CN117251788A

CN117251788A - State evaluation method, device, terminal equipment and storage medium

Info

Publication number: CN117251788A
Application number: CN202311215965.8A
Authority: CN
Inventors: 周冰钰; 高超; 张家前; 方振宇
Original assignee: Sunshine Zhiwei Technology Co ltd
Current assignee: Sunshine Zhiwei Technology Co ltd
Priority date: 2023-09-19
Filing date: 2023-09-19
Publication date: 2023-12-19

Abstract

The application discloses a state evaluation method, a state evaluation device, terminal equipment and a storage medium, wherein the state evaluation method comprises the following steps: acquiring target data of a photovoltaic inverter; inputting target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter. Based on the scheme, related data can be fully mined and utilized by combining quantitative parameters and qualitative parameters to construct an index system, the defect of single data type is overcome, a trained state evaluation model has more comprehensive state evaluation capability, and the accuracy and reliability of state evaluation of the photovoltaic inverter are improved.

Description

State evaluation method, device, terminal equipment and storage medium

Technical Field

The present disclosure relates to the field of photovoltaic power generation technologies, and in particular, to a state evaluation method, a state evaluation device, a terminal device, and a storage medium.

Background

As a key device of a solar power generation system, the running health state of the photovoltaic inverter directly affects the safety and stability of the power generation system, so that the requirement for performing state evaluation on the data of the photovoltaic inverter begins to appear.

The data related to the state evaluation of the photovoltaic inverter show large-scale, multi-source, multi-dimensional and other large data characteristics, but in the current state evaluation method of the photovoltaic inverter, the input and output of the photovoltaic inverter are simply measured, so that the state of the photovoltaic inverter is evaluated. Such a state evaluation method does not adequately mine and utilize the relevant data, resulting in a state result of the photovoltaic inverter that is not sufficiently accurate and reliable.

Disclosure of Invention

The main purpose of the application is to provide a state evaluation method, a state evaluation device, terminal equipment and a storage medium, and aims to solve the problem that a state result of a photovoltaic inverter is not accurate and reliable enough.

To achieve the above object, the present application provides a state evaluation method, including:

acquiring target data of a photovoltaic inverter;

inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter.

Optionally, before the step of obtaining the target data of the photovoltaic inverter, the method further includes:

acquiring at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter;

constructing a basic index system of the photovoltaic inverter based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter;

constructing a key index system of the photovoltaic inverter based on the basic index system of the photovoltaic inverter;

acquiring the sample data according to a key index system of the photovoltaic inverter;

and training the initial support vector machine model based on the sample data and the preset optimizing parameters to obtain the trained state evaluation model.

Optionally, the step of acquiring target data of the photovoltaic inverter includes:

acquiring original data of a photovoltaic inverter;

screening the original data according to the key parameter system of the photovoltaic inverter to obtain screened original data;

preprocessing the screened original data to obtain the target data.

Optionally, the step of constructing a basic index system of the photovoltaic inverter based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter includes:

Analyzing at least one quantitative parameter of the photovoltaic inverter based on a preset first analysis rule to obtain a sensitivity factor corresponding to each of the at least one quantitative parameter;

analyzing at least one qualitative parameter of the photovoltaic inverter based on a preset second analysis rule to obtain degradation factors corresponding to the at least one qualitative parameter;

and constructing and obtaining a basic index system of the photovoltaic inverter based on the at least one quantitative parameter, the sensitive factor corresponding to the at least one quantitative parameter, the at least one qualitative parameter and the degradation factor corresponding to the at least one qualitative parameter.

Optionally, the at least one quantitative parameter belongs to a respective corresponding feature type, the number of the feature types is at least one, the at least one quantitative parameter includes at least one quantitative parameter sample, the step of analyzing the at least one quantitative parameter of the photovoltaic inverter based on a preset first analysis rule to obtain a sensitivity factor corresponding to the at least one quantitative parameter includes:

according to the quantitative parameter samples corresponding to at least one quantitative parameter under each characteristic type, calculating to obtain the average intra-class distance and the average inter-class distance corresponding to at least one quantitative parameter under each characteristic type;

And calculating to obtain the sensitivity factor corresponding to at least one quantitative parameter under each characteristic type based on the average intra-class distance and the average inter-class distance.

Optionally, the at least one qualitative parameter belongs to a respective corresponding characteristic type, the at least one qualitative parameter includes a qualitative parameter actual value, a qualitative parameter threshold value, and a qualitative parameter factory value, the step of analyzing the at least one qualitative parameter of the photovoltaic inverter based on a preset second analysis rule to obtain a degradation factor corresponding to the at least one qualitative parameter includes:

and calculating the degradation factors corresponding to the at least one qualitative parameter under each characteristic type based on a preset offset calculation rule according to the qualitative parameter actual value, the qualitative parameter threshold value and the qualitative parameter factory value corresponding to the at least one qualitative parameter under each characteristic type.

Optionally, the step of constructing the key index system of the photovoltaic inverter based on the basic index system of the photovoltaic inverter includes:

screening at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter based on a basic index system of the photovoltaic inverter to obtain at least one preliminary screening quantitative parameter and at least one preliminary screening qualitative parameter;

Constructing and obtaining a normalized kernel matrix based on a preset Gaussian kernel function, quantitative parameter samples corresponding to the at least one preliminary screening quantitative parameter and qualitative parameter samples corresponding to the at least one preliminary screening qualitative parameter;

constructing and obtaining a corresponding feature matrix and a feature vector matrix based on the normalized kernel matrix;

constructing a target matrix based on the normalized kernel matrix, the feature matrix and the feature vector matrix;

and constructing a key index system of the photovoltaic inverter based on the target matrix.

Optionally, the step of training the initial support vector machine model based on the sample data and the preset optimizing parameter to obtain the trained state evaluation model includes:

preprocessing the sample data to obtain a training sample and a test sample;

training the initial support vector machine model based on the training sample;

judging whether the initial support vector machine model meets a preset cross verification condition or not;

if the initial support vector machine model meets the cross verification condition, training the initial support vector machine model is stopped, and a state evaluation model to be tested is obtained;

If the initial support vector machine model does not meet the cross verification condition, training the initial support vector machine model based on the training sample and preset optimizing parameters until the initial support vector machine model meets the cross verification condition, stopping training the initial support vector machine model, and obtaining the state evaluation model to be tested;

testing the state evaluation model to be tested based on the test sample;

and if the state evaluation model to be tested passes the test, obtaining the trained state evaluation model.

Optionally, the training the initial support vector machine model based on the training sample and a preset optimizing parameter until the initial support vector machine model meets the cross verification condition, and before the step of obtaining the state evaluation model to be tested, further includes:

at least one optimizing parameter is initialized based on a preset wolf algorithm.

Optionally, after the step of inputting the target data to the trained state evaluation model to obtain the state evaluation information of the photovoltaic inverter, the method further includes:

If the state evaluation information of the photovoltaic inverter accords with a preset alarm condition, determining an alarm category corresponding to the state evaluation information of the photovoltaic inverter;

and pushing corresponding alarm information according to the alarm category.

The embodiment of the application also provides a state evaluation device, which comprises:

the acquisition module is used for acquiring target data of the photovoltaic inverter;

the evaluation module is used for inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter.

