Grinding mill early warning method and system based on key component temperature rise trend analysis

Abstract

The application relates to the technical field of temperature measurement, in particular to a mill early warning method and system based on key component temperature rise trend analysis, which comprises the following steps: collecting temperature data and load data of key components of a mill; constructing an operation state characteristic index based on the fluctuation degree of the load data; adjusting a weight coefficient decay rate according to the operation state characteristic index, weighting and averaging the load data based on the weight coefficient to obtain effective load; statistically obtaining a reference characteristic value of the key components in each load interval; constructing a heat dissipation efficiency correction coefficient based on the difference between the current temperature data of the key components and an environmental reference temperature; multiplying the reference characteristic value of the load interval to which the current effective load belongs by the heat dissipation efficiency correction coefficient to obtain an expected temperature rise rate reference, and combining a measured temperature rise rate to calculate a trend deviation degree; comparing the trend deviation degree with preset multistage threshold values, and executing early warning actions of corresponding levels. The application improves early warning accuracy and reliability.

Description

Technical Field

[0001] This invention relates to the field of temperature measurement technology. In particular, it relates to a mill early warning method and system based on the analysis of temperature rise trends in key components. Background Technology

[0002] Key components of a mill, such as the main motor windings, bearings, and reducers, generate heat under prolonged high-load conditions. If the temperature rises abnormally and is not addressed promptly, it can lead to equipment damage or even shutdown. Therefore, early warning methods based on the temperature rise trend analysis of key components can identify potential faults in advance, enabling a shift from reactive to preventative maintenance, which is essential for ensuring the continuous and stable operation of equipment.

[0003] Existing mill temperature early warning methods mainly employ static threshold alarms or simple load-temperature linear correlation models. Static threshold methods set a fixed upper temperature limit, triggering an alarm when the measured temperature exceeds this limit. However, this method ignores the normal impact of load fluctuations on temperature: under high load conditions, the increased heat generation of the equipment and the resulting temperature rise are normal physical phenomena, but the static threshold method easily misinterprets this as a fault, leading to a high false alarm rate; under low load fault conditions, the equipment may experience abnormal temperature rises due to localized wear or poor cooling, but because the overall load is low, the absolute temperature value has not yet reached the fixed threshold, making it easy for this method to miss real faults.

[0004] A simple load-temperature linear correlation model attempts to establish a linear relationship between load and temperature, estimating the expected temperature based on the current load and then comparing it with the measured temperature for judgment. However, this method does not consider the thermal hysteresis effect of temperature changes; that is, after a load change, there is a time delay in the temperature response, and heat needs a certain amount of time to accumulate before it is reflected in the temperature. Because this physical law is ignored, the model suffers from a time-domain mismatch between load and temperature when the load fluctuates rapidly, leading to a systematic deviation between the expected and measured temperatures, thus affecting the accuracy and reliability of the early warning.

[0005] Therefore, there is a need in the field for a mill early warning method and system based on the analysis of the temperature rise trend of key components to solve the problem of low accuracy and reliability of the existing mill early warning methods. Summary of the Invention

[0006] To address the technical problem of low accuracy and reliability of existing mill early warning methods, the present invention provides solutions in the following aspects.

[0007] In the first aspect, the mill early warning method based on the temperature rise trend analysis of key components includes: S1: Collect temperature data of key components of the mill and load data of the mill, and perform preprocessing; S2: Construct an operating status characteristic index based on the fluctuation of load data within a preset time window; S3: Set the weight coefficient, adjust the decay rate of the weight coefficient according to the running status characteristic index, and perform a weighted average of the load data within the preset compensation window based on the weight coefficient to obtain the effective load; wherein, the weight coefficient decays as the time distance between the acquisition time and the current time within the compensation window increases; S4: Obtain the effective load range and divide it into multiple load intervals. Based on the historical temperature data of each key component, calculate the upper limit of the normal temperature rise rate of the key component in each load interval and use it as the reference characteristic value of the load interval. Based on the difference between the current temperature data of the key component and the preset ambient reference temperature, construct a heat dissipation efficiency correction coefficient that decreases as the difference increases. Determine the load interval to which the effective load belongs at the current moment, and multiply the reference characteristic value of the key component in the load interval by the heat dissipation efficiency correction coefficient to obtain the expected temperature rise rate reference of the key component. S5: Calculate the trend deviation based on the measured temperature rise rate and the expected temperature rise rate of the key components at the current moment; S6: Compare the trend deviation with the preset multi-level thresholds and execute the corresponding level of warning action.

[0008] Preferably, step S2 includes: acquiring a first load data sequence within a preset time window ending at the current time; calculating the sum of the average value of the first load data sequence and a preset first zero-prevention constant, and using it as the first sum; calculating the ratio of the standard deviation of the first load data sequence to the first sum to obtain the operating state characteristic index.

