Photovoltaic device fault diagnosis method, device, equipment, medium and program product

Abstract

The present disclosure relates to a photovoltaic device fault diagnosis method, device, equipment, medium and program product, comprising: determining multi-source heterogeneous data of a photovoltaic device and performing data standardization preprocessing; extracting abnormal features based on the multi-source heterogeneous data after standardization preprocessing, mapping to a pre-constructed fault knowledge graph to obtain an abnormal feature vector; performing reasoning based on the abnormal feature vector and the fault knowledge graph, performing fault identification and root cause analysis on the photovoltaic device to obtain a fault diagnosis result; based on a pre-constructed health degree evaluation index system, evaluating the fault diagnosis result to obtain a device health degree score of the photovoltaic device, determining a device maintenance time window, and generating preventive maintenance information. The present disclosure effectively realizes fault diagnosis of the photovoltaic device through multi-source heterogeneous data and fault identification and root cause analysis based on the fault knowledge graph.

Description

Technical Field

[0001] This disclosure relates to the field of photovoltaic fault diagnosis technology, and in particular to a photovoltaic equipment fault diagnosis method, apparatus, equipment, medium and program product. Background Technology

[0002] Driven by the global "dual carbon" strategy, photovoltaic power generation has become the mainstay of clean energy. However, the power generation losses caused by photovoltaic failures and other reasons are huge. Existing photovoltaic equipment fault diagnosis mainly relies on single-type data for diagnosis, which is not accurate enough. Moreover, existing fault diagnosis mostly uses simple rule combinations or machine learning models, which lacks the ability to logically reason about complex faults, cannot achieve effective root cause analysis, and is difficult to adapt to the management and control needs of large-scale photovoltaic power plants. Summary of the Invention

[0003] In order to at least partially solve one of the technical problems in the related technologies, this disclosure provides a method, apparatus, equipment, medium and program product for diagnosing photovoltaic equipment faults.

[0004] To achieve the above objectives, the first aspect of this exemplary embodiment provides a photovoltaic equipment fault diagnosis method, comprising: Identify the multi-source heterogeneous data of photovoltaic equipment and perform data standardization preprocessing; Based on standardized preprocessed multi-source heterogeneous data, abnormal features are extracted and mapped to a pre-constructed fault knowledge graph to obtain an abnormal feature vector. Based on the abnormal feature vector and the fault knowledge graph, reasoning is performed to identify faults and perform root cause analysis on the photovoltaic equipment, thereby obtaining fault diagnosis results. The fault diagnosis results are evaluated based on a pre-built health assessment index system to obtain the equipment health score of the photovoltaic equipment. Based on the historical records of the equipment health score, the equipment maintenance time window is determined and preventive maintenance information is generated.

[0005] In some exemplary embodiments, the multi-source heterogeneous data includes real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data; The process of determining the multi-source heterogeneous data of photovoltaic equipment and performing data standardization preprocessing includes: The multi-source heterogeneous data is processed by protocol conversion based on a standard format. Based on the analysis target time point and time window, the multi-source heterogeneous data is time-aligned. A device coding system and spatial topology model are constructed, a unique identifier is assigned to the photovoltaic device, and the infrared thermal imaging data and the visible light image data are mapped to the corresponding photovoltaic device for spatial alignment. Data quality checks are performed on the data collection points, and quality labels are attached to the data collection points based on the data quality checks. The data quality checks include value range checks, jump checks, and missing value checks.

[0006] In some exemplary embodiments, the fault knowledge graph includes six types of core entities and eight types of core relationships; The six core entities include equipment type entity, failure mode entity, symptom characteristic entity, environmental factor entity, treatment measure entity, and detection method entity; The eight core relationships include relationships that are manifested as relationships, induced relationships, processed relationships, occur in relationships, are detected in relationships, cause relationships, mutually exclusive relationships, and co-occurrence relationships. The pre-built fault knowledge graph includes: Based on equipment technical documents, the triplet relationship between equipment type, failure mode and treatment measures is determined through text parsing and natural language processing. Based on statistical analysis of historical failure cases, the confidence level and influence intensity parameters of the eight core relationships were determined. The eight core relationships were verified and supplemented based on structured questionnaires and knowledge interviews. An online learning and updating mechanism is established. Based on the online learning and updating mechanism, after the fault handling loop is completed, the six types of core entities and the eight types of core relationships are learned and updated. The fault knowledge graph is stored in a graph database.

[0007] In some exemplary embodiments, the step of extracting abnormal features based on standardized preprocessed multi-source heterogeneous data and mapping the abnormal features to a pre-constructed fault knowledge graph to obtain an abnormal feature vector includes: Based on standardized preprocessed real-time electrical operation data, electrical anomaly detection is performed. Based on the electrical anomaly detection results, a set of electrical symptom features is mapped. The electrical anomaly detection includes threshold detection, lateral comparison detection, and temporal pattern detection. Based on standardized preprocessed infrared thermal imaging data, infrared anomaly detection is performed. Based on the infrared anomaly detection results, a set of infrared symptom features is mapped. The infrared anomaly detection includes thermal image preprocessing, component region segmentation, and thermal anomaly detection. Based on standardized preprocessed visible light image data, a deep learning target detection model is used for defect identification and defect localization. Based on the defect identification and defect localization, a set of visible light symptom features is mapped. Based on the standardized preprocessed electrical characteristic test data, a measured electrical characteristic curve is generated, and the measured electrical characteristic curve is corrected to the standard test conditions to obtain a standard electrical characteristic curve. Based on the standard electrical characteristic curve, feature parameters are extracted and curve shape anomalies are identified. Based on the curve shape anomaly identification, a set of symptom features for electrical characteristic tests is obtained by mapping. Based on standardized preprocessed environmental meteorological data, meteorological anomaly detection is performed through correlation analysis, and a set of meteorological symptom features is mapped. The meteorological anomaly detection includes environmental condition labeling and environmental correlation degree calculation. Based on the electrical symptom feature set, the infrared symptom feature set, the visible light symptom feature set, the electrical characteristic test symptom feature set, and the meteorological symptom feature set, the abnormality quantification value and detection confidence of each symptom feature are determined; Based on the data channel identifier, the symptom characteristics, the abnormality quantification value, and the detection confidence level, a multi-channel abnormal feature vector is determined; If the same symptom feature is detected simultaneously by several data channels, evidence fusion is performed on the multi-channel confidence scores to obtain a fused confidence score. The detection confidence score is then replaced by the fused confidence score to determine the multi-channel abnormal feature vector.

[0008] In some exemplary embodiments, the step of reasoning based on the abnormal feature vector and the fault knowledge graph to perform fault identification and root cause analysis on the photovoltaic equipment and obtain fault diagnosis results includes: Based on the set of activated symptom features in the abnormal feature vector, a reverse query is performed through the relationship in the fault knowledge graph to determine the fault mode entity that generates the activated symptom features, and a candidate fault set is generated based on the fault mode entity. For each candidate fault pattern in the candidate fault set, the fault matching degree is calculated to determine the symptom coverage, symptom matching strength and environmental factor support. Based on the symptom coverage, the symptom matching strength and the environmental factor support, the matching degree score between each candidate fault pattern and the activated symptom feature is determined. Based on a preset matching degree threshold, candidate fault modes with high matching degree are determined. Based on the candidate fault modes with high matching degree, fault chains are traced in the opposite direction of the relationship in the fault knowledge graph to determine the upstream root cause fault mode and identify the root cause of the fault. Based on the mutual exclusion relationships in the fault knowledge graph and the matching score, unreasonable fault combinations in the candidate fault modes are eliminated. Based on the root cause of the fault, through spatiotemporal correlation aggregation and root cause correlation aggregation, several alarm events with the same root cause are aggregated into a main alarm event and several sub-alarm events to obtain the fault diagnosis result.