The embodiment of the application also provides a terminal device, which comprises a memory, a processor and a state evaluation program stored on the memory and capable of running on the processor, wherein the state evaluation program realizes the steps of the state evaluation method when being executed by the processor.

The embodiments of the present application also propose a computer-readable storage medium, on which a state evaluation program is stored, which when executed by a processor implements the steps of the state evaluation method as described above.

The state evaluation method, the state evaluation device, the terminal equipment and the storage medium provided by the embodiment of the application are used for acquiring target data of the photovoltaic inverter; inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter. Based on the scheme, related data can be fully mined and utilized by combining quantitative parameters and qualitative parameters to construct an index system, the defect of single data type is overcome, a trained state evaluation model has more comprehensive state evaluation capability, and the accuracy and reliability of state evaluation of the photovoltaic inverter are improved.

Drawings

FIG. 1 is a schematic diagram of functional modules of a terminal device to which a state evaluation device of the present application belongs;

FIG. 2 is a flowchart of a first exemplary embodiment of a state evaluation method according to the present application;

FIG. 3 is a flowchart illustrating a second exemplary embodiment of a state evaluation method according to the present application;

FIG. 4 is a flowchart of a third exemplary embodiment of a state evaluation method according to the present application;

FIG. 5 is a flowchart of a fourth exemplary embodiment of a state evaluation method according to the present application;

FIG. 6 is a flowchart illustrating a fifth exemplary embodiment of a status evaluation method according to the present application;

FIG. 7 is a flowchart of a sixth exemplary embodiment of a state evaluation method according to the present application;

FIG. 8 is a flowchart of a seventh exemplary embodiment of a state evaluation method according to the present application;

FIG. 9 is a flowchart illustrating an eighth exemplary embodiment of a state evaluation method according to the present application;

FIG. 10 is a schematic diagram of model training involved in the state evaluation method of the present application;

FIG. 11 is a flowchart of a ninth exemplary embodiment of a state evaluation method according to the present application;

fig. 12 is a flowchart of a tenth exemplary embodiment of a state evaluation method according to the present application.

The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.

Detailed Description

It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

The main solutions of the embodiments of the present application are: acquiring target data of a photovoltaic inverter; inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter. Based on the scheme, related data can be fully mined and utilized by combining quantitative parameters and qualitative parameters to construct an index system, the defect of single data type is overcome, a trained state evaluation model has more comprehensive state evaluation capability, and the accuracy and reliability of state evaluation of the photovoltaic inverter are improved.

Specifically, referring to fig. 1, fig. 1 is a schematic functional block diagram of a terminal device to which the state evaluation device of the present application belongs. The state evaluation device may be a device independent of the terminal device, which is capable of state evaluation, and which may be carried on the terminal device in the form of hardware or software. The terminal equipment can be an intelligent mobile terminal with a data processing function such as a mobile phone and a tablet personal computer, and can also be a fixed terminal equipment or a server with a data processing function.

In this embodiment, the terminal device to which the state evaluation apparatus belongs includes at least an output module 110, a processor 120, a memory 130, and a communication module 140.

The memory 130 stores an operating system and a state evaluation program, and the state evaluation device may acquire target data of the photovoltaic inverter; inputting target data into a trained state evaluation model, and storing information such as state evaluation information of the obtained photovoltaic inverter in the memory 130; the output module 110 may be a display screen or the like. The communication module 140 may include a WIFI module, a mobile communication module, a bluetooth module, and the like, and communicates with an external device or a server through the communication module 140.

Wherein the state evaluation program in the memory 130, when executed by the processor, performs the steps of:

acquiring target data of a photovoltaic inverter;

Further, the state evaluation program in the memory 130, when executed by the processor, further performs the steps of:

acquiring original data of a photovoltaic inverter;

preprocessing the screened original data to obtain the target data.

preprocessing the sample data to obtain a training sample and a test sample;

training the initial support vector machine model based on the training sample;

testing the state evaluation model to be tested based on the test sample;

and pushing corresponding alarm information according to the alarm category.

According to the scheme, the target data of the photovoltaic inverter are obtained; inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter. In the embodiment, related data can be fully mined and utilized by combining quantitative parameters and qualitative parameters to construct an index system, the defect of single data type is overcome, a trained state evaluation model has more comprehensive state evaluation capability, and the accuracy and reliability of state evaluation of the photovoltaic inverter are improved.

Referring to fig. 2, a first embodiment of a state evaluation method of the present application provides a flowchart, where the state evaluation method includes:

step S10, obtaining target data of the photovoltaic inverter.

In particular, as a key device of a solar power generation system, the running health state of the photovoltaic inverter directly affects the safety and stability of the power generation system, so that the need for performing state evaluation on the data of the photovoltaic inverter begins to appear. In the long-term operation process of the photovoltaic inverter, the problems of abnormal operation state and the like often occur, and the traditional maintenance method mainly comprises periodic manual inspection and post-inspection maintenance. However, these methods are not only costly to service, but also waste service resources and affect the stability of the overall power generation system.

With the continuous development of the Internet of things and big data technology, novel detection equipment and sensors are widely applied, so that the photovoltaic inverter data show big data characteristics of large scale, multiple sources, multiple dimensions and the like. Although the industry has begun to conduct simple state monitoring and evaluation of photovoltaic inverter data, collection and mining becomes difficult due to the large scale of data. This causes problems of inaccuracy in the mining of the data features of the device, coarseness in the state evaluation, and insufficient real-time performance of the fault diagnosis. These drawbacks prevent efficient development of photovoltaic inverter status maintenance work and limit the establishment of reasonable maintenance strategies. In the current state evaluation method of the photovoltaic inverter, data of external characteristics can be obtained by measuring input and output of a photovoltaic inverter system, but to understand the dynamic rule of the photovoltaic inverter, internal state variables are needed to describe the state, and the states cannot be directly measured in general.

For this reason, the present embodiment proposes a new state evaluation method for a photovoltaic inverter. First, target data required for state evaluation needs to be acquired, and the target data may include data of quantitative parameters and data of qualitative parameters.

The quantitative parameters of the photovoltaic inverter generally refer to parameters which can be represented by specific numerical values, such as voltage, current, power and the like, and are used for describing the operation performance and state of the inverter. The data of the quantitative parameter included in the target data may be one or more of the following: (1) input direct current voltage and current: the direct current voltage and current of the photovoltaic battery string input by the inverter. (2) Outputting alternating voltage and current: alternating voltage and current output by the inverter. (3) Maximum Power Point (MPP) voltage and current: the optimal point for inverter operation corresponds to the maximum output power. (4) Efficiency is that: the energy conversion efficiency of the inverter means the conversion efficiency from direct current to alternating current. (5) Temperature: the temperature inside the inverter includes the temperature of the radiator, the electronic component, and the like. (6) Frequency: the alternating current frequency output by the inverter. (7) Run time: the length of time the inverter is operated.

