A method and system of controlling a surveying instrument
By using a word segmentation module and a feature-semantic model to identify the category and semantics of control information for measuring instruments, and generating instruction codes, the problem of brand differences and universality in existing measuring instrument control methods is solved, and simple and accurate instrument control is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN CITY SIGLENT TECH
- Filing Date
- 2025-10-11
- Publication Date
- 2026-06-09
AI Technical Summary
Existing measuring instruments suffer from significant brand differences in control methods, high learning costs, and a lack of versatility and flexibility, failing to meet the needs of complex application scenarios such as multi-instrument integration and remote collaboration.
The system adopts a hierarchical design with agent modules and reasoning modules. It extracts feature word groups of control information through word segmentation module, identifies categories and semantics using feature-semantic model, and generates instruction codes through mapping module and functional agent module to realize the control of measuring instrument.
This eliminates the need for users to memorize complex command formats, lowers the operational threshold, improves the accuracy and robustness of control information, and enhances compatibility and interoperability between different instruments.
Smart Images
Figure CN120930636B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the fields of artificial intelligence and instrument automation control technology, specifically to a method and system for controlling measuring instruments. Background Technology
[0002] With the widespread application of measuring instruments in industry, scientific research, and other fields, the performance and capabilities of instruments from different manufacturers are constantly improving. However, this also brings about problems such as diverse control methods and complex interfaces. Currently, measuring instruments typically provide internal control software, enabling functions such as parameter setting, task initiation, and data acquisition through graphical interfaces or remote commands. These control methods are highly specialized, relying on customized interfaces or command systems provided by the manufacturer, requiring users to possess corresponding professional knowledge.
[0003] However, this "fixed control" model has several problems: on the one hand, different brands of instruments have large differences in the implementation of control commands, interface protocols and operating logic, which leads to high learning costs for users when changing equipment or operating across platforms; on the other hand, the control logic lacks universality and flexibility, and cannot meet the needs of complex application scenarios such as multi-instrument integration and remote collaboration.
[0004] Therefore, it is necessary to propose a universal control method or system that can unify and abstract the control process in order to reduce the operating threshold of instruments and improve the compatibility and interoperability between different instruments. Summary of the Invention
[0005] The main technical problem addressed in this application is to propose a simpler and more universal system and method for controlling and measuring instruments.
[0006] According to a first aspect, one embodiment provides a system for controlling a measuring instrument, comprising:
[0007] The proxy module includes a word segmentation module, a mapping module, and at least one functional proxy module.
[0008] The word segmentation module acquires control information input by the user, extracts feature word groups from the control information, and the feature word groups include at least one feature word; the word segmentation module inputs each feature word into a preset feature-semantic model, and determines the category and semantics of the control information according to the preset feature-semantic model.
[0009] The mapping module determines the functional proxy module corresponding to the control information based on the category and semantics of the control information.
[0010] The functional proxy module acquires the control information and performs format conversion based on the category and semantics of the control information to generate the first instruction code;
[0011] The inference module is used to acquire the first instruction code, perform code conversion on the first instruction code to generate a second instruction code, and send the second instruction code to the agent module to execute the control information and realize the control of the measuring instrument.
[0012] In one embodiment, the word segmentation module extracts feature word groups from the control information, including:
[0013] In the control information, starting with any word, n words are extracted consecutively to form an initial feature word, where n is a natural number;
[0014] When the initial feature character belongs to the preset dictionary in the word segmentation module, the initial feature character is output as the feature character to be judged;
[0015] When the initial feature character does not belong to the preset dictionary in the word segmentation module, n+1 characters are extracted consecutively to update the initial feature character;
[0016] Obtain the probability of each feature word to be judged in the control information, and determine the feature word group based on the probability of each feature word.
[0017] In one embodiment, the word segmentation module determines the category and semantics of the control information based on the preset feature-semantic model, including:
[0018] The preset feature-semantic model includes at least one feature word table, and the feature words in the feature word group are searched in each feature word table;
[0019] The selected feature word table is determined based on the weight of the feature word in each feature word table, and the category and semantics of the control information are determined based on the selected feature word table.
[0020] In one embodiment, each feature word table includes semantic information; the step of determining a selected feature word table based on the weight of the feature word appearing in each feature word table, and determining the category and semantics of the control information based on the selected feature word table, includes:
[0021] In each feature word table, the feature word table that covers the feature words in the feature word group that exceed the first weight is the selected feature word table, and the category of the control information is determined according to the selected feature word table;
[0022] In the selected feature word table, the semantic information of the feature words that exceed the second weight in the feature word group is the selected semantic information, and the semantics of the control information is determined based on the selected semantic information.
[0023] In one embodiment, the word segmentation module determines the category and semantics of the control information based on the preset feature-semantic model, including:
[0024] The preset feature-semantic model includes a feature model and a semantic model;
[0025] The feature model includes at least one feature word table; feature words in the feature word group are searched in each feature word table; in each feature word table, the feature word table that covers feature words in the feature word group that have a first weight is the selected feature word table, and the category of the control information is determined based on the selected feature word table;
[0026] The semantic model includes semantic information; among the semantic information, the semantic information that covers the feature words in the feature word group that have a weight greater than the third weight is selected semantic information, and the semantics of the control information is determined based on the selected semantic information.
[0027] In one embodiment, the function proxy module includes an interface function proxy module for operating the interface of the measuring instrument, a document function proxy module for calling a preset knowledge base, and a control function proxy module for directly controlling the measuring instrument.
[0028] In one embodiment, when the selected feature word table is one, the control information is sent to the interface function proxy module, the document function proxy module, or the control function proxy module;
[0029] When the selected feature word table and / or the selected semantic information are not unique, the control information is sent to the document function agent module.
