A mine crushing equipment energy-saving operation and maintenance scheme generation method and system

Abstract

The application relates to the technical field of intelligent operation and maintenance of mine equipment, and discloses a mine crushing equipment energy-saving operation and maintenance scheme generation method and system, which comprises the following steps: obtaining initial description information for a target device; searching for a plurality of candidate standard state features corresponding to a feature item to be clarified, and extracting dimension information for distinguishing the plurality of candidate standard state features; generating a guide instruction and outputting, obtaining supplementary feedback information; determining a target standard state feature from the plurality of candidate standard state features; replacing or associating the feature item to be clarified in the initial description information into the target standard state feature, and generating a maintenance scheme for the target device based on the initial description information after replacement or association. Through construction of a maintenance knowledge conversion framework based on interactive feature clarification and dynamic working condition constraints, the application realizes automatic generation of a standard maintenance scheme which is accurate and executable from a fuzzy and unstructured field fault description.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology for mining equipment, and in particular to a method and system for generating energy-saving operation and maintenance solutions for mining crushing equipment. Background Technology

[0002] Mining crushing equipment (such as jaw crushers, cone crushers, and gyratory crushers) is a core energy-consuming device in the mineral processing flow, and its operating status directly affects production efficiency and unit power consumption. In actual production, the equipment operates under high load, strong impact, and high dust conditions for extended periods, leading to frequent abnormal events. Currently, the maintenance process typically begins with verbal or written descriptions of abnormal phenomena by on-site personnel, collected via telephone, work orders, or mobile devices. However, these descriptions are highly subjective and colloquial; for example, different personnel might describe the same bearing vibration as "abnormal bearing noise," "frame shaking," or "stone jamming sound." This non-standardized description poses a significant challenge to fault diagnosis.

[0003] More seriously, there is a many-to-many mapping between abnormal phenomena and inherent faults in crushing equipment. A vague term such as "high temperature" may correspond to multiple physical causes, such as overheating of the liner friction, overcurrent temperature rise in the winding, increased oil temperature, or subjective sensations experienced by personnel. This semantic uncertainty causes serious ambiguity when the system maps natural language to standard fault codes. If "periodic impact sound" is misjudged as "ingress of uncrushable material" instead of the actual "pitting of gear teeth," the subsequent maintenance plan—including inspection procedures, spare parts lists, and safe operating procedures—will be completely deviated from reality. This not only wastes time and spare parts resources but may also lead to increased crushing power consumption instead of reduced consumption due to improper disassembly or adjustment that damages the original energy consumption optimization parameters of the equipment, or even cause secondary damage or safety accidents.

[0004] Furthermore, the operation and maintenance decisions for mining crushing equipment exhibit significant energy-saving sensitivity. Reasonable maintenance can restore crushing efficiency and reduce power consumption per unit processing volume, while excessive or delayed maintenance exacerbates energy loss. Currently, there is a lack of a systematic method that can automatically disambiguate and accurately generate optimal operation and maintenance solutions that balance equipment recovery and energy consumption, starting from non-standardized descriptions. Traditional models relying on human experience are prone to misjudgment and neglect energy consumption factors. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention discloses a method and system for generating energy-saving operation and maintenance schemes for mining crushing equipment. The aim is to solve the technical problem of low accuracy and poor reliability in generating maintenance schemes due to the ambiguity and non-standardization of equipment fault descriptions in existing technologies.

[0006] The technical solution of the present invention is as follows: In a first aspect, this invention discloses a method for generating an energy-saving operation and maintenance scheme for mining crushing equipment, the method comprising: Obtain initial description information for the target device, which includes at least one feature item to be clarified; Based on the preset mapping association rules, multiple candidate standard state features corresponding to the feature items to be clarified are retrieved, and dimensional information used to distinguish multiple candidate standard state features is extracted. Based on dimensional information, guide instructions are generated and output to obtain supplementary feedback information for the feature items to be clarified; Based on the supplementary feedback information, the target standard state feature is determined from multiple candidate standard state features; Replace or associate the features to be clarified in the initial description information with the target standard state features, and generate a maintenance plan for the target equipment based on the replaced or associated initial description information.

[0007] This technical solution establishes a transformation path from vague, unstructured initial descriptions to precise, standardized fault characteristics. Through proactive interactive clarification, rather than passive guessing, it fundamentally solves the problem of diagnostic errors caused by information ambiguity, significantly improving the accuracy and reliability of maintenance solutions.

[0008] Furthermore, after obtaining the initial description information for the target device, the process also includes: The initial description information is compared with a pre-defined list of fuzzy words. If there are words in the initial description information that match the fuzzy word list, then the matched words are identified as features to be clarified.

[0009] Furthermore, before retrieving multiple candidate standard state features corresponding to the feature item to be clarified according to the preset mapping association rules, the process also includes: Acquire the current operating condition data of the target device. The current operating condition data is used to characterize the current energy output status or operation cycle stage of the target device. The steps of retrieving multiple candidate standard state features corresponding to the feature item to be clarified, based on preset mapping and association rules, and extracting dimensional information used to distinguish multiple candidate standard state features include: By using current operating condition data to limit the boundaries of mapping association rules and eliminating redundant features that do not have the possibility of physical existence under the current operating conditions, the target mapping range is obtained. Within the target mapping range, retrieve multiple candidate standard state features corresponding to the feature terms to be clarified, and extract dimensional information used to distinguish multiple candidate standard state features.