In addition, qualitative parameters of the photovoltaic inverter are descriptive parameters, which cannot be directly represented by numerical values, and are usually expressed by characters or categories, such as fault types, working states and the like. The data of the qualitative ratings included in the target data may be one or more of the following: (1) fault type: indicating possible faults of the inverter, such as overvoltage, overcurrent, short-circuit, etc. (2) Operating state: whether the inverter is in a normal working state, a standby state, a fault state and the like currently. (3) And (3) stopping and recording: and recording information such as downtime and reasons of the inverter. (4) Early warning information: early warning information that the inverter may malfunction. (5) Operating environment: the inverter is subjected to environmental conditions such as weather, temperature, humidity, etc. (6) Maintaining a record: the maintenance history of the inverter includes records of maintenance, service, and the like. (7) Instruction manual: the instruction manual may describe various modes of operation of the inverter, wiring patterns, functional specifications of buttons and display screens, meanings of alarm codes, and the like. The comprehensive analysis of the quantitative parameters and the qualitative parameters can more comprehensively understand the state and the performance of the inverter, and help evaluate the health condition and the running condition of the inverter.

It should be noted that the target data should be obtained in time to ensure that the state evaluation information obtained finally is as synchronous as possible with the actual photovoltaic inverter operation state. In addition, the target data conforms to the input specifications of the state assessment model.

Step S20, inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter.

And inputting target data required to be subjected to state evaluation into a trained state evaluation model. In this process, the trained state evaluation model analyzes and processes the input target data according to the previous training experience and learned rules, thereby generating state evaluation information about the state of the photovoltaic inverter.

The quantitative parameters directly reflect the performance and the running condition of the photovoltaic inverter, the qualitative parameters describe various attributes of the photovoltaic inverter, an index system suitable for the photovoltaic inverter is constructed by combining the quantitative parameters and the qualitative parameters, and sample data closely related to the state evaluation of the photovoltaic inverter can be obtained by utilizing the index system; in addition, by selecting proper optimizing parameters, the model can be ensured to better fit data, and the model has higher prediction capability; in addition, the initial support vector machine model can be selected as a training object in consideration of the advantages that the support vector machine model has high robustness, is suitable for nonlinear data, is insensitive to abnormal values, has strong generalization performance, good interpretability and the like. Therefore, under the advantages of the index system, the sample data, the optimizing parameters, the support vector machine model and the like, the trained state evaluation model can more comprehensively know the state of the photovoltaic inverter, and high-quality state evaluation information is given.

It can be appreciated that the state evaluation information reflects the internal state and the external state of the photovoltaic inverter, providing a more comprehensive reference for the operation and maintenance decision of the photovoltaic inverter.

Further, referring to fig. 3, a flow chart is provided in a second embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 2, step S10, before obtaining the target data of the photovoltaic inverter, further includes:

And step S01, at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter are obtained.

Specifically, in order to enable the trained state evaluation model to have a more comprehensive state evaluation capability, at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter need to be obtained in advance. It is understood that the quantitative or qualitative parameters related to the different characteristics or indexes of the photovoltaic inverter are also different, so that the model training is not necessarily performed by using all the quantitative or qualitative parameters among at least one quantitative parameter and at least one qualitative parameter of the obtained photovoltaic inverter.

And step S02, constructing a basic index system of the photovoltaic inverter based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter.

Specifically, a basic index system of the photovoltaic inverter is constructed based on at least one quantitative parameter and at least one qualitative parameter of the obtained photovoltaic inverter. The basic index system comprises at least one quantitative parameter and a corresponding sensitive factor respectively, and comprises at least one qualitative parameter and a corresponding degradation factor respectively. The sensitive factors are used for strengthening key quantitative parameters and weakening non-key quantitative parameters; the degradation factors are used to strengthen critical qualitative ratings and weaken non-critical qualitative ratings.

It can be understood that under the basic index system, the influence of quantitative parameters and qualitative parameters on state evaluation is quantized, so that basis can be provided for state evaluation and monitoring of the photovoltaic inverter, and more accurate and effective operation analysis and maintenance decision can be realized.

And S03, constructing a key index system of the photovoltaic inverter based on the basic index system of the photovoltaic inverter.

Specifically, as the basic index system of the photovoltaic inverter quantifies the influence of each quantitative parameter and each qualitative parameter on the state evaluation, part of key quantitative parameters and key qualitative parameters can be selected on the basis of the basic index system, and a key index system of the photovoltaic inverter is constructed.

More specifically, the process of selecting a part of the key quantitative parameters and the key qualitative parameters may refer to the sensitivity factor and the degradation factor. Wherein, the value of the sensitive factor is positively correlated with the sensitivity of the quantitative parameter, and the larger the value of the sensitive factor is, the larger the sensitivity of the quantitative parameter is; the smaller the value of the sensitivity factor, the smaller the sensitivity of the quantitative parameter. The magnitude of the degradation factor is positively correlated with the degradation degree of the qualitative parameter, is negatively correlated with the operation state of the photovoltaic inverter, and the smaller the magnitude of the degradation factor is, the lower the degradation degree of the qualitative parameter is, and the better the operation state of the photovoltaic inverter is; the greater the value of the degradation factor, the greater the degree of degradation of the qualitative parameter, and the worse the operating state of the photovoltaic inverter.

And step S04, acquiring the sample data according to a key index system of the photovoltaic inverter.

Specifically, through the selection process of the foregoing steps, the key index system includes at least one key quantitative parameter and at least one key qualitative parameter, where the at least one key quantitative parameter and the at least one key qualitative parameter are critical to the state evaluation of the photovoltaic inverter.

On the basis of acquiring multiple internal and external operation data of the photovoltaic inverter, screening the multiple internal and external operation data according to a key index system to obtain sample data related to at least one key quantitative parameter and sample data related to at least one key qualitative parameter.

And step S05, training the initial support vector machine model based on the sample data and the preset optimizing parameters to obtain the trained state evaluation model.

Specifically, based on obtaining sample data related to at least one key quantitative parameter and sample data related to at least one key qualitative parameter, the initial support vector machine model is continuously trained based on the sample data related to the at least one key quantitative parameter and the sample data related to the at least one key qualitative parameter until training is finished, and a trained state evaluation model can be obtained.

The initial support vector machine model is constructed based on a preset support vector machine algorithm, and the Support Vector Machine (SVM) is a machine learning algorithm used for classification and regression tasks. When an initial support vector machine model is constructed, a support vector machine algorithm is selected, parameters are set according to characteristics of sample data and problem requirements, and then parameters of the model are adjusted through the sample data, so that the initial support vector machine model can better classify or predict the sample data in a regression mode. After training, a trained state estimation model can be obtained, and it can be appreciated that the trained state estimation model is an improved SVM model.

According to the scheme, at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter are obtained; constructing a basic index system of the photovoltaic inverter based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter; constructing a key index system of the photovoltaic inverter based on the basic index system of the photovoltaic inverter; acquiring the sample data according to a key index system of the photovoltaic inverter; and training the initial support vector machine model based on the sample data and the preset optimizing parameters to obtain the trained state evaluation model. In the embodiment, a basic index system is constructed by acquiring quantitative and qualitative parameters of the photovoltaic inverter, so that a key index system is formed. Sample data is obtained through the key index system, an initial support vector machine model is trained based on the sample data and optimizing parameters, and finally a trained state evaluation model is obtained. Therefore, the generalization capability of the trained state evaluation model can be improved based on the steps, so that the photovoltaic inverter has more comprehensive state evaluation capability, and accurate state evaluation of the photovoltaic inverter is realized.

