A fault monitoring system and method for a hydroelectric generating unit cluster

Abstract

This invention relates to the field of generator set monitoring technology, specifically to a fault monitoring system and method for a cluster of hydro-generator units. The method identifies potentially problematic units based on fluctuations in current and output power at the time of analysis; it obtains the ideal load increase power based on the operating power distribution of the problematic units relative to other normal units, and filters out faulty key units by assessing the matching degree of load distribution transfer; it verifies combined fault conditions by combining the load distribution of the problematic units under preliminary screening without key units, thus obtaining the fault monitoring results. This invention verifies unit fault conditions by analyzing the power load decrease of abnormal units in single and combined situations, as well as the matching degree of load distribution with other units, enabling more accurate fault location within a multi-unit cluster.

Description

Technical Field

[0001] This invention relates to the field of generator set monitoring technology, specifically to a fault monitoring system and method for a cluster of hydro-generator sets. Background Technology

[0002] A hydro-turbine generator cluster refers to a configuration in which multiple hydro-turbine generator units operate in parallel within the same hydropower station or hydropower system. Typically, the function of these generator units is to convert the kinetic energy of water flow into electrical energy, thereby increasing the system's total output power and stability. When multiple units operate in parallel, they can be flexibly configured and adjusted according to different water flow conditions, power generation demands, and power system requirements to provide stable power output and maintain the overall stability of the system.

[0003] Because a cluster of hydro-turbine generator units connects multiple units in parallel, and these generators operate in parallel via a common busbar, the connection impedance and busbar impedance create multiple current paths. If a unit experiences an operational fault, its current will be diverted. Even non-faulty units will experience significant fault-induced current. In this case, directly detecting the fault by checking abnormal voltage and current values ​​will lead to errors in locating the faulty unit, thus affecting subsequent fault maintenance of the entire cluster. Summary of the Invention

[0004] To address the technical problem that existing technologies, which directly detect faults by analyzing abnormal voltage and current values ​​of generator units, can lead to errors in locating and judging faulty units, thus affecting subsequent fault maintenance of the entire generator cluster, this invention aims to provide a fault monitoring system and method for a cluster of hydro-generator units. The specific technical solution adopted is as follows:

[0005] This invention provides a fault monitoring method for a cluster of hydro-turbine generator units, the method comprising:

[0006] Acquire the current data and output power of each unit in the hydro-generator cluster;

[0007] Based on the peak fluctuations and output power deviations of each unit in the historical time series current data, select units of interest; based on the output power decline of individual units of interest and the output power distribution of non-units of interest, obtain the ideal load increase power of non-units of interest corresponding to the units of interest.

[0008] For each unit of interest, the load transfer matching anomaly degree of the unit of interest is obtained based on the deviation between the actual output power growth and the ideal load power of all non-units of interest, as well as the matching situation with the output power loss of the unit of interest; and the key units are identified among all units of interest based on the load transfer matching anomaly degree.

[0009] If there are no key units, the fault monitoring results are determined based on the power load transfer and allocation verification of the non-key units corresponding to the combined key units.

[0010] Furthermore, the method for acquiring the units of interest includes:

[0011] For any given unit, determine the peak period of the current data of that unit in the historical time series;

[0012] During each peak period, the difference between the maximum current data and the expected current data is used as the degree of change; the time difference between the time corresponding to the maximum current data and the initial time of the peak period is used as the time consumption; the ratio between the degree of change and the time consumption of each peak period is used as the fluctuation index of each peak period; and the average of the fluctuation indices of all peak periods is used as the current fault significance of the unit.

[0013] The difference between the unit's expected power and its current output power is used as the unit's power deviation index; combined with the unit's current fault significance and power deviation index, the unit's attention level is determined.

[0014] Units with a level of attention exceeding a preset attention threshold will be designated as units under attention.

[0015] Furthermore, the method for obtaining the ideal load increase power includes:

[0016] For any unit of interest, the difference between the current output power and the rated power of the unit of interest is taken as the power loss of the unit of interest.

