Veneer reeling machine production task dynamic scheduling method and system based on equipment state perception

Abstract

The invention provides a veneer reeling machine production task dynamic scheduling method and system based on equipment state perception, and relates to the technical field of production scheduling, and the method comprises the steps: obtaining an equipment real-time database; performing feature extraction on the equipment real-time database to obtain an equipment real-time feature set; analyzing the equipment real-time feature set by applying a preset threshold rule, outputting a preliminary health level state, and updating the equipment real-time feature set based on the preliminary health level state; based on the updated real-time feature set of the equipment, deep evaluation of the health state of the equipment is realized, and a comprehensive state diagnosis result of the equipment is obtained; and obtaining a production task, and constructing a dynamic scheduling scheme in combination with the equipment comprehensive state diagnosis result. The technical problems that in the prior art, efficient and reliable scheduling of production tasks of the plate rolling machine is difficult to achieve, and production plan interruption and overall efficiency reduction are easily caused by sudden equipment failures are solved.

Description

Technical Field

[0001] The present application relates to the technical field of production scheduling, and in particular to a method and system for dynamically scheduling production tasks of a plate rolling machine based on equipment status perception. Background Art

[0002] Plate rolling machines, core equipment for sheet metal forming, are widely used in shipbuilding, pressure vessels, energy equipment, rail transit, and other fields to produce key structural components such as cylinders, cones, and curved plates. With the rapid development of high-end equipment manufacturing, the requirements for forming accuracy, efficiency, and equipment reliability in the plate rolling process are increasing. The stable operation of the plate rolling machine directly affects production progress and product quality.

[0003] Currently, factories typically use fixed scheduling or empirical scheduling methods to arrange plate rolling machine production tasks, but existing methods have significant problems: First, scheduling relies on static equipment status assessments and cannot perceive changes in equipment health in real time, resulting in a mismatch between task allocation and actual equipment capabilities; second, there is a lack of real-time monitoring of key plate rolling machine components, resulting in delayed fault warnings and frequent sudden downtime that disrupts production plans; in addition, scheduling optimization only considers task priority and delivery time, and does not combine dynamic adjustments with real-time equipment status, which can easily lead to the superposition of high-load tasks and accelerated equipment degradation. Among them, the most prominent problem is that existing scheduling methods do not accurately assess the health status of plate rolling machines, resulting in sudden equipment failures after scheduling. Not only can tasks not be completed on schedule, but emergency repairs can also further delay subsequent production, seriously affecting overall efficiency.

[0004] In summary, existing technologies make it difficult to achieve efficient and reliable scheduling of plate rolling machine production tasks. Sudden equipment failures can easily lead to production plan interruptions and overall efficiency decline. Therefore, a method is urgently needed to solve the above problems. Through accurate health assessment and intelligent task optimization, production efficiency can be maximized while ensuring stable equipment operation. Summary of the Invention

[0005] The present disclosure provides a method and system for dynamic scheduling of plate rolling machine production tasks based on equipment status perception, which is used to solve the technical problems in the existing technology that it is difficult to achieve efficient and reliable scheduling of plate rolling machine production tasks, and that sudden equipment failures may easily lead to production plan interruptions and overall efficiency reduction.

[0006] According to a first aspect of the present disclosure, a method for dynamically scheduling production tasks of a plate rolling machine based on equipment status perception is provided, comprising: A sensor network is deployed at key locations of the plate rolling machine to continuously collect raw physical data of the plate rolling machine's operation. The raw physical data is preprocessed to obtain a real-time equipment database, which includes equipment number, timestamp, vibration information, temperature information, and current information. The preprocessing includes data cleaning and preliminary filtering. Performing feature extraction on the real-time database of the device to obtain a real-time feature set of the device, wherein the feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction; Applying preset threshold rules to analyze the device real-time feature set, outputting a preliminary health level status, and updating the device real-time feature set based on the preliminary health level status; Based on the updated real-time feature set of the equipment, an in-depth assessment of the equipment health status is carried out to obtain the comprehensive equipment status diagnosis results; Obtain production tasks and build a dynamic scheduling plan based on the comprehensive equipment status diagnosis results.

[0007] According to a second aspect of the present disclosure, a plate rolling machine production task dynamic scheduling system based on equipment status perception is provided, comprising: A data acquisition module is used to deploy a sensor network at key locations of the plate rolling machine, continuously collect raw physical data of the plate rolling machine's operation, pre-process the raw physical data, and obtain a real-time equipment database. The real-time equipment database includes equipment number, timestamp, vibration information, temperature information, and current information. The pre-processing includes data cleaning and preliminary filtering; A multi-source data feature extraction module, which is used to extract features from the real-time database of the device to obtain a real-time feature set of the device, wherein the feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction; a device health status preliminary assessment module, configured to analyze the device real-time feature set using preset threshold rules, output a preliminary health level status, and update the device real-time feature set based on the preliminary health level status; An in-depth equipment health diagnosis module, which is used to perform an in-depth assessment of the equipment health status based on the updated real-time equipment feature set and obtain a comprehensive equipment status diagnosis result; An intelligent dynamic scheduling decision module is used to obtain production tasks and build a dynamic scheduling plan based on the equipment comprehensive status diagnosis results.

