing the unit AGC performance evaluation method, and combining the actual response of the unit AGC, proposing a unit AGC response characteristic evaluation and importance sorting method, the unit with good regulation performance is sorted in the front, which can obtain better CPS index when participating in AGC regulation, when the CPS index of the system is deteriorated, the unit sorted in the front is preferentially called to improve the CPS performance of the system.

[0005] In an embodiment of the present application, the instruction generation process of the unit participating in AGC regulation is analyzed from the grid frequency and tie-line power input, and the power of each unit participating in regulation is analyzed and generated, which is implemented as follows:

[0006] The grid frequency deviation and tie-line power deviation of the grid sampling period are collected, and the ACE value ACE of the current regional control deviation ACE is calculated in the ACE sampling period P , the current Δf and the ACE integral value ACE within a period I , the ACE integral value is calculated in the ACE sampling period, and the ACE P (k) and ACE I (k) of k moment are calculated as follows:

[0007] ACE P (k)=ΔP T (k)-10B i Δf(k)

[0008]

[0009] In the formula, ΔP T (k) is the deviation of the tie-line exchange power at k moment from the planned value, i.e. the tie-line power deviation, Δf(k) is the grid frequency deviation at k moment, ACE(m) is the ACE sampling value at m moment, and T ACE is the ACE sampling period.

[0010] The total regulation power ΔP G (k) at k moment is calculated by the following formula:

[0011]

[0012] The total regulation power ΔP G(k) After each sequence value, the actual total regulation power instruction received by the unit is aligned with the calculated total regulation power in the time scale, and the abnormal values and missing values in the data are removed;

[0013] If the number of units that can be called at time k is N AGC , then the calculated total regulation power should be completely allocated to N AGC units, that is, it satisfies:

[0014] R(k) = ΔP G (k)

[0015] According to the actual situation, the actual total regulation power instruction received by the unit is much lower than the total regulation power demand calculated according to the current AGC control strategy based on the CPS evaluation standard, that is, it does not satisfy ΔP G (k) = R(k).

[0016] In an embodiment of the present application, the abnormal values include the case that the calculated total regulation power far exceeds the predetermined conventional value and the actual total regulation power instruction value received by the unit is opposite to the calculated total regulation power demand.

[0017] In an embodiment of the present application, the evaluation model for establishing the difference between the unit response characteristics and the instruction is specifically implemented as follows:

[0018] (1) Influence of the upper and lower limits of the unit instruction

[0019] According to the historical operation data of the power grid, when the total regulation power is calculated according to the current AGC control strategy based on the CPS evaluation standard, the power distribution mode is allocated to the callable units in turn according to the calling order; therefore, if the number of units that can be called in a control period is limited, the range of the total regulation power instruction L that the unit can receive at the corresponding time is:

[0020]

[0021] In the formula, is the lower limit of the regulation of the unit i, is the upper limit of the regulation of the unit i, N AGC is the number of callable units, and if the total regulation power demand ΔP G of the unit exceeds the range of the above formula, the corresponding regulation power demand cannot be completely allocated;

[0022] (2) Influence of the upper and lower limits of the unit output

[0023] When the regulation power instruction received by the unit i exceeds the limit value of the upper and lower limits of the output of the unit i, that is, , the regulation power demand of the unit i cannot be completely allocated;

[0024] (3) The impact of the unit regulation speed

[0025] The control command cycle of the unit is variable, which is determined by the AGC control cycle and the actual control command of the unit; in each AGC control command cycle, if the unit has responded to the last control command, the current control command is issued; if the unit has not responded to the last control command, the current control command is not issued; there are two criteria for whether the unit has responded to the last control command: ① according to the adjustment increment corresponding to the last control command and the given unit response speed, calculate how much time is needed to respond to the corresponding control command, and the calculated time has passed; ② although the calculated time has not arrived, the actual output of the unit has reached the control target; according to the actual operation data of the unit, when the unit is continuously called, the number of unit adjustments in a period of time is counted, and the control command cycle of the unit can be calculated; the control command cycle of the unit is affected by the unit regulation speed, when the unit regulation speed is fast, the last issued control command can be completed within the predetermined time, and the unit can be put into the next AGC regulation, while when the unit regulation speed is slow, the unit needs to execute a regulation instruction in more than the predetermined time, then the unit needs to wait for time when it is put into the next regulation; when the number of callable units is limited in a period of time, and the control command cycle of the unit is longer, it will lead to the difficulty of fully allocating the total regulation power to the unit for execution in the corresponding period of time;

[0026] Based on (1)-(3), the reasons why the actual total regulation power instruction received by the unit is less than the total regulation power demand are as follows:

[0027] 1) When the number of callable units is limited, the actual total regulation power instruction received by the unit cannot meet the total regulation power demand of the system, leading to the difficulty of fully executing the actual total regulation power instruction;

[0028] 2) Different types of units have different regulation upper and lower limits, which affect the response of the unit to the actual total regulation power instruction;

[0029] 3) When the unit receives the actual total regulation power instruction, it is restricted by the upper and lower limits of the unit output, leading to the difficulty of fully executing the actual total regulation power instruction;

[0030] 4) The regulation speed of the unit directly affects the control command cycle of the unit, which needs more time to execute an actual total regulation power instruction, leading to the small number of actual total regulation power instructions received in the limited control cycle of the unit;

[0031] 5) When the number of callable units is limited in a period of time and the control command cycle of the unit is longer, it leads to the difficulty of fully allocating the actual total regulation power instruction to the unit for execution.

[0032] In an embodiment of the present application, the method for evaluating the performance of the unit AGC is optimized, and a method for evaluating the response characteristics of the unit AGC and sorting the importance is proposed in combination with the actual response of the unit AGC. The units with high ranking indicate that they have good adjustment performance and can obtain better CPS indicators for the system when participating in AGC adjustment. When the CPS indicators of the system are deteriorated, the units with high ranking are preferentially called to improve the CPS performance of the system. The specific implementation is as follows:

[0033] (1) AGC unit adjustment amount optimization and unit importance sorting method

[0034] When the actual total adjustment power instruction is distributed, the adjustment power instruction that the unit i can receive has a range limit, and the adjustment range of the unit i is recorded as:

[0035]

[0036] In the formula, is the lower limit of the adjustment of the unit i, is the upper limit of the adjustment of the unit i. Meanwhile, the influence of the upper and lower limits of the unit output also needs to be considered. Therefore, the actual output increment that the unit i can bear when participating in adjustment is:

[0037]

[0038] When the unit receives the rising instruction, the adjustable output increment of the unit is When the unit receives the falling instruction, the adjustable output increment of the unit is When the unit reaches the upper or lower limit of the output, the unit will no longer be put into AGC operation. Therefore, the adjustment amount of the unit when called needs to be increased.

