Centralized control system alarm signal matching method and system based on signal segmentation

Abstract

The invention belongs to the technical field of electric power station centralized control systems, and particularly relates to a signal segmentation-based centralized control system alarm signal matching method and system. The method comprises the following steps: extracting feature words similar to various feature words of any typical alarm signal from a real-time alarm signal, wherein the various feature words of the typical alarm signal are formed by segmenting the typical alarm signal; matching the typical alarm signal with the highest total similarity with the real-time alarm signal with the real-time alarm signal; wherein the mode of calculating the total similarity between a certain typical alarm signal and the real-time alarm signal is that the comprehensive similarity between any feature word of the typical alarm signal and feature words of the same kind of the real-time alarm signal is calculated, and the comprehensive similarity is obtained according to the pinyin similarity and the character string similarity; and obtaining the total similarity of the typical alarm signal and the real-time alarm signal according to the comprehensive similarity of all feature words of the typical alarm signal and the real-time alarm signal. And the alarm signal matching precision is high.

Description

Technical Field

[0001] The present invention belongs to the technical field of centralized control systems for electric power stations, and in particular relates to a method and system for matching alarm signals of a centralized control system based on signal segmentation. Background Art

[0002] The new generation of multi-format power station centralized control system is positioned to provide panoramic monitoring, remote operation and maintenance, production operation, auxiliary trading and other business functions for wind farms, photovoltaic power stations, energy storage power stations, booster stations and other stations. A centralized control system equipment monitoring system must be connected to dozens to hundreds of substations, and the number of devices and data monitored per person by the monitoring team is huge. It is difficult for the monitoring team members to identify the key signals representing the operating status of the equipment from the generated signals, or it is difficult to link these signals to related equipment failures. Once the equipment fails, it is likely to affect the equipment's operating efficiency, especially for power generation plants, which directly affects the power generation efficiency and causes losses to the owner.

[0003] In order to solve the above problems, the Chinese invention patent application with application publication number CN117591899A proposes a technical means to associate the alarm information with the equipment location by calculating the similarity between the alarm signal and the keywords in the equipment location. However, in this scheme, the cosine distance algorithm is used to calculate the text similarity when performing the similarity calculation. Only one similarity algorithm is used, which only considers the similarity between texts, and does not consider other information besides text, such as semantics, voice, etc. Therefore, using only the text similarity algorithm cannot fully and accurately reflect the similarity between the alarm signal and the setting point, which may cause association errors and affect the normal operation of the centralized control system. Summary of the invention

[0004] The purpose of the present invention is to provide a centralized control system alarm signal matching method and system based on signal segmentation, so as to solve the technical problem that when only text similarity is used to associate equipment and alarm signals in the prior art, the similarity calculation is not comprehensive and accurate enough, thus causing association errors and affecting the normal operation of the centralized control system.

[0005] In order to solve the above technical problems, the present invention provides a centralized control system alarm signal matching method based on signal segmentation, the method comprising:

[0006] Extracting characteristic words of the same type as various characteristic words of typical alarm signals from the real-time alarm signal, wherein various characteristic words of the typical alarm signal are obtained by segmenting the typical alarm signal;

[0007] The typical alarm signal with the highest total similarity to the real-time alarm signal is matched with the real-time alarm signal; wherein, the method of calculating the total similarity between a typical alarm signal and the real-time alarm signal is as follows: respectively calculating the comprehensive similarity between any feature word of the typical alarm signal and the same feature word of the real-time alarm signal, wherein the comprehensive similarity is obtained based on the pinyin similarity and the string similarity; and obtaining the total similarity between the typical alarm signal and the real-time alarm signal based on the comprehensive similarity of all feature words of the typical alarm signal and the real-time alarm signal.

[0008] Furthermore, the string similarity is: twice the number of characters in the intersection of the string of the characteristic word of the typical alarm signal and the string of the characteristic word of the corresponding real-time alarm signal divided by the sum of the number of characters in the two strings.

[0009] Furthermore, the method for obtaining the comprehensive similarity according to the pinyin similarity and the string similarity is: performing weighted summation of the pinyin similarity and the string similarity according to their respective corresponding weights to obtain the comprehensive similarity.

[0010] Furthermore, the string similarity weight and the pinyin similarity weight are adjusted according to their respective average accuracy rates, and the higher the average accuracy rate, the higher the similarity weight.