Further, referring to fig. 4, a flow chart is provided in a third embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 3, step S10, obtaining target data of the photovoltaic inverter is further refined, which includes:

step S101, obtaining raw data of a photovoltaic inverter.

Specifically, before the state of the photovoltaic inverter is evaluated, raw data of the photovoltaic inverter needs to be acquired, and the raw data may be multiple internal and external operation data of the photovoltaic inverter. Typically, the scale of the raw data is relatively large, and some of the raw data does not actually participate in the state evaluation process of the photovoltaic inverter.

For example, the raw data may include: 1. current and voltage data: including Direct Current (DC) input current and voltage and Alternating Current (AC) output current and voltage, which reflect the conversion and output of electrical energy. 2. Power data: the method comprises the steps of including direct current input power and alternating current output power and evaluating electric energy conversion efficiency of the photovoltaic inverter. 3. Temperature data: temperature data inside and outside the photovoltaic inverter is used to monitor thermal management and possible overheating conditions of the device. 4. Frequency data: and the alternating current output frequency data is used for ensuring that the electric energy output by the photovoltaic inverter accords with the standard frequency. 5. Working state data: including the switching state, operating mode, fault state, etc. of the photovoltaic inverter for evaluating the operating state and performance of the device. 6. Direct current input and alternating current output waveform data: current and voltage waveform data for analyzing power quality and detecting possible waveform distortion. 7. Communication data: communication data with other devices or monitoring systems for remote monitoring and control of the photovoltaic inverter. 8. Event log: the event and fault information of the device are recorded for fault diagnosis and maintenance. 9. Environmental condition data: including ambient condition data such as light, temperature, humidity, etc., which can affect the performance of the photovoltaic inverter. 10. Timestamp data: the time stamps of the data acquisitions are recorded for time series analysis and data alignment. 11. Working mode: describing the current working mode of the photovoltaic inverter, such as normal operation, fault state, shutdown and the like. 12. Fault code: if the photovoltaic inverter is in a fault state, including the corresponding fault code and description, is used to diagnose the problem. 13. Maintenance/inspection record: recording the date and time of repair, inspection or maintenance activities; describing the type of maintenance activities, such as periodic inspection, fault maintenance, preventive maintenance, etc.; the detailed record of the maintenance or inspection content includes the inspected project, the found problem, the measures taken at one time, etc.; describing the results of maintenance activities, including whether the problem is resolved, whether performance is improved, etc.; if the photovoltaic inverter has failed, the relevant fault code, maintenance measures and maintenance date are recorded.

The foregoing examples of the original data are merely for facilitating understanding of the content of the original data, and are not limiting of the original data. The raw data may include one or more of the above example data, or may include data other than the above example data. It is understood that the raw data may include both quantitative and qualitative data.

And step S102, screening the original data according to the key parameter system of the photovoltaic inverter to obtain screened original data.

In particular, the key index system comprises at least one key quantitative parameter and at least one key qualitative parameter, which are critical for the state evaluation of the photovoltaic inverter.

On the basis of obtaining the original data of the photovoltaic inverter, screening the original data according to a key index system to obtain screened original data. Wherein the raw data after screening comprises data related to at least one key quantitative parameter and data related to at least one key qualitative parameter.

Step S103, preprocessing the screened original data to obtain the target data.

Specifically, in order for the screened raw data to meet the input specifications of the trained state evaluation model, the screened raw data needs to be preprocessed.

The preprocessing may include one or more of the following steps: (1) data cleaning: and checking whether incomplete or erroneous data such as missing values, abnormal values and the like exist in the data, and then repairing, deleting or filling the data according to the situation. (2) Data conversion: the data is transformed to meet the requirements of the model. For example, continuous data is normalized (the mean value is 0, the standard deviation is 1) or normalized (the data is scaled to a specific range). (3) Feature selection: the selection of the appropriate features or attributes may be based on the importance, relevance, etc. of the features to reduce unnecessary dimensions, as desired for the problem. (4) Feature extraction: more meaningful features are extracted from the raw data. For example, statistical features are extracted from time-series data, texture features are extracted from images, and the like. (5) Data balancing: in dealing with classification problems, it may be necessary to deal with cases of class imbalance, such as balancing the number of samples of different classes by undersampling or oversampling. (6) And (3) data coding: the classification data is converted to a digital representation so that the model can be processed. For example, category labels are thermally encoded one-time. (7) Data dimension reduction: by reducing the data dimension, the computational complexity and risk of model overfitting is reduced. Common methods include Principal Component Analysis (PCA) and the like.

After preprocessing the screened original data, target data can be obtained. It will be appreciated that the goal of the preprocessing is to clean, transform and normalize the data, and the process of preprocessing can help to improve the performance and stability of the model, enabling the model to better adapt to real data and make accurate state evaluations.

According to the scheme, the original data of the photovoltaic inverter are obtained; screening the original data according to the key parameter system of the photovoltaic inverter to obtain screened original data; preprocessing the screened original data to obtain the target data. In this embodiment, the original data is screened and preprocessed based on the key parameter system, so that the target data which meets the input specification of the trained state evaluation model can be obtained, and a more reliable state evaluation result is obtained based on the target data.

Further, referring to fig. 5, a flow chart is provided according to a fourth embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 3, step S02, based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter, a basic index system of the photovoltaic inverter is further refined, which includes:

Step S021, analyzing at least one quantitative parameter of the photovoltaic inverter based on a preset first analysis rule to obtain the corresponding sensitive factor of the at least one quantitative parameter.

Specifically, in order to reveal the importance of a certain quantitative parameter to the state evaluation of the photovoltaic inverter, a mode of calculating the sensitivity factor may be adopted, and at least one quantitative parameter of the photovoltaic inverter is analyzed based on a preset first analysis rule, so as to obtain the sensitivity factor corresponding to each of the at least one quantitative parameter. Wherein the sensitivity factor is represented in the form of a numerical value.

Step S022, analyzing at least one qualitative parameter of the photovoltaic inverter based on a preset second analysis rule to obtain degradation factors corresponding to the at least one qualitative parameter.

Specifically, in order to reveal the importance of a certain qualitative parameter to the state evaluation of the photovoltaic inverter, a mode of calculating the degradation factor may be adopted, and at least one qualitative parameter of the photovoltaic inverter is analyzed based on a preset second analysis rule, so as to obtain the degradation factor corresponding to each of the at least one qualitative parameter. Wherein the degradation factor is expressed in the form of a numerical value.

And step S023, constructing and obtaining a basic index system of the photovoltaic inverter based on the at least one quantitative parameter, the sensitive factor corresponding to each of the at least one quantitative parameter, the at least one qualitative parameter and the degradation factor corresponding to each of the at least one qualitative parameter.

Specifically, a basic index system of the photovoltaic inverter can be constructed and obtained based on at least one quantitative parameter, a sensitive factor corresponding to the at least one quantitative parameter, at least one qualitative parameter and a degradation factor corresponding to the at least one qualitative parameter. The magnitude of the sensitive factor is positively correlated with the sensitivity of the quantitative parameter, and is used for strengthening the key quantitative parameter and weakening the non-key quantitative parameter; the magnitude of the degradation factor is positively correlated with the degradation degree of the qualitative parameter, and is used for strengthening the key qualitative parameter and weakening the non-key qualitative parameter.

