Power dispatching system and method based on voice recognition

Abstract

The invention relates to the technical field of power dispatching, in particular to a power dispatching system and method based on voice recognition. The system comprises a voice transfer device used for acquiring scheduling text data based on scheduling voice data; the work ticket reading device is used for acquiring instruction text data based on the work ticket data; and the matching device is used for matching the scheduling text data and the instruction text data and outputting a matching result. The method is realized based on the system. The misoperation possibility of regulation and control services can be remarkably reduced, and the safe operation level of a power grid is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and more specifically, to a power dispatching system and method based on speech recognition. Background Art

[0002] In the field of power, with the maturity of intelligent speech technology, the demand for and application scenarios of intelligent speech have gradually increased. In terms of basic theoretical research, for example, Zhang Xi, Huang Yusong, etc. have carried out research on the application of artificial intelligence technology to power dispatching automation systems. The research results show that AI technology can complete power dispatching stably and scientifically; Zhao Qichun compared the power system fault diagnosis based on AI technology with the conventional circuit system diagnosis method and found that AI technology can accurately locate the circuit fault location and give the optimal diagnosis plan; Sun Qiuye, etc. analyzed the application status of AI technology in power systems and found that artificial intelligence technology is a necessary means to solve complex system decision-making and control; Liang Chun, etc. designed a neural network model based on AI technology for the complex problem of ship power system fault diagnosis to achieve fast diagnosis and gave the fault diagnosis process of the inference engine; Tao Yubo, etc. proposed a model and method for short-term wind power prediction using artificial intelligence technology.

[0003] In the current dispatching order issuing link, it still mainly relies on the operation personnel to communicate through dialogue. The order issuing personnel will issue the dispatching order by voice based on the dispatching order content recorded in the work ticket; due to the lack of supervision of voice order technology and the lack of mutual verification between voice orders and dispatching orders, there is a risk of incorrect order issuing without personnel supervision. Summary of the Invention

[0004] The present invention provides a power dispatching system and method based on speech recognition, which can overcome certain or some defects of the prior art.

[0005] According to a power dispatching system and method based on speech recognition of the present invention, it includes: A speech transcription device, which is used to obtain dispatching text data based on dispatching speech data; A work ticket reading device, which is used to obtain instruction text data based on work ticket data; and A matching device, which is used to match the dispatching text data and the instruction text data and output a matching result.

[0006] Preferably, the speech transcription device includes, A speech acquisition unit, which is used to acquire dispatching speech data; and A speech transcription unit, which is used to transcribe the dispatching speech data to obtain dispatching text data.

[0007] Preferably, the work ticket reading device includes: a text acquisition unit for acquiring work ticket data; and a text recognition unit for recognizing the work ticket data to obtain instruction text data.

[0008] Preferably, the matching device includes: a word segmentation unit for segmenting the dispatching text data to obtain a first word sequence and for segmenting the instruction text data to obtain a second word sequence; a comparison unit for comparing the first word sequence and the second word sequence to obtain a matching result; and an output unit for outputting the matching result.

[0009] Preferably, the matching device further includes a first vocabulary database and a second vocabulary database. The first vocabulary database is used to store professional vocabulary related to electricity, and the second vocabulary database is used to store natural language vocabulary related to general natural language; The word segmentation unit obtains the first word sequence and the second word sequence based on the first vocabulary database and the second vocabulary database.

[0010] Preferably, the comparison unit performs separate comparisons on the words in the first word sequence and the second word sequence that belong to the first vocabulary database and the second vocabulary database, and obtains the matching result based on the fusion of the results of the separate comparisons.

[0011] In addition, the present invention also provides a power dispatching method based on speech recognition, which includes: processing the dispatching speech data based on a speech transcription device to obtain dispatching text data; processing the work ticket data based on the work ticket reading device to obtain instruction text data; matching the dispatching text data and the instruction text data based on the matching device and outputting a matching result.

[0012] Preferably, the matching of the dispatching text data and the instruction text data based on the matching device and outputting a matching result includes: segmenting the dispatching text data based on the word segmentation unit to obtain a first word sequence, and segmenting the instruction text data to obtain a second word sequence; comparing the first word sequence and the second word sequence based on the comparison unit to obtain a matching result; outputting the matching result based on the output unit.

[0013] Preferably, the segmenting of the dispatching text data based on the word segmentation unit to obtain a first word sequence and the segmenting of the instruction text data to obtain a second word sequence includes: Based on the first vocabulary database and the second vocabulary database, obtain the first word sequence and the second word sequence.

[0014] Preferably, the comparison unit compares the first word sequence and the second word sequence to obtain a matching result, including: Separate and individual comparisons are made based on the words in the first word sequence and the second word sequence that belong to the first vocabulary database and the second vocabulary database, and a matching result is obtained based on the fusion of the results of the individual comparisons.

