A substation robot inspection data quality evaluation method, system and device

Abstract

The application discloses a transformer substation robot inspection data quality evaluation method, system and equipment, relates to the technical field of robot power inspection, and comprises the following steps: for any inspection data in the inspection data set, determining a corresponding first parameter attribute based on a first preset decision tree; determining a corresponding second parameter attribute based on a second preset decision tree; determining a corresponding third parameter attribute based on a third preset decision tree; determining a corresponding fourth parameter attribute based on a fourth preset decision tree; determining a corresponding data quality score based on a fifth preset decision tree; determining a data quality comprehensive score corresponding to the inspection data set according to the data quality score corresponding to each inspection data in the inspection data set; and sending the inspection data set to a transformer substation monitoring background when the data quality comprehensive score meets a preset quality requirement. The application realizes comprehensive quality evaluation of inspection data at the robot end.

Description

Technical Field

[0001] This invention relates to the field of robotic power inspection technology, and in particular to a method, system and equipment for evaluating the quality of substation robotic inspection data. Background Technology

[0002] Inspection is a crucial method commonly used in substations to promptly identify equipment faults and potential safety hazards. With the rapid development of artificial intelligence, robotic inspection has become the most important inspection method in substations due to its flexibility and weather-independent nature. While robotic inspection simplifies inspection work and brings opportunities for power grid development, the massive amounts of diverse and inconsistent quality of inspection data it generates significantly increase the data transmission and processing pressure on both the inspection robot and the substation monitoring backend, placing an unbearable burden on the monitoring system.

[0003] As the front end for data transmission during inspections, inspection robots are primarily used as simple tools for daily monitoring and data collection, with limited capacity for data analysis. Exploring ways to improve the inspection robot's data analysis capabilities is an effective way to alleviate the data transmission and processing pressure on the monitoring backend. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and equipment for evaluating the quality of substation robot inspection data, which enables comprehensive quality evaluation of inspection data at the inspection robot end.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] In a first aspect, the present invention provides a method for evaluating the quality of substation robot inspection data, applied to inspection robots, the method comprising:

[0007] Obtain the inspection dataset obtained from a single substation inspection by the inspection robot;

[0008] For any inspection data in the inspection dataset, a corresponding first parameter attribute is determined based on a first preset decision tree; the first parameter attribute is whether the data is timely or untimely.

[0009] Based on the second preset decision tree, the corresponding second parameter attribute is determined; the second parameter attribute is whether the data is complete or incomplete.

[0010] Based on the third preset decision tree, the corresponding third parameter attribute is determined; the third parameter attribute is whether the data is duplicated or the data is not duplicated.

[0011] Based on the fourth preset decision tree, the corresponding fourth parameter attribute is determined; the fourth parameter attribute is whether the data is accurate or the data is inaccurate.

[0012] Based on the fifth preset decision tree, the corresponding data quality score is determined according to the first parameter attribute, the second parameter attribute, the third parameter attribute and the fourth parameter attribute;

[0013] Based on the data quality score corresponding to each inspection data in the inspection dataset, a comprehensive data quality score corresponding to the inspection dataset is determined, and when the comprehensive data quality score meets the preset quality requirements, the inspection dataset is sent to the substation monitoring backend.

[0014] Secondly, this invention provides a substation robot inspection data quality evaluation system, applied to inspection robots, the system comprising:

[0015] The data acquisition module is used to acquire the inspection dataset obtained by the inspection robot in a single inspection of a substation.

[0016] The first evaluation module is used to determine the corresponding first parameter attribute for any inspection data in the inspection dataset based on a first preset decision tree; the first parameter attribute is whether the data is timely or untimely.

[0017] The second evaluation module is used to determine the corresponding second parameter attribute based on the second preset decision tree; the second parameter attribute is whether the data is complete or incomplete.

[0018] The third evaluation module is used to determine the corresponding third parameter attribute based on the third preset decision tree; the third parameter attribute is whether the data is duplicated or not.

[0019] The fourth evaluation module is used to determine the corresponding fourth parameter attribute based on the fourth preset decision tree; the fourth parameter attribute is whether the data is accurate or inaccurate.

