Energy efficiency diagnosis method, system and medium for distributed photovoltaic power generation equipment

Abstract

The present application proposes a method, system and medium for energy efficiency diagnosis of distributed photovoltaic power generation equipment, which includes: obtaining energy efficiency fault knowledge of distributed photovoltaic power generation equipment and generating an energy efficiency diagnosis knowledge base of the main body; obtaining historical operation data of each device and determining the reference interval of detection data of each device under different working conditions; collecting real-time data of each device and performing abnormal detection on the real-time data based on the corresponding reference interval; if abnormal data is detected, determining the signs of the abnormal data, and performing reasoning diagnosis through the corresponding probability relationship between the signs preset in the main body energy efficiency diagnosis knowledge base and the fault mode, and determining the abnormal mode and fault of the distributed photovoltaic power generation equipment; sending the diagnosis results and maintenance strategies to the client of the operation and maintenance personnel. The method can quickly locate the faulty distributed photovoltaic power generation equipment and the specific fault, and provide corresponding maintenance strategies to reduce the operation and maintenance costs and improve the operation and maintenance efficiency.

Description

Technical Field

[0001] The present application relates to the field of photovoltaic power generation technology, and in particular to an energy efficiency diagnosis method, system and medium for distributed photovoltaic power generation equipment. Background Art

[0002] At present, it is common to generate electricity in areas with abundant sunlight resources through distributed photovoltaic power generation technology, and it is connected to the grid or used for self-use. This type of distributed photovoltaic power station is usually divided into an area for unified management, so it is different from a centralized photovoltaic power station. The various power generation units of a centralized photovoltaic power station are concentrated in a flat area, and are generally open spaces without obstructions, so the environmental conditions of each power generation device in the area are similar, while the environmental conditions of each power generation device in a distributed photovoltaic power station area vary greatly. For example, the environmental conditions of the power generation equipment on the top of the mountain and the power generation equipment at the foot of the mountain in the regional photovoltaic power station in the southwestern mountainous area are completely different. Therefore, the operation and maintenance of this type of distributed photovoltaic power station is relatively complex, but distributed photovoltaic power stations usually need to control labor costs and develop intelligent operation and maintenance systems to reduce operation and maintenance costs.

[0003] In the related technology, the abnormal state monitoring and diagnosis methods of centralized photovoltaic power stations mainly include the following three types: The first method is to use sensors to detect the current, voltage or electrical characteristics of the photovoltaic string, as well as infrared and other characteristics to detect faults. The second method focuses on analyzing the power loss of the power station. For example, a supervision and fault detection method is proposed based on power loss analysis. By analyzing the losses in the system and comparing the error deviation of DC current and voltage, fault identification is achieved. The third method determines the fault by comparing the real-time operation data with the model output obtained by historical data modeling.

[0004] However, the applicant found that due to the dispersion of distributed photovoltaic power station power generation equipment and relatively low construction costs, when applying the abnormal state monitoring and diagnosis methods of centralized photovoltaic power stations to distributed photovoltaic power station power generation equipment, the first two methods mentioned above lack suitable monitoring instruments and cannot be applied to the technical requirements of fault detection methods for distributed photovoltaic power generation with huge differences in environmental conditions, and the third intelligent method also lacks training samples. Therefore, the abnormal state monitoring and diagnosis methods of photovoltaic power stations in the current related technologies, such as the most widely used early warning method of horizontal comparison between a series of devices such as detecting abnormal discrete rate of combiner box strings and abnormal inverter efficiency, are not applicable in the operation and maintenance of distributed photovoltaic power stations. Summary of the invention

[0005] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0006] To this end, the first purpose of this application is to propose an energy efficiency diagnosis method for distributed photovoltaic power generation equipment, which can quickly locate the information of the faulty generator set and the corresponding specific fault, and provide corresponding health maintenance strategies, and support operation and maintenance personnel to locate abnormal photovoltaic power generation devices through mobile tools, query the unit failure mode and maintenance measures, and grasp the operating conditions of all generator sets in the distributed power station area, saving manpower, reducing the equipment and data costs required for operation and maintenance, and improving operation and maintenance efficiency.

[0007] The second object of this application is to propose an energy efficiency diagnostic system for distributed photovoltaic power generation equipment;

[0008] A third object of the present application is to provide a non-transitory computer-readable storage medium.

[0009] To achieve the above-mentioned purpose, the first embodiment of the present application is to propose an energy efficiency diagnosis method for distributed photovoltaic power generation equipment, the method comprising the following steps:

[0010] Analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate an ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment;

[0011] Acquire historical operation data of each photovoltaic power generation device, and determine a reference interval of detection data of each photovoltaic power generation device under different working conditions according to the historical operation data;

[0012] Collecting real-time operating data of each photovoltaic power generation device, and performing abnormality detection on the real-time operating data based on the corresponding reference interval;

[0013] If the real-time operating data of any photovoltaic power generation equipment is abnormal data, the signs of the abnormal data are determined, and the any distributed photovoltaic power generation equipment is diagnosed by reasoning through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base to determine the abnormal mode and fault mode of any distributed photovoltaic power generation equipment;

[0014] The diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the main body energy efficiency diagnosis knowledge base are sent to the mobile client of the operation and maintenance personnel.

[0015] Optionally, in one embodiment of the present application, an ontological energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment is generated, including: converting the energy efficiency fault knowledge into a structured language recognizable by a computer; based on the structured language, modeling is performed through a preset open source tool to generate an ontology-based energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation group; historical fault data of each of the distributed photovoltaic power generation equipment is obtained, and based on the Bayesian theorem, a priori probability assignment is performed on the fault modes corresponding to different symptoms to generate the corresponding probability relationship between the preset symptoms and the fault modes.