According to the scheme, at least one quantitative parameter of the photovoltaic inverter is analyzed based on a preset first analysis rule, so that the sensitivity factors corresponding to the at least one quantitative parameter are obtained; analyzing at least one qualitative parameter of the photovoltaic inverter based on a preset second analysis rule to obtain degradation factors corresponding to the at least one qualitative parameter; and constructing and obtaining a basic index system of the photovoltaic inverter based on the at least one quantitative parameter, the sensitive factor corresponding to the at least one quantitative parameter, the at least one qualitative parameter and the degradation factor corresponding to the at least one qualitative parameter. In the embodiment, a basic index system of the photovoltaic inverter is constructed by introducing the sensitive factors and the degradation factors, under the basic index system, the influence of quantitative parameters and qualitative parameters on state evaluation is quantized, the basis can be provided for state evaluation and monitoring of the photovoltaic inverter, and more accurate and effective operation analysis and maintenance decision can be realized.

Further, referring to fig. 6, a flow chart is provided in a fifth embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 5, the at least one quantitative parameter is assigned to a feature type corresponding to each at least one quantitative parameter, the at least one quantitative parameter includes at least one quantitative parameter sample, and in step S021, based on a preset first analysis rule, the at least one quantitative parameter of the photovoltaic inverter is analyzed to obtain a further refinement of a sensitivity factor corresponding to each at least one quantitative parameter, where the further refinement includes:

step S0211, according to the quantitative parameter samples corresponding to at least one quantitative parameter under each characteristic type, calculating to obtain the average intra-class distance and the average inter-class distance corresponding to at least one quantitative parameter under each characteristic type.

In particular, feature types refer to different types or classes of features or attributes, which can be selected and defined according to specific circumstances to better describe the state features of the photovoltaic inverter. Each feature type may have an effect on the state of the photovoltaic inverter, so that its corresponding quantitative parameter needs to be analyzed in the state evaluation.

In-class distance for the j-th quantitative parameter of the i-th characteristic type of the photovoltaic inverter and average in-class distance for the j-th quantitative parameter

Wherein D is _ij Class inner distance of jth quantitative parameter of ith feature typeSeparating; n (N) _i The number of samples of the quantitative parameter for the ith feature type in a fault; f (f) _ij (m)、f _ij (n) an mth sample and an nth sample in a jth quantitative parameter of an ith feature type, respectively; d, d ² Is f _ij (m) and f _ij (n) square of the distance d between; n is the number of feature types;

calculating average inter-class distance D of jth quantitative parameter of ith characteristic type of inverter _j ：

Wherein,the m-th sample average value and the n-th sample average value in the j-th quantitative parameter of the i-th feature type are respectively.

Step S0212, based on the average intra-class distance and the average inter-class distance, calculating to obtain the sensitivity factor corresponding to at least one quantitative parameter under each characteristic type.

Specifically, the average intra-class distance is calculatedAverage inter-class distance D _j Then, the sensitivity factor alpha of the jth quantitative parameter can be further calculated _j ：

Wherein alpha is _j The larger the j-th quantitative parameter is, the more sensitive the j-th quantitative parameter is, and the higher the rationality of selecting the j-th quantitative parameter for the state evaluation of the photovoltaic inverter is.

According to the scheme, the average intra-class distance and the average inter-class distance corresponding to at least one quantitative parameter under each characteristic type are calculated according to the quantitative parameter samples corresponding to the at least one quantitative parameter under each characteristic type; and calculating to obtain the sensitivity factor corresponding to at least one quantitative parameter under each characteristic type based on the average intra-class distance and the average inter-class distance. In the embodiment, the sensitivity factor corresponding to the quantitative parameter is calculated based on the average intra-class distance and the average inter-class distance, and is used for strengthening the key quantitative parameter and weakening the non-key quantitative parameter, so that more accurate and effective operation analysis and maintenance decision can be realized.

Further, referring to fig. 7, a flowchart is provided in a sixth embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 6, the at least one qualitative parameter is assigned to a respective corresponding feature type, where each of the at least one qualitative parameter includes a qualitative parameter actual value, a qualitative parameter threshold value, and a qualitative parameter factory value, and in step S022, at least one qualitative parameter of the photovoltaic inverter is analyzed based on a preset second analysis rule, so as to obtain further refinement of a degradation factor corresponding to each of the at least one qualitative parameter, where the further refinement includes:

Step S0221, calculating and obtaining degradation factors corresponding to at least one qualitative parameter under each characteristic type according to a qualitative parameter actual value, a qualitative parameter threshold value and a qualitative parameter factory value corresponding to the at least one qualitative parameter under each characteristic type respectively and based on a preset offset calculation rule.

In particular, feature types refer to different types or classes of features or attributes, which can be selected and defined according to specific circumstances to better describe the state features of the photovoltaic inverter. Each feature type may have an effect on the state of the photovoltaic inverter, so that its corresponding qualitative parameter needs to be analyzed in the state evaluation.

Aiming at a certain qualitative parameter of the photovoltaic inverter, a calculation formula of a degradation factor beta is adopted under a preset offset calculation rule, wherein the calculation formula is as follows:

wherein x is _i Is the actual value of qualitative parameter x _k Is a qualitative parameter threshold value, x _b Is a qualitative parameter factory value. X is x _i -x _k For a first type of offset, x, of the actual value of the qualitative rating with respect to the threshold value of the qualitative rating _b -x _k For the second class of offset of the qualitative parameter relative to the qualitative parameter threshold, the degradation factor beta obtained by dividing the first class of offset by the second class of offset is used for representing the relative degradation degree between the actual state and the factory state related to the qualitative parameter, and the smaller the degradation factor beta of the qualitative parameter is, the better the equipment state of the photovoltaic inverter is.

According to the scheme, the degradation factor corresponding to the at least one qualitative parameter under each characteristic type is obtained through calculation based on a preset offset calculation rule according to the qualitative parameter actual value, the qualitative parameter threshold value and the qualitative parameter factory value corresponding to the at least one qualitative parameter under each characteristic type. In this embodiment, the degradation factor corresponding to the qualitative parameter is calculated based on the actual value of the qualitative parameter, the qualitative parameter threshold value, and the factory value of the qualitative parameter, where the degradation factor is used to strengthen the critical qualitative parameter, weaken the non-critical qualitative parameter, and help to achieve more accurate and effective operation analysis and maintenance decision.

Further, referring to fig. 8, a flow chart is provided in a seventh embodiment of the state evaluation method of the present application, based on the embodiment shown in fig. 3, step S03, based on the basic index system of the photovoltaic inverter, constructs a key index system of the photovoltaic inverter to be further refined, which includes:

step S031, screening at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter based on a basic index system of the photovoltaic inverter to obtain at least one preliminary screening quantitative parameter and at least one preliminary screening qualitative parameter.

Specifically, under the basic index system of the photovoltaic inverter, the influence of quantitative parameters and qualitative parameters on state evaluation is quantified. Therefore, based on the sensitivity factor and the degradation factor as references, a part of quantitative parameters and qualitative parameters can be selected as the preliminary screening quantitative parameters and the preliminary screening qualitative parameters. Wherein, the quantity of the preliminary screening quantitative parameter and the preliminary screening qualitative parameter are at least one.

