Household equipment control method and system

Abstract

The invention discloses a household equipment control method and system. The method comprises the following steps: acquiring a household equipment control text output by an LLM model; recognizing an equipment object text and an instruction related text in the home equipment control text based on a text recognition algorithm; according to an equipment protocol rule corresponding to a current home scene, determining a home equipment protocol corresponding to the equipment object text; and according to the household equipment protocol and the instruction related text, analyzing and generating an equipment control instruction, and sending the equipment control instruction to corresponding household equipment for execution. Therefore, full-link intelligent conversion from natural language to precise equipment control can be realized, the accuracy and compatibility of home equipment control are improved, and the equipment misoperation risk caused by text analysis errors is reduced.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for controlling home appliances. Background Technology

[0002] With the rapid application of intelligent algorithms in the smart home field, users are increasingly valuing precise control of all devices throughout the house via natural language. A key technical challenge is accurately converting the unstructured control text output by LLM (Local Language Manager) into executable device commands to avoid misoperation. Existing technologies typically output fixed-format commands directly from the LLM or rely on single predefined rules to parse natural language, using simple keyword matching or limited protocol mapping to generate control commands for voice or text control. However, existing solutions lack fine-grained recognition of device objects and command content in the LLM output text, as well as dynamic parsing of current home scenario protocol rules. This makes it difficult to achieve high-accuracy end-to-end command conversion in complex environments with multiple devices and protocols. Commonly used templated or coarse-grained parsing strategies cannot adapt to the diverse expressions of LLM output, resulting in insufficient accuracy and compatibility in device control. Errors in text parsing can easily lead to misoperation or command failure, limiting the reliability and practical usability of natural language control. Therefore, existing technologies have shortcomings that urgently need to be addressed. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a home appliance control method and system that can realize the intelligent conversion of natural language into precise device control, improve the accuracy and compatibility of home appliance control, and reduce the risk of device misoperation caused by text parsing errors.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a method for controlling home appliances, the method comprising: Obtain the home appliance control text output by the LLM model; Based on the text recognition algorithm, the device object text and instruction-related text in the home device control text are identified; Based on the device protocol rules corresponding to the current home scenario, determine the home device protocol corresponding to the device object text; Based on the home appliance protocol and the related instruction text, device control instructions are parsed and generated, and then sent to the corresponding home appliance for execution.

[0005] As an optional implementation, in the first aspect of the present invention, the step of identifying the device object text and instruction-related text in the home appliance control text based on the text recognition algorithm includes: Based on a text type classifier, determine whether the home appliance control text belongs to structured text or natural language text; If the home appliance control text is structured text, the device object text and instruction-related text corresponding to the home appliance control text are determined based on the preset structured tag recognition rules; If the home appliance control text is natural language text, the device object text and instruction-related text in the home appliance control text are matched and determined based on the template matching algorithm.

[0006] As an optional implementation, in the first aspect of the invention, the step of matching and determining the device object text and instruction-related text in the home appliance control text based on the template matching algorithm includes: The control text for the home appliances is segmented into multiple text words; For each arbitrary combination of multiple adjacent text words, based on multiple historical text templates, determine a first probability that the text combination belongs to device object text and a second probability that it belongs to instruction-related text. The text combination with the highest probability is determined as the device object text; The text combination with the highest probability is identified as the instruction-related text.

[0007] As an optional implementation, in the first aspect of the invention, determining the first probability that the text combination belongs to device object text and the second probability that it belongs to instruction-related text based on a plurality of historical text templates includes: Calculate the average text similarity between the text combination and the text template of each historical device object to obtain the first probability; The average text similarity between the text combination and each historical instruction-related text template is calculated to obtain the second probability; the text similarity is obtained by predicting the input text pairs through a trained home appliance control text similarity prediction model; the home appliance control text similarity prediction model is trained through a training dataset of multiple training text pairs for home appliance control and corresponding text similarity annotations.

[0008] As an alternative implementation, in a first aspect of the invention, the home appliance control text is obtained by the LLM model after receiving and processing the user's input home control requirements.