[0016] Optionally, in one embodiment of the present application, a reference interval of the detection data of each of the photovoltaic power generation equipment under different operating conditions is determined based on the historical operating data, including: performing steady-state screening on the historical operating data to screen out non-steady-state data in the historical operating data; using irradiance and temperature as boundary conditions, dividing the operating conditions of each of the photovoltaic power generation equipment through a K-means clustering algorithm; eliminating fault data under each operating condition through a multivariate Gaussian mixture model to generate normal data training samples; training a long short-term memory artificial neural network LSTM through the normal data training samples, and determining the reference interval corresponding to the boundary conditions of each of the photovoltaic power generation equipment under each operating condition based on the predicted value output by the long short-term memory artificial neural network LSTM.

[0017] Optionally, in one embodiment of the present application, based on the parameter prediction values ​​output by the long short-term memory artificial neural network LSTM, a reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each operating condition is determined, including: for each parameter, obtaining multiple parameter prediction values ​​output by the long short-term memory artificial neural network LSTM, and the corresponding actual values; calculating the difference between each parameter prediction value and the actual value, taking the parameter prediction value with the largest difference with the actual value as the target parameter prediction value; constructing the reference interval with the target parameter prediction value as the center and twice the difference between the target parameter prediction value and the actual value as the interval length.

[0018] Optionally, in one embodiment of the present application, after sending the diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the ontology energy efficiency diagnosis knowledge base to the mobile client of the operation and maintenance personnel, it also includes: adding a fault record on the mobile client of the operation and maintenance personnel according to the fault mode that occurred; updating the posterior probability of the corresponding probability relationship between the preset symptom and the fault mode according to the acquired fault record.

[0019] Optionally, in one embodiment of the present application, adding a fault record on the mobile client of the operation and maintenance personnel includes: receiving the occurred fault mode, the cause of the fault, the fault symptoms and the corresponding maintenance measures sent by the mobile client; updating the posterior probability of the corresponding probability relationship between the preset symptoms and the fault mode according to the acquired fault record, including: judging whether the occurred fault mode exists, if the occurred fault mode exists, updating the posterior probability; if the occurred fault mode does not exist, adding the occurred fault mode to the ontology energy efficiency diagnosis knowledge base.

[0020] To achieve the above purpose, the second embodiment of the present application further proposes an energy efficiency diagnosis system for distributed photovoltaic power generation equipment, comprising the following modules:

[0021] A generation module is used to analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate an ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment;

[0022] A determination module, used to obtain historical operation data of each photovoltaic power generation device, and determine a reference interval of detection data of each photovoltaic power generation device under different working conditions according to the historical operation data;

[0023] an abnormality detection module, used to collect the real-time operation data of each photovoltaic power generation device, and perform abnormality detection on the real-time operation data based on the corresponding reference interval;

[0024] An inference diagnosis module is used to determine the signs of the abnormal data if the real-time operation data of any photovoltaic power generation equipment is abnormal data, and to perform inference diagnosis on any distributed photovoltaic power generation equipment through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base, so as to determine the abnormal mode and the fault mode of any distributed photovoltaic power generation equipment;

[0025] The sending module is used to send the diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the main body energy efficiency diagnosis knowledge base to the mobile client of the operation and maintenance personnel.

[0026] Optionally, in one embodiment of the present application, a generation module is specifically used to: convert the energy efficiency fault knowledge into a structured language recognizable by a computer; based on the structured language, modeling is performed through a preset open source tool to generate an ontology-based energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation group; historical fault data of each of the distributed photovoltaic power generation equipment is obtained, and based on the Bayesian theorem, a priori probability assignment is performed on the fault modes corresponding to different symptoms to generate the corresponding probability relationship between the preset symptoms and the fault modes.

[0027] Optionally, in one embodiment of the present application, a determination module is specifically used to: perform steady-state screening on the historical operating data to screen out non-steady-state data in the historical operating data; use irradiance and temperature as boundary conditions to divide the operating conditions of each photovoltaic power generation device through a K-means clustering algorithm; eliminate fault data under each operating condition through a multivariate Gaussian mixture model to generate normal data training samples; train a long short-term memory artificial neural network LSTM through the normal data training samples, and determine the reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each operating condition based on the predicted value output by the long short-term memory artificial neural network LSTM.

[0028] The technical solution provided by the embodiment of the present application brings at least the following beneficial effects: the present application forms the fault knowledge of abnormal energy efficiency level of photovoltaic generator set through FMEA and FTA analysis methods, and then based on the concept of ontology, expresses the unstructured fault knowledge of photovoltaic generator set through computer-recognizable structured language to form an open ontology knowledge base. When performing energy efficiency diagnosis on real-time data, through the distributed photovoltaic generator set energy efficiency diagnosis and health maintenance process of "abnormal detection-determining signs-diagnostic reasoning-locating faults-health maintenance decision-mobile user reading", abnormal detection is performed according to the benchmark interval determined by the environmental characteristics of each device, and then the abnormal information corresponding to the signs is input into the energy efficiency diagnosis knowledge base, and the reasoning diagnosis work based on the Bayesian model is carried out, so that the faulty generator set information and the corresponding specific faults can be quickly located, and the corresponding health maintenance strategy can be provided. In addition, through the mobile client interface, the operation and maintenance personnel can access the relevant unit information according to the generator set number through the mobile terminal device, query the unit failure mode and maintenance measures, master the operation status of all generator sets in the distributed power station area, make maintenance plans in advance, reduce the operation and maintenance costs, and improve the operation and maintenance efficiency.

[0029] In order to implement the above embodiments, the third aspect of the present application also proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the energy efficiency diagnosis method of the distributed photovoltaic power generation equipment in the above embodiments is implemented.