[0008] One or more technical solutions provided in the present disclosure have at least the following technical effects or advantages: deploying a sensor network at key locations of a plate rolling machine to continuously collect raw physical data from the plate rolling machine's operation, preprocessing the raw physical data to obtain a real-time equipment database, the real-time equipment database including equipment number, timestamp, vibration information, temperature information, and current information, the preprocessing including data cleaning and preliminary filtering; performing feature extraction on the real-time equipment database to obtain a real-time equipment feature set, the feature extraction including vibration information feature extraction, temperature information feature extraction, and current information feature extraction; applying preset threshold rules to analyze the real-time equipment feature set to output a preliminary health level status, and updating the real-time equipment feature set based on the preliminary health level status; performing an in-depth assessment of the equipment health status based on the updated real-time equipment feature set to obtain a comprehensive equipment status diagnosis result; obtaining production tasks, and constructing a dynamic scheduling solution based on the comprehensive equipment status diagnosis result. This solves the technical problems in the prior art of difficulty in achieving efficient and reliable scheduling of plate rolling machine production tasks, and the susceptibility to production plan interruptions and overall efficiency reduction due to sudden equipment failures. It has achieved the technical effects of accurately assessing the health status of equipment in real time, dynamically optimizing the allocation of production tasks, preventing sudden equipment failures, ensuring the continuous and stable execution of production plans, and improving overall production efficiency.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the present disclosure or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without any creative work.

[0011] Figure 1 A flow chart of a method for dynamically scheduling plate rolling machine production tasks based on equipment status perception provided in an embodiment of the present application; Figure 2 A schematic diagram of the structure of a dynamic scheduling system for plate rolling machine production tasks based on equipment status perception provided in an embodiment of the present application.

[0012] Explanation of the accompanying symbols: data acquisition module 11, multi-source data feature extraction module 12, equipment health status preliminary assessment module 13, equipment health in-depth diagnosis module 14, intelligent dynamic scheduling decision module 15. DETAILED DESCRIPTION

[0013] The following description of exemplary embodiments of the present disclosure is provided in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0014] Embodiment 1, the present disclosure provides a method for dynamically scheduling plate rolling machine production tasks based on equipment status perception, hereby referring to Figure 1 For illustration, the methods include: S1: Deploy a sensor network at key locations of the plate rolling machine to continuously collect raw physical data of the plate rolling machine's operation, pre-process the raw physical data, and obtain a real-time database of the equipment. The real-time database of the equipment includes equipment number, timestamp, vibration information, temperature information, and current information. The pre-processing includes data cleaning and preliminary filtering. Furthermore, this step S1 also includes: Integrate M groups of multi-source sensors to build a sensor network to obtain the original physical data of M plate rolling machines during operation. Each group of multi-source sensors includes a vibration sensor, a temperature sensor, and a current sensor. Perform data cleaning on the acquired raw physical data and remove obviously erroneous sampling points in the raw physical data based on rigid rules and dynamic statistical rules; Perform preliminary filtering on the cleaned raw physical data, and apply band-pass filtering and low-pass filtering to remove environmental interference; Using device numbers and timestamps as spatiotemporal tags, the pre-processed raw physical data is integrated to build a real-time device database.

[0015] Specifically, a sensor network is constructed by integrating M sets of multi-source sensors to capture vibration, temperature, and current information from M plate rolling machines during operation. Each set of multi-source sensors includes a vibration sensor, a temperature sensor, and a current sensor. The vibration sensor is a triaxial vibration accelerometer, secured to the midsection housing of the main transmission gearbox using a magnetic base, with the axis aligned with the input and output shafts. The installation torque is strictly controlled to 12 N·m to ensure a secure fit and avoid deformation of the base due to overtightening. The temperature sensor uses a pre-embedded PT100 probe embedded in the temperature measurement hole of the main motor bearing seat and the hydraulic cylinder seal, directly contacting the surface of the measured component. Thermal grease fills the gap to ensure good heat conduction. The current sensor uses a Rogowski coil current transformer installed on the main motor power input line. All sensors are connected to the edge computing gateway via industrial-grade shielded cables and synchronized using the IEEE1588 precision time protocol. This ensures that data from different sources has a unified time reference and timestamp error is controlled within 1 millisecond. The raw physical data collected directly through the sensor network is packaged and transmitted to the central server in a 1-minute cycle. Each data packet contains the device number, timestamp, vibration information, temperature information, and current information.

[0016] The acquired raw physical data is preprocessed, including data cleaning and preliminary filtering. Data cleaning requires two steps. The first step is to establish a rigid rule base based on the physical characteristics of the sensor and the safety threshold of the equipment. Specifically, the vibration sensor independently sets a ±50g range threshold for each axis. Sampling points that exceed this range are directly judged as abnormal samples and immediately eliminated. The effective range of the temperature sensor is set to 0-150°C. Data that exceeds the threshold is considered invalid and is eliminated after a secondary verification based on the equipment ambient temperature log. The current information obtained by the current sensor is set to an effective value range of 20% to 130% of the rated current, eliminating zero values ​​and extreme points caused by short circuits or open circuits. After completing the first data cleaning, for sampling points that still have abnormalities within the range, a sliding time window statistical analysis method is used to implement secondary data cleaning based on dynamic statistical rules. Using a 30-second time window, the mean μ and standard deviation σ of the effective value of the vibration information within the window are calculated in real time. If the Z-score value of the single-point vibration amplitude exceeds ±3σ three times in a row, it is judged as transient electromagnetic interference or mechanical collision and is eliminated. The change rate between adjacent temperature information is calculated. When the slope exceeds 5°C / second, the device start and stop status log is linked. If the device is in operation, it is retained; if it is in standby mode, it is eliminated. The moving average of the effective value of the current information is calculated within a 1-minute window. If the single-point deviation exceeds ±15% of the mean and persists for more than 5 cycles, it is judged as an abnormal sampling and the data is eliminated.