[0010] Furthermore, the steps of retrieving multiple candidate standard state features corresponding to the feature terms to be clarified within the target mapping range, and extracting dimensional information used to distinguish the multiple candidate standard state features, include: Within the target mapping range, a preset first information structure corresponding to the feature item to be clarified is obtained by querying. The first information structure stores the identifier of the feature item to be clarified, multiple candidate standard state features associated with the identifier of the feature item to be clarified, and specific attributes or operating conditions required to distinguish different candidate standard state features. Specific attributes or operating conditions in the first information structure obtained from the query are identified as dimension information.

[0011] Furthermore, the steps of generating and outputting guiding instructions based on dimensional information to obtain supplementary feedback information for the feature terms to be clarified include: Based on dimensional information, the corresponding list of optional answers is extracted from the preset option library, and a guiding instruction containing one or more optional answers corresponding to specific attributes or operating conditions is generated. Output guidance instructions to guide the user to choose from one or more optional answers, and provide the user's selection results as supplementary feedback.

[0012] Furthermore, the options corresponding to the guidance instructions include options that cannot be directly observed; After obtaining supplementary feedback information for the features to be clarified, it also includes: If the supplementary feedback information indicates that the option cannot be directly observed, then the observation constraints for the feature to be clarified are identified based on the equipment model of the target device. Based on the observation constraints, obtain second feature information that is correlated with the feature term to be clarified and can be perceived under the observation constraints; The guidance instructions are regenerated and output based on the second feature information to obtain feedback results on the second feature information, and the feedback results are used as new supplementary feedback information.

[0013] Furthermore, based on supplementary feedback information, the steps for determining the target standard state feature from multiple candidate standard state features include: Based on the supplementary feedback information, the candidate standard state feature with the highest correlation to the supplementary feedback information is identified using the preset state deduction logic, and then determined as the target standard state feature.

[0014] Furthermore, after replacing or associating the features to be clarified in the initial descriptive information with the target standard state features, it also includes: Check if there are any unclarified features in the initial description information after replacement or association; If they exist, the steps of retrieving multiple candidate standard state features corresponding to the feature items to be clarified according to the preset mapping association rules are repeated to carry out an iterative interactive information clarification process until all the feature items to be clarified in the initial description information are converted into target standard state features.

[0015] Furthermore, the steps for generating a repair plan for the target device based on the replaced or associated initial description information include: Based on the initial description information after replacement or association, standardized fault information for the target device is generated. Based on standardized fault information, a search and matching process is performed in a pre-set maintenance knowledge base to obtain at least one candidate fault mode and at least one candidate maintenance action associated with each candidate fault mode. For at least one candidate maintenance action retrieved, combine and logically verify the constraints stored in the maintenance knowledge base to generate at least one candidate maintenance solution; Output at least one candidate repair solution, and associate the output with the basis information for deriving the candidate repair solution from the standardized fault information.

[0016] Secondly, the present invention also discloses a maintenance scheme generation system for mining crushing equipment, used to perform the steps in the aforementioned method, including: The information acquisition module is used to acquire initial description information for the target device, which includes at least one feature item to be clarified. The feature retrieval module is used to retrieve multiple candidate standard state features corresponding to the feature item to be clarified according to the preset mapping association rules, and extract dimensional information used to distinguish multiple candidate standard state features. The guidance feedback module is used to generate and output guidance instructions based on dimensional information in order to obtain supplementary feedback information for the feature items to be clarified. The feature determination module is used to determine the target standard state feature from multiple candidate standard state features based on supplementary feedback information; The solution generation module is used to replace or associate the features to be clarified in the initial description information with target standard state features, and generate a maintenance solution for the target equipment based on the replaced or associated initial description information.

[0017] This technical solution provides a physical device capable of implementing the above methods, solidifying the method flow into functional modules and providing specific system architecture support for the practical application of the technology.

[0018] In summary, this invention provides a method and system for generating energy-saving operation and maintenance solutions for mining crushing equipment. The method establishes a proactive, interactive information clarification mechanism. When the system identifies vague descriptions such as "abnormal noise" or "heating," it does not perform speculative matching. Instead, it searches the knowledge base to find multiple precise and standardized state characteristics that might correspond to the vague term, and extracts key dimension information distinguishing these states, such as the frequency of the sound, the specific temperature value, or the operating condition. Subsequently, based on this dimension information, the system automatically generates guiding questions and presents them to the user, transforming the complex diagnostic process into simple and clear multiple-choice questions. Through simple feedback, the user can help the system eliminate ambiguity and pinpoint the unique target standard state characteristic. This process successfully transforms the originally vague and unstructured natural language input into structured data that the machine can accurately understand and process. Based on this highly defined standardized information, the system's subsequent retrieval and matching in the maintenance knowledge base becomes extremely accurate, enabling the generation of logically sound, accurate, and highly executable maintenance solutions. This fundamentally solves a series of chain problems caused by inaccurate input information, significantly improving the efficiency and quality of equipment maintenance. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for generating an energy-saving operation and maintenance scheme for mining crushing equipment, as provided in an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of a mining crushing equipment maintenance solution generation system provided in an embodiment of the present invention.

[0021] Labeling Explanation: 210, Information Acquisition Module; 220, Feature Retrieval Module; 230, Guidance Feedback Module; 240, Feature Determination Module; 250, Solution Generation Module. Detailed Implementation

[0022] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of this invention. The components of this invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0023] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] In real-world industrial production and equipment maintenance scenarios, a common challenge stems from the ambiguity in describing abnormal equipment conditions. Take a mining machinery unit as an example: when a field inspector submits a maintenance request via a handheld terminal stating, "Pump B at pump station 2 is making a bit of noise," this information becomes the starting point of the entire maintenance process. However, the description "a bit of noise" carries significant uncertainty. Experienced technicians might guess based on past experience, but for an automated maintenance diagnostic system or an inexperienced new employee, this information is almost unusable. This "noise" could be a sharp friction sound from poor lubrication of the rolling bearing, a hissing sound similar to boiling water due to cavitation caused by low pressure at the pump inlet, or a periodic impact sound caused by misalignment of the coupling between the pump and motor. These three scenarios correspond to drastically different root causes and maintenance solutions. If the system incorrectly identifies cavitation noise as a bearing problem and generates a work order to replace the bearing, the maintenance personnel will not only fail to solve the problem but also waste valuable downtime and spare parts resources. The underlying fault remains, and it may even exacerbate equipment damage.