[0030] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0032] Figure 1 A flow chart of an energy efficiency diagnosis method for distributed photovoltaic power generation equipment proposed in an embodiment of the present application;

[0033] Figure 2 A flowchart of a method for determining a reference interval proposed in an embodiment of the present application;

[0034] Figure 3 A schematic diagram of a flow chart of a specific energy efficiency diagnosis method for distributed photovoltaic power generation equipment proposed in an embodiment of the present application;

[0035] Figure 4 A schematic diagram of the structure of an energy efficiency diagnostic system for distributed photovoltaic power generation equipment proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0036] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0037] It should be noted that in the later stage of distributed photovoltaic operation and maintenance, the intelligent operation and maintenance system in the relevant technology only stops at locating the abnormality of one or more parameters and issuing warnings to remind the operation and maintenance personnel to carry out maintenance. This intelligent operation and maintenance system cannot meet the actual needs of distributed photovoltaic operation and maintenance. Therefore, this application proposes an energy efficiency diagnosis method that can intelligently locate specific fault modes and provide corresponding maintenance strategies for operation and maintenance personnel.

[0038] Among them, considering that the differences between the various power generation equipment in the distributed photovoltaic power station are too large and the subsequent operation and maintenance costs are limited, there are certain differences in the power generation power, voltage, current and other data between different generator sets in the regional power station, and they cannot be compared centrally. This application determines the respective operating parameter benchmark intervals and establishes their respective energy efficiency diagnosis knowledge bases based on the differences between the various generator sets in the distributed photovoltaic power station area. In addition, the current new energy intelligent operation and maintenance system stops at status monitoring and fault warning, and this application also proposes methods for fault diagnosis and health maintenance. Traditional fault diagnosis methods are usually divided into several categories based on data, knowledge, signals, models, etc. This application combines data and knowledge for the particularity of distributed photovoltaic power generation groups, and completes the real-time diagnosis process of distributed photovoltaic power generation groups "abnormal detection-determination of signs-diagnostic reasoning-fault location-health maintenance decision".

[0039] The following describes an energy efficiency diagnosis method and system for distributed photovoltaic power generation equipment proposed in an embodiment of the present invention with reference to the accompanying drawings.

[0040] Figure 1 This is a flow chart of an energy efficiency diagnosis method for distributed photovoltaic power generation equipment proposed in an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0041] Step 101, analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate an ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment.

[0042] The distributed photovoltaic power generation equipment to be diagnosed can be the various generator sets in the distributed photovoltaic power station or the various devices in the unit, which can be set according to the needs of operation and maintenance. Failure Mode and Effects Analysis (FMEA) and Fault Tree Analysis (FTA) are two methods for analyzing complex systems to find potential failures.

[0043] In the embodiment of the present application, the distributed photovoltaic power generation group is subjected to mechanism analysis and equipment analysis through FMEA and FTA analysis methods, and the common related fault knowledge that causes abnormal energy efficiency level of photovoltaic power generation equipment is sorted out. The above-mentioned related fault knowledge is the energy efficiency fault knowledge of distributed photovoltaic power generation equipment. Since the FMEA and FTA analysis methods are complementary in analysis methods, the energy efficiency fault knowledge of distributed photovoltaic power generation equipment can be analyzed more fully and comprehensively through these two methods.

[0044] Furthermore, since the energy efficiency fault knowledge of the generator set obtained through the analysis method is unstructured, in order to facilitate subsequent diagnosis, in an embodiment of the present application, based on the concept of ontology, an ontology energy efficiency diagnosis knowledge base of distributed photovoltaic power generation equipment is generated according to the obtained energy efficiency fault knowledge.

[0045] Among them, the energy efficiency diagnosis knowledge ontology of distributed photovoltaic power generation equipment is a set of terms that semantically represents the knowledge of the causal chain of fault evolution of distributed photovoltaic power generation equipment. The ontology energy efficiency diagnosis knowledge base of distributed photovoltaic power generation equipment contains various energy efficiency fault knowledge, for example, it can include various types of knowledge such as fault mode, fault symptoms, fault causes, fault impact and maintenance measures.

[0046] In specific implementation, as a possible implementation method, the energy efficiency fault knowledge can be first converted into a structured language that can be recognized by a computer, and then based on the structured language, modeling can be performed through a preset open source tool to generate an energy efficiency diagnosis knowledge base of distributed photovoltaic power generation groups based on ontology. At the same time, the historical fault data of each distributed photovoltaic power generation equipment can be obtained, and the prior probability of the fault mode corresponding to different symptoms can be assigned based on the Bayesian theorem to generate the corresponding probability relationship between symptoms and fault modes.

[0047] Specifically, the preset open source tool can be "protégé", or other tools can be selected as needed. In this embodiment, the unstructured photovoltaic power generation group related fault knowledge is expressed in a structured language that can be recognized by a computer, and the open source tool protégé is used for modeling to form an ontology-based distributed photovoltaic power generation group energy efficiency diagnosis knowledge base. When using the open source tool protégé for ontology modeling, it is also possible to use the historical failure cases of distributed photovoltaic power generation equipment, combined with the Bayesian principle, and use data attributes to complete the prior probability assignment of "symptom-failure mode".

[0048] Step 102 : acquiring historical operation data of each photovoltaic power generation device, and determining a reference interval of detection data of each photovoltaic power generation device under different working conditions according to the historical operation data.

[0049] The historical operation data may be the historical operation data of each photovoltaic power generation device under the operating state within a preset time, for example, in the past year. In the embodiment of the present application, a benchmark model may be trained based on the data under the normal operating state in the historical operation data, that is, each device monitors the operating parameter operating benchmark interval under different working conditions, different irradiation and different temperature conditions. This facilitates subsequent abnormal detection of real-time data based on the benchmark interval.