[0009] As an optional implementation, in the first aspect of the present invention, determining the home device protocol corresponding to the device object text according to the device protocol rules corresponding to the current home scenario includes: Based on the home control requirements and the user's user parameters, determine the current home scene and the corresponding home device area; Based on the current home scene and the corresponding home device area, multiple corresponding device protocol rules are matched from the preset device protocol database; Based on the device similarity algorithm, the home device protocol corresponding to the device object text is determined according to the device protocol rules.

[0010] As an optional implementation, in the first aspect of the present invention, determining the home device protocol corresponding to the device object text based on the device similarity algorithm and according to the device protocol rules includes: For each device protocol rule, obtain multiple application device descriptions from the historical application records of that device protocol rule; Calculate the average similarity between the device object text and the text description of each application device to obtain the similarity parameter corresponding to the device protocol rule; The device protocol rule with the highest similarity parameter is determined as the home device protocol corresponding to the device object text.

[0011] As an optional implementation, in the first aspect of the present invention, the step of parsing and generating device control instructions based on the home appliance protocol and the instruction-related text, and sending them to the corresponding home appliance for execution, includes: The home device protocol is input into the trained protocol formatting instruction prediction model to obtain the corresponding multiple instruction structure parts; For each instruction structure part, based on the preset fill-in prompt word rules, the instruction structure part and the instruction-related text are combined to form fill-in prompt words; The fill prompt is input into the LLM model to obtain the fill instruction corresponding to the instruction structure part; Combine the fill instructions corresponding to all the instruction structure parts to obtain the device control instructions and send them to the corresponding home appliances for execution.

[0012] A second aspect of this invention discloses a home appliance control system, the system comprising: The acquisition module is used to acquire the home appliance control text output by the LLM model; The recognition module is used to identify the device object text and instruction-related text in the home device control text based on a text recognition algorithm; The determination module is used to determine the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scene; The parsing module is used to parse and generate device control commands based on the home device protocol and the instruction-related text, and send them to the corresponding home device for execution.

[0013] As an optional implementation, in a second aspect of the invention, the identification module identifies the device object text and instruction-related text in the home appliance control text using a text recognition algorithm in the following specific ways: Based on a text type classifier, determine whether the home appliance control text belongs to structured text or natural language text; If the home appliance control text is structured text, the device object text and instruction-related text corresponding to the home appliance control text are determined based on the preset structured tag recognition rules; If the home appliance control text is natural language text, the device object text and instruction-related text in the home appliance control text are matched and determined based on the template matching algorithm.

[0014] As an optional implementation, in a second aspect of the invention, the identification module, based on a template matching algorithm, determines the specific method by which it matches and determines the device object text and instruction-related text in the home appliance control text, including: The control text for the home appliances is segmented into multiple text words; For each arbitrary combination of multiple adjacent text words, based on multiple historical text templates, determine a first probability that the text combination belongs to device object text and a second probability that it belongs to instruction-related text. The text combination with the highest probability is determined as the device object text; The text combination with the highest probability is identified as the instruction-related text.

[0015] As an optional implementation, in a second aspect of the invention, the specific method by which the recognition module determines, based on multiple historical text templates, the first probability that the text combination belongs to device object text and the second probability that it belongs to instruction-related text includes: Calculate the average text similarity between the text combination and the text template of each historical device object to obtain the first probability; The average text similarity between the text combination and each historical instruction-related text template is calculated to obtain the second probability; the text similarity is obtained by predicting the input text pairs through a trained home appliance control text similarity prediction model; the home appliance control text similarity prediction model is trained through a training dataset of multiple training text pairs for home appliance control and corresponding text similarity annotations.

[0016] As an alternative implementation, in a second aspect of the invention, the home appliance control text is obtained by the LLM model after receiving and processing the user's input home control requirements.

[0017] As an optional implementation, in a second aspect of the invention, the determining module determines the specific method of the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scenario, including: Based on the home control requirements and the user's user parameters, determine the current home scene and the corresponding home device area; Based on the current home scene and the corresponding home device area, multiple corresponding device protocol rules are matched from the preset device protocol database; Based on the device similarity algorithm, the home device protocol corresponding to the device object text is determined according to the device protocol rules.

