A Method for Inspection and Fault Diagnosis of New Energy Power Stations Based on the Internet of Things and Artificial Intelligence

Abstract

This application relates to the field of data processing technology and discloses a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence. The method includes: acquiring high-precision map data of the new energy power station; collecting equipment information of the equipment to be inspected; planning inspection routes for intelligent inspection equipment; and conducting inspections of the new energy power station based on the inspection routes. During inspection, real-time image data of the equipment within the new energy power station is collected, and periodic operational data of the equipment to be inspected is collected. A cloud server uses artificial intelligence technology to determine whether the equipment to be inspected has malfunctioned, and if a malfunction occurs, it also identifies the corresponding malfunction type, cause, and location. Real-time operational data and inspection results of the new energy power station are viewed through terminal devices. This application enables efficient inspection while saving costs, and also accurately diagnoses malfunctions and provides solutions.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular to a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence. Background Technology

[0002] With the rapid development of the new energy industry, the operation and maintenance management of new energy power plants such as wind farms is facing problems such as low efficiency and high costs. Traditional inspection methods rely on manual labor, which is time-consuming and labor-intensive, and it is difficult to monitor and diagnose equipment faults in real time.

[0003] A similar prior art patent application, CN118333257A, provides an inspection service management system based on artificial intelligence and the Internet of Things. The method includes an inspection service module, an inspection anomaly module, and a path customization module. The inspection service module and the inspection anomaly module are set up to control the inspection equipment to perform inspection services on the equipment to be serviced. The status of the equipment to be serviced is intelligently analyzed through periodic detection, and the equipment with inspection anomalies is marked in a timely manner. The path customization module is set up to intelligently analyze the relationship between the equipment with inspection anomalies and other equipment to be serviced, and the inspection service path of the inspection equipment is re-customized through correlation analysis to dynamically optimize the inspection service path and ensure rational inspection services for the equipment to be serviced.

[0004] Similar existing technologies include Chinese patent application CN106056281A, which provides an IoT-based inspection management method. The method includes: a) synchronizing data from a central database to an inspection terminal; b) generating a QR code containing the inspection point address information and the inspection items for that inspection point and sending it to the corresponding inspection point; c) the inspection terminal obtaining and displaying the inspection items by scanning the QR code; d) the inspection terminal recording and / or modifying the inspection data and generating inspection result information, which is then stored locally; e) when the inspection terminal goes online, synchronizing its local data to the central database.

[0005] However, neither of the technical solutions in the above two documents took into account the problem of increased inspection costs and reduced inspection efficiency caused by an excessive number of inspection devices.

[0006] Therefore, this invention provides a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence. Summary of the Invention

[0007] This application provides a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence, which can improve inspection efficiency, reduce inspection costs, and improve the accuracy of fault diagnosis.

[0008] Firstly, this application provides a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence, the method comprising:

[0009] Step S1: Obtain high-precision map data of the new energy power station. The high-precision map data includes equipment location, road network and obstacles. Collect equipment information of the equipment to be inspected, plan inspection routes for intelligent inspection equipment, and inspect the new energy power station based on the inspection routes.

[0010] Step S2: During the inspection, real-time image data of the equipment in the new energy power station is collected and sent to the cloud server. Smart sensors are installed on each piece of equipment to be inspected in the new energy power station to periodically collect the operating data of the equipment to be inspected. The operating data includes at least power generation, voltage, current and temperature, and the operating data is sent to the cloud server.

[0011] Step S3: The cloud server uses artificial intelligence to perform in-depth mining and analysis on the operating data and the equipment image data to determine whether the equipment to be inspected has a fault. If a fault occurs, it also identifies the corresponding fault type, fault cause and fault location.

[0012] Step S4: The real-time operation data and inspection results of the new energy power station are displayed in real time using visualization technology through terminal equipment. Historical operation data and historical inspection results can be viewed, and filtering can be performed by time and equipment type. When a fault is detected, an audible and visual warning will be issued in a timely manner, and the corresponding fault location, fault cause and response plan will be displayed.

[0013] In conjunction with the first aspect, in the first implementation of the first aspect of this application, planning an inspection route for the intelligent inspection equipment includes:

[0014] Step S11: Obtain the optimal number of intelligent inspection devices as the first number through calculation and comparison;

[0015] Step S12: Calculate the first distance from the inspection starting point to all uninspected equipment to be inspected, take the first number of smallest first distances corresponding to the equipment to be inspected as the first inspection target, and take the first inspection target as the first inspection target of each of the intelligent inspection devices.

[0016] Step S13: For each of the intelligent inspection devices, after inspecting one device to be inspected, the device information and first information corresponding to the device to be inspected are sent to the cloud server. Based on the cloud server, all uninspected devices to be inspected are obtained. Then, the second distance from the currently inspected device to all uninspected devices to be inspected is calculated. The device to be inspected corresponding to the smallest second distance is taken as the next inspection target of the intelligent inspection device. This step is repeated until all devices to be inspected have been inspected.