[0050] In one embodiment of the present application, in order to more clearly explain the method of determining the basic interval, the present application also proposes a specific method for determining the reference interval. Figure 2 A flowchart of a method for determining a reference interval proposed in an embodiment of the present application is shown in FIG. Figure 2 As shown, the method includes:

[0051] Step 201, perform steady-state screening on historical operation data to filter out non-steady-state data in the historical operation data.

[0052] Among them, non-steady-state data is data generated by photovoltaic power generation equipment under abnormal operating conditions due to sudden situations such as shading or covering. In the embodiment of the present application, the operation data of distributed photovoltaic power generation groups in the past year are screened in a steady state to eliminate non-steady-state data caused by sudden situations such as shading and covering.

[0053] It should be noted that due to the large differences between the distributed photovoltaic devices in a distributed photovoltaic power station, such as different environments, etc., in view of the differences between the devices, in one embodiment of the present application, an energy efficiency fault diagnosis knowledge base for each distributed photovoltaic device can be established. By mining the knowledge of the historical operating data of the distributed photovoltaic devices, it is convenient to more accurately determine the benchmark range for the current distributed photovoltaic device itself in the future, thereby improving the accuracy of the diagnosis.

[0054] Step 202 , using irradiance and temperature as boundary conditions, and using a K-means clustering algorithm to divide the operating conditions of each photovoltaic power generation device.

[0055] Specifically, the irradiance and temperature are used as boundary conditions, and the working conditions are divided based on the K-means clustering algorithm to determine the different working conditions of the photovoltaic power generation equipment. As an example, the silhouette criterion can be used to determine the optimal clustering number K. The working conditions are divided because historical data under various working conditions can be collected to facilitate the generation of benchmark intervals for different working conditions. In addition, fault diagnosis can be performed in combination with the working conditions in the future to improve the accuracy of diagnosis.

[0056] Step 203: Remove the fault data under each working condition through a multivariate Gaussian mixture model to generate normal data training samples.

[0057] Among them, the multivariate Gaussian mixture model combines parameter estimation and non-parametric estimation methods and is a semi-parametric probability density estimation method. When the number of sub-models is large enough, the multivariate Gaussian mixture model can approach any continuous distribution with higher accuracy, retaining normal data in the middle of the target in the probability distribution and eliminating abnormal data.

[0058] The fault data to be eliminated may be data under various abnormal operating conditions, such as operating data of the photovoltaic power generation equipment during the fault period.

[0059] Step 204 , training the long short-term memory artificial neural network LSTM by using normal data training samples, and determining the reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each working condition according to the predicted value output by the long short-term memory artificial neural network LSTM.

[0060] Specifically, the data obtained in the above steps under normal operating conditions are used as training data to train the long short-term memory artificial neural network LSTM. Since the parameter types in the training data correspond to the parameter types in the actual operating data, after the long short-term memory artificial neural network LSTM is trained, multiple predicted values ​​output by the long short-term memory artificial neural network LSTM can be obtained for each parameter to be detected, and the actual value actually detected can be obtained, and the reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each working condition can be determined based on the predicted value and the actual value.

[0061] As an example, for each parameter, first obtain multiple parameter prediction values ​​output by the long short-term memory artificial neural network LSTM and the corresponding actual values, then calculate the difference between each parameter prediction value and the actual value, compare the differences, sort them according to the size of the difference, and use the parameter prediction value with the largest difference from the actual value as the target parameter prediction value. Take the target parameter prediction value as the center and construct a reference interval with twice the difference between the target parameter prediction value and the actual value as the interval length, that is, the reference interval is (target parameter prediction value-maximum difference, target parameter prediction value+maximum difference).

[0062] Thus, the reference interval corresponding to the full boundary conditions under each working condition is determined.

[0063] Step 103 : collecting the real-time operation data of each photovoltaic power generation device, and performing abnormality detection on the real-time operation data based on the corresponding reference interval.

[0064] Specifically, the collected real-time operating data of each photovoltaic power generation device is compared with the corresponding reference interval, and whether the operating data exceeds the reference interval is detected in real time. If the real-time operating data is not within the reference interval, the data can be determined to be abnormal data.

[0065] Step 104, if the real-time operating data of any photovoltaic power generation equipment is abnormal data, the signs of the abnormal data are determined, and any distributed photovoltaic power generation equipment is diagnosed by reasoning through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base to determine the abnormal mode and fault mode of any distributed photovoltaic power generation equipment.

[0066] Among them, the signs of abnormal data are evidence for identifying the failure mode and cause. For example, the signs may be "the output voltage of the generator set drops" and "the output voltage of the generator set fluctuates greatly", etc.

[0067] In the embodiment of the present application, the sign of abnormal data is input into the main body energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment, and the distributed photovoltaic power generation equipment that generated the abnormal data is diagnosed by reasoning through the corresponding probability relationship between the sign and the fault mode preset in the main body energy efficiency diagnosis knowledge base in the above steps. That is, through the assigned prior probability of "sign-fault mode", the fault mode corresponding to the sign is determined according to the input sign, so as to realize the reasoning diagnosis work through the Bayesian model. As an example, according to the assigned prior probability of "sign-fault mode", the fault mode with the highest probability corresponding to the current sign is determined as the diagnosed conclusion fault mode.

