A power generation terminal equipment operation and maintenance system and method based on a large model

Abstract

This invention relates to the technical field of operation and maintenance of power generation terminal equipment, and particularly to a system and method for operation and maintenance of power generation terminal equipment based on a large model. This invention synchronously acquires real-time data from power generation terminal equipment through a sliding window mechanism, making the acquired real-time data more effective. It constructs a decision tree by training the historical operating parameters and historical fault types of the power generation terminal equipment during its historical operation, effectively building the decision tree. Furthermore, the decision tree, based on real-time data, can more accurately output fault types. The operation and maintenance status is determined based on the matching degree between the output fault type and the actual fault type, and adjustment parameters are determined based on the operation and maintenance status. This invention strengthens the collaborative analysis of multi-source data and improves cross-modal correlation capabilities by adjusting parameters based on the operation and maintenance status, thereby improving the accuracy of power generation equipment operation and maintenance.

Description

Technical Field

[0001] This invention relates to the technical field of operation and maintenance of power generation terminal equipment, and in particular to a power generation terminal equipment operation and maintenance system and method based on a large model. Background Technology

[0002] The operation and maintenance (O&M) technology of power generation terminal equipment has undergone a transformation from manual inspection to intelligent O&M. Early methods relied primarily on periodic manual inspections, paper records, and simple instrument checks, resulting in low efficiency and a high risk of missed inspections. With the application of sensor technology and SCADA systems, real-time monitoring and data acquisition of equipment status became possible. In recent years, by combining IoT platforms, edge computing, and digital twin technology, a closed-loop system of status perception, data analysis, and fault early warning has been constructed, and non-contact detection methods such as drone inspections and infrared thermal imaging have been introduced. Current technological trends center on AI-driven predictive maintenance, using machine learning to analyze historical data to predict faults. Combined with AR remote collaboration and robotic automated inspections, this forms a collaborative "cloud-edge-device" intelligent O&M model, significantly improving equipment reliability and O&M economics.

[0003] Chinese Patent Publication No. CN110112771 B discloses a method and apparatus for detecting the operating status of a generator in a nuclear power plant. The method determines the generator set's status based on the on / off states of the generator output switch and the main transformer high-voltage side switch. When the generator is operating normally, the generator set is in a disconnected state, and the generator's line current is greater than the current under minimum plant auxiliary load, the method determines that the generator is in an islanded operating state. Compared with existing technologies, this invention accurately determines whether the generator is in an islanded operating state through the generator output switch, the main transformer high-voltage side switch, and the generator's line current status.

[0004] It is evident that the existing technology has the following problems: the existing technology analyzes multi-source data in isolation and lacks cross-modal correlation capabilities, resulting in a low accuracy rate in the operation and maintenance of power generation equipment. Summary of the Invention

[0005] To address this issue, the present invention provides a power generation terminal equipment operation and maintenance system and method based on a large model, which overcomes the problem of low accuracy in the operation and maintenance of power generation equipment due to isolated analysis of multi-source data and lack of cross-modal correlation capabilities in the prior art.

[0006] To achieve the above objectives, the present invention provides a power generation terminal equipment operation and maintenance system based on a large model, comprising:

[0007] The real-time data acquisition unit is used to synchronously acquire infrared thermal images and vibration phase signals generated by several sensor monitoring points set on the power generation terminal equipment through a sliding window mechanism, and to collect electrical operating parameters. All data are time-aligned according to the set sliding window length.

[0008] The historical data acquisition unit is used to acquire historical operating parameters and historical fault types of the power generation terminal equipment during historical operation. The historical operating parameters include infrared thermal images, vibration phase signals, and electrical operating parameters.

[0009] The model training unit is connected to the historical data acquisition unit and is used to train a decision tree based on the data acquired by the historical data acquisition unit. The training parameters that need to be set when building the decision tree include the maximum depth when training the constraint tree structure, the minimum number of samples in the leaf nodes, the number of training iterations of the decision tree, and the initial learning rate.

[0010] The model output unit is connected to the real-time data acquisition unit and the historical data acquisition unit respectively. It inputs the data acquired by the real-time data acquisition unit into the decision tree and outputs the fault type of the power generation terminal equipment during abnormal operation.