[0018] As an optional implementation, in a second aspect of the invention, the determining module determines the specific method of the home device protocol corresponding to the device object text based on a device similarity algorithm and according to the device protocol rules, including: For each device protocol rule, obtain multiple application device descriptions from the historical application records of that device protocol rule; Calculate the average similarity between the device object text and the text description of each application device to obtain the similarity parameter corresponding to the device protocol rule; The device protocol rule with the highest similarity parameter is determined as the home device protocol corresponding to the device object text.

[0019] As an optional implementation, in a second aspect of the present invention, the specific method by which the parsing module parses and generates device control instructions based on the home appliance protocol and the instruction-related text, and sends them to the corresponding home appliance for execution, includes: The home device protocol is input into the trained protocol formatting instruction prediction model to obtain the corresponding multiple instruction structure parts; For each instruction structure part, based on the preset fill-in prompt word rules, the instruction structure part and the instruction-related text are combined to form fill-in prompt words; The fill prompt is input into the LLM model to obtain the fill instruction corresponding to the instruction structure part; Combine the fill instructions corresponding to all the instruction structure parts to obtain the device control instructions and send them to the corresponding home appliances for execution.

[0020] A third aspect of the present invention discloses another home appliance control system, the system comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the home appliance control method disclosed in the first aspect of the present invention.

[0021] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the home appliance control method disclosed in the first aspect of the present invention.

[0022] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: This invention acquires the home device control text output by the LLM model and extracts the device object text and instruction-related text based on the text recognition algorithm. It then combines the current home scenario device protocol rules to parse and generate device control instructions and send them for execution. This enables the end-to-end intelligent conversion from natural language to precise device control, improving the accuracy and compatibility of home device control and reducing the risk of device misoperation due to text parsing errors. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a home appliance control method disclosed in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of the structure of a home appliance control system disclosed in an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of another home appliance control system disclosed in an embodiment of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0029] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0030] This invention discloses a home appliance control method and system. By acquiring the home appliance control text output by an LLM model and extracting device object text and instruction-related text based on a text recognition algorithm, and then parsing and generating device control instructions according to the current home scenario device protocol rules, the method achieves end-to-end intelligent conversion from natural language to precise device control. This improves the accuracy and compatibility of home appliance control and reduces the risk of device misoperation due to text parsing errors. Detailed explanations follow.

[0031] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a home appliance control method disclosed in an embodiment of the present invention. Figure 1 The described home appliance control method can be applied to data processing systems / data processing devices / data processing servers (wherein the server includes local processing servers or cloud processing servers). For example... Figure 1 As shown, the home appliance control method may include the following operations: 101. Obtain the home appliance control text output by the LLM model.

[0032] Optionally, the home appliance control text is received and processed by the LLM model after the user inputs the home control requirements, and then output as the output.

[0033] Optionally, the LLM model can be a home control-specific model finely tuned from Qwen-14B-Chat, ChatGLM3-6B, or LLaMA-2-13B, and this invention does not impose any limitations on it.

[0034] Optionally, the control text for the home appliance can be natural language or structured instructions such as "adjust the living room light to 50% warm light", "turn off the bedroom air conditioner", or "turn on the kitchen air purifier in high power mode". This invention does not limit this.

[0035] 102. Based on text recognition algorithms, identify device object text and instruction-related text in home appliance control text. 103. Based on the device protocol rules corresponding to the current home scene, determine the home device protocol corresponding to the device object text. 104. Based on the home appliance protocols and related instruction texts, parse and generate device control instructions and send them to the corresponding home appliances for execution.

[0036] As can be seen, the above-described embodiments of the invention obtain the home device control text output by the LLM model and extract the device object text and instruction-related text based on the text recognition algorithm. Combined with the current home scene device protocol rules, they generate device control instructions and send them for execution. This enables the end-to-end intelligent conversion from natural language to precise device control, improves the accuracy and compatibility of home device control, and reduces the risk of device misoperation caused by text parsing errors.

[0037] As an optional embodiment, the step described above, which involves identifying device object text and instruction-related text in home appliance control text based on a text recognition algorithm, includes: Based on a text type classifier, determine whether the control text for home appliances belongs to structured text or natural language text; If the home appliance control text is structured text, the device object text and instruction-related text corresponding to the home appliance control text are determined based on the preset structured tag recognition rules; If the home appliance control text is natural language text, the device object text and instruction-related text in the home appliance control text are matched and determined based on the template matching algorithm.