[0068] It should be noted that in the embodiments of the present application, from the perspective of energy efficiency, the relevant abnormal operation modes and the above-mentioned fault modes that cause abnormal energy efficiency of the power generation device can also be analyzed, wherein the abnormal mode is a mode in which the photovoltaic power generation equipment is abnormal but no obvious fault occurs, for example, the voltage output by the photovoltaic power generation equipment does not reach the set voltage, etc. The embodiments of the present application detect the abnormal mode of the photovoltaic power generation equipment, and the photovoltaic power generation equipment can be maintained in advance to ensure the normal operation of the equipment. As one possible implementation method, the abnormal mode with the highest probability corresponding to the current sign can be determined as the abnormal mode diagnosed as the conclusion based on the prior probability of the pre-assigned "sign-abnormal mode". The specific implementation method is not limited here.

[0069] It should also be noted that in actual applications, different faults may have the same symptom, that is, the same symptom may correspond to different fault modes. In order to further improve the accuracy of energy efficiency diagnosis, in one embodiment of the present application, the operating conditions can also be used as supplementary information to assist in reasoning diagnosis, that is, according to the operating conditions determined in the above steps, faults with a lower probability of occurrence under the operating conditions are eliminated to obtain more accurate diagnostic results.

[0070] Step 105: Send the diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the ontology energy efficiency diagnosis knowledge base to the mobile client of the operation and maintenance personnel.

[0071] Specifically, the ontology knowledge base and the collected data of the embodiment of the present application are stored on the server. In addition, a mobile client interface can be developed so that the operation and maintenance personnel can access the information of the relevant equipment according to the number of the photovoltaic power generation equipment through the mobile terminal device, query the unit failure mode and maintenance measures, and grasp the operation status of all photovoltaic power generation equipment in the distributed power station area. In addition, the health maintenance strategy corresponding to the current failure mode stored in the ontology knowledge base can also be transmitted to the mobile client, so that the operation and maintenance personnel can consult the maintenance strategy for maintenance.

[0072] In one embodiment of the present application, a fault record can also be added on the mobile terminal of the operation and maintenance personnel based on the currently determined fault mode, and the posterior probability of the corresponding probability relationship between the symptoms and the fault mode stored in the ontology knowledge base can be updated based on the current fault, so as to update the relevant knowledge in the knowledge base according to the historical fault conditions, and further improve the accuracy of diagnosis in the subsequent diagnosis process.

[0073] As one possible implementation method, firstly, a fault record is added on the mobile client of the operation and maintenance personnel according to the fault mode that occurred, wherein the added fault record includes but is not limited to the fault mode that occurred, the cause of the fault, the fault symptoms and the corresponding maintenance measures, etc. In specific implementation, the operation and maintenance personnel can input the above fault record information by checking or filling in the form according to the instructions on the mobile terminal, and can also automatically analyze and sort out the fault records through the application installed on the mobile terminal, and send them to the background server of the energy efficiency diagnosis system of the distributed photovoltaic power generation equipment.

[0074] Then, the posterior probability of the corresponding probability relationship between the preset symptom and the fault mode is updated according to the acquired fault record. In specific implementation, the background algorithm of the energy efficiency diagnosis system updates the posterior probability according to the current fault. In one embodiment of the present application, it is first determined whether the fault mode exists. If the fault mode exists, the posterior probability is updated. If the fault mode does not exist, the fault mode is added to the energy efficiency diagnosis knowledge base.

[0075] As a result, the final diagnosis results and the corresponding health maintenance strategies in the ontology knowledge base will be transmitted to the mobile client. The operation and maintenance personnel can directly view the operating status of the specified unit and make relevant maintenance plans. In addition, the knowledge base is continuously updated according to the fault records, and the posterior probability can be updated according to the current fault to improve the accuracy of the diagnosis.

[0076] In summary, the energy efficiency diagnosis method of the distributed photovoltaic power generation equipment of the embodiment of the present application forms the fault knowledge of abnormal energy efficiency level of photovoltaic power generation group through FMEA and FTA analysis methods, and then based on the concept of ontology, the unstructured photovoltaic power generation group related fault knowledge is expressed through a computer-recognizable structured language to form an open ontology knowledge base. When the real-time data is diagnosed for energy efficiency, the distributed photovoltaic power generation group energy efficiency diagnosis and health maintenance process of "abnormal detection-determining signs-diagnostic reasoning-locating faults-health maintenance decision-mobile user reading" is used. The abnormal detection is performed according to the reference interval determined by the environmental characteristics of each device, and then the abnormal information corresponding to the signs is input into the energy efficiency diagnosis knowledge base, and the reasoning diagnosis work based on the Bayesian model is carried out, so that the faulty generator group information and the corresponding specific faults can be quickly located, and the corresponding health maintenance strategy is provided. In addition, through the mobile client interface, the operation and maintenance personnel can access the relevant unit information according to the generator group number through the mobile terminal device, query the unit failure mode and maintenance measures, master the operation status of all generator groups in the distributed power station area, make maintenance plans in advance, reduce the operation and maintenance costs, and improve the operation and maintenance efficiency.

[0077] In order to more clearly illustrate the energy efficiency diagnosis method of the distributed photovoltaic power generation equipment in the embodiment of the present application, a specific embodiment of energy efficiency diagnosis of distributed photovoltaic power generation equipment is described in detail below. Figure 3 A schematic flow chart of a specific energy efficiency diagnosis method for distributed photovoltaic power generation equipment proposed in an embodiment of the present application.