[0009] Preferably, the method for calculating the weight coefficient in step S3 includes: setting a basic time constant and an adjustment coefficient; calculating the product of the adjustment coefficient and the operating state characteristic index; calculating the sum of the product and the basic time constant to obtain the final time constant; calculating the ratio of the time distance to the final time constant; using the negative of the ratio as the exponent and the natural constant e as the base to construct a weight exponent function; and using the obtained weight exponent function value as the weight coefficient.

[0010] Preferably, the method for obtaining the reference characteristic value of the load range in step S4 includes: for each key component, based on the health status data selected from its historical temperature data, the normal temperature rise rate in each load range is statistically analyzed, and the 95th percentile is taken as the upper limit of the normal temperature rise rate of the load range, i.e., the reference characteristic value of the load range; wherein, the health status data refers to the temperature data of the mill operating under normal conditions with no faults, no abnormal wear, and normal cooling system operation.

[0011] Preferably, the step S4 of constructing a heat dissipation efficiency correction coefficient that decreases as the difference increases includes: setting an adjustment coefficient, multiplying the difference between the current temperature data of the key component and the preset ambient reference temperature by the adjustment coefficient, using the negative number of the resulting product as the exponent, using the natural constant e as the base, constructing a heat dissipation efficiency correction function, and using the resulting heat dissipation efficiency correction function value as the heat dissipation efficiency correction coefficient.

[0012] Preferably, step S5 includes: for each key component, calculating the difference between the temperature data at the current time and the previous time, and calculating the ratio of the difference to the sampling interval to obtain the measured temperature rise rate; calculating the absolute difference between the measured temperature rise rate and the expected temperature rise rate benchmark; calculating the sum of the expected temperature rise rate benchmark and the preset second zero constant as the second sum; and calculating the ratio of the absolute difference to the second sum to obtain the trend deviation.

[0013] Preferably, step S6 includes: setting multiple threshold levels, including a first threshold, a second threshold, and a third threshold, wherein the first threshold is less than the second threshold and less than the third threshold; triggering a prompt-level warning when the trend deviation of a key component is greater than or equal to the first threshold and less than the second threshold; triggering a warning-level warning when the trend deviation of a key component is greater than or equal to the second threshold and less than the third threshold; and triggering a shutdown action when the trend deviation of a key component is greater than or equal to the third threshold.

[0014] Preferably, the conditions for triggering the shutdown action also include: when the temperature data of the critical component at the current moment reaches the system's preset safety limit, the shutdown action is triggered.

[0015] Preferably, the temperature data of the key components includes the main motor winding temperature, the main motor bearing temperature, the reducer oil temperature, the cylinder end bearing temperature, and the mechanical seal temperature; the load data includes the main motor current.

[0016] Secondly, a mill early warning system based on the analysis of temperature rise trends of key components includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned mill early warning method based on the analysis of temperature rise trends of key components is implemented.

[0017] The present invention has the following effects: 1. This invention quantifies the operational stability of equipment by constructing an operational status characteristic index, and dynamically adjusts the weight decay rate of the thermal hysteresis compensation window accordingly. It calculates the effective load to eliminate the influence of thermal hysteresis, so that the temperature rise rate benchmark matches the current operating conditions, reduces the distortion of temperature rise trend judgment caused by load fluctuations, and significantly reduces the false alarm rate.

[0018] 2. This invention establishes independent segmented statistical mapping models for different key components, enabling each key component to have independent baseline characteristic values ​​and temperature correction logic. This allows it to adapt to the differences in thermophysical characteristics of monitoring points of different key components, avoiding misjudgment of components with large thermal inertia and underreporting of components with small thermal inertia, thus achieving targeted early warning performance.

[0019] 3. Since higher temperatures lead to faster heat dissipation, the actual temperature rise rate naturally slows down. Therefore, this invention introduces a heat dissipation efficiency correction coefficient based on the current temperature data. This heat dissipation efficiency correction coefficient decreases as the temperature of key components increases, so that the expected temperature rise rate benchmark is reduced accordingly at high temperatures. This solves the problem that the benchmark value of the traditional model deviates from the actual thermal equilibrium state under high temperature steady-state conditions, and effectively avoids false alarms or missed alarms under high temperature conditions. Attached Figure Description

[0020] Figure 1 This is a flowchart of steps S1-S6 in a mill early warning method based on the temperature rise trend analysis of key components according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0022] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] Reference Figure 1 The mill early warning method based on the temperature rise trend analysis of key components includes steps S1-S6, as follows: S1: Collect temperature data of key components of the mill and load data of the mill, and perform preprocessing; The temperature of the three-phase windings of the main motor is collected in real time at a sampling frequency of 1Hz by a temperature resistance temperature sensor integrated inside the main motor. The maximum value of the three-phase winding temperature is taken as the main motor winding temperature, which is used to characterize the thermal state of the winding of the main motor that is most prone to thermal failure under the current load and heat dissipation conditions.