[0039] By calculating the average difference between each unit and the target adjustment power when participating in AGC adjustment in history, a reference benchmark for the adjustment amount of the unit is obtained. Meanwhile, the fluctuation degree of the adjustment shortage of the unit needs to be considered, that is, the variance of the historical adjustment difference. The unit with large fluctuation degree needs more adjustment amount to stabilize the power grid. Different units have different importance in participating in AGC adjustment. Therefore, a sorting index is constructed to comprehensively consider the adjustment shortage size, fluctuation degree and importance of the unit when participating in adjustment. The units are sorted according to the calculated sorting index, and the units with high ranking are preferentially considered to increase the adjustment range and adjustment capacity of each adjustment.

[0040] (2) CPS and unit performance correlation analysis and unit calling order sorting method

[0041] Factors affecting CPS are related not only to the magnitude of the unit's regulating power but also to the unit's regulating performance when executing regulating commands. When the unit executes regulating power commands, indicators reflecting the unit's regulating characteristics include regulating rate, regulating accuracy, and response time. To analyze the correlation between CPS and the unit's regulating rate, regulating accuracy, and response time, it is first necessary to quantitatively evaluate each indicator and align it with the CPS1 indicator on a time scale. Considering that the CPS assessment cycle is 15 minutes, each regulating indicator is calculated and its correlation with CPS is analyzed on a 15-minute cycle. Different regulating characteristics are assigned weights based on the degree of influence of changes in regulating characteristics on CPS1, and a unit evaluation index is constructed through a weighted approach. When a unit is invoked for AGC regulation, units with higher importance are given priority.

[0042] In one embodiment of the present invention, the process of ranking the generating units based on calculated ranking indicators, taking into account the magnitude of the regulation deficit, the degree of fluctuation, and the importance of the units when participating in regulation, is specifically implemented as follows:

[0043] Suppose that when the control system issues the k-th total regulation power command, the total regulation power demand calculated based on the current AGC control strategy formulated according to the CPS assessment standard is ΔP. G (k), if the control cycle has a total of N AGC The actual increase in output that these generating units can undertake when participating in regulation is as follows:

[0044]

[0045] If |ΔP G If (k)|>|ΔP(k)|, it means that a regulation deficit occurred in the k-th adjustment. This regulation deficit is distributed to the units called in the k-th adjustment in an average manner.

[0046]

[0047] According to the above formula, the regulation deficit of a unit relative to the regulation demand when participating in AGC regulation is as follows: when the actual regulation power command received by the unit is less than the regulation power demand, the unit is under-regulated; when the unit meets the regulation demand, the regulation deficit is 0. In each regulation, the regulation deficit generated by the unit reflects its performance in AGC regulation. The regulation deficit values ​​generated by different units in these periods are calculated, and these data are statistically analyzed to calculate the mean and variance of the regulation deficit for each unit.

[0048]

[0049]

[0050] Among them, R i,jThe regulation shortage of the unit i participating in the jth regulation is a regulation shortage average μ i The average level of all regulation shortages, reflecting the average deviation degree of the unit relative to the power demand to be regulated, a regulation shortage variance The dispersion degree of the regulation shortage data, used to evaluate the fluctuation degree of the regulation shortage of the unit participating in the AGC regulation;

[0051] Meanwhile, different units have different participation degrees, and the frequently called units are the key units of the AGC regulation, and the performance thereof should be paid more attention to, therefore, the participation degree of the unit is described by the regulation mileage contributed by the unit in a period of time, and a total regulation mileage of the unit i is:

[0052]

[0053] The proportion of the regulation mileage of the unit to the total regulation mileage is taken as a contribution degree index of the unit:

[0054]

[0055] The greater the corresponding proportion is, the more times the unit receives the regulation power instruction, and the more responsibility the unit bears in the frequency modulation auxiliary service market;

[0056] In order to comprehensively consider the regulation shortage average, variance and contribution degree of the unit, the regulation shortage average, variance and contribution degree are obtained by a weighted manner to obtain a comprehensive performance index, and before weighting, the indexes need to be standardized to ensure that they have similar scales before weighting.

[0057] In an embodiment of the present application, the specific manner of standardizing the indexes is as follows:

[0058] The min-max standardization method is adopted to standardize the indexes, and the data comparability is improved; and an importance ranking index of the unit in capacity optimization is obtained i The calculation method is as follows:

[0059]

[0060] Wherein, alpha, beta and gamma are respectively the weighting coefficients of the expectation and the variance, alpha+beta+gamma=1, The standardized average, variance and contribution degree indexes.

[0061] In an embodiment of the present application, in step (2), the meanings and calculation methods of the regulation indexes are as follows:

[0062] 1) Regulation response rate V

[0063] The regulation speed reflects the speed of the unit output responding to the control command. The ratio of the actual power regulation amount of the unit to the regulation time is the regulation speed. The calculation method is:

[0064]

[0065] V i,j is the rising or falling regulation speed of the unit i executing the jth regulation power instruction; T si,j is the regulation time of the unit i executing the jth regulation power instruction; P ei,j and P si,j are the actual output of the unit at the start and end time of the jth regulation power instruction, respectively. ei,j

[0066] The total regulation response speed of the system in the examination period is the sum of the regulation response speeds of the units participating in AGC. The average value of the regulation response speed of the units in the historical regulation of AGC represents the regulation response speed of the corresponding unit. The calculation method of the total regulation response speed of the system is:

[0067]

[0068] N AGC is the number of units participating in regulation in an examination period, and N i is the number of times the unit i is called.

[0069] 2) Regulation accuracy E

[0070] The regulation accuracy refers to the deviation of the actual output of the unit from the target output at the end of a regulation power instruction. The calculation method of the regulation accuracy is:

[0071] E i,j = |C i,j -P ei,j -P si,j |

[0072] E i,j is the regulation accuracy of the unit i executing the jth regulation power instruction; C i,j is the corresponding instruction value; and the average regulation accuracy of all units in an examination period represents the overall regulation accuracy of the units:

[0073]

[0074] 3) Response time T

[0075] The response time refers to the time from the issuance of the AGC instruction to the actual output of the unit jumping out of the dead zone and the direction being consistent with that of the AGC instruction. The average response time of the online running units in the examination period is:

[0076] ​

[0077] T i,j Response time of the i-th unit for executing the j-th adjustment power instruction.

