Man-machine conversation method and system of digital energy-saving pump

Abstract

The invention belongs to the field of man-machine conversation, and particularly relates to a man-machine conversation method and system for a digital energy-saving pump, and the method comprises the steps: firstly generating a voice statement containing a plurality of words according to conversation voice input by a user, and then combining with a preset digital energy-saving pump standard word library and a digital energy-saving pump instruction library, and through unique correlation analysis, generating a voice statement containing a plurality of words; accurately determining a control instruction in the conversation voice; wherein the standard word library of the digital energy-saving pump covers all words and expanded words thereof in the instruction library, and the expanded words are composed of all out-of-order words of the words, so that the accuracy of instruction recognition is greatly improved; and finally, the determined control instruction is executed, so that efficient and intelligent man-machine conversation interaction is realized in a low-computing-power digital energy-saving pump scene, and the long-term instruction identification problem in the field is effectively solved.

Description

Technical Field

[0001] The invention belongs to the field of human-computer dialogue, and in particular relates to a human-computer dialogue method and system for a digital energy-saving pump. Background Art

[0002] In the fields of industry, agriculture and urban water supply, the intelligence of water pump operation control is crucial. Although human-computer dialogue technology provides a direction for improving its management level, it faces many challenges when used for digital energy-saving pump control. For example, most of the existing human-computer dialogue methods use voice recognition and then combine deep semantic analysis to perform semantic similarity matching, but this matching requires advanced computing power support devices to support the corresponding calculation process. However, due to the limitations of energy saving and configuration devices, it is impossible to configure advanced deep learning algorithms for advanced voice calculation processes on energy-saving pumps. Therefore, how to use low-computing power equipment on energy-saving pumps to achieve fast and effective human-computer dialogue is one of the important problems that the existing technology needs to solve. For example, the Chinese patent with authorization announcement number CN110555095B discloses a human-computer dialogue method and device, and the Chinese patent with authorization announcement number CN114691852B discloses a human-computer dialogue system and method. The above existing technologies have many defects. They all use more advanced algorithm models in semantic analysis and deep learning matching to achieve high-precision voice dialogue, but these methods require advanced processing equipment and high-configuration computing power, which is impossible for the energy-saving pump corresponding to this application. Therefore, the corresponding methods cannot be migrated to the human-computer dialogue process of the energy-saving pump corresponding to this application.

[0003] To this end, the present invention provides a human-machine dialogue method and system for a digital energy-saving pump. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention proposes a human-computer dialogue method and system for a digital energy-saving pump. The method first generates a voice sentence containing multiple words based on the dialogue voice input by the user, and then combines the preset digital energy-saving pump standard vocabulary and the digital energy-saving pump instruction library to accurately determine the control instructions in the dialogue voice through unique association analysis; wherein, the digital energy-saving pump standard vocabulary covers all words in the instruction library and their extended words, and these extended words are composed of all the words in random order, which greatly improves the accuracy of instruction recognition; finally, the determined control instructions are executed, thereby realizing efficient and intelligent human-computer dialogue interaction in the low-computing power digital energy-saving pump scenario, effectively solving the instruction recognition problem that has long been faced in this field.

[0005] To achieve the above object, the present invention provides the following technical solutions: A human-machine dialogue method for a digital energy-saving pump, comprising: Generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; Determine the control instructions in the dialogue voice in combination with a preset digital energy-saving pump standard word library and a digital energy-saving pump instruction library; Execute the control instruction, wherein the digital energy-saving pump standard word library includes all words in the digital energy-saving pump instruction library and extended words of all words, and the extended words include all disorderly arranged words of the words.

[0006] Specifically, the steps for constructing the digital energy-saving pump standard vocabulary include: Based on all the control instruction texts in the digital energy-saving pump instruction library, the control instruction word vector is obtained through the word segmentation algorithm; Based on the control instruction word vector, the occurrence frequency of each key character in the effective text word vector is obtained through a statistical algorithm. According to the occurrence frequency of each key character, the keyword frequency level is obtained and arranged in descending order; According to the arrangement result, the key characters with the highest keyword frequency level and the remaining characters in the control instruction word vector are arranged in forward or reverse order to form key character phrases, and the key character phrases that are the same as the control instruction text are placed in the first place to obtain the first keyword expansion phrase, and a front mapping is set in the first keyword expansion phrase; Remove the key characters with the highest keyword frequency level, repeat the process of obtaining the first keyword expansion phrase, and obtain the keyword expansion phrases corresponding to the remaining key characters; According to the keyword frequency level, all keyword expansion phrases corresponding to each control instruction text are stored in nodes through the prefix tree algorithm to obtain the standard vocabulary of the digital energy-saving pump.