[0011] An analysis unit, connected to the model output unit, is used to determine the operation and maintenance status based on the matching degree between the fault type output by the model output unit and the actual fault type within a preset time period. It also adjusts the maximum depth of the training constraint tree structure in the training decision tree based on the operation and maintenance status, and adjusts the number of training iterations of the decision tree when the number of adjustments equals the preset number and the operation and maintenance status is still unqualified.

[0012] Furthermore, the analysis unit is also used to plot a matching degree-preset time period curve based on the matching degree acquired within multiple preset time periods. The above steps are performed when the operation and maintenance state is determined to be a first state, where the matching degree is less than a first preset matching degree and greater than a second preset matching degree. The analysis unit is also used to determine whether to adjust the maximum depth when training the constraint tree structure based on the average of the absolute values ​​of the slopes corresponding to several nodes of the curve. The analysis unit is also used to increase the maximum depth when training the constraint tree structure based on the ratio of the sample size acquired within the preset time period to the preset sample size, and the increase in the maximum depth when training the constraint tree structure is proportional to the ratio.

[0013] Furthermore, the analysis unit is also used to increase the maximum depth based on the minimum number of samples at the leaf node, and the increase in the maximum depth is proportional to the minimum number of samples at the leaf node.

[0014] Furthermore, the analysis unit is also used to repeatedly adjust the maximum depth of the training constraint tree structure at least once until the number of adjustments reaches a preset number or the number of adjustments is less than a preset number, at which point the operation and inspection state is qualified. This involves obtaining the average value of the absolute values ​​of the slopes of the adjusted matching degree - preset time period curve, and performing repeated adjustments when the average value is less than or equal to a preset average value. The criterion for a qualified operation and inspection state is that the matching degree is greater than or equal to the first preset matching degree. The analysis unit is also used to obtain the optimized matching degree when the number of adjustments equals the preset number, and to determine the adjusted operation and inspection state based on the optimized matching degree. The analysis unit is also used to adjust the number of training iterations of the decision input based on the optimized matching degree, wherein the adjustment is performed when the adjusted operation and inspection state is a first state.

[0015] Furthermore, the analysis unit is also used to increase the number of training iterations of the decision tree based on the ratio of the optimized matching degree to the first preset matching degree, and the increase in the number of training iterations of the decision tree is inversely proportional to the ratio.

[0016] Furthermore, the analysis unit is also used to reduce the initial learning rate based on the number of training iterations of the decision tree, and the reduction in the initial learning rate is proportional to the number of training iterations.

[0017] Furthermore, the analysis unit is also used to calculate the average value of the absolute value of the slope of the matching degree-preset time period curve plotted based on the matching degree re-detected after the above adjustment steps, so as to obtain the optimized average value; the analysis unit is also used to plot the rotor current-time curve based on the rotor current of the generator in each preset time period acquired by the real-time data acquisition unit, wherein the above steps are performed when the operation and maintenance state is determined to be the second state or the optimized average value is less than or equal to the preset average value, the second state being that the matching degree is less than or equal to the second preset matching degree; the analysis unit is also used to reduce the sliding window length based on the variance of the absolute value of the slope of the rotor current-time curve, and the reduction in the sliding window length is proportional to the variance.

[0018] Furthermore, the analysis unit is also used to reduce the sliding step size based on the sliding window length, and the reduction in the sliding step size is inversely proportional to the sliding window length.

[0019] Furthermore, the analysis unit is also used to increase the range reference of the infrared sensor based on the sliding window length, and the increase in the range reference of the infrared sensor is inversely proportional to the sliding window length.

[0020] To achieve the above objectives, the present invention provides a method for operation and maintenance of power generation terminal equipment based on a large model, comprising:

[0021] Through a sliding window mechanism, infrared thermal images and vibration phase signals generated by several sensor monitoring points set on the power generation terminal equipment are acquired in real time and synchronously, and electrical operating parameters are collected. All real-time data are time-aligned according to the set sliding window length.

[0022] The historical operating parameters and fault types of the power generation terminal equipment during the historical operation process are obtained. The historical operating parameters include infrared thermal images, vibration phase signals and electrical operating parameters.