[0024] Temperature sensors installed in the main motor bearing housing collect the temperatures of the front and rear bearings of the main motor at a sampling frequency of 1Hz, and take the maximum value as the main motor bearing temperature. This is used to characterize the health of the main motor mechanical transmission chain, that is, to characterize the worst thermal state of the entire main motor bearing assembly, ensuring that no abnormal temperature rise of any bearing is missed, thereby most sensitively capturing potential faults in the mechanical transmission chain.

[0025] Temperature sensors installed on the bearing housings at the reducer and mill cylinder end are used to collect the reducer oil temperature and the temperatures of the front and rear bearings at the cylinder end at a sampling frequency of 1Hz. The maximum value is taken as the cylinder end bearing temperature. The reducer oil temperature reflects the state of the lubricating oil film inside the reducer and the heat generated by gear meshing; the cylinder end bearing temperature reflects the friction and load experienced by the bearings during mill cylinder rotation. The reducer oil temperature and the cylinder end bearing temperature together characterize whether the lubrication of the mechanical transmission system is adequate and whether the grinding load is within the normal range.

[0026] The temperature of the mechanical seal is collected at a sampling frequency of 1Hz by a temperature sensor installed in the mechanical seal cavity to characterize the sealing, lubrication and cooling effects.

[0027] The main controller reads the inverter feedback signal and collects the main motor current at a sampling frequency of 1Hz to characterize the heat source intensity and operating load rate of the grinding operation.

[0028] This yields temperature data: main motor winding temperature, main motor bearing temperature, reducer oil temperature, cylinder end bearing temperature, and mechanical seal temperature; as well as load data: main motor current.

[0029] The 3σ principle is used to detect outliers in temperature and load data, eliminating spike noise that exceeds physical limits caused by momentary sensor failures or electromagnetic interference, and retaining valid signals that conform to the thermal inertia of the equipment to ensure the physical authenticity of the data source.

[0030] A sliding window filter with a window size of 5 is used to smooth the temperature and load data after cleaning, suppressing high-frequency quantization noise and fluid pulsation interference from the sensor, and retaining the low-frequency effective signal that reflects the trend of heat load change.

[0031] Using a sampling frequency of 1Hz as a reference, multi-source heterogeneous data are synchronized and aligned to ensure that temperature data and load data within the same control cycle have a unified time reference, thus eliminating correlation analysis errors caused by time asynchrony.

[0032] This yields the pre-processed temperatures of the main motor windings, main motor bearings, reducer oil, cylinder end bearings, mechanical seal, and main motor current. The following steps all utilize the pre-processed temperature and load data.

[0033] S2: Construct an operating status characteristic index based on the degree of fluctuation of load data within a preset time window.

[0034] Obtain the first load data sequence within a preset time window ending at the current time; calculate the sum of the average value of the first load data sequence and a preset first zero-prevention constant, and use this sum as the first sum value; calculate the ratio of the standard deviation of the first load data sequence to the first sum value to obtain the operating state characteristic index. The specific formula is as follows: In the formula, Indicates the current time Operating status characteristic index; Indicates the current time The standard deviation of the main motor current sequence within a preset time window with the endpoint as the endpoint; Indicates the current time The average value of the main motor current sequence within a preset time window with the endpoint as the endpoint; This represents the first zero-prevention constant. To prevent calculation overflow when the average value of the load data is zero and the denominator is zero, a first zero-prevention constant is added to the denominator. , The value should be determined based on the current rating of the equipment, and should be much smaller than the normal load current. For example, for a medium-sized mill with a rated current of 500A, A value of 1.0 can be used; for small mills with a rated current of less than 100A, A value of 0.1 can be used; in this embodiment, the rated current of the mill is 600A, therefore Take 1.0.

[0035] It should be noted that while the main motor current is used as the representative parameter for the load data in this embodiment, it is not the only option. In other embodiments, the load data can also be replaced by the active power signal or output torque signal of the main motor.

[0036] The length of the time window is preset according to the mill model. For example, for large mills, the length of the time window can be 120 to cover a longer load response cycle; for small mills, the length of the time window can be 30 to improve the response speed of status recognition.

[0037] The construction of the operating status characteristic index integrates two dimensions: absolute and relative load fluctuations. The standard deviation of the first load data sequence is used as the numerator to quantify the absolute fluctuation amplitude of the load, conforming to the basic laws of mechanical motion stability. The mean of the first load data sequence is used as the denominator to normalize the absolute fluctuation into a relative volatility, thereby eliminating misjudgments caused by the large absolute value of normal fluctuations under high load conditions, and achieving dimensionlessness. Dividing the two values ​​provides a unified and comparable standard for the fluctuation characteristics under different load levels.