[0007] Specifically, the steps of setting the pre-mapping include: A pre-mapping is established between the key character word group after the first one in the keyword expansion phrase and the first key character word group. When the key character word group found is any key character word group after the first one, the first key character word group is used for search feedback through the pre-mapping.

[0008] Specifically, the steps of performing node storage by using the prefix tree algorithm include: According to the keyword expansion phrase corresponding to each control instruction text, the key character is used as the root node, and according to the position information of the remaining characters in the keyword expansion word, the mounting sequence chain of the remaining character storage nodes is determined; According to the position information of the keyword in the keyword expansion word, a connection relationship between the key character and the node mounting sequence chain is established to obtain a storage node tree corresponding to the keyword expansion word.

[0009] Furthermore, the connection relationship between the keyword and the node mounting sequence chain is established as follows: Assuming that the key character corresponding to the current keyword expansion word is in the third position, the remaining characters are kept under the corresponding tree node, and the third tree node is set to an empty node, and the corresponding empty node is connected to the root node to obtain the connection relationship between the keyword and the node mounting sequence chain.

[0010] Specifically, the step of performing node storage by using the prefix tree algorithm also includes: The key characters that are the same as the control instruction text are grouped into words and hot-coded as the standard hot-coding of the keyword expansion words; At the same time, the remaining words of the first keyword expansion phrase are adjusted according to the character position information in the corresponding composition, and the hot coding vector corresponding to each remaining word of the first keyword expansion phrase is obtained; According to the standard hot coding of the keyword expansion words and the hot coding vectors corresponding to each remaining group of words, a character position hot coding set corresponding to the first keyword expansion word group is obtained; Based on the process of obtaining the character position hot coding set corresponding to the first keyword expansion phrase, the character position hot coding set of the keyword expansion phrase corresponding to the remaining key characters is obtained, and each character position hot coding is embedded in the node mounting sequence chain of the corresponding keyword expansion word.

[0011] Specifically, the steps of constructing the digital energy-saving pump standard vocabulary library also include: Obtain the historical voice sentences corresponding to the historical users' different dialects or multi-language control voices and the number of correct control command text searches; According to the historical voice sentences, the correct control instruction text and the search frequency, a character co-occurrence matrix in the historical voice sentences is obtained by combining the keyword expansion phrase through a text segmentation algorithm; The character co-occurrence matrix is ​​constructed by hot coding the character positions of the corresponding keyword expansion words in the historical speech sentences; Based on the character co-occurrence matrix, the frequencies of occurrence of historical keyword expansion phrases, colloquial keyword expansion phrases and corresponding character position hot codes are obtained, and the colloquial keyword expansion phrases and corresponding character position hot codes are added to the corresponding keyword expansion phrase set.

[0012] Specifically, the steps of constructing the digital energy-saving pump standard vocabulary library also include: According to the frequency of occurrence of historical keyword expansion phrases and corresponding character position hot coding, the frequency of each keyword expansion word is obtained; The frequency of each keyword expansion word obtained is built into the corresponding keyword expansion word in the digital energy-saving pump standard word library; A frequency threshold is set. When the frequency of the corresponding keyword expansion word is greater than the frequency threshold, the corresponding keyword expansion word is mapped to the frequent word node configured in the digital energy-saving pump standard word library.

[0013] Specifically, the step of determining the control instruction in the conversation voice includes: According to the voice sentence acquired in real time, based on the process of acquiring the keyword expansion phrase, a real-time voice sentence keyword expansion phrase set is obtained; According to the keyword expansion phrase corresponding to each keyword frequency level in the real-time voice sentence keyword expansion phrase set, a cyclic parallel path search is performed through a multi-threaded search queue to obtain a real-time standard control instruction keyword expansion phrase set; According to the real-time standard control instruction keyword expansion phrase set combined with the digital energy-saving pump instruction library, the real-time standard control instruction text is obtained.

