An AI self-learning operation-based dry oil lubrication control system

Abstract

The present application relates to the technical field of industrial equipment lubrication control, in particular to a dry oil lubrication control system based on AI self-learning operation, which comprises an energy collection module for generating a lubrication energy residual feature set by comparing power and temperature, a hidden variable construction module for merging deviated sections to generate a state section identification result, a state recognition module for analyzing oil pressure, vibration and noise to generate an abnormal density identification result, a priority adjustment module for shortening the specific point oil supply interval according to the abnormal density, and a valve control reconstruction module for reorganizing the opening timing to generate a dry oil lubrication control result. The present application builds an energy residual basis through power and friction consumption verification, forms a deviation criterion and identifies imbalance signs in combination with thermal response lag law, forms a state section according to continuous deviation to enhance the hierarchical expression of evolution, determines the oil supply priority order in combination with the abnormal density level of oil pressure, vibration and noise, matches the oil supply interval and valve control timing to reconstruct the execution link, and improves the flow output consistency and lubrication control precision.

Description

Technical Field

[0001] This invention relates to the field of industrial equipment lubrication control technology, and in particular to a dry oil lubrication control system based on AI self-learning operation. Background Technology

[0002] The field of industrial equipment lubrication control technology encompasses the related technologies for lubrication supply, condition monitoring, and control and regulation of friction pairs during the operation of mechanical equipment. The core content of this technology includes lubrication medium delivery methods, lubrication point layout principles, lubrication cycle setting methods, acquisition of operating status parameters, and the formulation of control logic based on parameter changes. The overall technical system revolves around equipment wear control and operational stability. It acquires information such as temperature, vibration, noise, pressure, and operating time through sensors, and combines this with control units to regulate lubrication pumps, solenoid valves, and oil delivery pipelines. This field also involves on-site anti-interference design, ensuring stable operation of equipment under electromagnetic interference and mechanical shock environments, and the collaborative control and long-term operational adaptation of multi-lubrication point systems.

[0003] Among them, the AI-based self-learning dry lubrication control system refers to a control system that analyzes and adjusts the dry lubrication process by introducing a self-learning mechanism. The main technical issues it addresses include multi-source operating data acquisition, lubrication status identification, dynamic adjustment of lubrication strategies, and linkage control of actuators. Specifically, it involves acquiring bearing and friction pair operating information through vibration sensors, temperature sensors, and friction noise acquisition devices; performing hierarchical preprocessing on the acquired data and establishing a baseline data model; assigning dynamic weights to various parameters according to different working conditions for calculation and analysis to form a lubrication demand determination result; transmitting the determination result to the programmable controller through control logic, which controls the opening and closing of the solenoid valve and drives the lubrication pump to supply dry lubricating oil; simultaneously establishing independent data models for each lubrication point and performing parallel iterative updates; and continuously correcting lubrication parameters based on grease usage time, changes in ambient temperature, and equipment vibration status to achieve continuous adjustment and closed-loop control of the dry lubrication process.

[0004] Existing technologies rely on baseline models and parameter weight calculations to determine lubrication requirements. In scenarios where equipment operating conditions fluctuate over a long period or multiple influencing factors intertwine, different parameter changes may correspond to the same judgment result. The judgment process focuses more on the comprehensive evaluation of the parameter set, making it difficult to reflect the correlation and evolution characteristics between operating energy and friction response. This results in a lack of effective identification criteria for some hidden anomalies. At the same time, the data models of each lubrication point are updated independently and participate in control decisions separately. They lack effective characterization of the continuous characteristics and aggregation trends of abnormal states over time. When anomalies gradually spread or risks continue to accumulate, deviations may occur between the direction of oil supply adjustment and the actual risk area, thereby affecting the targeting and coordination of lubrication control. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a dry oil lubrication control system based on AI self-learning operation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a dry oil lubrication control system based on AI self-learning operation, the system comprising: The energy harvesting module reads the current and voltage of the main drive motor in sequence, aligns and converts the input power sequence in time, compares adjacent power states and writes power change identifiers, coaxially identifies temperature changes, combines material thermal hysteresis to determine friction consumption changes and verify the direction, and generates a lubrication energy residual feature set. The latent variable construction module filters energy deviation time points based on the lubrication energy residual feature set, calls temperature change identifiers on the same time axis, compares the deviation direction with the temperature state, marks the same direction, opposite direction or no response, merges continuous and identical relationships and divides into segments, and generates lubrication state segment identifier results. The status recognition module, based on the lubrication status segment identification results, calls the corresponding segment oil pressure and impact acceleration data, compares the operating range point by point, extracts and merges abnormal time sequence nodes, writes oil pressure, impact or noise abnormality identification, divides the density level according to the interval of abnormal time sequence nodes, and generates lubrication abnormality density identification results. The priority adjustment module locates the location of density change and abnormal time period based on the lubrication abnormal density identification result, associates the abnormal source with the lubrication point, compares the associated and unassociated oil supply intervals to determine the lubrication point to be adjusted, shortens its oil supply interval when it is highly dense or transitions to high density, and generates a lubrication oil supply interval configuration sequence. The valve control reconfiguration module calls the adjusted oil supply interval in the lubrication oil supply interval configuration sequence, forms the opening sequence according to the correspondence between the lubrication point and the solenoid valve and the distribution valve, corrects the position of inconsistent oil supply interval sorting, extends the position of the unstable interval, and generates dry oil lubrication control results by comparing the actual and target flow changes.

[0007] The working principle and advantages of this invention are as follows: In this invention, a basis for characterizing lubrication energy residuals is constructed by verifying the relationship between the direction of input power change and the direction of friction consumption change. An energy deviation criterion is formed by combining the material thermal response hysteresis law, enabling anomaly identification to be based on the correlation characteristics of the energy transfer link and the thermal response link. This allows for the discovery of potential imbalance signs during the evolution of friction states. State segments are formed through continuous deviation relationships, and the continuous characteristics of these segments are extracted, enhancing the hierarchical expression of the lubrication state evolution process. Furthermore, the density of abnormal nodes formed by oil pressure, vibration, and friction noise determines the priority order of oil supply adjustment, allowing lubrication resources to be dynamically configured according to the combined risks of pressure hindrance, mechanical shock, and triboacoustic response. Simultaneously, the execution link is reconstructed based on the matching relationship between the oil supply interval and the valve opening sequence, improving the consistency of flow output and the accuracy of lubrication control, and promoting a continuous and coordinated state in the dry oil delivery process. Attached Figure Description

[0008] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart illustrating the energy harvesting module of the present invention. Figure 3 This is a flowchart illustrating the process of obtaining the latent variable construction module of the present invention. Figure 4 This is a flowchart illustrating the acquisition process of the state recognition module of the present invention. Figure 5 This is a flowchart illustrating the acquisition process of the priority adjustment module of the present invention. Figure 6 This is a flowchart illustrating the acquisition process of the valve-controlled reconfiguration module of the present invention. Detailed Implementation

[0009] The following detailed explanation illustrates the specific implementation methods: The basic implementation examples are as follows: Figure 1 As shown: A dry lubrication control system based on AI self-learning operation, comprising: The energy acquisition module reads the input current and voltage values ​​of the main drive motor according to the operating sequence of the dry oil lubrication device. It aligns and converts the values ​​into an input power sequence based on the acquisition time. It compares adjacent power states point by point along the sequence and writes the power increase, decrease, or stabilization into the input power change identifier. It collects the temperature data of the bearing friction area and connects it to the same time axis. It identifies the temperature increase, decrease, or stabilization state point by point and forms a temperature change identifier. It imports the identifier into the preset material thermal characteristic correspondence. It determines the friction consumption change state according to the temperature change direction and the material thermal response hysteresis relationship. It performs direction verification on the input power change identifier and the friction consumption change state along the same time axis. It retains the state with consistent direction as the normal energy correspondence state and marks the state with inconsistent direction or no response as the energy deviation state, generating a lubrication energy residual feature set. The lubrication energy residual feature set includes energy deviation level, energy deviation duration interval, and energy deviation distribution category.