[0078] In an embodiment of the present application, to determine the key factors affecting the CPS1 indicator, the correlation between the unit adjustment rate, adjustment accuracy, response time and CPS1 is calculated based on historical data by a correlation analysis method: first, the grid AGC unit operation data is collected, and the CPS1 of each evaluation period and the adjustment rate, adjustment accuracy and response time of the units participating in AGC adjustment are calculated; further, a single-variable correlation analysis is performed on each adjustment performance and CPS1, and the basic steps of the correlation analysis are as follows:

[0079] ①Pearson coefficient is used to describe the correlation between different adjustment performances and CPS, and the Pearson coefficient calculation formula of two variables X and Y is as follows:

[0080]

[0081] In the formula, and are the mean values of X and Y, x i and y i are the i-th values in X and Y, respectively, and the value range of p(X, Y) is [-1, 1], p>0 indicates that the two groups of variables are positively correlated, p<0 indicates that the two groups of variables are negatively correlated, and the larger |p| is, the stronger the correlation is.

[0082] ②Pearson coefficient is used to describe the linear correlation between unit performance and CPS1, and since the dimension and variation range of CPS1 and unit performance are quite different, in order to more obviously reflect the association rules between unit performance and CPS1, conditional probability is used to describe the relationship between unit performance and CPS1, and the specific method is as follows: different intervals are divided for different performance indicators after quantization, and the conditional probability distribution of CPS1 greater than a certain value in different intervals is obtained, assuming that the statistics X=(x1, x2,...x n ) is the average of CPS1 every 15 minutes, Y=(y1, y2,...y n ) is the adjustment performance on the corresponding time scale, and when the adjustment performance range is [p1, p2], the conditional probability of CPS1 greater than a certain value c is:

[0083]

[0084] The conditional probability can reflect the probability trend of CPS1 greater than a certain value when the unit adjustment performance is in different intervals, and the adjustment characteristics that have a significant impact on CPS1 are used as the basis for the importance ranking of the unit, and when the unit participates in AGC adjustment, the system will obtain a better CPS1 value.

[0085] The application further provides a unit AGC performance ranking and key unit screening system based on unit data mining, comprising a memory, a processor and computer program instructions stored in the memory and capable of being executed by the processor, when the processor executes the computer program instructions, the method steps as described above can be realized.

[0086] Compared with the prior art, the application has the following beneficial effects: the unit AGC performance ranking and key unit screening method based on unit data mining can flexibly adjust the adjustment strategy according to the real-time operation performance of the unit, so that the AGC adjustment system main station scheduling strategy of the power system can adapt to the flexible operation demand of the power system under different environments. BRIEF DESCRIPTION OF DRAWINGS

[0087] Figure 1 The sum of the actual adjustment instructions received by the unit is compared with the total adjustment demand under the CPS control strategy.

[0088] Figure 2 The adjustment instruction probability distribution of pumped storage units.

[0089] Figure 3 The adjustment instruction probability distribution of thermal power units.

[0090] Figure 4 The adjustment instruction probability distribution of hydropower units.

[0091] Figure 5 The comparison of the adjustment instructions of #5 unit of a hydropower plant under different conditions.

[0092] Figure 6 The adjustment instruction distribution of #5 unit of a hydropower plant.

[0093] Figure 7 The adjustment instruction distribution of #4 unit of a thermal power plant in a certain city of a certain province.

[0094] Figure 8 The flowchart of the method of the application. DETAILED DESCRIPTION

[0095] The technical solutions of the application will be specifically described below with reference to the drawings.

[0096] It should be pointed out that the following detailed description is exemplary and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs.

[0097] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0098] As Figure 8 indicated in the description, the application provides a method for unit AGC performance ranking and key unit screening based on unit data mining. First, the mechanism analysis of the instruction generation process of the unit participating in AGC regulation is carried out. From the input of the grid frequency and the tie-line power, the power of each unit participating in the regulation is analyzed and generated. Second, an evaluation model of the difference between the unit response characteristics and the instruction is established. Third, the unit AGC performance evaluation method is optimized, and the unit AGC response characteristic evaluation and importance ranking method is proposed combined with the actual response of the unit AGC. The unit ranking in the front indicates that it has good regulation performance and can obtain better CPS indicators for the system when participating in AGC regulation. When the CPS indicators of the system are monitored to deteriorate, the unit ranking in the front is preferentially considered to improve the CPS performance of the system. The following is the specific implementation process of the application.

[0099] 1. AGC unit response characteristic analysis

[0100] 1.1 AGC unit received instruction analysis based on mechanism algorithm

[0101] In order to study the relationship between the tie-line CPS performance indicators and the unit AGC regulation performance, first, the relationship between the actual received regulation instruction of the unit under the current provincial grid control strategy and the regulation demand that the unit should bear calculated according to the CPS control strategy of the provincial grid is analyzed, so as to further analyze whether there is improvement space for the current control strategy.

[0102] According to the historical operation data of the provincial grid, the actual received regulation instruction of each unit with a period of 8 seconds can be calculated. The calculation method of the regulation instruction received by unit i at time k is as follows:

[0103] ΔP Gi = instruction value - current value (1)

[0104] Suppose that there are N AGC units participating in regulation in a control period, then the total regulation instruction received by the unit at time k is:

[0105]

[0106] According to the control strategy of a certain provincial power grid, the total regulation power demand with a period of 8 seconds can be calculated according to the total regulation instruction calculation method in the previous section.

[0107] The calculation method is as follows:

[0108] The frequency deviation and tie-line power deviation collected by a certain provincial power grid with a period of 1 second are collected, and the current ACE (area control deviation) value ACE is calculated with a period of 8 seconds. P , the current Δf and the ACE integral value ACE within a period. I The ACE integral value is calculated with a period of 8 seconds, and the ACE P (k) and ACE I (k) calculation method is as follows:

[0109] ACE P (k) = ΔP T (k) - 10B i Δf(k) (3)

[0110]

[0111] In the formula, ΔP T (k) is the deviation of tie-line exchange power at time k from the planned value, Δf(k) is the frequency deviation at time k, ACE(m) is the ACE sampling value at time m, T ACE is the ACE sampling period, and the grid sampling period is 1s.

[0112] The total regulation power ΔP G (k) at time k is calculated by the following formula:

[0113]

[0114] After calculating the above sequence values, the actual instruction value received by the unit is aligned with the total regulation power in the time scale, and the abnormal values and missing values in the data are processed. The abnormal values mainly include the total regulation power far exceeding the conventional value and the sum of the regulation power instructions received by the unit and the total regulation power demand being opposite in sign. The time period when the abnormal values and missing values appear is not used as analysis data, and it is excluded.