[0014] Specifically, the steps of performing cyclic parallel path search through a multi-threaded search queue include: According to the keyword expansion words in each keyword expansion phrase in the real-time voice sentence keyword expansion phrase set, construct each keyword expansion word search queue; According to the frequency of historical keyword expansion words, a keyword expansion word search queue containing a frequency greater than a frequency threshold is marked as a priority search to obtain a priority search queue; Based on the priority search queue, the node mounting sequence chain corresponding to the frequent word nodes and the infrequent words under the standard word library of the digital energy-saving pump is searched through parallel threads to obtain the first real-time standard control instruction keyword expansion phrase set; If the number of the first real-time standard control instruction keyword expansion phrase set is greater than one, a second search is performed on the first real-time standard control instruction keyword expansion phrase set through the remaining search queues in the keyword expansion phrase search queue to obtain the real-time standard control instruction keyword expansion phrase set.

[0015] A human-machine dialogue system for a digital energy-saving pump, comprising: an acquisition module and an instruction generation module; An acquisition module, used to generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; The instruction generation module is used to determine the control instructions in the dialogue voice by combining a preset digital energy-saving pump standard vocabulary library and a digital energy-saving pump instruction library.

[0016] Compared with the prior art, the present invention has the following beneficial effects: In view of the deficiencies in the prior art, the present invention, in terms of instruction recognition, makes use of a digital energy-saving pump standard vocabulary library containing instruction library words and their out-of-order extended words, combined with character co-occurrence matrices, position vectors and extended word frequency analysis constructed using historical multilingual and dialect voice data, thereby greatly improving the accuracy of instruction recognition and being able to effectively cope with complex input situations such as accents and disordered word order; in terms of computing power requirements, it bypasses the reliance on advanced deep learning algorithms, adopts database keyword search, statistical analysis and other low computing power consumption methods, and narrows the search range by setting frequency thresholds, prioritizing matching of high-frequency words and other means, significantly reducing the amount of computation, which is in line with the low computing power operating environment of energy-saving pumps; in particular, the present application is based on a dynamic optimization mechanism of historical data, which can continuously update the frequency of extended words, so that the vocabulary library keeps up with changes in user language habits, while optimizing the storage structure and improving the efficiency of vocabulary library management and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flow chart of a human-machine dialogue method for a digital energy-saving pump of the present invention; Figure 2 This is a module diagram of a human-machine dialogue system for a digital energy-saving pump of the present invention. DETAILED DESCRIPTION

[0018] Example 1

[0019] Most of the existing human-computer dialogue methods use high-precision deep learning algorithms to perform corresponding semantic recognition and matching, but this also requires a lot of memory and computing power, which increases the cost of implementation. Digital energy-saving pumps are mainly used for pumping irrigation, which requires a lot of energy consumption. Therefore, it is impossible to configure memory and computing devices that support high-precision computing to perform human-computer dialogue operations. For this, please refer to Figure 1 , an embodiment provided by the present invention: a human-computer dialogue method for a digital energy-saving pump, the steps comprising: S1. Generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; Furthermore, in this embodiment, the voice sentence is generated by calling the pre-trained voice and language recognition model. Here, only the corresponding recognition interface is called, and the corresponding voice and language recognition model is built into the edge computing node for voice recognition.

[0020] Furthermore, the speech and language recognition model in this embodiment is obtained by fine-tuning the pre-trained Whisper-Tiny model using Chinese dialects, Mandarin and foreign speech from different regions. It can recognize different dialects and different languages ​​and convert the recognized language into Chinese.

[0021] This process uses Chinese dialects, Mandarin and foreign speech in different regions to fine-tune the pre-trained Whisper-Tiny model, so that the model can recognize different dialects and different languages ​​and convert them into Chinese; this greatly improves the versatility and adaptability of the system, can meet the needs of users with different language backgrounds, and expands the scope of use of the product; for example, different dialects have differences in pronunciation, intonation, and vocabulary usage. During the fine-tuning process, the model can gradually capture these differences and optimize its own parameters, so as to accurately recognize different voice inputs and convert them into unified Chinese expressions for subsequent processing. In addition, in actual application scenarios, such as the control scenario of digital energy-saving pumps, the model is deployed on the edge computing node. When the user inputs the conversational voice, the edge node can process it immediately without uploading the voice data to the cloud; this not only speeds up the recognition speed, but also reduces the network cost and delay caused by data transmission. For some scenarios with high real-time requirements and inconvenient data transmission, this method can effectively ensure the stable operation of the system while reducing the high cost caused by relying on cloud computing.