[0023] The decision tree is constructed by training based on the data obtained from the historical operating parameters and historical fault types. The training parameters to be set when constructing the decision tree include the maximum depth when training the constraint tree structure, the minimum number of samples in the leaf nodes, the number of training iterations of the decision tree, and the initial learning rate.

[0024] The acquired real-time data is input into the decision tree, and the fault type of the power generation terminal equipment during abnormal operation is output.

[0025] The operation and maintenance status is determined based on the matching degree between the fault type output within a preset time period and the actual fault type. The maximum depth of the training constraint tree structure in the training decision tree is adjusted based on the operation and maintenance status. The number of training iterations of the decision tree is adjusted when the number of adjustments is equal to the preset number and the operation and maintenance status is still unqualified.

[0026] Compared with existing technologies, the advantages of this invention are as follows: This invention synchronously acquires real-time data from power generation terminal equipment through a sliding window mechanism, making the acquired real-time data more effective; and by constructing a decision tree based on the historical operating parameters and historical fault types of the power generation terminal equipment during its historical operation, it can effectively build the decision tree, which, based on real-time data, can more accurately output the fault type; furthermore, it determines the operation and maintenance status based on the matching degree between the output fault type and the actual fault type, and determines the adjustment parameters based on the operation and maintenance status. This invention enhances cross-modal correlation capabilities through collaborative analysis of multi-source data using a model, thereby improving the accuracy of power generation equipment operation and maintenance.

[0027] Furthermore, this invention determines the fluctuation of matching degree by averaging the absolute values ​​of the slopes of the matching degree-preset time period curve, and determines the cause of the non-compliance of the operation and maintenance status based on the fluctuation of matching degree to determine whether to adjust the maximum depth when training the constraint tree structure. This allows for more accurate adjustment of the maximum depth when training the constraint tree structure, thereby more accurate data prediction and further improving the accuracy of power generation equipment operation and maintenance.

[0028] Furthermore, by adjusting the maximum depth of the training constraint tree structure by the minimum number of leaf nodes, the present invention can prevent the tree from becoming too refined and noise, and retain the ability to distinguish rare faults when there are few critical fault samples, thereby further improving the accuracy of equipment operation and maintenance.

[0029] Furthermore, this invention determines whether to adjust the training iteration coefficients of the decision tree by repeatedly adjusting the operation and maintenance status. This can solve the problem of unqualified operation and maintenance status caused by underfitting of the gradient boosting tree in the decision tree after training based on the original number of training iterations, thereby further improving the efficiency of operation and maintenance of power generation terminal equipment.

[0030] Furthermore, by optimizing the ratio of the matching degree to the first preset matching degree, the present invention adjusts the number of training iterations of the decision tree, which makes the prediction results of the model more accurate, thereby further improving the accuracy of monitoring the operation and maintenance status of power generation terminal equipment, and further improving the efficiency of operation and maintenance of power generation terminal equipment.

[0031] Furthermore, by adjusting the initial learning rate through the number of training iterations of the decision tree, this invention can avoid ignoring detailed patterns due to increasing the number of training iterations, making the model's prediction results more accurate and thus further improving the efficiency of operation and maintenance of power generation terminal equipment.

[0032] Furthermore, the present invention determines whether to adjust the sliding window length by re-detecting the average value of the absolute value of the slope of the adjusted curve. By adjusting the sliding window length, there is no time delay when the decision tree detects the fault type, thereby matching the detected fault type with the actual time period, and further improving the accuracy of operation and maintenance of power generation terminal equipment.

[0033] Furthermore, by adjusting the sliding window length and then the sliding step size, the present invention enables more precise capture of key changes in current, resulting in more accurate data acquisition and further improving the prediction results of the decision tree, thereby further improving the accuracy of operation and maintenance of power generation terminal equipment.

[0034] Furthermore, after adjusting the sliding window and sliding step size, the sampling data will become saturated. Therefore, adjusting the range reference of the infrared sensor of the obtained infrared thermal image during high-frequency monitoring can reduce the sampling database, increase more effective data, and make the training and prediction results of the decision tree more accurate. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the operation and maintenance system for power generation terminal equipment based on a large model, according to an embodiment of the present invention.

[0036] Figure 2 This is a flowchart illustrating the steps of the operation and maintenance method for power generation terminal equipment based on a large model, as described in an embodiment of the present invention.