[0015] The present invention has the following beneficial effects: Based on a power dispatching system of the present disclosure, on the basis of the traditional telephone voice repeat of safety guardianship, pre-issuance and errata correction of manual operation tickets, and operation guardianship system, voice recognition verification is added, the dispatching order is real-time input into the system, and after transcription, it is synchronously text-verified with the operation ticket. At the same time, voice comparison verification is performed on the order-giving and repeat-reading voices to ensure that the dispatching instruction issuance and reception links are accurate and error-free; it can significantly reduce the possibility of misoperations in dispatching operations and improve the safe operation level of the power grid. At the same time, in the case of continuous increase in the scale of the power grid, continuous complexity of the grid structure, and high-speed growth of the load, it can also reduce the risk of misordering, thereby further reducing the power outage time in the event of an accident and narrowing the power outage scope of electricity users, and improving the power consumption comfort of social users; Based on voice recognition and semantic understanding technologies, in the scenario of dispatching operation instruction tickets, it can perform high-precision voice transcription on the real-time voice instructions of the on-duty dispatcher, combine text consistency comparison technology to judge whether there is a discrepancy between the voice instructions and the original ticket, and at the same time, combine semantic understanding technology and dispatching operation ticket rules and specifications to perform real-time supervision and guardianship warning on the correctness of the voice instructions. Brief Description of the Drawings

[0016] Figure 1 is a schematic diagram of a power dispatching system of the present disclosure; Figure 2 is a schematic diagram of a voice transcription device in a power dispatching system of the present disclosure; Figure 3 is a schematic diagram of a work ticket reading device in a power dispatching system of the present disclosure; Figure 4 is a schematic diagram of a matching device in a power dispatching system of the present disclosure. Detailed Embodiments

[0017] To further understand the content of the present invention, the present invention will be described in detail in combination with embodiments. It should be understood that the embodiments are only for explaining the present invention and not for limiting it.

[0018] As seen in Figure 1 , this embodiment provides a power dispatching system based on voice recognition, which includes: A speech transcription device for obtaining scheduling text data based on scheduled voice data; A work ticket reading device for obtaining instruction text data based on work ticket data; and A matching device for matching the scheduling text data and the instruction text data and outputting a matching result.

[0019] A power dispatching system based on the present disclosure can, on the basis of traditional safety supervision such as telephone voice repeat, pre-issuance and errata correction of manual operation tickets, and operation supervision system, add voice recognition verification, input scheduling commands into the system in real time, perform synchronous text verification with the operation ticket after transcription, and at the same time perform voice comparison verification on the order and the repeated voice to ensure that the dispatching instruction issuance and reception links are accurate and error-free. It can significantly reduce the possibility of misoperation in dispatching operations and improve the safe operation level of the power grid. At the same time, in the case of continuous increase in the scale of the power grid, continuous complexity of the grid structure, and high-speed growth of the load, it can also reduce the risk of misissuing orders, thereby further reducing the power outage time in the event of an accident and narrowing the power outage scope of electricity users, and improving the power consumption comfort of social users.

[0020] As seen in Figure 2 , the speech transcription device of the present disclosure can include, for example, A voice collection unit for collecting and obtaining scheduled voice data; and A voice transcription unit for transcribing the scheduled voice data to obtain scheduling text data.

[0021] Based on the above, the acquisition of scheduling text data can be preferably realized.

[0022] In the present disclosure, the voice collection unit can obtain data from the real-time voice stream of the current dispatching telephone system. In specific implementation, it can adopt the following methods: First, obtain the real-time voice stream in a mirroring manner from the dispatching telephone switch system; Second, connect to the dispatching telephone system in the way of an external recording service. The dispatching telephone system can split a path of the real-time voice stream to the recording service, and the recording service can collect and obtain the scheduled voice data.

[0023] In the present disclosure, the voice transcription unit can be implemented based on the current voice transcription model with a framework such as a convolutional neural network (CNN) or a recurrent neural network (RNN).

[0024] As seen in Figure 3 , the work ticket reading device in the present disclosure can include, for example, A text collection unit for collecting work ticket data; and A text recognition unit for recognizing the work ticket data to obtain instruction text data.

[0025] Based on the above, the acquisition of instruction text data can be preferably achieved.

[0026] In the present disclosure, the text acquisition unit can be implemented based on an image acquisition device, and the corresponding text recognition unit can be implemented based on an image processing algorithm model.

[0027] As seen in Figure 4 , the matching device in the present disclosure can include, for example, a word segmentation unit, which is used to segment the scheduling text data to obtain a first word sequence, and is also used to segment the instruction text data to obtain a second word sequence; a comparison unit, which is used to compare the first word sequence and the second word sequence to obtain a matching result; and an output unit, which is used to output the matching result.