[0022] S2. Determine the control instructions in the dialogue voice in combination with a preset digital energy-saving pump standard vocabulary and a digital energy-saving pump instruction library; wherein the digital energy-saving pump standard vocabulary includes all words in the digital energy-saving pump instruction library and the expanded words of all words, and the expanded words include all the disorderly arranged words of the words.

[0023] Furthermore, in this embodiment, the digital energy-saving pump standard word library includes all disordered words in the instruction library as extended words, which significantly increases the possibility of matching control instructions; even when the user's input voice has disordered word order, the system can still accurately identify the correct control instruction, improving the system's fault tolerance for user input. For example, the standard instruction is "start the water pump", and the user may say "start the water pump". Since the word library contains disordered extended words such as "start the water pump", the system can identify it as a valid control instruction during the matching process, thereby accurately executing the corresponding operation and improving the user experience.

[0024] Furthermore, the steps of constructing the digital energy-saving pump standard vocabulary in this embodiment include: Based on all the control instruction texts in the digital energy-saving pump instruction library, the control instruction word vector is obtained through the word segmentation algorithm; Based on the control instruction word vector, the occurrence frequency of each key character in the effective text word vector is obtained through a statistical algorithm. According to the occurrence frequency of each key character, the keyword frequency level is obtained and arranged in descending order; According to the arrangement result, the key characters with the highest keyword frequency level and the remaining characters in the control instruction word vector are arranged in forward or reverse order to form key character phrases, and the key character phrases that are the same as the control instruction text are placed in the first place to obtain the first keyword expansion phrase, and a front mapping is set in the first keyword expansion phrase; Furthermore, in this embodiment, the key characters are arranged in a forward or reverse order, specifically: Assume that the corresponding control instruction word vector is "turn on the water pump"; unpack "turn on the water pump" to obtain the four words "open", "start", "water" and "pump", and randomly arrange and combine these four words in forward or reverse order, such as "turn on", "start", "turn on the water pump", "water pump start", "turn on the pump", "pump start", etc., among which "turn on the water pump" is the same key character word group as the control instruction word vector, and is placed first in the keyword expansion phrase; the remaining words are combined with "turn on the water pump" to form the corresponding keyword expansion phrase.

[0025] Furthermore, the keyword expansion phrase in this embodiment is constructed from a plurality of keyword expansion words.

[0026] Furthermore, the step of setting the pre-mapping in this embodiment includes: A pre-mapping is established between the key character word group after the first one in the keyword expansion phrase and the first key character word group. When the key character word group found is any key character word group after the first one, the first key character word group is used for search feedback through the pre-mapping.

[0027] Remove the key characters with the highest keyword frequency level, repeat the process of obtaining the first keyword expansion phrase, and obtain the keyword expansion phrases corresponding to the remaining key characters; According to the keyword frequency level, all keyword expansion phrases corresponding to each control instruction text are stored in nodes through the prefix tree algorithm to obtain the standard vocabulary of the digital energy-saving pump.

[0028] Furthermore, the step of performing node storage by using a prefix tree algorithm in this embodiment includes: According to the keyword expansion phrase corresponding to each control instruction text, the key character is used as the root node, and according to the position information of the remaining characters in the keyword expansion word, the mounting sequence chain of the remaining character storage nodes is determined; According to the position information of the keyword in the keyword expansion word, a connection relationship between the key character and the node mounting sequence chain is established to obtain a storage node tree corresponding to the keyword expansion word; The key characters that are the same as the control instruction text are grouped into words and hot-coded as the standard hot-coding of the keyword expansion words; At the same time, the remaining words of the first keyword expansion phrase are adjusted according to the character position information in the corresponding composition, and the hot coding vector corresponding to each remaining word of the first keyword expansion phrase is obtained; According to the standard hot coding of the keyword expansion words and the hot coding vectors corresponding to each remaining group of words, a character position hot coding set corresponding to the first keyword expansion word group is obtained; Based on the process of obtaining the character position hot coding set corresponding to the first keyword expansion phrase, the character position hot coding set of the keyword expansion phrase corresponding to the remaining key characters is obtained, and each character position hot coding is embedded in the node mounting sequence chain of the corresponding keyword expansion word.