[0037] Figure 3 This is a flowchart illustrating the process of determining the operation and maintenance status as the first state based on the matching degree and the pre-stored preset matching degree in an embodiment of the present invention.

[0038] Figure 4 This is a flowchart illustrating the process of determining the operation and maintenance status as the second state based on the matching degree and the pre-stored preset matching degree in an embodiment of the present invention. Detailed Implementation

[0039] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0040] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0041] It should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0042] Please see Figure 1 As shown, it is a schematic diagram of the operation and maintenance system of the power generation terminal equipment based on a large model according to an embodiment of the present invention.

[0043] The system includes a real-time data acquisition unit, a historical data acquisition unit, a model training unit, a model output unit, and an analysis unit.

[0044] The real-time data acquisition unit is used to synchronously acquire infrared thermal images and vibration phase signals generated by several sensor monitoring points set on the power generation terminal equipment through a sliding window mechanism, and to collect electrical operating parameters. All data are time-aligned according to the set sliding window length.

[0045] The historical data acquisition unit is used to acquire historical operating parameters and historical fault types of the power generation terminal equipment during historical operation. The historical operating parameters include infrared thermal images, vibration phase signals, and electrical operating parameters.

[0046] The model training unit is connected to the historical data acquisition unit and is used to train a decision tree based on the data acquired by the historical data acquisition unit. The training parameters that need to be set when building the decision tree include the maximum depth when training the constraint tree structure, the minimum number of samples in the leaf nodes, the number of training iterations of the decision tree, and the initial learning rate.

[0047] The model output unit is connected to the real-time data acquisition unit and the historical data acquisition unit respectively. It is used to input the data acquired by the real-time data acquisition unit into the decision tree and output the fault type of the power generation terminal equipment during abnormal operation.

[0048] The analysis unit is connected to the model output unit. It is used to determine the operation and maintenance status based on the matching degree between the fault type output by the model output unit and the actual fault type within a preset time period, and to adjust the maximum depth of the training constraint tree structure in the training decision tree based on the operation and maintenance status. When the number of adjustments is equal to the preset number and the operation and maintenance status is still unqualified, the number of training iterations of the decision tree is adjusted.

[0049] Specifically, the core mechanisms of the sliding window include window partitioning and dynamic adjustment, and multi-source data alignment. Window partitioning and dynamic adjustment involve dividing the data stream into fixed-length or variable-length intervals. The window slides forward as new data arrives, ensuring continuous data coverage while avoiding redundant calculations. Adjusting the window size and sliding step directly affects data freshness and processing latency. Multi-source data alignment allows the sliding window to align data from different sources using timestamps or sequence numbers, ensuring temporal consistency within the window. These will not be elaborated further.

[0050] Please see Figure 2 The diagram shown is a flowchart illustrating the steps of the operation and maintenance method for power generation terminal equipment based on a large model according to an embodiment of the present invention.

[0051] The operation and maintenance process of power generation terminal equipment based on a large model is as follows:

[0052] S1, through the real-time data acquisition unit using the sliding window mechanism, synchronously acquires infrared thermal images and vibration phase signals generated by several sensor monitoring points set on the power generation terminal equipment, and collects electrical operating parameters. All data are time-aligned according to the set sliding window length.

[0053] S2, acquires historical operating parameters and historical fault types of the power generation terminal equipment during historical operation through the historical data acquisition unit. The historical operating parameters include infrared thermal images, vibration phase signals and electrical operating parameters.

[0054] S3, a decision tree is constructed by training the model training unit connected to the historical data acquisition unit based on the data acquired by the historical data acquisition unit. The training parameters to be set when constructing the decision tree include the maximum depth when training the constraint tree structure, the minimum number of samples in the leaf nodes, the number of training iterations of the decision tree, and the initial learning rate.

[0055] S4, the model output unit, which is connected to the real-time data acquisition unit and the historical data acquisition unit respectively, inputs the data acquired by the real-time data acquisition unit into the decision tree, and outputs the fault type of the power generation terminal equipment during abnormal operation;

[0056] S5, the analysis unit connected to the model output unit determines the operation and maintenance status based on the matching degree between the fault type output by the model output unit and the actual fault type within a preset time period, and adjusts the maximum depth of the training constraint tree structure in the training decision tree based on the operation and maintenance status, and adjusts the number of training iterations of the decision tree when the number of adjustments is equal to the preset number and the operation and maintenance status is still unqualified.