[0038] Optionally, the text type classifier can be a 2-layer BiLSTM (256 units) + fully connected classification head, trained on 100,000 labeled texts with an accuracy of 99.2%, but this invention does not limit it.

[0039] Optionally, the structured tag recognition rule can be a regular expression matching "[device:xxx][command:xxx]" or JSON key-value pair extraction, and this invention does not limit it.

[0040] Optionally, the template matching algorithm can be a multi-template cosine similarity Top-K matching based on Sentence-BERT plus slot filling, which is not limited in this invention.

[0041] As can be seen, through the above optional embodiments, the text type classifier determines whether the control text is structured or natural language, and the structured label recognition rules and template matching algorithms are used respectively to determine the device object text and instruction-related text, thereby achieving accurate bimodal parsing of different types of LLM output text, improving the robustness and universality of text recognition, and reducing the risk of parsing failure due to differences in text format.

[0042] As an optional embodiment, the step described above, which involves matching and determining the device object text and instruction-related text in the home appliance control text based on a template matching algorithm, includes: The control text for home appliances is segmented into multiple text words; For each arbitrary combination of multiple adjacent text words, based on multiple historical text templates, determine the first probability that the text combination belongs to device object text and the second probability that it belongs to instruction-related text. The text combination with the highest probability is identified as the device object text; The text combination with the second highest probability is identified as the instruction-related text.

[0043] Optionally, the word segmentation can be performed using Jieba Chinese word segmentation with stop word filtering or BERT-Tokenizer word segmentation; this invention does not impose any limitations on this method.

[0044] Optionally, the text combination uses a sliding window (window size 3-8 words), which is not limited in this invention.

[0045] As can be seen, through the above optional embodiments, by segmenting the text into words and determining the probability that the text combination belongs to the device object text and the probability that it belongs to the instruction-related text based on multiple historical text templates, the device object text and instruction-related text are determined, thereby achieving accurate text role positioning based on historical template matching, improving the accuracy and interpretability of natural language control text parsing, and reducing the risk of key information extraction errors caused by misjudgment of combinations.

[0046] As an optional embodiment, the step described above, determining the first probability that the text combination belongs to device object text and the second probability that it belongs to instruction-related text based on multiple historical text templates, includes: Calculate the average text similarity between the text combination and the text template of each historical device object to obtain the first probability; The average text similarity between the text combination and the text template related to each historical instruction is calculated to obtain the second probability.

[0047] Optionally, text similarity is obtained by predicting the input text pairs using a pre-trained home appliance control text similarity prediction model.

[0048] Optionally, the home appliance control text similarity prediction model is trained using a training dataset of multiple training text pairs for home appliance control and corresponding text similarity annotations.

[0049] Optionally, the home appliance control text similarity prediction model can be a RoBERTa-base Chinese version + Siamese network structure, with cosine similarity regression after dual-tower encoding, trained on 150,000 similar pairs of home appliance instructions, with MSE < 0.012. This invention does not limit the model.

[0050] Optionally, the historical text template is derived from the user's real command database of the past 30 days, but this invention does not limit it.

[0051] As can be seen, through the above optional embodiments, the device object text and instruction-related text are determined by calculating the average similarity between adjacent text combinations and historical device object / instruction templates, thereby achieving accurate text role positioning based on historical template matching, improving the accuracy and interpretability of natural language control text parsing, and reducing the risk of key information extraction errors caused by misjudgment of combinations.

[0052] As an optional embodiment, the step above, determining the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scenario, includes: Based on home control needs and user parameters, determine the current home scene and corresponding home device areas; Based on the current home scene and the corresponding home device area, multiple corresponding device protocol rules are matched from the preset device protocol database; Based on the device similarity algorithm, the home device protocol corresponding to the device object text is determined according to the device protocol rules.

[0053] Optionally, the user parameters may include a home floor plan, device registry, and user preferences (such as the common term "lamp" referring to the living room lamp), but this invention does not limit these parameters.