[0011] Furthermore, the calculation process of the pinyin similarity includes: obtaining similarity scores of two pinyins at the same position in the pinyin sequence of the typical alarm signal characteristic word and the corresponding real-time alarm signal characteristic word, summing up all similarity scores and then dividing by the maximum number of pinyins in the two pinyin sequences;

[0012] The similarity scores of two pinyins at the same position in two pinyin sequences are determined according to the initial consonants, final consonants and tones of the two pinyins.

[0013] Furthermore, the types of the feature words include equipment, signal and behavior.

[0014] Furthermore, the real-time alarm signal is a signal in the same format as a typical alarm signal.

[0015] Furthermore, the process of obtaining the total similarity between the typical alarm signal and the real-time alarm signal based on the comprehensive similarity of all feature words of the typical alarm signal and the real-time alarm signal includes: adding the comprehensive similarities of all feature words of the typical alarm signal and the real-time alarm signal to obtain the total similarity between the typical alarm signal and the real-time alarm signal.

[0016] Furthermore, the method of segmenting the typical alarm signal to form various characteristic words of the typical alarm signal is: using a hidden Markov model or a conditional random field model to segment the typical alarm signal to form various characteristic words.

[0017] Furthermore, the interpretation, warning and handling strategy of the typical alarm signal matching the real-time alarm signal is used as the interpretation, warning and handling suggestion of the real-time alarm signal.

[0018] The present invention is an improved invention creation, and its beneficial effects are as follows: the centralized control system alarm signal matching method based on signal segmentation of the present invention, when matching the typical alarm signal and the real-time alarm signal, uses the weighted sum of the string similarity and the pinyin similarity to obtain the comprehensive similarity to evaluate the matching degree between the two feature words. Compared with using a single similarity calculation method, the present invention obtains the comprehensive similarity from both the string and the voice when performing the similarity calculation, which is more comprehensive and improves the accuracy of the similarity calculation, thereby improving the accuracy of the total similarity calculation between the real-time alarm signal and the typical alarm signal, and can more accurately match the real-time alarm signal and the typical alarm signal, determine the equipment that issues the fault alarm, and ensure the normal operation of the centralized control system.

[0019] To solve the above technical problems, the present invention also provides a centralized control system alarm signal matching system based on signal segmentation, including a processor, which implements the steps described in the centralized control system alarm signal matching method based on signal segmentation of the present invention when executing a computer program.

[0020] The present invention is an improved invention creation, and its beneficial effects are as follows: the centralized control system alarm signal matching system based on signal segmentation of the present invention, when matching typical alarm signals and real-time alarm signals, uses the weighted sum of string similarity and pinyin similarity to obtain a comprehensive similarity to evaluate the matching degree between two feature words. Compared with using a single similarity calculation method, the present invention obtains a comprehensive similarity from both string and voice aspects when performing similarity calculation, which takes a more comprehensive approach and improves the accuracy of similarity calculation, thereby improving the accuracy of total similarity calculation between real-time alarm signals and typical alarm signals, and can more accurately match real-time alarm signals and typical alarm signals, determine the equipment that issues a fault alarm, and ensure the normal operation of the centralized control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a typical alarm signal processing schematic diagram of a method embodiment of the present invention;

[0022] Figure 2 It is a flow chart of a centralized control system alarm signal matching method based on signal segmentation according to a method embodiment of the present invention. DETAILED DESCRIPTION

[0023] The centralized control system alarm signal matching method and system based on signal segmentation of the present invention, when matching typical alarm signals with real-time alarm signals, uses the weighted sum of string similarity and pinyin similarity to obtain a comprehensive similarity to evaluate the degree of matching between two feature words. Compared with using a single similarity calculation method, the present invention obtains a comprehensive similarity from both string and voice aspects when performing similarity calculation, which takes a more comprehensive approach and improves the accuracy of similarity calculation, thereby improving the accuracy of the total similarity calculation between real-time alarm signals and typical alarm signals. It is able to more accurately match real-time alarm signals with typical alarm signals, determine the equipment that issues a fault alarm, and ensure the normal operation of the centralized control system.

[0024] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0025] Method Example:

[0026] The centralized control system alarm signal matching method based on signal segmentation of the present invention comprises the following steps:

[0027] Step 1: Segment typical alarm signals into different feature words to form a feature word library.