[0010] Please see Figure 2 Specifically, the energy harvesting module includes: The electrical parameter processing submodule reads the input current and voltage values ​​of the main drive motor according to the operating sequence of the dry lubrication device. It aligns the corresponding records with the acquisition time as a reference, converts the current and voltage values ​​according to the corresponding time points, and organizes them into an input power sequence. It then sequentially associates each acquisition time point and arranges them in the order of acquisition. It compares adjacent power states point by point along the sequence, writing the power increase, decrease, or stabilization to the corresponding identifier, generating a sequence of input power changes. Specifically, as follows: The input current and voltage values ​​of the main drive motor are read according to the operating sequence of the dry lubrication device. The sampling hardware uses voltage transformers and current transformers installed in the distribution cabinet. The signals are converted into digital electrical parameter data through the analog input channel of the control system. The corresponding records are aligned with the time stamp of the common system clock to extract the analog voltage signal of the main drive motor in real time. and current analog signal Set the sampling frequency to 50Hz. The voltage value read at any given time is 380.5V, and the current value is 42.6A. Using multiplication and set operations, the voltage and current values ​​are substituted into the three-phase AC power calculation formula: ,in This indicates the real-time input active power of the main drive motor. The value is the three-phase electrical constant, and is approximated as 1.732 in actual calculations. Indicates the line voltage value. Indicates the line current value. This represents the power factor correction factor, set to 0.85, and the calculation yields... Input power at time It is 23.9kW, and then in The voltage value is 381.2V and the current value is 43.3A. Calculations are then performed. Input power at all times The calculated power values ​​at each time point are 24.3kW. Store the data in a dynamic array and perform subtraction operations on the power values ​​at adjacent time points in sequence. ,in The change in power Power at the current time point, A benchmark value for judging power fluctuations is set based on the power at the previous time point. The value is 0.05kW, and this benchmark value is calibrated with reference to 0.2% of the motor's no-load loss. when Since 0.4 is greater than 0.05, the corresponding time point is recorded as the value "1" as a symbolic representation of the power enhancement state. If continuously collected... The power at time t is 24.32kW. Calculate the current difference. If the power is less than the reference value by 0.05kW, the position will be marked as "0" as a symbolic representation of the power stability state. The power output dropped to 23.8kW at that moment; Calculate the current difference If its absolute value is greater than 0.05kW, the corresponding identifier will be recorded as the value "-1" as a symbolic representation of the power reduction state. The above state identifiers will be arranged on the same time axis in the order of acquisition to generate a sequence of input power changes.

[0011] The heat dissipation determination submodule calls the time axis corresponding to the input power change sequence, collects temperature data of the bearing friction area and connects it to the same collection time point, identifies the temperature rise, fall or stable state point by point, forms a temperature change identifier, obtains the corresponding relationship of bearing material thermal properties and thermal response hysteresis relationship, and determines the state of increased, decreased or unchanged friction consumption according to the direction of temperature change and the order of hysteresis correspondence, thus obtaining the friction consumption change sequence; as detailed below: The time axis corresponding to the input power change sequence is called, and the temperature data of the bearing friction area is collected and connected to the same collection time point. The data acquisition equipment adopts a PT100 platinum resistance temperature sensor embedded in the core bearing area of ​​the spindle bearing housing. The 4 to 20 mA standard current signal is output through an SBWR series temperature transmitter with an accuracy class of 0.2 and connected to the SM331 analog input card of the control system. The physical sampling interval of the temperature data is set to 2.5 seconds. Time point alignment is completed by performing moving average downsampling processing on the preceding electrical parameter power data. Read surface temperature data continuously It was 58.42℃, and in Read surface temperature data continuously The temperature is 58.45℃. A differential calculation is performed on adjacent temperature points to obtain the rate of temperature change, and a temperature fluctuation judgment threshold is set. The threshold is 0.02℃. This threshold is calibrated based on the 0.1-grade full-scale accuracy of the temperature sensor and the average high-frequency noise level in the field. The relationship between the current temperature and the threshold of 0.02℃ is as follows: Since the current difference exceeds the threshold range, the current temperature state is determined to be rising and recorded as the character "H". If the absolute value of the difference is within 0.02℃, the temperature change is determined to be insignificant noise and recorded as the character "S" as a symbol of a stable state. If the difference is negative and the absolute value exceeds 0.02℃, the current temperature state is determined to be falling and recorded as the character "L" as a symbol of a falling state. When faced with inconsistent situations where high-frequency temperature disturbances cause frequent alternations in direction, dead zone suppression processing is performed, forcibly classifying points where fluctuations are confined within the threshold into a stable state. The thermal conductivity parameter of the GCr15 bearing steel used in the bearing is retrieved as 40.1 W / (m·K), and the specific heat capacity parameter is retrieved as 0.46 kJ / (kg·℃), combined with the standard density value of the material. The thermal response hysteresis time step was calculated using a one-dimensional transient thermal conduction and diffusion mathematical equation with a physical heat conduction distance of 15 mm. It consists of 8 acquisition cycles, that is, in The observed temperature change state at any time essentially corresponds to The frictional heat changes over time, among which... Indicates the number of lag periods. The current observation time is indicated by shifting the temperature change indicator sequence as a whole along the time axis in the past direction by 8 scales. According to the direction of temperature change and the corresponding order of lag, the shifted temperature rise indicator "H" is matched with the state of increased friction consumption, the temperature decrease indicator "L" is matched with the state of decreased friction consumption, and the temperature stability indicator "S" is matched with the state of unchanged friction consumption, thus forming a friction consumption change sequence with the same step size as the power change sequence.

[0012] The direction verification submodule, based on the input power change sequence and the friction consumption change sequence, verifies the direction of input power change and the direction of friction consumption change point by point along the same time axis. Points with consistent directions are registered as the normal energy state, while points with inconsistent directions or no corresponding response are registered as energy deviation states. The module collects the state identifiers, deviation directions, and state duration sequences for each time point according to the acquisition time, generating a lubrication energy residual feature set; specifically as follows: Based on the input power change sequence and the friction loss change sequence, the direction of input power change and the direction of friction loss change are checked point by point along the same time axis. The central processing unit of the control host performs logical verification through the bit matrix comparison operator and reads... The power change at a given time point is marked as "1" and the corresponding friction loss change is marked as "increase". At this point, a logical comparison operation is performed. Since the increase in power and the increase in friction loss are physically and logically within a positive correlation range for energy conversion, this... The time point is recorded as the normal energy state and assigned the value "0". When it is read... When the power change at a given time point is marked as "-1" while the friction consumption change is marked as "increase", a logical anisotropic comparison operation is performed. It is determined that the decrease in power and the increase in heat consumption are contradictory. This state represents a mechanical dry friction or sudden lubrication failure due to decreased input electrical energy and abnormally increased local frictional heating. The time point is recorded as an energy deviation state and assigned the value "1", while the deviation direction at that point is recorded as "negative deviation". If the power at a given time point is marked as "1" but the friction consumption is marked as "unchanged," and no response is generated for more than three sampling periods, this state represents increased electrical energy but mechanical jamming or energy diversion in the non-lubricated area of ​​the bearing without heat feedback. This is also classified as an energy deviation state and assigned the value "2," with the deviation direction recorded as "no response deviation." The distribution of each deviation state point on the time axis is statistically analyzed, and the number of consecutive points with the same deviation attribute is calculated to determine the state duration order. arrive When "negative deviation" occurs continuously, the continuous step size is recorded as 6. The status identifiers, deviation direction values, and continuous logical count values ​​of all time points are encapsulated in a structured manner to generate a lubrication energy residual feature set. It should be noted that, during the adaptation process for specific operating conditions, the main drive motor power parameter is used to assist in characterizing the correspondence between changes in equipment energy input and frictional thermal response. If the system identifies operating conditions such as the influence of external heat sources, low-speed conveying, high-temperature materials near lubrication points, or grease evaporation and carbonization due to heat, making it difficult for changes in the main drive motor power to directly characterize the lubrication state, the weight of the main drive motor power parameter in lubrication state identification will be adjusted. Temperature changes, impact acceleration changes, and frictional noise changes will be used as the main judgment criteria, and the main drive motor power parameter will be used as an auxiliary verification quantity for energy response deviation in the calculation.