[0017] In conjunction with the first aspect, in the second implementation of the first aspect of this application, after the intelligent inspection device obtains the next inspection target, it includes:

[0018] Step S131: Send the device information and second information of the intelligent inspection device corresponding to the next inspection target to the cloud server;

[0019] Step S132: If a fault occurs during the inspection process, the intelligent inspection device will send the device information and third information of the corresponding device to be inspected to the cloud server.

[0020] Step S133: The cloud server's storage unit stores a first list, a second list, and a third list. After receiving the first information, the cloud server deletes the device to be inspected corresponding to the first information from the second list and adds it to the third list. After receiving the second information, the cloud server deletes the device to be inspected corresponding to the second information from the first list and adds it to the second list. After receiving the third information, the cloud server removes the device to be inspected corresponding to the third information from the third list and adds it to the first list.

[0021] In conjunction with the first aspect, in the third implementation of the first aspect of this application, obtaining the optimal number of the intelligent inspection devices as the first number through calculation and comparison includes:

[0022] Step S111: Set a second quantity and a third quantity. The second quantity refers to the minimum number of intelligent inspection devices, and the third quantity refers to the maximum number of intelligent inspection devices. Obtain historical inspection data of the intelligent inspection devices, and calculate the inspection completion rate of the intelligent inspection devices based on the historical inspection data.

[0023] Step S112: When the number of intelligent inspection devices is the second number, plan inspection routes for the second number of intelligent inspection devices, obtain the inspection distance of each intelligent inspection device based on the inspection route, obtain the corresponding inspection completion rate based on the inspection distance, calculate the overall completion rate based on the inspection completion rate, and also obtain the inspection time corresponding to the largest inspection distance, and use the inspection time as the inspection time of the corresponding inspection route.

[0024] Step S113: Increment the second quantity by one, repeat step S112 to obtain a new overall completion rate, until the second quantity equals the third quantity, and then execute step S114;

[0025] Step S114: Obtain the second quantity corresponding to the overall completion rate being greater than a preset first threshold, and obtain multiple inspection schemes corresponding to the second quantity, and also obtain the inspection time corresponding to the inspection scheme, and take the second quantity corresponding to the inspection scheme with the smallest inspection time as the optimal quantity.

[0026] In conjunction with the first aspect, in the fourth implementation of the first aspect of this application, determining whether the device has malfunctioned includes:

[0027] Step S31: Obtain normal equipment image data of new energy power stations collected in history, and use the equipment image data of two adjacent collection cycles as learning data to train the first model;

[0028] Step S32: Obtain real-time collected device image data, and also obtain the device image data collected in the previous collection cycle as the first image data. Input the first image data into the first model, and input the predicted image data from the first model.

[0029] Step S33: Compare the real-time collected device image data and the predicted image data, calculate the first difference between the two, and if the first difference is greater than a preset second threshold, determine that the device to be inspected has a first fault.

[0030] In conjunction with the first aspect, in the fifth implementation of the first aspect of this application, determining whether the device has malfunctioned further includes:

[0031] Step S34: Obtain historically collected operational data, including normal operation data and abnormal operation data. Create a digital twin model of the equipment to be inspected based on the operational data and equipment-related information. Use the digital twin model to simulate the faults that occur in the equipment to be inspected. During the simulation of fault events in the digital twin model, collect relevant abnormal operation data and abnormal operation data.

[0032] Step S35: Use artificial intelligence technology to analyze all abnormal operating data and abnormal operation data, generate a fault storage table based on the analysis results, use normal operating data and abnormal operating data as second learning data to train the second model, input the real-time acquired operating data into the second model, and let the second model determine whether the equipment to be inspected has a second fault.

[0033] Step S36: In the event of a second fault, based on the real-time collected operational data, query the fault storage table for fault cases that are the same as or similar to the current fault, and obtain a solution.

[0034] In conjunction with the first aspect, in the sixth implementation of the first aspect of this application, artificial intelligence technology is used to analyze all abnormal operation data and abnormal data, and a fault storage table is generated based on the analysis results, including:

[0035] The abnormal operating data generated when an actual fault occurs is called the first abnormal data, and the abnormal operating data generated when a simulated fault occurs is called the second abnormal data. The second abnormal data is classified, and the fault cause and fault type corresponding to the second abnormal data of the same type are obtained. The first abnormal data is classified into the type corresponding to the second abnormal data based on similarity. Historical maintenance records are obtained, and corresponding response solutions are obtained for different fault types. The abnormal operating data, the corresponding fault cause, fault type and response solution are associated and stored in the fault storage table.

