A substation voice control method and control system based on large model technology

Abstract

The present application relates to the field of speech recognition technology, and in particular to a substation voice control method and control system based on a large model technology, the method comprising obtaining a control voice instruction, determining a corresponding target dictionary group based on the control voice instruction, and converting the control voice corresponding to the control voice instruction into control text based on the target dictionary group; extracting keywords from the control text based on a preset natural language recognition large model to obtain key control text; performing logical conversion on the key control text to obtain control logic, and controlling the target control device based on the control logic. The present application facilitates improving the accuracy of recognizing voice instructions.

Description

Technical Field

[0001] The present application relates to the field of speech recognition technology, and in particular to a substation speech control method and control system based on large model technology. Background Art

[0002] Substations, as the core hub of the power grid system, not only play a key role in converting electricity from high voltage to low voltage, ensuring the safe delivery of electricity to thousands of households and all walks of life, but also bear the heavy responsibility of intelligent distribution, fine control and solid protection of the power system. Driven by the wave of information technology, artificial intelligence and big data, the power industry is accelerating towards a new era of intelligence. Among them, the innovative model of voice-controlled substations has emerged and led the innovation of substation operation methods. Compared with the traditional model that relied on heavy manual operations, the introduction of voice commands has greatly simplified the control process of internal equipment in substations, and at the same time, it is easy to reduce the probability of risks caused by human errors.

[0003] However, if voice commands are used to control the internal equipment of the substation, errors in the voice recognition process may cause the wrong operation command to be executed, which may lead to internal equipment failure, power outages, or even safety accidents. Therefore, in the process of using voice commands to control the internal equipment of the substation, the accuracy of voice command recognition is particularly critical. Summary of the invention

[0004] In order to improve the accuracy of voice command recognition and thus improve the precision of voice control operations, the present application provides a substation voice control method and control system based on large model technology.

[0005] In the first aspect, the present application provides a substation voice control method based on large model technology, which adopts the following technical solution:

[0006] A voice control method for a substation based on a large model technology, comprising:

[0007] Acquire a control voice instruction, determine a corresponding target dictionary group based on the control voice instruction, and convert the control voice corresponding to the control voice instruction into control text based on the target dictionary group;

[0008] Based on a preset natural language recognition large model, keyword extraction is performed on the control text to obtain key control text;

[0009] The key control text is logically transformed to obtain control logic, and the target control device is controlled based on the control logic.

[0010] By adopting the above technical solution, the corresponding target dictionary group is selected for voice conversion through the control voice command, rather than using a fixed dictionary group for voice conversion, which is convenient for improving the voice conversion rate while also improving the accuracy of the voice conversion results. In addition, the use of a preset natural language recognition large model service to extract keywords from the control text can reduce the redundancy of the control text, thereby facilitating the selection of more accurate and comprehensive results from the control text, and further facilitating the improvement of the accuracy of the subsequent determination of the control logic. By improving the accuracy of converting the control voice into the control text, and the accuracy of converting the control text into the control logic, the accuracy of voice recognition can be improved, thereby facilitating the improvement of the accuracy of language-based control of equipment in the substation.

[0011] In a possible implementation, the determining the corresponding target dictionary group based on the control voice instruction includes:

[0012] After detecting the control voice command, obtaining the pre-set parameters, wherein the pre-set parameters include the control time, the control environment and the device status;

[0013] Determine a first dictionary group from a historical dictionary library based on the control moment, and determine a first dictionary layer based on the first dictionary group and a preset data layer, wherein the first dictionary group includes a plurality of first data items and a data association relationship between each first data item;

[0014] Determine a second dictionary group from the historical dictionary library based on the control environment, and determine a second dictionary layer based on the second dictionary group and a preset data layer, wherein the second dictionary group includes a plurality of second data items and a data association relationship between each second data item;

[0015] Determine a third dictionary group from the historical dictionary library based on the device state, and determine a third dictionary layer based on the third dictionary group and a preset data layer, wherein the third dictionary group includes a plurality of third data items and a data association relationship between each third data item;

[0016] Based on the first dictionary layer, the second dictionary layer, and the third dictionary layer, a target dictionary group is determined.

[0017] By adopting the above technical scheme, the control voice instructions are analyzed to determine the pre-parameters related to the control voice instructions. Based on the pre-parameters, the applicable scenarios and applicable dictionaries of the control voice instructions can be predicted. Finally, the target dictionary group corresponding to the control voice instructions is jointly determined based on the dictionary layers corresponding to multiple pre-parameters, so as to improve the adaptability between the target dictionary group and the control voice instructions, thereby improving the accuracy of the control voice conversion results.