[0078] like Figure 3 As shown in the figure, when using this method to diagnose energy efficiency, firstly, for the real-time operation data of each power generation equipment obtained through on-site communication in the distributed photovoltaic power station, the FMEA and FTA analysis methods are used to analyze the mechanism and equipment of the distributed photovoltaic power generation equipment, and sort out the common related fault knowledge that causes abnormal energy efficiency level of photovoltaic power generation units. With the help of the concept of ontology, the unstructured photovoltaic power generation unit related fault knowledge is expressed in a structured language that can be recognized by computers, and the open source tool protégé is used to model, forming an ontology-based distributed photovoltaic power generation unit energy efficiency diagnosis knowledge base. When using the open source tool protégé for ontology modeling, according to historical fault cases, combined with the Bayesian principle, the data attributes are used to complete the prior probability assignment of "symptom-fault mode". The operation and maintenance staff can add fault records on the mobile terminal according to the subsequent fault mode, and the background algorithm updates the posterior probability according to the current fault.

[0079] Then, for each generator set, according to its historical operating data under normal operation in the past year, the benchmark model is trained, and the irradiance is used as the benchmark interval for determining the operating parameters of each monitoring operation parameter under different irradiation and temperature conditions, and the fault data is used for verification to ensure the accuracy of the benchmark interval. The operating data of distributed photovoltaic generator sets in the past year are screened in a steady state, and non-steady-state data caused by sudden situations such as shading and coverage are eliminated. Then, the operating data during the fault period is eliminated to form a normal data training sample. In the specific implementation, the operating condition division based on the K-means clustering algorithm with irradiance and temperature as boundary conditions is used, and the optimal clustering number K is determined by the silhouette criterion. The local optimal algorithm multivariate Gaussian mixture model is used to eliminate abnormal data for historical data under each operating condition.

[0080] It should be noted that the multivariate Gaussian mixture model combines parameter estimation and non-parametric estimation methods and is a semi-parametric probability density estimation method. When the number of sub-models is large enough, the multivariate Gaussian mixture model can approach any continuous distribution with higher accuracy, and its probability distribution is shown in the following formula:

[0081]

[0082] Where X is the L-dimensional parameter data column vector, X = [x1x2,...,x L ] T ; k is the number of sub-models in the multivariate Gaussian mixture model; ω k is the weight coefficient of the kth sub-model, and φ(X|θ k ) represents the Gaussian probability density function of the kth sub-model, and its specific calculation formula is as follows:

[0083]

[0084] The multivariate Gaussian mixture model uses the Expectation Maximization (EM) algorithm to estimate parameters. The EM algorithm is an iterative algorithm for parameter estimation of probability models with latent variables. k and μ k is the covariance matrix and mean, and the value is estimated by the EM algorithm. The objective function of the EM algorithm is:

[0085]

[0086] The iterative process of the EM algorithm is as follows: select the initial values ​​of μ and ∑, then calculate the posterior probability of the corresponding k-th model according to formula (4), and finally iteratively calculate μ and ∑ in the model according to formula (5), (6), and (7). When the difference between the objective functions of two adjacent iterations is less than 10 -5When , stop the iteration.

[0087]

[0088]

[0089]

[0090]

[0091] Furthermore, the LSTM long short-term memory artificial neural network is trained based on the sample data to predict the parameters x p With the actual value x t The length of the reference interval is twice the maximum difference value sigma. Determine the reference interval (x p -sigma,x p +sigma), detects in real time whether the running data exceeds the benchmark range and completes the abnormality detection work.

[0092] It should also be noted that LSTM originates from recurrent neural network (RNN), which is a type of feedback neural network, and improves the gradient problem that occurs in long-term memory in RNN. In addition to inheriting the standard recurrent layer of RNN, threshold mechanisms such as memory gate and forget gate are introduced to control the memory and forgetting of information through the weight matrix. When processing long-time historical data, it can ensure that useful information in historical data can be remembered. The basic principle is that the original RNN hidden layer has only one state h, which is more sensitive to short-term states. On this basis, the LSTM network adds a long-term state quantity c to save the long-term state and transfer the state together with h. The standard module of the LSTM network contains three parts of state output: long-term state c t 、Short-term state h t And the current state output y t These three parts contain the memory and forgetting information of historical information in the long and short term, which are transmitted to the next neuron of the LSTM network together with the current state output to ensure the memory and propagation of time series information. The specific method of information regulation is to add three control gates, input gate, forget gate and output gate, in the state propagation path, corresponding to the three stages inside the LSTM network.

[0093] (1) Forgetting the long-term state: Obtain the forget gate z by calculation f Output c of the previous timing state t-1 Processing is performed to control which information in the long-term state in the time sequence state needs to be forgotten. These long-term information that is remembered and the memory information generated at the current moment form a new unit state c t Pass in the LSTM network unit at the next moment

[0094] (2) Input selection memory stage: By t , and the h passed in at the previous moment t-1 Get the current input information matrix z, through the gate signal z i Selective memory and long-term state after selective forgetting t -1 Working together to update long-term memory information to form c t ;

[0095] (3) Output stage: The output stage includes y t 、h t and c t , where h t and c t As short-term and long-term memory states, y is passed to the next time state. t According to the current short-term state h t The change is obtained as the output of the current state.

[0096] Furthermore, the corresponding symptoms of the fault mode are input into the energy efficiency diagnosis knowledge base to carry out the subsequent reasoning diagnosis work based on the Bayesian model. Among them, the functional requirements of ontology query and ontology-based data access can be realized through the SPARQL language. By using the SPARQL language to define the association mode of information, the target information that meets the user can be retrieved in the ontology. The typical query statement of SPARQL is as follows:

[0097] SELECT? x WHERE{? x rdfs:subClassOf:F01.}

[0098] The meaning of this formula is to query all subclasses of fault F01. SPARQL also supports multiple matches and literal matches. In the group graph pattern query, each triple pattern must be matched successfully. Compared with the database SQL language, SPARQL does not require the mastery of complex query logic, but directly faces the knowledge involved in the query problem itself. It is suitable for non-computer professional users to design query mechanisms, which improves the applicability of the present application method.