[0009] In some exemplary embodiments, the fault diagnosis results are evaluated based on a pre-built health assessment index system to obtain an equipment health score for the photovoltaic equipment. Based on the historical records of the equipment health score, an equipment maintenance time window is determined, and preventive maintenance information is generated, including: A four-dimensional health assessment index system, including performance, reliability, status, and risk dimensions, is pre-constructed. The fault diagnosis results are evaluated based on the four-dimensional health assessment index system, and the scores of each dimension are calculated. The equipment health score of the photovoltaic equipment is obtained through the preset dimension weights. Based on the historical records of the equipment health score, a health score change curve is fitted. Based on the health score change curve, the predicted time point when the equipment health score drops to a preset health score threshold is predicted. Based on the predicted time point, the equipment maintenance time window is determined. Based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information of the processing measures, preventive maintenance work order suggestions are generated.

[0010] Based on the same inventive concept, a second aspect of the exemplary embodiments of this disclosure provides a photovoltaic equipment fault diagnosis device, comprising: The determination module is used to determine the multi-source heterogeneous data of photovoltaic equipment and perform data standardization preprocessing. The first module is used to extract abnormal features based on standardized preprocessed multi-source heterogeneous data, and map the abnormal features to a pre-constructed fault knowledge graph to obtain an abnormal feature vector. The second module is used to perform reasoning based on the abnormal feature vector and the fault knowledge graph, to identify faults and perform root cause analysis on the photovoltaic equipment, and to obtain fault diagnosis results. The generation module is used to evaluate the fault diagnosis results based on a pre-built health assessment index system, obtain the equipment health score of the photovoltaic equipment, determine the equipment maintenance time window based on the historical records of the equipment health score, and generate preventive maintenance information.

[0011] The apparatus provided in the exemplary embodiments of this disclosure includes multi-source heterogeneous data such as real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data. The determining module is specifically used for: The multi-source heterogeneous data is processed by protocol conversion based on a standard format. Based on the analysis target time point and time window, the multi-source heterogeneous data is time-aligned. A device coding system and spatial topology model are constructed, a unique identifier is assigned to the photovoltaic device, and the infrared thermal imaging data and the visible light image data are mapped to the corresponding photovoltaic device for spatial alignment. Data quality checks are performed on the data collection points, and quality labels are attached to the data collection points based on the data quality checks. The data quality checks include value range checks, jump checks, and missing value checks.

[0012] The apparatus provided in the exemplary embodiments of this disclosure includes a first obtaining module, specifically used for: Based on standardized preprocessed real-time electrical operation data, electrical anomaly detection is performed. Based on the electrical anomaly detection results, a set of electrical symptom features is mapped. The electrical anomaly detection includes threshold detection, lateral comparison detection, and temporal pattern detection. Based on standardized preprocessed infrared thermal imaging data, infrared anomaly detection is performed. Based on the infrared anomaly detection results, a set of infrared symptom features is mapped. The infrared anomaly detection includes thermal image preprocessing, component region segmentation, and thermal anomaly detection. Based on standardized preprocessed visible light image data, a deep learning target detection model is used for defect identification and defect localization. Based on the defect identification and defect localization, a set of visible light symptom features is mapped. Based on the standardized preprocessed electrical characteristic test data, a measured electrical characteristic curve is generated, and the measured electrical characteristic curve is corrected to the standard test conditions to obtain a standard electrical characteristic curve. Based on the standard electrical characteristic curve, feature parameters are extracted and curve shape anomalies are identified. Based on the curve shape anomaly identification, a set of symptom features for electrical characteristic tests is obtained by mapping. Based on standardized preprocessed environmental meteorological data, meteorological anomaly detection is performed through correlation analysis, and a set of meteorological symptom features is mapped. The meteorological anomaly detection includes environmental condition labeling and environmental correlation degree calculation. Based on the electrical symptom feature set, the infrared symptom feature set, the visible light symptom feature set, the electrical characteristic test symptom feature set, and the meteorological symptom feature set, the abnormality quantification value and detection confidence of each symptom feature are determined; Based on the data channel identifier, the symptom characteristics, the abnormality quantification value, and the detection confidence level, a multi-channel abnormal feature vector is determined; If the same symptom feature is detected simultaneously by several data channels, evidence fusion is performed on the multi-channel confidence scores to obtain a fused confidence score. The detection confidence score is then replaced by the fused confidence score to determine the multi-channel abnormal feature vector.

[0013] The apparatus provided in the exemplary embodiments of this disclosure, wherein the second obtaining module is specifically used for: Based on the set of activated symptom features in the abnormal feature vector, a reverse query is performed through the relationship in the fault knowledge graph to determine the fault mode entity that generates the activated symptom features, and a candidate fault set is generated based on the fault mode entity. For each candidate fault pattern in the candidate fault set, the fault matching degree is calculated to determine the symptom coverage, symptom matching strength and environmental factor support. Based on the symptom coverage, the symptom matching strength and the environmental factor support, the matching degree score between each candidate fault pattern and the activated symptom feature is determined. Based on a preset matching degree threshold, candidate fault modes with high matching degree are determined. Based on the candidate fault modes with high matching degree, fault chains are traced in the opposite direction of the relationship in the fault knowledge graph to determine the upstream root cause fault mode and identify the root cause of the fault. Based on the mutual exclusion relationships in the fault knowledge graph and the matching score, unreasonable fault combinations in the candidate fault modes are eliminated. Based on the root cause of the fault, through spatiotemporal correlation aggregation and root cause correlation aggregation, several alarm events with the same root cause are aggregated into a main alarm event and several sub-alarm events to obtain the fault diagnosis result.

[0014] The apparatus provided in the exemplary embodiments of this disclosure, wherein the generation module is specifically used for: A four-dimensional health assessment index system, including performance, reliability, status, and risk dimensions, is pre-constructed. The fault diagnosis results are evaluated based on the four-dimensional health assessment index system, and the scores of each dimension are calculated. The equipment health score of the photovoltaic equipment is obtained through the preset dimension weights. Based on the historical records of the equipment health score, a health score change curve is fitted. Based on the health score change curve, the predicted time point when the equipment health score drops to a preset health score threshold is predicted. Based on the predicted time point, the equipment maintenance time window is determined. Based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information of the processing measures, preventive maintenance work order suggestions are generated.

[0015] Based on the same inventive concept, a third aspect of the exemplary embodiments of this disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect.

[0016] Based on the same inventive concept, a fourth aspect of the exemplary embodiments of this disclosure provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implement the steps of the method as described in the first aspect.

[0017] Based on the same inventive concept, a fifth aspect of the exemplary embodiments of this disclosure provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method as described in the first aspect.

[0018] The technical solution provided in this disclosure has the following advantages compared with the prior art: This disclosure achieves fault diagnosis for photovoltaic (PV) equipment by identifying multi-source heterogeneous data and performing data standardization preprocessing; extracting abnormal features from the standardized preprocessed data and mapping them to a pre-constructed fault knowledge graph to obtain abnormal feature vectors; performing inference based on the abnormal feature vectors and the fault knowledge graph to identify faults and conduct root cause analysis, resulting in fault diagnosis results; and evaluating the fault diagnosis results based on a pre-constructed health assessment index system to obtain a health score for the PV equipment, determine the equipment maintenance time window, and generate preventative maintenance information. This disclosure effectively achieves fault diagnosis for PV equipment through multi-source heterogeneous data and fault identification and root cause analysis based on a fault knowledge graph. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0020] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0022] Figure 1A schematic flowchart of a photovoltaic equipment fault diagnosis method provided for an exemplary embodiment of this disclosure; Figure 2 A schematic diagram of a photovoltaic equipment fault diagnosis device provided as an exemplary embodiment of this disclosure; Figure 3 A schematic diagram of the structure of an electronic device provided as an exemplary embodiment of the present disclosure. Detailed Implementation

[0023] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0024] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0025] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0026] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0027] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0028] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0029] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0030] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0031] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0032] To address the aforementioned issues, this disclosure provides an exemplary embodiment of a photovoltaic equipment fault diagnosis method, which will be described below with reference to specific exemplary embodiments.