[0115] If the number of AGC units that can be called at time k is N AGC units, the total regulation power calculated should be completely distributed to N AGC units and executed by the units, that is, it satisfies:

[0116] R(k) = ΔP G (k) (6)

[0117] The comparison between the sum of the actual regulation power received by the units in each control period from 0:00 to 6:00 on a certain day and the total regulation power demand under the CPS control strategy is shown in Fig. 1. Figure 1 It can be seen from the figure that in most periods, the total regulation power received by the units is far lower than the total regulation power demand calculated according to the current control strategy, i.e., ΔP G (k) = R(k), so although the current power grid in a certain province has formulated the corresponding AGC control strategy according to the CPS assessment standard, the instructions are often difficult to fully execute.

[0118] 1.2 Analysis of the difference between the instructions received by the AGC units and the actual response characteristics

[0119] According to the analysis in the previous section, in most periods, the total regulation received by the units is far lower than the total regulation demand calculated according to the current control strategy, although the current power grid in a certain province has formulated the corresponding AGC control strategy according to the CPS assessment standard, the instructions are often difficult to fully execute. To analyze the main reasons for this phenomenon, a more in-depth analysis of the actual operating characteristics of the units is needed.

[0120] (1) The influence of the upper and lower limits of the instructions received by the units

[0121] According to the historical operating data of a certain power grid, when the AGC generates the total regulation power, the power is allocated to the callable units in turn according to the calling order. For example, Figures 2 to 4 Fig. 2 shows the distribution histogram of the regulation instructions received by different types of units in a certain month. It can be seen that the upper and lower limits of the instructions of pumped storage and hydropower units are mostly distributed in [-10MW, 10MW], and the upper and lower limits of the instructions of thermal power units are mostly distributed in [-15MW, 15MW]. Therefore, if the number of callable units is limited, the range of the total regulation instructions L that the units can receive at this moment is:

[0122]

[0123] where, is the lower limit of the regulation of unit i, is the upper limit of the regulation of unit i, and N AGC is the number of callable units. If the total regulation demand ΔP G of the units exceeds this range, the regulation demand cannot be fully allocated.

[0124] (2) The influence of the upper and lower limits of the output of the units

[0125] From the histogram of the probability distribution of the received AGC commands, it can be seen that most of the time the received commands are distributed within the upper and lower limits, and a small part of the commands are within the upper and lower limit range. In one case, the total adjustment power of the unit needs to be adjusted at that time, and in another case, the unit is limited by the upper and lower limits of the output. The actual adjustment characteristics of #5 unit of a certain hydropower plant are analyzed (such as Figure 5 ), the output range of the unit is [100MW, 200MW], and the upper and lower limits of the AGC command are [-10MW, 10MW]. As Figure 5 (a) is the curve of the unit command value and the current value at the time when the unit normally receives the adjustment command. The adjustment command received by the unit at these times is the upper limit of the unit adjustment (10MW) and the lower limit value (-10MW); Figure 5 (b) is the comparison of the command value and the current value when the command value is the upper limit of the unit output. The current output of the unit is in the range of [190MW, 200MW], and due to the influence of the upper limit of the unit output (200MW), the adjustment command of the unit is less than the upper limit of the unit adjustment (10MW); similarly, Figure 5 (c) is the current output range of the unit in the range of [100MW, 110MW], and the current output of the unit is affected by the lower limit of the unit output (100MW). When receiving the load reduction command, the adjustment command of the unit is less than the lower limit of the unit adjustment (-10MW).

[0126] (3) Influence of unit adjustment rate

[0127] The control command period of the unit is variable, which is determined by the AGC control period and the actual control command of the unit. In each AGC control command period, if the unit has responded to the last control command, the current control command can be issued; if the unit has not responded to the last control command, the current control command is not issued. There are two criteria for whether the unit has responded to the last control command: according to the adjustment increment corresponding to the last control command and the given unit response rate to calculate how much time is needed to respond to the control command, which has passed; although the time has not arrived, the actual output of the unit has reached the control target. According to the actual operation data of the unit, the number of unit adjustments in a period of time is counted when the unit is continuously called, and the control command period of the unit can be roughly calculated. Take the AGC adjustment data of #5 unit of a certain hydropower plant and #4 unit of a certain provincial and certain city thermal power plant from 0:00 to 2:00 on a certain day as an example, as follows Figure 6 、 7The figure shows the received instruction situation of two units from 0:00 to 2:00. The average time for the #5 unit of a certain hydropower plant to execute each adjustment instruction is relatively short. Within two hours, it received 162 AGC adjustment instructions, and the control command cycle was about 45 seconds. The #4 unit of a certain thermal power plant in a certain city has a slower adjustment rate. It takes about two minutes to execute one adjustment instruction, so within two hours, it only received 54 AGC adjustment instructions, and the control command cycle was about 2 minutes.

[0128] The control command cycle of a unit is mainly affected by the unit's adjustment rate. When the unit's adjustment rate is fast, the unit can complete the previous instruction in a short time and can be put into the next AGC adjustment. When the adjustment rate is slow, the unit needs to execute one adjustment instruction in a longer time, and the unit needs to wait longer before being put into the next adjustment. When the number of available units is limited and the overall control command cycle is long, it will be difficult to fully allocate the total adjustment power to the units within a certain period of time.

[0129] This section mainly analyzes the main reasons why the current AGC adjustment instructions are difficult to be fully executed based on the actual operating characteristics of the units. The main reasons why the actual adjustment instructions received by the units are less than the total adjustment demand are as follows:

[0130] When the number of available units is limited, the total adjustment instructions received by the units may not meet the total adjustment demand of the system, resulting in insufficient execution of the instructions.

[0131] Different types of units have different adjustment upper and lower limits, which further affect the response of the units to the adjustment instructions.

[0132] When a unit receives an adjustment instruction, it is subject to the constraints of the output upper and lower limits, which may result in the unit being unable to fully execute the instruction, especially when the current output of the unit is close to or reaches its output upper limit.

[0133] The adjustment rate of a unit directly affects the control command cycle. Units with faster adjustment rates can respond to control commands more frequently, while units with slower adjustment rates may not be able to fully respond to instructions within the control cycle. Units with slow adjustment rates may need longer time to execute one adjustment instruction, resulting in fewer instructions received within a limited control cycle.

[0134] When the number of available units is limited within a certain period of time and the overall control command cycle is long, it may be difficult to fully allocate the total adjustment power to the units for execution, resulting in insufficient execution.