[0038] The operating state characteristic index directly reflects changes in the physical operating state of the mill, and its value has a clear physical meaning. When the load data is completely constant, i.e., the equipment is in an ideal steady state, the index approaches zero, indicating that the system is in a highly stable grinding period or an idle standby state. In this case, subsequent steps will accelerate the decay rate of the thermal hysteresis compensation weight based on the operating state characteristic index, making the effective load closer to the instantaneous load, thereby ensuring that the benchmark model maintains high sensitivity to steady-state changes and avoiding missed detections during stable operation.

[0039] When load data fluctuations intensify or equipment is in a transient state during startup or shutdown, the operating status characteristic index is significantly greater than zero, indicating that the system is experiencing sudden high-dynamic load characteristics. In this case, subsequent steps will slow down the decay rate of the thermal hysteresis compensation weight based on the operating status characteristic index, allowing the effective load to retain more historical load information. This smooths out the interference of instantaneous fluctuations on the baseline modeling, avoids false alarms during startup due to insufficient thermal hysteresis compensation, and achieves high robustness in early warning performance of the system.

[0040] Through the above methods, the operational state characteristic index constructed in this step provides a basis for state-adaptive regulation of subsequent thermal hysteresis compensation, effectively solving the contradictory requirement of the early warning system to have high sensitivity in the stable period and anti-interference capability in the transient period.

[0041] S3: Set the weight coefficient, adjust the decay rate of the weight coefficient according to the running status characteristic index, and perform a weighted average of the load data in the preset compensation window based on the weight coefficient to obtain the effective load; wherein, the weight coefficient decays as the time distance between the acquisition time and the current time in the compensation window increases.

[0042] In the analysis of temperature rise trends in key components of a mill, a significant physical time difference exists between load changes and temperature response, known as the thermal hysteresis effect. Traditional methods directly correlate instantaneous load with current temperature, ignoring the physical law that heat accumulation requires time. This leads to a time-domain mismatch between the model baseline and the measured temperature when the load fluctuates rapidly. To address this, this solution constructs an effective load to align the time domain of the load input and the temperature response.

[0043] The system acquires a second load data sequence within a preset compensation window, ending at the current time. It calculates the time distance between the acquisition time of each load data point within the compensation window and the current time, and calculates the weighting coefficient for each load data point based on the time distance and the operating state characteristic index. The effective load is obtained by weighting the second load data sequence based on these weighting coefficients. The weighting coefficients decrease as the time distance increases, and the rate of decrease is adjusted according to the operating state characteristic index.

[0044] The step of calculating the weight coefficient of each load data based on the time distance and the operating state characteristic index includes: setting a basic time constant and an adjustment coefficient, calculating the product of the adjustment coefficient and the operating state characteristic index, calculating the sum of the product and the basic time constant to obtain the final time constant; calculating the ratio of the time distance to the final time constant, taking the negative of the ratio as the exponent, and taking the natural constant as the base to construct a weight exponential function, and the obtained value of the weight exponential function is taken as the weight coefficient. The specific formula is expressed as follows: In the formula, represents the weight coefficient that varies with the time distance and the operating state characteristic index . represents the time step index, that is, the time distance between the collection moment of the load data in the compensation window and the current moment , when it is , it represents the current moment, when it is , it represents the farthest historical collection moment in the compensation window, which is dimensionless. represents the length of the compensation window, whose value is preset according to the mill model and the thermal inertia characteristics of key components. For example, for a large-model mill with large thermal inertia, the heat accumulation period is long, can be set to 1800; for a small-model mill with small thermal inertia, the response speed requirement is high, can be set to 180. represents the final time constant, , represents the basic time constant. When , , at this time the weight coefficient attenuates at the basic speed. In this embodiment, , the value of can be set according to the thermal response speed of the equipment in the steady state. A larger value is taken for equipment with large thermal inertia, and a smaller value is taken for equipment with small thermal inertia; represents the adjustment coefficient. In this embodiment, , the value of can be set according to the system's sensitivity requirement for load fluctuation. A larger value is taken when strong smoothing anti-interference is required, and a smaller value is taken when high-sensitivity response is required; and the specific values of can be obtained through offline calibration or empirical setting according to the equipment model and on-site working conditions. is the natural constant.

[0045] The calculation formula of the effective load is expressed as follows: In the formula, Indicates the current time The effective payload; Indicates the time step index; Indicates the length of the compensation window; Indicates distance over time and operating status characteristic index The changing weighting coefficients; Indicates the time of data collection The load data, in this embodiment, refers to the main motor current.