[0033] This disclosure provides an exemplary embodiment of a photovoltaic equipment fault diagnosis method, such as... Figure 1 As shown, it includes: Step S101: Determine the multi-source heterogeneous data of the photovoltaic equipment and perform data standardization preprocessing.

[0034] In some exemplary embodiments, the multi-source heterogeneous data includes real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data; The process of determining the multi-source heterogeneous data of photovoltaic equipment and performing data standardization preprocessing includes: The multi-source heterogeneous data is processed by protocol conversion based on a standard format. Based on the analysis target time point and time window, the multi-source heterogeneous data is time-aligned. A device coding system and spatial topology model are constructed, a unique identifier is assigned to the photovoltaic device, and the infrared thermal imaging data and the visible light image data are mapped to the corresponding photovoltaic device for spatial alignment. Data quality checks are performed on the data collection points, and quality labels are attached to the data collection points based on the data quality checks. The data quality checks include value range checks, jump checks, and missing value checks.

[0035] In some exemplary embodiments, multi-source heterogeneous data is acquired based on different data channels, with different data channels corresponding to different data acquisition methods, data formats, and time resolutions.

[0036] In some exemplary embodiments, the data structure, data format, and update frequency of multi-source heterogeneous data differ. By standardizing preprocessing, the multi-source heterogeneous data is unified and standardized, thus solving the information dimension limitations of diagnosis from a single data source.

[0037] In some exemplary embodiments, data standardization preprocessing includes format unification, time alignment, spatial alignment, and data quality labeling.

[0038] In some exemplary embodiments, the equipment coding system is a coding system that uses a hierarchical structure of power plants, arrays, subarrays, inverters, strings, and components as its framework, and assigns a globally unique identifier to each photovoltaic device to achieve standardized management and rapid hierarchical traceability of photovoltaic devices.

[0039] In some exemplary embodiments, the spatial topology model is based on the actual physical layout of the photovoltaic power plant, and establishes a model of equipment coding system and mapping relationship between equipment and spatial coordinates, so as to accurately map the defect areas identified in infrared thermal imaging data and visible light image data to specific components, strings and inverters.

[0040] As an example, real-time electrical operation data is acquired through channel A. This data includes the inverter's AC side active power, reactive power, and power generation; the DC side string voltage, current, and power; the AC side voltage, current, and frequency; and the operating status signals (operation / shutdown / fault) of the inverter, transformer substation, and combiner box, as well as circuit breaker opening and closing signals and various alarm events actively reported by photovoltaic equipment. The real-time electrical operation data is acquired in real-time through the power plant's SCADA system's open interface (database interface or API), with a sampling frequency ranging from 1 second to 15 minutes. A single large power plant can generate hundreds of millions of data points daily. The data format includes structured time-series data and event-based alarm data.

[0041] As an example, infrared thermal imaging data is acquired via channel B. This data includes images of the surface temperature distribution of photovoltaic modules obtained by an infrared thermal imager mounted on a UAV, as well as periodic scan data from fixed infrared monitoring devices within the station. The spatial resolution of the images is no less than 640×512 pixels, and the temperature measurement accuracy is no less than ±2℃. The data format is an unstructured thermal image file (such as FLIR RAW or TIFF format), with accompanying metadata tags including shooting time, GPS location, ambient temperature, wind speed, etc.

[0042] As an example, visible light image data is acquired through channel C. This data includes frontal images of photovoltaic modules obtained by a high-resolution visible light camera mounted on a drone, as well as high-resolution satellite orthophotos. The resolution of the visible light images is no less than 1 cm / pixel, and they are used to identify visible defects on the module surface, such as microcracks, fragmentation, dirt and dust accumulation, obstructions (bird droppings, leaves), and junction box abnormalities. The data format is JPEG or TIFF image files.

[0043] As an example, electrical characteristic test data is acquired through channel D. This data includes current-voltage characteristic curves at the component or string level, insulation resistance test data, and grounding resistance test data. The current-voltage characteristic curves contain key characteristic parameters such as open-circuit voltage Voc, short-circuit current Isc, maximum power point voltage Vmpp and current Impp, and fill factor FF. The electrical characteristic test data can be acquired by a dedicated IV curve tester or the built-in scanning function of a smart combiner box, with an update frequency ranging from daily to monthly.

[0044] As an example, environmental meteorological data is acquired through channel E. This data includes total irradiance (GHI) (W / m²), diffuse irradiance (DHI) (W / m²), module backsheet temperature, ambient temperature, relative humidity, wind speed, wind direction, etc., which are monitored in real time by the meteorological station within the photovoltaic power station, as well as external numerical weather prediction (NWP) data. The meteorological data is acquired in real time from the on-site meteorological station via Modbus RTU / TCP or IEC104 protocol, with a sampling frequency of 1 to 5 minutes.

[0045] As an example, a pluggable protocol adapter framework can be used to uniformly convert heterogeneous data from different sources (Modbus, OPC UA, IEC104, HTTP REST API, MQTT, etc.) into the platform's internal standard JSON format.

[0046] As an example, different data channels have different sampling frequencies (from seconds to months). Therefore, based on the target analysis time point and time window, time alignment is performed on the multi-source heterogeneous data; to achieve the target analysis time point. Set the time window for each data channel around the central point. Time window for data channel A Time window of data channel B Time window of data channel C Time window of data channel D Time window of data channel E ; within their respective time windows Select the time point from the target analysis point The most recent valid data points are correlated and fused, and aligned with time; among them, the target time point is analyzed. It can be the moment the event occurs or a preset moment to be diagnosed.

[0047] As an example, quality labels are automatically attached to each data acquisition point (data recording units acquired from data channels, photovoltaic devices, and time points) using the following rules: Value range check: Power value out of range , mark ; Jump check: If the rate of change of data at adjacent time points exceeds the corresponding physical range (e.g., the rate of change of power per minute exceeds 50% of the rated power), it should be marked. ; Freeze test: continuous The values ​​of all data collection points are the same, labeled ; Missing data check: If no data is obtained within the preset time interval, mark it as missing. ; Based on this, it is marked as , The data is not used in subsequent fault diagnosis analysis and is marked as... The data needs to be cross-validated with data from other data channels before use.

[0048] In some exemplary embodiments, a fault knowledge graph is pre-built.

[0049] In some exemplary embodiments, the fault knowledge graph includes six types of core entities and eight types of core relationships; The six core entities include equipment type entity, failure mode entity, symptom characteristic entity, environmental factor entity, treatment measure entity, and detection method entity; The eight core relationships include relationships that are manifested as relationships, induced relationships, processed relationships, occur in relationships, are detected in relationships, cause relationships, mutually exclusive relationships, and co-occurrence relationships. The pre-built fault knowledge graph includes: Based on equipment technical documents, the triplet relationship between equipment type, failure mode and treatment measures is determined through text parsing and natural language processing. Based on statistical analysis of historical failure cases, the confidence level and influence intensity parameters of the eight core relationships were determined. The eight core relationships were verified and supplemented based on structured questionnaires and knowledge interviews. An online learning and updating mechanism is established. Based on the online learning and updating mechanism, after the fault handling loop is completed, the six types of core entities and the eight types of core relationships are learned and updated. The fault knowledge graph is stored in a graph database.