[0030] According to a second aspect, one embodiment provides a method for controlling a measuring instrument, comprising:
[0031] Obtain control information input by the user;
[0032] Extract the feature word group of the control information, wherein the feature word group includes at least one feature word;
[0033] Each feature word is input into a preset feature-semantic model, and the category and semantics of the control information are determined based on the preset feature-semantic model.
[0034] The format is converted based on the category and semantics of the control information to generate the first instruction code;
[0035] The first instruction code is converted into a second instruction code, and the control information is executed according to the second instruction code to control the measuring instrument.
[0036] In one embodiment, determining the category and semantics of the control information based on the preset feature-semantic model includes:
[0037] The preset feature-semantic model includes at least one feature word table, and the feature words in the feature word group are searched in each feature word table; wherein, each feature word table includes semantic information;
[0038] In each feature word table, the feature word table that covers the feature words in the feature word group that have a weight greater than the first weight is the selected feature word table, and the category of the control information is determined based on the selected feature word table;
[0039] In the selected feature word table, the semantic information of the feature words in the feature word group that exceed the second weight is selected semantic information, and the semantics of the control information is determined based on the selected semantic information.
[0040] According to a third aspect, one embodiment provides a computer-readable storage medium storing a computer program that can be executed by a processor to implement the methods described in any of the above embodiments.
[0041] According to the method and system for controlling a measuring instrument according to the above embodiments, the system includes an agent module and an inference module. The agent module includes a word segmentation module, a mapping module, and at least one functional agent module. The word segmentation module is used to acquire control information input by the user, extract feature word groups from the control information, and input the feature words into a preset feature-semantic model to determine the category and semantics of the control information. The mapping module determines the functional agent module corresponding to the control information based on the category and semantics of the control information. The functional agent module acquires the control information and performs format conversion based on the category and semantics to generate a first instruction code. The inference module is used to acquire the first instruction code generated by the functional agent module, perform further code conversion on it to generate a second instruction code, and return the second instruction code to the agent module, which then issues the instruction to control the measuring instrument. In the method and system for controlling a measuring instrument of this application, the user can directly input control information through natural language, and the category and semantics can be automatically identified through the word segmentation module and the feature-semantic model, eliminating the need for the user to memorize complex instrument instruction formats and significantly reducing the operational threshold. Furthermore, the system achieves a precise mapping from user semantics to instrument functions through a dual recognition mechanism of category and semantics, avoiding ambiguity caused by single keyword matching and improving the accuracy and robustness of control information. Attached Figure Description
[0042] Figure 1 This is a schematic diagram of the structure of a system for controlling a measuring instrument in one embodiment. Figure 1 ;
[0043] Figure 2 This is a schematic diagram of the structure of a system for controlling a measuring instrument in one embodiment. Figure 2 ;
[0044] Figure 3 This is a schematic diagram of the structure of a system for controlling a measuring instrument in one embodiment. Figure 3 ;
[0045] Figure 4 This is a schematic diagram of the structure of a system for controlling a measuring instrument in one embodiment. Figure 4 ;
[0046] Figure 5 Deployment of the agent module and the inference module in one embodiment Figure 1 ;
[0047] Figure 6 Deployment of the agent module and the inference module in one embodiment Figure 2 ;
[0048] Figure 7 Deployment of the agent module and the inference module in one embodiment Figure 3 ;
[0049] Figure 8 Deployment of the agent module and the inference module in one embodiment Figure 4 ;
[0050] Figure 9 Method flow of the method for controlling a measuring instrument in another embodiment Figure 1 ;
[0051] Figure 10 Method flow of the method for controlling a measuring instrument in another embodiment Figure 2 . Detailed Implementation
[0052] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments. Similar elements in different embodiments are referred to by related similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the present application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present application are not shown or described in the specification. This is to avoid obscuring the core parts of the present application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.
[0053] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.
[0054] The serial numbers assigned to components in this document, such as "first" and "second," are used only to distinguish the described objects and have no sequential or technical meaning. The terms "connection" and "linkage" used in this application, unless otherwise specified, include both direct and indirect connections (linkages).
[0055] Please refer to Figure 1 One embodiment provides a system for controlling a measuring instrument, including an agent module 110 and an inference module 120.
[0056] In one embodiment, the proxy module 110 is used to identify the category and semantics of control information according to a preset control information identification pattern, and convert the control information into a corresponding first instruction code based on the determined category and semantics. The reasoning module 120 is used to obtain the first instruction code, perform code conversion on the first instruction code to generate a second instruction code, and send the second instruction code to the proxy module 110. The proxy module 110 executes the control information corresponding to the second instruction code according to the second instruction code, thereby enabling the measuring instrument to realize the function corresponding to the control information.
[0057] It should be noted that the category of control information refers to which type of control command is set in the measuring instrument to which the control information belongs, and the semantics of control information refers to the specific meaning of the control information, that is, what operation the measuring instrument specifically controls to perform.
[0058] Please refer to Figure 2 In one embodiment, the proxy module 110 includes a word segmentation module 111, a mapping module 112, and at least one functional proxy module 113.
[0059] In one embodiment, the word segmentation module 111 acquires control information input by the user and extracts feature word groups from the control information, wherein each feature word group includes at least one feature word. The word segmentation module 111 inputs each feature word into a preset feature-semantic model and determines the category and semantics of the control information based on the preset feature-semantic model. The category and semantics of the control information can be determined simultaneously by the feature-semantic model, or they can be determined sequentially by the feature-semantic model.