[0046] when When the load is relatively large, i.e., when there are significant load fluctuations, the final time constant increases, the weight decays more slowly, and the effective load retains more historical load information, thus smoothing out the interference of instantaneous fluctuations on the baseline modeling; when When the weight is smaller, meaning the equipment is in a stable state, the final time constant decreases, and the weight decays more rapidly, making the effective load closer to the current instantaneous load and ensuring that the baseline model remains highly sensitive to steady-state changes. Besides constructing a weight exponential function, other forms of weight functions such as linear decay functions or Gaussian functions can also be used, all of which fall within the scope of this invention.

[0047] Payload It characterizes the actual heat input intensity acting on key components, eliminating the time mismatch between instantaneous load fluctuations and temperature response. When the load is stable and the operating state characteristic index approaches zero, the effective load approaches the current instantaneous load, indicating that the system is in thermal equilibrium. When the load fluctuates drastically and the operating state characteristic index is significantly greater than zero, the effective load retains more historical high load information, indicating that the system is in the process of heat accumulation and can automatically compensate for the lag in temperature rise.

[0048] S4: Obtain the effective load range and divide it into multiple load intervals. Based on the historical temperature data of each key component, calculate the upper limit of the normal temperature rise rate of the key component in each load interval and use it as the reference characteristic value of the load interval. Based on the difference between the current temperature data of the key component and the preset ambient reference temperature, construct a heat dissipation efficiency correction coefficient that decreases as the difference increases. Determine the load interval to which the effective load belongs at the current moment, and multiply the reference characteristic value of the key component in the load interval by the heat dissipation efficiency correction coefficient to obtain the expected temperature rise rate reference of the key component.

[0049] In the analysis of temperature rise trends of key components in a mill, different key components, such as the main motor windings, main motor bearings, and reducer oil temperatures, have drastically different heat capacities, heat dissipation coefficients, and heat conduction paths. Using a single model can lead to sluggish responses from components with high thermal inertia, while components with high thermal sensitivity are prone to false alarms. Therefore, this solution independently constructs a dynamic expected temperature rise rate benchmark for each key component.

[0050] First, determine the effective load range. This range can be obtained by offline statistical analysis of historical normal operation data, specifically the minimum and maximum effective load values; if historical data is unavailable, it can also be preset based on the no-load current and rated current specified in the equipment's factory specifications. Divide the effective load range into... A discrete load range. The value is preset according to the equipment model and monitoring accuracy requirements. For example, it can be divided into ten load ranges to cover the entire operating condition from no-load to full-load. The length of each load range is calculated by dividing the measurement range by 1 / 2. Calculated.

[0051] Determine the load range to which the effective load belongs at the current moment, and set the indicator function of this load range to 1, while setting the indicator function of the other load ranges to 0, thereby achieving segmented matching of operating conditions.

[0052] For each critical component, based on health status data selected from its historical temperature data, the normal temperature rise rate within each load range is statistically analyzed, and the 95th percentile is taken as the baseline characteristic value for that load range. The baseline characteristic value represents the maximum allowable normal temperature rise rate of the critical component within that load range under healthy conditions. Health status data refers to data obtained when the mill is operating without faults, abnormal wear, and with the cooling system functioning normally.

[0053] Based on the difference between the current temperature data of key components and the preset ambient reference temperature, a heat dissipation efficiency correction function is constructed. This function decreases as the difference increases, for example, using a linear decreasing form or an exponential decay form. The value of the heat dissipation efficiency correction function is used as the heat dissipation efficiency correction coefficient. In practical applications, the range of values ​​for the heat dissipation efficiency correction coefficient is... This is to prevent the reference from amplifying abnormally or failing under extreme temperatures.

[0054] The ambient reference temperature is taken as the initial temperature at the beginning of equipment startup, before a significant temperature rise occurs. Specifically, it is obtained by continuously collecting temperature data of each key component for 30 seconds before the equipment is started from a cold state, and taking the average temperature during this period as the ambient reference temperature.