[0028] Based on the above, the acquisition of the matching result can be achieved based on the comparison of the first word sequence and the second word sequence. This method can take into account the particularity in the power dispatching process. Specifically, in the power dispatching process, there will be a large number of words related to the power industry and words with specific meanings related to the power network that is currently the dispatching object; based on this method, the text data can be converted into word sequences for comparison and matching, so as to better adapt to the power dispatching field and improve the matching accuracy.

[0029] In the present disclosure, the matching device further includes a first vocabulary database and a second vocabulary database. The first vocabulary database is used to store professional vocabulary related to electricity, and the second vocabulary database is used to store natural language vocabulary related to general natural language; The word segmentation unit obtains the first word sequence and the second word sequence based on the first vocabulary database and the second vocabulary database.

[0030] Based on the above, the comparison and matching considering the professionalism of the power dispatching industry can be preferably achieved, and the accuracy of word segmentation can be improved.

[0031] In the present disclosure, the comparison unit separately compares the words in the first word sequence and the second word sequence that belong to the first vocabulary database and the second vocabulary database, and obtains the matching result based on the fusion of the results of the separate comparisons.

[0032] Based on the above, the further improvement of the matching result can be achieved.

[0033] In addition, based on the above power dispatching system, another aspect of the present disclosure is to provide a power dispatching method based on speech recognition, which includes: Processing the scheduling speech data based on a speech transcription device to obtain scheduling text data; Process the work ticket data based on the work ticket reading device to obtain the instruction text data; Match the dispatching text data and the instruction text data based on the matching device, and output the matching result.

[0034] Among them, the matching of the dispatching text data and the instruction text data based on the matching device to output the matching result can include: Segment the dispatching text data based on the word segmentation unit to obtain the first word sequence, and segment the instruction text data to obtain the second word sequence; Compare the first word sequence and the second word sequence based on the comparison unit to obtain the matching result; Output the matching result based on the output unit.

[0035] Among them, the segmentation of the dispatching text data based on the word segmentation unit to obtain the first word sequence, and the segmentation of the instruction text data to obtain the second word sequence can include: Obtain the first word sequence and the second word sequence based on the first vocabulary database and the second vocabulary database.

[0036] Among them, the comparison of the first word sequence and the second word sequence based on the comparison unit to obtain the matching result can include: Separate and compare the words belonging to the first vocabulary database and the second vocabulary database in the first word sequence and the second word sequence respectively, and obtain the matching result based on the fusion of the results of the separate comparisons.

[0037] As a specific example of the present disclosure: After obtaining the dispatching text data A and the instruction text data B; the word segmentation unit can first perform word segmentation processing on the dispatching text data A and the instruction text data B, and then obtain the first word sequence and the second word sequence ; Taking the first word sequence as an example, the following specific description is given. The word segmentation unit first matches the word order of the dispatching text data A with the first vocabulary database to obtain the vocabulary sequence in the dispatching text data A that belongs to the first vocabulary database ; After that, the word segmentation unit uses the vocabulary sequence as the segmentation point, and divides the dispatching text data A into multiple text segments based on the word order of the dispatching text data A, and any text segment does not contain the vocabulary in the vocabulary sequence ; After that, the word segmentation unit sequentially matches the multiple text segments with the second vocabulary database to obtain the vocabulary sequence in the dispatching text data A that belongs to the second vocabulary database ; Finally, according to the word order of the scheduling text data A, the word segmentation unit sequentially sorts the words in the vocabulary sequence and the vocabulary sequence ; The second word sequence is obtained in the same way as the first word sequence , which will not be elaborated in this article; After obtaining the first word sequence , the vocabulary sequence corresponding to the first word sequence and the vocabulary sequence , the second word sequence , and the vocabulary sequence corresponding to the second word sequence (the obtaining method is the same as that of the vocabulary sequence ) and the vocabulary sequence (the obtaining method is the same as that of the vocabulary sequence ), the comparison unit can perform semantic comparison and matching based on the following Based on the word order of the vocabulary sequence , reconstruct the word order of the vocabulary sequence ; specifically, obtain the text similarity between each word in the vocabulary sequence and each word in the vocabulary sequence . Take the word with the largest text similarity and reaching the first similarity threshold (for example, 75%) at the corresponding word order position of the word in the vocabulary sequence corresponding to the vocabulary sequence as the word at the corresponding word order position of the reconstructed vocabulary sequence . The matching result includes the words that are more or missing in the vocabulary sequence compared with the vocabulary sequence , and the words whose similarity with the corresponding words does not reach the second similarity threshold (for example, 95%); For example, if the vocabulary sequence is {a, b, c} and the vocabulary sequence is {e, f, g}, the similarities between words a, b, c, d and words e, f, g, h are e f g a 60% 90% 20% b 70% 80% 100% c 50% 15% 20% Then the reconstructed vocabulary sequence is {null value, a, b}. At this time, the matching result includes the extra word a, the missing word e, and the word f that does not reach the second similarity threshold; After that, obtain from the vocabulary sequence the words located in the reconstructed vocabulary sequence the vocabulary set between every two adjacent words, obtained from the word sequence in the word sequence the vocabulary set between every two adjacent words, and the vocabulary set from the word sequence is used to obtain the text similarity with the vocabulary set from the word sequence in a one-to-one correspondence in order. The matching result also includes the vocabulary set from the word sequence where the similarity does not reach the third similarity threshold (e.g., 80%); After obtaining all the matching results, the output unit outputs all the output results as text, and when the matching result is not null, an instruction for triggering a warning action is generated.