[0017] The cleaned raw physical data undergoes preliminary filtering to remove environmental interference. For vibration information, a bandpass filter is applied to preserve the characteristic frequencies of bearings and gears while filtering out environmental interference. A fourth-order Butterworth bandpass filter is used, with a strictly limited passband frequency of 5Hz-5kHz. The low-frequency cutoff is set to 5Hz to eliminate low-frequency interference such as foundation vibration and overall equipment shake. The high-frequency cutoff is set to 5kHz to suppress high-frequency electromagnetic noise such as motor brush noise and inverter switching harmonics. Forward and backward bidirectional filtering is used to eliminate signal phase offset and ensure accurate time location of impact events. Mirror extension is applied to the beginning and end of the data segment to avoid endpoint distortion during the filtering process. For temperature and current information, a low-pass filter is applied to preserve the thermal inertia and load fluctuation characteristics of the equipment while filtering out high-frequency noise. A Chebyshev I-type low-pass filter is used with a cutoff frequency of 1Hz and a passband ripple controlled within 0.5dB. The temperature signal needs to filter out sensor circuit thermal noise and electromagnetic coupling interference while preserving the slowly changing trend of the oil temperature. The current signal needs to eliminate inverter carrier interference while preserving the characteristics of motor load fluctuations. Apply moving average to the filtered data to further smooth random fluctuations and improve trend recognition.

[0018] The cleaned and filtered raw physical data is reorganized using the device number and timestamp as time and space tags to form a real-time database of the device.

[0019] S2: performing feature extraction on the real-time database of the device to obtain a real-time feature set of the device, wherein the feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction; Furthermore, this step S2 also includes: Obtaining vibration information, temperature information, and current information from a real-time database of the device; Vibration information features are extracted through multi-dimensional analysis methods in the time domain and frequency domain. The extracted feature information includes vibration effective value, vibration peak value, vibration kurtosis and high-frequency energy proportion; The trend analysis method is used to extract temperature information features, and the extracted feature information includes absolute temperature value and temperature rise rate; Extract current information features based on the electrical and mechanical coupling mechanism. The extracted feature information includes the effective value of current and the three-phase current imbalance. All feature information is integrated into a structured real-time feature set of the device according to the device number.

[0020] Specifically, the real-time database of the equipment is obtained, and features of the three types of information in the database, namely vibration, temperature, and current, are extracted. The raw data is converted into indicators that reflect the operating status of key components of the equipment, providing a basis for subsequent health status assessment.

[0021] The main purpose of vibration information feature extraction is to identify mechanical faults such as imbalance, wear, looseness, asymmetry, and impact damage in bearings, gears, and shaft systems. The specific operation process includes: Time domain feature extraction. Directly analyze the vibration waveform to obtain the vibration effective value, vibration peak value and vibration kurtosis of the vibration information. The formulas are: Vibration effective value: ; Among them, RMS represents the effective value of the vibration signal, which is used to characterize the average intensity of the vibration energy, N represents the number of sampling points, and y i Represents the amplitude of the i-th vibration signal sampling point, where i is the index value. The RMS value reflects the overall level of vibration energy and is the most basic and stable indicator for assessing overall equipment degradation. A continuously increasing RMS value usually indicates a developing fault.

[0022] Vibration peak: ; Among them, Peak represents the peak value of the vibration signal, which is used to capture the maximum intensity of the instantaneous impact, max is the function of the selected maximum value, |y i | Represents the absolute value of the signal at the sampling point. The vibration peak can capture transient impact intensity. It is very sensitive to localized damage. A single elevated vibration peak may indicate an impact event.

[0023] Vibration Kurtosis: ; Among them, Kurtosis is the kurtosis of the vibration signal, which is used to quantify the sharpness of the impact signal, N is the total number of sampling points, and y i represents the amplitude of the i-th vibration signal sampling point, where i is the index, η represents the signal mean, and σ represents the signal standard deviation. Vibration kurtosis is extremely sensitive to shock signals and serves as an early warning indicator of faults. In the early stages of a fault, shock signals are mixed with background vibration, and while the RMS value may not change much, the vibration kurtosis value will increase significantly. As the fault progresses and shocks increase, the vibration kurtosis value may actually decrease.

[0024] Frequency domain feature extraction. Vibration information is converted from the time domain to the frequency domain through Fast Fourier Transform (FFT), a spectrum is obtained, and frequency domain feature extraction is performed to obtain the high-frequency energy ratio of the vibration information. The formula is: High frequency band energy ratio: ; Among them, HFR represents the proportion of high-frequency band energy, which is used to represent the proportion of high-frequency friction or impact noise. The numerator in the formula represents the high-frequency energy, the numerator represents the full-band energy, k represents the frequency band number, k max Indicates the maximum sequence number, k min represents the minimum sequence number, f represents the frequency band boundary, X(f k) represents the signal amplitude, and M is the window length of FFT analysis.

[0025] Specifically, to obtain the energy ratio of the high frequency band, it is necessary to first obtain the frequency resolution Δf through the FFT spectrum. The formula is: , where f s Represents the sampling rate, M is the window length of FFT analysis. Based on the obtained frequency resolution Δf, combined with the determined frequency band boundary f min With f max Calculate the frequency band number k min With k max , the formula is , where f is the frequency band boundary. For example, assuming the sampling rate is 20 kHz and the FFT analysis window length M is 2048, then Δf is approximately 9.77 Hz. Based on the commonly used engineering values, f is determined as min is 4000Hz and f max is 20000Hz, and k is calculated min 410,k max is 1024, so the calculation formula for HFR becomes The resulting high-frequency energy ratio can be used to monitor the degree of wear degradation. As component wear increases, the high-frequency friction and impact noise energy tends to increase, causing this ratio to rise.

