Hydropower station remote fault early warning method and system based on big data analysis

Abstract

The invention discloses a hydropower station remote fault early warning method and system based on big data analysis, and the method comprises the steps: collecting data through a sensor, and dynamically adjusting the data collection frequency and parameters of the sensor through a machine learning algorithm; using the processed data to construct a fault identification early warning model, and identifying a potential fault; based on the identified potential fault, combining a high-precision space-time positioning technology to construct a fault phase positioning model; and establishing an intelligent decision-making model by using the analyzed fault and the fault position, and providing an optimal fault processing scheme. And the accurate positioning of the fault is realized. The accuracy and timeliness of fault early warning are remarkably improved, missing detection and false detection are reduced, and the safe operation level and economic benefits of the hydropower station are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring and maintenance, and in particular to a remote fault early warning method and system for a hydropower station based on big data analysis. Background Art

[0002] With the continuous growth of electricity demand and the complexity of power systems, the safe and reliable operation of hydropower stations as important clean energy production units is particularly important. Traditional fault detection and early warning methods for hydropower stations mainly rely on manual inspections and regular maintenance, which is not only inefficient but also prone to missed detection and false detection. In recent years, with the rapid development of the Internet of Things, sensor technology and big data analysis technology, intelligent monitoring and fault early warning of hydropower stations have become a research hotspot. It has become an effective means of fault early warning to monitor various operating parameters of hydropower stations in real time, such as voltage, current, vibration, temperature, etc., and combine big data analysis and machine learning algorithms to analyze and predict the operating status of equipment in real time. However, the existing technology still has many shortcomings in data acquisition, fault identification and early warning accuracy, and it is urgent to further optimize and improve.

[0003] The existing fault warning technology for hydropower stations has the following shortcomings: First, the data collection frequency and parameter settings are often fixed and lack dynamic adjustment capabilities, resulting in the collected data being inaccurate or untimely when the equipment operating status changes, affecting the accuracy of fault identification. Second, the existing fault identification models usually only consider a single data source and fail to fully utilize the complementarity and correlation of multi-source data, resulting in limited fault identification effects. Third, fault location technology mainly relies on traditional geographic information systems (GIS), and its positioning accuracy and real-time performance are difficult to meet the needs of complex hydropower station environments. Finally, the existing technology lacks intelligent and automated means in the decision-making of fault handling solutions, and mainly relies on manual experience and judgment, and cannot give full play to the advantages of big data and machine learning technologies. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: data collection is not dynamic enough, the fault identification model is single, the fault location accuracy is insufficient, and the decision-making scheme has a low degree of intelligence.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a remote fault early warning method for a hydropower station based on big data analysis, comprising:

[0007] Collect data through sensors and use machine learning algorithms to dynamically adjust the sensor's data collection frequency and parameters;

[0008] Using the processed data, a fault identification and early warning model is constructed to identify potential faults;

[0009] Based on the identified potential faults, a fault phase location model is constructed in combination with high-precision spatiotemporal location technology;

[0010] Using the analyzed faults and fault locations, an intelligent decision-making model is established to provide the best fault handling solution.

[0011] As a preferred solution of the method for remote fault early warning of a hydropower station based on big data analysis described in the present invention, wherein: collecting data through sensors includes collecting voltage and current data of the hydropower station through voltage sensors and current sensors to monitor the operating status of the power system;

[0012] Vibration sensors are used to monitor vibrations in generators, turbines and other mechanical components;

[0013] Use displacement sensors to monitor the displacement of key components;

[0014] Collect temperature data of the equipment and the surrounding environment through temperature sensors;

[0015] Use humidity sensors to monitor the humidity of the environment where the equipment is located;

[0016] Monitor water and air pressure data of hydropower stations through pressure sensors;

[0017] Use oil quality sensors to monitor the quality of lubricating oil, transformer oil, etc.;

[0018] Monitoring of gas composition around generators and transformers through gas sensors;

[0019] Locate mobile devices and key components through GPS sensors;

[0020] Use noise sensors to monitor the noise level of equipment during operation;

[0021] Collect historical data through the hydropower station database.