[0060] It should be noted that user-input control information includes text input, voice recognition, virtual mouse / virtual remote adaptation, and intent recognition. Text input refers to the user directly entering command text; voice recognition means the user speaks a command, which the system recognizes and converts into text; the virtual mouse means the system translates voice or text commands into actions such as clicking / dragging / inputting on the interface. For example, if the user says "Click to start measurement," the agent module 110 will map it to the instruction to click the "Start" button on the instrument interface; if the measuring instrument has a remote desktop, web console, or dedicated GUI control interface, virtual remote adaptation is achieved through a simulated remote control protocol. The system converts natural language commands into remote protocol operation commands to control the measuring instrument; intent recognition means the system directly recognizes the user's goals and needs through natural language understanding, rather than executing commands word for word. For example, if the user says "Save the results," the system recognizes the intent as "Save the current test data."
[0061] In one embodiment, the mapping module 112 can determine which functional proxy module 113 the control information belongs to based on the category and semantics of the control information, and then send the control information to the corresponding functional proxy module 113.
[0062] In one embodiment, the corresponding function proxy module 113 obtains the control information, performs format conversion based on the function category of the control information in the measuring instrument and the measurement function of the control information in the measuring instrument, and generates a first instruction code.
[0063] In summary, the system for controlling and measuring instruments provided in this application includes a proxy module 110 and an inference module 120. The proxy module 110 comprises a word segmentation module 111, a mapping module 112, and at least one functional proxy module 113. The word segmentation module 111 extracts features from the user-input control information and identifies the category and semantics of the control information based on a feature-semantic model. The mapping module 112 determines the functional category and measurement function corresponding to the control information based on the category and semantics, and selects a suitable functional proxy module 113. The functional proxy module 113 then converts the control information into a first instruction code. After obtaining the first instruction code, the inference module 120 performs code conversion to generate a second instruction code and returns this code to the proxy module 110, which then executes the corresponding operation to drive the measuring instrument to complete the required function.
[0064] Through the hierarchical design of the word segmentation module 111, mapping module 112, function proxy module 113 and inference module 120 in this application, the step-by-step processing of the input control information from semantic parsing to function execution is realized, which is convenient for expansion and maintenance. Moreover, the system uses the feature-semantic model to identify the category and semantics of the control information, enabling the measuring instrument to understand natural language or complex instructions, and improving the level of interaction intelligence. Thus, the user only needs to input the control information without understanding the underlying instructions to complete the control of the measuring instrument, reducing the usage difficulty.
[0065] In one embodiment, to improve the accuracy of the word segmentation module 111 in extracting the characteristic word groups of the control information, in the control information of this application, starting from any word, n (n is a natural number) consecutive words are extracted to form an initial characteristic word. If the initial characteristic word is in the preset dictionary of the word segmentation module 111, the initial characteristic word is output as the to-be-judged characteristic word. If the initial characteristic word is not in the preset dictionary of the word segmentation module 111, n + 1 consecutive words are extracted to update the initial characteristic word, and so on in a loop until all the to-be-judged characteristic words in the control information are extracted. After all the to-be-judged characteristic words in the control information are extracted, the characteristic word group is determined according to the probability of each judged characteristic word.
[0066] In one embodiment, in the control information, if the initial characteristic word obtained by continuously extracting n words belongs to the preset dictionary of the word segmentation module 111, and at the same time, the initial characteristic word obtained by continuously extracting n + 1 words also belongs to the preset dictionary of the word segmentation module 111, then the initial characteristic words obtained twice are also output as the to-be-judged characteristic words, and it is also necessary to judge the probability of the to-be-judged characteristic words obtained in this case.
[0067] Specifically, taking "open channel 1" as an example, all the possible words obtained from the dictionary are "hit", "open", "open"; "through", "path", "channel"; "1" (numbers are generally treated as separate words, and may also be combined with "channel" into "channel 1"), so the to-be-judged characteristic words obtained from this control information are "hit", "open", "open", "through", "path", "channel" and "1". Obtain the probability of each to-be-judged characteristic word. For example, the probability of "open" is greater than that of "hit" and "open", and the probability of "channel" is greater than that of "through" and "path". Then, it can be determined that the characteristic word group extracted from the control information "open channel 1" is "open", "channel", "1".
[0068] In one embodiment, to improve the accuracy of the word segmentation module 111 in extracting feature word groups of control information, this application can also perform word segmentation based on a deep learning model, including: vectorizing each character (characters include letters and numbers) in the control information to obtain a number matrix of a set dimension; adjusting the dimension value of each character in the control information according to the vectors of adjacent characters; determining the label of each character in the control information according to the characters after the dimension value adjustment; and finally completing the concatenation of feature word groups according to the labels.
[0069] It should be noted that computers can only process numbers and cannot directly understand Chinese characters. Therefore, each character in the control information needs to be converted into a numerical form before it can be processed by a deep learning model. Dimension is the number of numbers in a vector. This application specifies that each character is represented by 300 numbers, so each character will correspond to a 300-dimensional vector (the 300 numbers are not randomly chosen but obtained through training). The higher the dimension, the more information the vector can express.
[0070] In one embodiment, after obtaining the vector of each character, the deep learning model dynamically adjusts the numerical values of each dimension of the vector corresponding to each character according to the context of each character, thereby determining the label of each character in the control information. In this way, the feature word group in the control information can be determined based on the label of each character.
[0071] It should be noted that the tags include B, M, E, and S. B represents the first word of a word, M represents the middle word of a word, E represents the last word of a word, and S represents a single word.
[0072] In one embodiment, during the process of determining the category and semantics of control information based on a preset feature-semantic model in the word segmentation module 111, the feature-semantic model includes at least one feature word table, and each feature word table includes semantic information. The word segmentation module 111 searches for feature words in feature word groups in each feature word table, determines the selected feature word table based on the weight of the feature words appearing in each feature word table, and determines the category and semantics of the control information based on the selected feature word table. Specifically, in each feature word table, the feature word table that covers feature words in the feature word group that have a weight exceeding the first weight is the selected feature word table, and the category of control information is determined based on the selected feature word table; in the selected feature word table, the semantic information that covers feature words in the feature word group that have a weight exceeding the second weight is the selected semantic information, and the semantics of the control information is determined based on the selected semantic information.