[0038] Through the above, based on speech recognition and semantic understanding technologies, in the scenario of dispatching operation instruction tickets, high-precision speech transcription of the real-time voice instructions of the on-duty dispatcher can be performed. Combining with the text consistency comparison technology, it can be judged whether there is a discrepancy between the voice instructions and the original ticket. At the same time, combining semantic understanding technology with dispatching operation ticket rules and specifications, the correctness of the voice instructions can be supervised and monitored in real time and early warnings can be given.

[0039] It is easy to understand that those skilled in the art can combine, split, reorganize, etc. the embodiments of the present application based on one or several embodiments provided by the present application to obtain other embodiments, and these embodiments do not exceed the protection scope of the present application.

[0040] The above schematically describes the present invention and its implementation manners. This description is not restrictive, and what is shown in the embodiments is only part of the implementation manners of the present invention. The actual structure is not limited thereto. Therefore, if those of ordinary skill in the art are inspired by it and design similar structural manners and embodiments without creative efforts without departing from the purpose of the present invention's creation, they should all fall within the protection scope of the present invention.

Claims

1. A power dispatching system based on speech recognition, comprising: A speech transcription device for obtaining dispatching text data based on dispatching speech data; A work ticket reading device for obtaining instruction text data based on work ticket data; and A matching device for matching the dispatching text data and the instruction text data and outputting a matching result.

2. The power dispatching system based on speech recognition according to claim 1, characterized in that: The speech transcription device includes A speech acquisition unit for acquiring dispatching speech data; and A speech transcription unit for transcribing the dispatching speech data to obtain dispatching text data.

3. The power dispatching system based on speech recognition according to claim 1, wherein: The work ticket reading device includes A text acquisition unit for acquiring work ticket data; and A text recognition unit for recognizing the work ticket data to obtain instruction text data.

4. The power dispatching system based on speech recognition according to claim 1, wherein: The matching device includes A word segmentation unit for segmenting the dispatching text data to obtain a first word sequence, and for segmenting the instruction text data to obtain a second word sequence; A comparison unit for comparing the first word sequence and the second word sequence to obtain a matching result; and An output unit for outputting the matching result.

5. The power dispatching system based on speech recognition according to claim 4, characterized in that: The matching device further includes a first vocabulary database and a second vocabulary database. The first vocabulary database is used to store professional vocabulary related to electricity, and the second vocabulary database is used to store natural language vocabulary related to general natural language; The word segmentation unit obtains the first word sequence and the second word sequence based on the first vocabulary database and the second vocabulary database.

6. The power dispatching system based on speech recognition according to claim 5, wherein: The comparison unit separately compares the words in the first word sequence and the second word sequence that belong to the first vocabulary database and the second vocabulary database, and obtains a matching result based on the fusion of the results of the separate comparisons.

7. A power dispatching method based on speech recognition, comprising: Processing the dispatching speech data based on the speech transcription device to obtain dispatching text data; Processing the work ticket data based on the work ticket reading device to obtain instruction text data; Matching the dispatching text data and the instruction text data based on the matching device and outputting a matching result.

8. The method according to claim 7, wherein The step of matching the dispatching text data and the instruction text data based on the matching device and outputting a matching result includes: Segmenting the dispatching text data based on the word segmentation unit to obtain a first word sequence, and segmenting the instruction text data to obtain a second word sequence; Comparing the first word sequence and the second word sequence based on the comparison unit to obtain a matching result; Outputting the matching result based on the output unit.

9. The method according to claim 8, wherein The step of segmenting the dispatching text data based on the word segmentation unit to obtain a first word sequence, and segmenting the instruction text data to obtain a second word sequence includes: Obtaining the first word sequence and the second word sequence based on the first vocabulary database and the second vocabulary database.

10. The method according to claim 8, wherein The step of comparing the first word sequence and the second word sequence based on the comparison unit to obtain a matching result includes: Separately comparing the words in the first word sequence and the second word sequence that belong to the first vocabulary database and the second vocabulary database, and obtaining a matching result based on the fusion of the results of the separate comparisons.

Images