[0036] In conjunction with the first aspect, in the sixth implementation of the first aspect of this application, querying the fault storage table for fault cases that are the same as or similar to the current fault includes:

[0037] Retrieve all abnormal operation data for each type from the fault storage table. Normalize the abnormal operation data belonging to the same type. Take the average value of each data parameter in the normalized abnormal operation data. Combine all the average values ​​into feature data. Repeat this step to obtain feature data corresponding to all types. Normalize the abnormal operation data collected in real time to generate target data. Calculate the similarity between the target data and multiple feature data. Use the response plan corresponding to the fault type of the feature data with the highest similarity as the response plan for the target data.

[0038] The technical solution provided in this application collects equipment information with inspection devices, obtains the optimal number of intelligent inspection devices based on the equipment information through comparison and calculation, and plans inspection routes for the intelligent inspection devices. This ensures that the intelligent inspection devices make the best or optimal choice under the current state at each step, thereby achieving a balance between maximizing the inspection rate and minimizing the number of inspections. During the inspection process, real-time image data of the equipment within the new energy power station is collected, and the operating data of the equipment to be inspected is periodically collected. The equipment image data and operating data are sent to a cloud server, which determines whether the equipment to be inspected has malfunctioned. In the event of a malfunction, the corresponding malfunction type, cause, and solution are identified, enabling timely detection of equipment anomalies or malfunctions, improving the accuracy and timeliness of fault diagnosis, and providing reference solutions for maintenance personnel. Real-time operation data and inspection results of the new energy power station are displayed in real-time using visualization technology through terminal devices, helping relevant personnel to view the operating status and inspection results of the new energy power station in real time and perform maintenance management based on this information, thereby improving the operating efficiency and reliability of the new energy power station. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a schematic diagram of an embodiment of the new energy power station inspection and fault diagnosis method based on the Internet of Things and artificial intelligence in this application. Detailed Implementation

[0041] This application provides a method for inspection and fault diagnosis of new energy power stations based on the Internet of Things and artificial intelligence. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0042] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the new energy power station inspection and fault diagnosis method based on the Internet of Things and artificial intelligence in this application includes:

[0043] Step S1: Obtain high-precision map data of the new energy power station. The high-precision map data includes equipment location, road network and obstacles. Collect equipment information of the equipment to be inspected, plan inspection routes for intelligent inspection equipment, and carry out inspection of the new energy power station based on the inspection routes.

[0044] Specifically, to plan inspection routes for intelligent inspection equipment, enabling it to inspect various devices at new energy power stations using optimal routes, thereby improving inspection efficiency and saving costs, high-precision map data of the new energy power stations is first acquired. Equipment information, including location, model, and name, is also collected. Based on the locations of the devices to be inspected within the power station, the optimal number of intelligent inspection devices is determined through comparison and calculation. Inspection routes are then planned for the intelligent inspection equipment, ensuring that each step makes the best or optimal choice under the current conditions. This allows the intelligent inspection equipment to achieve a balance between maximizing the inspection rate and minimizing the number of inspections. Inspecting new energy power stations using this method improves inspection efficiency, reduces costs, and acquires image data of each device to provide data support for subsequent fault diagnosis.

[0045] Step S2: During the inspection, collect real-time image data of the equipment in the new energy power station and send the equipment image data to the cloud server. Install smart sensors on each piece of equipment to be inspected in the new energy power station and periodically collect the operating data of the equipment to be inspected. The operating data includes at least power generation, voltage, current and temperature, and send the operating data to the cloud server.

[0046] Specifically, in order to obtain relevant data on each piece of equipment to be inspected within the new energy power station, during the inspection process of the intelligent inspection equipment, real-time image data of each piece of equipment to be inspected within the new energy power station is collected and sent to the cloud server. This allows the cloud server to determine whether each piece of equipment to be inspected has experienced a primary fault based on the image data. The primary fault refers to a fault that can be detected by visual inspection. In the case of wind turbines, it can determine whether the wind turbines have detached, damaged, cracked, or tilted conditions. Furthermore, by installing intelligent sensors on each piece of equipment to be inspected, the operating data of the equipment to be inspected is periodically collected and sent to the cloud server. This facilitates the cloud server's analysis of the operating data to determine whether the equipment to be inspected has experienced a secondary fault. The secondary fault refers to operational faults of the equipment to be inspected, such as mechanical faults or power generation faults.

[0047] Step S3: The cloud server uses artificial intelligence to perform in-depth mining and analysis of the operating data and equipment image data to determine whether the equipment to be inspected has a fault. If a fault occurs, it also identifies the corresponding fault type, fault cause and fault location.