[0020] The fifth evaluation module is used to determine the corresponding data quality score based on the fifth preset decision tree and according to the first parameter attribute, the second parameter attribute, the third parameter attribute and the fourth parameter attribute;

[0021] The comprehensive evaluation module is used to determine the comprehensive data quality score corresponding to the inspection dataset based on the data quality score corresponding to each inspection data in the inspection dataset, and to send the inspection dataset to the substation monitoring backend when the comprehensive data quality score meets the preset quality requirements.

[0022] Thirdly, the present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform a method for evaluating the quality of substation robot inspection data.

[0023] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0024] This invention discloses a method, system, and device for evaluating the quality of substation robot inspection data. Multiple decision trees are established. At the inspection robot's end, attribute decisions are made on the inspection data obtained from a single substation inspection to obtain a first parameter attribute, a second parameter attribute, a third parameter attribute, and a fourth parameter attribute, thereby determining the data's completeness, accuracy, repeatability, and timeliness. Then, a data quality score is determined based on a fifth preset decision tree, leading to a comprehensive data quality score for the inspection dataset. This invention considers the four aspects of data repeatability, completeness, accuracy, and timeliness, and uses preset decision trees for comprehensive quality evaluation, resulting in higher efficiency and accuracy. Based on the obtained comprehensive data quality score, qualified inspection datasets are sent to the substation monitoring backend, thereby reducing the data transmission and processing pressure on the monitoring backend to some extent for fault identification and health management of substation equipment. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating the substation robot inspection data quality evaluation method of the present invention.

[0027] Figure 2 This is a decision tree diagram showing the timeliness of the data on phase A temperature of the upper section of the surge arrester of transformer #1 in this invention example;

[0028] Figure 3 This is a data integrity decision tree diagram of the A-phase temperature of the upper section of the surge arrester of transformer #1 in this invention example;

[0029] Figure 4 This is a decision tree diagram showing the data repeatability of phase A temperature in the upper section of the surge arrester of transformer #1 in this invention example;

[0030] Figure 5 This is a decision tree diagram showing the data accuracy of phase A temperature in the upper section of the surge arrester of transformer #1 in this invention example;

[0031] Figure 6 This is a data quality scoring decision tree diagram of the A-phase temperature of the upper section of the surge arrester of transformer #1 in this invention example. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] This invention provides a method, system, and device for evaluating the quality of substation robot inspection data. It evaluates the data integrity, accuracy, repeatability, and timeliness of the robot inspection data using a decision tree. Combining the evaluation results of these four indicators, the decision tree determines the quality score for each inspection data point. Then, by accumulating the quality scores of each inspection data point, a comprehensive data quality score for the inspection data collected by the robot is obtained. The qualified inspection data can be applied to the substation monitoring backend for fault identification and health management of substation equipment, thereby reducing the pressure on backend data transmission and processing.

[0034] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] Example 1

[0036] like Figure 1 As shown, this invention provides a method for evaluating the quality of substation robot inspection data, applied to inspection robots. The method includes:

[0037] Step 100: Obtain the inspection dataset obtained by the inspection robot during a single substation inspection, so as to facilitate subsequent automated analysis of the data collected during a single inspection on the inspection robot. Specific analysis indicators include data timeliness, data completeness, data repeatability, and data accuracy.

[0038] Step 200: For any inspection data in the inspection dataset, determine the corresponding first parameter attribute based on the first preset decision tree; the first parameter attribute is whether the data is timely or untimely.

[0039] Specifically, the construction rule of the first preset decision tree is: whether the difference between the actual collection time and the preset collection time of the inspection data is greater than a time difference threshold. When the difference between the actual collection time and the preset collection time of the inspection data is greater than the time difference threshold, the corresponding first parameter attribute is that the data is not timely, that is, the data is not timely data; when the difference between the actual collection time and the preset collection time of the inspection data is less than or equal to the time difference threshold, the corresponding first parameter attribute is that the data is timely, that is, the data is timely data.

[0040] When the inspection data is timely, the timeliness of the data is calculated using the following formula:

[0041]

[0042] in, Indicates the timeliness of the data. This indicates the actual time when the inspection data was collected. The preset collection time for the inspection data is indicated by TQ, where TQ represents the time difference threshold; the inspection data x ij (k) represents the value of the j-th inspection parameter collected by the i-th robot during the k-th inspection of the substation, where i = 1, 2, ..., A, j = 1, 2, ..., B, A represents the total number of robots, and B represents the number of inspection parameters in a single inspection. A and B can be set according to actual needs.