[0022] As a preferred solution of the method for remote fault warning of a hydropower station based on big data analysis described in the present invention, the method of dynamically adjusting the data acquisition frequency and parameters of the sensor using a machine learning algorithm includes selecting a random forest model as the basis of the machine learning algorithm, designing a multi-source data fusion mechanism, fusing the data of different sensors into a comprehensive value, and introducing an exponential decay function to represent the change of data acquisition over time, which is expressed as follows:

[0023]

[0024] Among them, wi represents the weight in data fusion, F(X,t) represents the dynamically adjusted data collection frequency and parameter function, T represents the number of trees in the random forest model, x i (t) represents the data of the ith data source at time t, f0e -λt represents the exponential decay function of the data acquisition frequency, g k( X ) represents the dynamic adjustment function of the sensor parameters, k represents the parameters in the Laplace transform, and F(X,t) represents the data processing function.

[0025] As a preferred solution of the method for remote fault warning of a hydropower station based on big data analysis described in the present invention, wherein: using the processed data to build a fault identification and warning model to identify potential faults includes: based on the processed data, through reinforcement learning and generative adversarial network methods, building a fault identification and warning model, the formula is expressed as:

[0026]

[0027] Design a discriminator output and state-action value function that combines the generative adversarial network to verify the warning results. The formula is expressed as:

[0028]

[0029] Where s represents the current state of the system, a represents the action taken in state s, r(s,a) represents the reward for taking action a in state s, γ represents the discount factor, P(s′|s,a) represents the probability of taking action a in state s to transfer to state s′, G represents the generator, D represents the discriminator, z represents the input noise of the generator, x represents the real data, E represents the expectation, and w i represents the weight, x i (t) represents the i-th input data at time t, f0 represents the initial value, λ represents the exponential decay coefficient, g k (X) represents the effect of function g on data X, k is the attenuation constant, represents the state-action value function after warning, α represents the weight coefficient of the state-action value function, β represents the weight coefficient of the state-action value function after warning, and η represents the weight coefficient of the discrimination result;

[0030] The value range of Q(s,a) is from -1 to 1, and the closer it is to 1, the higher the probability of failure;

[0031] The value range of V is from negative infinity to positive infinity. The larger the value, the higher the accuracy of the warning result.

[0032] As a preferred solution of the method for remote fault warning of a hydropower station based on big data analysis described in the present invention, wherein: based on the identified potential faults, combined with high-precision spatiotemporal positioning technology, a fault phase positioning model is constructed, including combining a spatiotemporal convolutional network and high-precision spatiotemporal positioning technology to construct a fault phase positioning model, and the formula is expressed as:

[0033] Temporal convolution operation:

[0034] H t =σ(W t *X)

[0035] Spatial convolution operation:

[0036] H st =σ(W s *H t )

[0037] The high-precision spatiotemporal positioning formula is expressed as:

[0038] L(t)=f(GPS(t),IMU(t))

[0039]

[0040] Design a verification formula to verify the fault location result. The formula is expressed as:

[0041] L=E[(yy true ) 2 ]

[0042] Where X represents the spatiotemporal data matrix, W t represents the temporal convolution kernel, W s represents the spatial convolution kernel, σ represents the activation function, L(t) represents the high-precision spatiotemporal positioning function, GPS(t) represents the GPS data at time t, IMU(t) represents the IMU data at time t, H t represents the temporal convolution feature, H st represents the spatiotemporal convolution feature, θ represents the model parameter, y represents the fault phase location result, represents the fault phase location prediction value, y true represents the actual fault location, L represents the loss function, and E represents the expectation;

[0043] The value range of y is from -1 to 1, and the closer it is to 1, the higher the certainty of the fault location;

[0044] The value range of L is from 0 to positive infinity. The closer it is to 0, the higher the accuracy of the positioning result.