[0054] Optionally, the device protocol can be ZigBee HA, Tuya IR, Yeelight LAN, or Xiaomi Mijia protocol; this invention does not limit the specific protocol.

[0055] Optionally, the device similarity algorithm uses multi-field cosine similarity of device description text and protocol metadata, which is not limited in this invention.

[0056] As can be seen, through the above optional embodiments, by matching the corresponding device protocol rules according to the current home scene and user parameters, and determining the corresponding protocol of the device object text based on device similarity, accurate protocol selection based on scene adaptation is achieved, improving the compatibility of different home environments and devices, and reducing the risk of instruction parsing failure due to protocol mismatch.

[0057] As an optional embodiment, the step above, determining the home device protocol corresponding to the device object text based on the device similarity algorithm and according to device protocol rules, includes: For each device protocol rule, retrieve multiple application device descriptions from the historical application records of that device protocol rule; Calculate the average similarity between the device object text and the text description of each application device to obtain the similarity parameter corresponding to the device protocol rule; The device protocol rule with the highest similarity parameter is identified as the home device protocol corresponding to the device object text.

[0058] As can be seen, through the above optional embodiments, the optimal device protocol rule is determined by calculating the average similarity between the device object text and the historical application device description, thereby achieving accurate protocol matching based on historical usage experience, improving the pertinence and reliability of protocol selection, providing a correct protocol basis for subsequent instruction parsing, and reducing the risk of invalid control instructions due to incorrect protocol selection.

[0059] As an optional embodiment, the above steps, including parsing and generating device control commands based on home appliance protocols and instruction-related text, and sending them to the corresponding home appliances for execution, include: The home appliance protocol is input into the trained protocol formatting instruction prediction model to obtain the corresponding multiple instruction structure parts; For each instruction structure part, based on the preset fill-in prompt word rules, the instruction structure part and the instruction-related text are combined to form fill-in prompt words; Input the fill prompt into the LLM model to obtain the fill instruction corresponding to the instruction structure part; Combine the fill instructions corresponding to all instruction structure parts to obtain device control instructions and send them to the corresponding home appliances for execution.

[0060] Optionally, the protocol formatting instruction prediction model can be a T5-large model, which is fine-tuned by SFT of 10k protocol → structured instruction pairs to output a JSON-formatted instruction skeleton. This invention does not limit this model.

[0061] Optionally, the rule for filling in the prompt words is "based on the user's intent '{instruction-related text}', fill in the following Yeelight protocol fields: {structural part}", which is not limited in this invention.

[0062] Optionally, the combined device control command is a fully executable JSON or binary command, which is issued through the corresponding gateway. This invention does not impose any limitations on this.

[0063] As can be seen, through the above optional embodiments, by inputting the home device protocol into the protocol formatting instruction prediction model to generate the instruction structure part, and then combining the instruction-related text to assemble and fill in the prompt words into the LLM model to complete the instruction filling and combine to generate the final device control instruction, the accurate instruction structure generation based on protocol guidance is realized, improving the standardization and execution success rate of control instructions, and reducing the risk of device refusal to execute due to instruction format inconsistencies.

[0064] The solution in the embodiments of the present invention is illustrated by a specific example: At 20:15 on August 10, 2025, Mr. Wang said, "Turn the bedroom light to warm light and lower the brightness."

[0065] The LLM model outputs the text: "Turn the bedroom lights to warm light and lower the brightness."

[0066] The text type classifier identifies it as natural language text.

[0067] Sliding window + template matching, identifies: Device object text: "Bedroom light"; Command text: "Adjust to warm light, lower the brightness"; Based on Mr. Wang's home floor plan, the system confirmed that the "bedroom light" corresponds to the Yeelight bedroom ceiling light (device ID: 0x12345678). The protocol library shows that Yeelight LAN protocol has the highest similarity score of 0.96, so this protocol is selected.

[0068] Protocol formatting instruction prediction model output structure: { "id": 1, "method": "set_ct_abx", "params": [COLOR_TEMP, "smooth", 500], "method2": "set_bright", "params2": [BRIGHTNESS, "smooth", 500] } The system maps "warm light" to 3000K and "lower brightness" to 35% based on user history preferences. After assembling and filling the prompt words, it calls LLM to fill again, and finally generates a complete instruction and sends it out through the Yeelight gateway.