[0055] For each critical component, the indicator function value of the load range to which the current effective load belongs is multiplied by the reference characteristic value of the critical component in that load range. The resulting product is then multiplied by the heat dissipation efficiency correction factor for that critical component to obtain the reference expected temperature rise rate for that critical component. This is expressed by the following formula: In the formula, Indicates key components At the present moment The expected rate of temperature rise is the benchmark; Indicates the number of load ranges; Indicates the current time payload The indicator function value of the load range, when the effective load Falling into When there is a load range ,otherwise ; Indicates key components In the The baseline characteristic value of each load range, i.e., the maximum normal temperature rise rate; Indicates key components At the present moment The heat dissipation efficiency correction factor , Indicates key components At the present moment Temperature data, Indicates the ambient reference temperature. This represents the adjustment coefficient, used to control the rate at which the heat dissipation efficiency correction coefficient decreases as the temperature difference increases. The following principles apply to the selection of values: larger values ​​are used for key components with low thermal inertia and sensitive temperature response, and smaller values ​​are used for key components with high thermal inertia and slow temperature response. In this embodiment, the adjustment coefficient corresponding to the main motor winding is 0.012, the adjustment coefficient corresponding to the main motor bearing is 0.008, the adjustment coefficient corresponding to the reducer oil temperature is 0.004, the adjustment coefficient corresponding to the drum end bearing is 0.006, and the adjustment coefficient corresponding to the mechanical seal temperature is 0.015. Represents the natural constant.

[0056] Because higher temperatures in critical components result in greater temperature differences with the surrounding environment, leading to faster heat dissipation and a slower temperature rise rate, the permissible temperature rise rate should be correspondingly reduced. If this factor is ignored and the baseline characteristic value is used directly as the criterion, normal slowdowns in temperature rise under high-temperature conditions may be misjudged as abnormal, leading to missed alarms. Therefore, a heat dissipation efficiency correction coefficient is introduced to adjust the baseline characteristic value.

[0057] This embodiment uses an exponential decay function to correct the baseline characteristic value: when the temperature of the critical component is equal to the ambient reference temperature, the temperature difference is 0, the heat dissipation efficiency correction coefficient is 1, and the baseline characteristic value is not adjusted; when the temperature of the critical component increases, the temperature difference between the temperature data of the critical component and the ambient reference temperature is positive, the heat dissipation efficiency correction coefficient is less than 1, the baseline characteristic value is scaled, so that the expected temperature rise rate baseline is appropriately reduced; moreover, as the temperature of the critical component increases, the rate of decrease of the heat dissipation efficiency correction coefficient will gradually slow down, that is, the magnitude of the improvement in heat dissipation capacity will gradually decrease, which is completely consistent with the actual heat dissipation law.

[0058] When the temperature of a critical component is abnormally high, the heat dissipation efficiency correction factor is limited to a lower limit of 0.1 to prevent the expected temperature rise rate benchmark from being set too low, thus preventing the system from ever triggering an alarm. Conversely, when the temperature of a critical component is lower than the ambient reference temperature (such as when the equipment is first started), the heat dissipation efficiency correction factor is greater than 1, allowing the expected temperature rise rate benchmark to be appropriately adjusted upwards from the benchmark characteristic value. This avoids false alarms due to an excessively low expected temperature rise rate benchmark, while an upper limit of 2.0 ensures that the upward adjustment is not too large. Through these corrections, the expected temperature rise rate benchmark can more realistically reflect the normal thermal behavior of the equipment under different temperature conditions.

[0059] The expected temperature rise rate benchmark represents the expected healthy thermal state of key components under current load and serves as a reference line for fault diagnosis. When the equipment is in normal grinding operation and the load is stable, if the measured temperature rise rate is less than or equal to the expected temperature rise rate benchmark, it indicates that the component's heat generation and dissipation are in dynamic equilibrium, and the system determines it to be in a healthy state. When the equipment is in the early stages of a fault or the component's performance is declining, if the measured temperature rise rate is greater than this benchmark, it indicates that the component has independent faults such as abnormal friction, insulation aging, or cooling failure, and the system determines it to be an abnormal trend.

[0060] Because each key component possesses independent baseline characteristic parameters and temperature correction logic, the model can autonomously adjust its judgment criteria based on the thermal inertia and thermal state of each component. For components with high thermal inertia and slow temperature changes (such as gearbox oil temperature), the model adopts a relatively gradual baseline change trend to avoid misjudgments due to slow temperature response; for components with low thermal inertia and sensitive temperature changes (such as motor windings), the model employs a faster baseline response mechanism to avoid missing real faults due to rapid temperature fluctuations. In this way, this method achieves differentiated early warning for different component characteristics, significantly improving the accuracy and reliability of the system.

[0061] S5: Calculate the trend deviation based on the measured temperature rise rate and the expected temperature rise rate of the key components at the current moment.

[0062] After obtaining the baseline expected temperature rise rate for key components, it is necessary to quantify the difference between the measured temperature rise rate and the baseline expected temperature rise rate to provide a unified standard indicator for subsequent early warning judgments. To this end, this step constructs a trend deviation rate. This trend deviation rate, through normalization processing, eliminates the impact of differences in temperature magnitudes among different key components, achieving a unified early warning standard for multiple components.