[0045] As a preferred solution of the method for remote fault warning of a hydropower station based on big data analysis described in the present invention, wherein: using the analyzed faults and fault locations, an intelligent decision-making model is established to provide the optimal fault handling solution, including combining the analyzed value thresholds of Q(s, a), V, y and L, and using knowledge graphs and natural language processing (NLP) technology to realize the automatic extraction and application of expert knowledge, while introducing association rule mining and causal reasoning algorithms, and using historical data to realize the automatic analysis and inference of the root cause of the fault, the formula is expressed as:

[0046]

[0047] Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge graph, R represents the association rule, C represents the causal relationship, D represents the historical data, P represents the fault handling plan, and f represents the decision function.

[0048] As a preferred solution of the method for remote fault warning of a hydropower station based on big data analysis described in the present invention, wherein: using the analyzed faults and fault locations to establish an intelligent decision-making model, providing the best fault handling solution also includes using Q(s, a) and y to identify and locate the fault, using V and L to verify the warning results, using the knowledge graph K and NLP technology to extract expert knowledge, and generating a preliminary fault handling solution, the formula is expressed as:

[0049] K={(E,R,T)}

[0050] P1=NLP ( K)

[0051] The association rules R and causal relationships C are used to optimize and verify the preliminary plan. The formula is expressed as follows:

[0052] R={(A→B, support, confidence)}

[0053] Use the causal relationship C to reason and analyze the fault:

[0054] C={(X,Y)|X→Y}

[0055] P2=ApplyRandCtoP1 applies R and C to P

[0056] Use historical data D to verify and improve the optimized solution:

[0057] P = optmizeP2withD

[0058] Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge relationship, D represents historical data, P represents the fault handling plan, f represents the decision function, E represents the entity, T represents the type, A and B represent events, support represents support, and confidence represents confidence.

[0059] When Q(s,a)>0.8 and y>0.8, the system identifies and locates a high-probability fault and starts the fault handling process;

[0060] When V>10 and L<0.1, verify the accuracy of the warning result and confirm the fault;

[0061] Use knowledge graph and NLP technology to extract expert knowledge and generate a preliminary fault handling solution P1;

[0062] Optimize and verify P1 through association rule mining and causal reasoning to generate the optimization solution P2;

[0063] P2 is further optimized in combination with historical data, and the optimal fault handling solution P is finally generated.

[0064] A remote fault early warning system for a hydropower station based on big data analysis, wherein:

[0065] The data acquisition module collects data through sensors and uses machine learning algorithms to dynamically adjust the data acquisition frequency and parameters of the sensors;

[0066] The fault identification and early warning module uses the processed data to build a fault identification and early warning model to identify potential faults;

[0067] The fault phase location module builds a fault phase location model based on the identified potential faults and combined with high-precision time-space location technology;

[0068] The intelligent decision-making module uses the analyzed faults and fault locations to establish an intelligent decision-making model and provide the best fault handling solution.

[0069] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0070] A computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps of any one of the methods of the present invention.

[0071] Beneficial effects of the present invention: The present invention provides a remote fault warning method for a hydropower station based on big data analysis. By dynamically adjusting the data acquisition frequency and parameters of the sensor, a fault identification and warning model of multi-source data fusion is constructed, and high-precision spatiotemporal positioning technology is combined to achieve accurate positioning of the fault. By using knowledge graphs and natural language processing technology, expert knowledge is automatically extracted and applied, and an intelligent decision-making model is established in combination with association rule mining and causal reasoning algorithms to provide the best fault handling solution. The accuracy and timeliness of fault warnings are significantly improved, missed detections and false detections are reduced, and the safe operation level and economic benefits of hydropower stations are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0073] Figure 1 An overall flow chart of a remote fault warning method for a hydropower station based on big data analysis provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0074] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0075] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides a remote fault early warning method for a hydropower station based on big data analysis, comprising:

[0076] S1: Collect data through sensors and use machine learning algorithms to dynamically adjust the data collection frequency and parameters of sensors.