[0135] These conclusions help to understand the reasons for the insufficient execution of adjustment instructions by units in the AGC system of a certain provincial power grid, and provide some direction for evaluating the comprehensive performance of units and optimizing control strategies.

[0136] 2 Based on data mining CPS main influencing factor analysis and comprehensive performance sorting method

[0137] According to the above analysis, a major problem of the current AGC control system is that the AGC unit has limited regulation capacity, and the total regulation demand of the system cannot be fully allocated in most control periods. The most effective way to solve this problem is to purchase more frequency modulation capacity and let more units participate in AGC regulation. However, considering the purchase cost of regulation capacity and the mechanism of the frequency modulation auxiliary service market, it is difficult to mobilize the enthusiasm of frequency modulation power plants to participate in AGC in the short term. In order to improve the CPS index of the power grid, the regulation capacity of the main frequency modulation resource should be fully utilized, the regulation amount of the unit that frequently participates in AGC regulation should be appropriately increased, and the shortage of actual regulation and total regulation demand should be further reduced. At the same time, when the regulation amount of the unit is not much different, the main factor affecting CPS is the regulation performance of AGC unit. By analyzing the key performance indicators affecting CPS, the importance of the unit is sorted, and more regulation capacity is reserved for the unit with high importance degree and priority is given to these units in allocation, which can make the system obtain better CPS index. Therefore, the strategy to improve CPS can be analyzed from two aspects:

[0138] 1) According to the historical operation data, the regulation shortage of the unit when participating in regulation is calculated. This index can reflect the difference between the target regulation power and the actual regulation power of the unit when participating in regulation. By appropriately increasing the regulation amount of the unit according to this reference value, the regulation shortage can be reduced;

[0139] 2) The regulation characteristics of different frequency modulation resources are different. By using correlation analysis method, the key regulation performance affecting CPS can be found out, and the importance sorting index of the unit can be constructed according to the importance degree of the regulation performance. The sorting result can be used as the calling order of the unit, which can better improve the CPS index.

[0140] 2.1 AGC unit regulation amount optimization and unit importance sorting method

[0141] Under the current dispatching strategy, when the regulation command is allocated, the regulation command that the unit i can receive will have a range limit, and the regulation range is:

[0142]

[0143] In the formula, is the lower limit of the regulation of unit i, is the upper limit of the regulation of unit i, and at the same time, the influence of the upper and lower limits of the unit output should also be considered. Therefore, the actual output increment that the unit i can undertake when participating in regulation is:

[0144]

[0145] When the unit receives an ascent command, the unit can adjust the output increment to be When the unit receives a descent command, the unit can adjust the output increment to be When a generator unit reaches its upper or lower output limit, it can no longer be put into AGC (Automatic Control) operation. The main reason for this problem is insufficient regulation capacity. For such units, more regulation capacity needs to be reserved in the day-ahead plan. Therefore, to solve the problem that the actual regulation of the unit cannot meet the regulation demand, the regulation amount when the unit is called upon can be appropriately increased. For units that are put into AGC operation on the same day, and those that exit AGC operation at certain times due to reaching their upper or lower output limits, their regulation capacity should be appropriately increased.

[0146] By calculating the average difference between the target regulation power and the historical AGC (Automatic Generation Control) power of each generating unit, a reference benchmark for increasing the unit's regulation volume can be established. Simultaneously, the volatility of the unit's regulation deficit, i.e., the historical regulation variance, needs to be considered. Units with greater volatility may require more regulation to stabilize the grid. Different generating units have varying degrees of importance in AGC regulation; identifying which units are key regulation resources, whose regulation volumes have the greatest impact on grid stability and CPS (Cycles Per Power) indicators, requires prioritizing the optimization of these units' regulation volumes. Based on these considerations, a ranking index can be constructed, comprehensively considering the size and volatility of the unit's regulation deficit during regulation participation, as well as the unit's importance. The units are ranked according to the calculated ranking index, with priority given to increasing the regulation range and capacity of the top-ranked units in each regulation cycle.

[0147] Let the total regulation demand calculated according to the CPS control strategy when the control system issues the k-th regulation command be ΔP. G (k), if the control cycle has a total of N AGC The actual increase in output that these generating units can undertake when participating in regulation is as follows:

[0148]

[0149] If |ΔP G If (k)|>|ΔP(k)|, it indicates that there is a regulation deficit in this adjustment, and this regulation deficit is distributed to the units in this call in an average manner:

[0150]

[0151] According to the above analysis, the regulation shortage of the unit relative to the regulation demand can be calculated when the unit participates in the AGC regulation instruction. When the actual instruction of the unit is less than the regulation demand, the unit is under-regulated; otherwise, the unit meets the regulation demand, and the regulation shortage is 0. In each regulation event, the regulation shortage generated by the unit reflects its performance in the automatic generation control regulation. The regulation shortage values generated by different units in these periods can be calculated, and statistical analysis can be performed on these data to calculate the mean and variance of the regulation shortage of each unit.

[0152]

[0153] wherein R i,j is the regulation shortage of unit i participating in the jth regulation. The mean μ i of the regulation shortage represents the average level of all regulation shortages, reflecting the average deviation of the unit relative to the regulation demand. The variance σ2 of the regulation shortage represents the dispersion of the regulation shortage data, which is used to evaluate the fluctuation degree of the regulation shortage of the unit participating in the AGC regulation.

[0154] At the same time, different units have different participation degrees, and frequently called units are key units for AGC regulation, whose performance should be paid more attention to. Therefore, the participation degree of the unit can be described by the total regulation mileage of the unit in a period of time, and the total regulation mileage of unit i is:

[0155]

[0156] The proportion of the frequency modulation mileage of the unit to the total frequency modulation mileage is taken as the contribution degree index of the unit:

[0157]

[0158] The larger the proportion, the more instructions the unit receives, and the more responsibility the unit undertakes in the frequency modulation auxiliary service market.

[0159] In order to comprehensively consider the mean, variance and contribution degree of the regulation shortage of the unit, the mean, variance and contribution degree of the regulation shortage can be obtained by weighting. Before weighting, the indexes need to be standardized to ensure that they have similar scales before weighting. The min-max standardization method is used to standardize the indexes, which improves the comparability of the data. The importance ranking index λ i of the unit in capacity optimization is obtained.

[0160]

[0161] wherein α, β, γ are the weighting coefficients of the expectation and variance, respectively, and α+β+γ=1, The mean, variance and contribution degree after standardization. This method not only considers the size and fluctuation of the regulation shortage when the unit is called, but also considers the importance of the unit to the AGC system, which can be used to evaluate the regulation performance of the unit.