[0026] The main purpose of temperature information feature extraction is to identify problems such as bearing overheating, motor overheating, lubrication failure, poor cooling, and uneven load. The main features extracted include absolute temperature value and temperature rise rate, and the formulas are: Absolute temperature value: ; Among them, T current represents the absolute temperature value, P represents the moving average window size, T j Represents the original sampling value of the sensor, j is the index value, C cal Represents the correction coefficient, which is set to 0.385 here. offset Represents environmental compensation. Absolute temperature is the most direct basis for overheating alarms. Different levels of alarm thresholds can be set to determine the operating status of the plate rolling machine.

[0027] Temperature rise rate: ; Among them, R rise represents the temperature rise rate, T t represents the current temperature, Δt is the time window, T t-Δt represents historical temperature values. The temperature rise rate can detect anomalies earlier. During normal equipment startup or load increase, the temperature rise is gradual. A sudden increase in the temperature rise rate often indicates an impending serious problem, such as lubrication failure, cooling system malfunction, or severe friction.

[0028] The main purpose of current information feature extraction is to identify motor electrical faults, power quality issues, and mechanical load anomalies reflected by motor current. The main features extracted include the current RMS value and the three-phase current imbalance, and the formulas are: Current RMS: ; Among them, I RMS RMS current represents the effective value of current and is a key indicator of load severity. N represents the number of sampling points in a cycle, Ir represents the instantaneous current sampling value, and r is the index value. The instantaneous current sampling value can reflect the magnitude of the motor load. If the RMS current value continuously exceeds the rated value, it indicates motor overload. Combined with vibration and temperature, the cause of the overload can be determined.

[0029] Three-phase current imbalance: ; Among them, I A , I B , I C Represents the effective value of the three-phase current, which is calculated according to the effective value formula of the current. I avg = represents the three-phase average current, and Unbalance represents the degree of imbalance. The three-phase current imbalance can be used to diagnose electrical faults such as unbalanced power supply voltage, asymmetric motor windings, and broken rotor bars.

[0030] All the feature information extracted from the above features is integrated into an information table. Index values ​​are assigned based on the plate rolling machine number, and timestamps are added based on the original data. This ultimately yields a real-time feature set for the equipment. This feature set includes vibration RMS values, peak values, kurtosis, and high-frequency energy percentage data for vibration information; absolute temperature values ​​and temperature rise rates for temperature information; and current RMS values ​​and three-phase current imbalance data for current information, arranged in order. The data is then categorized using the equipment number and timestamp as spatiotemporal labels, storing all feature information for each of the M plate rolling machines every minute.

[0031] S3: Analyze the device real-time feature set by applying a preset threshold rule, output a preliminary health level status, and update the device real-time feature set based on the preliminary health level status; Furthermore, this step S3 also includes: Setting threshold limits for each feature in the real-time feature set of the device, including normal limits and fault limits; The comprehensive health index is calculated based on the device's real-time feature set and threshold limits. The specific formula is: ; Among them, HI(t) represents the comprehensive health index at the current moment, i is the index value, U i (t) represents the current actual value of the i-th feature, L i,normal represents the normal limit of the i-th feature, L i,fault represents the fault limit of the i-th feature, ω i represents the weight of the i-th feature, and α represents the nonlinear adjustment coefficient; The device's current preliminary health status is determined based on the comprehensive health index. When HI(t) ≥ 0.85, the device is considered to be at level 1 health. When 0.6 ≤ HI(t) < 0.85, the device is considered to be at level 2 warning. When HI(t) < 0.6, the device is considered to be at level 3 fault. The equipment whose preliminary health level status is determined to be level 3 fault level will be shut down, and the real-time feature set of the equipment will be updated at the same time, and all data under the shut down equipment number will be deleted.

[0032] Specifically, the preset threshold rules are applied to analyze the feature data in the real-time feature set of the equipment to determine the health status of the plate rolling machine in terms of vibration, temperature and current.

[0033] In terms of vibration characteristic information, the normal limit of vibration RMS is set to 2.5g, the fault limit is 4.0g, the normal limit of vibration peak Peak is set to 5.0g, the fault limit is 8.0g, the normal limit of vibration Kurtosis is set to 3.5, the fault limit is 4.5, the normal limit of high-frequency energy ratio HFR is set to 8%, and the fault limit is 15%.

[0034] In terms of temperature information, set the absolute temperature value T current The normal limit is 70°C, the fault limit is 85°C, and the temperature rise rate R is set. rise The normal limit is 0.2°C / min and the fault limit is 0.5°C / min.

[0035] Regarding current information, set the effective current value I RMS The normal limit is 1.1I, where I is the rated current, and the fault limit is 1.2I. The normal limit of the three-phase current imbalance is set to 5%, and the fault limit is 10%.

[0036] The comprehensive health index of each plate rolling machine is calculated based on the thresholds set by the above parameters. The specific formula is: ; Among them, HI(t) represents the comprehensive health index at the current moment, i is the index value, U i(t) represents the current actual value of the i-th feature, L i,normal represents the normal limit of the i-th feature, L i,fault represents the fault limit of the i-th feature, ω i Represents the weight of the i-th feature, satisfying all ω i The sum of is 1, and α represents the nonlinear adjustment coefficient, which is used to control the sensitivity after exceeding the threshold, and is usually set to 1.5.

