A method and system for intelligent interaction of power grid control business based on speech recognition technology

Abstract

The present invention belongs to the technical field of power grid control, and specifically relates to a method and system for intelligent interaction of power grid control business based on voice recognition technology. The system for intelligent interaction of power grid control business based on voice recognition technology comprises a voice recognition module: used for collecting voice commands issued by power grid control operation personnel, and converting them into recognizable entries, and at the same time, notifying voice information to the power grid control operation personnel; a vocabulary module: used for storing power grid control commands, and converting the power grid control commands into classification entries, and at the same time selecting keywords from the classification entries, and displaying the classification entries corresponding to the power grid control commands through the keywords. In the case where the voice command is still being interacted, the required classification entry has already appeared. At this time, the classification entry can be directly selected, thereby reducing the time of voice command interaction, and further improving the efficiency of power grid control business.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to a method and system for intelligent interaction of power grid control services based on speech recognition technology. Background Art

[0002] With the development of economy, people's demand for electricity is getting higher and higher. In order to meet the power supply demand, the complexity of the power grid structure is also increasing, and the task burden of control centers at all levels is also increasing. In order to ensure the real-time and safety of power grid control operations, power grid control operators need to quickly obtain and process more information in a short time. Intelligent voice interaction technology can help people quickly obtain and process information. In order to reduce the burden on power grid control operators, intelligent voice interaction technology has now been applied to power grid control business.

[0003] Intelligent voice interaction technology requires voice recognition, but voice recognition has a continuity, that is, the power grid control and operation personnel cannot interrupt the voice command during human-computer communication. However, when performing power grid control and operation, there will be a sequence according to the severity of the event. When an emergency occurs, it is necessary to suspend the ongoing voice command and control the power grid corresponding to the emergency. However, since the voice command cannot be paused, it is necessary to abandon the interrupted voice command when handling an emergency. When the emergency is handled, the interrupted voice command needs to be re-entered, which is not conducive to the rapid operation of the power grid control business, but also increases the workload of the power grid control and operation personnel. Summary of the invention

[0004] The purpose of the present invention is to provide a method and system for intelligent interaction of power grid control services based on voice recognition technology, so as to solve the problem that voice commands in existing power grid control services based on voice recognition technology may be interrupted as mentioned in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for intelligent interaction of power grid control business based on speech recognition technology, the method for intelligent interaction of power grid control business based on speech recognition technology is as follows:

[0006] Step 1: construct a power grid control command vocabulary, classify the entries in the power grid control command vocabulary, obtain the classified entries, and then select the keywords of the classified entries in the power grid control command vocabulary;

[0007] Step 2: Collect and recognize voice commands, then convert the voice commands into recognizable entries, and then compare the recognizable entries with keywords. When the recognizable entries contain keywords, call up the classification entries corresponding to the keywords and display them;

[0008] Step 3: When the voice command of the emergency event needs to overwrite the voice command of the common event, the recognizable entries corresponding to the voice command of the common event are retained and the current page is saved, and then a new page is switched for the voice command of the emergency event to overwrite;

[0009] Step 4: After the voice command interaction for emergency events is completed, the page corresponding to the voice command for common events is switched back, and the keywords and recognizable entries corresponding to the voice command for common events are displayed. The power grid control and operation personnel continue the voice interaction based on the displayed keywords.

[0010] Step 5: After the voice interaction is completed, the classification entries and recognizable entries corresponding to the voice command are displayed. The grid control and operation personnel check the displayed classification entries and recognizable entries. When the classification entry is correct, the classification entry is confirmed; when the recognizable entry is correct, the recognizable entry is confirmed.

[0011] Preferably, the classification terms are numbered when they are displayed, and when the displayed classification terms include the required classification terms, the power grid control operator can select the classification term by selecting the number.

[0012] Preferably, the classification entries are counted, and each time a classification entry is confirmed by a power grid control operator, the count of the classification entry is increased by one. When multiple classification entries are displayed simultaneously, they are arranged in sequence from top to bottom according to the size of the counting results.

[0013] Preferably, the power grid control operator confirms that before confirming the classification terms, when it is found that the classification terms contain unnecessary keywords, the unnecessary keywords can be deleted and the keywords can be re-entered by voice.

[0014] A system for intelligent interaction of power grid control business based on speech recognition technology, comprising a speech recognition module: used to collect speech commands issued by power grid control operators and convert them into recognizable terms, and also to inform the power grid control operators of speech information;

[0015] Thesaurus module: used to store power grid control commands and convert them into classification terms, select keywords from the classification terms, and then associate the keywords with the classification terms;

[0016] Comparison module: used to compare the recognizable terms with the keywords and confirm whether the recognizable terms contain the keywords. When the recognizable terms contain the keywords, the comparison module will circle the classification terms corresponding to the included keywords.