[0057] Please see Figure 3 The diagram shows a flowchart illustrating the process of determining the operation and maintenance status as the first state based on the matching degree and a pre-stored preset matching degree in an embodiment of the present invention. The analysis unit in this embodiment is further configured to plot a matching degree-preset time period curve based on the matching degrees acquired over multiple preset time periods. The above steps are performed when the operation and maintenance status is determined to be the first state, where the matching degree is less than a first preset matching degree and greater than a second preset matching degree. The analysis unit is also configured to determine whether to adjust the maximum depth of the training constraint tree structure based on the average of the absolute values ​​of the slopes corresponding to several nodes of the curve. Furthermore, the analysis unit is configured to increase the maximum depth of the training constraint tree structure based on the ratio of the sample size acquired within the preset time period to the preset sample size, and the increase in the maximum depth of the training constraint tree structure is proportional to the ratio.

[0058] Specifically, the matching degree L0 can be divided into a first preset matching degree L1 and a second preset matching degree L2. In the matching degree standard, the first preset matching degree L1 = 0.9 (i.e., 90%) and the second preset matching degree L2 = 0.75 (i.e., 75%). It should be noted that in other embodiments, the values ​​of L1 and L2 can also be determined according to the operation and maintenance requirements of the power generation terminal equipment. The comparison process based on the matching degree L with L1 and L2 is as follows:

[0059] If the matching degree L is greater than or equal to the first preset matching degree L1, it indicates that the operation and maintenance status of the power generation terminal equipment is qualified.

[0060] If the matching degree L is less than the first preset matching degree L1 and greater than the second preset matching degree L2, the current operation and maintenance status is determined as the first status;

[0061] If the matching degree L is less than the second preset matching degree L2, the current operation and maintenance status will be determined as the second status.

[0062] Specifically, if the operation and maintenance status is determined to be the first state, the matching degree within multiple preset time periods is obtained, and a matching degree-preset time period curve is plotted. The average value of the absolute values ​​of the slopes corresponding to several nodes of the curve is calculated. If the average value is less than or equal to the preset average value, it indicates that the matching degree fluctuates less. This is because the small sample size means that rare faults are not detected, resulting in underfitting of a single decision tree and thus bias in data prediction. Therefore, the maximum depth of training the constraint tree structure in the training decision tree is adjusted based on the ratio of the obtained sample size to the preset sample size. The preset average value is 1.1%.

[0063] Specifically, if the ratio of the obtained sample size to the preset sample size is Q0 = 0.85, then the comparison process based on the ratio Q and the preset ratio Q0 is as follows:

[0064] If the ratio Q is less than or equal to the preset ratio Q0, the maximum depth when training the constraint tree structure will be adjusted to 1.07 times the original maximum depth.

[0065] If the ratio Q is greater than the preset ratio Q0, the maximum depth when training the constraint tree structure will be adjusted to 1.19 times the original maximum depth.

[0066] Specifically, to prevent the tree from excessively refining noise and to retain the ability to distinguish rare faults when there are few critical fault samples, the maximum depth is adjusted based on the minimum number of leaf nodes. The preset minimum number of leaf nodes R0 = 7. The comparison process between the minimum number of leaf nodes R and the preset minimum number of leaf nodes R0 is as follows:

[0067] If the minimum number of leaf nodes R is less than or equal to the preset minimum number of leaf nodes R0, then the maximum depth when training the constraint tree structure will be adjusted to 1.04 times the value after the sample size adjustment.

[0068] If the minimum number of leaf nodes R is less than or equal to the preset minimum number of leaf nodes R0, then the maximum depth when training the constraint tree structure will be adjusted to 1.21 times the value after the sample size adjustment.