[0037] The specific weight values ​​in the formula are ω1=0.15, ω2=0.10, ω3=0.20, ω4=0.10, ω5=0.15, ω6=0.15, ω7=0.10, and ω8=0.05. ω1 represents the effective vibration value weight, reflecting the overall mechanical load; ω2 represents the peak vibration weight, reflecting sensitivity to transient shocks; ω3 represents vibration kurtosis, a core indicator of early failure; ω4 represents the high-frequency energy proportion, indicating the long-term trend of wear and degradation; ω5 represents the absolute temperature value, directly related to equipment safety; ω6 represents the temperature rise rate, which is key to predicting sudden failures; ω7 represents the effective current value, reflecting the electrical load; and ω8 represents the three-phase current imbalance, reflecting sensitivity to motor symmetry.

[0038] Finally, the current preliminary health level status of the plate rolling machine is judged according to the calculated comprehensive health index. When HI(t)≥0.85, the preliminary health level status is judged to be the first level health level. When 0.6≦HI(t)<0.85, the preliminary health level status is judged to be the second level warning level. When HI(t)<0.6, the preliminary health level status is judged to be the third level fault level.

[0039] When the preliminary health level is judged to be a level three fault level, the plate rolling machine will be stopped directly and a maintenance alarm will be activated. The plate rolling machine will no longer enter the subsequent scheduling process, and all data under the stopped equipment number in the equipment real-time feature set will be deleted.

[0040] S4: Based on the updated real-time feature set of the device, an in-depth assessment of the device health status is performed to obtain a comprehensive diagnosis result of the device status; Furthermore, this step S4 also includes: Read the updated device real-time feature set, calculate the moving average and trend slope of each feature information, and construct a feature enhancement vector set; Dynamically adjust the threshold limit of each feature according to the moving mean, and dynamically adjust the weight distribution of each feature according to the trend slope; The membership functions of the three states are set based on the adjusted threshold limits, and the membership degree of the current feature is calculated. The membership functions include the trapezoidal function of the healthy state, the triangular function of the warning state, and the Z-type function of the fault state. The specific expressions are: healthy: ; Warning: ; Fault: ; Among them, μ normal (x) represents the membership of feature x to the health state, μ warning (x) represents the membership of feature x to the warning state, μ fault (x) represents the membership of feature x to the fault state, x represents the original value of the feature, L normal Represents the normal limit, L warning Represents the warning limit, L mid Represents the attenuation limit, L fault represents the fault limit, where L warning =L normal +0.3(L fault- L normal ), L mid =(L normal +L fault ) / 2; The final confidence level of the device in the three states is calculated using the weighted average method. The specific calculation formula is: ; Among them, Belief(A) represents the confidence of the device as a whole in state A, ω i Represents the feature weight of the i-th feature after dynamic adjustment according to the trend slope, μ i (A) represents the membership of the i-th feature; Based on the confidence level of the final equipment in the three areas of health, warning, and failure, the comprehensive equipment status diagnosis result of the plate rolling machine at this time is determined.

[0041] Specifically, read the updated device real-time feature set and obtain all feature data of each device in the previous 60 minutes based on the current time. For each type of feature information, calculate its moving average and trend slope, and finally obtain the feature enhancement vector. The formulas are: Moving Average: ; Among them, Avg i Represents the moving mean of the i-th feature, representing the mean of the feature information in the past 60 minutes at the current time t, i is the index value, t is the time, m represents the time sequence number, u i (m) represents the original value of the i-th feature at time m.

[0042] Trend slope: ; Among them, Slope i Represents the trend slope of the i-th feature, i is the index value, tm Represents the timestamp, u i (m) represents the original value of the i-th feature at time m, and the trend slope can determine whether the feature continues to deteriorate.

[0043] The feature enhancement vector of each feature is constructed based on the original value, moving mean, and trend slope of the feature. Taking the effective value RMS of vibration as an example, if the RMS value at the current moment is 3.2g, the calculated moving mean is 2.8g, and the trend slope is 0.01g / min, then the final feature enhancement vector generated for the effective value of vibration is [original RMS=3.2, mean=2.8, slope=0.01]. Therefore, each plate rolling machine equipment will eventually obtain 8 feature enhancement vectors at the current time, corresponding to the eight original feature data, and then construct a feature enhancement vector set.

[0044] The threshold limits of each feature are dynamically adjusted based on the moving mean value contained in the feature enhancement vector set, and the weight distribution of each feature data is dynamically adjusted based on the trend slope. Based on the adjusted threshold limits and weight distribution, fuzzy membership calculations and device confidence calculations are performed on the feature data to determine the comprehensive device status diagnosis results.

[0045] For the characteristic information, membership functions of three states are set, including the trapezoidal function of the healthy state, the triangular function of the warning state, and the Z-type function of the fault state. The threshold range of each function is defined according to the threshold limit after dynamic adjustment. For example, taking the effective value of vibration as an example, if the moving mean is calculated to be 2.8g, the threshold limit is dynamically adjusted at this time, the new normal limit is 2.8g, and the new fault limit is 4.0+(2.8-2.5)=4.3g. Based on the two threshold limits, the warning limit and the attenuation limit are further set, and finally the specific thresholds of the three types of functions are delineated. The membership degree of the current feature is calculated according to the set membership function. The specific expression is: healthy: ; Warning: ; Fault: ; Among them, μ normal (x) represents the membership of feature x to the health state, μ warning (x) represents the membership of feature x to the warning state, μ fault (x) represents the membership of feature x to the fault state, x represents the original value of the feature, L normal Represents the normal limit, L warning Represents the warning limit, L mid Represents the attenuation limit, L fault represents the fault limit, where L warning =L normal +0.3(Lfault- L normal ), L mid =(L normal +L fault ) / 2.