[0017] Each non-focused unit is sequentially designated as an analysis unit. The ratio of the maximum output power of the analysis unit to the maximum power of the unit cluster is used as the power ratio coefficient of the analysis unit. The product of the current margin load of the analysis unit and the power ratio coefficient is used as the load allocation coefficient of the analysis unit.

[0018] The ratio of the load allocation factor of the analyzed unit to the sum of the load allocation factors of all non-focused units is used as the load allocation weight of the analyzed unit.

[0019] The product of the load allocation weight of the analyzed unit and the power loss degree is taken as the ideal load increase power of the analyzed unit corresponding to the unit of interest.

[0020] Furthermore, the method for obtaining the load transfer matching anomaly degree includes:

[0021] For any unit of interest, obtain the actual increase in the current output power of each non-unit of interest corresponding to the unit of interest; calculate the power difference between the actual increase in the output power of each non-unit of interest corresponding to the unit of interest and the ideal load increase power, and use the sum of all power differences as the normal theoretical deviation of the unit of interest.

[0022] The difference between the sum of the actual growth of all non-focused units corresponding to the focus unit and the power loss of the focus unit is taken as the allocation deviation of the focus unit.

[0023] By combining the normal theoretical deviation and the distribution deviation of the unit of interest, the load transfer matching anomaly of the unit of interest is obtained.

[0024] Furthermore, the method for acquiring the key generating units includes:

[0025] Units with load transfer matching anomalies less than a preset fault threshold are designated as key units.

[0026] Furthermore, if there are no key generating units, the fault monitoring results are determined based on the power load transfer and allocation verification of the non-key generating units corresponding to the combined key generating units, including:

[0027] Obtain different combination schemes for the units of interest;

[0028] When there exists a combination scheme that satisfies the same amount of output power increase of all non-focused units under the combination scheme as the same amount of output power decrease of the focus units under the combination scheme, and satisfies the same amount of output power increase of each non-focused unit as the same amount of ideal load increase of the focus units under the combination scheme, the focus units of the combination scheme are designated as key units.

[0029] When a key unit exists, the fault detection result is recorded as a fault exists.

[0030] Furthermore, the method for obtaining the peak time period includes:

[0031] Obtain the peak point of the historical time-series current data for this unit;

[0032] For any peak point, the time interval of the peak point is defined as the consecutive times with positive slopes before the peak point; and the time interval of the peak point is defined as the consecutive times with negative slopes after the peak point.

[0033] The time period consisting of the time corresponding to the peak point and all other time periods is taken as the peak time period of the peak point.

[0034] Furthermore, the margin load is the difference between the maximum output power of the analysis unit and the current output power.

[0035] Furthermore, the different combinations of the units of interest include:

[0036] Select two or more non-overlapping units from all units of interest, and treat each selection as a combination scheme.

[0037] The present invention also provides a fault monitoring system for a cluster of hydro-generator units, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0038] The present invention has the following beneficial effects:

[0039] This invention identifies potential monitoring units by analyzing fluctuations in current and output power of generating units at a given time. After screening for possible abnormal units, it considers load distribution and transfer, and obtains the ideal load increase power based on the operating power distribution of the monitoring units relative to other normal units. The matching degree of load distribution and transfer further refines the screening of key units with faults. After analyzing a single fault, it considers the possibility of shared load transfer among multiple faulty units. By verifying the combined fault situation through load transfer distribution of combined monitoring units under initial screening without key units, it obtains more accurate and reliable fault monitoring results. This invention verifies unit fault conditions by analyzing the power load decrease of abnormal units in single and combined cases and the matching degree of load distribution with other units, enabling more accurate fault location within multi-unit groups. Attached Figure Description

[0040] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 A flowchart illustrating a fault monitoring method for a cluster of hydro-generator units according to an embodiment of the present invention;

[0042] Figure 2 This is a schematic diagram of a cluster distribution of hydro-generator units provided in one embodiment of the present invention. Detailed Implementation

[0043] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a fault monitoring system and method for a hydro-generator cluster proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0045] The following description, in conjunction with the accompanying drawings, details the specific scheme of the fault monitoring system and method for a cluster of hydro-generator units provided by the present invention.