[0069] Three seconds later, the bedroom light switched to a soft 3000K warm light at 35% brightness. Mr. Wang did not need to mention the "Yeelight" brand or specific parameters to achieve precise control.

[0070] If the user then says, "Make it dimmer," the system will generate an incremental command of -15% brightness based on context memory, thus achieving continuous and natural dialogue control.

[0071] Example 2 Please see Figure 2 , Figure 2 This is a schematic diagram of a home appliance control system disclosed in an embodiment of the present invention. Figure 2 The described home appliance control system can be applied to data processing systems / data processing devices / data processing servers (wherein, the server includes local processing servers or cloud processing servers). For example... Figure 2 As shown, the home appliance control system may include: Module 201 is used to acquire the home appliance control text output by the LLM model.

[0072] The recognition module 202 is used to identify device object text and instruction-related text in home device control text based on text recognition algorithms. The determination module 203 is used to determine the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scene. The parsing module 204 is used to parse and generate device control commands based on the home device protocol and instruction-related text, and send them to the corresponding home device for execution.

[0073] As can be seen, the above-described embodiments of the invention obtain the home device control text output by the LLM model and extract the device object text and instruction-related text based on the text recognition algorithm. Combined with the current home scene device protocol rules, they generate device control instructions and send them for execution. This enables the end-to-end intelligent conversion from natural language to precise device control, improves the accuracy and compatibility of home device control, and reduces the risk of device misoperation caused by text parsing errors.

[0074] As an optional embodiment, the recognition module, based on a text recognition algorithm, identifies the specific methods by which it identifies device object text and instruction-related text within home appliance control text, including: Based on a text type classifier, determine whether the control text for home appliances belongs to structured text or natural language text; If the home appliance control text is structured text, the device object text and instruction-related text corresponding to the home appliance control text are determined based on the preset structured tag recognition rules; If the home appliance control text is natural language text, the device object text and instruction-related text in the home appliance control text are matched and determined based on the template matching algorithm.

[0075] As can be seen, through the above optional embodiments, the text type classifier determines whether the control text is structured or natural language, and the structured label recognition rules and template matching algorithms are used respectively to determine the device object text and instruction-related text, thereby achieving accurate bimodal parsing of different types of LLM output text, improving the robustness and universality of text recognition, and reducing the risk of parsing failure due to differences in text format.

[0076] As an optional embodiment, the recognition module uses a template matching algorithm to determine the specific method of matching and identifying device object text and instruction-related text in home device control text, including: The control text for home appliances is segmented into multiple text words; For each arbitrary combination of multiple adjacent text words, based on multiple historical text templates, determine the first probability that the text combination belongs to device object text and the second probability that it belongs to instruction-related text. The text combination with the highest probability is identified as the device object text; The text combination with the second highest probability is identified as the instruction-related text.

[0077] As can be seen, through the above optional embodiments, by segmenting the text into words and determining the probability that the text combination belongs to the device object text and the probability that it belongs to the instruction-related text based on multiple historical text templates, the device object text and instruction-related text are determined, thereby achieving accurate text role positioning based on historical template matching, improving the accuracy and interpretability of natural language control text parsing, and reducing the risk of key information extraction errors caused by misjudgment of combinations.

[0078] As an optional embodiment, the identification module determines, based on multiple historical text templates, the specific method for determining a first probability that the text combination belongs to device object text and a second probability that it belongs to instruction-related text, including: Calculate the average text similarity between the text combination and the text template of each historical device object to obtain the first probability; The average text similarity between the text combination and each historical instruction-related text template is calculated to obtain the second probability; optionally, the text similarity is obtained by predicting the input text pairs using a pre-trained home appliance control text similarity prediction model; the home appliance control text similarity prediction model is trained using a training dataset of multiple training text pairs for home appliance control and corresponding text similarity annotations.

[0079] As can be seen, through the above optional embodiments, the device object text and instruction-related text are determined by calculating the average similarity between adjacent text combinations and historical device object / instruction templates, thereby achieving accurate text role positioning based on historical template matching, improving the accuracy and interpretability of natural language control text parsing, and reducing the risk of key information extraction errors caused by misjudgment of combinations.