[0050] In some exemplary embodiments, a pre-built fault knowledge graph is used to systematically organize fault domain knowledge scattered in equipment manuals, historical maintenance records, expert experience and literature, providing a knowledge base for subsequent fault diagnosis.

[0051] In some exemplary embodiments, the equipment type entity is used to represent the types of equipment in a photovoltaic power plant; the failure mode entity is used to represent the types of failures that may occur in the equipment; the symptom characteristic entity is used to represent the abnormal behavior that can be observed when a failure occurs; the environmental factor entity represents the external environmental conditions that may induce or aggravate the failure; the treatment measure entity represents the repair or response plan for a specific failure; and the detection method entity represents the detection means used to verify the failure or quantify the degree of failure.

[0052] In some exemplary embodiments, the relationships are expressed as follows: A fault mode entity is related to a symptom feature entity, representing the symptom features exhibited by the fault mode; an induced relationship is related to an environmental factor entity and a fault mode entity, representing a fault mode that may be induced or exacerbated under environmental conditions; a treatment relationship is related to a treatment measure entity and a fault mode entity, representing a fault mode that can be repaired by the treatment measure; an occurrence relationship is related to a fault mode entity and an equipment type entity, representing the type of equipment on which the fault mode occurs; a detection relationship is related to a detection method entity and a fault mode entity, representing a fault mode that can be verified or quantified by the detection method; a causal relationship is related to a fault mode entity, representing the causal transmission relationship between fault modes, used to subsequently establish a fault chain; a mutual exclusion relationship is related to a fault mode entity, representing that two fault modes will not occur simultaneously under specific conditions, used to exclude unreasonable diagnostic combinations in the subsequent fault diagnosis process; and a co-occurrence relationship is related to a symptom feature entity, used to represent symptom features that appear together, used to assist in the identification of fault modes.

[0053] In some exemplary embodiments, knowledge sources include structured knowledge extraction, historical case mining, expert knowledge encoding, and continuous knowledge updating mechanisms.

[0054] In some exemplary embodiments, the equipment technical documentation includes technical manuals, fault code tables, and maintenance guides provided by the equipment manufacturer.

[0055] As an example, equipment type entities include photovoltaic modules (different types such as monocrystalline silicon, polycrystalline silicon, and thin film), inverters (string inverters, centralized inverters, and microinverters), combiner boxes (smart combiner boxes and ordinary combiner boxes), box-type transformers, support systems, cables, and connectors; each equipment type entity contains attributes such as: equipment category code, set of rated parameters (rated power, rated voltage, rated current, efficiency, etc.), manufacturer, and model specifications.

[0056] As an example, fault mode entities include component-level fault mode entities, inverter-level fault mode entities, combiner box-level fault mode entities, etc.; component-level fault mode entities include, but are not limited to, hot spot effect, PID decay (potential-induced decay), microcracks, snail trails, backplane cracking, junction box burnout, bypass diode failure, EVA yellowing, glass breakage, and component detachment; inverter-level fault mode entities include, but are not limited to, IGBT module breakdown, DC bus overvoltage, AC overcurrent, grounding fault, communication interruption, cooling fan failure, and capacitor aging; combiner box-level fault mode entities include, but are not limited to, fuse blowout, surge protector damage, loose terminals, and overheating; each fault mode entity contains attributes: fault code, fault name, fault description, severity level (fatal / critical / moderate / minor), and typical occurrence probability.

[0057] As an example, symptom-characteristic entities include electrical symptom-characteristic entities, thermal imaging symptom-characteristic entities, visible light symptom-characteristic entities, and electrical characteristic symptom-characteristic entities; electrical symptom-characteristic entities include low string current, low string voltage, reduced inverter efficiency, excessive power attenuation, and abnormal short-circuit current; thermal imaging symptom-characteristic entities include individual hot spots (local temperature 10°C higher than the surrounding area), strip-shaped thermal anomalies (overheated temperature of the entire row / column), overheating of junction boxes, and overheating of connectors; visible light symptom-characteristic entities include glass breakage, backplane blistering / cracking, severe dirt, visible snail trails, and yellowing / discoloration of EVA; electrical characteristic symptom-characteristic entities include stepped distortion of the IV curve, decreased fill factor FF, low open-circuit voltage Voc, increased series resistance Rs, and decreased parallel resistance Rsh.

[0058] As an example, environmental factors include high irradiance (>1000W / m²), high temperature environment (module temperature >70℃), high humidity (>85%RH), salt spray environment, sandstorm, hail, lightning, blizzard load, strong wind (>15m / s), local shading (tree shadow, bird droppings, dust accumulation), etc.

[0059] As an example, the handling measures entities include component replacement, bypass diode replacement, junction box repair / replacement, connector re-crimping, component cleaning, inverter module replacement, heat dissipation system cleaning, cable repair, tightening torque check and retightening, surge protector replacement, etc.; each handling measure includes attributes: measure code, required tool list, estimated man-hours, required spare parts list, and operational safety level.

[0060] As an example, detection methods include IV curve testing, infrared thermal imaging scanning, EL (electroluminescence) testing, insulation resistance measurement, grounding resistance measurement, visual inspection, and online voltage and current monitoring.

[0061] As an example, the relationship is as follows: hot spot effect → manifests as → single hot spot, hot spot effect → manifests as → low string current; each relationship has the following attributes: confidence (indicating the probability of the fault exhibiting this symptom, with a value of 0~1) and severity_mapping (describing the correspondence between the severity of the symptom and the severity of the fault).

[0062] As an example, the induced relationship is: local shading → induced → hot spot effect; high humidity → induced → PID attenuation. Relationship attributes include: influence strength weight (0~1) and description of the mechanism of action.

[0063] As an example, the relationship is: component replacement → processing → glass breakage, component cleaning → processing → dirt obscuring; the relationship attributes include: repair effectiveness (0~1) and priority.

[0064] As an example, the following relationships occur: PID decay → occurs in → P-type monocrystalline silicon module; the following relationship is detected: IV curve test → detected in → bypass diode failure; the following relationship causes: bypass diode failure → causes → hot spot effect, loose terminal block → causes → connector overheating → causes → junction box burnout.

[0065] As an example, fault cases that have been closed-loop processed are extracted from the power plant's historical operation and maintenance work order system. Each fault case includes the fault occurrence time, fault phenomenon description, detection data, fault cause determination, handling measures, and handling results. Through statistical analysis, the confidence level and influence strength parameters of each relationship are determined. Taking the manifestation relationship as an example, the confidence level of the manifestation relationship is calculated as follows: ; in, This represents the confidence level of the relationship, with a range of values. , indicating the first Failure Mode Entities With the Species of symptoms The relationship between them; Indicates the first Failure Mode Entities The first occurrence in historical failure cases Species of symptoms The number of times; Indicates the first Failure Mode Entities The total number of historical failure cases.

[0066] As an example, an online learning and updating mechanism for a fault knowledge graph is established. Whenever the operation and maintenance system completes a fault handling loop, the system automatically compares the fault, symptoms, causes, and handling mapping relationships in the case with the existing knowledge in the knowledge graph. If a new fault mode or new relationship appears, a knowledge update suggestion is generated and submitted to experts for review. If it is an existing relationship, the corresponding confidence and statistical parameters are automatically updated.

[0067] As an example, a fault knowledge graph is stored using a graph database (such as Neo4j) as the medium; entities are stored as graph nodes, relationships are stored as graph edges, and attributes are stored as attribute fields of nodes or edges; complex graph traversal and path query based on the Cypher query language are supported, providing knowledge retrieval capabilities for subsequent fault reasoning.