[0046] Please see Figure 1 The diagram illustrates a fault monitoring method for a cluster of hydro-generator units according to an embodiment of the present invention. The method includes the following steps:

[0047] S1: Obtain the current data and output power of each unit in the hydro-generator cluster.

[0048] A hydro-turbine generator cluster typically consists of multiple identical or different types of hydro-turbine generator units, with each turbine's corresponding generator connected in parallel to output current. The primary function of a hydro-turbine generator cluster is to generate electricity to meet power loads. It provides a stable power output through the parallel connection of multiple turbines. (See also: [link to relevant documentation]). Figure 2 The diagram illustrates a cluster distribution of hydro-generator units according to an embodiment of the present invention.

[0049] Because the hydro-generator cluster meets the corresponding power output requirements by connecting branch lines in parallel, and considering the current distribution characteristics of parallel connection, if a unit fails, other non-faulty units will also experience current fluctuations. The fault current is not concentrated in the faulty unit, but exists in some of the non-faulty units, causing the fault detection to be misjudged as other units.

[0050] In this embodiment, to more accurately locate potentially faulty generating units, real-time current data and output power of different units are monitored. In typical hydro-generator clusters, there are corresponding fault load power redistribution strategies. Therefore, monitoring the real-time output power load transfer of different units facilitates subsequent analysis of the operating status and load distribution matching of multiple units to locate the faulty unit. It should be noted that data acquisition is a technique well-known to those skilled in the art, achieved by installing corresponding sensors. The specific installation and acquisition process is not limited or described here.

[0051] S2: Select units of interest based on the peak fluctuations of current data and the degree of deviation of output power in the historical time series of each unit; obtain the ideal load increase power of non-interested units corresponding to the units of interest based on the degree of output power decline of individual units of interest and the output power distribution of non-interested units.

[0052] Since the fault characteristics of the faulty unit are not obvious due to current shunting, the possible turbines of concern are initially identified by combining the current data and output power of different turbines. Then, based on the operating load distribution of other units under the concern unit, a preliminary load distribution analysis is conducted.

[0053] Hydroelectric generator sets are generally driven by water flow to rotate a turbine, which in turn drives a generator via a linkage shaft, and in conjunction with the excitation system, outputs current. However, during operation, short circuits or grounding faults can cause a surge in fault current. At the same time, the current sharing effect caused by parallel operation will distribute the fault current to other normal units, resulting in significant current surges in multiple units.

[0054] Therefore, in this embodiment of the invention, the units of interest that may fail are first determined by the current data and output power of different units. The method for obtaining the units of interest includes:

[0055] First, for any given generator unit, determine the peak period of its historical current data. The fault current of a generator unit refers to the sharp increase in operating current caused by a fault in the generator unit or power system, such as a short circuit or grounding at the fault point. Power systems are typically designed with protection systems that trigger when the current reaches a certain value, quickly disconnecting the faulty circuit and thus limiting the duration of the fault current.

[0056] Therefore, the fault current generally exhibits a sharp peak. In this embodiment of the invention, the peak point of the current data in the historical time series of the unit is obtained. For any peak point, the time period before the peak point with a continuous positive slope is taken as the time period of the peak point, and the time period after the peak point with a continuous negative slope is taken as the time period of the peak point. The entire peak change period is obtained by continuously increasing the time periods before the peak point and continuously decreasing the time periods after the peak point. The time period consisting of the time corresponding to the peak point and all the time periods is taken as the peak time period of the peak point.

[0057] It should be noted that the acquisition of peak value and slope is a well-known technical method familiar to those skilled in the art. Peak value acquisition can be achieved using peak detection algorithms, etc., and will not be limited or elaborated here.

[0058] Furthermore, during each peak period, the difference between the maximum current data and the expected current data is used as the degree of abrupt change; a larger degree of abrupt change indicates a more drastic change in the operating current. The time difference between the moment corresponding to the maximum current data and the initial moment of the peak period is used as the time duration; a smaller time duration indicates a shorter time spent responding to this abrupt change.