[0162] 2.2CPS and unit performance correlation analysis and unit calling order sorting method

[0163] The factors affecting CPS are not only related to the size of the unit regulation power, but also related to the regulation performance of the unit when executing the regulation instruction. The main indicators reflecting the regulation characteristics of the unit when executing the regulation instruction include the unit regulation rate, regulation accuracy and response time. In order to analyze the correlation between CPS and unit regulation rate, regulation accuracy and response time, it is necessary to quantitatively evaluate each indicator and align it with the CPS1 indicator in the time scale. Considering that the evaluation period of CPS is 15 minutes, this paper calculates each regulation indicator and analyzes its correlation with CPS every 15 minutes. The meaning and calculation method of each indicator are as follows:

[0164] (1) Regulation response rate V

[0165] The regulation rate reflects the speed of the unit output response to the control command. The ratio of the actual power regulation amount of the generator unit to the used regulation time is the regulation rate. The calculation method is

[0166]

[0167] In the formula: V i,j is the rising (falling) regulation rate of unit i executing the jth regulation instruction; T si,j , T ei,j represent the start and end time of this regulation instruction, P si,j , P ei,j are the actual output of the unit at the corresponding time.

[0168] The total regulation response rate of the system in the evaluation period is the sum of the regulation response rates of the units put into AGC. The average value of the regulation response rate in the historical regulation of AGC units represents the regulation response rate of the unit, then the calculation method of the total regulation response rate of the system is:

[0169]

[0170] Where, N AGC is the number of units participating in regulation in a evaluation period, N i is the number of times unit i is called, the same below.

[0171] (2) Regulation accuracy E

[0172] Regulation accuracy refers to the deviation between the actual output and the target output of the unit at the end of a regulation command. The calculation method for regulation accuracy is as follows:

[0173] E i,j =|C i,j -(P ei,j -P si,j (18)

[0174] E i,j The adjustment accuracy for the j-th instruction executed by unit i; C i,j This corresponds to the instruction value. The overall adjustment accuracy of the unit is characterized by the average adjustment accuracy of all units within an assessment cycle.

[0175]

[0176] (3) Response time T

[0177] Response time refers to the time from the issuance of the AGC command to the actual output of the unit exiting the dead zone and aligning with the direction of the AGC command. The average response time of the units operating online during the assessment period is as follows:

[0178]

[0179] T i,j Let be the response time for unit i to execute the j-th instruction.

[0180] To identify the key factors influencing CPS1, this study, based on historical data, uses correlation analysis to examine the correlation between the regulation rate, regulation accuracy, response time, and CPS1 of the power grid AGC units. The basic method is as follows: First, operational data of the power grid AGC units are collected, and the CPS1 for each assessment cycle (15 minutes), as well as the regulation rate, regulation accuracy, and response time of the units put into AGC regulation, are calculated. Further, univariate correlation analysis is performed on each regulation performance parameter and CPS1. The basic steps of the correlation analysis are as follows:

[0181] 1) The Pearson coefficient is used to describe the correlation between different regulation performances and CPS. The formula for calculating the Pearson coefficient of the two variables (X,Y) is as follows:

[0182]

[0183] In the formula, and Let x and y be the means of x and y, respectively. The range of ρ(X,Y) is [-1,1]. ρ>0 indicates that the two sets of variables are positively correlated, ρ<0 indicates that the two sets of variables are negatively correlated, and the larger the value of |ρ|, the stronger the correlation.

[0184] 2) Pearson coefficient is used to describe the linear correlation between unit performance and CPS1, since the dimension and variation range of CPS1 and unit performance are quite different, in order to more obviously reflect the correlation between unit performance and CPS1, conditional probability is used to describe the relationship between unit performance and CPS1, and the specific method is as follows: different intervals are divided for different performance indexes after quantization, the conditional probability distribution of CPS1 greater than a certain value in different intervals is obtained, assuming that X=(x1, x2,...x n ) is the average value of CPS1 every 15 min, Y=(y1, y2,...y n ) is the regulation performance on the corresponding time scale, when the regulation performance range is [p1, p2], the conditional probability of CPS1 greater than c is:

[0185]

[0186] The conditional probability can reflect the probability change trend of CPS1 greater than a certain value when the unit regulation performance is in different intervals, and the regulation characteristics that have a significant impact on CPS1 can be used as the basis for unit importance ranking, and the system will obtain better CPS1 value when the unit participates in AGC regulation.

[0187] The weights of different regulation characteristics can be set according to the influence degree of the change of the regulation characteristics on CPS1, and the unit evaluation index is constructed by weighting, and the unit with high importance degree is preferentially considered when it is called as AGC regulation.

[0188] The application also provides a unit AGC performance ranking and key unit screening system based on unit data mining, which comprises a memory, a processor and computer program instructions stored in the memory and capable of being executed by the processor, when the processor executes the computer program instructions, the method steps described above can be realized.

[0189] Those skilled in the art should understand that the embodiments of the application can be provided as a method, a system or a computer program product. Therefore, the application can be in the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the application can be in the form of a computer program product implemented on one or more computer usable storage media containing computer usable program code (including but not limited to disk storage, CD-ROM, optical storage, etc.).

[0190] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0191] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0192] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0193] The above description is only preferred embodiments of the present application, and is not intended to limit the present application to other forms described above. Any person skilled in the art can make modifications or improvements to the above-described embodiments based on the technical content disclosed above, and the modifications or improvements are equivalent embodiments within the scope of the present application. However, any simple modifications, equivalent changes and improvements made to the above embodiments without departing from the technical solution of the present application are still within the protection scope of the present application.