[0013] The latent variable construction module filters energy deviation time points based on the energy deviation state in the lubrication energy residual feature set and calls the temperature change identifiers under the same time axis. It compares the deviation direction with the temperature rise, fall, or stable state point by point. When the deviation direction is consistent with the temperature change direction, it is marked as a same-direction relationship. When the two directions are opposite, it is marked as an opposite-direction relationship. When the temperature change is stable or no corresponding response is formed, it is marked as a no-response relationship. It reads the comparison relationship of each energy deviation time point in chronological order and compares the current energy deviation time point relationship with the previous energy deviation time point relationship. Energy deviation time points that maintain the same relationship continuously are grouped into state segments. The duration of each state segment is recorded. The duration of adjacent state segments is compared and mapped to the preset interval range to complete the segment division and generate the lubrication state segment identifier result. The lubrication status segment identification results include same-direction segments, opposite-direction segments, and no-response segments.

[0014] Please see Figure 3 Specifically, the latent variable building blocks include: The temperature deviation matching submodule, based on the energy deviation state in the lubrication energy residual feature set, filters energy deviation time points, calls temperature change identifiers under the same time axis, and reads the deviation direction and temperature rise, fall, or stable state according to the time point. Those with the same direction are recorded as a same-direction relationship, those with opposite directions are recorded as a reverse-direction relationship, and those with stable temperature or no corresponding response are recorded as a no-response relationship. These are then written into the relationship code according to the time sequence to generate a temperature deviation relationship code; specifically as follows: Based on the energy deviation status in the lubrication energy residual feature set, energy deviation time points are screened, and database retrieval commands running on an industrial PC are used to retrieve data from the historical database table related to the previous energy deviation time period. to The associated, translated and aligned sequence of friction loss changes was used to extract the energy deviation direction identifiers for each sampling point within that time period. and the corresponding alignment temperature status indicator ,in Indicates the first The deviation direction encoding of each sampling point Indicates the first The temperature change status of each sampling point after translation and alignment, Taking a specific point in time as an example, the energy deviation direction is recorded as the value "-1", indicating a negative deviation, while the corresponding temperature change is marked as "H", indicating an increasing state. A direction consistency logic comparison is performed to determine the physical relationship between the numerical sign and the character meaning. If the deviation direction is positive and the temperature is marked "H", or the deviation direction is negative and the temperature is marked "L", then it is determined to be a same-direction relationship, and the value "1" is written to the relationship encoding array. If the deviation direction is negative and the temperature is marked "H", or the deviation direction is positive and the temperature is marked "L", then it is determined to be a reverse relationship, and the value "2" is written to the relationship encoding array. When faced with noise waveforms where frequent fluctuations in temperature change direction lead to inconsistent continuity, the highest frequency attribute within the current sliding window is extracted as the filtering benchmark. Since the deviation direction is negative and the temperature is rising, the two present a contradiction in energy conversion logic, are determined to be inverse relationships, and are recorded with the code "2". If extracted... The temperature at a given time point is labeled "S," clearly indicating that the temperature is within a stable range, meaning the temperature fluctuation is within the preset 0.02℃ threshold. This indicates that the preceding thermal response lag relationship failed to form an effective feedback within the current observation window, thus it is judged as a no-response relationship and recorded as code "3." to The above logical judgment is performed at all points within the time period, and the resulting value "2,2,2,2,2,2" is stored in the partial temperature relationship matrix according to the time sequence to generate the partial temperature relationship code.

[0015] The relation merging submodule, based on the temperature deviation relation encoding, reads the corresponding relations of each energy deviation time point in chronological order, compares the current time point relation with the previous time point relation, performs sequential merging for time points with consecutive and consistent relation codes, ends the current merging at the relation code conversion position, and starts the subsequent merging, records the start time, end time, and corresponding relation category of each segment, and extracts the duration of each segment according to the start and end times to obtain the state segment duration; as detailed below: Based on the temperature deviation coding, the corresponding relationships of each energy deviation time point are read sequentially. The PLC controller uses pointer offset addressing to start from the physical address of the coding matrix. Begin extracting the encoded value bit by bit, and set the index of the current processed bit to 0. Read The encoded value of the bit And simultaneously read its adjacent preceding one. coded value ,in Represents the base address of the matrix. Indicates the current sampling point number. Indicates the current relation encoding. Indicates the previous relation encoding, performs an equality logic check, if equal If the current time point is determined to be in the state continuation period, the merge counter is maintained. Perform accumulation operation ,in A counter representing consecutive consistent nodes, in arrive Taking the consecutive encoding "2" as an example, in A comparison revealed that it was similar to With consistent encoding, the merge count increases from 1 to 2. The code "3" was read at the time point, and... If the comparison result of the code "2" is not equal, the merging operation of the current code "2" is immediately stopped. Combined with the aforementioned 2.5-second temperature sampling interval, the... The corresponding actual time of 25 seconds is recorded as the start timestamp of the current segment. ,Will The corresponding actual time of 37.5 seconds is recorded as the end timestamp. Then, the corresponding relationship category "reverse relationship" is stored in the segment attribute field, and the time difference calculation function is called to calculate... and The difference between the two, i.e., 37.5 seconds minus 25 seconds equals 12.5 seconds, gives the absolute duration of the physical time of this state segment, which is then used to calculate the time. The counting restarts from the starting point marked as the next merge segment. The above step-by-step scanning and segmentation are performed on the entire sequence of the partial temperature relationship encoding. Each encoding mutation position is used as a logical breakpoint. The duration values ​​of each independent segment are extracted in sequence to obtain the duration of the state segment.

[0016] The time-series demarcation submodule calls the state segment duration, reads the duration and relationship category of adjacent state segments, obtains the upper and lower boundaries corresponding to the preset interval range, compares the duration of each segment with the interval boundary according to the time sequence, writes the same segment identifier for state segments falling into the same interval, and writes the corresponding category identifier for state segments crossing the interval boundary. It then sequentially associates the segment start and end times, duration, and relationship category to generate the lubrication state segment identifier result; specifically as follows: The system calls the state segment duration, reads the duration and relationship category of adjacent state segments, pre-sets the segment division boundaries based on the single rated oil discharge cycle of the dry oil lubrication pump and the thermal balance constant of the mechanical system, and sets the lower limit of the short-term disturbance boundary. The response time is 3 seconds, which is the middle limit of the normal response boundary. The maximum long-term offset is 20 seconds. It lasts for 60 seconds, of which , , These represent time span boundary values ​​for different levels. The duration of the first segment, 12.5 seconds, is extracted and compared with the boundary values. Since 12.5 seconds falls within the range... Within this range, it is identified as a "transition response segment" and assigned the identifier "SEG_01". If the duration of the second segment is 45 seconds, its value falls within the range... Within 60 seconds, it is identified as a "stable deviation segment" and assigned the identifier "SEG_02". If the duration exceeds 60 seconds, it is assigned the identifier "SEG_03" as a "risk evolution segment". For extremely short segments with a duration of less than 3 seconds, they are identified as system data transmission jitter or instantaneous spikes, and boundary smoothing is performed. Independent segment creation is abandoned, and the relationship category of the narrow pulse is forcibly merged into the adjacent longer segment. During the determination process, the relationship categories of each segment determined in the off-temperature matching submodule are associated, such as "reverse". The segment with a "relationship" and a duration of 12.5 seconds is denoted as "R_SEG_01", and the segment with a "no response relationship" and a duration of 45 seconds is denoted as "N_SEG_02". A five-dimensional vector is created, which includes the start timestamp, end timestamp, duration value, relationship category code, and final segment level identifier. Level labeling and attribute association are performed on all segments in the entire lubrication operation process in sequence. The labeled sequence is arranged in chronological order and encapsulated into a structured data object to generate the lubrication status segment identification result.