[0043] Step 300: Based on the second preset decision tree, determine the corresponding second parameter attribute; the second parameter attribute is whether the data is complete or incomplete.

[0044] Specifically, the construction rules for the second preset decision tree are: whether the data is empty (φ), whether it is garbled data, and whether it is invalid (null) data. When the inspected data is invalid, the corresponding second parameter attribute is incomplete data, meaning the data is not complete. When the inspected data is not invalid but empty (or null), the corresponding second parameter attribute is incomplete data, meaning the data is not complete. When the inspected data is not invalid, not empty, but garbled, the corresponding second parameter attribute is incomplete data, meaning the data is not complete. When the inspected data is not invalid, not empty, and not garbled, the corresponding second parameter attribute is complete data, meaning the data is complete.

[0045] Step 400: Based on the third preset decision tree, determine the corresponding third parameter attribute; the third parameter attribute is data duplication or data non-duplication.

[0046] Specifically, the third preset decision tree is as follows: when the inspection data is numerical data and satisfies the first repeatability evaluation rule, the corresponding third parameter attribute is data repetition; when the inspection data is numerical data and does not satisfy the first repeatability evaluation rule, the corresponding third parameter attribute is data non-repetition; when the inspection data is text data and satisfies the second repeatability evaluation rule, the corresponding third parameter attribute is data repetition; when the inspection data is text data and does not satisfy the second repeatability evaluation rule, the corresponding third parameter attribute is data non-repetition. The first repeatability evaluation rule is that the difference between the inspection data and the previous inspection data is zero; the second repeatability evaluation rule is that the text similarity between the inspection data and the previous inspection data is greater than a preset similarity threshold.

[0047] The steps for calculating text similarity include:

[0048] (1) For the inspection data (i.e. the kth inspection) and the previous inspection data (i.e. the (k-1)th inspection), word segmentation and stop word removal are performed in sequence to obtain the set of effective feature words for inspection and the set of effective feature words for the previous inspection.

[0049] (2) The set of valid feature words for the inspection and the set of valid feature words for the previous inspection are merged and deduplicated to obtain the set of inspection words; the set of inspection words corresponding to the j-th inspection parameter in the k-th and k-1-th inspections is: in, Let p represent the p-th inspection word in the set of inspection words corresponding to the j-th inspection parameter during the k-th and k-1-th inspections, where p = 1, 2, ..., P.

[0050] (3) Calculate the word frequency of each word in the inspection word set based on the inspection data and the previous inspection data to obtain the inspection word frequency vector. and the word frequency vector from the last inspection in, Represents the set of inspection terms The number of times the p-th inspected word appears in the set of effective feature words in the k-th inspection of the j-th inspection parameter. Represents the set of inspection terms The number of times the p-th inspected word appears in the set of effective feature words in the (k-1)-th inspection of the j-th inspection parameter, where p = 1, 2, ..., P.

[0051] (4) After normalizing the inspected word frequency vector and the previous inspected word frequency vector using L2-norm, the text similarity is calculated using cosine similarity. The calculation formula is as follows:

[0052]

[0053] in, and These are the word frequencies of the p-th inspected word in both normalized versions.

[0054] Step 500: Based on the fourth preset decision tree, determine the corresponding fourth parameter attribute; the fourth parameter attribute is whether the data is accurate or inaccurate.

[0055] Specifically, when the inspection data is numerical data, the construction rules for the fourth preset decision tree are: whether the data precision is sufficient and whether the data is within the specified range; when the inspection data is text data, the construction rules for the fourth preset decision tree are: whether the data is within the specified language set and whether the data length meets the requirements. Taking numerical data as an example, when the inspection data is within the specified range and the data precision is sufficient, the fourth parameter attribute is data accuracy; when the inspection data is within the specified range but the data precision is insufficient, the fourth parameter attribute is data inaccuracy; when the inspection data is outside the specified range, the fourth parameter attribute is data inaccuracy, that is, the data is not accurate data.

[0056] Step 600: Based on the fifth preset decision tree, determine the corresponding data quality score according to the first parameter attribute, the second parameter attribute, the third parameter attribute, and the fourth parameter attribute.