[0046] The fuzzy membership of multiple features is converted into a comprehensive device status diagnosis result. First, the membership is normalized to facilitate subsequent calculations. Second, the final confidence level of the device in the three states is calculated using the weighted average method. Feature weights are introduced into the calculation process. The feature weights at this time are no longer the weights used in the previous comprehensive health index calculation, but are updated based on the trend slope calculation results. The specific calculation formula is: ; Among them, Belief(A) represents the confidence of the device as a whole in state A, ω i Represents the feature weight of the i-th feature, μ i (A) represents the membership of the i-th feature.

[0047] Based on the confidence level of the final equipment in the three areas of health, warning, and failure, the comprehensive equipment status diagnosis result of the plate rolling machine at this time is determined.

[0048] S5: Obtain production tasks and build a dynamic scheduling plan based on the equipment comprehensive status diagnosis results.

[0049] Furthermore, this step S5 also includes: Obtain a list of production tasks to be scheduled, the list including task numbers, machining accuracy requirements, estimated time, delivery date, and priority attributes; Based on the comprehensive status diagnosis results of the equipment, preliminary task filtering is performed to eliminate tasks that cannot be completed by any equipment at the current stage, retain other tasks, and generate a candidate task set; Obtain a set of candidate tasks and calculate the comprehensive matching score between the tasks that pass the initial screening and the device. The specific formula is as follows: ; Among them, Score represents the comprehensive matching degree between production tasks and equipment, S Precision Quantify the degree of fit between task requirements and equipment capabilities, the specific value is , where TaskAccuracy represents task accuracy, EquipmentPrecision represents equipment accuracy; S Health Reflects the intensity of tasks that the device can safely carry in its current state. The specific value is , where E is the health status; S DeliveryTime Indicates that tasks with a deadline are prioritized. The specific value is , where CurrentTime represents the current time and DeliveryDeadline represents the delivery deadline; S Cost It represents the consideration of economic constraints such as energy and mold change, with a value of 0.8 during peak processing and 1.0 during off-peak processing; A genetic algorithm is used to optimize task allocation and time scheduling. Chromosome encoding is performed based on the candidate task set and equipment information to determine the gene structure. Each gene represents a task allocation unit, including equipment number, task number, start time, and load percentage. Initialize the population, randomly generate 100 scheduling plans, and introduce the previously calculated comprehensive matching score; After the comprehensive matching degree is introduced, all schemes are subjected to selection, crossover, and mutation operations. The selection operation retains the top 30% of individuals with the best fitness and directly advances to the next generation. The remaining 70% is selected through roulette. The crossover operation randomly selects two parent individuals and exchanges some gene sequences. The mutation operation randomly adjusts the device number, start time, or load of a gene. A new population is generated through selection, crossover, and mutation operations, and iterative calculations are performed to gradually improve until convergence or 50 generations of iterations are reached, and finally a dynamic scheduling solution is output.

[0050] Specifically, a list of production tasks to be scheduled is obtained based on the factory's production order. This list includes the task number, machining accuracy requirements, estimated time, delivery date, and priority. All production tasks are verified for completeness and timeliness, and expired or unavailable tasks are removed to ensure the reliability of the input data.

[0051] Based on the comprehensive equipment status diagnosis results, a preliminary task filtering process is performed to identify tasks that the equipment can currently perform and eliminate tasks that no equipment can currently complete. Specifically, healthy equipment can undertake all tasks, but load monitoring is required. Early warning equipment must limit task types and loads, prohibiting tasks that require processing accuracy exceeding the equipment's current capabilities. Faulty equipment is only allowed to perform low-priority tasks or undergo maintenance. This stage is implemented using a rules engine. Finally, a candidate task set is generated based on the initial screening results.

[0052] Obtain a set of candidate tasks and calculate the comprehensive matching score between the tasks that pass the initial screening and the device. The specific formula is as follows: ; Among them, Score represents the comprehensive matching degree between production tasks and equipment, S Precision Quantify the degree of fit between task requirements and equipment capabilities, the specific value is , where TaskAccuracy represents task accuracy, EquipmentPrecision represents equipment accuracy; SHealth Reflects the intensity of tasks that the device can safely carry in its current state. The specific value is , where E is the health status; S DeliveryTime Indicates that tasks with a deadline are prioritized. The specific value is , where CurrentTime represents the current time and DeliveryDeadline represents the delivery deadline; S Cost It represents the consideration of economic constraints such as energy and mold change. The value is 0.8 during peak processing hours and 1.0 during non-peak processing hours.

[0053] A genetic algorithm optimizes task allocation and scheduling. The specific steps are as follows: chromosomes are encoded based on production task and equipment information to determine the gene structure. Each gene represents a task allocation unit, including the equipment number, task number, start time, and load percentage. A population is initialized, and 100 scheduling scenarios are randomly generated. Each scenario must ensure that the equipment load does not exceed the upper limit allowed by the health status, that task times do not overlap, and that materials are readily available. A previously calculated comprehensive matching score is applied to the equipment and task numbers involved in each scenario. Selection, crossover, and mutation operations are performed on all scenarios after the comprehensive matching score is applied. The selection process retains the top 30% of individuals with the highest fitness and advances directly to the next generation, while the remaining 70% are selected through a roulette wheel. The crossover process randomly selects two parent individuals and exchanges some of their gene sequences. The mutation process randomly adjusts the equipment number, start time, or load of a gene. A new population is generated through selection, crossover, and mutation, and iterative calculations are performed until convergence or 50 generations have been reached. Finally, a Pareto-optimal solution set is generated, consisting of multiple non-dominated scheduling scenarios for decision-making. Staff select the scenario that best suits production conditions.