[0025] Firstly, please see Figure 1 This invention provides a method for generating an energy-saving operation and maintenance scheme for mining crushing equipment, the method comprising: S1. Obtain initial description information for the target device, the initial description information containing at least one feature item to be clarified; S2. Based on the preset mapping association rules, retrieve multiple candidate standard state features corresponding to the feature items to be clarified, and extract the dimensional information used to distinguish multiple candidate standard state features. S3. Generate and output guidance instructions based on dimensional information to obtain supplementary feedback information for the feature items to be clarified; S4. Based on the supplementary feedback information, determine the target standard state feature from multiple candidate standard state features; S5. Replace or associate the features to be clarified in the initial description information with the target standard state features, and generate a maintenance plan for the target equipment based on the replaced or associated initial description information.

[0026] Specifically, initial description information refers to the raw, unprocessed fault description provided by users, such as field maintenance personnel or equipment operators, through various input methods. This information is typically natural language text and may contain colloquialisms, slang, subjective feelings, and uncertain quantifiers, such as "Pump B at pump station 2B is a bit noisy" in the aforementioned example.

[0027] Features requiring clarification refer to words or phrases in the initial descriptive information that are ambiguous in meaning and have multiple possible interpretations. They are the key to the uncertainty of information, such as the word "loud" in "a little loud".

[0028] Candidate standard state characteristics refer to multiple precise, standardized, and unambiguous engineering and technical terms that may correspond to a characteristic item to be clarified. Each standard state characteristic represents a specific, measurable physical phenomenon or equipment condition. For example, for "noise," candidate standard state characteristics may include "early wear noise of rolling bearings," "cavitation noise at pump inlet," "vibration noise due to misalignment of couplings," and "resonance noise due to loose protective cover," etc.

[0029] Dimensional information refers to a set of key attributes or conditions used to distinguish and identify the characteristics of different candidate standard states. These dimensions form the basis for making effective questions. For example, to distinguish the above-mentioned sounds, dimensional information may include the pitch of the sound (high frequency, mid frequency, low frequency), the rhythm of the sound (continuous, intermittent, periodic), and the correlation between the sound and the operating conditions of the equipment (occurring only at startup, occurring only under high load), etc.

[0030] Guiding instructions are interactive commands generated by the system based on dimensional information, used to ask users questions to obtain supplementary information. They typically appear in the form of multiple-choice questions, true / false questions, or guided fill-in-the-blank questions, aiming to translate professional diagnostic logic into questions that ordinary users can understand and answer.

[0031] Supplementary feedback information is the user's response to the guidance instructions, that is, the user's selection results or input content.

[0032] The target standard state feature refers to the standard state feature that the system ultimately determines from multiple candidate standard state features based on the user's supplementary feedback information, and that best matches the actual situation.

[0033] In a specific embodiment, the execution process of the energy-saving operation and maintenance scheme generation method for mining crushing equipment is as follows. First, the system receives initial description information submitted by on-site personnel via a mobile application through equipment management software deployed on the factory server. The content is "Pump B of Pump Station No. 2 is making a bit of noise." After receiving this text information, the system enters the information processing flow.

[0034] The system analyzes the text and identifies the feature term "loudness" that needs clarification. Subsequently, the system accesses its internal knowledge base and retrieves information based on preset mapping and association rules. These rules are represented in the database as an association table, establishing a correspondence between fuzzy words and standard features. The system finds that the feature term "loudness" corresponds to three candidate standard state features: "early wear noise of rolling bearings," "cavitation noise at pump inlet," and "vibration noise due to misalignment of couplings." Simultaneously, the system extracts the dimensional information used to distinguish these three candidate features from the rules: "sound pitch."

[0035] Next, the system generates a guidance instruction based on the "sound pitch" dimension. This instruction is pushed to the on-site personnel who submitted the repair request through the mobile application interface in the form of a question: "Which of the following descriptions is the sound you heard closest to?", with three options: "A. Sharp, metallic friction sound", "B. Continuous, hissing sound like boiling water", and "C. Dull, rhythmic impact sound".

[0036] Based on their actual listening experience near the equipment, the on-site personnel selected option A on the mobile application interface. This selection result "A" was sent back to the server as supplementary feedback.

[0037] After receiving the supplementary feedback, the system on the server, based on its internal reasoning logic, associated option A with the candidate standard state characteristic of "early wear noise in rolling bearings." Because a sharp metallic friction sound is a typical manifestation of bearing wear, the system determined "early wear noise in rolling bearings" as the target standard state characteristic for this event.

[0038] Finally, the system internally transforms the original description "Pump B at Pump Station No. 2 is making some noise" into a standardized fault description: "Equipment 'Pump B at Pump Station No. 2' exhibits 'early wear noise in rolling bearing'." Based on this precise and unambiguous information, the system further retrieves and generates targeted maintenance solutions from the maintenance knowledge base, such as: "Step 1: Shut down the machine and disconnect the power; Step 2: Remove the bearing cover on the pump body coupling side; Step 3: Inspect and replace the specified model of rolling bearing; Step 4: Reassemble and perform alignment calibration; Step 5: Start the equipment and monitor the operating sound and temperature to confirm that the abnormal noise has been eliminated." Through this interactive clarification approach, the method transforms a vague initial description that may lead to misjudgment into precise, standardized information that can be directly used to generate the correct maintenance plan, effectively avoiding resource waste and maintenance delays caused by guesswork and information uncertainty in traditional processes.