[0057] The fifth preset decision tree is:

[0058] (1) When the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is inaccurate data, the first value is marked as the corresponding data quality score.

[0059] (2) When the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is non-repeating data, and the fourth parameter attribute is inaccurate data, the second value is marked as the corresponding data quality score.

[0060] (3) When the first parameter attribute is timely data and the second parameter attribute is incomplete data, the third value is marked as the corresponding data quality score. It should be noted that if the data is incomplete, the data is definitely inaccurate. In this case, there is no need to further evaluate its data repeatability and accuracy; it can be uniformly judged as data repeatability and inaccuracy, with the corresponding data quality score being...

[0061] (4) When the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is accurate data, the fourth value is marked as the corresponding data quality score.

[0062] (5) When the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is non-repeating data, and the fourth parameter attribute is accurate data, the fifth value is marked as the corresponding data quality score.

[0063] (6) When the first parameter attribute is that the data is not timely, the value of 0 is marked as the corresponding data quality score; in other words, when the inspection data is not timely, regardless of the evaluation results of its data integrity, data repeatability and data accuracy, its data quality score is 0.

[0064] The first, second, third, fourth, and fifth values ​​are all determined based on the timeliness of the inspection data. The first value is data timeliness + 1; the second value is data timeliness + 2; the third value is data timeliness; the fourth value is data timeliness + 2; and the fifth value is data timeliness + 3.

[0065] The first preset decision tree, the second preset decision tree, the third preset decision tree, the fourth preset decision tree, and the fifth preset decision tree are all generated by the ID3 algorithm;

[0066] The steps for generating a decision tree using the ID3 algorithm include:

[0067] 1) Divide the target inspection sample set into multiple classes and calculate the information entropy Info(D) for each class. The calculation formula is as follows:

[0068]

[0069] Where, p l Let L be the probability that a sample in the target inspection sample set D belongs to the l-th class, and L represent the total number of classes.

[0070] 2) Based on the attributes of each sample in the target inspection sample set, calculate the information entropy Info(D,E) of the corresponding attributes. h The calculation formula is:

[0071]

[0072] Among them, D a Attribute E in the target inspection sample set D h A subset of samples with 'a' values.

[0073] 3) Based on the information entropy of each class and the information entropy of the attribute, calculate the information gain Gain(D,E) of the attribute relative to the target inspection sample set. h The calculation formula is as follows:

[0074] Gain(D,E h =Info(D) - Info(D,E) h ).

[0075] 4) Select the attribute with the largest information gain as the branch node in sequence to complete the construction of each level of decision tree.

[0076] Step 700: Based on the data quality score corresponding to each inspection data point in the inspection dataset, determine the overall data quality score corresponding to the inspection dataset, and when the overall data quality score meets the preset quality requirements, send the inspection dataset to the substation monitoring backend. Assume the inspection data... Data quality score The formula for calculating the comprehensive data quality score of the inspection data collected by the i-th inspection robot during the k-th inspection is:

[0077]

[0078] In one example, the overall data quality score can be set as follows: Excellent (≥90), Good (≥75 <90), Average (≥60 <75), and Poor (<60). Then, the inspection dataset rated as Excellent (i.e., meeting the preset quality requirements) can be sent to the substation monitoring backend, allowing the backend to directly use this dataset for subsequent processing.

[0079] In another specific example, 25 sets of inspection data from inspection robot A in a substation were used to verify the effectiveness of the method described above. Table 1 shows the 25 sets of inspection data from inspection robot A in a substation, of which 23 sets are training samples and the remaining 2 sets are test samples.

[0080] Table 1. Inspection data of inspection robot A in a substation.

[0081]

[0082]

[0083] Regarding data timeliness, taking the A-phase temperature of the upper section of the surge arrester of transformer #1 as an example, Table 2 provides the basic data on its timeliness. Then, combining the first 23 training samples in Table 2, the ID3 algorithm is used to generate a data timeliness decision tree for the inspection parameter of the A-phase temperature of the upper section of the surge arrester of transformer #1, as detailed below. Figure 2 As shown. Finally, using Figure 2 The data timeliness decision tree in the data determines that the remaining two sets of data are timely data.