[0073] It should be noted that the first weight is used to define the criteria for determining the category of control information. For example, when the coverage of a feature word in the control information in a certain feature word table reaches or exceeds the first weight, the feature word table is determined as the selected feature word table, and the category of the control information is determined accordingly. The value of the first weight ranges from 50% to 70%, and can also be set to 60% to ensure that misjudgments are avoided due to a single high-frequency feature word appearing by chance, while also ensuring sufficient reliability in category determination. The second weight is used to define the criteria for determining semantic information. For example, in the selected feature word table, when a feature word in the covered feature word group that exceeds the second weight has a corresponding relationship with a certain semantic information, the semantic information is determined as the selected semantic information, thereby obtaining the specific semantics of the control information. The value of the second weight ranges from 30% to 50%, and can also be set to 40% to ensure more flexible semantic-level determination and the ability to distinguish different specific semantics within the same category.
[0074] In one embodiment, this application sets up an interface operation feature word table, a help document feature word table, and an instrument control feature word table in the feature-semantic model for measuring instruments.
[0075] In one embodiment, the feature words and semantic information included in the interface operation feature word table are shown in Table 1: The feature words in the interface operation feature word table include display, hide, zoom in, zoom out, full screen, window, curve, cursor, and grid, etc., and the semantics corresponding to these feature words are display interface, adjust interface, and waveform observation mode; The feature words in the interface operation feature word table also include open, close, switch, return, menu, tab, page, channel, and display, etc., and the semantics corresponding to these feature words are interface navigation and switching; The feature words in the interface operation feature word table also include save, load, import, export, screenshot, file, store, and record, etc., and the semantics corresponding to these feature words are data or interface saving and loading; The feature words in the interface operation feature word table also include start, stop, pause, continue, refresh, reset, default, exit, and print, etc., and the semantics corresponding to these feature words are basic operation instructions.
[0076] Table 1. User Interface Operation Feature Table
[0077] Feature words Semantic information Show, hide, zoom in, zoom out, full screen, window, curve, cursor, grid Interface display, interface adjustment, and waveform observation methods Open, close, switch, return, menu, tab, page, channel, display Interface navigation and switching Save, load, import, export, screenshot, file, store, record Data or interface saving and loading Start, Stop, Pause, Continue, Refresh, Reset, Default, Exit, Print Basic operation instructions
[0078] In one embodiment, the feature words and semantic information included in the help document feature word table are shown in Table 2: The feature words in the help document feature word table include help, document, manual, guide, instruction, and tutorial, etc., and the semantics of these feature words are to open or call help resources; The feature words in the help document feature word table include find, search, locate, directory, and index, etc., and the semantics of these feature words are to retrieve documents or instructions; The help document feature word table also includes error, warning, code, prompt, exception, and common problems, etc., and the semantics of these feature words are to query error information or solutions; The help document feature word table also includes version, update, feature introduction, technical specifications, and user manual, etc., and the semantics of these feature words are to obtain the instrument version or feature description.
[0079] Table 2 Help Document Feature Font Table
[0080] Feature words Semantic information Help, documentation, manuals, guides, instructions, tutorials Open or invoke help resources Find, search, locate, directory, index Document or description retrieval Errors, warnings, codes, tips, exceptions, common problems Query error messages or solutions Version, Updates, Features, Technical Specifications, User Guide Get instrument version or function description
[0081] In one embodiment, the feature words and semantic information included in the instrument control feature word table are shown in Table 3: The feature words in the instrument control feature word table include channel, CH1, CH2, input, output, coupling, on, off, display, and hide, etc., and the semantics of these feature words are controlling the channel switch and coupling mode; the instrument control feature word table also includes trigger, rising edge, falling edge, automatic, single, delay, trigger level, and trigger source, etc., and the semantics of these feature words are configuring trigger conditions; the instrument control feature word table also includes time base, sampling rate, time, number of sampling points, record length, and storage depth, etc., and the semantics of these feature words are controlling sampling and time base; the instrument control feature word table also includes time base, sampling rate, time, number of sampling points, record length, and storage depth, etc., and the semantics of these feature words are controlling sampling and time base; the instrument control feature word table also includes channel, CH1, CH2, input, output, coupling, on, off, display, and hide, etc. The character list also includes amplitude, voltage, peak value, root mean square, reference level, gain, attenuation, power, and dBm, etc., and the semantics of these character words are voltage / power measurement and adjustment; the instrument control character list also includes frequency, center frequency, bandwidth, resolution bandwidth, scan width, span, and harmonics, etc., and the semantics of these character words are frequency / bandwidth control; the instrument control character list also includes waveform, signal, FFT, spectrum, period, pulse, DC, and AC, etc., and the semantics of these character words are signal and frequency domain analysis; the instrument control character list also includes measurement, statistics, average, peak value, phase, delay, jitter, and power spectrum, etc., and the semantics of these character words are automatic measurement.