[0059] Therefore, the ratio between the degree of change and the time consumption of each peak period is used as the fluctuation index of each peak period. The larger the fluctuation index, the sharper the corresponding current fluctuation, and the more likely the peak current is a fault current. Combined with the average of the fluctuation index of all peak periods in the time series, it is used as the current fault significance of the unit. The larger the current fault significance, the more likely the unit is to have a fault current.

[0060] However, the presence of fault current does not necessarily indicate a fault in the unit. The shunt effect can cause fault current to appear in normally functioning units as well, thus requiring further assessment. Generally, the operation of a faulty unit will be affected, leading to a decrease in its output power. Conversely, units receiving shunt current are less affected in terms of output power due to the presence of protection devices. Therefore, further analysis should be conducted in conjunction with power deviation.

[0061] Furthermore, the difference between the expected power and the current output power of the generator set is used as the power deviation index. A larger power deviation index indicates a greater decrease in the real-time output power and a more significant power drop. It is understood that the expected power is the output power of the generator set under normal conditions. In this embodiment of the invention, considering that the output power increases with the rotor speed of the motor, and the two are positively correlated, a motor speed-output power model is established based on the motor's own speed parameters. The generator set's electronic speed is input into the model to obtain the expected power. In other embodiments of the invention, the expected power can be set according to the specific implementation scenario requirements, and no restrictions are imposed here.

[0062] By combining the current fault significance and power deviation index of the unit, the attention level of the unit is obtained. In this embodiment of the invention, the product of the current fault significance and power deviation index of the unit is used as the attention level of the unit. The higher the attention level, the higher the probability that the unit is the fault-causing unit.

[0063] After normalizing the attention levels, units that may experience malfunctions are marked. Units with attention levels exceeding a preset attention threshold are designated as units of concern. In this embodiment, the preset attention threshold is set to 0.85, but the specific value can be adjusted by the implementer and is not limited here. It should be noted that normalization is a well-known technique among those skilled in the art, and the choice of normalization can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0064] The focus on the unit is a preliminary screening for anomalies based on the different operating states of the units. However, the operating state of a hydro-generator unit is related to its load-bearing capacity. It is not possible to accurately determine whether there is a fault based solely on the abnormality of the operating state. For example, the measurement signals may be affected by high-frequency interference during the acquisition and transmission process, resulting in misjudgment or missed detection. Therefore, it is also necessary to conduct a comprehensive evaluation by combining the functional output results of the hydro-generator unit.

[0065] If a generator unit fails, its power generation capacity will be limited, making it unable to handle the corresponding load and causing a decrease in its output power. However, due to the parallel connection of different generator units, the normal generator units will maintain normal load output. To maintain output power stability, different generator units within the cluster employ corresponding load sharing strategies for the failed unit. Under this strategy, the power lost by the failed unit is transferred to other normal generator units, appropriately increasing the output power of the normal units to fill the gap. Therefore, this can be verified based on the load transfer situation among different generator units within the cluster.

[0066] Based on this characteristic, functional testing of the fault conditions of the units of interest can be performed. First, it is necessary to analyze the load that should be allocated to each normal unit. Preferably, in this embodiment of the invention, the method for obtaining the ideal load increase power includes:

[0067] Firstly, for any unit of interest, the difference between the current output power and the rated power of the unit of interest is taken as the power loss of the unit of interest. Generally, the output power of a unit that actually fails will decrease to a relatively high degree.

[0068] Each non-focused unit is then analyzed as a separate unit, and the analysis is performed when the focus unit fails. The general unit load allocation strategy allocates load based on the maximum output capacity of different units and the real-time margin.