[0017] The status recognition module, based on the status segment identifier in the lubrication status segment identifier result, calls the dry oil distributor outlet oil pressure data, bearing housing vibration impact acceleration data, and bearing friction area friction noise data within the corresponding time range. It reads the oil pressure operation status, impact acceleration operation status, and friction noise operation status according to the time axis, and compares them point by point with their respective operation ranges to extract abnormal oil pressure time points, abnormal impact time points, and abnormal noise time points. These are then merged into an abnormal time sequence node set according to time sequence. An abnormal source identifier is written for each abnormal time sequence node. The time interval between adjacent abnormal time sequence nodes is extracted and compared with a preset interval threshold to classify the corresponding density level. Within the time range corresponding to each status segment identifier, the density level is read sequentially according to time sequence. The location where the density level changes is marked as the density change location, and associated with the abnormal source combination category to generate the lubrication abnormal density recognition result. The results of lubrication anomaly density identification include low-density anomaly regions, medium-density anomaly regions, high-density anomaly regions, and anomaly source combination categories. The anomaly source combination categories include oil pressure anomaly category, impact anomaly category, friction noise anomaly category, oil pressure and impact combined anomaly category, oil pressure and noise combined anomaly category, impact and noise combined anomaly category, and oil pressure, impact and noise combined anomaly category.

[0018] Please see Figure 4 Specifically, the status recognition module includes: The range verification submodule, based on the state segment identifier in the lubrication state segment identifier result, calls the dry oil distributor outlet oil pressure data, bearing housing vibration impact acceleration data, and bearing friction area friction noise data within the corresponding time range. It reads the oil pressure operating status, impact acceleration operating status, and friction noise operating status along the time axis, obtains the upper and lower boundaries of their respective operating ranges, and checks point by point whether the collected status exceeds the corresponding boundary. The time points exceeding the boundary are registered as oil pressure abnormal time points, impact abnormal time points, and noise abnormal time points, respectively, generating an abnormal time point sequence. The details are as follows: Based on the state segment identifier in the lubrication state segment identifier result, retrieve the segment stored in the real-time database that is under the identifier "R_SEG_01" in the "reverse relationship transition response segment". to The time range is clearly defined as 25 to 37.5 seconds on the actual physical time axis. Pressure data is acquired online by a resistance pressure transmitter installed at the end of the dry oil distributor's main pipeline. Vibration data is captured at high frequency by a piezoelectric accelerometer vertically fixed to the non-drive end housing of the bearing housing via a magnetic base. Friction noise data is acquired by an industrial microphone or acoustic sensor located on the outside of the bearing housing and facing the friction pair area. The transmitter, accelerometer, and acoustic sensor signals are respectively sent to the PLC via a 4 to 20 mA current transmitter module and a high-speed data acquisition card. The analog signal value of the dry oil outlet pressure during this time period is extracted from the end of the distributor. Simultaneously, the calculated value of the original impact acceleration sampled signal after time-domain root mean square operation is extracted. The noise operation status is extracted by bandpass filtering, background noise subtraction, and root mean square sound pressure level calculation of the friction noise signal. Set the lower boundary of the normal operating range of hydraulic pressure. The pressure is 2.0 MPa, and the upper boundary is... The pressure is 18.0 MPa. This boundary is used for safety calibration with reference to 90% of the rated relief valve opening pressure of the dry oil pump, and the lower boundary of the normal operating range of impact acceleration is set. 0g, upper boundary The upper boundary is 3.5g. This upper boundary is determined based on the rated impact limit corresponding to the vibration intensity standard ISO10816 for bearings at the current speed, and sets the lower boundary of the normal operating range for friction noise. The upper boundary is the background noise baseline value. The noise threshold is the background noise baseline value increased by 8dB, where This is the actual oil pressure. This is the actual impact acceleration. This represents the actual friction noise sound pressure level. and This is the oil pressure threshold. and Vibration threshold and To determine the friction noise threshold, read the data corresponding to 30 seconds of actual physical time. The oil pressure at that moment is 19.2 MPa, the effective acceleration value at that moment is 4.2 g, and the friction noise sound pressure level at that moment is 76 dB. If the current background noise baseline value is 65 dB and the noise threshold is 73 dB, then a numerical comparison is performed. It is determined that 19.2 is greater than 18.0, 4.2 is greater than 3.5, and 76 is greater than 73. Therefore, it is determined that the oil pressure, impact, and friction noise states at that moment are all in an out-of-bounds state. Register the time point as an abnormal oil pressure and associate it with the "R_SEG_01" segment identifier, and at the same time... Registered as an abnormal time point of impact, and The time point recorded as an abnormal noise event is 32.5 seconds past the corresponding actual physical time. If the oil pressure is constantly detected to be 17.5 MPa, the acceleration to be 3.8 g, and the friction noise sound pressure level to be 75 dB, then... The time points were registered as impact anomalies and noise anomalies. The above boundary check operation was performed on all sampling points in the section. The identified out-of-bounds time points were encapsulated in chronological order to generate an anomaly time point sequence.

[0019] The node orchestration submodule calls the abnormal time point sequence, merges oil pressure abnormal time points, impact abnormal time points, and noise abnormal time points according to the acquisition time, retains the original time and state segment of each node, writes oil pressure abnormality, impact abnormality, or noise abnormality identifier according to the node source, arranges all nodes in chronological order, calculates the time interval between adjacent abnormal time sequence nodes, obtains the preset abnormal node interval threshold, and maps each time interval to the threshold division interval to obtain the abnormal node density level; the details are as follows: Call the abnormal time point sequence to read the oil pressure abnormal time points identified in the "R_SEG_01" segment. Impact Anomaly Time Point and abnormal noise times , , In the PLC's data block storage area, a timestamp deduplication and merging operation is performed. Multiple types of anomalies occurring at the same time are merged into one anomaly timing node through a logical OR operator. The original actual physical time of the node (30 seconds, 32.5 seconds, 35 seconds) and its corresponding "R_SEG_01" segment index are retained. In the node attribute data structure field, "PGN" is written to represent the triple anomaly identifier of oil pressure, impact, and noise, or "GN" is written to represent the dual anomaly identifier of impact and noise, or "G" is written to represent the single impact anomaly identifier. Here, "P" represents oil pressure anomaly, "G" represents impact anomaly, and "N" represents friction noise anomaly. All merged nodes are arranged in ascending order of time. Extract the time difference between adjacent nodes sequentially. ,in This indicates the time interval between adjacent abnormal nodes. This represents the actual number of seconds that the current node's time is extended. The actual number of seconds extended from the previous node time, in and For example, the actual physical time interval is calculated to be 2.5 seconds. A preset abnormal node interval threshold system is obtained, and the density level division interval is set as follows: when the time interval is within... When the time interval is determined to be "high density" and assigned the value "level 3", it is considered "high density". When the time interval is greater than 12.0 seconds, it is judged as "medium density" and assigned the value "level 2". When the time interval is greater than 12.0 seconds, it is judged as "low density" and assigned the value "level 1". The above time thresholds are calibrated by referring to 25% (i.e., 2.4 seconds, rounded up to 3.0 seconds) and 125% (i.e., 12.0 seconds) of the complete mechanical reciprocating motion cycle of the dry oil distributor valve core, which is 9.6 seconds. Since 2.5 seconds falls within the range of 2.5 seconds, the time interval is determined as "medium density" and assigned the value "level 2". Within the interval, The density level of the time point is registered as "Level 3", if the subsequent corresponding actual physical time is 45 seconds. An anomaly occurred at a certain point in time and was related to the previous node. The 35-second interval is 10.0 seconds, because 10.0 seconds falls on... Within the interval, then The density level of each time point is registered as "Level 2". The combined results of the oil pressure abnormal time point, impact abnormal time point and noise abnormal time point in the same abnormal time sequence node are written into the abnormal source combination category. The time sequence position of each node, the abnormal source combination category and the corresponding density level value are mapped and associated to obtain the density level of the abnormal node.