[0063] For each key component, calculate the difference between the temperature data at the current moment and the previous moment, and calculate the ratio of this difference to the sampling interval to obtain the measured temperature rise rate; calculate the absolute difference between the measured temperature rise rate and the expected temperature rise rate benchmark; calculate the sum of the expected temperature rise rate benchmark and the preset second zero-prevention constant as the second sum; calculate the ratio of the absolute difference to the second sum to obtain the trend deviation. The specific formula is as follows: In the formula, Indicates key components At the present moment The degree of deviation from the trend; Indicates key components At the present moment The measured rate of temperature rise, , Indicates key components At the present moment Temperature data, such as the current temperature of the main motor bearing. Indicates key components In the previous moment Temperature data, Indicates the sampling interval; Indicates key components At the present moment The expected rate of temperature rise is the benchmark; This represents the second zero-prevention constant, in this embodiment. Taking 0.01 is far smaller than the normal temperature rise rate and will not have a substantial impact on the calculation results of the trend deviation. In practical applications, It can also be adjusted within the range of 0.001 to 0.1 according to the on-site working conditions.

[0064] The numerator of the trend deviation is the absolute value of the difference between the measured temperature rise rate and the expected temperature rise rate benchmark, directly reflecting the degree to which the current temperature rise behavior deviates from the healthy state. Using absolute values ​​means the system can simultaneously monitor two abnormal modes: one is excessively rapid temperature rise, corresponding to overheating faults; the other is excessively slow temperature rise, corresponding to sensor drift or abnormal load loss. The denominator is normalized using the expected temperature rise rate benchmark, eliminating differences in judgment standards caused by different temperature change magnitudes among different components, achieving dimensionlessness, and providing a unified deviation scale for all key components.

[0065] When the trend deviation When the temperature approaches 0, it indicates that the measured temperature rise is highly consistent with the expected temperature rise rate, and the component is in good condition; when the trend deviates... A value significantly greater than 0 indicates that the component exhibits an independent failure trend, such as abnormal friction, insulation aging, or cooling failure. The trend deviation directly reflects the degree of deterioration in the health status of critical components and will serve as the core input indicator for the next step of early warning determination.

[0066] S6: Compare the trend deviation with the preset multi-level thresholds and execute the corresponding level of warning action.

[0067] The purpose of this step is to compare the calculated trend deviation with the preset multi-level thresholds, and to execute the corresponding level of early warning or protection action according to the degree of deviation, so as to achieve a progressive risk response from prompt to shutdown.

[0068] Three thresholds are preset, each increasing progressively: the first threshold... Second threshold and the third threshold ,and The trend deviation of each key component is compared with the above thresholds, and the following hierarchical logic is executed: The first level is the alert level: when At this point, it indicates that the mill is showing a slight abnormal trend but has not yet reached a dangerous level. The system then displays a prompt on the human-machine interface, recording the abnormal parameters and the time of occurrence, but does not trigger audible or visual alarms, providing operators with an initial risk warning.

[0069] The second level is the warning level: when When the temperature rises significantly, it indicates a clear abnormal trend in the mill, requiring timely intervention. At this point, the system activates an audible and visual alarm and highlights the abnormal parameters on the human-machine interface, prompting operators to arrange for equipment inspection as soon as possible. These abnormal parameters include: the name of the component experiencing the abnormality, the current measured temperature rise rate, the current trend deviation value, and the corresponding warning threshold. For example, if the trend deviation of the main motor winding reaches 0.85, exceeding the warning threshold (second threshold) of 0.7, the interface will highlight information such as "Main Motor Winding," "Measured Temperature Rise Rate 1.25℃ / min," "Trend Deviation 0.85," and "Exceeds Warning Threshold 0.7," enabling operators to quickly locate the fault source and take appropriate measures.

[0070] The third level is the shutdown level: when When the temperature data of any critical component reaches the system's preset safety limit (the safety limit is set according to the technical parameters provided by the equipment manufacturer), it indicates a serious risk of equipment failure. At this time, the system will execute a forced shutdown interlock protection, immediately cutting off equipment operation to prevent equipment damage or production accidents.

[0071] The specific value can be determined as follows: During the offline calibration phase of the system, the trend deviation at each time point is calculated based on historical health data, and the 90th percentile is taken as the value. 95th percentile as 99th percentile as If no historical data is available, initial values ​​can be set based on engineering experience, for example... Take 0.3, Take 0.7, Set the value to 1.2, and adjust it based on the actual performance after the system is running.

[0072] This sub-step, through a tiered early warning mechanism, enables differentiated responses to abnormal trends of varying severity. This avoids the impact of excessive alarms on normal production while ensuring timely shutdown protection in the event of serious malfunctions, demonstrating a complete risk management chain from prompting to warning to shutdown.

[0073] This application also discloses a mill early warning system based on the temperature rise trend analysis of key components. The system includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the mill early warning method based on the temperature rise trend analysis of key components according to the above embodiments of the present invention is implemented.