[0069] Therefore, the ratio of the maximum output power of the analysis unit to the maximum power of the unit cluster is used as the power ratio coefficient of the analysis unit, and the product of the current margin load of the analysis unit and the power ratio coefficient is used as the load allocation coefficient of the analysis unit. In this embodiment of the invention, the margin load is the difference between the maximum output power of the analysis unit and its current output power. Based on the cluster's load allocation strategy, the proportion that normal units should bear when receiving power transferred from faulty units is calculated using the maximum power. This is used to subsequently determine the amount of transferred power that each normal unit should theoretically receive. The larger the load allocation coefficient, the stronger the load-bearing capacity of this analysis unit.

[0070] The load allocation coefficient of the analyzed unit is further used as the ratio between the load allocation coefficients of all non-focused units and the sum of their values ​​as the load allocation weight of the analyzed unit. This weight is then normalized using the overall allocation data. The larger the weight, the greater the loss load allocated to the analyzed unit. Finally, the theoretical increase in load weight is obtained by combining this with the output power loss of the focus units.

[0071] Finally, the product of the load allocation weight and the power loss degree of the analyzed unit is used as the ideal load increase power of the analyzed unit corresponding to the unit of interest. Based on the power loss degree of the unit of interest and the load allocation coefficient of each normal unit, the amount of transferred power that each normal unit should theoretically receive from the faulty unit is obtained, which serves as a benchmark for comparison and verification with the actual transferred power.

[0072] S3: For each unit of interest, based on the deviation between the actual output power growth and the ideal load power of all non-interested units, and the matching situation of the output power loss of the units of interest, the load transfer matching anomaly degree of the units of interest is obtained; based on the load transfer matching anomaly degree, key units are identified among all units of interest.

[0073] For any monitored generating unit, assuming it experiences a fault, its lost output power will be transferred to other units. That is, if at a given moment, the actual increase in load for a normal generating unit equals its allocated load, and the sum of the increases in load for all normal generating units equals the total load loss of that unit, then a fault has occurred. Otherwise, no fault has occurred. By monitoring and obtaining the actual increase in load power for each normal generating unit, the difference between the actual increase and the theoretical increase can be compared to further verify and determine the faulty generating unit.

[0074] Therefore, by comparing the actual load increase received by other normal units under the unit of concern with the theoretically calculated load increase, and by matching the power loss of the unit of concern with the actual total received power of normal units, the degree of anomaly in the actual load transfer matching is comprehensively evaluated. Preferably, in this embodiment of the invention, the method for obtaining the degree of load transfer matching anomaly includes:

[0075] First, for any unit of interest, the actual increase in the current output power of each non-unit of interest corresponding to that unit of interest is obtained to reflect the actual load change. In this embodiment of the invention, the actual increase can be obtained by the difference between the output power at the current moment and the output power during off-peak periods.

[0076] After calculating the power difference between the actual increase and the ideal load increase power for each non-focused unit corresponding to the focus unit, the sum of all power differences is taken as the normal theoretical deviation of the focus unit. The normal theoretical deviation characterizes the difference between the actual load increase and the theoretical amount for each normal unit. The smaller the value, the smaller the difference and the more reasonable the allocation.

[0077] Furthermore, the difference between the sum of the actual growth of all non-units corresponding to the unit of concern and the power loss of the unit of concern is used as the allocation deviation of the unit of concern. The sum of the actual growth reflects the actual total load growth of all normal units, and the allocation deviation characterizes the difference between the total load loss and the total allocation. The smaller the value, the more balanced the load loss and allocation of the faulty unit is under this allocation strategy, reflecting that the corresponding unit of concern is more likely to be a faulty unit.

[0078] Therefore, by combining the normal theoretical deviation and the allocation deviation of the unit of concern, the load transfer matching anomaly of the unit of concern is obtained. In this embodiment of the invention, the product of the normal theoretical deviation and the allocation deviation is taken as the load transfer matching anomaly of the unit of concern. The smaller the load transfer matching anomaly, the more reasonable the matching is, and the higher the possibility that the unit of concern is a faulty unit.

[0079] Therefore, key units are marked by load transfer matching anomaly. In this embodiment of the invention, units with load transfer matching anomaly less than a preset fault threshold are designated as key units. Key units are potential faulty units. The preset fault threshold can be set to 0.2. The specific value can be adjusted by the implementer according to the specific implementation situation, and no restriction is imposed here.