[0039] In a preferred embodiment, in order to achieve automated identification of features to be clarified in the initial description information, after obtaining the initial description information for the target device, the method further includes: The initial description information is compared with a pre-defined list of fuzzy words. If there are words in the initial description information that match the fuzzy word list, then the matched words are identified as features to be clarified.

[0040] In the above process, the system is able to proactively identify the need for clarification of the word "sound" because it has a pre-configured list of ambiguous terms. This list is a collection that stores all ambiguous and unclear words commonly used in equipment maintenance. The construction of this list can be based on statistical analysis of large amounts of historical maintenance records, or on summarization by domain experts.

[0041] An example of a specific list of vague terms may include the following words: loud, unusual noise, loud sound, shout, heat, hot, high temperature, shake, vibrate, sway, leak, seep, drip, weak, powerless, slow, stuck, pause, alarm, trip, etc.

[0042] When the system receives the initial description "Pump B at Pump Station No. 2 is making a bit of a noise," it scans the text character by character or according to word segmentation results, matching it against each word in the fuzzy word list. In this example, the system detects that the word "noise" in the text perfectly matches the word "noise" in the list. Therefore, the system automatically marks the word "noise" as a feature item to be clarified, triggering subsequent retrieval and interaction processes. If the initial description is "Pump B's casing is very hot," the system will similarly identify it as a feature item to be clarified by matching the word "hot" in the list, and further inquire about the specific circumstances of "hot," such as whether it is a feeling by touch or a specific temperature reading, thus distinguishing between subjective judgment and over-temperature alarm measured by sensors. This automated recognition mechanism based on a preset list allows the entire clarification process to be triggered seamlessly and efficiently, without manual intervention to determine which part of the information is ambiguous, greatly improving processing efficiency.

[0043] In a preferred embodiment, in a real industrial environment, equipment anomalies are often closely related to its operating status. To make the clarification process more intelligent and avoid asking users meaningless questions in the current scenario, before retrieving multiple candidate standard state features corresponding to the feature item to be clarified according to preset mapping association rules, the method further includes: Acquire the current operating condition data of the target device. The current operating condition data is used to characterize the current energy output status or operation cycle stage of the target device. The steps of retrieving multiple candidate standard state features corresponding to the feature item to be clarified, based on preset mapping and association rules, and extracting dimensional information used to distinguish multiple candidate standard state features include: By using current operating condition data to limit the boundaries of mapping association rules and eliminating redundant features that do not have the possibility of physical existence under the current operating conditions, the target mapping range is obtained. Within the target mapping range, retrieve multiple candidate standard state features corresponding to the feature terms to be clarified, and extract dimensional information used to distinguish multiple candidate standard state features.

[0044] Current operating condition data refers to a set of parameters that reflect the real-time operating status of equipment. This data can typically be obtained from the factory's distributed control system (DCS), programmable logic controller (PLC), or sensors directly connected to the equipment. For a pump, its operating condition data may include motor speed, outlet flow rate, outlet pressure, motor operating current, and equipment start / stop status.

[0045] Continuing with the aforementioned scenario, suppose that when the system receives the description "Pump B is making some noise," it not only receives the text information but also synchronously obtains the current operating status data of Pump B through the interface. The data shows "Motor speed: 0 RPM." This means that the pump is currently in a stopped state.

[0046] At this point, after retrieving the three candidate standard state features corresponding to "noise" (early wear noise of rolling bearings, cavitation noise at the pump inlet, and vibration noise from misaligned couplings), the system does not immediately generate a question. It first performs a boundary-limiting step. In the system's knowledge base, each candidate standard state feature is associated with its physical preconditions. For example, the conditions for "cavitation noise at the pump inlet" are "motor speed > preset threshold (e.g., 500 RPM)" and "inlet pressure < vaporization pressure." Since the currently acquired operating data is "motor speed: 0 RPM," which does not meet the necessary conditions for cavitation, the system determines that the candidate feature "cavitation noise at the pump inlet" has no physical possibility of existence under the current operating conditions and removes it from the candidate list. Similarly, "vibration noise from misaligned couplings" usually only becomes apparent when the equipment is running. After such removal, the target mapping range may only have "early wear noise of rolling bearings" as the most likely candidate, because even when the machine is stopped, the sound of a damaged bearing can be heard by manually turning the machine.

[0047] In this scenario, the system might not even need to ask the user any further questions, directly identifying "early wear noise in rolling bearings" as the target standard condition characteristic. Alternatively, if other abnormal noises might occur even when the machine is stopped, such as "internal foreign object obstruction," the system's generated guidance instructions will only include options related to these possibilities, excluding those related to "cavitation." This mechanism of pre-screening using real-time operating data significantly improves the accuracy and efficiency of the interaction, avoiding presenting users with questions that don't reflect the actual situation on-site, making the entire diagnostic process more professional and intelligent.

[0048] To enable the systematic management and efficient querying of fuzzy words, standard features, and the dimensional information required to distinguish them, this invention further proposes a specific data organization method. The steps of retrieving multiple candidate standard state features corresponding to the feature item to be clarified within the target mapping range, and extracting the dimensional information used to distinguish the multiple candidate standard state features, include: Within the target mapping range, a preset first information structure corresponding to the feature item to be clarified is obtained by querying. The first information structure stores the identifier of the feature item to be clarified, multiple candidate standard state features associated with the identifier of the feature item to be clarified, and specific attributes or operating conditions required to distinguish different candidate standard state features. Specific attributes or operating conditions in the first information structure obtained from the query are identified as dimension information.