[0048] Specifically, to more quickly and accurately determine whether each piece of equipment to be inspected has malfunctioned, the cloud server pre-trains a first model based on historical equipment image data. Based on this first model and real-time collected current equipment image data, it determines whether the equipment to be inspected has experienced a first malfunction. It also trains a second model based on historical operational data. Based on this second model and real-time collected operational events, it determines whether the equipment to be inspected has experienced a second malfunction and obtains the corresponding event type. Event types include normal events and fault events. Since there are various types of fault events, each fault event includes different fault types. Subsequently, based on historically collected fault information, the corresponding fault cause can be identified, and based on the equipment information corresponding to the fault information, the corresponding fault location can be obtained. This provides assistance to relevant personnel in handling the fault, enabling rapid fault resolution. Through these methods, equipment anomalies or malfunctions can be detected in a timely manner, improving the accuracy and timeliness of fault diagnosis. Furthermore, it provides maintenance personnel with accurate fault diagnosis reports and maintenance suggestions.

[0049] Step S4: The real-time operation data and inspection results of the new energy power station are displayed in real time using visualization technology through terminal equipment. Historical operation data and historical inspection results can be viewed, and filtering can be performed by time and equipment type. When a fault is detected, an audible and visual warning will be issued in a timely manner, and the corresponding fault location, fault cause and response plan will be displayed.

[0050] Specifically, after the cloud server collects operational data and equipment image data, it determines whether each piece of equipment to be inspected has malfunctioned through step S3. If a malfunction is detected, the corresponding operational status, i.e., the type of malfunction of the equipment to be inspected, is statistically collected in real time. When staff view the data through terminal devices, the real-time operational data and inspection results of the new energy power station are displayed in real time using visualization technology. Historical operational data and historical inspection results can also be viewed, and filtering by time and equipment type is supported. When a malfunction is detected, an audible and visual warning is issued in a timely manner, and the corresponding malfunction location, cause, and response plan are displayed. Through the above method, relevant staff can view the operational status and inspection results of the new energy power station in real time through terminal devices, and perform maintenance management based on this information, thereby improving the operational efficiency and reliability of the new energy power station.

[0051] In one specific embodiment, planning an inspection route for the intelligent inspection equipment includes the following steps:

[0052] Step S11: Obtain the optimal number of intelligent inspection devices as the first quantity through calculation and comparison;

[0053] Step S12: Calculate the first distance from the inspection starting point to all uninspected equipment to be inspected, take the first number of smallest first distances corresponding to the equipment to be inspected as the first inspection target, and take the first inspection target as the first inspection target of each intelligent inspection device.

[0054] Step S13: For each intelligent inspection device, after inspecting one device to be inspected, send the device information and first information corresponding to the device to be inspected to the cloud server. Based on the cloud server, obtain all uninspected devices to be inspected. Then calculate the second distance from the currently inspected device to all uninspected devices to be inspected. Take the device to be inspected with the smallest second distance as the next inspection target of the intelligent inspection device. Repeat this step until all devices to be inspected have been inspected.

[0055] Specifically, to reduce inspection costs, the optimal number of intelligent inspection devices is determined through calculation and comparison. These devices may be inspection robots or drones, which are relatively expensive. Using too few devices may lead to a higher failure rate due to prolonged operation, and frequent malfunctions during inspections will significantly reduce efficiency. On the other hand, using too many devices, while reducing the load on individual devices, increases costs, and may not significantly improve efficiency. Therefore, it is necessary to comprehensively consider factors such as cost, efficiency, and device failure rate to determine the optimal number of intelligent inspection devices. This ensures that inspection tasks are completed efficiently with the minimum number of devices. The specific method for determining the optimal number will be explained later.

[0056] After obtaining the optimal number, inspection routes are planned for that optimal number of intelligent inspection devices. First, the first distance from the inspection starting point to all uninspected devices is calculated. The devices corresponding to the smallest of the first distances are taken as the first inspection targets. These first inspection targets are then used as the first inspection targets for each intelligent inspection device. For example, if the first number is 3, the 3 smallest first distances are obtained, and the devices corresponding to these 3 smallest first distances are identified. These devices are then used as the first inspection targets for each intelligent inspection device. After an intelligent inspection device completes the inspection of one device, it moves the corresponding device to the next inspection target. The corresponding equipment information and the first information are sent to the cloud server. The cloud server can record the equipment to be inspected based on the first information. Then, the intelligent inspection equipment obtains the remaining uninspected equipment, calculates the second distance from the currently inspected equipment to all uninspected equipment, and obtains the equipment to be inspected corresponding to the minimum second distance as the next inspection target of the intelligent inspection equipment. This step is repeated until all the equipment to be inspected in the new energy power station has been inspected. Through the above method, the intelligent inspection equipment can select the nearest equipment to be inspected at each step, thereby improving inspection efficiency.