[0029] The process first determines the keyword frequency level through a statistical algorithm and constructs an extended phrase based on it, so that phrases related to high-frequency keywords have a higher priority when searching, reducing unnecessary search paths. The prefix tree algorithm uses the keyword as the root node and constructs a storage structure based on the character position information. It can quickly locate along the character path when searching, avoiding traversing the entire vocabulary. For example, when searching for instructions related to "startup", starting from the "start" node of the prefix tree, the relevant phrases are quickly found according to the character position mounting sequence chain; hot coding further encodes the character position features of the phrase, so that the system can more quickly determine the combination order of the phrase when searching. Combined with the pre-map, when encountering incomplete or inconsistent phrases, it can also quickly find the accurate match through the mapping relationship, thereby greatly speeding up the search speed.

[0030] The clear vocabulary structure in this process enables developers to intuitively understand how the vocabulary is organized; for example, when a new control instruction needs to be added, the new instruction and its expansion words can be reasonably integrated into the digital energy-saving pump standard vocabulary based on the existing keyword frequency calculation, expansion phrase construction, and prefix tree storage rules; the combination of character position hot encoding and node mounting sequence chain makes the characteristics of each character position in each expansion phrase more obviously different, making it easier and clearer to find the corresponding out-of-order expansion phrase when searching.

[0031] Furthermore, the step of constructing the digital energy-saving pump standard word library in this embodiment also includes: Obtain the historical voice sentences corresponding to the historical users' different dialects or multi-language control voices and the number of correct control command text searches; According to the historical voice sentences, the correct control instruction text and the search frequency, a character co-occurrence matrix in the historical voice sentences is obtained by combining the keyword expansion phrase through a text segmentation algorithm; The character co-occurrence matrix is ​​constructed by hot coding the character positions of the corresponding keyword expansion words in the historical speech sentences; Based on the character co-occurrence matrix, the frequencies of occurrence of historical keyword expansion phrases, colloquial keyword expansion phrases and corresponding character position hot codes are obtained, and the colloquial keyword expansion phrases and corresponding character position hot codes are added to the corresponding keyword expansion phrase set.

[0032] Furthermore, the colloquial keyword expansion phrase in this embodiment is constructed based on the non-standardized control instruction text obtained by analyzing each time a voice instruction is received, for example: In scenarios where different dialects and multiple languages ​​correspond: Considering that historical users have different dialects or multilingual control voices, real-time voice sentences may be expressed in various dialects or languages. For example, the command "start the water pump" may be expressed as "turn on the pump" in some dialects, or have corresponding expressions in other languages. At this time, the system needs to understand these diverse expressions based on the historical different dialects and multilingual voice data included in the construction of the digital energy-saving pump standard vocabulary.

[0033] In the scenario where colloquial and informal expressions correspond: In actual conversations, users may use colloquial, concise or non-standard expressions. For example, "make the pump faster" may be a colloquial way of saying "increase the speed of the water pump." The system needs to identify such non-standard instruction expressions through flexible matching of keyword expansion phrases, which requires that keyword expansion phrases not only include different permutations and combinations of standard instructions, but also cover some common colloquial expressions.

[0034] In complex contexts and scenarios corresponding to omissions: In actual conversations, users may give instructions in complex contexts or omit some information because the context is already known. For example, in a situation where a water pump has already been discussed, the user may only say "start". At this time, the system needs to combine the previous conversation history and the relevant information in the digital energy-saving pump standard vocabulary to accurately identify the complete control instruction as "start the water pump".

[0035] According to the frequency of occurrence of historical keyword expansion phrases and corresponding character position hot coding, the frequency of each keyword expansion word is obtained; The frequency of each keyword expansion word obtained is built into the corresponding keyword expansion word in the digital energy-saving pump standard word library; A frequency threshold is set. When the frequency of the corresponding keyword expansion word is greater than the frequency threshold, the corresponding keyword expansion word is mapped to the frequent word node configured in the digital energy-saving pump standard word library.