[0017] Region coherence module: used to ensure the continuity of voice commands. The region coherence module contains multiple separate regions. One voice command corresponds to one region. It is also responsible for associating the regions corresponding to a complete voice command.

[0018] Display module: used to select the classified terms circled by the comparison module from the vocabulary module and display them, and can also display the recognizable terms and keywords;

[0019] Verification module: The power grid control personnel select the required classification terms from the displayed classification terms and confirm that they are command terms, or select recognizable terms and confirm that they are command terms.

[0020] Preferably, the verification module further includes a counting module, which is responsible for counting the number of times the classification terms are confirmed, and the display module arranges the classification terms up and down according to the number of times the classification terms are confirmed when displaying the classification terms.

[0021] Preferably, the display module also includes a modification module, which is responsible for deleting and modifying the displayed keywords and can also modify the recognizable entries.

[0022] Preferably, the vocabulary module also includes an update module, which is responsible for updating the vocabulary module. When the power grid control personnel select a recognizable entry and confirm it as a command entry, the recognizable entry is compared with the displayed classification entry. When the overlap rate between the two is 80%, the recognizable entry is added to the vocabulary module as a new classification entry.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1) The present invention uses page overlay so that the voice commands of common events do not need to be abandoned, and the interaction of the voice commands of emergency events can also be carried out. When the interaction of the voice commands of emergency events is completed, the voice commands of common events are re-extracted, and then the interaction of the voice commands of common events continues. Since the voice commands of common events are not abandoned, the interaction time of voice commands can be saved, thereby improving the efficiency of power grid control business and reducing the workload of power grid control operators;

[0025] 2) The classification entries corresponding to the power grid control commands are displayed through keywords. When the voice command is still being interacted, the required classification entry has already appeared. At this time, the classification entry can be directly selected, thereby reducing the time of voice command interaction and improving the efficiency of power grid control services. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1It is a schematic diagram of the module flow of the present invention;

[0027] Figure 2 is the normalization formula of the pooling layer of the present invention;

[0028] Figure 3 A mapping formula for the linear combination of the fully connected layers of the present invention;

[0029] Figure 4 The experimental analysis formula is collected for data collection of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] In the description of the present invention, it is necessary to understand that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0032] Example:

[0033] See also Figure 1-4 The present invention provides a technical solution: an intelligent interaction method for power grid control business based on speech recognition technology, the intelligent interaction method for power grid control business based on speech recognition technology is as follows:

[0034] Step 1: construct a power grid control command word library, which stores existing power grid control commands, and classifies the entries in the power grid control command word library. Different power grid control commands have different functions. By classifying them according to their functions, classified entries with different functions can be obtained, and then keywords of the classified entries in the power grid control command word library are selected. The number of keywords is determined according to the length of the classified entries. One classified entry contains at least two keywords;

[0035] Step 2: Collect and recognize voice commands, and then convert the voice commands into recognizable entries. Since different regions have different dialects, different conversion modules are designed according to different regions. Then the recognizable entries are compared with keywords. When the recognizable entries contain keywords, the classification entries corresponding to the keywords are called out and displayed. A page turning key is set on the display page to turn the displayed pages. The more keywords, the fewer classification entries are displayed.

[0036] When recognizing voice commands, an intention recognition algorithm based on convolutional neural network and compressed delay network is used to convert voice commands into recognizable terms. Then, the waveform features of the recognizable terms voice commands and the text content of the recognizable terms are used as the input of the convolutional neural network. At the same time, a delay unit is introduced to reduce the time cost of the algorithm. Positioning and recognition interaction are performed through voice key points. The convolutional neural network includes input layer, convolution layer, pooling layer, fully connected layer and input layer. The normalization formula of the pooling layer is as follows: Figure 2 , where x1 is the input feature of the convolution layer, σ B and are the mean error and variance set by the BN layer, γ and β are two parameters that can be automatically learned by the network, f is the activation parameter of the BN layer, and finally x4 is used as the input of the BN layer and enters the next layer;

[0037] The fully connected layer linearly combines the inputs of all units in the previous layer and performs the mapping formula. Figure 3 , where w is the weight vector of the fully connected layer, and b is the bias constant with the fully connected layer;

[0038] The experimental result analysis formula for data collection is as follows: Figure 4 , where F1 is used as a metric, P as precision, R as call rate, FPR as false alarm rate, FNR as missed alarm rate, and the recognition accuracy of speech key points is measured by the false alarm rate and missed alarm rate of key points. The false alarm rate is the ratio of the number of true key points detected to the actual number of key points, and the missed alarm rate is the ratio of the number of true key points not detected to the number of detected key points;