[0082] Table 3 Instrument Control Character List
[0083] Feature words Semantic information Channel, CH1, CH2, Input, Output, Coupling, On, Off, Show, Hide Control channel switching and coupling method Trigger, rising edge, falling edge, automatic, single, delay, trigger level, trigger source Configure trigger conditions Time base, sampling rate, time, number of sampling points, record length, storage depth Control sampling and time base Amplitude, Voltage, Peak Value, RMS, Reference Level, Gain, Attenuation, Power, dBm Voltage / Power Measurement and Adjustment Frequency, center frequency, bandwidth, resolution bandwidth, scan width, span, harmonics Frequency / bandwidth control Waveform, signal, FFT, spectrum, period, pulse, DC, AC Signal and frequency domain analysis Measurement, statistics, average, peak value, phase, delay, jitter, power spectrum Automatic measurement
[0084] In one embodiment, each feature character in the interface operation feature character table, help document feature character table, and instrument control feature character table carries a weight value, which can be manually set or obtained through training. The word segmentation module 111 searches for each feature character in the feature character group in the interface operation feature character table, help document feature character table, and instrument control feature character table one by one. If the number of feature characters covered in a certain feature character table exceeds a first weight, then that feature character table is selected as the final category. Within the selected feature character table, different feature characters correspond one-to-one with specific semantics. If the feature character coverage rate under a certain semantic exceeds a second weight, then that semantic is selected.
[0085] In another embodiment, when the word segmentation module 111 performs the task of determining the category and semantics of the control information according to the preset feature-semantic model, the feature-semantic model consists of two models, namely the feature model and the semantic model.
[0086] In one embodiment, a feature model is used to determine the category to which control information belongs, and it includes at least one feature word table. Specifically, the feature word table includes an interface operation feature word table, a help document feature word table, and an instrument control feature word table.
[0087] In one embodiment, the feature model searches for feature words in feature word groups within each feature word table and calculates weights based on the coverage of feature words in each feature word table. If the number of feature words covered in a certain feature word table exceeds a first weight, then that feature word table is determined as the selected feature word table. The category of control information can be determined based on the selected feature word table.
[0088] In one embodiment, a semantic model is used to parse the semantics of control information under a determined category. The semantic model includes semantic information, with each semantic information corresponding to a set of feature words. When more than one feature word in the feature word set covers a certain semantic information, the semantic information is determined to be selected, and the semantics of the control information is determined accordingly.
[0089] It should be noted that the third weight is used to define the criteria for determining semantic information under the semantic model. If the third weight is set too low, incorrect matching may occur due to the appearance of individual feature words, reducing the accuracy of semantic recognition; if the third weight is set too high, some valid semantic information may not be recognized, affecting the system's ability to cover control information. Therefore, this application sets the value range of the third weight to 20%–40%, or it can be set to 30%. In this way, the recognition of semantic information will not be misjudged due to the accidental appearance of a small number of feature words, and it can still correctly parse the corresponding semantic information even when the user-input control information contains omissions, simplifications, or colloquial expressions.
[0090] Using the above method, the word segmentation module 111 acquires the control information input by the user and processes it through a preset feature-semantic model. The feature-semantic model includes multiple feature word tables, including an interface operation feature word table, a help document feature word table, and an instrument control feature word table. The word segmentation module 111 determines the category and semantics of the control information by matching the feature words in each feature word table. For example, when the input is "open channel 1", the word segmentation module 111 detects that the feature words "channel" and "open" in the control information have high weights in the instrument control feature word table, thus classifying the control information as instrument control and determining the semantics as "control channel switch and coupling method" based on the feature word matching results.
[0091] In one embodiment, the mapping module 112 receives the category and semantics of the control information output by the word segmentation module 111, and uses this information to comprehensively determine the function proxy module 113 corresponding to the control information.
[0092] In one embodiment, when the selected feature word table is an interface operation feature word table, the category of the control information is an interface operation class. When the category of the control information is an interface operation class, the mapping module 112 determines, based on the semantic information, that the corresponding function proxy module 113 is the interface function proxy module 113a.
[0093] In one embodiment, when the selected feature word table is a help document feature word table, the category of the control information is the help document class. When the category of the control information is the help document class, the mapping module 112 determines, based on the semantic information, that the function proxy module 113 corresponding to the control information is the document function proxy module 113b.
[0094] In one embodiment, when the selected feature word table is an instrument control feature word table, the category of the control information is instrument control. When the category of the control information is instrument control, the mapping module 112 determines the corresponding function proxy module 113 as control function proxy module 113c based on the semantic information.
[0095] Through the above process, the mapping module 112 can directly map the categories and semantics output by the word segmentation module 111 to the corresponding functional proxy module 113, thereby realizing the rapid and accurate diversion and execution of natural language control information.
[0096] Please refer to Figure 3 In one embodiment, the function proxy module 113 includes an interface function proxy module 113a for operating the interface of the measuring instrument, a document function proxy module 113b for calling a preset knowledge base, and a control function proxy module 113c for directly controlling the measuring instrument.
[0097] In one embodiment, the function proxy module 113 converts the control information according to a preset format to form standardized formatted information. This formatted information adopts a unified data structure and includes at least the following elements: original control information, function category, measurement function, and parameter information. The original control information is natural language text input by the user; the function category is the function category corresponding to the control information in the measuring instrument; the measurement function is the measurement function corresponding to the control information in the measuring instrument; and the parameter information is the relevant parameters obtained through semantic parsing.
[0098] In one embodiment, after acquiring control information, the function proxy module 113 performs format conversion based on the formatted information to generate a first instruction code. The first instruction code is an instruction-based representation used to drive internal function calls of the measuring instrument, and its encoding rules are set according to the control interface of the measuring instrument. The function proxy module 113 sends the first instruction code to the inference module 120, and the inference module 120 calls the specific functions of the measuring instrument based on the first instruction code, thereby realizing the automated conversion of natural language instructions to instrument function calls.