[0018] In a possible implementation, determining a target dictionary group based on the first dictionary layer, the second dictionary layer, and the third dictionary layer includes:

[0019] Overlaying the first dictionary layer, the second dictionary layer and the third dictionary layer to obtain an overlay dictionary layer;

[0020] Dividing the superimposed dictionary layer into multiple sub-superimposed dictionary domains, each sub-superimposed dictionary domain includes a sub-first dictionary layer, a sub-second dictionary layer and a sub-third dictionary layer;

[0021] Matching the data items corresponding to the sub-first dictionary layer, the sub-second dictionary layer and the sub-third dictionary layer in each sub-overlay dictionary domain, and determining the data item matching value corresponding to each sub-overlay dictionary domain;

[0022] According to the data item matching value corresponding to each sub-overlay dictionary domain and the preset matching value, a target sub-overlay dictionary domain is determined, and a target dictionary group is determined based on the target sub-overlay dictionary domain.

[0023] By adopting the above technical scheme, by superimposing the first dictionary layer, the second dictionary layer and the third dictionary layer to form a superimposed dictionary layer, it is convenient to comprehensively consider and analyze the dictionary layers corresponding to different pre-parameters, and by dividing the superimposed dictionary layer into regions, it is convenient to improve the accuracy and pertinence when analyzing data items. In addition, by matching the data items contained in different dictionary layers in the sub-superimposed dictionary domain, it is convenient to analyze the degree of association between different data items in the same sub-superimposed dictionary domain. Finally, the target dictionary group is determined from multiple sub-superimposed dictionary domains through data item matching values, which is convenient to improve the reliability when determining the target dictionary group, thereby facilitating the improvement of the accuracy when converting the regulated speech.

[0024] In a possible implementation, converting the control voice corresponding to the control voice instruction into control text based on the target dictionary group includes:

[0025] Performing audio segmentation on the control speech to obtain a plurality of segmented syllables;

[0026] Based on the target dictionary group, at least one initial control field corresponding to each segmented syllable is determined, and based on the position of each segmented syllable in the control speech, at least one initial control field corresponding to each segmented syllable is superimposed to obtain an initial control text;

[0027] Determine a data item matrix corresponding to each segmented syllable according to at least one initial control field corresponding to each segmented syllable, and perform data matrix integration on each data item matrix to obtain an integrated data item matrix, wherein the integrated data item matrix includes a plurality of matrix elements;

[0028] Acquire historical conversion data, and determine an element value of each matrix element based on the historical conversion data;

[0029] A target matrix element is determined based on each matrix element and a preset element value, and the control text is determined based on an initial control field corresponding to the target matrix element.

[0030] By adopting the above technical solution, the control speech is segmented into audio segments, that is, the control speech is divided into multiple independent and easy-to-process segmented syllables, and each segmented syllable is analyzed and processed, so as to improve the accuracy of analyzing and processing the control speech. In addition, after identifying and analyzing the initial control field corresponding to each segmented syllable from the target dictionary group, the matrix element values ​​composed of different initial control fields are determined based on the historical conversion data, so as to facilitate combining the initial control fields corresponding to different segmented syllables, and quantify the probability of each combination appearing in the historical conversion data, so as to facilitate improving the speed and accuracy of determining the control text. Among them, by converting the initial control fields corresponding to different segmented syllables into matrix data, it is also convenient to improve the convenience of analyzing and processing the initial control fields.

[0031] In a possible implementation manner, after determining the element value of each matrix element based on the historical conversion data, the method further includes:

[0032] According to the element value of each matrix element, multiple matrix elements are sorted to obtain an element sequence, and a control text list is determined based on the element sequence and an initial control field corresponding to each matrix element;

[0033] If repeated control language instructions are received within a preset time period, the control text is updated based on the control text list.

[0034] By adopting the above technical solution and arranging different control texts, when the user is less satisfied with the pushed control texts, the relevant control texts can be quickly pushed through the control text list, thereby facilitating improving user satisfaction.

[0035] In a possible implementation, before acquiring the control voice instruction, the method further includes:

[0036] When the wake-up feature is received, the receiving monitoring timer is started and the receiving monitoring duration is obtained;

[0037] If no control voice command is obtained and the receiving monitoring time is longer than the preset receiving time, the receiving monitoring service is closed.

[0038] By adopting the above technical solution, by setting the receiving monitoring time and monitoring whether it exceeds the preset receiving time, the receiving monitoring service can be automatically closed when no effective control voice command is received, so as to avoid being in a state of waiting to receive commands for a long time, thereby reducing resource consumption.

[0039] In a second aspect, the present application provides a control system, which adopts the following technical solution:

[0040] A control system, comprising:

[0041] at least one processor;

[0042] Memory;

[0043] At least one application, wherein the at least one application is stored in a memory and configured to be executed by at least one processor, and the at least one application is configured to: execute the above-mentioned substation voice control method based on large model technology.

[0044] In a third aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution:

[0045] A computer-readable storage medium includes: a computer program that can be loaded by a processor and execute the above-mentioned substation voice control method based on large model technology.

[0046] In a fourth aspect, the present application provides a computer program product, which adopts the following technical solution:

[0047] A computer program product includes a computer program, and when the computer program is executed by a processor, the substation voice control method based on the large model technology is implemented.