[0028] This embodiment uses a method based on a statistical language model, such as a hidden Markov model (HMM) or a conditional random field (CRF). The typical alarm signal is segmented into three parts: equipment, signal, and behavior, and a feature word library is formed based on the feature words of these three parts. The specific signal segmentation method is as follows:

[0029] sigMaxP(Dev|Sig|St,S)

[0030] Where: S represents the input typical alarm signal; Dev, Sig, and St represent the segmentation result sequences, that is, the signal is divided into three parts: device, signal, and behavior.

[0031] The meaning of this formula is to find the word segmentation scheme that maximizes P(Dev|Sig|St,S). This probability value reflects the maximum optimization value of the word segmentation scheme Dev|Sig|St when a typical alarm signal S is given. When the value is the largest, it is the optimal segmentation scheme.

[0032] Algorithms such as hidden Markov model and conditional random field are common Chinese word segmentation algorithms. When applied in this embodiment, when the algorithm is executed, some common names of equipment, signals, and plants, and fixed separation patterns of equipment, plants, and signals are taken as special rules of the Chinese word segmentation algorithm to obtain an effect that is closer to the new energy power system and design.

[0033] Step 2: Construct a relationship diagram between the interpretation, warning and disposal strategies of typical models.

[0034] Maintain the association between typical alarm signals and signal interpretations, and equipment abnormality warning and disposal. Based on the typical alarm signals and signal interpretations, and equipment abnormality disposal strategies, use association rule mining to find the association rules between signal patterns and warnings and disposals, and establish a relationship diagram between typical alarm signals and interpretations, warnings, and disposal strategies.

[0035] In this embodiment, a relationship diagram between the interpretation of typical models, warnings and disposal strategies is established, which facilitates the operation and maintenance personnel to analyze and process the alarm signal after the typical model and the real-time alarm signal are successfully matched.

[0036] Steps 1 and 2 are the processing for typical alarm signals. The processing relationship diagram is shown in the figure below: Figure 1 shown.

[0037] Step 3: Obtain the real-time alarm signal of the centralized control system and extract the effective signal from it.

[0038] Since different centralized control systems have their own specific equipment naming methods, after obtaining the real-time alarm signal, it is necessary to first de-specialize the real-time alarm signal, obtain the alarm signal in a universal format after processing, and retain the part that best matches the typical alarm signal as the valid signal. The specific method for retaining the valid signal is:

[0039] The real-time alarm signal after de-specialization is divided into different feature words. For example, the alarm signal information of the real-time monitoring system is usually XXX station XXX interval XXX equipment XXXX signal action (reset); then the above keywords are used to automatically generate regular matching rules, and the different feature words are segmented to obtain a valid signal composed of multiple segmented feature words, such as: [{% station} & {% interval %} & {% equipment %} & {% signal %} && {% action % | % reset %}], and this rule is used to identify valid signals. The feature words in the valid signal should include the feature words in the feature word library.

[0040] Step 4: Calculate the total similarity between the effective signal of the real-time alarm signal and all typical alarm signals respectively, and determine the matching signal of the real-time alarm signal.

[0041] The signal can be expressed in the form of a string, so calculating the similarity between the effective signal and the typical alarm signal is actually calculating the similarity between the two strings. The similarity of the present invention adopts the comprehensive similarity of the string similarity and the pinyin similarity. The specific calculation method includes:

[0042] 1. Calculate string similarity.

[0043] The string similarity can be calculated by cosine similarity, Jaccard similarity, Levenshtein similarity, etc. In this embodiment, the n-gram algorithm is used to calculate the string similarity. The specific calculation method is as follows:

[0044] S = 2*|A∩B| / (|A|+|B|)

[0045] Where: S is the string similarity, A and B represent two strings, which are the feature words in the effective alarm signal and the corresponding feature words in the typical alarm signal respectively; |A| and |B| represent the number of characters in string A and string B respectively; |A∩B| represents the number of characters in the intersection of string A and string B.

[0046] This formula reflects the ratio of the number of characters in the intersection of two strings A and B to the sum of the number of characters in the two strings A and B. Character similarity is between 0 and 1, and the larger the value, the more similar the two strings are. When the characters in the two strings A and B are exactly the same, the character similarity S is 1.

[0047] 2. Calculate the pinyin similarity.

[0048] A1 and B1 are the pinyin sequences corresponding to the character strings A and B respectively, A1 = [a1, a2, ..., an] and B1 = [b1, b2, ..., bm].