[0057] In one specific embodiment, after the intelligent inspection device obtains the next inspection target, it further performs the following steps:

[0058] Step S131: Send the device information and second information of the intelligent inspection device corresponding to the next inspection target to the cloud server;

[0059] Step S132: If a fault occurs during the inspection process, the intelligent inspection equipment will send the equipment information and third-party information of the corresponding equipment to be inspected to the cloud server.

[0060] Step S133: The cloud server's storage unit stores a first list, a second list, and a third list. After receiving the first information, the cloud server deletes the device to be inspected corresponding to the first information from the second list and adds it to the third list. After receiving the second information, the cloud server deletes the device to be inspected corresponding to the second information from the first list and adds it to the second list. After receiving the third information, the cloud server removes the device to be inspected corresponding to the third information from the third list and adds it to the first list.

[0061] Specifically, since more than one intelligent inspection device is used, if the next inspection device detected by the intelligent inspection device is one that has already been inspected, or if the next inspection device detected by the intelligent inspection device is a device that another intelligent inspection device is about to inspect, multiple intelligent inspection devices will repeatedly inspect the same device. To avoid this situation, a first list, a second list, and a third list are maintained in the cloud server. The first list stores the device information of all devices that have not been inspected or locked. The second list stores the device information of devices that have not been inspected but have been locked. The third list stores the device information of devices that have been inspected.

[0062] The first message indicates that the corresponding equipment to be inspected has been inspected; the second message indicates that the corresponding equipment to be inspected has been locked by the intelligent inspection equipment; and the third message indicates that the equipment to be inspected has been unlocked by the intelligent inspection equipment.

[0063] During the inspection process, the intelligent inspection equipment may encounter a malfunction. At this time, the intelligent inspection equipment may not have finished inspecting the current equipment to be inspected. If it finds that it has a malfunction, it will send the equipment information and third-party information of the corresponding equipment to be inspected to the cloud server.

[0064] The above methods can prevent repeated inspections when the intelligent inspection equipment is inspecting the equipment to be inspected at the new energy power station. They also take into account the situation when the intelligent inspection equipment malfunctions, so that each piece of equipment to be inspected can be reliably inspected, thus improving inspection efficiency.

[0065] It is important to note that when assigning devices to be inspected to each intelligent inspection device for the first time, the device information and second information of the corresponding devices to be inspected should be sent to the cloud server to ensure that the cloud server can always record the inspection status of each device to be inspected.

[0066] In one specific embodiment, the optimal number of intelligent inspection devices is obtained by calculation and comparison as the first number, specifically including the following steps:

[0067] Step S111: Set the second quantity and the third quantity. The second quantity refers to the minimum number of intelligent inspection devices and the third quantity refers to the maximum number of intelligent inspection devices. Obtain the historical inspection data of the intelligent inspection devices and calculate the inspection completion rate of the intelligent inspection devices based on the historical inspection data.

[0068] Step S112: When the number of intelligent inspection devices is the second number, plan inspection routes for the second number of intelligent inspection devices, obtain the inspection distance of each intelligent inspection device based on the inspection route, obtain the corresponding inspection completion rate based on the inspection distance, calculate the overall completion rate based on the inspection completion rate, and also obtain the inspection time corresponding to the maximum inspection distance, and use the inspection time as the inspection time of the corresponding inspection route.

[0069] Step S113: Increment the second quantity by one, repeat step S112, obtain the new overall completion rate, until the second quantity equals the third quantity, and execute step S114.

[0070] Step S114: Obtain the second quantity corresponding to the overall completion rate being greater than the preset first threshold, and obtain the inspection plan corresponding to multiple second quantities. Also obtain the inspection time corresponding to the inspection plan, and take the second quantity corresponding to the inspection plan with the smallest inspection time as the optimal quantity.

[0071] Specifically, in order to obtain the optimal number of intelligent inspection devices, the minimum number can be set to 1, and the maximum number can be the historical maximum number of intelligent inspection devices, or it can be set based on the inspection budget. Since intelligent inspection devices may malfunction during the execution of tasks, causing inspection interruptions, and the inspection completion rate will decrease as the inspection distance of intelligent inspection devices increases, the inspection completion rate of intelligent inspection devices can be statistically analyzed based on historical inspection data, that is, to obtain the relationship between the inspection completion rate and the inspection distance.

[0072] Next, assuming the number of intelligent inspection devices is the second number, the method in the previous step is used to plan the inspection routes for the second number of intelligent inspection devices. Based on the inspection routes, the corresponding inspection distance is obtained. Based on the relationship between the inspection distance and the inspection completion rate, the corresponding inspection completion rate is obtained. Based on the inspection completion rate of each intelligent inspection device, the overall completion rate is calculated. Assuming that the moving speed of the intelligent inspection devices is the same, the inspection time required by the intelligent inspection device corresponding to the maximum inspection distance is taken as the inspection time of the overall inspection task.