[0048] In summary, the present application includes at least one of the following beneficial technical effects:

[0049] By selecting the corresponding target dictionary group for voice conversion through control voice instructions instead of using a fixed dictionary group for voice conversion, it is convenient to improve the voice conversion rate while also improving the accuracy of the voice conversion results. In addition, using the preset natural language recognition large model service to extract keywords from the control text can reduce the redundancy of the control text, thereby facilitating the selection of more accurate and comprehensive results from the control text, and further facilitating the improvement of the accuracy of the subsequent determination of the control logic. By improving the accuracy of converting the control voice into the control text, and the accuracy of converting the control text into the control logic, the accuracy of voice recognition can be improved, thereby facilitating the improvement of the accuracy of language-based control of equipment in the substation.

[0050] By performing audio segmentation on the control speech, that is, dividing the control speech into multiple independent and easy-to-process segmented syllables, and analyzing and processing each segmented syllable, it is convenient to improve the accuracy of analyzing and processing the control speech. In addition, after identifying and analyzing the initial control field corresponding to each segmented syllable from the target dictionary group, the matrix element values ​​composed of different initial control fields are determined based on historical conversion data, so as to facilitate combining the initial control fields corresponding to different segmented syllables, and quantify the probability of each combination appearing in the historical conversion data, so as to facilitate improving the speed and accuracy of determining the control text. Among them, by converting the initial control fields corresponding to different segmented syllables into matrix data, it is also convenient to improve the convenience of analyzing and processing the initial control fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of a method for voice control of a substation based on a large model technology in an embodiment of the present application;

[0052] Figure 2 is a flow chart of a method for determining a target dictionary group in an embodiment of the present application;

[0053] Figure 3 It is a flowchart of a method for controlling text conversion in an embodiment of the present application;

[0054] Figure 4 It is a structural schematic diagram of a control system in an embodiment of the present application. DETAILED DESCRIPTION

[0055] The following is combined with Figure 1-4 This application is described in further detail.

[0056] After reading this specification, those skilled in the art may make non-creative modifications to this embodiment as needed, but such modifications are protected by patent law as long as they are within the scope of the claims of this application.

[0057] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0058] It should be noted that in the optional embodiments of the present application, the object information and other related data involved, when the embodiments in the present application are applied to specific products or technologies, need to obtain the permission or consent of the object, and the collection, use and processing of the relevant data need to comply with the relevant laws, regulations and standards of the relevant countries and regions. In other words, if the embodiments of the present application involve data related to the object, it needs to be obtained with the authorization and consent of the object, the authorization and consent of the relevant departments, and in compliance with the relevant laws, regulations and standards of the country and region. If personal information is involved in the embodiments, the acquisition of all personal information needs to obtain the consent of the individual. If sensitive information is involved, the separate consent of the information subject needs to be obtained. The embodiments also need to be implemented with the authorization and consent of the object.

[0059] Specifically, the embodiment of the present application provides a substation voice control method based on large model technology, which is executed by a control system, and the control system can be a server or a terminal device, wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited thereto. The terminal device and the server can be directly or indirectly connected via wired or wireless communication, and the embodiment of the present application does not limit this.

[0060] refer to Figure 1 , Figure 1 1 is a flow chart of a method for voice control of a substation based on a large model technology in an embodiment of the present application, the method comprising steps S110-S130, wherein:

[0061] Step S110: obtaining a control voice instruction, determining a corresponding target dictionary group based on the control voice instruction, and converting the control voice corresponding to the control voice instruction into control text based on the target dictionary group.

[0062] Specifically, the control voice command is an audio stream sent by the relevant operation and maintenance staff through the microphone. The relevant operation and maintenance staff are the personnel who need to control the equipment in the substation. By issuing the control voice command, any device in the substation can be controlled by voice. In order to improve the effectiveness of the control voice command, the relevant operation and maintenance personnel need to wake up the control system before issuing the control voice command, which can include:

[0063] When the wake-up feature is received, the receiving monitoring timer is started and the receiving monitoring duration is obtained; if the control voice command is not obtained and the receiving monitoring duration is higher than the preset receiving duration, the receiving monitoring service is closed.

[0064] Specifically, since there are often various sounds in daily life, if the wake-up operation feature is not set, the control system needs to keep in monitoring state at all times. In addition to increasing resource consumption, it may also be disturbed by background noise, thereby misidentifying the control voice command. Therefore, relevant operation and maintenance personnel are required to wake up the control system through the wake-up feature before issuing the control voice command. The wake-up feature can be "Little A Little A" or "ABCD". The specific wake-up feature is not specifically limited in the embodiment of the present application and can be set according to relevant technical personnel.