[0039] Step 3: When the voice command of the emergency event needs to overwrite the voice command of the common event, here it means that when the voice command of the common event has not been input, the recognizable entry corresponding to the voice command of the common event is retained and the current page is saved, here it means retaining the display page corresponding to the voice command of the common event, and then switching to a new page for the voice command of the emergency event to overwrite. The overwriting here is only an overlay on the display layer. When the page is removed, the display page corresponding to the voice command of the common event can be redisplayed;

[0040] Step 4: After the voice command interaction of the emergency event is completed, switch back to the page corresponding to the voice command of the common event, and display the keywords and recognizable entries corresponding to the voice command of the common event. The display of keywords and recognizable entries can facilitate the power grid control and operation personnel to clearly understand the position where the voice command of the common event is cut off. The power grid control and operation personnel continue to perform voice interaction according to the displayed keywords. At this time, when the power grid control and operation personnel broadcast the voice command again, after extracting the new keywords, they will be arranged behind the existing keywords and searched as keywords of the same voice command;

[0041] Step 5: After the voice interaction is completed, the classification entries and recognizable entries corresponding to the voice command are displayed. The power grid control and operation personnel check the displayed classification entries and recognizable entries to confirm whether the displayed classification entries and recognizable entries are correct. When the classification entry is correct, the classification entry is confirmed; when the recognizable entry is correct, the recognizable entry is confirmed.

[0042] When the classification entries are displayed, they are labeled. When the displayed classification entries include the required classification entries, the power grid control operator can select the classification entry by selecting the number, by voice, or by key selection. Since the classification entry is a complete power grid control command, the required entry will appear in the classification entry before the identifiable entries are fully displayed. At this time, the required entry can be quickly selected by the number, thereby quickly completing the power grid control.

[0043] The classification entries are counted. Each time a classification entry is confirmed by the power grid control operator, the count of the classification entry is increased by one. When multiple classification entries are displayed at the same time, they are arranged in order from top to bottom according to the size of the counting results. The more times a classification entry appears, the more times the power grid control command corresponding to the classification entry is used. Since the size of the display page is limited, the corresponding classification entry is displayed according to the number of times the power grid control command is used. This makes it easy to display the required classification entries in a limited display page.

[0044] The grid control and operation personnel confirm that before confirming the classification entries, when it is found that the classification entries contain unnecessary keywords, the unnecessary keywords can be deleted and the keywords can be re-entered by voice. Since the existing voice technology cannot accurately recognize all voices, there will be erroneous recognizable entries, and then the wrong keywords will be selected. Modifying and deleting the keywords can avoid the required classification entries from being eliminated.

[0045] A system for intelligent interaction of power grid control business based on speech recognition technology, including a speech recognition module: used to collect speech commands issued by power grid control operators and convert them into recognizable entries, and also to inform the power grid control operators of speech information, and adopt speech recognition modules according to different regions, so that the speech recognition modules are suitable for the dialects of the region;

[0046] Thesaurus module: used to store power grid control commands and convert power grid control commands into classification terms, or convert classification terms into power grid control commands. At the same time, keywords are selected from the classification terms, and then the keywords are associated with the classification terms. Classification terms associated with the keywords are found through the keyword interface.

[0047] Comparison module: used to compare the recognizable terms with the keywords and confirm whether the recognizable terms contain the keywords. When the recognizable terms contain the keywords, the comparison module will circle the classification terms corresponding to the included keywords.

[0048] Regional coherence module: used to ensure the continuity of voice commands. The regional coherence module contains multiple separate regions. One voice command corresponds to one region. When a voice command is interrupted during input, the input voice command occupies one region. When the interrupted voice command is input again, a new region is established. The two regions are associated with each other to form a complete voice command.

[0049] Display module: used to select the classified terms circled by the comparison module from the word library module and display them. It can also display recognizable terms and keywords. Through the display module, it is convenient for power grid control and operation personnel to understand the specific content of the voice command, and it is also convenient for power grid control and operation personnel to quickly find the required classified terms;

[0050] Verification module: The grid control personnel select the required classification terms from the displayed classification terms and confirm them as command terms, or select recognizable terms and confirm them as command terms. The displayed classification terms and command terms are not necessarily accurate, so the grid control personnel need to verify them.

[0051] The verification module also includes a counting module, which is responsible for counting the number of times the classification entries are confirmed. When displaying the classification entries, the display module arranges the classification entries up and down according to the number of times they are confirmed.