[0074] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0075] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A mill early warning method based on the temperature rise trend analysis of key components, characterized in that, include: S1: Collect temperature data of key components of the mill and load data of the mill, and perform preprocessing; S2: Construct an operating status characteristic index based on the fluctuation of load data within a preset time window; S3: Set the weight coefficient, adjust the decay rate of the weight coefficient according to the running status characteristic index, and perform a weighted average of the load data within the preset compensation window based on the weight coefficient to obtain the effective load; wherein, the weight coefficient decays as the time distance between the acquisition time and the current time within the compensation window increases; S4: Obtain the effective load range and divide it into multiple load intervals. Based on the historical temperature data of each key component, calculate the upper limit of the normal temperature rise rate of the key component in each load interval and use it as the reference characteristic value of the load interval. Based on the difference between the current temperature data of the key component and the preset ambient reference temperature, construct a heat dissipation efficiency correction coefficient that decreases as the difference increases. Determine the load interval to which the effective load belongs at the current moment, and multiply the reference characteristic value of the key component in the load interval by the heat dissipation efficiency correction coefficient to obtain the expected temperature rise rate reference of the key component. S5: Calculate the trend deviation based on the measured temperature rise rate and the expected temperature rise rate of the key components at the current moment; S6: Compare the trend deviation with the preset multi-level thresholds and execute the corresponding level of warning action.

2. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, Step S2 includes: acquiring a first load data sequence within a preset time window ending at the current time; calculating the sum of the average value of the first load data sequence and a preset first zero-prevention constant, and using it as the first sum; calculating the ratio of the standard deviation of the first load data sequence to the first sum to obtain the operating state characteristic index.

3. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, The method for calculating the weighting coefficient in step S3 includes: setting a base time constant and an adjustment coefficient; calculating the product of the adjustment coefficient and the operating state characteristic index; calculating the sum of this product and the base time constant to obtain the final time constant; calculating the ratio of the time distance to the final time constant; using the negative of this ratio as the exponent; and using the natural constant... Using the base as the base, construct a weighted exponential function, and use the resulting weighted exponential function value as the weight coefficient.

4. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, The method for obtaining the reference characteristic value of the load range in step S4 includes: for each key component, based on the health status data selected from its historical temperature data, the normal temperature rise rate in each load range is statistically analyzed, and the 95th percentile is taken as the upper limit of the normal temperature rise rate of the load range, which is the reference characteristic value of the load range; wherein, the health status data refers to the temperature data of the mill operating under normal conditions with no faults, no abnormal wear, and normal cooling system operation.

5. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, The step S4 of constructing a heat dissipation efficiency correction coefficient that decreases as the difference increases includes: setting an adjustment coefficient, multiplying the difference between the current temperature data of the key component and the preset environmental reference temperature by the adjustment coefficient, using the negative of the product as the exponent, and using the natural constant... Using this as the base, a heat dissipation efficiency correction function is constructed, and the resulting heat dissipation efficiency correction function value is used as the heat dissipation efficiency correction coefficient.

6. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, Step S5 includes: for each key component, calculating the difference between the temperature data at the current time and the previous time, and calculating the ratio of the difference to the sampling interval to obtain the measured temperature rise rate; calculating the absolute difference between the measured temperature rise rate and the expected temperature rise rate benchmark; calculating the sum of the expected temperature rise rate benchmark and the preset second zero constant as the second sum; and calculating the ratio of the absolute difference to the second sum to obtain the trend deviation.

7. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, Step S6 includes: setting multiple threshold levels, including a first threshold, a second threshold, and a third threshold, wherein the first threshold is less than the second threshold and less than the third threshold; triggering a prompt-level warning when the trend deviation of a critical component is greater than or equal to the first threshold and less than the second threshold; triggering a warning-level warning when the trend deviation of a critical component is greater than or equal to the second threshold and less than the third threshold; and triggering a shutdown action when the trend deviation of a critical component is greater than or equal to the third threshold.

8. The mill early warning method based on the temperature rise trend analysis of key components according to claim 7, characterized in that, The conditions for triggering a shutdown action also include: when the temperature data of a critical component at the current moment reaches the system's preset safety limit, a shutdown action is triggered.

9. The mill early warning method based on the temperature rise trend analysis of key components according to claim 1, characterized in that, The temperature data of the key components include the main motor winding temperature, the main motor bearing temperature, the reducer oil temperature, the cylinder end bearing temperature, and the mechanical seal temperature; the load data includes the main motor current.

10. A mill early warning system based on the temperature rise trend analysis of key components, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the mill early warning method based on the analysis of temperature rise trends of key components according to any one of claims 1-9.

Images