[0068] Step S102: Based on the standardized preprocessed multi-source heterogeneous data, extract abnormal features and map the abnormal features to the pre-constructed fault knowledge graph to obtain abnormal feature vectors.

[0069] In some exemplary embodiments, the step of extracting abnormal features based on standardized preprocessed multi-source heterogeneous data and mapping the abnormal features to a pre-constructed fault knowledge graph to obtain an abnormal feature vector includes: Based on standardized preprocessed real-time electrical operation data, electrical anomaly detection is performed. Based on the electrical anomaly detection results, a set of electrical symptom features is mapped. The electrical anomaly detection includes threshold detection, lateral comparison detection, and temporal pattern detection. Based on standardized preprocessed infrared thermal imaging data, infrared anomaly detection is performed. Based on the infrared anomaly detection results, a set of infrared symptom features is mapped. The infrared anomaly detection includes thermal image preprocessing, component region segmentation, and thermal anomaly detection. Based on standardized preprocessed visible light image data, a deep learning target detection model is used for defect identification and defect localization. Based on the defect identification and defect localization, a set of visible light symptom features is mapped. Based on the standardized preprocessed electrical characteristic test data, a measured electrical characteristic curve is generated, and the measured electrical characteristic curve is corrected to the standard test conditions to obtain a standard electrical characteristic curve. Based on the standard electrical characteristic curve, feature parameters are extracted and curve shape anomalies are identified. Based on the curve shape anomaly identification, a set of symptom features for electrical characteristic tests is obtained by mapping. Based on standardized preprocessed environmental meteorological data, meteorological anomaly detection is performed through correlation analysis, and a set of meteorological symptom features is mapped. The meteorological anomaly detection includes environmental condition labeling and environmental correlation degree calculation. Based on the electrical symptom feature set, the infrared symptom feature set, the visible light symptom feature set, the electrical characteristic test symptom feature set, and the meteorological symptom feature set, the abnormality quantification value and detection confidence of each symptom feature are determined; Based on the data channel identifier, the symptom characteristics, the abnormality quantification value, and the detection confidence level, a multi-channel abnormal feature vector is determined; If the same symptom feature is detected simultaneously by several data channels, evidence fusion is performed on the multi-channel confidence scores to obtain a fused confidence score. The detection confidence score is then replaced by the fused confidence score to determine the multi-channel abnormal feature vector.

[0070] In some exemplary embodiments, threshold detection is performed by setting static and dynamic thresholds for each battery parameter based on the device's rated parameters and operating procedures; lateral comparison detection is performed by calculating the relative deviation index of each device by utilizing the local lateral comparison relationship between several strings under the same inverter or between several inverters under the same array; and time-series pattern detection is performed by identifying abrupt anomalies (such as parameters suddenly dropping to zero), gradual anomalies (such as parameters continuously and slowly decreasing), and periodic anomalies (such as parameter decreases occurring at a specific time) based on a sliding window time series anomaly detection algorithm.

[0071] In some exemplary embodiments, when an anomaly is detected in data channels A to D, the environmental data before and after the time point of the anomaly are traced back, and the temporal correlation between each environmental factor and the anomaly event is calculated; if a specific environmental condition exists within the time window of the anomaly (such as a lightning event that occurred within 24 hours before the anomaly), then the environmental factor is included as a potential triggering factor in subsequent reasoning.

[0072] In some exemplary embodiments, multiple channels refer to data channels for acquiring data.

[0073] As an example, the set of electrical symptoms includes: {low string current, low string voltage, reduced inverter efficiency, excessive power attenuation, excessively high equipment temperature, communication interruption, ...}, with each symptom accompanied by its detection confidence and quantified abnormality value; the set of infrared symptoms includes: {individual hot spots, strip-shaped thermal anomalies, junction box overheating, connector overheating, overall high temperature}; the set of visible light symptoms includes: {glass breakage, visible cracks, snail trails, backplane abnormalities, severe dirt, EVA yellowing, component detachment}; the set of electrical characteristic test symptoms includes: {IV curve step distortion, decreased fill factor, low open circuit voltage, increased series resistance, decreased parallel resistance, excessive power attenuation}; and the set of meteorological symptoms includes: {environmental condition status labels, environmental correlation}.

[0074] As an example, key characteristic parameters are extracted from standard electrical characteristic curves (IV curves) to identify curve shape anomalies and abnormal patterns. These key characteristic parameters include open-circuit voltage. and its measured open-circuit voltage With nameplate value deviation ratio Short-circuit current and its deviation ratio Maximum power and its deviation ratio Fill factor Equivalent series resistance (Calculated using hyperbolic fitting method) Equivalent parallel resistance (Estimated by the inverse of the slope near the short-circuit point of the IV curve); Abnormal patterns include step-like distortion, overall downward shift, and slope anomalies.

[0075] As an example, a multi-channel abnormal feature vector is determined based on the data channel identifier, the activated symptom features, the abnormality quantification value, and the detection confidence level: ; in, Represents a multi-channel anomaly feature vector; Indicates the first One activated symptom feature; Indicates the first Detection confidence of each activated symptom feature; Indicates the first The quantification of the degree of abnormality of each activated symptom feature; Indicates the first The data channel identifier (A / B / C / D / E) corresponding to each activated symptom feature; This represents the total number of activated symptom features.

[0076] As an example, if the same symptom feature is detected simultaneously by several data channels, the confidence scores from multiple channels can be fused using the Dempster-Shafer evidence fusion theory: ; in, This represents the overall confidence level after fusing the confidence levels from multiple channels; Indicates the first The confidence level of independent detection of the same symptom feature for each data channel; Indicates the degree of disbelief for all data channels. The product of two products; This represents the conflict coefficient.

[0077] In the above formula This indicates the joint probability that the symptom was not detected across all data channels; the conflict coefficient. This formula is used to correct inconsistencies between different data channels; it enables multiple channels to corroborate each other, thereby improving the reliability of symptom characteristics.

[0078] As an example, multiple channels are determined based on data channels A, B, C, D, and E.

[0079] Step S103: Based on the abnormal feature vector and the fault knowledge graph, reasoning is performed to identify faults and perform root cause analysis on the photovoltaic equipment to obtain fault diagnosis results.

[0080] In some exemplary embodiments, the step of reasoning based on the abnormal feature vector and the fault knowledge graph to perform fault identification and root cause analysis on the photovoltaic equipment and obtain fault diagnosis results includes: Based on the set of activated symptom features in the abnormal feature vector, a reverse query is performed through the relationship in the fault knowledge graph to determine the fault mode entity that generates the activated symptom features, and a candidate fault set is generated based on the fault mode entity. For each candidate fault pattern in the candidate fault set, the fault matching degree is calculated to determine the symptom coverage, symptom matching strength and environmental factor support. Based on the symptom coverage, the symptom matching strength and the environmental factor support, the matching degree score between each candidate fault pattern and the activated symptom feature is determined. Based on a preset matching degree threshold, candidate fault modes with high matching degree are determined. Based on the candidate fault modes with high matching degree, fault chains are traced in the opposite direction of the relationship in the fault knowledge graph to determine the upstream root cause fault mode and identify the root cause of the fault. Based on the mutual exclusion relationships in the fault knowledge graph and the matching score, unreasonable fault combinations in the candidate fault modes are eliminated. Based on the root cause of the fault, through spatiotemporal correlation aggregation and root cause correlation aggregation, several alarm events with the same root cause are aggregated into a main alarm event and several sub-alarm events to obtain the fault diagnosis result.

[0081] As an example, a collection of activated symptom features based on multi-channel abnormal feature vectors. In the fault knowledge graph, a reverse query is performed along the relationships to identify fault mode entities that generate activated symptom features, thus generating a candidate fault set. : ; in, Represents the set of candidate faults; This indicates a relationship; Represents a fault knowledge graph.