Claims

1. A method for unit AGC performance ranking and key unit screening based on unit data mining, characterized in that, Firstly, the mechanism analysis of the instruction generation process of the unit participating in AGC regulation is carried out, and the power that each unit should participate in the regulation is analyzed and generated from the input of the grid frequency and the tie-line power, and the specific implementation is as follows: Collect the grid frequency deviation and tie-line power deviation of the grid sampling period, and calculate the current area control deviation ACE value ACE with ACE sampling period P , current Δf and ACE integral value ACE within a period I , ACE integral value is calculated with ACE sampling period, ACE of k moment P (k), ACE I (k) calculation method is as follows: ACE P (k) = ΔP T (k) - 10B i Δf(k) ΔP T (k) is the deviation of tie-line interchange power from the scheduled value at time k, Δf(k) is the deviation of grid frequency at time k, ACE(m) is the ACE sample value at time m, T ACE is the ACE sampling period; The total regulating power ΔP at time k is calculated by the following equation G (k): The total regulating power ΔP is calculated G (k) After each sequence value, the actual total regulating power instruction received by the unit is aligned with the calculated total regulating power in the time scale, and the abnormal values and missing values in the data are removed; If the number of units available at time k is N AGC The total regulating power calculated should be distributed completely to the N AGC units, i.e. it satisfies: R(k) = ΔP G (k) According to the actual situation, it is known that the actual total regulating power instruction received by the unit is much lower than the total regulating power demand calculated according to the current AGC control strategy based on the CPS assessment standard, that is, ΔP G (k) = R(k); Secondly, an evaluation model of the difference between the unit response characteristics and the instruction is established; Finally, the AGC performance evaluation method of the unit is optimized, and the AGC response characteristic evaluation and importance sorting method of the unit are proposed combined with the actual response of the unit AGC. The unit in the front of the sorting indicates that it has good regulation performance and can obtain better CPS index for the system when participating in AGC regulation. When the CPS index of the system is monitored to be deteriorated, the unit in the front of the sorting is preferentially considered to be called to improve the CPS performance of the system. The specific implementation is as follows: (1) AGC unit regulation amount optimization and unit importance sorting method When the actual total regulation power instruction is distributed, the regulation power instruction that the unit i can receive will have a range limit, and the regulation range is recorded as: is the lower limit of the adjustment of the unit i, is the upper limit of the adjustment of the unit i, and the actual output increment that the unit i can undertake when participating in the adjustment is: When the unit receives the up command, the unit can adjust the output increment ΔP i UP When the unit receives the down command, the unit can adjust the output increment ΔP i DN When the unit reaches the upper or lower limit of the output, the unit will not be able to be put into AGC operation any more, so the adjustment amount when the unit is called needs to be increased. By calculating the average difference between each unit participating in AGC regulation in history and the target regulation power, a reference benchmark for the regulation amount of the unit is obtained. At the same time, considering the fluctuation degree of the regulation shortage of the unit, the unit with large fluctuation degree needs more regulation amount to smoothly regulate the power grid. A sorting index is constructed, which comprehensively considers the regulation shortage size, fluctuation degree and importance of the unit when participating in regulation. According to the calculated sorting index, the units are sorted, and the units in the front of the ranking are preferentially considered to increase the regulation range and regulation capacity of each regulation; (2) CPS and unit performance correlation analysis and unit calling order sorting method When the unit executes the regulation power instruction, the indexes reflecting the regulation characteristics of the unit include the regulation rate, regulation accuracy and response time. The indexes are quantitatively evaluated, and are aligned with the CPS1 index in the time scale. The regulation indexes are calculated every 15 minutes, and the correlation between the regulation indexes and the CPS is analyzed. According to the influence degree of the change of the regulation characteristics on the CPS1, the weights of different regulation characteristics are set, and the unit evaluation index is constructed by weighting. When the unit is called as AGC regulation, the unit with high importance degree is preferentially considered.

2. The method of claim 1, wherein the method is characterized by, The abnormal value includes that the total regulation power calculated far exceeds the predetermined conventional value and the actual total regulation power instruction value received by the unit is opposite to the total regulation power demand calculated.

3. The method of claim 1, wherein the method further comprises: The evaluation model of the difference between the unit response characteristics and the instruction is established, and the specific implementation is as follows: (1) Influence of the upper and lower limits of the instruction received by the unit According to the historical operation data of the power grid, when the total regulation power is calculated according to the AGC control strategy based on the current CPS evaluation standard, the power distribution mode is distributed to the callable units in turn according to the calling order. Therefore, if the number of callable units is limited at a corresponding moment, the range of the total regulation power instruction L that the unit can receive is: wherein is the lower limit of the regulation of the unit i, is the upper limit of the regulation of the unit i, N AGC is the number of callable units, if the total regulation power requirement ΔP G exceeds the range of the above equation, the corresponding regulation power requirement cannot be allocated completely; (2) Influence of the upper and lower limits of the unit output When the adjustment power command received by the unit i exceeds the upper and lower limits of the output of the unit i, i.e. the adjustment power demand of the unit i cannot be allocated completely; (3) Influence of the regulation rate of the unit The control command cycle of the unit is variable, which is determined by the AGC control cycle and the actual control command of the unit; in each AGC control command cycle, if the unit has responded to the last control command, the current control command is issued; if the unit has not responded to the last control command, the current control command is not issued; there are two criteria for whether the unit has responded to the last control command: ① according to the adjustment increment corresponding to the last control command and the given unit response rate, calculate how much time is needed to respond to the corresponding control command, and the calculated time has passed; ② although the calculated time has not arrived, the actual output of the unit has reached the control target; according to the actual operation data of the unit, when the unit is continuously called, the number of unit adjustments in a period of time is counted, and the control command cycle of the unit can be calculated; the control command cycle of the unit is affected by the unit adjustment rate, when the unit adjustment rate is fast, the last issued control command can be completed within the predetermined time, and the unit can be put into the next AGC adjustment, while when the unit adjustment rate is slow, the unit needs to execute a adjustment instruction in more than the predetermined time, then the unit needs to wait for a period of time before entering the next adjustment; when the number of callable units in a period of time is limited and the control command cycle of the unit is long, it will be difficult to fully allocate the total adjustment power in the corresponding period of time to the unit for execution; Based on (1)-(3), the reasons why the actual total adjustment power instruction received by the unit is less than the total adjustment power demand are as follows: 1) When the number of callable units is limited, the actual total adjustment power instruction received by the unit cannot meet the total adjustment power demand of the system, resulting in that the actual total adjustment power instruction cannot be fully executed; 2) Different types of units have different adjustment upper and lower limits, which affect the response of the unit to the actual total adjustment power instruction; 3) When the unit receives the actual total adjustment power instruction, it is restricted by the upper and lower limits of the unit output, resulting in that the unit cannot fully execute the actual total adjustment power instruction; 4) The adjustment rate of the unit directly affects the control command cycle of the unit, which needs more time to execute an actual total adjustment power instruction, resulting in that the number of actual total adjustment power instructions received in a limited control cycle of the unit is small; 5) When the number of callable units in a period of time is limited and the control command cycle of the unit is long, it is difficult to fully allocate the actual total adjustment power instruction to the unit for execution.