[0049] This first information structure is essentially a structured data record, which can be a row in a database table or a data object in JSON or XML format. It organizes all the information related to a feature to be clarified, making it easier for a computer to process.

[0050] Below is an example of the first information structure for the feature to be clarified, "abnormal noise," presented in an easy-to-understand format: { "Identifier of Feature Items to be Clarified": "Abnormal Noise" "Associated candidate standard state features":[ {"Feature ID":"F001","Feature Name":"Rapid Bearing Early Wear Noise"}, {"Feature ID":"F002","Feature Name":"Pump Inlet Cavitation Noise","Constraint Condition":"Speed ​​Speed ​​> 500 RPM"}, {"Feature ID":"F003","Feature Name":"Coupling Misalignment Vibration Sound","Constraint Condition":"Speed ​​> 100 RPM"} ], "Distinguish between specific attributes or operating conditions required":[ { "Attribute ID":"A01", "Attribute Name":"Voice Pitch", Prompt text: "Please describe the tonal characteristics of the sound." }, { "Attribute ID":"A02", "Attribute Name": "Occurring Operating Condition", Prompt text: "Under what operating conditions is the sound most noticeable?" } ] } When the system needs to handle "abnormal noises," it directly queries and loads this structure. From the list of "associated candidate standard state features," the system can obtain all possible standard interpretations. From the list of "specific attributes or operating conditions required for differentiation," the system can determine what information needs to be obtained from the user to make a judgment, namely, the dimensions of "sound pitch" and "occurrence condition." This structured data storage method makes adding, modifying, and maintaining knowledge very convenient. For example, when a new type of abnormal noise is discovered, it is only necessary to add a new candidate standard state feature to this structure and update the corresponding differentiation attributes, without modifying the program's hard-coded logic.

[0051] To further enhance the user-friendliness of the interaction and the standardization of feedback information, this invention optimizes the generation method of guidance instructions. The steps of generating and outputting guidance instructions based on dimensional information to obtain supplementary feedback information for the feature items to be clarified include: Based on dimensional information, the corresponding list of optional answers is extracted from the preset option library, and a guiding instruction containing one or more optional answers corresponding to specific attributes or operating conditions is generated. Output guidance instructions to guide the user to choose from one or more optional answers, and provide the user's selection results as supplementary feedback.

[0052] This means that the system does not dynamically generate answer options for questions, but rather extracts them from a pre-set "option library." This option library stores standardized answer options for various "dimensional information" (i.e., specific attributes).

[0053] Continuing with the example of "voice tone" (attribute ID A01), after determining that a query for voice tone is needed, the system will use "A01" as an index to search the option library. The records in the option library associated with "A01" might look like this: { "Associated Attribute ID":"A01", "List of optional answers":[ {"Answer ID":"Opt01_1","Answer Text":"High-frequency sharp metallic friction sound","Association Feature ID":["F001"]}, {"Answer ID":"Opt01_2","Answer Text":"A continuous hissing sound similar to boiling water","Association Feature ID":["F002"]}, {"Answer ID":"Opt01_3","Answer Text":"Dull and rhythmic impact sound","Associated Feature ID":["F003"]} ] } Once the system obtains this list of possible answers, it can construct a standard multiple-choice question and present the "answer text" as options to the user. When the user selects an option, such as "high-frequency, sharp metallic friction sound," the system receives not only the text but, more importantly, its corresponding "Answer ID" (Opt01_1) and the pre-associated "Feature ID" (F001). In this way, the system can definitively infer that the target standard state feature is "early wear noise in rolling bearings." This method of generating questions based on an option library ensures the uniformity and standardization of the interactive interface, avoids secondary ambiguity caused by free text input by the user, and creates a closed loop in the entire information clarification process, from standardized questions to standardized answers to standardized conclusions, greatly guaranteeing the accuracy and automation of data processing.

[0054] In some complex field environments, even experienced maintenance personnel may face difficulties in observation. For example, when the pump requiring diagnosis is installed in a sealed, soundproof enclosure, or when the surrounding noise is so high that it is impossible to clearly hear the sound emitted by the equipment, the aforementioned interactive clarification based on sound characteristics will be hindered.

[0055] To address this situation, the present invention enhances the robustness of the interaction process. The options corresponding to the guidance instructions include those that cannot be directly observed; after obtaining supplementary feedback information for the feature to be clarified, it also includes: If the supplementary feedback information indicates that the option cannot be directly observed, then the observation constraints for the feature to be clarified are identified based on the equipment model of the target device. Based on the observation constraints, obtain second feature information that is correlated with the feature term to be clarified and can be perceived under the observation constraints; The guidance instructions are regenerated and output based on the second feature information to obtain feedback results on the second feature information, and the feedback results are used as new supplementary feedback information.

[0056] In a specific implementation scenario, on-site personnel reported a fault, with the initial description being "abnormal noise from pump B at pump station No. 2." The "abnormal noise" was identified as a feature requiring clarification.

[0057] Based on preset mapping and association rules, the system retrieves multiple candidate standard state features corresponding to "abnormal noise," such as "early wear noise of rolling bearings" and "cavitation noise of pump body." To distinguish these candidate features, the system extracts "pitch" as key dimension information and generates a guiding instruction: "Please determine the pitch characteristic of the abnormal noise? Options: A. High-frequency sharp; B. Low-frequency muffled; C. Cannot be heard / cannot be observed." Due to background noise interference in the pump room, on-site personnel were unable to accurately determine the pitch, so they selected option C: "Unable to hear clearly / Unable to observe." This selection, as supplementary feedback, triggered the system's roundabout diagnostic logic.