[0080] As an optional embodiment, the home appliance control text is received by the LLM model and processed after the user inputs the home control requirements, and then output as the output.

[0081] As can be seen, the above optional embodiments limit the details of how home device control text is recognized and output by the LLM model, so as to realize the user's automatic control of home devices through the LLM model, assist in the full-link intelligent conversion from natural language to precise device control, improve the accuracy and compatibility of home device control, and reduce the risk of device misoperation caused by text parsing errors.

[0082] As an optional embodiment, the determining module determines the specific method of the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scenario, including: Based on home control needs and user parameters, determine the current home scene and corresponding home device areas; Based on the current home scene and the corresponding home device area, multiple corresponding device protocol rules are matched from the preset device protocol database; Based on the device similarity algorithm, the home device protocol corresponding to the device object text is determined according to the device protocol rules.

[0083] As can be seen, through the above optional embodiments, by matching the corresponding device protocol rules according to the current home scene and user parameters, and determining the corresponding protocol of the device object text based on device similarity, accurate protocol selection based on scene adaptation is achieved, improving the compatibility of different home environments and devices, and reducing the risk of instruction parsing failure due to protocol mismatch.

[0084] As an optional embodiment, the determining module determines the specific method of the home device protocol corresponding to the device object text based on the device similarity algorithm and according to the device protocol rules, including: For each device protocol rule, retrieve multiple application device descriptions from the historical application records of that device protocol rule; Calculate the average similarity between the device object text and the text description of each application device to obtain the similarity parameter corresponding to the device protocol rule; The device protocol rule with the highest similarity parameter is identified as the home device protocol corresponding to the device object text.

[0085] As can be seen, through the above optional embodiments, the optimal device protocol rule is determined by calculating the average similarity between the device object text and the historical application device description, thereby achieving accurate protocol matching based on historical usage experience, improving the pertinence and reliability of protocol selection, providing a correct protocol basis for subsequent instruction parsing, and reducing the risk of invalid control instructions due to incorrect protocol selection.

[0086] As an optional embodiment, the parsing module parses and generates device control commands based on home appliance protocols and instruction-related text, and sends them to the corresponding home appliances for execution in a specific manner, including: The home appliance protocol is input into the trained protocol formatting instruction prediction model to obtain the corresponding multiple instruction structure parts; For each instruction structure part, based on the preset fill-in prompt word rules, the instruction structure part and the instruction-related text are combined to form fill-in prompt words; Input the fill prompt into the LLM model to obtain the fill instruction corresponding to the instruction structure part; Combine the fill instructions corresponding to all instruction structure parts to obtain device control instructions and send them to the corresponding home appliances for execution.

[0087] As can be seen, through the above optional embodiments, by inputting the home device protocol into the protocol formatting instruction prediction model to generate the instruction structure part, and then combining the instruction-related text to assemble and fill in the prompt words into the LLM model to complete the instruction filling and combine to generate the final device control instruction, the accurate instruction structure generation based on protocol guidance is realized, improving the standardization and execution success rate of control instructions, and reducing the risk of device refusal to execute due to instruction format inconsistencies.

[0088] Example 3 Please see Figure 3 , Figure 3 This is yet another home appliance control system disclosed in the embodiments of the present invention. Figure 3 The described home appliance control system is used in a data processing system / data processing device / data processing server (wherein, the server includes a local processing server or a cloud processing server). For example... Figure 3 As shown, the home appliance control system may include: Memory 301 storing executable program code; Processor 302 coupled to memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute the steps of the home appliance control method described in Embodiment 1.

[0089] Example 4 This invention discloses a computer read storage medium that stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps of the home appliance control method described in Embodiment 1.

[0090] Example 5 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the home appliance control method described in Embodiment 1.

[0091] The foregoing has described specific embodiments of this specification; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0093] For ease of description, the above devices are described in terms of function, divided into various units. Of course, in implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware components.