[0073] Increment the second quantity by one and repeat step S112 to obtain the corresponding overall completion rate until the second quantity equals the third quantity. Then execute step S114 to obtain the overall completion rate corresponding to the use of different quantities of intelligent inspection equipment to complete the inspection task.

[0074] A reasonable first threshold for overall completion rate is preset (e.g., 96%). Only when the overall completion rate is greater than the first threshold does it meet the preset requirements. The inspection plan corresponding to the second number of intelligent inspection devices with an overall completion rate greater than the first threshold is obtained. For example, the second number of intelligent inspection devices that meet the requirements is 4, 5, or 6. Since there is a possibility of failure during the execution of inspection tasks by intelligent inspection devices, the more intelligent inspection devices there are, the greater the probability of failure. Therefore, the inspection time corresponding to different numbers of intelligent inspection devices is compared. The one with the most intelligent inspection devices may not have the shortest inspection time. So, the inspection plan corresponding to the number of intelligent inspection devices is obtained when there are 4, 5, or 6 devices. Then, the inspection time of the corresponding inspection plan is obtained. The second number of inspection plans corresponding to the shortest inspection time is taken as the optimal number.

[0075] It should be noted that the overall completion rate can be obtained by multiplying the completion rates of each intelligent inspection device.

[0076] By using the methods described above, the number of intelligent inspection devices can be reduced as much as possible while ensuring that inspection tasks are completed efficiently, thereby saving inspection costs.

[0077] In one specific embodiment, determining whether the equipment to be inspected has a fault includes the following steps:

[0078] Step S31: Obtain normal equipment image data of new energy power stations collected in history, and use the equipment image data of two adjacent collection cycles as learning data to train the first model;

[0079] Step S32: Obtain real-time collected device image data, and also obtain the device image data collected in the previous collection cycle as the first image data. Input the first image data into the first model, and input the predicted image data into the first model.

[0080] Step S33: Compare the real-time collected device image data and the predicted image data, calculate the first difference between the two, and if the first difference is greater than the preset second threshold, determine that the device to be inspected has a first fault.

[0081] Specifically, to determine whether the equipment to be inspected has malfunctioned, the system first acquires historically collected image data of the equipment at the new energy power station under normal conditions, i.e., image data of the equipment when no malfunctions have occurred. Since the equipment image data is collected periodically, multiple adjacent collection cycles' equipment image data are used as learning data to train the first model. The first model can predict the equipment image data for the next collection cycle based on the equipment image data collected in the previous collection cycle. The system then acquires the equipment image data for the current collection cycle and the equipment image data collected in the previous collection cycle of the current collection cycle as the first image data. The first image data is input into the first model, and the first model outputs predicted image data. The system compares the real-time collected equipment image data with the predicted image data and calculates the first difference between the two. If the first difference is greater than a preset second threshold, it indicates that the difference between the equipment image data collected in the current collection cycle and the expected equipment image data for the current collection cycle has exceeded the preset threshold, and there is a high probability that a first malfunction has occurred, i.e., an appearance malfunction. Therefore, it is determined that the corresponding equipment to be inspected has a first malfunction. The above method can quickly and accurately determine whether the equipment to be inspected has a first malfunction.

[0082] In one specific embodiment, determining whether the device has malfunctioned further includes the following steps:

[0083] Step S34: Obtain historically collected operational data, including normal operation data and abnormal operation data. Create a digital twin model of the equipment to be inspected based on the operational data and equipment-related information. Use the digital twin model to simulate the faults that occur in the equipment to be inspected. During the process of simulating fault events in the digital twin model, collect relevant abnormal operation data and abnormal operation data.

[0084] Step S35: Use artificial intelligence technology to analyze all abnormal operating data and abnormal operation data, generate a fault storage table based on the analysis results, use normal operating data and abnormal operating data as second learning data to train the second model, input the real-time acquired operating data into the second model, and let the second model determine whether the equipment to be inspected has a second fault.

[0085] Step S36: In the event of a second fault, based on the real-time collected operational data, query the fault storage table for fault cases that are the same as or similar to the current fault, and obtain a solution.

[0086] Specifically, to determine whether a second fault has occurred in the equipment to be inspected, which refers to a non-visual fault such as a mechanical or electrical fault, historical operational data is collected. This operational data includes normal operation data and abnormal operation data. In reality, we perform timely maintenance on the equipment to be inspected in new energy power plants, which may result in less actual abnormal operation data. Therefore, a digital twin model of the equipment to be inspected is created based on the operational data and equipment-related information. The digital twin model is used to simulate the fault that occurred in the equipment to be inspected. During the simulation of the fault time in the digital twin model, relevant abnormal operation data is collected. Then, artificial intelligence technology is used to classify all abnormal operation data, grouping abnormal data corresponding to the same or similar faults into the same category. The classified abnormal operation data is stored in a fault storage table. After obtaining enough abnormal operation data, the normal operation data and abnormal operation data are used as second learning data to train a second model. The second model outputs whether a second fault has occurred based on the real-time collected equipment operation data. In the case of a second fault, based on the real-time collected operation data, fault cases that are the same or similar to the current fault are queried from the fault storage table to obtain a response plan.