[0065] For the same reason, after receiving the wake-up feature, the duration of the control system entering the monitoring state after awakening can be limited by comparing the receiving monitoring time and the preset receiving time. That is, it does not mean that the control voice command can be issued for a long time after the control system is awakened. Once the receiving monitoring time is higher than the preset receiving time, the control system will exit the monitoring state again, where the preset receiving time can be 10 seconds or 15 seconds. The specific time is not specifically limited in the embodiments of this application and can be set by relevant technical personnel. For example, if the wake-up feature is "Xiaoyou Xiaoyou", the relevant operation and maintenance staff need to call out "Xiaoyou Xiaoyou" through the microphone before issuing the control voice command, and need to use the microphone to issue the control voice command within a period not longer than the preset receiving time.

[0066] After successfully acquiring the control voice command, it is necessary to determine the corresponding target dictionary group based on the control voice, and then parse the control voice command based on the corresponding target dictionary group to obtain the control text corresponding to the control voice command, wherein different control voice commands correspond to different target dictionary groups. The target dictionary group contains multiple data items, and a wide range of data items can provide a relatively rich vocabulary resource for the conversion process, thereby facilitating the improvement of the accuracy of converting the control voice command into control text. In addition, the target dictionary group also includes the association between each data item. Through the association between the data items, the common usage collocations and grammatical rules between the data items can be determined, which is convenient for grammatical analysis in the process of converting the control voice command into control text. Therefore, the target dictionary group is particularly critical in the process of converting the control voice command into control text. For example, after the relevant operation and maintenance staff wake up the control system, the control text corresponding to the control voice command issued can be "Help me check the grounding current monitoring status of the main transformer core".

[0067] Step S120: Based on the preset natural language recognition model, keywords are extracted from the control text to obtain key control text.

[0068] Specifically, the process of building a preset natural language recognition large model may include:

[0069] Step 1: Sample data collection: The sample data is substation operation instruction data, including text, voice and other forms;

[0070] Step 2: Manually label sample features: extract key data, key text, key operations, etc. from the sample data through manual recognition, and manually label each sample data based on the recognition results;

[0071] Step 3: Model training: Use the sample data and the manual annotation content corresponding to each sample data to train the model to be trained until the model training results for the sample data are consistent with the corresponding manual annotations.

[0072] The construction process of the preset language recognition large model is not specifically limited in the embodiments of the present application, as long as the keyword extraction of the control text can be performed. For example, the control text is "Help me check the grounding current monitoring situation of the main transformer core". After the preset natural language recognition large model is used to extract keywords from the control text, the key control text obtained may be "main transformer core grounding current monitoring situation". By extracting keywords from the control text, it is convenient to reduce the redundancy of the control text. In addition, by extracting keywords from the control text, the accuracy of the speech recognition results can also be improved. For example, the control text is "Help me check the situation of the only patrol last week", where "only" is incorrectly recognized, and the correct recognition content should be "intelligent". By extracting keywords from the control text, the key control text obtained may be "last week's intelligent patrol situation".

[0073] Step S130: Performing logical transformation on the key control text to obtain control logic, and controlling the target control device based on the control logic.

[0074] Specifically, since different devices in the substation have different control execution standards, it is necessary to perform logical transformation on key control texts to ensure that each voice control instruction can be accurately understood and executed by the corresponding device, thereby reducing the risk of misoperation. The control system can establish communication with each device in the substation, that is, after determining the control logic, the control instruction can be determined according to the control logic, and then the control instruction can be sent to the corresponding device through wireless transmission or limited transmission to achieve control of any device in the substation. Perform logical transformation on the control text, that is, convert the control text into content including logical relationships, and the logical relationships include but are not limited to and, or, and, for example, the control text is "When the indoor temperature T is higher than 28 degrees, turn on the air conditioning cooling mode A", and the result obtained after logical transformation can be "If T>28, execute A".

[0075] For the embodiments of the present application, the corresponding target dictionary group is selected for voice conversion through the control voice instruction, rather than using a fixed dictionary group for voice conversion, which is convenient for improving the voice conversion rate while also improving the accuracy of the voice conversion results. In addition, the control text is extracted by using the preset natural language recognition large model service, which can reduce the redundancy of the control text, thereby facilitating the selection of more accurate and comprehensive results from the control text, and further facilitating the improvement of the accuracy of the subsequent determination of the control logic. By improving the accuracy of converting the control voice into the control text, and the accuracy of converting the control text into the control logic, the accuracy of voice recognition can be improved, thereby facilitating the improvement of the accuracy of language-based control of equipment in the substation.

[0076] Further, in order to improve the compatibility between the target dictionary group and the control voice instruction, the corresponding target dictionary group is determined based on the control voice instruction, which may specifically include steps S1 to S5, such as Figure 2 As shown, where:

[0077] Step S1: After detecting the control voice command, obtain the pre-set parameters, which include the control time, control environment and device status.

[0078] Specifically, the pre-set parameters are the corresponding control time, the control environment in the substation, and the device status of each device in the substation when the control voice command is obtained or detected. Since the time of obtaining different control voice commands is different, the pre-set parameters corresponding to different control voice commands are also different.