[0036] This process obtains the historical voice sentences corresponding to the control voices of different dialects or multilinguals of historical users and the number of correct control command text searches, combines the text segmentation algorithm with the keyword expansion phrase to build a character co-occurrence matrix, and then obtains the frequency of the keyword expansion words, which can make the digital energy-saving pump standard vocabulary better adapt to the language habits of different users, thereby improving the accuracy of command recognition. For example, the combination of specific words in some dialects is relatively fixed. By analyzing the character co-occurrence relationship in the historical voice sentences and building a character co-occurrence matrix, these language patterns can be captured. For example, in a certain regional dialect, the two characters "pump" and "open" often appear together in control instructions, and this relationship can be reflected by the character co-occurrence matrix. Based on this, the frequency of occurrence of the keyword expansion phrase and the corresponding character position hot encoding can be obtained, so that the frequency of occurrence of each keyword expansion word in actual use, that is, the frequency, can be more accurately understood. The frequency information is built into the digital energy-saving pump standard vocabulary. When recognizing real-time voice sentences, the system can give priority to keyword expansion words with high frequency, thereby improving the accuracy of command recognition. Secondly, in actual human-computer interaction, users tend to use colloquial expressions. For example, a user may say "turn on the pump" instead of the standard "start the water pump". By analyzing historical voice sentences, these colloquial keyword expansion phrases, such as "turn on the pump" and their corresponding character position hot encoding, are extracted and included in the vocabulary. In this way, when the system receives similar colloquial instructions, it can accurately identify them based on the information in the vocabulary, so that users do not need to deliberately use the standard instruction format, which improves the convenience and comfort of users and enhances the user interaction experience; in addition, when searching for control instructions, frequently used instructions should be able to be quickly located. By setting a frequency threshold, high-frequency keyword expansion words are stored in frequent word nodes. When the system receives real-time voice sentences for instruction matching, it first searches in frequent word nodes. Since the keyword expansion words in frequent word nodes are frequently used by users, matching items can be found faster, reducing search time and computing resource consumption. For example, if "start the water pump" is a frequently used instruction, after mapping it to a frequent word node, the system can prioritize searching in this node when identifying related instructions, greatly improving search efficiency. Finally, as time goes by and the number of users increases, users' language habits and commonly used instructions may change. This process continuously collects historical user voice data, continuously updates the character co-occurrence matrix, the frequency of keyword expansion words, and other information, and then adjusts the storage and mapping relationship of keyword expansion words in the digital energy-saving pump standard word library. For example, if a new colloquial expression is found to be gradually popular after a period of time, the system can analyze the new historical data, incorporate it into the word library, and optimize it accordingly, so that the word library can always adapt to the latest user needs.

[0037] Furthermore, the step of determining the control instruction in the conversation voice includes: According to the voice sentence acquired in real time, based on the process of acquiring the keyword expansion phrase, a real-time voice sentence keyword expansion phrase set is obtained; According to the keyword expansion phrase corresponding to each keyword frequency level in the real-time voice sentence keyword expansion phrase set, a cyclic parallel path search is performed through a multi-threaded search queue to obtain a real-time standard control instruction keyword expansion phrase set; Furthermore, the step of performing cyclic parallel path search through the multi-threaded search queue in this embodiment includes: According to the keyword expansion words in each keyword expansion phrase in the real-time voice sentence keyword expansion phrase set, construct each keyword expansion word search queue; According to the frequency of historical keyword expansion words, a keyword expansion word search queue containing a frequency greater than a frequency threshold is marked as a priority search to obtain a priority search queue; Based on the priority search queue, the node mounting sequence chain corresponding to the frequent word nodes and the infrequent words under the standard word library of the digital energy-saving pump is searched through parallel threads to obtain the first real-time standard control instruction keyword expansion phrase set; If the number of the first real-time standard control instruction keyword expansion phrase set is greater than one, a second search is performed on the first real-time standard control instruction keyword expansion phrase set through the remaining search queues in the keyword expansion phrase search queue to obtain the real-time standard control instruction keyword expansion phrase set.

[0038] According to the real-time standard control instruction keyword expansion phrase set combined with the digital energy-saving pump instruction library, the real-time standard control instruction text is obtained.

[0039] This process uses a multi-threaded search queue to perform cyclic parallel path search, which can significantly improve the efficiency of searching for control instructions, reduce the time required for searching, and enable the system to convert the user's conversational voice into standard control instructions more quickly; in the traditional search process, keyword expansion phrases are usually searched one by one in sequence, which is inefficient when processing large amounts of data. With the multi-threaded search queue, multiple search tasks can be performed simultaneously, and different keyword expansion phrases can be processed in parallel, making full use of the advantages of modern multi-core processors and speeding up the overall search speed.

[0040] In the above process, since the keyword expansion words with high frequency are more likely to be the control instructions that users want to express in actual use, putting them in the priority search queue can make the system prioritize these high-probability options; for example, when users often use instructions such as "start the water pump", the corresponding keyword expansion words have a high frequency. By prioritizing the search for this word, matching items can be found faster in the digital energy-saving pump standard vocabulary, reducing the search for low-frequency phrases and improving the search efficiency and accuracy. This priority search mechanism is based on the optimization of historical usage data. The system can allocate more computing resources to the most likely search path based on previous usage, reducing the waste of search time on uncommon or low-frequency phrases and improving the system's response speed to users' commonly used instructions.