[0020] The density positioning submodule, based on the density level of abnormal nodes, reads the density level of abnormal time-series nodes sequentially within the time range corresponding to each state segment identifier. It associates the abnormal source identifier, node time, and state segment, compares the current node density level with the previous node density level, registers the level-maintaining positions as continuing nodes, and marks the level-transition positions as density change positions. It also determines the abnormal source combination category based on oil pressure abnormality identifiers, impact abnormality identifiers, and noise abnormality identifiers. Finally, it sequentially aggregates the state segment identifier, density level, abnormal source combination category, and density change position to generate the lubrication abnormality density identification result; specifically as follows: Based on the security level of the abnormal node, in the "R_SEG_01" state segment... to On the timeline, the dense level numerical sequence of each abnormal time-series node is read bit by bit in chronological order. The current processing node is extracted using the array comparison register inside the PLC. Level value The source identifier "PGN" and its associated identifier, where "P" indicates abnormal oil pressure, "G" indicates abnormal impact, and "N" indicates abnormal friction noise, synchronously retrieve the previous node. Level value ,in and This indicates adjacent abnormal time-series nodes. and This represents the corresponding density level value. Perform a subtraction operation on the level value and take the absolute value. If the absolute value is 0, the current node is determined to be a continuation node, maintaining the current "high-density" state description. Upon entering the "N_SEG_02" section and detecting a change in density level from "Level 3" to "Level 1", a subtraction operation is performed to obtain an absolute value of 2. If this difference is not zero, then... This timestamp marks the location of density change, recording the change attributes from "high density" to "low density". At the same time, it retrieves the bearing housing vibration source identifier, oil pressure status distribution, and friction noise status distribution corresponding to this location. It combines the node index of the level jump, the level difference before and after the change, the corresponding status segment identifier, the abnormal source combination category, and the hardware source identifier of the abnormality into a multi-dimensional array, and counts the frequency and specific time distribution of density changes in each status segment. It outputs the physical time of each level switching point, the corresponding dense interval attribution information, and the abnormal source combination category in a structured manner to generate the lubrication abnormal density identification result. It should be noted that the oil pressure parameters in this scheme are used to reflect the blockage of the oil supply path, the status of the distributor outlet, or the feedback of lubrication execution, and participate in the identification of abnormal source combinations and abnormal density. If the correlation between oil pressure changes and the actual friction state of the lubrication point decreases under the current operating conditions, or if the oil pressure parameters are insufficient to characterize the lubrication state alone, the system adjusts the weight of the oil pressure parameters in abnormal density identification, and uses abnormal temperature, abnormal impact acceleration, and abnormal friction noise as the main judgment criteria.

[0021] The priority adjustment module determines the abnormal time period corresponding to the abnormal time sequence node based on the density change position in the abnormal density identification result, and associates the corresponding lubrication point with the abnormal source identifier. It calls the current oil supply interval data of each lubrication point, compares the oil supply interval of the associated lubrication point in the abnormal time period with the oil supply interval of the non-associated lubrication point in the same abnormal time period, and determines the lubrication point to be adjusted. When the density level of the abnormal time period corresponding to the lubrication point to be adjusted is high density, or the density level corresponding to the density change position changes from low density or medium density to high density, the lubrication point to be adjusted is shortened. The oil supply interval of the non-lubrication point to be adjusted remains unchanged. The adjusted oil supply intervals of each lubrication point are sorted according to time sequence to generate a lubrication oil supply interval configuration sequence. The lubrication supply interval configuration sequence includes the basic supply interval, the shortened supply interval, and the maintenance supply interval.

[0022] Please see Figure 5 Specifically, the priority adjustment module includes: The time period association submodule determines the abnormal time period corresponding to the abnormal time sequence node based on the density change position in the lubrication abnormal density identification result, and associates the corresponding lubrication point with the abnormal source identifier. It reads the time and source identifier of the abnormal node before and after the density change position, and groups the nodes with the same source and continuous time into the same abnormal time period. It filters the time periods that the source identifier can correspond to the lubrication point according to the lubrication point number and density level of each time period, and generates an abnormal time period point mapping. The process of grouping nodes with the same source and consecutive time into the same abnormal time period is as follows: The abnormal source identifier and node time of each abnormal time sequence node are read in the order of collection time. Abnormal time sequence nodes with the same abnormal source identifier and consecutive arrangement on the time axis are merged in sequence. The current merging ends when the abnormal source identifier changes or the time continuity is interrupted, forming the corresponding abnormal time period and recording the start time and end time; as follows: Based on the density change locations in the lubrication anomaly density identification results, the abnormal time periods corresponding to the abnormal time sequence nodes are determined, and the corresponding lubrication points are associated with the anomaly source identifier. Using the Modbus protocol register address mapping table configured within the controller, the hardware physical code is bidirectionally bound to the specific mechanical lubrication parts, and data is extracted from the address space of the data storage device. , , The original records of three consecutive abnormal nodes are retrieved, and the abnormal source identifier field corresponding to each node is read. The source is identified by "PGN" and mapped to the lubrication point of spindle bearing No. 1, where "P" indicates abnormal oil pressure, "G" indicates abnormal impact, and "N" indicates abnormal friction noise. The data is read sequentially by performing a step scan on the timing node set. and Perform a subtraction operation on the corresponding actual timestamp value. Obtain the preset continuity determination threshold. The threshold is 5.0 seconds, which is set with reference to the time constant of the main oil pump completing a single pressurized oil flow propulsion at standard viscosity. Indicates the time difference of abnormal nodes. This indicates the upper limit of the continuous span. A comparison is performed between 2.5 seconds and 5.0 seconds. Since 2.5 seconds is within the range... Within the interval, determine whether two nodes are continuous in the time dimension. and Grouped into the same set, then read The node then repeats the above comparison process to confirm its relationship with... The actual physical time interval is also 2.5 seconds and less than the threshold. When scanning to... When calculating a node, its relationship with... The actual physical time interval is 45 seconds minus 35 seconds, which equals 10.0 seconds. Since 10.0 seconds exceeds the preset continuity threshold of 5.0 seconds, thus breaking the time continuity, a merge termination action is executed. The corresponding 30 seconds are recorded as the start timestamp of the current abnormal time period. ,Will The corresponding 35 seconds are recorded as the end timestamp. ,in and These represent the boundary times of the abnormal period, and are associated with the corresponding density level value "3" and the abnormal source combination category in this data structure. The lubrication point number 1 and the abnormal source combination category are then linked to this... The time period is used to perform bidirectional index binding. At the same time, lubrication points No. 2 and No. 3, which do not have records of abnormal oil pressure, abnormal impact or abnormal noise in the corresponding segment, are removed. All groups formed by logical merging are merged, including start and end times, source point numbers, density levels and abnormal source combination categories, to generate abnormal time period point mapping.

[0023] The point identification submodule calls the point mapping during abnormal time periods to obtain the current oil supply interval of each lubrication point. It sorts the oil supply intervals of associated and unassociated lubrication points according to the abnormal time period, keeping the time range, collection order and lubrication point number consistent. It compares the oil supply interval status of the two types of lubrication points within the same abnormal time period, registers the lubrication points that are associated with the abnormal source and have different interval status as objects to be adjusted, and associates them with the corresponding density level to obtain a list of oil supply objects. The process of comparing the oil supply interval status of two types of lubrication points within the same abnormal time period is as follows: The oil supply intervals of associated and unassociated lubrication points are arranged according to the same time range and collection order. The oil supply interval status is read and compared point by point. When there is an inconsistency in the oil supply interval status, the corresponding lubrication point is marked as an object to be adjusted, and associated with the density level of the corresponding abnormal time period; as detailed below: The abnormal time period point mapping is invoked to obtain the current oil supply interval of each lubrication point, and the basic oil supply cycle setting value of lubrication point No. 1 is retrieved from the D200 section of the PLC's retainable data register. Synchronous retrieval during the same abnormal period The oil supply interval values ​​for the non-associated objects, namely lubrication points 2 and 3. and ,in , , The current oil supply interval in seconds for different numbered lubrication points is represented. A comparison matrix indexed by the collection order is established. The "PGN" abnormal attribute associated with lubrication point 1 is horizontally assembled with its 300-second oil supply interval. Simultaneously, the normal attribute of point 2 is assembled with its 300-second interval. A point-by-point comparison is performed to check the out-of-bounds records of oil pressure, impact acceleration, and friction noise at point 1 during the abnormal period. It is confirmed that it is in a high-density state between 30 and 35 seconds, and the abnormal source combination category is the combination of oil pressure, impact, and noise abnormality. In contrast, the control group point 2 did not produce out-of-bounds values ​​for oil pressure, impact, or friction noise within the same time range. A logical XOR operation is performed on the bit array to determine the lubrication demand status of point 1 and the operation of point 2. Significant inconsistencies exist in the stable state. Even though the original oil supply interval is 300 seconds, the oil supply response priority shifts due to the intervention of abnormal density and abnormal source combination categories. Lubrication point No. 1 is extracted from the global point list and stored in the adjustment buffer area, and its adjustment trigger bit is marked with the value "1". At the same time, it is associated with the level value "3" and abnormal source combination category corresponding to this point in the abnormal density identification result. For points No. 2, No. 3, etc., which have no abnormal association, their adjustment trigger bits are cleared to zero and retained in their original sequence positions. All lubrication point numbers with trigger bits of the value "1" are summarized, and their current oil supply interval value, the state segment identifier, the associated abnormal time span, and the abnormal source combination category are attached to them to obtain the oil supply object list.