[0082] As an example, symptom coverage: ; in, Indicates the failure mode; Indicates the fault mode Symptom coverage; Indicates the fault mode In fault knowledge graph A set of symptom features associated with the disease; This indicates a subset of the currently activated symptoms that belongs to the feature set of that symptom; This indicates the number of elements in the set.

[0083] Symptom match strength: ; in, Indicates the strength of symptom match; Indicates the first The overall confidence level of a symptom after multi-channel confidence fusion; Representing fault modes in a fault knowledge graph Symptoms The confidence level of the relationship.

[0084] Environmental factor support level: ; in, Indicates the degree of support for environmental factors; This represents the set of currently active environmental factor labels; This indicates the first environmental factor label in the set. One environmental factor; Representing environmental factors in a fault knowledge graph Induced Failure Mode Relationship confidence; This indicates taking the maximum value.

[0085] Match score: ; in, Indicates the match score; Weights representing symptom coverage; Weights representing the strength of symptom matching; The weights representing the degree of support for environmental factors.

[0086] In the above formula, symptom coverage Used to represent failure modes The actual observed probability of typical faults, and the range of values. The higher the value, the more severe the failure mode. The more actual manifestations; the stronger the symptom match. Used to represent failure modes With current symptoms Matching reliability, value range Environmental factor support indicates the current environment's influence on the failure mode. The degree of induction, and the range of values. The higher the environmental factor support, the more likely the current environment is to induce this failure mode. ; ,in, The preset value can be 0.4. The preset value can be 0.4. The preset value can be 0.2; the matching score is used to comprehensively quantify fault modes. The degree of matching with the current data; the higher the matching score, the more likely the fault is to be the root cause.

[0087] As an example, the preset matching score ranges from 0 to 100, and the preset matching threshold is 60. Matching scores higher than the matching threshold of 60 are recorded as high matching scores.

[0088] As an example, for fault modes with high matching degree, a graph traversal is performed in the reverse direction of the causal relationship in the fault knowledge graph to search for whether there is an upstream fault mode. This makes the fault chain If a fault chain exists, further examine the upstream root cause failure mode. The system checks whether the symptoms match those currently activated to identify the root cause of the fault. For example, when the currently activated symptom set is detected as {single hot spot, low string current, IV curve step distortion}: The candidate fault set includes: hot spot effect, component microcracks, bypass diode failure, etc. Tracing the cause relationship reveals: Bypass diode failure → causes → hot spot effect (because the failed bypass diode cannot protect the shielded battery string, causing the battery string to become a load, consume power, and generate hot spots). Based on environmental factor analysis, if the current environmental condition of "partial shading" exists, the fault chain will be further extended to: partial shading → induce → long-term overload of bypass diode → cause → bypass diode failure → cause → hot spot effect; The output fault root cause is "bypass diode failure" (direct cause) + "partial shading" (inducing factor), not just the surface fault phenomenon of "hot spot effect". At the same time, unreasonable fault combinations are eliminated by using mutual exclusion relationships. For example, if "component complete open circuit" and "string overcurrent" are both judged as candidate faults, they are marked as mutually exclusive in the knowledge graph, and the one with the lower matching degree is eliminated.

[0089] As an example, Indicates a relationship, Indicates induced relationship, Indicates handling relationships, Indicates occurrence in relation, Indicates detection of relationships, Indicates the initiation of a relationship, Indicates mutual exclusion. Indicates a co-occurrence relationship.

[0090] As an example, spatiotemporal correlation aggregation: alarm events within a time window (preset 30 minutes) and spatial correlation range (same inverter or adjacent strings) are aggregated into a group; root cause correlation aggregation: for alarm events within the same group, causal chain analysis is performed through the causal relationship in the fault knowledge graph, and alarm events with the same fault root cause are aggregated into a main alarm (displaying root cause information) and several sub-alarms (displaying specific symptoms).

[0091] Step S104: Evaluate the fault diagnosis results based on the pre-constructed health assessment index system to obtain the equipment health score of the photovoltaic equipment. Based on the historical records of the equipment health score, determine the equipment maintenance time window and generate preventive maintenance information.

[0092] In some exemplary embodiments, the fault diagnosis results are evaluated based on a pre-built health assessment index system to obtain an equipment health score for the photovoltaic equipment. Based on the historical records of the equipment health score, an equipment maintenance time window is determined, and preventive maintenance information is generated, including: A four-dimensional health assessment index system, including performance, reliability, status, and risk dimensions, is pre-constructed. The fault diagnosis results are evaluated based on the four-dimensional health assessment index system, and the scores of each dimension are calculated. The equipment health score of the photovoltaic equipment is obtained through the preset dimension weights. Based on the historical records of the equipment health score, a health score change curve is fitted. Based on the health score change curve, the predicted time point when the equipment health score drops to a preset health score threshold is predicted. Based on the predicted time point, the equipment maintenance time window is determined. Based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information of the processing measures, preventive maintenance work order suggestions are generated.

[0093] In some exemplary embodiments, the indicators of each dimension are normalized to a score of 0 to 100, and then the health of the photovoltaic equipment is obtained through preset dimension weights.

[0094] In some exemplary embodiments, the health status change curve is fitted based on a linear regression or exponential decay model.

[0095] As an example, calculate the equipment health score of photovoltaic equipment: ; in, This indicates the device's health score; Indicates the weight of the performance dimension; This indicates a score for the performance dimension; Indicates the weight of the reliability dimension; This indicates a score for the reliability dimension; Represents the weights of the state dimension; Represents the score for the state dimension; Indicates the weight of the risk dimension; This indicates the score for the risk dimension.

[0096] In the above formula, the weights of the performance dimension The weight can be preset to 0.35; this is the weight of the reliability dimension. The weight can be preset to 0.25; the weight of the state dimension. The weight can be preset to 0.25; the weight of the risk dimension. It can be preset to 0.15.

[0097] As an example, the device health rating is divided into the following levels: 90-100 points are healthy, 70-89 points are sub-healthy (recommend attention is recommended), 50-69 points are abnormal (recommend arranging an examination), and 0-49 points are serious (recommend immediate action is recommended).

[0098] As an example, the preset health score threshold is 70 points, and the predicted time point when the device health score drops below 70 points is given.

[0099] As an example, based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information (working hours, spare parts requirements) of the processing measures, preventive maintenance work order suggestions are generated. The preventive maintenance work order suggestions include the suggested maintenance time, maintenance content, list of required tools and spare parts, estimated working hours and costs, thereby realizing the transformation from scheduled maintenance to on-demand maintenance and from passive inspection to predictive maintenance.

[0100] The beneficial effects of this disclosure are: This disclosure integrates five types of data: real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data, to reflect the equipment status from different dimensions. Through confidence level fusion, it achieves multi-evidence collaborative diagnosis, effectively reducing the misdiagnosis rate and missed diagnosis rate.

[0101] This disclosure improves the adaptability of fault diagnosis by using protocol conversion processing, time alignment, spatial alignment, and data quality checks to make fault diagnosis compatible with data sources of different devices, protocols, and data formats.

[0102] This disclosure establishes the relationship between failure modes, symptom characteristics, environmental factors and treatment measures by constructing a failure knowledge graph, thereby enabling tracing the source along the causal chain.

[0103] This disclosure aggregates several alarm events with the same root cause into a single main alarm by using spatiotemporal correlation aggregation and root cause correlation aggregation based on fault knowledge graphs, thereby avoiding a large number of redundant alarms and improving fault handling efficiency.

[0104] This disclosure improves the ability to identify and diagnose fault modes through online learning and continuous updating of a fault knowledge graph.