[0077] The voltage and current data of the hydropower station are collected through voltage sensors and current sensors to monitor the operating status of the power system; the vibration of generators, turbines and other mechanical parts is monitored through vibration sensors; the displacement of key components is monitored using displacement sensors; the temperature data of the equipment and the surrounding environment is collected through temperature sensors; the humidity of the environment where the equipment is located is monitored using humidity sensors; the water pressure and air pressure data of the hydropower station are monitored through pressure sensors; the quality of lubricating oil, transformer oil, etc. is monitored using oil quality sensors; the gas composition around generators and transformers is monitored through gas sensors; mobile equipment and key components are located through GPS sensors; noise sensors are used to monitor the noise level of equipment during operation; and historical data is collected through the hydropower station database.

[0078] The random forest model is selected as the basis of the machine learning algorithm, and a multi-source data fusion mechanism is designed to fuse the data of different sensors into a comprehensive value, and an exponential decay function is introduced to represent the change of data collection over time. The formula is expressed as follows:

[0079]

[0080] Among them, w i represents the weight in data fusion, F(X,t) represents the dynamically adjusted data collection frequency and parameter function, T represents the number of trees in the random forest model, x i (t) represents the data of the ith data source at time t, f0e -λt represents the exponential decay function of the data acquisition frequency, g k( X ) represents the dynamic adjustment function of the sensor parameters, k represents the parameters in the Laplace transform, and F(X,t) represents the data processing function.

[0081] The operating status of a hydropower station is complex and changeable, and data collection with a single fixed frequency cannot fully and accurately reflect the true status of the equipment. By collecting a variety of data through sensors, such as voltage, current, vibration, displacement, temperature, humidity, pressure, etc., the operating status of the hydropower station can be fully monitored. By using machine learning algorithms to dynamically adjust the data collection frequency and parameters of sensors, the data collection strategy can be optimized in real time according to the operating status of the equipment, improving the real-time and accuracy of the data.

[0082] Furthermore, this dynamic adjustment mechanism can not only save energy consumption of sensors and extend their service life, but also ensure that more accurate data is collected at critical moments, thereby improving the sensitivity and accuracy of fault detection.

[0083] S2: Use the processed data to build a fault identification and warning model to identify potential faults.

[0084] Based on the processed data, a fault identification and warning model is constructed through reinforcement learning and generative adversarial network methods. The formula is expressed as:

[0085]

[0086] Design a discriminator output and state-action value function that combines the generative adversarial network to verify the warning results. The formula is expressed as:

[0087]

[0088] Where s represents the current state of the system, a represents the action taken in state s, r(s,a) represents the reward for taking action a in state s, γ represents the discount factor, P(s′|s,a) represents the probability of taking action a in state s to transfer to state s′, G represents the generator, D represents the discriminator, z represents the input noise of the generator, x represents the real data, E represents the expectation, and w i represents the weight, x i (t) represents the i-th input data at time t, f0 represents the initial value, λ represents the exponential decay coefficient, g k (X) represents the effect of function g on data X, k is the attenuation constant, represents the state-action value function after warning, α represents the weight coefficient of the state-action value function, β represents the weight coefficient of the state-action value function after warning, and η represents the weight coefficient of the discrimination result.

[0089] The value range of Q9s,a) is from -1 to 1, and the closer it is to 1, the higher the probability of failure.

[0090] The value range of V is from negative infinity to positive infinity. The larger the value, the higher the accuracy of the warning result.

[0091] After preprocessing, the large amount of data collected by the sensor can remove noise and irrelevant information, making the data more representative and accurate. By building a fault identification and warning model, more accurate fault identification can be performed based on these processed data. The use of reinforcement learning and generative adversarial network (GAN) methods can improve the accuracy and real-time performance of fault identification in the continuous learning and optimization of the model.