[0084] Table 2. Expert Review Results on the Timeliness of Data on Phase A Temperature of the Upper Section of the Surge Arrester in Substation #1

[0085]

[0086]

[0087] Regarding data integrity, Table 3 provides the basic data on the temperature of phase A of the upper section of the surge arrester of transformer #1. Then, combining the first 23 training samples in Table 3, the ID3 algorithm is used to generate its data integrity decision tree, as detailed below. Figure 3 As shown. Finally, using Figure 3 The data integrity decision tree in the data determines that the remaining two sets of data are complete data.

[0088] Table 3. Expert Review Results on the Data Integrity of Phase A Temperature of the Upper Section of the Surge Arrester for Transformer #1

[0089]

[0090]

[0091] For data repeatability, the A-phase temperature of the upper section of the surge arrester of transformer #1 is numerical data. Therefore, the evaluation rule for its data repeatability is whether the difference between it and the data from the previous inspection is zero. Table 4 provides the basic data for its data repeatability. Then, combining the first 23 training samples in Table 4, the ID3 algorithm is used to generate its data repeatability decision tree, as follows: Figure 4 As shown. Finally, using Figure 4 The data repetition decision tree in the data determines that the remaining two sets of data are repetitive data and non-repetitive data, respectively.

[0092] Table 4. Expert Review Results on the Repeatability of Phase A Temperature Data in the Upper Section of the Surge Arrester of Transformer #1

[0093]

[0094]

[0095] Regarding data accuracy, the A-phase temperature of the upper section of the surge arrester of transformer #1 is numerical data. Therefore, the evaluation rules for its data accuracy include whether the data precision is sufficient and whether the data is within the specified range. Table 5 provides the basic data for its data accuracy. Then, combining the first 23 training samples in Table 5, the ID3 algorithm is used to generate its data accuracy decision tree, as follows: Figure 5 As shown. Finally, using Figure 5 The data accuracy decision tree in the data determines that the remaining two sets of data are accurate.

[0096] Table 5. Expert Review Results on the Accuracy of Phase A Temperature Data for the Upper Section of the Surge Arrester of Transformer #1

[0097]

[0098]

[0099] For data quality scoring, Table 6 provides the basic data for the A-phase temperature of the upper section of the surge arrester of transformer #1. Then, combining the first 23 training samples in Table 6, the ID3 algorithm is used to generate a data quality scoring decision tree, as detailed below. Figure 6 As shown. Finally, using Figure 6 The data quality scoring decision tree in the dataset determines the data quality scores for the remaining two data sets as Timeliness+2 and Timeliness+3, respectively. Furthermore, as shown in Table 1, the Timeliness of the remaining two data sets is 1; therefore, the data quality scores for the remaining two data sets are 3 and 4, respectively.

[0100] Table 6. Expert Evaluation Results of Data Quality Scoring for Phase A Temperature of Upper Section of Transformer #1 Surge Arrester

[0101]

[0102]

[0103] Similarly, the data quality scores for the remaining 67 detection parameters in the two sets of training samples can be obtained. By summing the data quality scores of these 68 detection parameters, the comprehensive data quality score for the remaining two sets of training samples can be obtained. Table 7 shows the data quality score results for the remaining two sets of training samples.

[0104] Table 7. Data quality score results for the two training samples.

[0105]

[0106]

[0107] The existing Chinese patents, CN201510200863.8 (application date: 2015-04-23) and CN201811543208.2 (application date: 2018-12-17) entitled "A Method and System for Evaluating the Data Quality of Transmission and Transformation Equipment," both target substation equipment, but focus on the condition monitoring data of the equipment, and this data is entirely numerical. In contrast, the method of this invention targets the robot inspection data of substation equipment, which includes numerical data such as transformer temperature, as well as textual data such as the presence or absence of abnormal noises and indicator light status. Therefore, the research object and technical solution of this invention differ from the aforementioned two technologies.

[0108] In summary, this invention evaluates the data integrity, accuracy, repeatability, and timeliness of substation equipment robot inspection data using a decision tree. Combining the evaluation results of these four indicators, the decision tree determines the quality score for each inspection data point. Then, by accumulating the quality scores of each inspection data point, a comprehensive data quality score for the inspection data collected by the inspection robot is obtained. This method can be used to evaluate the quality of various types of substation equipment inspection data collected by inspection robots. The evaluation results provide guidance for the collection of basic substation equipment data using inspection robots and for targeted quality improvement of the collected data. Furthermore, the qualified inspection data after evaluation can be applied to substation monitoring backends for fault identification and health management of substation equipment.