[0058] First, based on the equipment model of the target equipment "Pump B of Pump Station No. 2", the system queries the equipment archive for its installation and operating environment information and identifies the current observation constraint as "high background noise environment, low signal-to-noise ratio of sound characteristics".

[0059] Next, the system searches its knowledge base for a second feature that is strongly correlated with the current feature to be clarified ("abnormal noise") and is perceptible under the constraint of "high background noise". The knowledge base association rules indicate that "early wear of rolling bearings" faults may not only produce abnormal noise, but are usually accompanied by "abnormal vibration of bearing housing" and "increased temperature of bearing housing". Vibration and temperature can be effectively measured by instruments and are not affected by noise, so they are determined by the system to be feasible alternative observation points.

[0060] Subsequently, the system regenerates guidance instructions based on the perceptible second feature information. For example, it pushes a new question: "Please measure the effective value of the vertical vibration velocity of the pump coupling side bearing housing and select the range: A. less than 3.5 mm / s; B. 3.5 mm / s - 7.1 mm / s; C. greater than 7.1 mm / s." After the on-site personnel measure, they select C (greater than 7.1 mm / s).

[0061] The system uses the feedback results on vibration values ​​as new supplementary feedback information. Utilizing preset state deduction logic (for example, the correlation between significantly exceeding the standard vibration value and bearing wear is the highest), it reversely locks the most likely target standard state feature as "early wear noise of rolling bearings," thus successfully bypassing the obstacle of direct auditory observation and completing information clarification.

[0062] This mechanism enables the diagnostic process to intelligently find and utilize alternative paths for reasoning when information access is blocked, significantly enhancing the adaptability and robustness of the interactive process.

[0063] After the system collects supplementary feedback from users, a clear logic is needed to narrow down the final target from multiple candidates. The steps for determining the target standard state features from multiple candidate standard state features based on the supplementary feedback include: Based on the supplementary feedback information, the candidate standard state feature with the highest correlation to the supplementary feedback information is identified using the preset state deduction logic, and then determined as the target standard state feature.

[0064] This state deduction logic is the core of the system's decision-making process, and it can be implemented in various forms. A simple and direct approach is to pre-label the candidate standard state characteristics that each answer option points to when building the option library. As in the example above, the option "high-frequency sharp metallic friction sound" is directly associated with "early wear noise of rolling bearings".

[0065] In more complex scenarios, state deduction logic can be a system based on weighted scoring. Suppose that clarifying an "abnormal noise" involves two questions: Q1 is about the pitch, and Q2 is about the operating condition in which it occurred.

[0066] For Q1, if the user selects "high-frequency sharp sound", the answer could be "rolling bearing wear" +5 points or "coupling misalignment" +1 point.

[0067] For Q2, if the user selects "noticeable only at high speeds", the answer could be "rolling bearing wear" +2 points or "coupling misalignment" +5 points.

[0068] After the user answers the two questions, the system accumulates the scores for each candidate standard state feature. In this example, "rolling bearing wear" has a total score of 7 points, and "coupling misalignment" has a total score of 6 points. The system selects "rolling bearing wear," with the highest score, as the candidate standard state feature with the highest relevance and determines it as the target standard state feature. This decision-making process based on state deduction logic ensures that the transformation from user feedback to the final conclusion is based on preset, rigorous rules, rather than random guessing, thus guaranteeing the accuracy of the clarification results.

[0069] When dealing with complex faults, the initial description may contain multiple points of ambiguity. To ensure that all information is clarified, this invention introduces an iterative processing mechanism. After replacing or associating the features to be clarified in the initial description information with target standard state features, the process further includes: Check if there are any unclarified features in the initial description information after replacement or association; If they exist, the steps of retrieving multiple candidate standard state features corresponding to the feature items to be clarified according to the preset mapping association rules are repeated to carry out an iterative interactive information clarification process until all the feature items to be clarified in the initial description information are converted into target standard state features.

[0070] For example, an initial description might read: "Pump B at Pump Station No. 2 has been making a lot of noise lately, and its casing feels very hot to the touch." During the initial analysis, the system will compare this information with a list of fuzzy terms and identify two features that need clarification: "very loud" and "very hot."

[0071] The system first processes the first feature to be clarified, "loud noise". Through the aforementioned interactive clarification process, the system may successfully transform it into the target standard state feature, "pump inlet cavitation noise".

[0072] After completing this step, the system does not immediately end, but re-examines the processed information. It finds that the feature item "very hot" still exists and requires clarification. Therefore, the system starts a new clarification cycle, this time targeting "very hot".

[0073] The system will search for candidate standard status features related to "very hot", such as "overheating due to poor bearing lubrication", "overheating due to motor coil overload", or "excessively high pump medium temperature". Then, it will generate new guidance instructions, such as: "Regarding 'very hot', please confirm the specific situation: A. Only the bearing housing temperature is high; B. The temperature of the entire motor housing is rising uniformly; C. The pump inlet and outlet pipe temperatures are already very high."

[0074] After on-site personnel selected option B based on the actual situation, the system clarified "very hot" as "motor coil overload." At this point, the system rechecked the information and found that all features requiring clarification had been addressed. Thus, the entire iterative interactive information clarification process concluded. Ultimately, the original vague description was completely transformed into a standardized fault message with complete information: "Equipment 'Pump B of Pump Station No. 2' exhibits 'cavitation noise at the pump inlet,' accompanied by 'motor coil overload.'" This iterative mechanism ensures that no ambiguous point is overlooked, and can completely and accurately reconstruct the complex fault picture.