[0099] In one embodiment, when parsing the control information input by the user, the word segmentation module 111 determines the category and corresponding semantics of the control information based on a preset feature word table. The feature word table includes an interface operation feature word table, a help document feature word table, and an instrument control feature word table. If the word segmentation module 111 matches only one of the feature word tables after parsing the user-input control information, it indicates that the category of the control information is unique. At this time, the system directly determines the category of the control information based on the matched feature word table and sends the control information to the function proxy module 113 corresponding to that category. For example: when the control information matches the interface operation feature word table, the control information is sent to the interface function proxy module 113a to realize interface display adjustment, waveform observation mode switching, or interface navigation and saving operations; when the control information matches the help document feature word table, the control information is sent to the document function proxy module 113b to call the preset knowledge base and return help instructions; when the control information matches the instrument control feature word table, the control information is sent to the control function proxy module 113c to realize direct calling of the underlying control logic of the measuring instrument, such as channel switch, range adjustment, trigger setting, etc.
[0100] In one embodiment, if the control information matches multiple feature word lists simultaneously, resulting in the category not being uniquely determined, or if the control information matches a single feature word list but its semantic information is ambiguous and cannot uniquely correspond to a specific function in the measuring instrument, the system will send the control information to the document function proxy module 113b. The document function proxy module 113b can invoke a preset knowledge base and, combined with contextual semantic information and historical interaction records, further interpret and clarify the user input. For example, when the user inputs "save data," it may involve either "save waveform interface" or "save measurement data file." In this case, the system will first enter the document function proxy module 113b to provide the user with different possible interpretations and confirm the final execution path based on subsequent user feedback.
[0101] Please refer to Figure 4 In one embodiment, the reasoning module 120 includes an instrument interface recognition module 121, a question-and-answer module 122, and a programmable instruction encoding module 123.
[0102] In one embodiment, the instrument interface recognition module 121 is connected to the interface function proxy module 113a. The instrument interface recognition module 121 obtains the first instruction code and calls the interface-to-virtual coordinate transformation model to map the abstract operation semantics to specific virtual interface coordinates, operation paths, and action types (click, drag, switch). The instrument interface recognition module 121 further calls a preset knowledge base to retrieve interface layout rules, control positions, and operation constraints related to the operation. Based on the retrieval results, the instrument interface recognition module 121 checks and corrects the first instruction code. The corrected operation information is converted into a second instruction code, which clearly indicates the specific interface interaction action. The second instruction code is returned to the interface function proxy module 113a, which encapsulates it into formatted text with interface operation symbols and outputs it to the instrument interface, thereby completing the corresponding interface operation.
[0103] In one embodiment, the question-answering module 122 is connected to the document function proxy module 113b. The question-answering module 122 obtains a first instruction code and calls a preset knowledge base to retrieve semantic content and explanatory documents related to the query target. Based on the knowledge base retrieval results, the question-answering module 122 generates a matching answer semantic as the core information of the second instruction code. During the reasoning process, the question-answering module 122 performs consistency checks and supplementary explanations on the preliminary answer. For example, when explaining "bandwidth limitation," it will supplement the explanation that "after enabling bandwidth limitation, high-frequency components of the channel will be filtered out, which helps to reduce noise." The corrected second instruction code is formatted into formatted text with a knowledge response mark. The document function proxy module 113b receives the second instruction code and converts it into a user-readable natural language response, thereby fulfilling the user's knowledge query needs.
[0104] In one embodiment, the programmable instruction encoding module 123 is connected to the control function proxy module 113c. The programmable instruction encoding module 123 acquires a first instruction code and sends the formatting information corresponding to the first instruction code to a preset knowledge base. It then retrieves standard instruction rules and semantic mapping information matching the formatting information from the knowledge base. Based on the knowledge base retrieval results, the programmable instruction encoding module 123 checks and corrects the first instruction code, such as verifying whether the parameter range is reasonable and whether the instruction format conforms to the standard protocol, and makes corrections if necessary. The programmable instruction encoding module 123 then performs secondary formatting on the corrected code to generate a second instruction code, and returns the second instruction code to the control function proxy module 113c. The control function proxy module 113c encapsulates the second instruction code into formatted text with a type flag based on its own identifier. Finally, the formatted text is output as an instrument instruction to realize the measurement function corresponding to the control information.
[0105] Please refer to Figure 5 In one embodiment, the control terminal and the measuring instrument establish a connection via a local area network or a point-to-point Bluetooth network. This network structure ensures that control information can be transmitted in a low-latency, high-stability manner, thereby meeting the real-time and reliability requirements of the testing scenario. Since the measuring instrument itself has strong computing and storage capabilities (including video memory, RAM, and processor), the agent module 110 and inference module 120 can be directly deployed inside it without relying on external servers or the cloud, thereby improving the immediacy and security of control.
[0106] In one embodiment, the control terminal can be a PC, tablet, mobile phone, or other device with an interactive interface. The user inputs control information on the control terminal through a graphical interface, voice commands, or script commands. The control information is transmitted over the network and then enters the proxy module 110 of the measuring instrument. The proxy module 110, as the first-level processing unit in the measuring instrument, performs semantic analysis and classification of the control information. Based on the classification results, the proxy module 110 maps the information to corresponding functional proxies. Since the proxy module 110 has been clearly described in the above embodiments, it will not be repeated here. The classified instructions are sent to the corresponding reasoning module 120. The reasoning module 120 performs logical deduction and code generation based on the semantics, context, and knowledge base content of the input information. After reasoning is completed, the reasoning module 120 returns the result to the proxy module 110. Since the reasoning module 120 has been clearly described in the above embodiments, it will not be repeated here. After receiving the reasoning result, the proxy module 110 converts it into standardized instrument instructions conforming to the specifications of the measuring instrument receiving module 130. The converted instrument instructions are then sent to the measuring instrument receiving module 130. The receiving module 130 is a unit that directly interacts with the instrument's underlying hardware driver. It is responsible for parsing the instructions passed from the upper layer into specific hardware operations. The receiving module 130 sends the specific hardware operations to the execution module 140 to control the test instrument to complete the corresponding actions.