[0079] The control time is the time when the control voice command is received. At a fixed control time, the relevant operation and maintenance staff may issue similar commands, for example, issuing equipment update commands at 00:00. The control environment is the operating environment of the equipment in the substation. Different operating environments may affect the content of the voice control commands of the relevant operation and maintenance staff. For example, in hot weather, the relevant operation and maintenance staff may issue cooling commands to cool down the equipment in the substation. The equipment status is the status of the equipment in the substation during operation. Different equipment status may also affect the content of the instructions issued by the relevant operation and maintenance staff. For example, when the equipment status is in the maintenance status, the relevant operation and maintenance staff generally issue a recovery operation instruction. The equipment status includes but is not limited to operation, hot standby, cold standby, maintenance, etc. Therefore, the data items that may be used when converting the control voice in the control voice command into control text can be predicted based on the control time, control environment and equipment status to improve the adaptability between the target dictionary group and the control voice command.

[0080] Step S2: determining a first dictionary group from the historical dictionary library based on the control moment, and determining a first dictionary layer based on the first dictionary group and a preset data layer layer, wherein the first dictionary group includes a plurality of first data items and a data association relationship between each first data item.

[0081] Specifically, the historical dictionary library contains historical control texts corresponding to each historical control moment in the historical stage, that is, it contains available data items corresponding to each historical control moment. Therefore, by analyzing the historical control texts, it is convenient to filter out the first data items that may be used at the control moment from the historical dictionary library. Multiple first data items constitute a first dictionary group, wherein the duration corresponding to the historical stage can be 5 days before the control moment, or 7 days between the control moments. The specific duration is not specifically limited in the embodiment of the present application and can be set by relevant technical personnel.

[0082] In addition to including multiple first data items corresponding to the historical control stage, the first dictionary group also includes data association relationships between the first data items. The data association relationships between the first data items can be common data item combinations. For example, the first data items can be equipment, detection, data, upload, status, update, etc. Common data item combinations can be "equipment detection", "data upload", "status update", etc. The preset data layer can be a pre-established plane layer. First, according to the data association relationship between the first data items, the first data items in the first dictionary group can be organized to obtain the first dictionary binary tree corresponding to the first dictionary group, and then the first dictionary binary tree is imported into the preset data layer to obtain the first dictionary layer. Through the first dictionary layer, the first data items contained in the first dictionary group and the data association relationship between the first data items can be intuitively viewed.

[0083] Step S3: Determine a second dictionary group from the historical dictionary library based on the control environment, and determine a second dictionary layer based on the second dictionary group and the preset data layer. The second dictionary group includes multiple second data items and data association relationships between each second data item.

[0084] Specifically, the historical dictionary library also includes the corresponding historical control texts under each control environment in the historical stage. By analyzing the historical control texts, it is convenient to determine the second data items that may be used in the control environment. Multiple second data items constitute the second dictionary group. The implementation method of determining the second dictionary layer based on the second dictionary group and the preset data layer can refer to the method of determining the first dictionary layer based on the first dictionary group and the preset data layer in the above embodiment, which will not be repeated here.

[0085] Step S4: determining a third dictionary group from the historical dictionary library based on the device status, and determining a third dictionary layer based on the third dictionary group and the preset data layer, wherein the third dictionary group includes multiple third data items and data association relationships between each third data item.

[0086] Specifically, the historical dictionary library also includes the corresponding historical control texts under each device state in the historical stage. By analyzing the historical control texts, it is convenient to determine the third data items that may be used under the device state. Multiple third data items constitute a third dictionary group. The implementation method of determining the third dictionary layer based on the third dictionary group and the preset data layer can refer to the method of determining the first dictionary layer based on the first dictionary group and the preset data layer in the above embodiment, which will not be repeated here.

[0087] Among them, in order to facilitate unified analysis of multiple data items, the preset data layer levels used when constructing the first dictionary layer, the second dictionary layer and the third dictionary layer are consistent, that is, when the same data item exists in different dictionary layers, the data item is at the corresponding position of the different dictionary layers. For example, the first dictionary layer contains data item a, and the second dictionary layer also contains data item a. It is known that the data item position of data item a in the first dictionary layer is (2, 2), so it can be known that the data item position of data item a in the second dictionary layer is also (2, 2).

[0088] Step S5: Determine a target dictionary group based on the first dictionary level, the second dictionary level, and the third dictionary level.

[0089] Specifically, the first dictionary layer, the second dictionary layer, and the third dictionary layer can be merged or data items can be compared to determine the area where there are overlapping data items between the first dictionary layer, the second dictionary layer, and the third dictionary layer, and the corresponding target dictionary group can be determined based on the area where there are overlapping data items. The target dictionary group corresponding to the control voice instruction is jointly determined based on the dictionary layers corresponding to multiple pre-parameters, so as to improve the adaptability between the target dictionary group and the control voice instruction.