[0099] Finally, the final diagnosis results and the corresponding health maintenance strategies in the ontology knowledge base will be transmitted to the mobile client, and the operation and maintenance personnel can directly view the operating status of the specified unit and make relevant maintenance plans. In addition, the operation and maintenance personnel check or fill in the fault mode, cause of the fault, fault symptoms and related maintenance measures according to the instructions on the mobile terminal. If the fault mode already exists, the prior probability value in the relevant reasoning relationship in the Bayesian model will be adjusted after checking. If the fault mode occurs for the first time, the operation and maintenance personnel need to fill in the relevant information, and the system will automatically enter it into the ontology knowledge base.

[0100] In order to implement the above embodiment, the present application also proposes an energy efficiency diagnosis system for distributed photovoltaic power generation equipment. Figure 4 This is a schematic diagram of the structure of an energy efficiency diagnosis system for distributed photovoltaic power generation equipment proposed in an embodiment of the present application. Figure 4 As shown, the diagnostic system includes a generating module 100 , a determining module 200 , an abnormality detecting module 300 , an inference diagnostic module 400 and a sending module 500 .

[0101] Among them, the first acquisition module 100 is used to analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain the energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate the ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment.

[0102] The determination module 200 is used to obtain the historical operation data of each photovoltaic power generation device, and determine the reference interval of the detection data of each photovoltaic power generation device under different working conditions according to the historical operation data.

[0103] The abnormality detection module 300 is used to collect the real-time operation data of each photovoltaic power generation device and perform abnormality detection on the real-time operation data based on the corresponding reference interval.

[0104] The reasoning diagnosis module 400 is used to determine the signs of the abnormal data if the real-time operating data of any photovoltaic power generation equipment is abnormal data, and to perform reasoning diagnosis on any distributed photovoltaic power generation equipment through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base to determine the abnormal mode and fault mode of any distributed photovoltaic power generation equipment.

[0105] The sending module 500 is used to send the diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the main body energy efficiency diagnosis knowledge base to the mobile client of the operation and maintenance personnel.

[0106] Optionally, in one embodiment of the present application, the generation module 100 is specifically used to: convert energy efficiency fault knowledge into a structured language recognizable by a computer; based on the structured language, modeling is performed through a preset open source tool to generate an energy efficiency diagnosis knowledge base of a distributed photovoltaic power generation group based on ontology; historical fault data of each distributed photovoltaic power generation equipment is obtained, and based on the Bayesian theorem, a priori probability assignments are performed on the fault modes corresponding to different symptoms to generate a corresponding probability relationship between preset symptoms and fault modes.

[0107] Optionally, in one embodiment of the present application, the determination module 200 is specifically used to: perform steady-state screening on historical operating data to screen out non-steady-state data in the historical operating data; use irradiance and temperature as boundary conditions to divide the operating conditions of each photovoltaic power generation equipment through the K-means clustering algorithm; eliminate abnormal data under each operating condition through a multivariate Gaussian mixture model to generate normal data training samples; train a long short-term memory artificial neural network LSTM through the normal data training samples, and determine the reference interval corresponding to the boundary conditions of each photovoltaic power generation equipment under each operating condition based on the predicted value output by the long short-term memory artificial neural network LSTM.

[0108] Optionally, in one embodiment of the present application, the determination module 200 is also used to: for each parameter, obtain multiple parameter prediction values ​​output by the long short-term memory artificial neural network LSTM, and the corresponding actual values; calculate the difference between each parameter prediction value and the actual value, and take the parameter prediction value with the largest difference with the actual value as the target parameter prediction value; construct a reference interval with the target parameter prediction value as the center and twice the difference between the target parameter prediction value and the actual value as the interval length.

[0109] Optionally, in one embodiment of the present application, the sending module 500 is also used to add a fault record on the mobile client of the operation and maintenance personnel according to the fault mode that occurred; and update the posterior probability of the corresponding probability relationship between the preset symptoms and the fault mode according to the acquired fault record.

[0110] Optionally, in one embodiment of the present application, the sending module 500 is also used to receive the occurred fault mode, the cause of the fault, the fault symptoms and the corresponding maintenance measures sent by the mobile client; determine whether the occurred fault mode exists, and if the occurred fault mode exists, update the posterior probability; if the occurred fault mode does not exist, add the occurred fault mode to the ontology energy efficiency diagnosis knowledge base.

[0111] It should be noted that the above explanation of the embodiment of the energy efficiency diagnosis method of distributed photovoltaic power generation equipment is also applicable to the system of this embodiment, and will not be repeated here.

[0112] In summary, the energy efficiency diagnosis system of the distributed photovoltaic power generation equipment of the embodiment of the present application can quickly locate the faulty generator set information and the corresponding specific faults, and provide corresponding health maintenance strategies. In addition, through the mobile client interface, the operation and maintenance personnel can access the relevant unit information according to the generator set number through the mobile device, query the unit failure mode and maintenance measures, grasp the operation status of all generator sets in the distributed power station area, make maintenance plans in advance, reduce operation and maintenance costs, and improve operation and maintenance efficiency.

[0113] In order to implement the above embodiments, the present application also proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, an energy efficiency diagnosis method for distributed photovoltaic power generation equipment as described in any of the above embodiments is implemented.