[0087] In one specific embodiment, artificial intelligence technology is used to analyze all abnormal operation data and abnormal data, and a fault storage table is generated based on the analysis results. The specific steps include:

[0088] The abnormal operating data generated when an actual fault occurs is called the first abnormal data, and the abnormal operating data generated when a simulated fault occurs is called the second abnormal data. The second abnormal data is classified, and the fault cause and fault type corresponding to the second abnormal data of the same type are obtained. The first abnormal data is classified into the type corresponding to the second abnormal data based on similarity. Historical maintenance records are obtained, and corresponding response solutions are obtained for different fault types. The abnormal operating data, the corresponding fault cause, fault type and response solution are associated and stored in the fault storage table.

[0089] Specifically, artificial intelligence technology is used to compare the actual abnormal operation data with the abnormal operation data simulated in the digital twin model. Since the abnormal operation data generated by the simulated fault is set by humans, the corresponding fault cause and fault type can be obtained based on the abnormal operation data. The abnormal operation data generated when the fault actually occurs and the abnormal operation data generated by the simulation are compared for similarity. Similar abnormal operation data are classified into the same category. The fault cause and fault type corresponding to the abnormal operation data generated by the simulation are used as the fault cause and fault type of the corresponding type. Then, the corresponding fault type response plan is obtained based on historical maintenance records. The classified abnormal operation data, fault cause, fault type and response plan are associated and stored in the fault storage table so that the response plan related to the fault can be quickly and accurately retrieved in the future.

[0090] In one specific embodiment, querying the fault storage table for fault cases that are the same as or similar to the current fault includes the following steps:

[0091] Retrieve all abnormal operation data for each type from the fault storage table. Normalize abnormal operation data belonging to the same type. Take the average value of each data parameter in the normalized abnormal operation data. Combine all average values ​​into feature data. Repeat this step to obtain feature data corresponding to all types. Generate target data by normalizing the abnormal operation data collected in real time. Calculate the similarity between the target data and multiple feature data. Use the response plan corresponding to the fault type of the feature data with the highest similarity as the response plan for the target data, providing a reference for relevant personnel when performing fault repair.

[0092] Specifically, the similarity between the feature data of the abnormal operation data collected in real time and the feature data of each type of abnormal operation data after classification is calculated. The abnormal operation data and the feature data with the maximum similarity are classified into the same category, and the corresponding fault type is taken as the fault type of the corresponding real-time collected abnormal operation data. Based on the obtained fault type, the corresponding response plan is obtained. The above method can accurately and quickly help staff obtain the response plan for the second fault, so that the fault can be resolved quickly.

[0093] If the above-described methods are implemented as 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 this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0094] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A new energy station inspection and fault diagnosis method based on Internet of Things and artificial intelligence, characterized in that, The method comprises the following steps: S1, obtaining high-precision map data of the new energy station, the high-precision map data comprising device position, road network and obstacles, collecting device information of the to-be-inspected device, planning an inspection route for the intelligent inspection device, and inspecting the new energy station based on the inspection route, the intelligent inspection device being a robot or a drone; S2, collecting device image data in the new energy station in real time during the inspection, and sending the device image data to a cloud server, installing an intelligent sensor on each to-be-inspected device in the new energy station, periodically collecting operation data of the to-be-inspected device, the operation data comprising power generation, voltage, current and temperature, and sending the operation data to the cloud server; S3, the cloud server uses artificial intelligence to deeply mine and analyze the operation data and the device image data, and judges whether the to-be-inspected device has a fault, and in the case of a fault, identifies the corresponding fault type, fault cause and fault location; S4, using a terminal device to display the real-time operation data and the inspection result of the new energy station in real time using a visualization technology, and providing viewing of historical operation data and historical inspection results, and providing filtering according to time and device type, and in the case of a fault, timely issuing an audible and visual warning, and displaying the corresponding fault location, fault cause and countermeasures; Planning an inspection route for the intelligent inspection device comprises calculating and comparing to obtain an optimal number of intelligent inspection devices as a first number; calculating a first distance from the inspection starting point to all to-be-inspected devices that have not been inspected, taking the to-be-inspected devices corresponding to the first number of smallest first distances as first inspection targets, and taking the first inspection targets as the first inspection targets of each intelligent inspection device; for each intelligent inspection device, after inspecting one to-be-inspected device, sending the device information and first information of the to-be-inspected device to the cloud server, obtaining all to-be-inspected devices that have not been inspected based on the cloud server, and then calculating a second distance from the currently inspected to-be-inspected device to all to-be-inspected devices that have not been inspected, taking the to-be-inspected device corresponding to the smallest second distance as the next inspection target of the intelligent inspection device, and repeating the step until all to-be-inspected devices are completely inspected.