[0090] Furthermore, in order to improve the reliability of determining the target dictionary group, based on the first dictionary layer, the second dictionary layer and the third dictionary layer, determining the target dictionary group may specifically include:

[0091] The first dictionary layer, the second dictionary layer and the third dictionary layer are layered to obtain a superimposed dictionary layer; the superimposed dictionary layer is divided into regions to obtain a plurality of sub-superimposed dictionary domains, each of which includes a sub-first dictionary layer, a sub-second dictionary layer and a sub-third dictionary layer; the data items corresponding to the sub-first dictionary layer, the sub-second dictionary layer and the sub-third dictionary layer in each sub-superimposed dictionary domain are matched to determine the data item matching value corresponding to each sub-superimposed dictionary domain; a target sub-superimposed dictionary domain is determined according to the data item matching value corresponding to each sub-superimposed dictionary domain and a preset matching value, and a target dictionary group is determined based on the target sub-superimposed dictionary domain.

[0092] Specifically, since the first dictionary group, the second dictionary group and the third dictionary group are constructed using the same preset data layer, the first dictionary layer, the second dictionary layer and the third dictionary layer can be directly superimposed. In order to view the data items contained in all dictionary groups and the data association relationship between each data item through the superimposed dictionary layer, the transparency of each dictionary layer can be adjusted before superimposing multiple dictionary layers. When the superimposed dictionary layer is divided into regions, the superimposed dictionary layer can be evenly divided into 8 regions, or the superimposed dictionary layer can be evenly divided into 12 regions. The specific division method is not specifically limited in the embodiment of the present application.

[0093] After the superimposed dictionary layer is divided into regions, multiple sub-superimposed dictionary domains can be obtained, each of which contains a corresponding sub-first dictionary layer, a sub-second dictionary layer, and a sub-third dictionary layer. The data item matching value of each sub-superimposed dictionary domain can be determined by analyzing the distribution of data items in the sub-superimposed dictionary domain and the overlap of data items between different sub-dictionary layers. The higher the data item matching value, the greater the probability that the data item in the sub-superimposed dictionary domain is used in the control text conversion process under the pre-parameters corresponding to the control voice instruction. By matching the data items contained in different dictionary layers in the sub-superimposed dictionary domain, the degree of association between different data items in the same sub-superimposed dictionary domain can be analyzed.

[0094] After determining the data item matching value corresponding to each sub-overlay dictionary domain, the sub-overlay dictionary domain that is higher than the preset matching value can be determined as the target sub-overlay dictionary domain. The number of target sub-overlay dictionary domains can be multiple. Finally, the data items contained in each target sub-overlay dictionary domain are integrated to obtain a target dictionary group.

[0095] Further, in order to improve the speed and accuracy of determining the control text, when converting the control voice corresponding to the control voice instruction into the control text based on the target dictionary group, it can specifically include steps Sa1 to Sa5, such as Figure 3 As shown, where:

[0096] Step Sa1: Perform audio segmentation on the control speech to obtain multiple segmented syllables.

[0097] Specifically, when performing audio segmentation on the control speech, the segmentation processing can be performed according to a preset segmentation standard, wherein the preset segmentation standard can be segmentation processing based on pause positions, common syllable lengths, and preset terminators, etc. The specific segmentation method is not specifically limited in the embodiments of the present application, as long as the control speech can be segmented into audio segments.

[0098] Step Sa2: Based on the target dictionary group, determine at least one initial control field corresponding to each segmented syllable, and based on the position of each segmented syllable in the controlled speech, superimpose at least one initial control field corresponding to each segmented syllable to obtain initial control text.

[0099] Specifically, in addition to multiple data items, the target dictionary group also includes data item syllables corresponding to each data item. The syllables corresponding to different data items may be the same. Therefore, the same data item syllable may correspond to at least one data item. Each segmented syllable is compared with the data item syllable in the target dictionary group, and the target data item syllable corresponding to each segmented syllable is determined from the target dictionary group, and at least one data item corresponding to each target data item syllable is determined as at least one initial control field corresponding to the segmented syllable.

[0100] By adopting the above method, at least one initial control field corresponding to each segmented syllable can be determined, and then based on the data superposition technology, at least one initial control field corresponding to each segmented syllable is packaged and processed, and the packaged result is superimposed on the position of the corresponding segmented syllable in the control speech to obtain the initial control text, and the initial control text is a combination of the initial control fields corresponding to each segmented syllable. Since the segmented syllable is determined after the control speech is segmented, different segmented syllables have different positions in the control speech.

[0101] Step Sa3: Determine the data item matrix corresponding to each segmented syllable according to at least one initial control field corresponding to each segmented syllable, and perform data matrix integration on each data item matrix to obtain an integrated data item matrix, wherein the integrated data item matrix includes multiple matrix elements.

[0102] Specifically, since different segmented syllables contain at least one initial control field, there are multiple ways to combine the initial control fields corresponding to multiple segmented syllables, that is, the final generated control text may have multiple versions, wherein at least one initial control field corresponding to each segmented syllable can be determined as a data item matrix, and by operating on multiple data item matrices, it is convenient to achieve a full combination between at least one initial control field corresponding to different segmented syllables. The integrated data item matrix contains multiple matrix elements, each of which corresponds to a combination result.