[0054] In the second embodiment, based on the same invention concept as the method for dynamically scheduling plate rolling machine production tasks based on equipment status perception in the previous embodiment, this application also provides a dynamic scheduling system for plate rolling machine production tasks based on equipment status perception, please refer to the attached Figure 2 , the system comprising: Data acquisition module 11, which is used to deploy a sensor network at key locations of the plate rolling machine, continuously collect raw physical data of the plate rolling machine operation, pre-process the raw physical data, and obtain a real-time equipment database. The real-time equipment database includes equipment number, timestamp, vibration information, temperature information, and current information. The pre-processing includes data cleaning and preliminary filtering; The multi-source data feature extraction module 12 is used to extract features from the device real-time database to obtain a device real-time feature set. The feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction. The acquisition of the device real-time feature set also includes: Obtaining vibration information, temperature information, and current information from a real-time database of the device; Vibration information features are extracted through multi-dimensional analysis methods in the time domain and frequency domain. The extracted feature information includes vibration effective value, vibration peak value, vibration kurtosis and high-frequency energy proportion; The trend analysis method is used to extract temperature information features, and the extracted feature information includes absolute temperature value and temperature rise rate; Extract current information features based on the electrical and mechanical coupling mechanism. The extracted feature information includes the effective value of current and the three-phase current imbalance. Integrate all feature information into a structured real-time feature set of devices according to device numbers; a device health status preliminary assessment module 13, configured to analyze the device real-time feature set using a preset threshold rule, output a preliminary health level status, and update the device real-time feature set based on the preliminary health level status; The device health in-depth diagnosis module 14 is used to perform an in-depth assessment of the device health status based on the updated device real-time feature set to obtain a comprehensive device status diagnosis result; The intelligent dynamic scheduling decision module 15 is used to obtain production tasks and build a dynamic scheduling plan in combination with the equipment comprehensive status diagnosis result.

[0055] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0056] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A dynamic scheduling method for plate rolling machine production tasks based on equipment status perception is characterized by: The method comprises: A sensor network is deployed at key locations of the plate rolling machine to continuously collect raw physical data of the plate rolling machine's operation. The raw physical data is preprocessed to obtain a real-time equipment database, which includes equipment number, timestamp, vibration information, temperature information, and current information. The preprocessing includes data cleaning and preliminary filtering. Performing feature extraction on the real-time database of the device to obtain a real-time feature set of the device, wherein the feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction. Obtaining the real-time feature set of the device further includes: Obtaining vibration information, temperature information, and current information from a real-time database of the device; Vibration information features are extracted through multi-dimensional analysis methods in the time domain and frequency domain. The extracted feature information includes vibration effective value, vibration peak value, vibration kurtosis and high-frequency energy proportion; The trend analysis method is used to extract temperature information features, and the extracted feature information includes absolute temperature value and temperature rise rate; Extract current information features based on the electrical and mechanical coupling mechanism. The extracted feature information includes the effective value of current and the three-phase current imbalance. Integrate all feature information into a structured real-time feature set of devices according to device numbers; Applying preset threshold rules to analyze the device real-time feature set, outputting a preliminary health level status, and updating the device real-time feature set based on the preliminary health level status; Based on the updated real-time feature set of the equipment, an in-depth assessment of the equipment health status is carried out to obtain the comprehensive equipment status diagnosis results; Obtain production tasks and build a dynamic scheduling plan based on the comprehensive equipment status diagnosis results.

2. The method for dynamic scheduling of plate rolling machine production tasks based on equipment status perception according to claim 1 is characterized in that: Get the real-time database of the device, including: Integrate M groups of multi-source sensors to build a sensor network to obtain the original physical data of M plate rolling machines during operation. Each group of multi-source sensors includes a vibration sensor, a temperature sensor, and a current sensor. Perform data cleaning on the acquired raw physical data and remove obviously erroneous sampling points in the raw physical data based on rigid rules and dynamic statistical rules; Perform preliminary filtering on the cleaned raw physical data, and apply band-pass filtering and low-pass filtering to remove environmental interference; Using device numbers and timestamps as spatiotemporal tags, the pre-processed raw physical data is integrated to build a real-time device database.

3. The method for dynamically scheduling plate rolling machine production tasks based on equipment status perception according to claim 1 is characterized in that: Outputting a preliminary health level status, and updating the device real-time feature set based on the preliminary health level status, including: Setting threshold limits for each feature in the real-time feature set of the device, including normal limits and fault limits; The comprehensive health index is calculated based on the device's real-time feature set and threshold limits. The specific formula is: ； Among them, HI(t) represents the comprehensive health index at the current moment, i is the index value, U i (t) represents the current actual value of the i-th feature, L i,normal represents the normal limit of the i-th feature, L i,fault represents the fault limit of the i-th feature, ω i represents the weight of the i-th feature, and α represents the nonlinear adjustment coefficient; The device's current preliminary health status is determined based on the comprehensive health index. When HI(t) ≥ 0.85, the device is considered to be at level 1 health. When 0.6 ≤ HI(t) < 0.85, the device is considered to be at level 2 warning. When HI(t) < 0.6, the device is considered to be at level 3 fault. The equipment whose preliminary health level status is determined to be level 3 fault level will be shut down, and the real-time feature set of the equipment will be updated at the same time, and all data under the shut down equipment number will be deleted.