[0107] Please refer to Figure 6 In one embodiment, under normal circumstances, because the measuring instrument itself has strong computing and storage capabilities, it can simultaneously deploy the agent module 110 and the inference module 120 to achieve localized processing and ensure the real-time performance and security of command responses. However, in certain application scenarios, such as when the measuring instrument has limited hardware resources and cannot simultaneously support the operation of multiple combined modules (e.g., insufficient video memory, excessive CPU load, or limited memory), or when the test task itself requires calling a larger-scale knowledge base and more complex inference algorithms, exceeding the computing capacity of a single instrument, the system will adopt an independent deployment strategy for the inference module 120.
[0108] In one embodiment, when the measuring instrument and the inference module 120 are on the same local area network (LAN), the inference module 120 can be deployed as a LAN service on a dedicated server or high-performance computing node. When multiple measuring instruments need to be called across regions or centrally managed, the inference module 120 can be deployed on a server accessible via a wide area network (WAN) or a cloud platform to achieve remote inference services. Upon startup, the proxy module 110 reads or registers the service address of the inference module 120. After receiving control information and completing semantic classification, the proxy module 110 encapsulates the instructions into a unified protocol format and sends it to the registered service address of the inference module 120. The independently deployed inference module 120, upon receiving a request from the proxy module 110, invokes the knowledge base and inference engine to complete the processing and returns the inference result to the proxy module 110 in the form of a protocol response. The entire process ensures that even when the inference module 120 operates externally, its functional logic and inference accuracy remain consistent with those of locally deployed inference.
[0109] Please refer to Figure 7 In one embodiment, when the control requirements for the measuring instrument increase and the complexity of the inference process significantly increases, the proxy module 110 and the inference module 120 can be deployed independently. In addition to completing the traditional control information input and transmission functions, the control terminal will also undertake part of the data transmission work to the receiving module 130. That is, after the inference result is generated, the inference module 120 can directly return the result to the control terminal, which then connects to the receiving module 130 of the measuring instrument.
[0110] Please refer to Figure 8 In one embodiment, when the measuring instrument and the agent module 110 can interact directly, there is no need to rely on the control terminal as an intermediate forwarding node. In this case, the system can use a direct connection mode to complete the control process, thereby further reducing transmission latency and simplifying the communication link. The receiving module 130 of the measuring instrument can communicate directly with the agent module 110. After completing the classification and semantic parsing of the control information and the functional agent division, the agent module 110 can directly transmit formatted instructions or execution instructions converted by the inference module 120 to the receiving module 130 of the measuring instrument.
[0111] Please refer to Figure 9 Another embodiment provides a method for controlling a measuring instrument, comprising the following steps.
[0112] Step S1: Obtain the control information input by the user.
[0113] In one embodiment, the control information input by the user includes text input, speech recognition, virtual mouse / virtual remote adaptation, and intent recognition. Text input refers to the user directly entering command text; speech recognition refers to the user speaking commands, which the system then recognizes and converts into text; virtual mouse refers to the system translating speech or text commands into actions such as clicking / dragging / inputting on the interface. For example, if the user says "Click to start measurement," the agent module 110 will map it to the instruction to click the "Start" button on the instrument interface; if the measuring instrument has a remote desktop, web console, or dedicated GUI control interface, virtual remote adaptation is achieved through a simulated remote control protocol, where the system converts natural language commands into remote protocol operation commands to control the measuring instrument; intent recognition refers to the system directly recognizing the user's goals and needs through natural language understanding, rather than executing commands word by word. For example, if the user says "Help me save the results," the system recognizes the intent as "Save the current test data."
[0114] Step S2: Extract the feature word groups of control information.
[0115] In one embodiment, this application extracts n (n is a natural number) characters consecutively from the control information, starting with any character, to form an initial feature character. If the initial feature character is in the preset dictionary in the word segmentation module 111, the initial feature character is output as the feature character to be judged. If the initial feature character is not in the preset dictionary in the word segmentation module 111, n+1 characters are extracted consecutively to update the initial feature character, and this process is repeated until all the feature characters to be judged in the control information have been extracted. After all the feature characters to be judged in the control information have been extracted, a feature character group is determined according to the probability of each judgment feature character.
[0116] In one embodiment, in the control information, if the initial feature character obtained by continuously extracting n characters belongs to the preset dictionary in the word segmentation module 111, and the initial feature character obtained by continuously extracting n+1 characters also belongs to the preset dictionary in the word segmentation module 111, then the initial feature characters obtained in both cases will be output as the feature characters to be judged, and the probability of the feature characters to be judged obtained in this case will also need to be judged.
[0117] Specifically, taking "Open Channel 1" as an example, all possible words obtained from the dictionary are "hit", "open", "open"; "through", "path", "channel"; "1" (numbers are generally treated as separate words and may also be combined with "channel" to form "Channel 1"). Therefore, the candidate feature words obtained from this control information are "hit", "open", "open", "through", "path", "channel", and "1". Obtain the probabilities of each candidate feature word. For example, the probability of "open" is greater than that of "hit" and "open", and the probability of "channel" is greater than that of "through" and "path". Then, based on this, it can be determined that the feature word group extracted from the control information "Open Channel 1" is "open", "channel", "1".
[0118] In one embodiment, to improve the accuracy of the feature word group extraction of the word segmentation module 111 for control information, the present application can also perform word segmentation based on a deep learning model, including: performing vectorization conversion on each character (here the characters include letters and numbers) in the control information to obtain a digital matrix of a set dimension; adjusting the dimension values of each character in the control information according to the vectors of adjacent characters; determining the label of each character in the control information according to the character after the dimension value adjustment, and finally completing the splicing of the feature word group according to the label.