[0080] S4: If there are no key units, determine the fault monitoring results based on the power load transfer and allocation verification of the non-key units corresponding to the combined key units.

[0081] By identifying key units, verification of a single faulty unit within the unit cluster is completed. If, after verifying any key unit, no condition meeting the aforementioned threshold is found, it is possible that multiple units have failed. If multiple units fail, the load distribution of different faulty units will be superimposed under the corresponding allocation strategy. In this case, each normal unit will bear the load distribution from all the faulty units.

[0082] Therefore, in step S3 where no key units are identified, allocation verification is performed based on the combined units of interest to determine the possibility of a fault. In this embodiment of the invention, different combination schemes of units of interest are obtained, that is, two or more non-overlapping units of interest are selected from all units of interest, and each selection is considered as a combination scheme. For example, if there are three units of interest, A, B, and C, then the combination schemes include A+B, A+C, B+C, and A+B+C.

[0083] When a combination scheme exists that satisfies the same amount of output power increase for all non-focused units as the same amount of output power decrease for the focus units, and also satisfies the same amount of output power increase for each non-focused unit as the same amount of ideal load increase for the focus units, the fault detection result is recorded as a fault exists. This is equivalent to making a combined fault hypothesis based on the combined allocation. As an example, the expression for verifying the condition under a certain combination scheme is:

[0084] In the formula, Represented as the first combination scheme The actual increase in the output power of non-focused units. Represented as the first combination scheme The unit under attention corresponds to the first The ideal load increase power for each non-focused unit, and the actual load increase for each non-focused unit. It should satisfy the sum of the loads allocated to it by multiple faulty units. This represents the actual increase in output power of all non-focused units under the combined scheme. Represented as the first combination scheme The power loss of the unit of concern is relative to the total actual load transfer. It should satisfy the sum of the load loss transfer amounts of all faulty units.

[0085] If a combination meets the above conditions, it indicates that multiple monitoring units under this combination are faulty units, and therefore, the monitoring units under this combination are designated as key units. Monitoring times in which key units exist can be recorded as times of fault occurrence. If no verification conditions are met, the monitoring results at other times are used to recalculate the possible key monitoring units and re-verify them.

[0086] In this embodiment of the invention, during the monitoring of faulty units in a cluster of hydro-turbine generator units, if a fault is detected, the key unit is marked as the faulty unit. During maintenance, the faulty unit needs to be quickly disconnected from the grid while the remaining normal units continue to operate. The fault source is accurately located by relevant maintenance personnel to determine whether it is an electrical fault (stator / rotor problem), a mechanical fault (bearing, blade, gear), or a sensor monitoring fault. This saves maintenance costs while improving the accuracy and reliability of turbine fault diagnosis.

[0087] In summary, this invention identifies potential monitoring units by analyzing fluctuations in current and output power at a given time. After screening for possible abnormal units, it considers load distribution and transfer, and obtains the ideal load increase power based on the operating power distribution of the monitoring units relative to other normal units. The matching degree of load distribution and transfer further refines the screening of key units with faults. After analyzing a single fault, it considers the possibility of shared load transfer among multiple faulty units. By verifying the combined fault situation through load transfer distribution of combined monitoring units under initial screening without key units, it obtains more accurate and reliable fault monitoring results. This invention uses single and combined analysis of abnormal unit power load decline and the matching degree of load distribution with other units to conditionally verify unit faults, enabling more accurate fault location in multi-unit groups.