[0024] The sequence arrangement submodule, based on the list of lubrication objects, calls the density level and density change position of each lubrication point to be adjusted during the abnormal time period, determines whether the object to be adjusted is in high density or has changed from low density or medium density to high density, performs lubrication interval shortening for lubrication points that meet the conditions, and keeps the lubrication interval of non-lubrication points unchanged. The lubrication interval of each lubrication point is arranged according to the running time to generate a lubrication lubrication interval configuration sequence. The process of shortening the oil supply interval for lubrication points that meet the requirements is as follows: The objects to be adjusted are screened according to the density level and location of the corresponding abnormal time period. Those in a high-density state or undergoing a transition from low-density, medium-density to high-density are processed by decreasing the oil supply interval, while keeping the oil supply interval of unscreened lubrication points unchanged. Simultaneously, the adjusted oil supply intervals are rearranged according to the operating time sequence; the details are as follows: Based on the list of lubrication points, the density level and location of density changes during the abnormal time period are retrieved for each lubrication point to be adjusted, and the density level value of lubrication point No. 1 is read. and the corresponding density change location markers The level jump direction verification is performed within the floating-point arithmetic unit of the controller, by comparison. The level records before and after the node are verified in conjunction with the high-density judgment logic. If the current processing point itself is in a high-density state, or if it needs to perform advance adjustment due to the previous high-density zone, the real-time temperature of the friction area of ​​bearing No. 1 (58.45℃) transmitted by the PT100 sensor and the reference temperature are retrieved. The deviation value is read, and the excess amplitude of the friction noise sound pressure level relative to the background noise reference value during the abnormal time period is read. Combined with the abnormal density level, the fuel injection interval is shortened, and the floating point operation instruction is called to execute the formula. The adjustment coefficient satisfies the following relationship: ,in This indicates the adjusted new fuel injection interval. This indicates the original fuel injection interval before adjustment, with the current value being 300 seconds. This represents the calculated dimensionless shortening coefficient. Indicates the current real-time temperature and takes , The base reference temperature for normal bearing operation is set to [temperature value]. , Indicates the maximum permissible operating temperature of the bearing and sets it to . Substituting the data and performing multi-level algebraic operations yields: The upper limit threshold for adjustment is set to 0.4. If 0.23 is less than 0.4, it is taken as the effective adjustment amount. When the friction noise sound pressure level exceeds the preset noise increment threshold relative to the background noise reference value, it is determined that there is an abnormal friction acoustic response at the lubrication point, and this abnormal noise response is used as an auxiliary trigger condition for shortening the oil supply interval. When noise abnormality, impact abnormality, and oil pressure abnormality occur simultaneously within the same abnormal time period, the oil supply adjustment priority of the corresponding lubrication point is raised to the highest level within the current abnormal time period, and the oil supply interval is decreased to obtain the adjusted new interval. For lubrication points 1, the configuration parameters are overwritten with a value of 231 seconds, and the 231 seconds are written into the corresponding control register bit. For lubrication points 2 and 3, which are not to be adjusted, the subtraction operation is skipped and their original 300-second setting value is directly retained. According to the physical topology connection order of the lubrication points, the 231 seconds of point 1, the 300 seconds of point 2, and the 300 seconds of point 3 are linearly arranged according to the time sequence index to establish a time sequence table corresponding to the control logic of the dry oil pump solenoid valve. The adjusted time interval values ​​are reassembled and encapsulated according to the acquisition time order to generate the lubrication supply interval configuration sequence.

[0025] The valve control reconfiguration module calls the adjusted oil supply intervals of each lubrication point in the lubrication supply interval configuration sequence. According to the preset control correspondence between each lubrication point and the dry oil pump outlet solenoid valve and distribution valve, it arranges the opening order and forms a valve control opening sequence. It compares the order of the oil supply intervals of each lubrication point with the valve control opening sequence, and adjusts the positions where the order is inconsistent. It reads the opening time interval of adjacent solenoid valves and distribution valves after adjustment and compares it with the preset stable intervals collected during operation. It performs extension processing on the positions that do not meet the preset stable intervals. According to the extended valve control opening sequence, it reads the actual flow change during the lubrication oil delivery process and compares it point by point with the target flow change under the corresponding opening sequence to extract the flow deviation position and generate the dry oil lubrication control result. The results of dry lubrication control include flow deviation position, flow matching status, and valve control execution status.

[0026] Please see Figure 6 Specifically, the valve-controlled reconfiguration module includes: The valve sequence arrangement submodule calls the adjusted oil supply intervals of each lubrication point in the lubrication supply interval configuration sequence, obtains the preset control correspondence between each lubrication point and the dry oil pump outlet solenoid valve and distribution valve, associates the corresponding solenoid valves and distribution valves according to the oil supply interval sequence, arranges the opening order of each valve to form a valve-controlled opening sequence, compares the oil supply interval order with the valve-controlled opening sequence, adjusts the positions with inconsistent sequences to the corresponding lubrication point oil supply sequence, and generates the valve opening sequence; as detailed below: The system calls up the adjusted oil supply intervals of each lubrication point in the lubrication supply interval configuration sequence and extracts the oil supply interval of lubrication point 1 from the system priority configuration data table. Oil supply interval of lubrication point No. 2 and the oil supply interval of lubrication point No. 3 ,in , , Representing the new time parameters for each distribution valve branch, access the hardware group mapping address table of the dry oil lubrication device and read the first solenoid valve corresponding to lubrication point 1. Its end distribution valve The physical logic address is used to synchronously read the solenoid valves corresponding to lubrication points 2 and 3. , and its distribution valve and The control index, where and These represent different oil supply drive components in different channels. The bubble sort algorithm is executed in the PLC's structured text program to arrange 231, 300, and 300 in ascending order. The minimum value (231 seconds) for lubrication point 1 is identified as the minimum value within the array, and its corresponding control valve group is then selected. Assign the first priority, then read the same value from points 2 and 3 for 300 seconds, and execute the secondary sequence arrangement according to the step increment of the hardware interface physical number. Assign a second priority to enable it. Assign a third priority to enable, and establish a system clock based on the current system clock. For the time base, there is a time-series linked list, where This indicates the current system base timestamp, setting the trigger point for the opening command of valve group 1 at [time]. Set the preset opening point of valve group 2 to The initial valve-controlled opening sequence is constructed sequentially, and the PLC's internal cyclic scan register is retrieved. The existing valve instruction dequeue order is used to perform a bit-by-bit logical XOR comparison operation. The first valve recorded is The first position of the current calculation sequence is If the location is determined to be out of order, a hardware interrupt is triggered, and an address bus rewrite is performed. The control bit is forcibly moved to the head of the task queue, The oil supply sequence is then transferred to the next position in the corresponding lubrication point No. 1, generating the valve opening sequence.