[0092] Furthermore, this method can not only efficiently identify potential faults and prevent serious faults from occurring, but also improve the accuracy and credibility of early warning results through the GAN model, reducing false positives and missed negatives.

[0093] S3: Based on the identified potential faults, a fault phase location model is constructed in combination with high-precision spatiotemporal location technology.

[0094] Combining the spatiotemporal convolutional network and high-precision spatiotemporal positioning technology, a fault phase positioning model is constructed, and the formula is expressed as follows:

[0095] Temporal convolution operation:

[0096] H t =σ(W t *X)

[0097] Spatial convolution operation:

[0098] H st =σ(W s *H t )

[0099] The high-precision spatiotemporal positioning formula is expressed as:

[0100] L(t)=f(GPS(t),IMU(t))

[0101]

[0102] Design a verification formula to verify the fault location result. The formula is expressed as:

[0103] L=E[(yy true ) 2 ]

[0104] Where X represents the spatiotemporal data matrix, W t represents the temporal convolution kernel, W s represents the spatial convolution kernel, σ represents the activation function, L(t) represents the high-precision spatiotemporal positioning function, GPS(t) represents the GPS data at time t, IMU(t) represents the IMU data at time t, H t represents the temporal convolution feature, H st represents the spatiotemporal convolution feature, θ represents the model parameter, y represents the fault phase location result, represents the fault phase location prediction value, y true represents the actual fault location, L represents the loss function, and E represents the expectation.

[0105] The value range of y is from -1 to 1, and the closer it is to 1, the higher the certainty of the fault location.

[0106] The value range of L is from 0 to positive infinity. The closer it is to 0, the higher the accuracy of the positioning result.

[0107] After identifying a potential fault, accurately locating the location of the fault is the key to taking further measures. Combining spatiotemporal convolutional networks and high-precision spatiotemporal positioning technology can achieve accurate fault positioning. Spatiotemporal convolutional networks can analyze the characteristics of faults in time and space, while high-precision positioning technology can provide accurate location information.

[0108] Furthermore, this combination can not only quickly locate the fault location, reduce troubleshooting time, and improve maintenance efficiency, but also provide high-precision positioning results in complex environments, ensuring that maintenance personnel can quickly and accurately reach the fault point for processing.

[0109] S4: Use the analyzed faults and fault locations to establish an intelligent decision-making model and provide the best fault handling solution.

[0110] Combined with the analyzed value thresholds of Q(s,a), V, y and L, and using knowledge graph and natural language processing (NLP) technology, the automatic extraction and application of expert knowledge is realized. At the same time, association rule mining and causal reasoning algorithms are introduced, and historical data are used to realize the automatic analysis and inference of the root cause of the fault. The formula is expressed as follows:

[0111]

[0112] Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge graph, R represents the association rule, C represents the causal relationship, D represents the historical data, P represents the fault handling plan, and f represents the decision function.

[0113] Use Q(s,a) and y to identify and locate faults, use V and L to verify the warning results, use the knowledge graph K and NLP technology to extract expert knowledge, and generate a preliminary fault handling plan. The formula is expressed as:

[0114] K={(E,R,T)}

[0115] P1=NLP ( K)

[0116] The association rules R and causal relationships C are used to optimize and verify the preliminary plan. The formula is expressed as follows:

[0117] R={(A→B, support, confidence)}

[0118] Use the causal relationship C to reason and analyze the fault:

[0119] C={(X,Y)|X→Y}

[0120] P2=ApplyRandCtoP1 applies R and C to P

[0121] Use historical data D to verify and improve the optimized solution:

[0122] P = optmizeP2withD

[0123] Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge relationship, D represents historical data, P represents the fault handling plan, f represents the decision function, E represents the entity, T represents the type, A and B represent events, support represents support, and confidence represents confidence.

[0124] When Q(s,a)>0.8 and y>0.8, the system identifies and locates a high-probability fault and starts the fault handling process.

[0125] When V>10 and Verify the accuracy of the warning results and confirm the fault.