4. The method for dynamically scheduling plate rolling machine production tasks based on equipment status perception according to claim 1, characterized in that: Obtain comprehensive equipment status diagnostic results, including: Read the updated device real-time feature set, calculate the moving average and trend slope of each feature information, and construct a feature enhancement vector set; Dynamically adjust the threshold limit of each feature according to the moving mean, and dynamically adjust the weight distribution of each feature according to the trend slope; The membership functions of the three states are set based on the adjusted threshold limits, and the membership degree of the current feature is calculated. The membership functions include the trapezoidal function of the healthy state, the triangular function of the warning state, and the Z-type function of the fault state. The specific expressions are: healthy: ; Warning: ; Fault: ; Among them, μ normal (x) represents the membership of feature x to the health state, μ warning (x) represents the membership of feature x to the warning state, μ fault (x) represents the membership of feature x to the fault state, x represents the original value of the feature, L normal Represents the normal limit, L warning Represents the warning limit, L mid Represents the attenuation limit, L fault represents the fault limit, where L warning =L normal +0.3(L fault- L normal ), L mid =(L normal +L fault ) / 2; The final confidence level of the device in the three states is calculated using the weighted average method. The specific calculation formula is: ； Among them, Belief(A) represents the confidence of the device as a whole in state A, ω i Represents the feature weight of the i-th feature after dynamic adjustment according to the trend slope, μ i (A) represents the membership of the i-th feature; Based on the confidence level of the final equipment in the three areas of health, warning, and failure, the comprehensive equipment status diagnosis result of the plate rolling machine at this time is determined.

5. The method for dynamically scheduling plate rolling machine production tasks based on equipment status perception according to claim 1, characterized in that: Obtain production tasks and build a dynamic scheduling plan based on the equipment comprehensive status diagnosis results, including: Obtain a list of production tasks to be scheduled, the list including task numbers, machining accuracy requirements, estimated time, delivery date, and priority attributes; Based on the comprehensive status diagnosis results of the equipment, preliminary task filtering is performed to eliminate tasks that cannot be completed by any equipment at the current stage, retain other tasks, and generate a candidate task set; Obtain a set of candidate tasks and calculate the comprehensive matching score between the tasks that pass the initial screening and the device. The specific formula is as follows: ； Among them, Score represents the comprehensive matching degree between production tasks and equipment, S Precision Quantify the degree of fit between task requirements and equipment capabilities, the specific value is , where TaskAccuracy represents task accuracy, EquipmentPrecision represents equipment accuracy; S Health Reflects the intensity of tasks that the device can safely carry in its current state. The specific value is , where E is the health status; S DeliveryTime Indicates that tasks with a deadline are prioritized. The specific value is , where CurrentTime represents the current time and DeliveryDeadline represents the delivery deadline; S Cost It represents the consideration of economic constraints such as energy and mold change, with a value of 0.8 during peak processing and 1.0 during off-peak processing; A genetic algorithm is used to optimize task allocation and time scheduling. Chromosome encoding is performed based on the candidate task set and equipment information to determine the gene structure. Each gene represents a task allocation unit, including equipment number, task number, start time, and load percentage. Initialize the population, randomly generate 100 scheduling plans, and introduce the previously calculated comprehensive matching score; After the comprehensive matching degree is introduced, all schemes are subjected to selection, crossover, and mutation operations. The selection operation retains the top 30% of individuals with the best fitness and directly advances to the next generation. The remaining 70% is selected through roulette. The crossover operation randomly selects two parent individuals and exchanges some gene sequences. The mutation operation randomly adjusts the device number, start time, or load of a gene. A new population is generated through selection, crossover, and mutation operations, and iterative calculations are performed to gradually improve until convergence or 50 generations of iterations are reached, and finally a dynamic scheduling solution is output.

6. The dynamic scheduling system for plate rolling machine production tasks based on equipment status perception is characterized by: The system is used to implement the method for dynamically scheduling plate rolling machine production tasks based on equipment status perception as described in any one of claims 1 to 5, and the system includes: A data acquisition module is used to deploy a sensor network at key locations of the plate rolling machine, continuously collect raw physical data of the plate rolling machine's operation, pre-process the raw physical data, and obtain a real-time equipment database. The real-time equipment database includes equipment number, timestamp, vibration information, temperature information, and current information. The pre-processing includes data cleaning and preliminary filtering; A multi-source data feature extraction module is used to extract features from the device real-time database to obtain a device real-time feature set. The feature extraction includes vibration information feature extraction, temperature information feature extraction, and current information feature extraction. Obtaining the device real-time feature set also includes: Obtaining vibration information, temperature information, and current information from a real-time database of the device; Vibration information features are extracted through multi-dimensional analysis methods in the time domain and frequency domain. The extracted feature information includes vibration effective value, vibration peak value, vibration kurtosis and high-frequency energy proportion; The trend analysis method is used to extract temperature information features, and the extracted feature information includes absolute temperature value and temperature rise rate; Extract current information features based on the electrical and mechanical coupling mechanism. The extracted feature information includes the effective value of current and the three-phase current imbalance. Integrate all feature information into a structured real-time feature set of devices according to device numbers; a device health status preliminary assessment module, configured to analyze the device real-time feature set using preset threshold rules, output a preliminary health level status, and update the device real-time feature set based on the preliminary health level status; An in-depth equipment health diagnosis module, which is used to perform an in-depth assessment of the equipment health status based on the updated real-time equipment feature set and obtain a comprehensive equipment status diagnosis result; An intelligent dynamic scheduling decision module is used to obtain production tasks and build a dynamic scheduling plan based on the equipment comprehensive status diagnosis results.

Images