[0105] This disclosure establishes a four-dimensional health assessment index system, fits a health change curve to predict equipment maintenance, generates preventive maintenance work order suggestions, and optimizes the allocation of operation and maintenance resources.

[0106] It should be noted that the method of this disclosure embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this disclosure embodiment, and the multiple devices will interact with each other to complete the method described.

[0107] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0108] To achieve the above embodiments, this disclosure also proposes a photovoltaic equipment fault diagnosis device.

[0109] Figure 2 This is a schematic diagram of a photovoltaic equipment fault diagnosis device provided in an embodiment of the present disclosure. The device 200 can be implemented by software and / or hardware and is generally integrated into an electronic device.

[0110] like Figure 2 As shown, the device 200 includes: a determining module 201, a first obtaining module 202, a second obtaining module 203, and a generating module 204, wherein, Module 201 is used to determine the multi-source heterogeneous data of photovoltaic equipment and perform data standardization preprocessing. The first module 202 is used to extract abnormal features based on standardized preprocessed multi-source heterogeneous data, and map the abnormal features to a pre-constructed fault knowledge graph to obtain an abnormal feature vector. The second module 203 is used to perform reasoning based on the abnormal feature vector and the fault knowledge graph, to identify faults and perform root cause analysis on the photovoltaic equipment, and to obtain fault diagnosis results. The generation module 204 is used to evaluate the fault diagnosis results based on a pre-built health assessment index system, obtain the equipment health score of the photovoltaic equipment, determine the equipment maintenance time window based on the historical records of the equipment health score, and generate preventive maintenance information.

[0111] The apparatus provided in the exemplary embodiments of this disclosure includes multi-source heterogeneous data such as real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data. The determining module 201 is specifically used for: The multi-source heterogeneous data is processed by protocol conversion based on a standard format. Based on the analysis target time point and time window, the multi-source heterogeneous data is time-aligned. A device coding system and spatial topology model are constructed, a unique identifier is assigned to the photovoltaic device, and the infrared thermal imaging data and the visible light image data are mapped to the corresponding photovoltaic device for spatial alignment. Data quality checks are performed on the data collection points, and quality labels are attached to the data collection points based on the data quality checks. The data quality checks include value range checks, jump checks, and missing value checks.

[0112] The apparatus provided in the exemplary embodiments of this disclosure includes a first obtaining module 202, which is specifically used for: Based on standardized preprocessed real-time electrical operation data, electrical anomaly detection is performed. Based on the electrical anomaly detection results, a set of electrical symptom features is mapped. The electrical anomaly detection includes threshold detection, lateral comparison detection, and temporal pattern detection. Based on standardized preprocessed infrared thermal imaging data, infrared anomaly detection is performed. Based on the infrared anomaly detection results, a set of infrared symptom features is mapped. The infrared anomaly detection includes thermal image preprocessing, component region segmentation, and thermal anomaly detection. Based on standardized preprocessed visible light image data, a deep learning target detection model is used for defect identification and defect localization. Based on the defect identification and defect localization, a set of visible light symptom features is mapped. Based on the standardized preprocessed electrical characteristic test data, a measured electrical characteristic curve is generated, and the measured electrical characteristic curve is corrected to the standard test conditions to obtain a standard electrical characteristic curve. Based on the standard electrical characteristic curve, feature parameters are extracted and curve shape anomalies are identified. Based on the curve shape anomaly identification, a set of symptom features for electrical characteristic tests is obtained by mapping. Based on standardized preprocessed environmental meteorological data, meteorological anomaly detection is performed through correlation analysis, and a set of meteorological symptom features is mapped. The meteorological anomaly detection includes environmental condition labeling and environmental correlation degree calculation. Based on the electrical symptom feature set, the infrared symptom feature set, the visible light symptom feature set, the electrical characteristic test symptom feature set, and the meteorological symptom feature set, the abnormality quantification value and detection confidence of each symptom feature are determined; Based on the data channel identifier, the symptom characteristics, the abnormality quantification value, and the detection confidence level, a multi-channel abnormal feature vector is determined; If the same symptom feature is detected simultaneously by several data channels, evidence fusion is performed on the multi-channel confidence scores to obtain a fused confidence score. The detection confidence score is then replaced by the fused confidence score to determine the multi-channel abnormal feature vector.

[0113] The apparatus provided in the exemplary embodiments of this disclosure, wherein the second obtaining module 203 is specifically used for: Based on the set of activated symptom features in the abnormal feature vector, a reverse query is performed through the relationship in the fault knowledge graph to determine the fault mode entity that generates the activated symptom features, and a candidate fault set is generated based on the fault mode entity. For each candidate fault pattern in the candidate fault set, the fault matching degree is calculated to determine the symptom coverage, symptom matching strength and environmental factor support. Based on the symptom coverage, the symptom matching strength and the environmental factor support, the matching degree score between each candidate fault pattern and the activated symptom feature is determined. Based on a preset matching degree threshold, candidate fault modes with high matching degree are determined. Based on the candidate fault modes with high matching degree, fault chains are traced in the opposite direction of the relationship in the fault knowledge graph to determine the upstream root cause fault mode and identify the root cause of the fault. Based on the mutual exclusion relationships in the fault knowledge graph and the matching score, unreasonable fault combinations in the candidate fault modes are eliminated. Based on the root cause of the fault, through spatiotemporal correlation aggregation and root cause correlation aggregation, several alarm events with the same root cause are aggregated into a main alarm event and several sub-alarm events to obtain the fault diagnosis result.

[0114] The apparatus provided in the exemplary embodiments of this disclosure, wherein the generation module 204 is specifically used for: A four-dimensional health assessment index system, including performance, reliability, status, and risk dimensions, is pre-constructed. The fault diagnosis results are evaluated based on the four-dimensional health assessment index system, and the scores of each dimension are calculated. The equipment health score of the photovoltaic equipment is obtained through the preset dimension weights. Based on the historical records of the equipment health score, a health score change curve is fitted. Based on the health score change curve, the predicted time point when the equipment health score drops to a preset health score threshold is predicted. Based on the predicted time point, the equipment maintenance time window is determined. Based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information of the processing measures, preventive maintenance work order suggestions are generated.

[0115] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing this disclosure, the functions of each module can be implemented in one or more software and / or hardware.

[0116] The photovoltaic equipment fault diagnosis device provided in this disclosure can execute the photovoltaic equipment fault diagnosis method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of executing the method.

[0117] To implement the above embodiments, this disclosure also proposes a computer program product, including a computer program / instructions, which, when executed by a processor, implements the photovoltaic equipment fault diagnosis method in the above embodiments.

[0118] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure.

[0119] The following is a detailed reference. Figure 3 The diagram illustrates a structural schematic suitable for implementing the electronic device in the embodiments of this disclosure. The electronic device in the embodiments of this disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0120] like Figure 3 As shown, the electronic device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a memory 1008 into a random access memory (RAM) 1003. The RAM 1003 also stores various programs and data required for the operation of the electronic device. The processor 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0121] Typically, the following devices can be connected to the I / O interface 1005: input devices 1006 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 1007 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; memory devices 1008 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0122] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 1009, or installed from memory 1008, or installed from ROM 1002. When the computer program is executed by processor 1001, it performs the functions defined in the photovoltaic equipment fault diagnosis method of embodiments of this disclosure.

[0123] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0124] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0125] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0126] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the aforementioned photovoltaic equipment fault diagnosis method.