[0126] Use knowledge graph and NLP technology to extract expert knowledge and generate a preliminary fault handling plan P1.

[0127] P1 is optimized and verified through association rule mining and causal reasoning to generate the optimization solution P2.

[0128] P2 is further optimized in combination with historical data, and the optimal fault handling solution P is finally generated.

[0129] After the fault is identified and its location is determined, an intelligent decision support system is needed to provide the best fault handling solution. By combining the fault identification warning model value, warning result verification value, fault phase location result and fault phase location result error threshold, and using knowledge graph and natural language processing technology, the automatic extraction and application of expert knowledge can be realized. Through association rule mining and causal reasoning algorithms, the root cause of the fault can be automatically analyzed and inferred based on historical data.

[0130] Furthermore, this intelligent decision-making model can provide real-time, optimal fault handling solutions, reduce errors in human decision-making, and improve the efficiency and accuracy of fault handling. In addition, by automatically extracting expert knowledge, decisions can be made quickly, improving the overall intelligence level of the system.

[0131] On the other hand, this embodiment also provides a remote fault warning system for a hydropower station based on big data analysis, which includes:

[0132] The data acquisition module collects data through sensors and uses machine learning algorithms to dynamically adjust the data acquisition frequency and parameters of the sensors.

[0133] The fault identification and early warning module uses the processed data to build a fault identification and early warning model to identify potential faults.

[0134] The fault phase location module builds a fault phase location model based on the identified potential faults and combined with high-precision time and space location technology.

[0135] The intelligent decision-making module uses the analyzed faults and fault locations to establish an intelligent decision-making model and provide the best fault handling solution.

[0136] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0137] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0138] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0139] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0140] Example 2

[0141] The experiment was conducted in a large hydropower station. The main experimental equipment included voltage sensors, current sensors, vibration sensors, displacement sensors, temperature sensors, humidity sensors, pressure sensors, oil quality sensors, gas sensors, GPS sensors and noise sensors. Various sensors were installed at key equipment and nodes of the hydropower station. The sensor type and location were determined according to the operating characteristics of the equipment and historical fault data.

[0142] The collected data is preprocessed by the data processing unit to remove noise and outliers to ensure the accuracy and validity of the data. Then, the machine learning algorithm is used to dynamically adjust the data collection frequency and parameters of the sensor to ensure the real-time and adaptability of data collection.

[0143] The processed data is used to build a fault identification and early warning model. The model uses reinforcement learning and generative adversarial network (GAN) methods to improve the accuracy of fault identification through continuous learning and optimization. Based on the identified potential faults, a fault phase location model is built in combination with high-precision spatiotemporal positioning technology. The spatiotemporal data is processed through a spatiotemporal convolutional network, combined with GPS and IMU data, to achieve accurate fault location.

[0144] An intelligent decision-making model is established using the analyzed fault information and fault location. The model combines knowledge graph and natural language processing (NLP) technology to automatically extract and apply expert knowledge, and introduces association rule mining and causal reasoning algorithms to automatically analyze and infer the root cause of the fault based on historical data, and ultimately provide the optimal fault handling solution. The experimental results are shown in Table 1.

[0145] Table 1 Experimental data table

[0146]

[0147] The results of experimental data analysis show that the fault identification and early warning model of the present invention can effectively identify potential faults. For example, at a certain moment, the vibration parameters of the generator 1 and the turbine 1 increased significantly, and the current parameters fluctuated abnormally. The early warning model successfully identified the potential fault and issued an alarm in time.

[0148] In terms of fault location, by combining high-precision spatiotemporal positioning technology and spatiotemporal convolutional networks, the experiment achieved accurate positioning of the fault location. When an actual fault occurs, the location of the faulty equipment can be accurately determined through GPS and IMU data, helping operation and maintenance personnel to quickly find the fault point and handle it.