Step S032, based on a preset Gaussian kernel function, quantitative parameter samples corresponding to the at least one preliminary screening quantitative parameter and qualitative parameter samples corresponding to the at least one preliminary screening qualitative parameter, constructing and obtaining a normalized kernel matrix.

Specifically, based on at least one quantitative parameter sample corresponding to the preliminary screening quantitative parameters and at least one qualitative parameter sample corresponding to the preliminary screening qualitative parameters, an original kernel matrix K is constructed by adopting a preset Gaussian kernel function.

Then, the original kernel matrix K is normalized, and a normalized kernel matrix K' can be obtained.

Step S033, based on the normalized kernel matrix, constructing and obtaining a corresponding feature matrix and a feature vector matrix.

Specifically, the feature matrix K1 and the feature vector matrix K2 in the form of diagonal arrays can be constructed by performing calculation based on the normalized kernel matrix K'.

Step S034, constructing and obtaining a target matrix based on the normalized kernel matrix, the feature matrix and the feature vector matrix.

Specifically, a plurality of corresponding eigenvalues and a contribution rate corresponding to each eigenvalue are calculated based on the eigenvalue matrix K1, wherein the contribution rate is the sum of a certain eigenvalue divided by all eigenvalues.

Then, the feature values are ranked according to the order of the feature value contribution rate from large to small, the feature values ranked in front are selected, and the reserved feature quantity can be determined according to a preset threshold value or the accumulated contribution rate. A new matrix K11 is constructed based on the selected eigenvalues.

Then, the eigenvectors of the eigenvector matrix K2 are ordered according to the eigenvalue order of the matrix K11 to obtain a matrix K21. A new matrix K3 can be obtained by dividing each row of the matrix K21 by the eigenvalue of the corresponding matrix K11.

Finally, a target matrix KK can be obtained according to the normalized kernel matrix K' and the matrix K3:

KK＝K3*K'

and step 035, constructing a key index system of the photovoltaic inverter based on the target matrix.

Specifically, after the target matrix KK is obtained, each column (which may also be understood as each dimension) of the target matrix KK may be ordered. Further, the key indicators may be selected according to a plurality of columns in the sorted target matrix KK. For example, a first column and a second column in the target matrix KK would correspond to a key first index and a key second index, respectively.

The method can be used for selecting key indexes by combining the target matrix KK in an application scene of actual working conditions, and the key indexes are helpful for more accurately evaluating the state of the photovoltaic inverter and play an important role in actual application.

It is understood that the key index system is constructed based on the key index, and the key index has corresponding quantitative parameters and qualitative parameters, that is, the key index system includes the key quantitative parameters and the key qualitative parameters.

According to the scheme, at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter are screened based on a basic index system of the photovoltaic inverter, so that at least one preliminary screening quantitative parameter and at least one preliminary screening qualitative parameter are obtained; constructing and obtaining a normalized kernel matrix based on a preset Gaussian kernel function, quantitative parameter samples corresponding to the at least one preliminary screening quantitative parameter and qualitative parameter samples corresponding to the at least one preliminary screening qualitative parameter; constructing and obtaining a corresponding feature matrix and a feature vector matrix based on the normalized kernel matrix; constructing a target matrix based on the normalized kernel matrix, the feature matrix and the feature vector matrix; and constructing a key index system of the photovoltaic inverter based on the target matrix. In this embodiment, by selecting the key index from the basic index system in the above mode of improving the nonlinear mapping dimension reduction of the core, the key index system is further constructed, so that an effective reference can be provided for accurately evaluating the state of the photovoltaic inverter.

Further, referring to fig. 9, an eighth embodiment of the state evaluation method of the present application provides a flowchart, based on the embodiment shown in fig. 3, step S05, training the initial support vector machine model based on the sample data and the preset optimizing parameters, to obtain further refinement of the trained state evaluation model, which includes:

and step S051, preprocessing the sample data to obtain training samples and test samples.

Specifically, with reference to fig. 10, fig. 10 is a schematic diagram of model training related to the state evaluation method of the present application. After the sample data is acquired, the sample data may be preprocessed, the preprocessing being similar to step S103 of the third embodiment of the state evaluation method of the present application. Unlike step S103, the preprocessing procedure of the present embodiment adds a data dividing step, which refers to dividing sample data into training samples and test samples so as to use independent data in training and testing models.

Further, the training samples and the test samples can be normalized, so that the feature scale difference of the training samples and the test samples is eliminated in model training, convergence is accelerated, stability and performance are improved, and data consistency is maintained.

And step S052, training the initial support vector machine model based on the training sample.

Specifically, the training sample is used as the input of the initial support vector machine model, and the initial support vector machine model is adjusted and trained, so that the initial support vector machine model can be better adapted to actual data, and the prediction accuracy and performance of the model are improved.

And step S053, judging whether the initial support vector machine model meets a preset cross verification condition.

Specifically, after the training step is finished, it may be determined whether the initial support vector machine model satisfies a preset cross-validation condition, and the determining process involves cross-validation.

Cross-validation is a method of evaluating the performance of a machine learning model by dividing an existing data set into a plurality of subsets, then using a portion of the data set as a validation set a plurality of times, and the remaining portion as a training set, alternating in turn to evaluate the performance of the model on different data subsets. This helps to verify the generalization ability of the model under different data conditions, avoiding the problems of over-fitting or under-fitting. The most common cross-validation method is k-fold cross-validation, which divides the data set into k subsets, selecting one subset at a time as the validation set, the remaining k-1 subsets as the training set, repeating this process multiple times and calculating the average performance index to yield a more accurate assessment of the model performance.

It can be appreciated that if the initial support vector machine model passes the cross-validation, the cross-validation condition is satisfied; if the initial support vector machine model fails the cross-validation, the cross-validation condition is not satisfied.

And step S054, if the initial support vector machine model does not meet the cross verification condition, training the initial support vector machine model based on the training sample and preset optimizing parameters until the initial support vector machine model meets the cross verification condition, and stopping training the initial support vector machine model to obtain the state evaluation model to be tested.

Specifically, if the initial support vector machine model does not satisfy the cross-validation condition, training of the initial support vector machine model is continued. In the subsequent training, the optimizing parameters can be added on the basis of training samples as the input of the initial support vector machine model. The optimizing parameters can provide more adjusting options, influence the learning process and result of the initial support vector machine model, enable the initial support vector machine model to be better adapted to training samples, and improve the generalization capability, accuracy, stability and reliability of the initial support vector machine model.

In this way, training of the initial support vector machine model can be stopped after repeated training until the initial support vector machine model meets the cross validation condition, and a state evaluation model to be tested is obtained.

In some cases, after step S053, if the initial support vector machine model satisfies the cross-validation condition, training of the initial support vector machine model is stopped, resulting in a state evaluation model to be tested. At this time, the state evaluation model to be tested has better generalization performance, but still needs to further execute the test.

And step S055, testing the state evaluation model to be tested based on the test sample.

Specifically, a test sample is used as an input of a state evaluation model to be tested, and the state evaluation model to be tested is tested. Correspondingly, the state evaluation model to be tested outputs state evaluation information of the test sample according to the learned relation and rule.