2. The method of claim 1, wherein, After the intelligent inspection device obtains the next inspection target, it also performs the following steps: S131, sending the device information and second information of the intelligent inspection device corresponding to the next inspection target to the cloud server; S132, during the inspection, if a fault occurs, sending the device information and third information of the corresponding to-be-inspected device to the cloud server; S133, the cloud server has a first list, a second list and a third list stored in a storage unit, after receiving the first information, the cloud server deletes the to-be-inspected device corresponding to the first information from the second list and adds it to the third list, after receiving the second information, the cloud server deletes the to-be-inspected device corresponding to the second information from the first list and adds it to the second list, and after receiving the third information, the cloud server removes the to-be-inspected device corresponding to the third information from the third list and adds it to the first list.

3. The method of claim 2, wherein, The optimal number of intelligent inspection devices is obtained as the first number by calculation and comparison, including: S111, set the second number and the third number, the second number refers to the minimum number of intelligent inspection devices, the third number refers to the maximum number of intelligent inspection devices, obtain the historical inspection data of the intelligent inspection device, and calculate the inspection completion rate of the intelligent inspection device based on the historical inspection data; S112, in the case that the number of intelligent inspection devices is the second number, the inspection route is planned for the second number of intelligent inspection devices, the inspection distance of each intelligent inspection device is obtained based on the inspection route, the corresponding inspection completion rate is obtained based on the inspection distance, the overall completion rate is calculated based on the inspection completion rate, and the inspection time corresponding to the maximum inspection distance is obtained, which is taken as the inspection time corresponding to the inspection route; S113, add one to the second number, repeat S112, obtain a new overall completion rate, until the second number is equal to the third number, execute S114; S114, obtain the second number corresponding to the overall completion rate greater than the first threshold value, obtain the inspection scheme corresponding to the plurality of second numbers, and obtain the inspection time corresponding to the inspection scheme, the second number corresponding to the inspection scheme corresponding to the smallest inspection time is taken as the optimal number.

4. The method of claim 1, wherein, Determine whether the device is malfunctioning, including: S31, obtain the normal device image data of the new energy station collected historically, and take the device image data of two adjacent collection periods as learning data to train the first model; S32, obtain the real-time collected device image data, and obtain the device image data collected in the last collection period as the first image data, input the first image data into the first model, and input the predicted image data from the first model; S33, compare the real-time collected device image data and the predicted image data, calculate the first difference between the two, and if the first difference is greater than a preset second threshold, determine that the device to be inspected has a first fault.

5. The method of claim 1, wherein, Determine whether the device is malfunctioning, also including: S34, obtain the historical operation data, the operation data including normal operation data and abnormal operation data, create a digital twin model of the device to be inspected based on the operation data and device related information, simulate the fault of the device to be inspected using the digital twin model, and collect related abnormal operation data and abnormal operation data during the simulation of the fault event in the digital twin model; S35, analyze all abnormal operation data and abnormal operation data using artificial intelligence technology, generate a fault storage table based on the analysis result, take the normal operation data and abnormal operation data as second learning data, train a second model, input the real-time collected operation data into the second model, and determine whether the device to be inspected has a second fault based on the second model; S36, in the case of a second fault, query the same or similar fault cases from the fault storage table based on the real-time collected operation data, and obtain the response scheme.

6. The method according to claim 5, characterized in that, Analyze all abnormal operation data and abnormal operation data using artificial intelligence technology, generate a fault storage table based on the analysis result, including: The abnormal operation data generated when an actual fault occurs is referred to as first abnormal data, the abnormal operation data generated when a simulated fault occurs is referred to as second abnormal data, the second abnormal data is classified, the fault cause and the fault type corresponding to the second abnormal data belonging to the same type are obtained, the first abnormal data is divided into the types corresponding to the second abnormal data based on similarity, historical maintenance records are obtained, corresponding solutions for different fault types are obtained, and the abnormal operation data, the corresponding fault cause, the fault type and the solution are stored in a fault storage table.

7. The method according to claim 6, characterized in that, The same or similar fault cases as the current fault are queried from the fault storage table, including: All abnormal operation data of each type is obtained from the fault storage table, the abnormal operation data belonging to the same type is normalized, an average value of each data parameter in the normalized abnormal operation data is taken, all average values are combined into feature data, the feature data corresponding to all types is repeatedly obtained, target data is generated by normalizing the real-time collected abnormal operation data, the similarity between the target data and multiple feature data is calculated, and the solution corresponding to the fault type corresponding to the feature data corresponding to the maximum similarity is taken as the solution of the target data.

Images