[0049] The calculation formula for pinyin similarity is as follows: P = Σ(sim(ai,bi)) / max(n,m).

[0050] Where, sim(ai,bi) represents the pinyin similarity between pinyin ai and bi. When the values ​​of n and m are not equal, for example, n<m, an+1 does not exist and is recorded as null, sim(null,bn+1)=0; pinyin similarity can be defined based on features such as initials, finals, and tones. The similarity matrix is ​​defined based on the pronunciation features of pinyin. Σ(sim(ai,bi)) represents the sum of the similarity scores of all pinyin pairs. max(n,m) represents the maximum length of the two pinyin sequences A1 and B1.

[0051] The specific calculation method of sim(ai,bi) is: sum the initial consonant similarity, final consonant similarity and tone similarity of ai and bi and divide by 3. If the initial consonants of ai and bi are the same, the initial consonant similarity of ai and bi is 1, otherwise it is 0; if the final consonants of ai and bi are the same, the final consonant similarity of ai and bi is 1, otherwise it is 0; for tone similarity, the similarity is 1 if the tones are the same, and there are 6 cases in total if the tones are different: the similarity between the first and second tones, the similarity between the first and third tones, the similarity between the first and fourth tones, the similarity between the second and third tones, the similarity between the second and fourth tones, and the similarity between the third and fourth tones. Each case corresponds to a value between 0 and 1, and the similarity values ​​between different cases can be the same or different. The higher the tone similarity, the greater the corresponding similarity value.

[0052] The similarity score is between 0 and 1, indicating the similarity between two pinyin sequences. The closer the score is to 1, the more similar they are.

[0053] 3. Calculate the comprehensive similarity.

[0054] The comprehensive similarity calculation method is: X=s*S+p*P.

[0055] Where X represents the comprehensive similarity; s represents the string similarity weight, S represents the string similarity; p represents the pinyin similarity weight, and P represents the pinyin similarity.

[0056] In this embodiment, the initial values ​​of the string similarity weight s and the pinyin similarity weight p are the same and both are 1. During the use of the system, the actual signals are used to calculate the average accuracy Zs and Zp of the string similarity and the pinyin similarity, respectively, to update the initial weight values. If the average accuracy Zs of the string similarity is greater than the average accuracy Zp of the pinyin similarity, the string similarity weight s is increased and the pinyin similarity weight p is correspondingly reduced. In the process of updating the string similarity weight s and the pinyin similarity weight p, the sum of the two should be kept unchanged and always equal to the sum of their initial weights.

[0057] 4. Calculate the total similarity between all feature words in the feature word library.

[0058] Total similarity: X 总 =ΣXi. Where X 总 It is the sum of the similarities between the real-time alarm signal and all feature word libraries of the typical alarm signal, and Xi is the weighted similarity between a certain feature word of the real-time alarm signal and the typical alarm signal.

[0059] 5. Among all typical alarm signals, the one with the highest total similarity to the real-time alarm signal is taken as the matching signal of the real-time alarm signal.

[0060] Step 5: Generate interpretation, warning and handling suggestions of the real-time alarm signal based on the interpretation, warning and handling strategy of the matching signal of the real-time alarm signal.

[0061] In this embodiment, all typical alarm signals have established a relationship diagram between interpretation, warning and handling strategy, and the faulty equipment, the content of the fault signal and the fault handling suggestions are determined based on the relationship diagram of the matching signal.

[0062] Steps 3 to 5 are the matching process between real-time alarm signals and typical alarm signals, such as Figure 2 shown.

[0063] System Example:

[0064] A centralized control system alarm signal matching system based on signal segmentation of the present invention includes a processor, and the processor implements the steps of the centralized control system alarm signal matching method based on signal segmentation as described in the method embodiment of the present invention when executing a computer program. The specific process, principle, etc. of the method have been described in detail in the method embodiment, and will not be repeated in this embodiment.

[0065] The processor may be a microprocessor MCU, a programmable logic device FPGA or other processing device.