[0109] Example 2

[0110] To implement the technical solution in Embodiment 1 and achieve the corresponding functions and technical effects, this embodiment also provides a substation robot inspection data quality evaluation system, applied to inspection robots. The system includes:

[0111] The data acquisition module is used to acquire the inspection dataset obtained by the inspection robot during a single inspection of a substation.

[0112] The first evaluation module is used to determine the corresponding first parameter attribute for any inspection data in the inspection dataset based on a first preset decision tree; the first parameter attribute is whether the data is timely or untimely.

[0113] The second evaluation module is used to determine the corresponding second parameter attribute based on the second preset decision tree; the second parameter attribute is whether the data is complete or incomplete.

[0114] The third evaluation module is used to determine the corresponding third parameter attribute based on the third preset decision tree; the third parameter attribute is whether the data is duplicated or not.

[0115] The fourth evaluation module is used to determine the corresponding fourth parameter attribute based on the fourth preset decision tree; the fourth parameter attribute is whether the data is accurate or inaccurate.

[0116] The fifth evaluation module is used to determine the corresponding data quality score based on the fifth preset decision tree and according to the first parameter attribute, the second parameter attribute, the third parameter attribute, and the fourth parameter attribute.

[0117] The comprehensive evaluation module is used to determine the comprehensive data quality score corresponding to the inspection dataset based on the data quality score corresponding to each inspection data in the inspection dataset, and to send the inspection dataset to the substation monitoring backend when the comprehensive data quality score meets the preset quality requirements.

[0118] Example 3

[0119] This embodiment provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor runs the computer program to enable the electronic device to execute the substation robot inspection data quality evaluation method of Embodiment 1. Optionally, the above-mentioned electronic device may be a server.

[0120] In addition, this embodiment of the invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the substation robot inspection data quality evaluation method of Embodiment 1.

[0121] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0122] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for evaluating the quality of substation robot inspection data, characterized in that, Applied to inspection robots, the methods include: Obtain the inspection dataset obtained from a single substation inspection by the inspection robot; For any inspection data in the inspection dataset, a corresponding first parameter attribute is determined based on a first preset decision tree; the first parameter attribute is whether the data is timely or untimely. Based on the second preset decision tree, the corresponding second parameter attribute is determined; the second parameter attribute is whether the data is complete or incomplete. Based on the third preset decision tree, the corresponding third parameter attribute is determined; the third parameter attribute is whether the data is duplicated or the data is not duplicated. Based on the fourth preset decision tree, the corresponding fourth parameter attribute is determined; the fourth parameter attribute is whether the data is accurate or the data is inaccurate. Based on the fifth preset decision tree, the corresponding data quality score is determined according to the first parameter attribute, the second parameter attribute, the third parameter attribute and the fourth parameter attribute; The fifth preset decision tree is as follows: when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is inaccurate data, the first value is marked as the corresponding data quality score; when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is non-duplicated data, and the fourth parameter attribute is inaccurate data, the second value is marked as the corresponding data quality score; when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is accurate data, the fourth value is marked as the corresponding data quality score; when the first parameter attribute is untimely data, the value 0 is marked as the corresponding data quality score. The first value, the second value, the third value, the fourth value, and the fifth value are all determined based on the timeliness of the inspection data. Based on the data quality score corresponding to each inspection data in the inspection dataset, a comprehensive data quality score corresponding to the inspection dataset is determined, and when the comprehensive data quality score meets the preset quality requirements, the inspection dataset is sent to the substation monitoring backend.

2. The method for evaluating the quality of substation robot inspection data according to claim 1, characterized in that, The formula for calculating the timeliness of the data is: ； in, Indicates the timeliness of the data. This indicates the actual time when the inspection data was collected. This indicates the preset collection time for inspection data. Indicates the time difference threshold; the inspection data is the first... i The robot in the first k The first sample collected during the second inspection of the substation j The values ​​of each inspection parameter.

3. The method for evaluating the quality of substation robot inspection data according to claim 1, characterized in that, The first value is data timeliness +1; the second value is data timeliness +2; the third value is data timeliness; the fourth value is data timeliness +2; and the fifth value is data timeliness +3.