[0149] The intelligent decision-making model makes full use of knowledge graphs and NLP technology, combines historical data and expert knowledge, automatically generates fault handling solutions, and optimizes and verifies them through association rule mining and causal reasoning. Experimental data shows that the decision-making model not only provides the best fault handling solution, but also significantly improves the efficiency and accuracy of fault handling and reduces the errors caused by human judgment.

[0150] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A remote fault warning method for a hydropower station based on big data analysis, characterized in that: include: Collect data through sensors and use machine learning algorithms to dynamically adjust the data collection frequency and parameters of sensors; Using the processed data, a fault identification and early warning model is constructed to identify potential faults; Based on the identified potential faults, a fault phase location model is constructed in combination with high-precision spatiotemporal location technology; Using the analyzed faults and fault locations, an intelligent decision-making model is established to provide the best fault handling solution.

2. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 1, characterized in that: Collecting data through sensors includes collecting voltage and current data of the hydropower station through voltage sensors and current sensors to monitor the operating status of the power system; Vibration sensors are used to monitor vibrations in generators, turbines and other mechanical components; Use displacement sensors to monitor the displacement of key components; Collect temperature data of the equipment and the surrounding environment through temperature sensors; Use humidity sensors to monitor the humidity of the environment where the equipment is located; Monitor water and air pressure data of hydropower stations through pressure sensors; Use oil quality sensors to monitor the quality of lubricating oil, transformer oil, etc.; Monitoring of gas composition around generators and transformers through gas sensors; Locate mobile devices and key components through GPS sensors; Use noise sensors to monitor the noise level of equipment during operation; Collect historical data through the hydropower station database.

3. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 2, characterized in that: Using machine learning algorithms to dynamically adjust the data collection frequency and parameters of sensors includes selecting the random forest model as the basis of the machine learning algorithm, designing a multi-source data fusion mechanism, fusing the data from different sensors into a comprehensive value, and introducing an exponential decay function to represent the change of data collection over time. The formula is expressed as: Among them, w i represents the weight in data fusion, F(X,t) represents the dynamically adjusted data collection frequency and parameter function, T represents the number of trees in the random forest model, x i (t) represents the data of the ith data source at time t, f0e -λt represents the exponential decay function of the data acquisition frequency, g k (X) represents the dynamic adjustment function of the sensor parameters, k represents the parameters in the Laplace transform, and F(X,t) represents the data processing function.

4. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 3, characterized in that: Using the processed data, a fault identification and warning model is constructed to identify potential faults. Based on the processed data, a fault identification and warning model is constructed through reinforcement learning and generative adversarial network methods. The formula is expressed as: Design a discriminator output and state-action value function that combines the generative adversarial network to verify the warning results. The formula is expressed as: Where s represents the current state of the system, a represents the action taken in state s, r(s,a) represents the reward for taking action a in state s, γ represents the discount factor, P(s′|s,a) represents the probability of taking action a in state s to transfer to state s′, G represents the generator, D represents the discriminator, z represents the input noise of the generator, x represents the real data, E represents the expectation, and w i represents the weight, x i (t) represents the i-th input data at time t, f0 represents the initial value, λ represents the exponential decay coefficient, g k (X) represents the effect of function g on data X, k is the attenuation constant, represents the state-action value function after warning, α represents the weight coefficient of the state-action value function, β represents the weight coefficient of the state-action value function after warning, and η represents the weight coefficient of the discrimination result; The value range of Q(s,a) is from -1 to 1, and the closer it is to 1, the higher the probability of failure; The value range of V is from negative infinity to positive infinity. The larger the value, the higher the accuracy of the warning result.

5. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 4, characterized in that: Based on the identified potential faults, combined with high-precision spatiotemporal positioning technology, a fault phase positioning model is constructed, including combining spatiotemporal convolutional network and high-precision spatiotemporal positioning technology to construct a fault phase positioning model, which is expressed as follows: Temporal convolution operation: H t =s ( W t *X) Spatial convolution operation: H st =σ ( W s *H t ) The high-precision spatiotemporal positioning formula is expressed as: L(t)=f(GPS(t),IMU(t)) Design a verification formula to verify the fault location result. The formula is expressed as: L=E[(yy true ) 2 ] Where X represents the spatiotemporal data matrix, W t represents the temporal convolution kernel, W s represents the spatial convolution kernel, σ represents the activation function, L(t ) represents the high-precision spatiotemporal positioning function, GPS(t) represents the GPS data at time t, and IMU(t ) represents the IMU data at time t, H t represents the temporal convolution feature, H st represents the spatiotemporal convolution feature, θ represents the model parameter, y represents the fault phase location result, represents the fault phase location prediction value, y true represents the actual fault location, L represents the loss function, and E represents the expectation; The value range of y is from -1 to 1, and the closer it is to 1, the higher the certainty of the fault location; The value range of L is from 0 to positive infinity. The closer it is to 0, the higher the accuracy of the positioning result.

6. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 5, characterized in that: By using the analyzed faults and fault locations, an intelligent decision-making model is established to provide the optimal fault handling solution, including combining the analyzed Q(s,a), V, y and L value thresholds, and using knowledge graphs and natural language processing (NLP) technology to achieve automatic extraction and application of expert knowledge. At the same time, association rule mining and causal reasoning algorithms are introduced to use historical data to achieve automatic analysis and inference of the root cause of the fault. The formula is expressed as follows: Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge graph, R represents the association rule, C represents the causal relationship, D represents the historical data, P represents the fault handling plan, and f represents the decision function.

7. A method for remote fault early warning of a hydropower station based on big data analysis as claimed in claim 6, characterized in that: Using the analyzed faults and fault locations, an intelligent decision-making model is established to provide the best fault handling solution, which also includes using Q(s,a) and y to identify and locate faults, using V and L to verify the warning results, and using the knowledge graph K and NLP technology to extract expert knowledge and generate a preliminary fault handling solution. The formula is expressed as follows: K={(E,R,T)} P1=NLP ( K) The association rules R and causal relationships C are used to optimize and verify the preliminary plan. The formula is expressed as follows: R={(A→B, support, confidence)} Use the causal relationship C to reason and analyze the fault: C={(X,Y)|X→Y} P2=ApplyRandCtoP1 applies R and C to P Use historical data D to verify and improve the optimized solution: P = optmizeP2withD Among them, Q(s,a) represents the fault identification warning model value, V represents the warning result verification value, y represents the fault phase location result, L represents the error of the fault phase location result, K represents the knowledge relationship, D represents historical data, P represents the fault handling plan, f represents the decision function, E represents the entity, T represents the type, A and B represent events, support represents support, and confidence represents confidence. When Q(s,a)>0.8 and y>0.8, the system identifies and locates a high-probability fault and starts the fault handling process; When V>10 and L<0.1, verify the accuracy of the warning result and confirm the fault; Use knowledge graph and NLP technology to extract expert knowledge and generate a preliminary fault handling solution P1; Optimize and verify P1 through association rule mining and causal reasoning to generate the optimization solution P2; P2 is further optimized in combination with historical data, and the optimal fault handling solution P is finally generated.

8. A remote fault early warning system for a hydropower station based on big data analysis using the method according to any one of claims 1 to 7, characterized in that: The data acquisition module collects data through sensors and uses machine learning algorithms to dynamically adjust the data acquisition frequency and parameters of the sensors; The fault identification and warning module uses the processed data to build a fault identification and warning model to identify potential faults; The fault phase location module builds a fault phase location model based on the identified potential faults and combined with high-precision time-space location technology; The intelligent decision-making module uses the analyzed faults and fault locations to establish an intelligent decision-making model and provide the best fault handling solution.

9. A computer device comprising: Memory and processor; The memory stores a computer program, which is characterized in that when the processor executes the computer program, it implements the steps of a remote fault warning method for a hydropower station based on big data analysis.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a remote fault warning method for a hydropower station based on big data analysis are implemented.

Images