[0127] Electronic devices can be programmed with computer program code in one or more programming languages ​​or combinations thereof to perform the operations of this disclosure. These programming languages ​​include, but are not limited to, object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0129] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0130] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0131] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0133] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0134] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for diagnosing faults in photovoltaic equipment, characterized in that, include: Identify the multi-source heterogeneous data of photovoltaic equipment and perform data standardization preprocessing; Based on standardized preprocessed multi-source heterogeneous data, abnormal features are extracted and mapped to a pre-constructed fault knowledge graph to obtain an abnormal feature vector. Based on the abnormal feature vector and the fault knowledge graph, reasoning is performed to identify faults and perform root cause analysis on the photovoltaic equipment, thereby obtaining fault diagnosis results. The fault diagnosis results are evaluated based on a pre-built health assessment index system to obtain the equipment health score of the photovoltaic equipment. Based on the historical records of the equipment health score, the equipment maintenance time window is determined and preventive maintenance information is generated.

2. The method according to claim 1, characterized in that, The multi-source heterogeneous data includes real-time electrical operation data, infrared thermal imaging data, visible light image data, electrical characteristic test data, and environmental meteorological data; The process of determining the multi-source heterogeneous data of photovoltaic equipment and performing data standardization preprocessing includes: The multi-source heterogeneous data is processed by protocol conversion based on a standard format. Based on the analysis target time point and time window, the multi-source heterogeneous data is time-aligned. A device coding system and spatial topology model are constructed, a unique identifier is assigned to the photovoltaic device, and the infrared thermal imaging data and the visible light image data are mapped to the corresponding photovoltaic device for spatial alignment. Data quality checks are performed on the data collection points, and quality labels are attached to the data collection points based on the data quality checks. The data quality checks include value range checks, jump checks, and missing value checks.

3. The method according to claim 1, characterized in that, The fault knowledge graph includes six core entities and eight core relationships; The six core entities include equipment type entity, failure mode entity, symptom characteristic entity, environmental factor entity, treatment measure entity, and detection method entity; The eight core relationships include relationships that are manifested as relationships, induced relationships, processed relationships, occur in relationships, are detected in relationships, cause relationships, mutually exclusive relationships, and co-occurrence relationships. The pre-built fault knowledge graph includes: Based on equipment technical documents, the triplet relationship between equipment type, failure mode and treatment measures is determined through text parsing and natural language processing. Based on statistical analysis of historical failure cases, the confidence level and influence intensity parameters of the eight core relationships were determined. The eight core relationships were verified and supplemented based on structured questionnaires and knowledge interviews. An online learning and updating mechanism is established. Based on the online learning and updating mechanism, after the fault handling loop is completed, the six types of core entities and the eight types of core relationships are learned and updated. The fault knowledge graph is stored in a graph database.

4. The method according to claim 1, characterized in that, Based on the standardized preprocessed multi-source heterogeneous data, abnormal features are extracted and mapped to a pre-constructed fault knowledge graph to obtain an abnormal feature vector, including: Based on standardized preprocessed real-time electrical operation data, electrical anomaly detection is performed. Based on the electrical anomaly detection results, a set of electrical symptom features is mapped. The electrical anomaly detection includes threshold detection, lateral comparison detection, and temporal pattern detection. Based on standardized preprocessed infrared thermal imaging data, infrared anomaly detection is performed. Based on the infrared anomaly detection results, a set of infrared symptom features is mapped. The infrared anomaly detection includes thermal image preprocessing, component region segmentation, and thermal anomaly detection. Based on standardized preprocessed visible light image data, a deep learning target detection model is used for defect identification and defect localization. Based on the defect identification and defect localization, a set of visible light symptom features is mapped. Based on the standardized preprocessed electrical characteristic test data, a measured electrical characteristic curve is generated, and the measured electrical characteristic curve is corrected to the standard test conditions to obtain a standard electrical characteristic curve. Based on the standard electrical characteristic curve, feature parameters are extracted and curve shape anomalies are identified. Based on the curve shape anomaly identification, a set of symptom features for electrical characteristic tests is obtained by mapping. Based on standardized preprocessed environmental meteorological data, meteorological anomaly detection is performed through correlation analysis, and a set of meteorological symptom features is mapped. The meteorological anomaly detection includes environmental condition labeling and environmental correlation degree calculation. Based on the electrical symptom feature set, the infrared symptom feature set, the visible light symptom feature set, the electrical characteristic test symptom feature set, and the meteorological symptom feature set, the abnormality quantification value and detection confidence of each symptom feature are determined; Based on the data channel identifier, the symptom characteristics, the abnormality quantification value, and the detection confidence level, a multi-channel abnormal feature vector is determined; If the same symptom feature is detected simultaneously by several data channels, evidence fusion is performed on the multi-channel confidence scores to obtain a fused confidence score. The detection confidence score is then replaced by the fused confidence score to determine the multi-channel abnormal feature vector.

5. The method according to claim 1, characterized in that, The reasoning based on the abnormal feature vector and the fault knowledge graph, performing fault identification and root cause analysis on the photovoltaic equipment, and obtaining fault diagnosis results include: Based on the set of activated symptom features in the abnormal feature vector, a reverse query is performed through the relationship in the fault knowledge graph to determine the fault mode entity that generates the activated symptom features, and a candidate fault set is generated based on the fault mode entity. For each candidate fault pattern in the candidate fault set, the fault matching degree is calculated to determine the symptom coverage, symptom matching strength and environmental factor support. Based on the symptom coverage, the symptom matching strength and the environmental factor support, the matching degree score between each candidate fault pattern and the activated symptom feature is determined. Based on a preset matching degree threshold, candidate fault modes with high matching degree are determined. Based on the candidate fault modes with high matching degree, fault chains are traced in the opposite direction of the relationship in the fault knowledge graph to determine the upstream root cause fault mode and identify the root cause of the fault. Based on the mutual exclusion relationships in the fault knowledge graph and the matching score, unreasonable fault combinations in the candidate fault modes are eliminated. Based on the root cause of the fault, through spatiotemporal correlation aggregation and root cause correlation aggregation, several alarm events with the same root cause are aggregated into a main alarm event and several sub-alarm events to obtain the fault diagnosis result.

6. The method according to claim 1, characterized in that, The fault diagnosis results are evaluated based on a pre-built health assessment index system to obtain an equipment health score for the photovoltaic equipment. Based on the historical records of the equipment health score, a maintenance time window is determined, and preventive maintenance information is generated, including: A four-dimensional health assessment index system, including performance, reliability, status, and risk dimensions, is pre-constructed. The fault diagnosis results are evaluated based on the four-dimensional health assessment index system, and the scores of each dimension are calculated. The equipment health score of the photovoltaic equipment is obtained through the preset dimension weights. Based on the historical records of the equipment health score, a health score change curve is fitted. Based on the health score change curve, the predicted time point when the equipment health score drops to a preset health score threshold is predicted. Based on the predicted time point, the equipment maintenance time window is determined. Based on the equipment maintenance time window, the processing relationships in the fault knowledge graph, and the attribute information of the processing measures, preventive maintenance work order suggestions are generated.

7. A photovoltaic equipment fault diagnosis device, characterized in that, include: The determination module is used to determine the multi-source heterogeneous data of photovoltaic equipment and perform data standardization preprocessing. The first module is used to extract abnormal features based on standardized preprocessed multi-source heterogeneous data, and map the abnormal features to a pre-constructed fault knowledge graph to obtain an abnormal feature vector. The second module is used to perform reasoning based on the abnormal feature vector and the fault knowledge graph, to identify faults and perform root cause analysis on the photovoltaic equipment, and to obtain fault diagnosis results. The generation module is used to evaluate the fault diagnosis results based on a pre-built health assessment index system, obtain the equipment health score of the photovoltaic equipment, determine the equipment maintenance time window based on the historical records of the equipment health score, and generate preventive maintenance information.

8. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores a computer program / instruction thereon, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, characterized in that, Includes a computer program / instructions that, when executed by a processor, implement the steps of the method according to any one of claims 1 to 6.

Images