[0069] Specifically, the analysis unit in this embodiment of the invention is further used to repeatedly adjust the maximum depth of the training constraint tree structure at least once until the number of adjustments reaches a preset number or the number of adjustments is less than a preset number, at which point the operation and inspection state is qualified. This involves obtaining the average value of the absolute value of the slope of the adjusted matching degree - preset time period curve, and performing repeated adjustments when the average value is less than or equal to a preset average value. The criterion for a qualified operation and inspection state is that the matching degree is greater than or equal to the first preset matching degree. The analysis unit is also used to obtain the optimized matching degree when the number of adjustments equals the preset number, and to determine the adjusted operation and inspection state based on the optimized matching degree. Furthermore, the analysis unit is used to adjust the number of training iterations of the decision input based on the optimized matching degree, wherein the adjustment is performed when the adjusted operation and inspection state is a first state.

[0070] Specifically, the curve is redrawn based on the adjusted matching degree, and the average of the absolute values ​​of the slope of the curve is calculated. When the average value is less than or equal to the preset average value, the maximum depth is repeatedly adjusted until the number of adjustments reaches the preset number or the number of adjustments is less than the preset number and the operation and inspection status is qualified. At this time, the optimized matching degree when the number of adjustments is equal to the preset number is re-detected. If the operation and inspection status is still the first state based on the optimized matching degree, it indicates that the gradient boosting tree in the decision tree has underfitted after training based on the original number of training iterations, which makes the output result of the decision tree inaccurate. Then, the number of training iterations of the decision tree is adjusted based on the ratio of the optimized matching degree to the first preset matching degree.

[0071] Specifically, if the preset ratio T0 between the optimized matching degree and the first preset matching degree is 0.89, then the comparison process based on the ratio T and the preset ratio T0 is as follows:

[0072] If the ratio T is less than or equal to the preset ratio T0, the training iteration coefficient of the decision tree will be adjusted to 2.3 times the original number of training iterations.

[0073] If the ratio T is greater than the preset ratio T0, the training iteration coefficient of the decision tree will be adjusted to 1.2 times the original number of training iterations.

[0074] Specifically, in the gradient boosting framework, decision trees are added iteratively as weak learners. At this point, an initial learning rate is needed to control the contribution of each tree, which will not be elaborated further.

[0075] Specifically, the preset number of training iterations for the decision tree is U0 = 500, which is the preset total number of weak learners. The comparison process between the number of training iterations U and the preset number of training iterations U0 is as follows:

[0076] If the number of training iterations U is less than or equal to the preset number of training iterations U0, the initial learning rate will be adjusted to 0.94 times the original initial learning rate.

[0077] If the number of training iterations U is greater than the preset number of training iterations U0, the initial learning rate will be adjusted to 0.87 times the original initial learning rate.

[0078] Please see Figure 4 The diagram shows a flowchart illustrating the process of determining the operation and maintenance state as the second state based on the matching degree and a pre-stored preset matching degree in an embodiment of the present invention. The analysis unit in this embodiment is further used to calculate the average value of the absolute value of the slope of the matching degree-preset time period curve plotted after the above adjustment steps, to obtain an optimized average value; the analysis unit is also used to plot a rotor current-time curve based on the generator rotor current within each preset time period acquired by the real-time data acquisition unit, wherein the above steps are performed when the operation and maintenance state is determined to be the second state or the optimized average value is less than or equal to the preset average value, the second state being a matching degree less than or equal to the second preset matching degree; the analysis unit is also used to reduce the sliding window length based on the variance of the absolute value of the rotor current-time curve slope, and the reduction in the sliding window length is proportional to the variance.

[0079] Specifically, the adjusted matching degree is reacquired, and the matching degree-preset time period curve is redrawn based on the matching degree. The average value of the absolute value of the slope of the curve is calculated. If the average value of the absolute value of the slope corresponding to several nodes of the curve is still less than or equal to the preset average value, it indicates that there is a time delay in the fault type detection by the decision tree. That is, it should have been detected in the A time period, but it was detected in the A+1 time period. This will cause a large number of detected fault types to not match the actual time period. Then, the rotor current of the generator in each preset time period is obtained, and the rotor current-time curve is drawn. The sliding window length is reduced based on the variance of the absolute value of the curve slope, thereby solving the problem of detection delay.

[0080] Specifically, if the preset variance V0 = 0.5, the comparison process between the variance V and the preset variance V0 is as follows:

[0081] If the variance V is less than or equal to the preset variance V0, the sliding window length will be adjusted to 0.91 times the original sliding window length.