[0075] After all ambiguous information has been clarified and transformed into standardized fault information, the final stage of generating a repair plan begins. The steps for generating a repair plan for the target equipment based on the initial description information after replacement or association include: Based on the initial description information after replacement or association, standardized fault information for the target device is generated. Based on standardized fault information, a search and matching process is performed in a pre-set maintenance knowledge base to obtain at least one candidate fault mode and at least one candidate maintenance action associated with each candidate fault mode. For at least one candidate maintenance action retrieved, combine and logically verify the constraints stored in the maintenance knowledge base to generate at least one candidate maintenance solution; Output at least one candidate repair solution, and associate the output with the basis information for deriving the candidate repair solution from the standardized fault information.

[0076] Taking the results of the aforementioned iterative clarification as an example, the standardized fault information obtained by the system is: "Equipment 'Pump B of Pump Station No. 2' exhibits 'cavitation noise at the pump inlet', accompanied by 'overload heating of the motor coil'."

[0077] The system uses this standard information as keywords to search the maintenance knowledge base. The knowledge base is a structured database that stores fault symptoms, fault modes, maintenance actions, and the relationships between them.

[0078] The search results may look like the following: Matching "pump inlet cavitation noise" yields the candidate fault mode "inlet flow channel blockage or inlet valve not fully open", with the associated candidate maintenance actions being: "check and clean the pump inlet filter" and "check and confirm the inlet pipeline valve opening".

[0079] Matching "motor coil overload" yields the candidate fault mode "motor overload", and the associated candidate repair actions are: "use a clamp meter to measure the motor operating current", "check if the power supply voltage is stable", and "check if the motor cooling fan is intact and clean the dust".

[0080] Next, the system needs to combine these scattered maintenance actions into a logically coherent and executable maintenance plan. This is where the constraints stored in the maintenance knowledge base come into play. For example, constraints might stipulate: "Mechanical overload factors should be ruled out before performing any electrical measurements (such as current measurement)" and "Actions involving shutdown should be performed in a centralized manner."

[0081] Based on these constraints, the system performs combination and logic verification to generate a candidate maintenance plan: "Plan A: 1. (Safety first) Shut down, disconnect power, and tag; 2. Check and clean the pump inlet filter; 3. Check and confirm the opening of the inlet pipeline valve; 4. Check and clean the motor cooling fan; 5. (Post-operation verification) Power on and start the machine, and use a clamp meter to measure the motor operating current under rated conditions to confirm whether it has returned to the normal range." Finally, when presenting this repair plan to the user, the system also includes the following reasoning: "The basis for generating this plan is: 1. Through interactive clarification, the on-site phenomena were confirmed as 'cavitation noise' and 'motor overheating'. 2. 'Cavitation noise' points to a problem with the inlet flow channel, therefore, checking the filter and valves is recommended. 3. 'Motor overheating' points to overload; after ruling out possible mechanical causes, final confirmation is required by measuring the current. 4. The operation sequence follows the safety and logical principle of mechanical before electrical, and shutdown before verification." This method of providing supporting reasoning greatly enhances the credibility and transparency of the plan, facilitating review and decision-making by the maintenance team leader, and providing a clear background and thought process for the implementers.

[0082] Secondly, see Figure 2 The present invention also provides a maintenance scheme generation system for mining crushing equipment, used to perform the steps in the aforementioned method, including: Information acquisition module 210 is used to acquire initial description information for the target device, the initial description information containing at least one feature item to be clarified; The feature retrieval module 220 is used to retrieve multiple candidate standard state features corresponding to the feature item to be clarified according to the preset mapping association rules, and extract dimensional information used to distinguish multiple candidate standard state features. The guidance feedback module 230 is used to generate and output guidance instructions based on dimensional information in order to obtain supplementary feedback information for the feature items to be clarified. The feature determination module 240 is used to determine the target standard state feature from multiple candidate standard state features based on supplementary feedback information. The solution generation module 250 is used to replace or associate the feature items to be clarified in the initial description information with target standard state features, and generate a maintenance solution for the target equipment based on the replaced or associated initial description information.

[0083] In a specific system implementation, the functions of these modules can be specified as follows: The information acquisition module 210 is used to receive JSON format data sent via HTTP requests from mobile applications, web forms, or third-party systems. This module is responsible for parsing the incoming data and extracting the device identifier, the natural language description text entered by the user, and any accompanying device operating status data.

[0084] After receiving the feature items to be clarified from the information acquisition module 210, the feature retrieval module 220 performs a series of database query operations to retrieve the corresponding candidate standard state features, distinguishing dimensions, and related constraints. The implementation of this module can rely on SQL queries, graph database queries, or other efficient data retrieval technologies.

[0085] The guidance feedback module 230 dynamically generates a JSON or HTML structure for asking questions based on the dimensional information provided by the feature retrieval module 220 and the content in the option library, and returns it to the front-end application for rendering via API response. Simultaneously, this module also includes an interface for receiving user feedback, passing the user's selection results to the feature determination module 240.

[0086] The feature determination module 240 can be a simple rule engine or a more complex scoring or probability model. After receiving user feedback, this module executes an inference algorithm to calculate the candidate standard state features with the highest correlation to the feedback, and outputs the finally determined target standard state features.

[0087] The solution generation module 250 receives fully standardized fault information output by the feature determination module 240 and uses this as input to perform deep retrieval and matching in the maintenance knowledge base. This module contains a set of logical rules for combining and verifying maintenance actions, ensuring that the generated maintenance solution is reasonable and feasible in terms of operation sequence, resource requirements, and safety specifications. Finally, the module packages the generated maintenance solution along with its derivation basis and outputs it via API.

[0088] These modules work together to form a complete and automated processing chain, transforming a vague maintenance request into a precise and executable maintenance plan through structured interaction and intelligent reasoning, providing strong technical support for the intelligent operation and maintenance of industrial equipment.