[0114] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0115] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0116] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0117] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0118] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0119] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0120] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0121] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for energy efficiency diagnosis of distributed photovoltaic power generation equipment, characterized in that: The following steps are involved: Analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate an ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment; Obtain the historical operation data of each of the photovoltaic power generation equipment, and determine the reference interval of the detection data of each of the photovoltaic power generation equipment under different working conditions according to the historical operation data, wherein the determination of the reference interval of the detection data of each of the photovoltaic power generation equipment under different working conditions according to the historical operation data includes: performing steady-state screening on the historical operation data to screen out non-steady-state data in the historical operation data; using irradiance and temperature as boundary conditions, dividing the working conditions of each of the photovoltaic power generation equipment by the K-means clustering algorithm; eliminating the fault data under each working condition by the multivariate Gaussian mixture model to generate normal data training samples; training a long short-term memory artificial neural network LSTM by the normal data training samples, and determining the reference interval corresponding to the boundary conditions of each of the photovoltaic power generation equipment under each working condition according to the predicted value output by the long short-term memory artificial neural network LSTM; Collecting real-time operating data of each photovoltaic power generation device, and performing abnormality detection on the real-time operating data based on the corresponding reference interval; If the real-time operation data of any photovoltaic power generation equipment is abnormal data, the signs of the abnormal data are determined, and any distributed photovoltaic power generation equipment is diagnosed by reasoning through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base to determine the abnormal mode and fault mode of any distributed photovoltaic power generation equipment; The diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the main body energy efficiency diagnosis knowledge base are sent to the mobile client of the operation and maintenance personnel.

2. The diagnostic method according to claim 1, characterized in that The generating of the ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment comprises: Converting the energy efficiency fault knowledge into a structured language recognizable by a computer; Based on the structured language, modeling is performed through a preset open source tool to generate an energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation group based on ontology; The historical fault data of each of the distributed photovoltaic power generation equipment is obtained, and the prior probability of the fault modes corresponding to different symptoms is assigned based on the Bayesian theorem to generate the corresponding probability relationship between the preset symptoms and the fault modes.

3. The diagnostic method according to claim 1, characterized in that: Determining the reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each working condition according to the parameter prediction value output by the long short-term memory artificial neural network LSTM, includes: For each parameter, obtain multiple parameter prediction values ​​output by the long short-term memory artificial neural network LSTM and the corresponding actual values; Calculate the difference between each parameter prediction value and the actual value, and take the parameter prediction value with the largest difference from the actual value as the target parameter prediction value; The reference interval is constructed with the target parameter prediction value as the center and twice the difference between the target parameter prediction value and the actual value as the interval length.

4. The diagnostic method according to claim 1, after sending the diagnostic result and the maintenance strategy corresponding to the diagnostic result stored in the energy efficiency diagnostic knowledge base of the entity to the mobile client of the operation and maintenance personnel, further comprising: Adding a fault record on the mobile client of the operation and maintenance personnel according to the fault mode that occurred; The posterior probability of the corresponding probability relationship between the preset symptom and the fault mode is updated according to the acquired fault record.

5. The diagnostic method according to claim 4, characterized in that: Adding a fault record on the mobile client of the operation and maintenance personnel includes: Receiving the fault mode, fault cause, fault symptom and corresponding maintenance measures sent by the mobile client; The updating of the posterior probability of the corresponding probability relationship between the preset symptom and the fault mode according to the acquired fault record includes: Determine whether the fault mode exists, and if the fault mode exists, update the posterior probability; If the occurred fault mode does not exist, the occurred fault mode is added to the ontology energy efficiency diagnosis knowledge base.

6. An energy efficiency diagnostic system for distributed photovoltaic power generation equipment, characterized in that: include: A generation module is used to analyze the distributed photovoltaic power generation equipment to be diagnosed through failure mode and effect analysis FMEA and fault tree analysis FTA, obtain energy efficiency fault knowledge of the distributed photovoltaic power generation equipment, and generate an ontology energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation equipment; A determination module is used to obtain the historical operation data of each photovoltaic power generation device, and determine the reference interval of the detection data of each photovoltaic power generation device under different working conditions according to the historical operation data; wherein the determination module is specifically used to: perform steady-state screening on the historical operation data to screen out the non-steady-state data in the historical operation data; use irradiance and temperature as boundary conditions to divide the working conditions of each photovoltaic power generation device through the K-means clustering algorithm; eliminate the fault data under each working condition through the multivariate Gaussian mixture model to generate normal data training samples; train the long short-term memory artificial neural network LSTM through the normal data training samples, and determine the reference interval corresponding to the boundary conditions of each photovoltaic power generation device under each working condition according to the predicted value output by the long short-term memory artificial neural network LSTM; an abnormality detection module, used to collect the real-time operation data of each photovoltaic power generation device, and perform abnormality detection on the real-time operation data based on the corresponding reference interval; An inference diagnosis module is used to determine the signs of abnormal data if the real-time operation data of any photovoltaic power generation equipment is abnormal data, and to perform inference diagnosis on any distributed photovoltaic power generation equipment through the corresponding probability relationship between the signs and the fault modes preset in the ontology energy efficiency diagnosis knowledge base, and determine the abnormal mode and fault mode of any distributed photovoltaic power generation equipment; The sending module is used to send the diagnosis result and the maintenance strategy corresponding to the diagnosis result stored in the main body energy efficiency diagnosis knowledge base to the mobile client of the operation and maintenance personnel.

7. The diagnostic system according to claim 6, characterized in that The generating module is specifically used for: Converting the energy efficiency fault knowledge into a structured language recognizable by a computer; Based on the structured language, modeling is performed through a preset open source tool to generate an energy efficiency diagnosis knowledge base of the distributed photovoltaic power generation group based on ontology; The historical fault data of each of the distributed photovoltaic power generation equipment is obtained, and the prior probability of the fault modes corresponding to different symptoms is assigned based on the Bayesian theorem to generate the corresponding probability relationship between the preset symptoms and the fault modes.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the energy efficiency diagnosis method for distributed photovoltaic power generation equipment as described in any one of claims 1 to 5 is implemented.

Images