[0052] The display module also includes a modification module, which is responsible for deleting and modifying the displayed keywords, and can also modify the recognizable entries.

[0053] The vocabulary module also includes an update module, which is responsible for updating the vocabulary module. When the power grid control personnel select a recognizable entry and confirm it as a command entry, the recognizable entry is compared with the displayed classification entry. When the overlap rate between the two is 80%, the recognizable entry is added to the vocabulary module as a new classification entry.

[0054] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention; therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is limited by the attached claims rather than the above description. Therefore, it is intended to include all changes within the meaning and scope of the equivalent elements of the claims in the present invention, and any figure marks in the claims should not be regarded as limiting the claims involved.

[0055] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent interactive method for power grid control business based on speech recognition technology, characterized in that: The intelligent interaction method of power grid control business based on speech recognition technology is as follows: Step 1: construct a power grid control command vocabulary, classify the entries in the power grid control command vocabulary, obtain the classified entries, and then select the keywords of the classified entries in the power grid control command vocabulary; Step 2: Collect and recognize voice commands, then convert the voice commands into recognizable entries, and then compare the recognizable entries with keywords. When the recognizable entries contain keywords, call up the classification entries corresponding to the keywords and display them; Step 3: When the voice command of the emergency event needs to overwrite the voice command of the common event, the recognizable entries corresponding to the voice command of the common event are retained and the current page is saved, and then a new page is switched for the voice command of the emergency event to overwrite; Step 4: After the voice command interaction for emergency events is completed, the page corresponding to the voice command for common events is switched back, and the keywords and recognizable entries corresponding to the voice command for common events are displayed. The power grid control and operation personnel continue the voice interaction based on the displayed keywords. Step 5: After the voice interaction is completed, the classification terms and recognizable terms corresponding to the voice command are displayed, and the power grid control and operation personnel check the displayed classification terms and recognizable terms. When the classification terms are correct, the classification terms are confirmed, and when the recognizable terms are correct, the recognizable terms are confirmed; When the classification terms are displayed, the classification terms are numbered. When the displayed classification terms contain the required classification terms, the power grid control operator can select the classification terms by selecting the number; The classification terms are counted, and each time a classification term is confirmed by a power grid control operator, the count of the classification term is increased by one. When multiple classification terms are displayed simultaneously, they are arranged in order from top to bottom according to the size of the counting results.

2. According to claim 1, a method for intelligent interaction of power grid control services based on speech recognition technology is characterized by: The power grid control operator confirms that before confirming the classification terms, when it is found that the classification terms contain unnecessary keywords, the unnecessary keywords can be deleted and the keywords can be re-entered by voice.

3. A system for intelligent interaction of power grid control services based on speech recognition technology, characterized in that: It includes a voice recognition module: used to collect voice commands issued by power grid control and operation personnel, and convert the voice commands into recognizable terms, and also inform the power grid control and operation personnel of the voice information; Thesaurus module: used to store power grid control commands and convert them into classification terms, select keywords from the classification terms, and then associate the keywords with the classification terms; Comparison module: used to compare the recognizable terms with the keywords and confirm whether the recognizable terms contain the keywords. When the recognizable terms contain the keywords, the comparison module will circle the classification terms corresponding to the included keywords. Region coherence module: used to ensure the continuity of voice commands. The region coherence module contains multiple separate regions. One voice command corresponds to one region. It is also responsible for associating the regions corresponding to a complete voice command. Display module: used to select the classified terms circled by the comparison module from the vocabulary module and display the classified terms, while also displaying the recognizable terms and keywords; Verification module: The power grid control personnel selects the required classification terms from the displayed classification terms and confirms that the classification terms are command terms, or selects the recognizable terms and confirms that the recognizable terms are command terms; The verification module also includes a counting module, which is responsible for counting the number of times the classification terms are confirmed. When displaying the classification terms, the display module arranges the classification terms up and down according to the number of times they are confirmed.

4. The system for intelligent interaction of power grid control services based on speech recognition technology according to claim 3 is characterized by: The display module also includes a modification module, which is responsible for deleting and modifying the displayed keywords, and also modifies the identifiable entries.

5. The system for intelligent interaction of power grid control services based on speech recognition technology according to claim 3 is characterized by: The vocabulary module also includes an update module, which is responsible for updating the vocabulary module. When the power grid control personnel select a recognizable entry and confirm that the recognizable entry is a command entry, the recognizable entry is compared with the displayed classification entry. When the overlap rate between the two is 80%, the recognizable entry is added to the vocabulary module as a new classification entry.

Images