[0066] The alarm signal matching method and system of the centralized control system based on signal segmentation of the present invention adopts pinyin similarity and string similarity when calculating the similarity between the real-time alarm signal and the typical alarm signal and matching them, respectively obtaining the similarity between the corresponding feature words from the two perspectives of voice and string, and obtaining a more comprehensive and accurate comprehensive similarity, and then obtaining the total similarity between the real-time alarm signal and the typical alarm signal based on the comprehensive similarity, which can improve the accuracy of the alarm signal matching and ensure the stable operation of the centralized control system. Furthermore, in this embodiment, the improved Jaccard similarity is used to calculate the string similarity, which improves the accuracy of the string similarity, thereby improving the accuracy of the comprehensive similarity and the total similarity, and then improving the accuracy of the final signal matching.

Claims

1. A method for matching alarm signals of a centralized control system based on signal segmentation, characterized in that: The method includes: Extracting characteristic words of the same type as various characteristic words of typical alarm signals from the real-time alarm signal, wherein various characteristic words of the typical alarm signal are obtained by segmenting the typical alarm signal; The typical alarm signal with the highest total similarity to the real-time alarm signal is matched with the real-time alarm signal; wherein, the method of calculating the total similarity between a typical alarm signal and the real-time alarm signal is as follows: respectively calculating the comprehensive similarity between any feature word of the typical alarm signal and the same feature word of the real-time alarm signal, wherein the comprehensive similarity is obtained based on the pinyin similarity and the string similarity; and obtaining the total similarity between the typical alarm signal and the real-time alarm signal based on the comprehensive similarity of all feature words of the typical alarm signal and the real-time alarm signal.

2. The method for matching alarm signals of a centralized control system based on signal segmentation according to claim 1, characterized in that: The string similarity is: twice the number of characters in the intersection of the string of the characteristic word of the typical alarm signal and the string of the characteristic word of the corresponding real-time alarm signal divided by the sum of the number of characters in the two strings.

3. The method for matching alarm signals of a centralized control system based on signal segmentation according to claim 1, characterized in that: The method for obtaining the comprehensive similarity based on the pinyin similarity and the string similarity is: weighted summing the pinyin similarity and the string similarity according to their respective corresponding weights to obtain the comprehensive similarity.

4. The method for matching alarm signals of a centralized control system based on signal segmentation according to claim 3 is characterized in that: The string similarity weight and pinyin similarity weight are adjusted according to their respective average accuracy rates. The higher the average accuracy rate, the higher the similarity weight.

5. The method for matching alarm signals of a centralized control system based on signal segmentation according to claim 1, characterized in that: The calculation process of the pinyin similarity includes: obtaining the similarity scores of two pinyins at the same position in the pinyin sequence of the typical alarm signal characteristic word and the corresponding real-time alarm signal characteristic word, summing up all the similarity scores and then dividing by the maximum number of pinyins in the two pinyin sequences; The similarity scores of two pinyins at the same position in two pinyin sequences are determined according to the initial consonants, final consonants and tones of the two pinyins.

6. The method for matching alarm signals of a centralized control system based on signal segmentation according to any one of claims 1 to 5, characterized in that: The types of feature words include equipment, signal and behavior.

7. The method for matching alarm signals of a centralized control system based on signal segmentation according to any one of claims 1 to 5, characterized in that: The real-time alarm signal is a signal in the same format as a typical alarm signal.

8. The method for matching alarm signals of a centralized control system based on signal segmentation according to any one of claims 1 to 5, characterized in that: The process of obtaining the total similarity between the typical alarm signal and the real-time alarm signal according to the comprehensive similarity of all feature words of the typical alarm signal and the real-time alarm signal includes: adding the comprehensive similarities of all feature words of the typical alarm signal and the real-time alarm signal to obtain the total similarity between the typical alarm signal and the real-time alarm signal.

9. The method for matching alarm signals of a centralized control system based on signal segmentation according to any one of claims 1 to 5, characterized in that: The method for forming various characteristic words of the typical alarm signal by segmenting the typical alarm signal is: using a hidden Markov model or a conditional random field model to segment the typical alarm signal into various characteristic words.

10. The method for matching alarm signals of a centralized control system based on signal segmentation according to any one of claims 1 to 5, characterized in that: The interpretation, warning and handling strategies of typical alarm signals that match the real-time alarm signals are used as the interpretation, warning and handling suggestions of the real-time alarm signals.

11. A centralized control system alarm signal matching system based on signal segmentation, comprising a processor, characterized in that: The processor implements the steps of the centralized control system alarm signal matching method based on signal segmentation as described in any one of claims 1 to 10 when executing the computer program.

Images