[0089] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for generating an energy-saving operation and maintenance scheme for mining crushing equipment, characterized in that, The method includes: Obtain initial description information for the target device, the initial description information containing at least one feature item to be clarified; According to the preset mapping association rules, multiple candidate standard state features corresponding to the feature item to be clarified are retrieved, and dimensional information used to distinguish the multiple candidate standard state features is extracted. Based on the dimensional information, a guidance instruction is generated and output to obtain supplementary feedback information for the feature item to be clarified; Based on the supplementary feedback information, the target standard state feature is determined from the plurality of candidate standard state features; The feature items to be clarified in the initial description information are replaced or associated with the target standard state features, and a maintenance plan for the target device is generated based on the replaced or associated initial description information.

2. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 1, characterized in that, After obtaining the initial description information for the target device, the process further includes: The initial description information is compared with a preset list of fuzzy words. If there is a word in the initial description information that matches the list of fuzzy words, then the matched word is determined as the feature item to be clarified.

3. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 1, characterized in that, Before retrieving multiple candidate standard state features corresponding to the feature item to be clarified according to preset mapping association rules, the method further includes: Acquire the current operating condition data of the target device, wherein the current operating condition data is used to characterize the current energy output state or operation cycle stage of the target device; The step of retrieving multiple candidate standard state features corresponding to the feature item to be clarified according to a preset mapping association rule, and extracting dimensional information for distinguishing the multiple candidate standard state features includes: The mapping association rules are bounded using the current operating condition data, and redundant features that do not have a physical possibility of existence under the current operating condition are eliminated to obtain the target mapping range. Within the target mapping range, retrieve multiple candidate standard state features corresponding to the feature item to be clarified, and extract dimensional information used to distinguish the multiple candidate standard state features.

4. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 3, characterized in that, The step of retrieving multiple candidate standard state features corresponding to the feature item to be clarified within the target mapping range, and extracting dimensional information for distinguishing the multiple candidate standard state features includes: Within the target mapping range, a preset first information structure corresponding to the feature item to be clarified is obtained by querying. The first information structure stores the feature item identifier to be clarified, multiple candidate standard state features associated with the feature item identifier to be clarified, and specific attributes or operating conditions required to distinguish different candidate standard state features. The specific attributes or operating conditions in the first information structure obtained from the query are determined as the dimension information.

5. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 4, characterized in that, The step of generating and outputting guidance instructions based on the dimensional information to obtain supplementary feedback information for the feature item to be clarified includes: Based on the dimensional information, a list of corresponding optional answers is extracted from a preset option library, and a guiding instruction containing one or more optional answers corresponding to the specific attribute or operating condition is generated; Output the guidance instruction to guide the user to select from one or more optional answers, and use the user's selection result as the supplementary feedback information.

6. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 1, characterized in that, The options corresponding to the guidance instructions include options that cannot be directly observed; After obtaining the supplementary feedback information for the feature to be clarified, the method further includes: If the supplementary feedback information is the option that cannot be directly observed, then the observation constraints for the feature to be clarified are identified based on the device model of the target device. Based on the observation constraints, obtain second feature information that is correlated with the feature term to be clarified and can be perceived under the observation constraints; Based on the second feature information, the guidance instruction is regenerated and output to obtain feedback results for the second feature information, and the feedback results are used as new supplementary feedback information.

7. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 1, characterized in that, The step of determining the target standard state feature from the plurality of candidate standard state features based on the supplementary feedback information includes: Based on the supplementary feedback information, the candidate standard state feature with the highest correlation to the supplementary feedback information is determined using a preset state deduction logic, and then identified as the target standard state feature.

8. The method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 1, characterized in that, After replacing or associating the feature item to be clarified in the initial description information with the target standard state feature, the method further includes: Check whether there are any unclarified features to be clarified in the initial description information after replacement or association; If they exist, the step of retrieving multiple candidate standard state features corresponding to the feature item to be clarified according to the preset mapping association rules is repeated to perform an iterative interactive information clarification process until all the feature items to be clarified in the initial description information are converted into the target standard state features.

9. A method for generating an energy-saving operation and maintenance scheme for mining crushing equipment according to claim 8, characterized in that, The step of generating a repair plan for the target device based on the replaced or associated initial description information includes: Based on the replaced or associated initial description information, standardized fault information for the target device is generated. Based on the standardized fault information, a search and matching process is performed in a preset maintenance knowledge base to obtain at least one candidate fault mode and at least one candidate maintenance action associated with each of the candidate fault modes. For the retrieved candidate maintenance action, combine and logically verify the constraints stored in the maintenance knowledge base to generate at least one candidate maintenance solution. Output the at least one candidate repair solution, and output the basis information for deriving the candidate repair solution from the standardized fault information.

10. A system for generating maintenance plans for mining crushing equipment, used to execute the steps of the method according to any one of claims 1 to 9, characterized in that, include: The information acquisition module is used to acquire initial description information for the target device, the initial description information including at least one feature item to be clarified; The feature retrieval module is used to retrieve multiple candidate standard state features corresponding to the feature item to be clarified according to preset mapping association rules, and extract dimensional information used to distinguish the multiple candidate standard state features. The guidance feedback module is used to generate and output guidance instructions based on the dimensional information in order to obtain supplementary feedback information for the feature item to be clarified. The feature determination module is used to determine the target standard state feature from the plurality of candidate standard state features based on the supplementary feedback information; The solution generation module is used to replace or associate the feature items to be clarified in the initial description information with the target standard state features, and generate a maintenance solution for the target equipment based on the replaced or associated initial description information.

Images