[0094] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0096] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0098] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0099] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0100] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0101] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0102] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0103] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0104] Finally, it should be noted that the home appliance control method and system disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for controlling home appliances, characterized in that, The method includes: Obtain the home appliance control text output by the LLM model; Based on the text recognition algorithm, the device object text and instruction-related text in the home device control text are identified; Based on the device protocol rules corresponding to the current home scenario, determine the home device protocol corresponding to the device object text; Based on the home appliance protocol and the related instruction text, device control instructions are parsed and generated, and then sent to the corresponding home appliance for execution.

2. The home appliance control method according to claim 1, characterized in that, The text recognition algorithm identifies device object text and instruction-related text in the home appliance control text, including: Based on a text type classifier, determine whether the home appliance control text belongs to structured text or natural language text; If the home appliance control text is structured text, the device object text and instruction-related text corresponding to the home appliance control text are determined based on the preset structured tag recognition rules. If the home appliance control text is natural language text, the device object text and instruction-related text in the home appliance control text are matched and determined based on the template matching algorithm.

3. The home appliance control method according to claim 2, characterized in that, The template matching algorithm, which matches and determines the device object text and instruction-related text in the home appliance control text, includes: The control text for the home appliances is segmented into multiple text words; For each arbitrary combination of multiple adjacent text words, based on multiple historical text templates, determine a first probability that the text combination belongs to device object text and a second probability that it belongs to instruction-related text. The text combination with the highest probability is determined as the device object text; The text combination with the highest probability is identified as the instruction-related text.

4. The home appliance control method according to claim 3, characterized in that, The determination of a first probability that the text combination belongs to device object text and a second probability that it belongs to instruction-related text based on multiple historical text templates includes: Calculate the average text similarity between the text combination and the text template of each historical device object to obtain the first probability; The average text similarity between the text combination and each historical instruction-related text template is calculated to obtain the second probability; the text similarity is obtained by predicting the input text pairs through a trained home appliance control text similarity prediction model; the home appliance control text similarity prediction model is trained through a training dataset of multiple training text pairs for home appliance control and corresponding text similarity annotations.

5. The home appliance control method according to claim 1, characterized in that, The home appliance control text is received and processed by the LLM model after the user inputs the home control requirements, and then output as the result.

6. The home appliance control method according to claim 5, characterized in that, The step of determining the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scenario includes: Based on the home control requirements and the user's user parameters, determine the current home scene and the corresponding home device area; Based on the current home scene and the corresponding home device area, multiple corresponding device protocol rules are matched from the preset device protocol database; Based on the device similarity algorithm, the home device protocol corresponding to the device object text is determined according to the device protocol rules.

7. The home appliance control method according to claim 6, characterized in that, The device similarity algorithm, based on the device protocol rules, determines the home device protocol corresponding to the device object text, including: For each device protocol rule, obtain multiple application device descriptions from the historical application records of that device protocol rule; Calculate the average similarity between the device object text and the text description of each application device to obtain the similarity parameter corresponding to the device protocol rule; The device protocol rule with the highest similarity parameter is determined as the home device protocol corresponding to the device object text.

8. The home appliance control method according to claim 1, characterized in that, The step of parsing and generating device control commands based on the home appliance protocol and the related instruction text, and sending them to the corresponding home appliance for execution, includes: The home device protocol is input into the trained protocol formatting instruction prediction model to obtain the corresponding multiple instruction structure parts; For each instruction structure part, based on the preset fill-in prompt word rules, the instruction structure part and the instruction-related text are combined to form fill-in prompt words; The fill prompt is input into the LLM model to obtain the fill instruction corresponding to the instruction structure part; Combine the fill instructions corresponding to all the instruction structure parts to obtain the device control instructions and send them to the corresponding home appliances for execution.

9. A home appliance control system, characterized in that, The system includes: The acquisition module is used to acquire the home appliance control text output by the LLM model; The recognition module is used to identify the device object text and instruction-related text in the home device control text based on a text recognition algorithm; The determination module is used to determine the home device protocol corresponding to the device object text based on the device protocol rules corresponding to the current home scene; The parsing module is used to parse and generate device control commands based on the home device protocol and the instruction-related text, and send them to the corresponding home device for execution.

10. A home appliance control system, characterized in that, The system includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the home appliance control method as described in any one of claims 1-8.

Images