4. The method of claim 1, wherein the method further comprises: The adjustment shortage size, fluctuation degree and importance of the unit when participating in adjustment are comprehensively considered, and the units are sorted according to the calculated sorting index, which is implemented as follows: Let the total adjustment power demand calculated according to the AGC control strategy based on the current CPS evaluation standard be ΔP when the control system issues the kth total adjustment power command G (k), if the total control period has N AGC units participate in adjustment, the actual output increment that these units can bear: If |ΔP G If (k)|>|ΔP(k)|, it means that a regulation deficit occurred in the k-th adjustment. This regulation deficit is distributed to the units called in the k-th adjustment in an average manner. According to the above formula, the adjustment shortage of the unit relative to the adjustment demand when participating in AGC adjustment is calculated, and the unit is under-adjusted when the actual received adjustment power instruction is less than the adjustment power demand; The unit meets the adjustment demand, and the adjustment shortage is 0; in each adjustment, the adjustment shortage of the unit reflects its performance in AGC adjustment, the adjustment shortage values of different units in these periods are calculated, and statistical analysis is performed on these data to calculate the mean and variance of the adjustment shortage of each unit: wherein R i,j is the regulation shortage of the unit i when participating in the jth regulation, and the regulation shortage average μ i represents the average level of all regulation shortages, reflecting the average deviation degree of the unit relative to the power demand that should be regulated, and the regulation shortage variance represents the dispersion degree of the regulation shortage data, and is used to evaluate the fluctuation degree of the regulation shortage of the unit when participating in the AGC regulation; Meanwhile, the participation degree of different units is different, the frequently called units are the key units of AGC adjustment, and the performance of the units should be paid more attention to, therefore, the participation degree of the units is described by the adjustment mileage contributed by the units in a period of time, the total adjustment mileage of the unit i: The proportion of the adjustment mileage of the unit to the total adjustment mileage is taken as the contribution degree index of the unit: The greater the corresponding proportion is, the more the number of times of receiving the adjustment power instruction of the unit is, and the more responsibility the unit bears in the frequency modulation auxiliary service market; In order to comprehensively consider the adjustment shortage mean, variance and contribution degree of the unit, the weighted method is used to obtain the comprehensive performance index by the mean, variance and contribution degree of the adjustment shortage; before weighting, the indexes need to be standardized to ensure that they have similar scales before weighting.

5. The method of claim 4, wherein the method further comprises: The specific way of standardizing each index is: The min-max standardization method is used to normalize each index, and the data comparability is improved; and an importance ranking index λ of the unit group in the capacity optimization time is obtained i The calculation method is as follows: Wherein, α, β, γ are the weighted coefficients of expectation and variance, respectively, and α + β + γ = 1, are the standardized mean, variance and contribution indicators.

6. The method of claim 4, wherein the method further comprises: In step (2), the meanings and calculation methods of each adjustment index are as follows: 1) Adjustment response rate V The adjustment rate reflects the speed of the unit output response to the control command, and the ratio of the actual power adjustment amount of the unit to the used adjustment time is the adjustment rate, and the calculation method is: In the formula, V i,j is the rising or falling adjustment rate of the i th unit for the j th adjustment power instruction; T si,j , T ei,j denote the start and end time of the j th adjustment power instruction, P si,j , P ei,j are the actual output of the unit at the corresponding time, respectively. The total adjustment response rate of the system in the examination period is the total of the adjustment response rates of the units put into AGC, and the average value of the adjustment response rate in the historical adjustment of the AGC unit is used to represent the adjustment response rate of the corresponding unit, and then the calculation method of the total adjustment response rate of the system is: where N AGC is the number of units participating in the regulation during the assessment period, N i is the number of times unit i is called upon. 2) Adjustment accuracy E The adjustment accuracy is the deviation between the actual output and the target output of the unit at the end of a power adjustment instruction, and the calculation method of the adjustment accuracy is E i,j =|C i,j -(P ei,j -P si,j )| E i,j is the adjustment accuracy of the unit i for executing the jth adjustment power instruction; C i,j is the corresponding instruction value; the average adjustment accuracy of all units in a review period represents the overall adjustment accuracy of the unit: 3) Response time T The response time is the time used from the AGC instruction to the actual output of the unit jumping out of the dead zone and the direction being consistent with that of the AGC instruction; the average response time of the online running unit in the examination period is: T i,j Response time for the i-th unit to execute the j-th adjustment power command.

7. The method of claim 6, wherein the method further comprises: In order to determine the key factors affecting the CPS1 index, based on the historical data, the correlation between the adjustment rate, adjustment accuracy, response time of the unit and CPS1 is calculated by the correlation analysis method: first, the AGC unit operation data of the power grid is collected, the CPS1 of each examination period and the adjustment rate, adjustment accuracy, response time of the unit put into AGC adjustment are calculated; further, the single variable correlation analysis is respectively performed on each adjustment performance and CPS1, and the steps of the correlation analysis are as follows: ①The Pearson coefficient is used to describe the correlation between different adjustment performances and CPS, and the calculation formula of the Pearson coefficient of two variables X and Y is: wherein and are the mean values of X and Y, respectively, x i , y i are the i-th values of X and Y, respectively, the value of p(X, Y) ranges from -1 to 1, p > 0 indicates that the two sets of variables are positively correlated, p < 0 indicates that the two sets of variables are negatively correlated, and the larger |p| is, the stronger the correlation is; ②Pearson coefficient is used to describe the linear correlation between unit performance and CPS1. Since the dimension and variation range of CPS1 and unit performance are quite different, conditional probability is used to describe the relationship between unit performance and CPS1. The specific method is as follows: different intervals are divided for different performance indicators after quantization, and the conditional probability distribution of CPS1 greater than a certain value in different intervals is obtained. Assuming that X=(x1, x2,...x n ) is the average of CPS1 every 15 min, and Y=(y1, y2,...y n ) is the regulation performance on the corresponding time scale, when the regulation performance range is [p1, p2], the conditional probability of CPS1 greater than a certain value c is: The conditional probability can reflect the probability trend of CPS1 being greater than a certain value when the unit adjustment performance is in different intervals, and the adjustment characteristics having a significant influence on CPS1 are taken as the basis for the importance ranking of the unit, and the system will obtain better CPS1 value when the unit participates in AGC adjustment.

8. A system for unit AGC performance ranking and key unit discrimination based on unit data mining, characterized in that, The computer program instructions stored in the memory and capable of being executed by the processor can realize the method steps of any one of claims 1-7 when the processor executes the computer program instructions. The computer program instructions stored in the memory and capable of being executed by the processor can realize the method steps of any one of claims 1-7 when the processor executes the computer program instructions.

Images