The test sample serves to verify the performance and generalization ability of the state assessment model to be tested. Through the performance on the test sample, the prediction capability of the state evaluation model to be tested on unseen data can be evaluated, and whether the model has better popularization capability is judged, so that the accuracy and the reliability of the state evaluation model to be tested in practical application are judged.

And step S056, if the state evaluation model to be tested passes the test, obtaining the trained state evaluation model.

Specifically, if the state evaluation information of the test sample output by the state evaluation model to be tested meets the preset requirement, the state of the photovoltaic inverter can be accurately simulated and evaluated, then it can be determined that the state evaluation model to be tested passes the test, and the state evaluation model to be tested passing the test is further used as a trained state evaluation model. It will be appreciated that the trained state estimation model is a modified SVM model whose optimal set of parameters can be expressed as (C, σ) ² ). Wherein C is trainingRegularization parameters, commonly referred to as soft spacing parameters, in the refined state estimation model are used to control the complexity and tolerance of the model. Smaller values of C lead to more relaxed intervals, tolerating some classification errors of the training data, and making the model more generalizable. Larger values of C may result in tighter intervals, forcing the model to better fit the training data, but may result in overfitting. Sigma (sigma) ² Is a bandwidth parameter of a kernel function, typically used with radial basis function (Radial Basis Function, RBF) kernels. The bandwidth parameters determine the shape of the kernel function, affecting the mapping of the data in high-dimensional space. Smaller sigma ² The values may result in steeper kernel functions and more focused mapping of data points in high dimensional space, potentially resulting in a model that is sensitive to noise. Larger sigma ² Values may result in smoother kernel functions, more scattered mapping of data points in high dimensional space, and may improve generalization ability of the model.

If the state evaluation information of the test sample output by the state evaluation model to be tested does not meet the preset requirement, the state evaluation model to be tested may need to be continuously trained, or the training step may be modified by adopting other measures so as to be converged.

According to the embodiment, through the scheme, specifically through preprocessing the sample data, a training sample and a test sample are obtained; training the initial support vector machine model based on the training sample; judging whether the initial support vector machine model meets a preset cross verification condition or not; if the initial support vector machine model meets the cross verification condition, training the initial support vector machine model is stopped, and a state evaluation model to be tested is obtained; if the initial support vector machine model does not meet the cross verification condition, training the initial support vector machine model based on the training sample and preset optimizing parameters until the initial support vector machine model meets the cross verification condition, stopping training the initial support vector machine model, and obtaining the state evaluation model to be tested; testing the state evaluation model to be tested based on the test sample; and if the state evaluation model to be tested passes the test, obtaining the trained state evaluation model. In the embodiment, the trained state evaluation model is ensured to have stronger generalization capability through the steps of preprocessing data, model training, cross validation, parameter optimization, model test and the like, the accuracy and the reliability of a prediction result are improved, and a reliable tool is provided for state evaluation of the photovoltaic inverter.

Further, referring to fig. 11, a flowchart is provided in a ninth embodiment of the state evaluation method according to the present application, based on the embodiment shown in fig. 9, step S054, training the initial support vector machine model based on the training samples and preset optimizing parameters until the initial support vector machine model meets the cross validation condition, and before obtaining the state evaluation model to be tested, further includes:

step S057, initializing at least one optimizing parameter based on a preset wolf algorithm.

Specifically, the optimizing parameters play a role in adjusting and optimizing the performance of the model in model training, so that the generalization capability of the model on new data can be improved, and the robustness and reliability of the model are improved.

For this purpose, the wolf algorithm is an optimization algorithm, and at least one optimization parameter may be initialized based on a preset wolf algorithm. The optimizing parameters may include one or more of scale, scaling factor, crossover probability, number of iterations, etc. in parameter optimization.

The process of initializing the optimization parameters may be: (1) initializing a parameter space: a range of values for each parameter is defined, which parameters affect the performance of the model. (2) Initializing wolf group: an initial set of wolf individuals, each individual representing a set of parameter values, was set. (3) Calculating the fitness: the fitness of each individual wolf is calculated using an objective function that represents the performance of the model on the training data. (4) Updating the gray wolf position: the position of the wolf is updated using algorithmic rules to better explore the parameter space, based on the relationship between fitness values and individual wolves. (5) Searching for the optimal individual: the optimal individual in the wolf group, i.e. the combination of parameters with the best performance, is determined from the fitness value. (6) Updating parameters: and using the parameters of the current optimal individual as new parameter values, and continuing to optimize the model. (7) Iterative optimization: repeating the steps 3 to 6 until the preset iteration times or convergence conditions are reached. (8) Obtaining optimal optimizing parameters: and after the iteration is finished, obtaining the optimal combination of the optimizing parameters for training the model.

It can be appreciated that, besides the wolf algorithm, algorithms such as grid search, random search, bayesian optimization, genetic algorithm, particle swarm optimization, simulated annealing, local optimization and the like can be used to set the optimizing parameters corresponding to the initial support vector machine model.

Compared with other algorithms for initializing optimizing parameters, the gray wolf algorithm has the following advantages: (1) the number of parameters of the gray wolf algorithm is smaller than that of other complex algorithms, and only a few parameters such as population size, iteration number and the like are usually required to be set, so that the optimization parameter is easier to adjust. (2) The gray wolf algorithm realizes global search of the parameter space by simulating the cooperative behavior of the gray wolf population, has certain advantages when the global optimal solution is found, and particularly has good performance in the high-dimensional parameter space. (3) The gray wolf algorithm has self-adaptability, can automatically adjust the search strategy according to the nature of the problem, and is suitable for a plurality of different types of problems, including continuous parameter optimization and discrete parameter optimization. (4) The gray wolf algorithm has good parallelism, can be easily operated in a multi-core processor or a distributed computing environment, and accelerates the searching process. (5) The gray wolf algorithm is generally less sensitive to the choice of the initial solution and therefore is somewhat robust and can handle the diversity of different problem areas.

The embodiment initializes at least one optimizing parameter based on the preset gray wolf algorithm through the scheme. In the embodiment, the gray wolf algorithm is adopted to automatically search the parameter space, optimize the model performance, improve the model accuracy and accelerate convergence, and effectively avoid the complexity and uncertainty of manually adjusting the parameters.

Further, referring to fig. 12, a flowchart is provided in a tenth embodiment of the state evaluation method according to the present application, based on the embodiment shown in fig. 2, step S20, after inputting the target data into the trained state evaluation model to obtain the state evaluation information of the photovoltaic inverter, further includes:

step S30, if the state evaluation information of the photovoltaic inverter accords with a preset alarm condition, determining an alarm category corresponding to the state evaluation information of the photovoltaic inverter.

In particular, in order to timely identify faults of the photovoltaic inverter and take measures to prevent possible fault enlargement and damage, an alarm may be issued when the state evaluation information of the photovoltaic inverter meets preset alarm conditions.

Specifically, the present embodiment adopts a multi-category alarm mode, and first, the alarm category corresponding to the state evaluation information of the photovoltaic inverter needs to be determined.