[0119] It should be noted that computers can essentially only process numbers and cannot directly understand Chinese characters. Therefore, each character in the control information needs to be converted into a numerical form before being handed over to the deep learning model for processing. Dimension refers to the number of numbers in a vector. The present application stipulates that each character is represented by 300 numbers, so each character will correspond to a 300-dimensional vector (the 300 numbers are not randomly selected but obtained through training). The higher the dimension, the more information the vector can express.
[0120] In one embodiment, after the deep learning model obtains the vector of each character, according to the context of each character's front and back, it dynamically adjusts the numerical values of each dimension of the vector corresponding to each character, thereby determining the label of each character in the control information. Based on this, the feature word group in the control information can be determined according to the label of each character.
[0121] It should be noted that the labels include B, M, E, and S. B represents the first character of a word, M represents the middle character of a word, E represents the last character of a word, and S represents a single character forming a word.
[0122] Step S3: Input each feature word into a preset feature-semantic model, and determine the category and semantics of the control information according to the preset feature-semantic model.
[0123] Please refer to Figure 10In one embodiment, when performing step S3 to input each feature word into a preset feature-semantic model and determine the category and semantics of the control information based on the preset feature-semantic model, the following steps are also included.
[0124] Step S31: In each feature word table, the feature word table that covers the feature words in the feature word group that have a weight greater than the first weight is the selected feature word table, and the category of control information is determined based on the selected feature word table.
[0125] In one embodiment, the feature-semantic model includes at least one feature word table, and each feature word table includes semantic information. The word segmentation module 111 searches for feature words in feature word groups within each feature word table, determines the selected feature word table based on the weight of each feature word's appearance in each feature word table, and determines the category and semantics of the control information based on the selected feature word table. Specifically, among the feature word tables, the feature word table that covers feature words in the feature word group with a weight exceeding a first weight is the selected feature word table, and the category of the control information is determined based on the selected feature word table.
[0126] Step S32: In the selected feature word table, the semantic information of the feature words in the feature word group that exceed the second weight is selected semantic information, and the semantics of the control information is determined based on the selected semantic information.
[0127] Step S4: Perform format conversion based on the category and semantics of the control information to generate the first instruction code.
[0128] Step S5: Perform code conversion on the first instruction code to generate the second instruction code.
[0129] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.
[0130] The above examples illustrate this application only to aid understanding and are not intended to limit its scope. Those skilled in the art to which this application pertains can make various simple deductions, modifications, or substitutions based on the ideas presented.
Claims
1. A system for controlling a measuring instrument, characterized in that, include: The proxy module includes a word segmentation module, a mapping module, and at least one functional proxy module. The word segmentation module acquires control information input by the user, extracts feature word groups from the control information, and the feature word groups include at least one feature word; the word segmentation module inputs each feature word into a preset feature-semantic model, and the preset feature-semantic model includes a feature model and a semantic model. The feature model includes a feature word table for interface operation, a feature word table for help documentation, and a feature word table for instrument control. Feature words in the feature word groups are searched within each feature word table. In each feature word table, the feature word table that covers the feature words in the feature word group that exceed the first weight is the selected feature word table, and the category of the control information is determined based on the selected feature word table; The semantic model includes semantic information; among the semantic information, the semantic information that covers the feature words in the feature word group that have a weight greater than the third weight is selected semantic information, and the semantics of the control information is determined based on the selected semantic information. The mapping module determines the functional proxy module corresponding to the control information based on the category and semantics of the control information. The functional proxy module acquires the control information and performs format conversion based on the category and semantics of the control information to generate the first instruction code; The inference module is used to acquire the first instruction code, perform code conversion on the first instruction code to generate a second instruction code, and send the second instruction code to the agent module to execute the control information and realize the control of the measuring instrument.
2. The system for controlling and measuring instruments as described in claim 1, characterized in that, The word segmentation module extracts feature word groups from the control information, including: In the control information, starting with any word, n words are extracted consecutively to form an initial feature word, where n is a natural number; When the initial feature character belongs to the preset dictionary in the word segmentation module, the initial feature character is output as the feature character to be judged; When the initial feature character does not belong to the preset dictionary in the word segmentation module, n+1 characters are extracted consecutively to update the initial feature character; Obtain the probability of each feature word to be judged in the control information, and determine the feature word group based on the probability of each feature word.
3. The system for controlling and measuring instruments as described in claim 2, characterized in that, The functional proxy module includes an interface function proxy module for operating the interface of the measuring instrument, a document function proxy module for calling a preset knowledge base, and a control function proxy module for directly controlling the measuring instrument.
4. The system for controlling and measuring instruments as described in claim 3, characterized in that, When only one feature word is selected, the control information is sent to the interface function proxy module, the document function proxy module, or the control function proxy module. When the selected feature word table and / or the selected semantic information are not unique, the control information is sent to the document function agent module.
5. A method for controlling a measuring instrument, characterized in that, include: Obtain control information input by the user; Extract the feature word group of the control information, wherein the feature word group includes at least one feature word; Each feature word is input into a preset feature-semantic model, which includes a feature model and a semantic model. The feature model includes a feature word table for interface operation, a feature word table for help documentation, and a feature word table for instrument control. Feature words in the feature word groups are searched within each feature word table. In each feature word table, the feature word table that covers the feature words in the feature word group that exceed the first weight is the selected feature word table, and the category of the control information is determined based on the selected feature word table; The semantic model includes semantic information; among the semantic information, the semantic information that covers the feature words in the feature word group that have a weight greater than the third weight is selected semantic information, and the semantics of the control information is determined based on the selected semantic information. The format is converted based on the category and semantics of the control information to generate the first instruction code; The first instruction code is converted into a second instruction code, and the control information is executed according to the second instruction code to control the measuring instrument.
6. A computer-readable storage medium, characterized in that, The medium stores a computer program that can be executed by a processor to implement the method as described in claim 5.