[0082] If the variance V is greater than the preset variance V0, the sliding window length will be adjusted to 0.79 times the original sliding window length.

[0083] Specifically, after adjusting the sliding window length, the sliding step size also needs to be reduced simultaneously to more accurately capture key changes in the current. Therefore, the sliding step size is adjusted by adjusting the sliding window length, where the sliding step size is always less than the sliding window length. The preset sliding window length W0 = 400ms. The comparison process between the sliding window length W and the preset sliding window length W0 is as follows:

[0084] If the sliding window length W is less than or equal to the preset sliding window length W0, the sliding step size will be adjusted to 0.72 times the original sliding step size.

[0085] If the sliding window length W is greater than the preset sliding window length W0, the sliding step size will be adjusted to 0.87 times the original sliding step size.

[0086] Specifically, after adjusting the sliding window and sliding step size, the sampling data will be saturated. Therefore, in high-frequency monitoring, the range reference of the infrared sensor is based on the infrared thermal image obtained by adjusting the sliding window length, so that the sampling data is more effective.

[0087] Specifically, if the preset sliding window length M0 = 350ms, the comparison process between the sliding window length W and the preset sliding window length M0 is as follows:

[0088] If the sliding window length W is less than or equal to the preset sliding window length M0, then the range reference of the infrared sensor will be adjusted to 1.98 times the original reference.

[0089] If the sliding window length W is greater than the preset sliding window length M0, the range reference of the infrared sensor will be adjusted to 1.57 times the original reference.

[0090] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0091] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A large model-based power generation terminal equipment operation and inspection system, characterized by, The method comprises the following steps: a real-time data acquisition unit is configured to acquire infrared thermal images and vibration phase signals generated by a plurality of sensor monitoring points arranged at a power terminal device in a sliding window mechanism, and to collect electrical operation parameters, wherein all the data are time-aligned according to a set sliding window length; a historical data acquisition unit is configured to acquire historical operation parameters and historical fault types of the power terminal device in a historical operation process, wherein the historical operation parameters include the infrared thermal images, the vibration phase signals and the electrical operation parameters; a model training unit is connected to the historical data acquisition unit and configured to train a decision tree based on the data acquired by the historical data acquisition unit, wherein the training parameters required for constructing the decision tree include a maximum depth when training a constraint tree structure, a minimum number of leaf nodes, a training iteration number of the decision tree and an initial learning rate; a model output unit is connected to the real-time data acquisition unit and the historical data acquisition unit, and is configured to input the data acquired by the real-time data acquisition unit into the decision tree and output a fault type of the power terminal device in an abnormal operation state; an analysis unit is connected to the model output unit and configured to determine an operation inspection state based on a matching degree of the fault type output by the model output unit and an actual fault type in a preset time period, and to adjust the maximum depth when training the constraint tree structure in the training decision tree based on the operation inspection state, and to adjust the training iteration number of the decision tree when the adjustment number is equal to a preset number and the operation inspection state is still unqualified; the analysis unit is further configured to draw a matching degree-pre-set time period curve based on the matching degrees in a plurality of preset time periods, wherein the above steps are performed when it is determined that the operation inspection state is a first state, and the first state is that the matching degree is less than a first preset matching degree and greater than a second preset matching degree; the analysis unit is further configured to determine whether to adjust the maximum depth when training the constraint tree structure based on an average value of absolute values of slopes of a plurality of nodes of the curve; the analysis unit is further configured to increase the maximum depth when training the constraint tree structure based on a ratio of a sample amount acquired in the preset time period to a preset sample amount, and the increase amplitude of the maximum depth when training the constraint tree structure is proportional to the ratio.

2. The large model-based power generation terminal device operation inspection system according to claim 1, characterized by, the analysis unit is further configured to increase the maximum depth when training the constraint tree structure based on the minimum number of leaf nodes, and the increase amplitude of the maximum depth when training the constraint tree structure is proportional to the minimum number of leaf nodes.