For example, the alarm categories corresponding to the state evaluation information are classified into four types Z1, Z2, Z3, and Z4. Wherein Z1 represents the normal running state of the photovoltaic inverter, and the running index of the photovoltaic inverter corresponding to Z1 is normal; z2 represents the general running state of the photovoltaic inverter, the running index data of the photovoltaic inverter corresponding to Z2 fluctuates, and operation and maintenance personnel need to continuously pay attention to the fluctuation condition of each index data; z3 represents an unstable running state of the photovoltaic inverter, the running index data of the photovoltaic inverter corresponding to Z3 deviates from a normal value, and an operation and maintenance person needs to analyze the abnormal data and make a corresponding maintenance plan to prevent faults; z4 represents the abnormal operation state of the photovoltaic inverter, and the key data of the operation index of the photovoltaic inverter corresponding to Z4 is abnormal, so that the corresponding index is required to be immediately analyzed and timely eliminated.

Step S40, pushing corresponding alarm information according to the alarm category.

Specifically, corresponding alert information may be pushed for different alert categories.

For example, for the four alarm categories Z1, Z2, Z3, and Z4, when the alarm category corresponding to the current state evaluation information of the photovoltaic inverter is Z1, the corresponding alarm identifier may be displayed in the background; when the alarm category corresponding to the current state evaluation information of the photovoltaic inverter is Z2, displaying a corresponding alarm mark in the background and emitting alarm sound; when the alarm category corresponding to the current state evaluation information of the photovoltaic inverter is Z3, a corresponding alarm mark can be displayed in the background, an alarm sound is sent out, and alarm information is pushed to a manager; when the alarm category corresponding to the current state evaluation information of the photovoltaic inverter is Z4, a corresponding alarm identifier can be displayed in the background, alarm sound is generated, alarm information is pushed to a manager and operation and maintenance personnel near the abnormal photovoltaic inverter, and the alarm information can comprise the position information of the abnormal photovoltaic inverter, so that the manager or the operation and maintenance personnel can go to the site in time to process the abnormality.

The foregoing examples merely illustrate that the alert information is pushed according to alert categories, and the core of the alert information is that the content of the alert information is associated with alert categories, and different alert categories correspond to different alert information.

According to the scheme, the alarm category corresponding to the state evaluation information of the photovoltaic inverter is determined if the state evaluation information of the photovoltaic inverter meets the preset alarm condition; and pushing corresponding alarm information according to the alarm category. In the embodiment, the corresponding alarm information is pushed pertinently on the basis of determining the alarm type, so that the problem of the photovoltaic inverter can be accurately indicated, the operation and maintenance response and repair are accelerated, and the fault loss of the photovoltaic inverter is effectively reduced.

In addition, an embodiment of the present application further provides a state evaluation device, where the state evaluation device includes:

The principle and implementation process of the state evaluation in this embodiment are referred to the above embodiments, and are not repeated herein.

In addition, the embodiment of the application also provides a terminal device, which comprises a memory, a processor and a state evaluation program stored on the memory and capable of running on the processor, wherein the state evaluation program realizes the steps of the state evaluation method when being executed by the processor.

Because the state evaluation program is executed by the processor and adopts all the technical schemes of all the embodiments, the state evaluation program at least has all the beneficial effects brought by all the technical schemes of all the embodiments and is not described in detail herein.

Furthermore, the embodiments of the present application also propose a computer-readable storage medium, on which a state evaluation program is stored, which when executed by a processor implements the steps of the state evaluation method as described above.

Compared with the prior art, the state evaluation method, the state evaluation device, the terminal equipment and the storage medium provided by the embodiment of the application are used for acquiring the target data of the photovoltaic inverter; inputting the target data into a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, wherein the trained state evaluation model is obtained by training an initial support vector machine model based on sample data and preset optimizing parameters of the photovoltaic inverter, the sample data is obtained based on an index system of the photovoltaic inverter, and the index system is obtained by constructing quantitative parameters and qualitative parameters of the photovoltaic inverter. Based on the scheme, related data can be fully mined and utilized by combining quantitative parameters and qualitative parameters to construct an index system, the defect of single data type is overcome, a trained state evaluation model has more comprehensive state evaluation capability, and the accuracy and reliability of state evaluation of the photovoltaic inverter are improved.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.

The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as above, including several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, a controlled terminal, or a network device, etc.) to perform the method of each embodiment of the present application.

The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the claims, and all equivalent structures or equivalent processes using the descriptions and drawings of the present application, or direct or indirect application in other related technical fields are included in the scope of the claims of the present application.

Claims

1. A state evaluation method, characterized in that the state evaluation method comprises:

acquiring target data of a photovoltaic inverter;

2. The method of claim 1, wherein prior to the step of obtaining the target data for the photovoltaic inverter, further comprising:

3. The method of state evaluation of claim 2, wherein the step of obtaining target data of the photovoltaic inverter comprises:

acquiring original data of a photovoltaic inverter;

preprocessing the screened original data to obtain the target data.

4. The state evaluation method according to claim 2, wherein the step of constructing a basic index system of the photovoltaic inverter based on at least one quantitative parameter and at least one qualitative parameter of the photovoltaic inverter comprises:

5. The state evaluation method according to claim 4, wherein the at least one quantitative parameter belongs to a respective corresponding feature type, the number of the feature types is at least one, the at least one quantitative parameter each includes at least one quantitative parameter sample, and the step of analyzing the at least one quantitative parameter of the photovoltaic inverter based on a preset first analysis rule to obtain a sensitivity factor corresponding to the at least one quantitative parameter each includes:

6. The state evaluation method according to claim 5, wherein the at least one qualitative parameter belongs to a respective corresponding feature type, the at least one qualitative parameter each includes a qualitative parameter actual value, a qualitative parameter threshold value, and a qualitative parameter factory value, and the step of analyzing the at least one qualitative parameter of the photovoltaic inverter based on a preset second analysis rule to obtain a degradation factor corresponding to the at least one qualitative parameter each includes:

7. The state evaluation method according to claim 2, wherein the step of constructing a key index system of the photovoltaic inverter based on the basic index system of the photovoltaic inverter includes:

8. The state estimation method according to claim 2, wherein the step of training the initial support vector machine model based on the sample data and the preset optimizing parameters to obtain the trained state estimation model includes:

preprocessing the sample data to obtain a training sample and a test sample;

training the initial support vector machine model based on the training sample;

if the initial support vector machine model does not meet the cross verification condition, training the initial support vector machine model based on the training sample and preset optimizing parameters until the initial support vector machine model meets the cross verification condition, and stopping training the initial support vector machine model to obtain a state evaluation model to be tested;

Testing the state evaluation model to be tested based on the test sample;

9. The state evaluation method according to claim 8, wherein the step of training the initial support vector machine model based on the training samples and preset optimizing parameters until the initial support vector machine model satisfies the cross validation condition, and before the step of obtaining the state evaluation model to be tested, further comprises:

10. The state evaluation method according to claim 1, wherein after the step of inputting the target data to a trained state evaluation model to obtain state evaluation information of the photovoltaic inverter, further comprising:

and pushing corresponding alarm information according to the alarm category.

11. A state evaluation device, characterized in that the state evaluation device comprises:

12. A terminal device, characterized in that it comprises a memory, a processor and a state evaluation program stored on the memory and executable on the processor, which state evaluation program, when executed by the processor, implements the steps of the state evaluation method according to any one of claims 1-10.

13. A computer-readable storage medium, on which a state evaluation program is stored, which, when executed by a processor, implements the steps of the state evaluation method according to any one of claims 1-10.