[0103] Step Sa4: Obtain historical conversion data, and determine the element value of each matrix element based on the historical conversion data.

[0104] Specifically, the historical conversion data includes multiple historical control texts corresponding to historical stages. The element value corresponding to each matrix element can be determined according to the number of times each combination result appears in the historical conversion data. The more times it appears in the historical conversion data, the higher the element value of the corresponding matrix element. There is a number correspondence between the number of appearances and the element value. Based on the number correspondence, the element values ​​corresponding to different numbers can be determined. The specific content of the number correspondence is not specifically limited in the embodiments of the present application and can be set by relevant technical personnel.

[0105] Step Sa5: Determine the target matrix element based on each matrix element and the preset element value, and determine the control text based on the initial control field corresponding to the target matrix element.

[0106] Specifically, the combination results corresponding to each matrix element can constitute an initial control text. In order to improve the accuracy of determining the final control text, it can be selected from multiple matrix elements according to preset element values. The target matrix element is a matrix element whose element value is higher than the preset element value. The number of target matrix elements is not specifically limited in the embodiments of the present application.

[0107] After identifying and analyzing the initial control fields corresponding to each segmented syllable from the target dictionary group, the matrix element values ​​composed of different initial control fields are determined based on the historical conversion data, so as to facilitate the combination of the initial control fields corresponding to different segmented syllables and quantify the probability of each combination appearing in the historical conversion data, so as to facilitate improving the speed and accuracy of determining the control text. Among them, by converting the initial control fields corresponding to different segmented syllables into matrix data, it is also convenient to improve the convenience of analyzing and processing the initial control fields.

[0108] Furthermore, in order to improve user satisfaction, the method provided in the embodiment of the present application further includes, after determining the element value of each matrix element based on the historical conversion data:

[0109] According to the element value of each matrix element, multiple matrix elements are sorted to obtain an element sequence, and a control text list is determined based on the element sequence and the initial control field corresponding to each matrix element; if repeated control voice instructions are received within a preset time period, the control text is updated based on the control text list.

[0110] Specifically, since the differences between different matrix elements may be small, and only one control text will be generated from the target matrix elements and pushed, when the relevant operation and maintenance personnel are dissatisfied with the push results, they may repeatedly issue the same control voice command in a short period of time. At this time, in order to reduce the amount of data calculation, a new control text can be generated from other target matrix elements or other matrix elements, and pushed again. Since the other target matrix elements or other matrix elements are all determined, the push rate will increase when the control text of the information is pushed again, thereby facilitating the improvement of the satisfaction of the relevant operation and maintenance personnel. In addition, since the similarity between other target matrix elements or other matrix elements and the target matrix elements is high, the satisfaction of relevant operation and maintenance personnel may be improved after re-pushing. For example, the control text corresponding to the original target matrix element is "turn on the lights of the third device and the fourth device in the second row", but the control intention corresponding to the control voice command of the relevant operation and maintenance staff may be "turn on the lights of the second device and the third device in the second row". At this time, if the equipment in the substation is controlled according to the control text corresponding to the target matrix element, the control result is that the third device and the fourth device in the second row in the substation are lit. However, the satisfaction of the relevant operation and maintenance staff with the control result may be reduced. At this time, by quickly selecting the next control text according to the control text list for push, the probability that the next control text is "turn on the lights of the second device and the third device in the second row" is relatively high. Therefore, re-pushing based on the control text list is convenient for improving the accuracy of the push results, that is, it is convenient to improve the control accuracy while also improving the satisfaction of relevant operation and maintenance personnel.

[0111] The element value corresponding to each matrix element in the element sequence can be from large to small, or from small to large. The specific arrangement order is not specifically limited in the embodiment of the present application. The control text list contains the initial control text corresponding to each matrix element. When a repeated control voice instruction is received within a preset time period, the control text is updated from high to low according to the element value corresponding to each initial control text. The number of updates is not higher than the preset update number, where the preset update number can be 3 times or 5 times, and the specific number is not specifically limited in the embodiment of the present application. The duration corresponding to the preset time period can be 2 seconds or 3 seconds after the control text is first pushed. The specific duration is not specifically limited in the embodiment of the present application. Repeated control voice instructions are voice instructions whose similarity with the control voice instruction is higher than the preset standard similarity.

[0112] The present application provides a control system, such as Figure 4 As shown, Figure 4The control system 400 shown includes: a processor 401 and a memory 403. The processor 401 and the memory 403 are connected, such as through a bus 402. Optionally, the control system 400 may also include a transceiver 404. It should be noted that in actual applications, the transceiver 404 is not limited to one, and the structure of the control system 400 does not constitute a limitation on the embodiments of the present application.