3. The large model-based power generation terminal device operation inspection system according to claim 2, characterized by, the analysis unit is further configured to repeatedly adjust the maximum depth when training the constraint tree structure at least once until the operation inspection state is qualified when the adjustment number reaches the preset number or the adjustment number is less than the preset number, and then stop adjusting, wherein an average value of absolute values of slopes of the matching degree-pre-set time period curve after adjustment is acquired, and the repeated adjustment operation is performed when the average value is less than or equal to a preset average value, and the standard for the operation inspection state being qualified is that the matching degree is greater than or equal to the first preset matching degree; the analysis unit is further configured to acquire an optimized matching degree when the adjustment number is equal to the preset number, and to determine an adjusted operation inspection state based on the optimized matching degree. The analysis unit is further configured to adjust the number of training iterations of the decision tree based on the optimized matching degree, wherein the adjusted operating inspection state is the first state.

4. The large model-based power generation terminal device operation inspection system according to claim 3, characterized by, The analysis unit is further configured to increase the number of training iterations of the decision tree based on a ratio of the optimized matching degree to the first preset matching degree, and the increase amplitude of the number of training iterations of the decision tree is inversely proportional to the ratio.

5. The large model-based power generation terminal device operation inspection system according to claim 4, characterized by, The analysis unit is further configured to decrease the initial learning rate based on the number of training iterations of the decision tree, and the decrease amplitude of the initial learning rate is proportional to the number of training iterations.

6. The large model-based power generation terminal device operation inspection system according to claim 5, characterized by, The analysis unit is further configured to calculate an average value of absolute values of slopes of a matching degree-preset time period curve drawn based on matching degrees re-detected after adjusting the initial learning rate, to obtain an optimized average value. The analysis unit is further configured to draw a rotor current-time curve based on rotor currents of the generator in each preset time period obtained by the real-time data acquisition unit, wherein the above steps are performed when it is determined that the operating inspection state is a second state or the optimized average value is less than or equal to the preset average value, and the second state is that the matching degree is less than or equal to the second preset matching degree. The analysis unit is further configured to decrease the length of the sliding window based on a variance of absolute values of slopes of the rotor current-time curve, and the decrease amplitude of the length of the sliding window is proportional to the variance.

7. The large model-based power generation terminal device operation inspection system according to claim 6, characterized by, The analysis unit is further configured to decrease the sliding step length based on the length of the sliding window, and the decrease amplitude of the sliding step length is inversely proportional to the length of the sliding window.

8. The large model-based power generation terminal device operation inspection system according to claim 7, characterized by, The analysis unit is further configured to increase the range reference of the infrared sensor based on the length of the sliding window, and the increase amplitude of the range reference of the infrared sensor is inversely proportional to the length of the sliding window.

9. A large model-based power generation terminal equipment operation and inspection method, characterized by, The method comprises: Real-time synchronous acquisition of infrared thermographs and vibration phase signals generated by a plurality of sensor monitoring points arranged at a power generation terminal device and collection of electrical operation parameters through a sliding window mechanism, wherein all real-time data are time-aligned according to a set sliding window length; Collection of historical operation parameters and historical fault types of the power generation terminal device in a historical operation process, wherein the historical operation parameters include infrared thermographs, vibration phase signals and electrical operation parameters; Training based on data obtained from the historical operation parameters and the historical fault types to construct a decision tree, wherein training parameters required for construction of the decision tree include a maximum depth when a tree structure is constrained during training, a minimum number of leaf nodes, a number of training iterations of the decision tree and an initial learning rate; Input of the obtained real-time data into the decision tree to output a fault type of the power generation terminal device in an abnormal operation state; Determination of an operating inspection state based on a matching degree between the output fault type and an actual fault type in a preset time period, and adjustment of a maximum depth when a tree structure is constrained during training of the decision tree based on the operating inspection state, and adjustment of the number of training iterations of the decision tree when the number of adjustments is equal to a preset number and the operating inspection state is still unqualified; Drawing of a matching degree-preset time period curve based on matching degrees in a plurality of preset time periods, wherein the above steps are performed when it is determined that the operating inspection state is a first state, and the first state is that the matching degree is less than a first preset matching degree and greater than a second preset matching degree. Determine whether to adjust the maximum depth of the training constraint tree structure based on an average of absolute values of slopes corresponding to a plurality of nodes of the curve; Increase the maximum depth of the training constraint tree structure based on a ratio of a sample quantity obtained within a preset time period to a preset sample quantity, and the increase amplitude of the maximum depth of the training constraint tree structure is proportional to the ratio.

Images