4. The method for evaluating the quality of substation robot inspection data according to claim 1, characterized in that, The first preset decision tree, the second preset decision tree, the third preset decision tree, the fourth preset decision tree, and the fifth preset decision tree are all generated by the ID3 algorithm; The steps for generating a decision tree using the ID3 algorithm include: The target inspection sample set is divided into multiple classes, and the information entropy of each class is calculated; Based on the attributes of each sample in the target inspection sample set, calculate the information entropy of the corresponding attributes; Based on the information entropy of each class and the information entropy of the attribute, calculate the information gain of the attribute relative to the target inspection sample set; The attribute with the largest information gain is selected as the branch node in turn to complete the construction of each level of the decision tree.

5. The method for evaluating the quality of substation robot inspection data according to claim 2, characterized in that, The third preset decision tree is: When the inspection data is numerical data and meets the first repeatability evaluation rule, the corresponding third parameter attribute is data repetition; the first repeatability evaluation rule is that the difference between the inspection data and the previous inspection data is zero. When the inspection data is numerical data and does not meet the first repeatability evaluation rule, the corresponding third parameter attribute is that the data is not repeated. When the inspection data is text data and meets the second repeatability evaluation rule, the corresponding third parameter attribute is data repetition; the second repeatability evaluation rule is that the text similarity between the inspection data and the previous inspection data is greater than a preset similarity threshold. When the inspection data is text-based and does not meet the second repeatability evaluation rule, the corresponding third parameter attribute is that the data is not repeated.

6. The method for evaluating the quality of substation robot inspection data according to claim 5, characterized in that, The steps for calculating text similarity include: For the inspection data and the previous inspection data, word segmentation and stop word removal are performed sequentially to obtain the set of effective feature words for inspection and the set of effective feature words for the previous inspection. The set of valid feature words for inspection and the set of valid feature words for the previous inspection are merged and deduplicated to obtain the set of inspection words. Calculate the word frequency of each word in the inspection word set based on the inspection data and the previous inspection data to obtain the inspection word frequency vector and the previous inspection word frequency vector; After normalizing the inspected word frequency vector and the previous inspected word frequency vector using L2-norm, the text similarity is calculated using cosine similarity.

7. A substation robot inspection data quality evaluation system, characterized in that, The system, applied to inspection robots, includes: The data acquisition module is used to acquire the inspection dataset obtained by the inspection robot in a single inspection of a substation. The first evaluation module is used to determine the corresponding first parameter attribute for any inspection data in the inspection dataset based on a first preset decision tree; the first parameter attribute is whether the data is timely or untimely. The second evaluation module is used to determine the corresponding second parameter attribute based on the second preset decision tree; the second parameter attribute is whether the data is complete or incomplete. The third evaluation module is used to determine the corresponding third parameter attribute based on the third preset decision tree; the third parameter attribute is whether the data is duplicated or not. The fourth evaluation module is used to determine the corresponding fourth parameter attribute based on the fourth preset decision tree; the fourth parameter attribute is whether the data is accurate or inaccurate. The fifth evaluation module is used to determine the corresponding data quality score based on the fifth preset decision tree and according to the first parameter attribute, the second parameter attribute, the third parameter attribute and the fourth parameter attribute; The fifth preset decision tree is as follows: when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is inaccurate data, the first value is marked as the corresponding data quality score; when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is non-duplicated data, and the fourth parameter attribute is inaccurate data, the second value is marked as the corresponding data quality score; when the first parameter attribute is timely data, the second parameter attribute is complete data, the third parameter attribute is duplicate data, and the fourth parameter attribute is accurate data, the fourth value is marked as the corresponding data quality score; when the first parameter attribute is untimely data, the value 0 is marked as the corresponding data quality score. The first value, the second value, the third value, the fourth value, and the fifth value are all determined based on the timeliness of the inspection data. The comprehensive evaluation module is used to determine the comprehensive data quality score corresponding to the inspection dataset based on the data quality score corresponding to each inspection data in the inspection dataset, and to send the inspection dataset to the substation monitoring backend when the comprehensive data quality score meets the preset quality requirements.

8. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory is used to store computer programs, and the processor runs the computer programs to enable the electronic device to perform the substation robot inspection data quality evaluation method according to any one of claims 1-6.

Images