[0113] Processor 401 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 401 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0114] The bus 402 may include a path to transmit information between the above components. The bus 402 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 402 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The fact that only one line is used in the diagram does not mean that there is only one bus or only one type of bus.

[0115] The memory 403 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0116] The memory 403 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 401. The processor 401 is used to execute the application code stored in the memory 403 to implement the contents shown in the above method embodiment.

[0117] The control system includes, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. It can also be a server, etc. Figure 4 The control system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0118] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment.

[0119] An embodiment of the present application provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the method in any of the above embodiments is implemented.

[0120] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0121] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A voice control method for a substation based on a large model technology, characterized in that: include: Acquire a control voice instruction, determine a corresponding target dictionary group based on the control voice instruction, and convert the control voice corresponding to the control voice instruction into control text based on the target dictionary group; Based on a preset natural language recognition large model, keyword extraction is performed on the control text to obtain key control text; Performing logical transformation on the key control text to obtain control logic, and controlling the target control device based on the control logic; Wherein, determining the corresponding target dictionary group based on the control voice instruction includes: After detecting the control voice command, obtaining the pre-set parameters, wherein the pre-set parameters include the control time, the control environment and the device status; Determine a first dictionary group from a historical dictionary library based on the control moment, and determine a first dictionary layer based on the first dictionary group and a preset data layer, wherein the first dictionary group includes a plurality of first data items and a data association relationship between each first data item; Determine a second dictionary group from the historical dictionary library based on the control environment, and determine a second dictionary layer based on the second dictionary group and a preset data layer, wherein the second dictionary group includes a plurality of second data items and a data association relationship between each second data item; Determine a third dictionary group from the historical dictionary library based on the device state, and determine a third dictionary layer based on the third dictionary group and a preset data layer, wherein the third dictionary group includes a plurality of third data items and a data association relationship between each third data item; determining a target dictionary group based on the first dictionary layer, the second dictionary layer, and the third dictionary layer; The step of determining a target dictionary group based on the first dictionary layer, the second dictionary layer, and the third dictionary layer includes: Overlaying the first dictionary layer, the second dictionary layer and the third dictionary layer to obtain an overlay dictionary layer; Dividing the superimposed dictionary layer into multiple sub-superimposed dictionary domains, each sub-superimposed dictionary domain includes a sub-first dictionary layer, a sub-second dictionary layer and a sub-third dictionary layer; Matching the data items corresponding to the sub-first dictionary layer, the sub-second dictionary layer and the sub-third dictionary layer in each sub-overlay dictionary domain, and determining the data item matching value corresponding to each sub-overlay dictionary domain; According to the data item matching value corresponding to each sub-overlay dictionary domain and the preset matching value, a target sub-overlay dictionary domain is determined, and a target dictionary group is determined based on the target sub-overlay dictionary domain.

2. According to the large model technology-based substation voice control method of claim 1, it is characterized in that: The step of converting the control voice corresponding to the control voice instruction into control text based on the target dictionary group includes: Performing audio segmentation on the control speech to obtain a plurality of segmented syllables; Based on the target dictionary group, at least one initial control field corresponding to each segmented syllable is determined, and based on the position of each segmented syllable in the control speech, at least one initial control field corresponding to each segmented syllable is superimposed to obtain an initial control text; Determine a data item matrix corresponding to each segmented syllable according to at least one initial control field corresponding to each segmented syllable, and perform data matrix integration on each data item matrix to obtain an integrated data item matrix, wherein the integrated data item matrix includes a plurality of matrix elements; Acquire historical conversion data, and determine an element value of each matrix element based on the historical conversion data; A target matrix element is determined based on each matrix element and a preset element value, and the control text is determined based on an initial control field corresponding to the target matrix element.

3. The substation voice control method based on large model technology according to claim 2 is characterized in that: After determining the element value of each matrix element based on the historical conversion data, the method further includes: According to the element value of each matrix element, multiple matrix elements are sorted to obtain an element sequence, and a control text list is determined based on the element sequence and an initial control field corresponding to each matrix element; If repeated control language instructions are received within a preset time period, the control text is updated based on the control text list.

4. The substation voice control method based on large model technology according to claim 1 is characterized in that: Before acquiring the control voice instruction, the method further includes: When the wake-up feature is received, the receiving monitoring timer is started and the receiving monitoring duration is obtained; If no control voice command is obtained and the receiving monitoring time is longer than the preset receiving time, the receiving monitoring service is closed.

5. A control system, characterized in that: The control system includes: at least one processor; Memory; At least one application, wherein the at least one application is stored in a memory and configured to be executed by at least one processor, and the at least one application is configured to: execute a substation voice control method based on large model technology as described in any one of claims 1-4.

6. A computer-readable storage medium, characterized in that: include: A computer program is stored which can be loaded by a processor and execute a voice control method for a substation based on a large model technology as described in any one of claims 1-4.

7. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the steps of a substation voice control method based on large model technology as described in any one of claims 1 to 4.

Images