[0027] The interval extension submodule, based on the valve opening sequence, reads the adjusted opening time intervals of adjacent solenoid valves and distribution valves, collects the corresponding time intervals of continuous and stable opening of each valve during the operation of the dry oil lubrication device, organizes them into preset stable intervals, compares the opening time intervals of each adjacent valve with the corresponding preset stable intervals, and sequentially extends the valves at positions with opening times less than the lower limit of the stable interval according to the subsequent valve sequence, while maintaining the original opening time at the remaining positions, thus obtaining a stable valve control timing sequence; as detailed below: Based on the valve opening sequence, read the adjusted opening time intervals of adjacent solenoid valves and distribution valves, and extract them from the timing chain list. Opening time and Opening time Perform subtraction to obtain the adjacent opening interval. Retrieve historical log records of the dry oil lubrication device under stable operation over the past 24 hours, extract the average physical time taken for the dry oil distributor piston in the main pipeline to complete a full one-way oil discharge stroke and release the remaining pressure in the pipeline below the 1.0 MPa safety reset value, and set this as the preset stable interval. This constant value is derived from physical engineering parameters such as the main pipeline length of 20 meters, the NLGI-2 grade grease cone penetration of 265, and the gear pump displacement of 2.0 ml / min, which are input into the empirical formula for fluid friction resistance. The calculation is performed, and the formula is an empirical expression fitted based on field engineering experimental data. The system coefficients in the formula implicitly include physical unit conversion factors between different operating conditions. The value is 0.5, representing a correction factor for local resistance and structure of the pipeline network, and calibrated through field testing. The main road is 20 meters long. The equivalent apparent viscosity correction factor for the grease is based on a cone penetration of 265° and measured by a rotational rheometer under standard operating conditions at 25°C, with a calibration value of 3.0. Given a gear pump with a rated discharge capacity of 2.0 ml / min, the result can be directly calculated by substituting the values ​​into the algebraic equation. seconds, of which and Indicates the absolute time of the initial opening of adjacent valves. This represents the calculated interval time. This indicates the minimum stable safety interval required for system pressure relief and reset. A conditional logic judgment is executed, comparing 10 seconds and 15 seconds. If the proposition that 10 seconds is less than 15 seconds is true, the instruction interval is identified as being within an unstable interference range that does not meet the unloading and reset requirements. Then, the timer increment algorithm is invoked to perform time axis delay reconstruction calculations. ,Will The control command trigger time was updated from 10 seconds to 15 seconds, and then read... The original opening time was 20 seconds; calculate its timeframe compared to the updated time. The actual time difference between them is 20 seconds minus 15 seconds, which equals 5 seconds. The comparison with 15 seconds is performed again, and it is determined that 5 seconds is less than 15 seconds. Continue to perform item-by-item recursive processing via timer instructions. ,in and This represents the revised scheduling timestamp. For subsequent valve opening points with intervals greater than or equal to 15 seconds, their original relative offsets remain unchanged. All safety timestamp data after the shift correction are statistically analyzed, and the updated timestamps are then used. The sequence is rewritten into the task scheduler storage area of ​​the control system to obtain a stable valve control timing sequence.

[0028] The flow verification submodule calls the stable valve control timing sequence, collects the actual flow rate changes of lubricating oil delivery according to the delayed opening time of each valve, obtains the target flow rate changes corresponding to the oil supply demand of each lubrication point, compares the actual flow rate and the target flow rate point by point according to the same opening time, and determines whether the direction and magnitude of the changes are consistent. Points where the direction or magnitude is inconsistent are registered as flow deviation locations, associated with valve number, lubrication point number, and opening time, and generate dry oil lubrication control results; specifically as follows: The stable valve control timing is invoked, and the actual flow rate change of lubricating oil is collected according to the delayed opening time of each valve. During the 0-15 second oil discharge window after opening, the actual oil flow rate data is extracted by cyclic counting using an inductive proximity sensor closely attached to the side of the piston indicator pin of distributor No. 1. The sensor sends the number of piston movements in pulse form to the high-speed counter module of the PLC. It reads that the piston completes two cycles within this window. Given that the geometric displacement of this specification of distributor valve in a single reciprocating cycle is... Performing multiplication operations yields the actual flow volume value. ; The design target flow rate requirement for lubrication point 1 under high-density and abnormally severe conditions was retrieved from the process parameter database. The target value corresponds to three complete piston cycle pulses, and the subtraction unit is called to perform the absolute difference calculation. Set the allowable threshold for system flow deviation. for This limit is calibrated in engineering based on a 5% physical hysteresis error of the proximity sensor combined with the compressibility coefficient of the lubricating grease. This indicates the actual displacement measured by the sensor. This indicates the target displacement set by the control. This indicates the upper limit of the allowable error, for comparison. and The numerical value is determined, and 0.6 is greater than 0.1. Therefore, the value is... The time point was recorded as the flow deviation location, and an algebraic sign direction comparison was performed. It was found that the actual measured flow rate was lower than the process target flow rate, and its specific deviation nature direction was recorded as "negative loss". This location was then associated with the hardware number of solenoid valve No. 1, the physical number of the lubrication point of spindle bearing No. 1, and the start-up time. Perform multidimensional data structure binding, for and During the fuel injection period, the above high-speed pulse flow acquisition and comparison operation is repeated. If the difference between the actual displacement of 1.2 cubic centimeters and the target displacement of 1.2 cubic centimeters is less than the threshold, then the period is marked as a normal state with matching flow. Finally, the valve index, time coordinate and deviation amplitude of all valves with flow deviation records are summarized to generate dry oil lubrication control results.

[0029] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A dry lubrication control system based on AI self-learning operation, characterized in that, The system includes: The energy harvesting module reads the current and voltage of the main drive motor in sequence, aligns and converts the input power sequence in time, compares adjacent power states and writes power change identifiers, coaxially identifies temperature changes, combines material thermal hysteresis to determine friction consumption changes and verify the direction, and generates a lubrication energy residual feature set. The latent variable construction module filters energy deviation time points based on the lubrication energy residual feature set, calls temperature change identifiers on the same time axis, compares the deviation direction with the temperature state, marks the same direction, opposite direction or no response, merges continuous and identical relationships and divides into segments, and generates lubrication state segment identifier results. The status recognition module, based on the lubrication status segment identification results, calls the corresponding segment oil pressure and impact acceleration data, compares the operating range point by point, extracts and merges abnormal time sequence nodes, writes oil pressure, impact or noise abnormality identification, divides the density level according to the interval of abnormal time sequence nodes, and generates lubrication abnormality density identification results. The priority adjustment module locates the location of density change and abnormal time period based on the abnormal density identification result, associates the abnormal source with the lubrication point, compares the associated and unassociated oil supply intervals, determines the lubrication point to be adjusted, shortens its oil supply interval when the density is high or when it turns to high density, and generates a lubrication oil supply interval configuration sequence.

2. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The lubrication energy residual feature set includes energy deviation level, energy deviation duration interval and energy deviation distribution category. The lubrication state segment identification results include unidirectional segments, reversible segments and no-response segments. The lubrication anomaly density identification results include low-density anomaly regions, medium-density anomaly regions, high-density anomaly regions and anomaly source combination categories. The anomaly source combination categories include oil pressure anomaly categories, impact anomaly categories, friction noise anomaly categories and their combination anomaly categories. The lubrication oil supply interval configuration sequence includes basic oil supply interval, shortened oil supply interval and maintained oil supply interval.

3. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The energy harvesting module includes: The electrical parameter processing submodule reads the input current and input voltage values ​​of the main drive motor according to the operating acquisition sequence of the dry oil lubrication device. It aligns the corresponding records with the acquisition time as the reference, converts the current and voltage values ​​according to the corresponding time points and organizes them into an input power sequence. It sequentially associates each acquisition time point and arranges them according to the acquisition order. It compares the adjacent power states point by point along the sequence, writes the power increase, decrease or stability into the corresponding identifier, and generates a sequence of input power change. The heat consumption determination submodule calls the time axis corresponding to the input power change sequence, collects temperature data of the bearing friction area and connects it to the same collection time point, identifies the temperature rise, fall or stable state point by point, forms temperature change identifier, obtains the corresponding relationship of bearing material thermal characteristics and thermal response hysteresis relationship, and determines the state of increased, decreased or unchanged friction consumption according to the temperature change direction and hysteresis corresponding order, and obtains the friction consumption change sequence. The direction verification submodule, based on the input power change sequence and the friction consumption change sequence, checks the input power change direction and the friction consumption change direction point by point along the same time axis. Points with consistent directions are registered as normal energy corresponding states, and points with inconsistent directions and points without corresponding responses are registered as energy deviation states. The module collects the state identifiers, deviation directions and state duration sequences of each time point according to the acquisition time, and generates a lubrication energy residual feature set.

4. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The latent variable construction module includes: The temperature mismatching submodule filters energy deviation time points based on the energy deviation state in the lubrication energy residual feature set, calls the temperature change identifier under the same time axis, reads the deviation direction and temperature rise, fall or stable state according to the time point, records the same direction as the same direction relationship, the opposite direction as the opposite relationship, and the stable temperature or no corresponding response as the no response relationship, writes the relationship code according to the time sequence, and generates the temperature mismatch relationship code. The relation merging submodule, based on the temperature deviation relation encoding, reads the corresponding relations of each energy deviation time point in chronological order, compares the current time point relation with the previous time point relation, performs sequential merging for time points with consecutive and consistent relation encodings, ends the current merging at the relation encoding conversion position and starts the subsequent merging, records the start time, end time and corresponding relation category of each segment, extracts the duration of each segment according to the start and end times, and obtains the state segment duration. The time-series delimitation submodule calls the state segment duration, reads the duration and relationship category of adjacent state segments, obtains the upper and lower boundaries corresponding to the preset interval range, compares the duration of each segment with the interval boundary according to the time sequence, writes the same segment identifier for state segments falling into the same interval, writes the corresponding category identifier for state segments crossing the interval boundary, and sequentially associates the start and end time, duration and relationship category of the segment to generate the lubrication state segment identifier result.

5. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The status recognition module includes: The range verification submodule, based on the state segment identifier in the lubrication state segment identifier result, calls the dry oil distributor outlet oil pressure data, bearing housing vibration impact acceleration data, and bearing friction area friction noise data within the corresponding time range. It reads the oil pressure operation status, impact acceleration operation status, and friction noise operation status according to the time axis, obtains the upper and lower boundaries of their respective operation ranges, and judges whether the collected status exceeds the corresponding boundary point by point. The time points of exceeding the boundary are registered as oil pressure abnormal time points, impact abnormal time points, and noise abnormal time points, respectively, and an abnormal time point sequence is generated. The node orchestration submodule calls the abnormal time point sequence, merges the oil pressure abnormal time point, impact abnormal time point and noise abnormal time point according to the collection time, retains the original time and state segment of each node, writes oil pressure abnormality identifier, impact abnormality identifier or noise abnormality identifier according to the node source, arranges all nodes in chronological order, calculates the time interval between adjacent abnormal time sequence nodes, obtains the preset abnormal node interval threshold, and maps each time interval to the threshold division interval to obtain the abnormal node density level. The density positioning submodule, based on the density level of the abnormal nodes, reads the density level of abnormal time-series nodes in chronological order within the time range corresponding to each state segment identifier, associates the abnormal source identifier, node time and its state segment, compares the current node density level with the previous node density level, registers the level-maintaining position as a continuing node, marks the level-transition position as the density change position, and determines the abnormal source combination category based on the oil pressure abnormality identifier, impact abnormality identifier, and noise abnormality identifier. It then sequentially aggregates the state segment identifier, density level, abnormal source combination category, and density change position to generate the lubrication abnormality density identification result.

6. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The priority adjustment module includes: The time period association submodule determines the abnormal time period corresponding to the abnormal time sequence node based on the density change position in the lubrication abnormal density identification result, and associates the corresponding lubrication point with the oil pressure abnormality identifier, impact abnormality identifier, or noise abnormality identifier. It reads the time and source identifier of the abnormal node before and after the density change position, and groups the nodes with the same source or the same source combination and continuous time into the same abnormal time period. According to the lubrication point number, density level and abnormal source combination category associated with each time period, it filters the time period that the source identifier can correspond to the lubrication point and generates an abnormal time period point mapping. The point identification submodule calls the point mapping of the abnormal time period to obtain the current oil supply interval of each lubrication point. It sorts the oil supply intervals of associated and unassociated lubrication points according to the abnormal time period, keeping the time range, collection order and lubrication point number consistent. It compares the oil supply interval status of the two types of lubrication points in the same abnormal time period, registers the lubrication points that are associated with the abnormal source and have different interval status as objects to be adjusted, and associates them with the corresponding density level to obtain a list of oil supply objects. The sequence arrangement submodule, based on the list of lubrication objects, calls the density level and density change position of each lubrication point to be adjusted during the abnormal time period, determines whether the object to be adjusted is in a high density or has changed from low density or medium density to high density, performs lubrication interval shortening processing on lubrication points that meet the conditions, and keeps the lubrication interval of non-lubrication points unchanged. The lubrication intervals of each lubrication point are arranged according to the running time to generate a lubrication lubrication interval configuration sequence.

7. The dry lubrication control system based on AI self-learning operation according to claim 6, characterized in that: The process of grouping nodes with the same source and consecutive time into the same abnormal time period is as follows: The abnormal source identifier and node time of each abnormal time sequence node are read in the order of collection time. The abnormal source identifier includes oil pressure abnormality identifier, impact abnormality identifier and noise abnormality identifier. Abnormal time sequence nodes with the same abnormal source identifier or the same abnormal source combination category and arranged continuously on the time axis are merged in sequence. The current merging ends when the abnormal source identifier changes, the abnormal source combination category changes, or the time continuity is interrupted, forming the corresponding abnormal time period and recording the start time and end time.

8. The dry lubrication control system based on AI self-learning operation according to claim 6, characterized in that: The process of comparing the oil supply interval status of two types of lubrication points within the same abnormal time period is as follows: The oil supply intervals of associated lubrication points and non-associated lubrication points are arranged in the same time range and collection order. The oil supply interval status is read point by point and compared. When there is an inconsistency in the oil supply interval status, the corresponding lubrication point is marked as the object to be adjusted, and the density level of the corresponding abnormal time period is associated. The process of shortening the oil supply interval for lubrication points that meet the requirements is as follows: The objects to be adjusted are screened according to the density level and density change location of the corresponding abnormal time period. For the objects to be adjusted that are in a high density state or have changed from low density, medium density to high density, the oil supply interval is reduced. The oil supply interval of the unscreened lubrication points remains unchanged. At the same time, the adjusted oil supply intervals are rearranged according to the running time sequence.

9. The dry lubrication control system based on AI self-learning operation according to claim 1, characterized in that: The system also includes: The valve control reconfiguration module calls the adjusted oil supply interval in the lubrication oil supply interval configuration sequence, forms the opening sequence according to the correspondence between the lubrication point and the solenoid valve and the distribution valve, corrects the position of inconsistent oil supply interval sorting, extends the position of the unstable interval, and generates dry oil lubrication control results by comparing the actual and target flow changes. The dry oil lubrication control results include flow deviation position, flow matching status, and valve control execution status.

10. The dry lubrication control system based on AI self-learning operation according to claim 9, characterized in that: The valve-controlled reconfiguration module includes: The valve sequence arrangement submodule calls the adjusted oil supply interval of each lubrication point in the lubrication oil supply interval configuration sequence, obtains the preset control correspondence between each lubrication point and the dry oil pump outlet solenoid valve and distribution valve, associates the corresponding solenoid valve and distribution valve according to the oil supply interval sequence, arranges the opening order of each valve to form the valve control opening sequence, compares the oil supply interval sorting with the valve control opening sequence, adjusts the positions with inconsistent sequences to the corresponding lubrication point oil supply sequence, and generates the valve opening sequence. The interval extension submodule reads the opening time interval of adjacent solenoid valves and distribution valves after adjustment based on the valve opening sequence, collects the corresponding time interval of continuous and stable opening of each valve during the operation of the dry oil lubrication device, organizes it into a preset stable interval, compares the opening time interval of each adjacent valve with the corresponding preset stable interval, and extends the position that is less than the lower limit of the stable interval item by item according to the subsequent valve sequence, while keeping the original opening time of the remaining positions to obtain a stable valve control timing sequence. The flow verification submodule calls the stable valve control timing, collects the actual flow rate change of lubricating oil delivery according to the delayed opening time of each valve, obtains the target flow rate change corresponding to the oil supply demand of each lubrication point, compares the actual flow rate and the target flow rate point by point according to the same opening time, and judges whether the change direction and change magnitude are consistent. The time point where the direction or magnitude is inconsistent is registered as the flow deviation position, associated with the valve number, lubrication point number and opening time, and generates dry oil lubrication control result.

Images