[0088] The present invention also provides a fault monitoring system for a cluster of hydro-generator units, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0089] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0090] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for failure monitoring of a cluster of hydroelectric generating units, characterized by, The method comprises: obtaining current data and output power of each unit in a water turbine generator set cluster; screening a concerned unit according to peak fluctuation of current data and output power deviation of each unit on historical time sequence, obtaining ideal load growth power of a concerned unit corresponding to a non-concerned unit according to output power drop of a single concerned unit and output power distribution of a non-concerned unit; for each concerned unit, obtaining load transfer matching abnormality of the concerned unit according to deviation of actual output power growth and ideal load growth power of all non-concerned units and matching condition of output power loss of the concerned unit, determining a key unit among all concerned units based on the load transfer matching abnormality; if there is no key unit, determining a fault monitoring result according to power load transfer distribution verification of non-concerned units corresponding to combined concerned units; the method for obtaining the ideal load growth power comprises: for any one concerned unit, taking difference between current output power and rated power of the concerned unit as power loss degree of the concerned unit; sequentially taking each non-concerned unit as an analysis unit, taking ratio of maximum output power of the analysis unit to maximum power of the unit cluster as power proportion coefficient of the analysis unit, and taking product of current margin load of the analysis unit and the power proportion coefficient as load distribution coefficient of the analysis unit; taking ratio between the load distribution coefficient of the analysis unit and sum of load distribution coefficients of all non-concerned units as load distribution weight of the analysis unit; taking product of the load distribution weight of the analysis unit and the power loss degree as ideal load growth power of the non-concerned unit corresponding to the concerned unit; the method for obtaining the load transfer matching abnormality comprises: for any one concerned unit, obtaining actual growth amount of current output power of each non-concerned unit corresponding to the concerned unit, calculating power difference between the actual growth amount and the ideal load growth power of each non-concerned unit corresponding to the concerned unit, and taking sum of all power differences as normal theoretical deviation degree of the concerned unit; taking difference between sum of actual growth amounts of all non-concerned units corresponding to the concerned unit and the power loss degree of the concerned unit as distribution deviation degree of the concerned unit; combining the normal theoretical deviation degree and the distribution deviation degree of the concerned unit, obtaining load transfer matching abnormality of the concerned unit.

2. The method of claim 1, wherein, the method for obtaining the concerned unit comprises: for any one unit, determining peak period of current data of the unit on historical time sequence; on each peak period, taking difference between maximum current data and expected current data as mutation degree, taking time difference between time corresponding to the maximum current data and initial time of the peak period as time consumption degree, taking ratio between the mutation degree and the time consumption degree of each peak period as fluctuation index of each peak period, and taking average value of fluctuation indexes of all peak periods as current fault salience of the unit; taking difference between expected power of the unit and output power at current time as power deviation index of the unit, and obtaining attention degree of the unit by combining the current fault salience and the power deviation index of the unit. The unit with the attention degree greater than the preset attention threshold is taken as the attention unit.

3. The method of claim 1, wherein, The method for obtaining the key unit comprises: The unit with the load transfer matching abnormality less than the preset fault threshold is taken as the key unit.

4. The method of claim 1, wherein, If there is no key unit, the fault monitoring result is determined according to the power load transfer distribution verification of the non-attention unit corresponding to the combined attention unit, comprising: Different combination schemes of the attention unit are obtained. When there is a combination scheme, the output power increase of all non-attention units under the combination scheme is consistent with the output power decrease of the attention unit under the combination scheme, and the output power increase of each non-attention unit is consistent with the ideal load increase power of the attention unit corresponding to the combination scheme, the attention unit of the combination scheme is taken as the key unit. When there is a key unit, the fault detection result is recorded as existing fault.

5. The method of claim 2, wherein, The method for obtaining the peak period comprises: The peak point of the current data on the historical time sequence of the unit is obtained. For any peak point, the time point of the period of the peak point is taken as the time point of the period of the peak point, and the time point of the period of the peak point is taken as the time point of the period of the peak point. The period composed of the time point corresponding to the peak point and all period time points is taken as the peak period of the peak point.

6. The method of claim 1, wherein, The margin load is the difference between the maximum output power of the analysis unit and the current output power.

7. The method of claim 1, wherein, Different combination schemes of the attention unit comprise: Two or more non-repeated attention units are selected from all attention units, and each selection is taken as a combination scheme.

8. A fault monitoring system for a cluster of hydroelectric generating units, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, The processor executes the computer program to realize the steps of the fault monitoring method of the hydro-generator unit cluster according to any one of claims 1-7.

Images