[0041] In the above process, when multiple possible real-time standard control instruction keyword expansion phrase sets are found through the priority search queue, there may be multiple results that meet the conditions. At this time, through the secondary search, the remaining search queue can be further screened to eliminate mismatches or inaccurate matching results. For example, the first search may generate multiple candidate results due to the similarity of some keyword expansion words. The secondary search can further refine and screen by checking other keyword expansion words, and finally find the control instruction that best meets the user's intention, avoiding misjudgment caused by the limitations of a single search and improving the accuracy of instruction recognition.

[0042] S3. Execute the control instruction to obtain a first effective control feedback voice and a control feedback interface.

[0043] Furthermore, in this embodiment, the corresponding operation process of the digital energy-saving pump is controlled according to the control instruction, and the corresponding first effective control feedback voice is generated according to the operation result, and fed back to the issuer of the control instruction. At the same time, the corresponding operation process is displayed in real time through the built-in control feedback interface of the digital energy-saving pump, and the displayed result is kept consistent with the feedback content of the first effective control feedback voice.

[0044] Furthermore, the digital energy-saving pump in this embodiment activates the human-machine dialogue interface under normal operation, checks the interface, icons, display and operation status, and interacts with the issuer of the corresponding control instruction.

[0045] Example 2 See also Figure 2 , another embodiment provided by the present invention: a human-machine dialogue system for a digital energy-saving pump, comprising: an acquisition module, an instruction generation module and a feedback module; An acquisition module, used to generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; The instruction generation module is used to determine the control instructions in the dialogue voice by combining a preset digital energy-saving pump standard vocabulary library and a digital energy-saving pump instruction library.

[0046] The feedback module obtains a first effective control feedback voice and a control feedback interface based on the control instruction in the dialogue voice.

[0047] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the scope of protection of the purpose of the present invention and the claims, and all of these are within the protection of the present invention.

[0048] If the disclosed technical solution involves personal information, the product using the disclosed technical solution has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the disclosed technical solution involves sensitive personal information, the product using the disclosed technical solution has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, when the personal information processing rules are notified by obvious signs / information, the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A human-computer dialogue method for a digital energy-saving pump, characterized in that: include: Generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; Determine the control instructions in the dialogue voice in combination with a preset digital energy-saving pump standard word library and a digital energy-saving pump instruction library; Execute the control instruction, wherein the digital energy-saving pump standard word library includes all words in the digital energy-saving pump instruction library and extended words of all words, and the extended words include all disorderly arranged words of the words.

2. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 1, characterized in that: The steps of constructing the digital energy-saving pump standard word library include: Based on all the control instruction texts in the digital energy-saving pump instruction library, the control instruction word vector is obtained through the word segmentation algorithm; Based on the control instruction word vector, the occurrence frequency of each key character in the effective text word vector is obtained through a statistical algorithm. According to the occurrence frequency of each key character, the keyword frequency level is obtained and arranged in descending order; According to the arrangement result, the key characters with the highest keyword frequency level and the remaining characters in the control instruction word vector are arranged in forward or reverse order to form key character phrases, and the key character phrases that are the same as the control instruction text are placed in the first place to obtain the first keyword expansion phrase, and a front mapping is set in the first keyword expansion phrase; Remove the key characters with the highest keyword frequency level, repeat the process of obtaining the first keyword expansion phrase, and obtain the keyword expansion phrases corresponding to the remaining key characters; According to the keyword frequency level, all keyword expansion phrases corresponding to each control instruction text are stored in nodes through the prefix tree algorithm to obtain the standard vocabulary of the digital energy-saving pump.

3. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 2, characterized in that: The step of setting the pre-mapping comprises: A pre-mapping is established between the key character word group after the first one in the keyword expansion phrase and the first key character word group. When the key character word group found is any key character word group after the first one, the first key character word group is used for search feedback through the pre-mapping.

4. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 3, characterized in that: The step of performing node storage by using a prefix tree algorithm includes: According to the keyword expansion phrase corresponding to each control instruction text, the key character is used as the root node, and according to the position information of the remaining characters in the keyword expansion word, the mounting sequence chain of the remaining character storage nodes is determined; According to the position information of the keyword in the keyword expansion word, a connection relationship between the key character and the node mounting sequence chain is established to obtain a storage node tree corresponding to the keyword expansion word.

5. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 4, characterized in that: The step of performing node storage by prefix tree algorithm also includes: The key characters that are the same as the control instruction text are grouped into words and hot-coded as the standard hot-coding of the keyword expansion words; At the same time, the remaining words of the first keyword expansion phrase are adjusted according to the character position information in the corresponding composition, and the hot coding vector corresponding to each remaining word of the first keyword expansion phrase is obtained; According to the standard hot coding of the keyword expansion words and the hot coding vectors corresponding to each remaining group of words, a character position hot coding set corresponding to the first keyword expansion word group is obtained; Based on the process of obtaining the character position hot coding set corresponding to the first keyword expansion phrase, the character position hot coding set of the keyword expansion phrase corresponding to the remaining key characters is obtained, and each character position hot coding is embedded in the node mounting sequence chain of the corresponding keyword expansion word.

6. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 5, characterized in that: The step of constructing the digital energy-saving pump standard word library also includes: Obtain the historical voice sentences corresponding to the historical users' different dialects or multi-language control voices and the number of correct control command text searches; According to the historical voice sentences, the correct control instruction text and the search frequency, a character co-occurrence matrix in the historical voice sentences is obtained by combining the keyword expansion phrase through a text segmentation algorithm; The character co-occurrence matrix is ​​constructed by hot coding the character positions of the corresponding keyword expansion words in the historical speech sentences; Based on the character co-occurrence matrix, the frequencies of occurrence of historical keyword expansion phrases, colloquial keyword expansion phrases and corresponding character position hot codes are obtained, and the colloquial keyword expansion phrases and corresponding character position hot codes are added to the corresponding keyword expansion phrase set.

7. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 6, characterized in that: The step of constructing the digital energy-saving pump standard word library also includes: According to the frequency of occurrence of historical keyword expansion phrases and corresponding character position hot coding, the frequency of each keyword expansion word is obtained; The frequency of each keyword expansion word obtained is built into the corresponding keyword expansion word in the digital energy-saving pump standard word library; A frequency threshold is set. When the frequency of the corresponding keyword expansion word is greater than the frequency threshold, the corresponding keyword expansion word is mapped to the frequent word node configured in the digital energy-saving pump standard word library.

8. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 7, characterized in that: The step of determining the control instruction in the conversation speech comprises: According to the voice sentence acquired in real time, based on the process of acquiring the keyword expansion phrase, a real-time voice sentence keyword expansion phrase set is obtained; According to the keyword expansion phrase corresponding to each keyword frequency level in the real-time voice sentence keyword expansion phrase set, a cyclic parallel path search is performed through a multi-threaded search queue to obtain a real-time standard control instruction keyword expansion phrase set; According to the real-time standard control instruction keyword expansion phrase set combined with the digital energy-saving pump instruction library, the real-time standard control instruction text is obtained.

9. A human-computer dialogue method for a digital energy-saving pump as claimed in claim 8, characterized in that: The step of performing cyclic parallel path search through a multi-threaded search queue comprises: According to the keyword expansion words in each keyword expansion phrase in the real-time voice sentence keyword expansion phrase set, construct each keyword expansion word search queue; According to the frequency of historical keyword expansion words, a keyword expansion word search queue containing a frequency greater than a frequency threshold is marked as a priority search to obtain a priority search queue; Based on the priority search queue, the node mounting sequence chain corresponding to the frequent word nodes and the infrequent words under the standard word library of the digital energy-saving pump is searched through parallel threads to obtain the first real-time standard control instruction keyword expansion phrase set; If the number of the first real-time standard control instruction keyword expansion phrase set is greater than one, a second search is performed on the first real-time standard control instruction keyword expansion phrase set through the remaining search queues in the keyword expansion phrase search queue to obtain the real-time standard control instruction keyword expansion phrase set.

10. A human-machine dialogue system for a digital energy-saving pump, which is used to implement a human-machine dialogue method for a digital energy-saving pump according to any one of claims 1 to 9, characterized in that: include: Acquisition module and instruction generation module; The acquisition module is used to generate a voice sentence according to the conversation voice input by the user; the voice sentence includes a plurality of words; The instruction generation module is used to determine the control instructions in the dialogue voice by combining a preset digital energy-saving pump standard vocabulary library and a digital energy-saving pump instruction library.

Images