Mill load analysis system based on multi-parameter model

Abstract

The invention relates to the technical field of mill load detection, in particular to a mill load analysis system based on a multi-parameter model, which realizes comprehensive and accurate mill load analysis through cooperative work of multiple modules. The data acquisition module adopts various sensors to comprehensively collect vibration and abrasive disc rotating speed data, and provides rich and accurate basic data for subsequent analysis; vibration data are collected by an acceleration sensor and a speed sensor; abrasive disc rotating speed data are obtained in real time through a rotating speed sensor at a specific position and are stored in combination with a historical memory module; the data analysis module performs time domain and frequency domain analysis on the vibration data, establishes a relation model with a load, analyzes the rotating speed of an abrasive disc by using a time sequence, and dynamically updates a threshold value to judge a load state; the multi-parameter metadata fusion and analysis mode can more comprehensively and accurately reflect the actual load condition of the mill, is helpful for timely finding potential fault hidden dangers, avoids equipment damage caused by abnormal load of the mill, and guarantees production continuity and stability.

Description

Technical Field

[0001] The present invention relates to the technical field of mill load detection, and particularly to a mill load analysis system based on a multi-parameter model. Background Art

[0002] In modern industrial production, as a key equipment for material grinding, mills are widely used in many fields such as mines, cement, and chemical industries. The stable operation of mills is crucial for production efficiency, product quality, and the economic benefits of enterprises. However, during the operation of mills, their load states are affected by various factors, such as material characteristics, wear degree of grinding discs, and feed rate, which makes it a challenging task to accurately monitor and analyze the mill load.

[0003] Traditional mill load monitoring methods often rely on a single parameter for analysis. For example, they only focus on the current or vibration amplitude of the mill. This single-parameter analysis method has great limitations and cannot comprehensively reflect the complex operating conditions of the mill. Judging the load only based on the current may ignore the influence of factors such as grinding disc wear and uneven material distribution on the operation of the mill. And simply relying on the vibration amplitude is difficult to accurately identify the subtle changes inside the mill caused by abnormal loads. Therefore, the analysis method based on a single parameter is prone to misjudging the mill load state and cannot detect potential fault hazards in time.

[0004] Therefore, the present invention proposes a mill load analysis system based on a multi-parameter model. Summary of the Invention

[0005] Technical Problems to be Solved

[0006] Aiming at the deficiencies of the prior art, the present invention provides a mill load analysis system based on a multi-parameter model, thereby solving the technical problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0008] A mill load analysis system based on a multi-parameter model includes the following modules:

[0009] S1. Vibration data: Install an acceleration sensor and a velocity sensor. The acceleration sensor can be piezoelectric or piezoresistive, which is used to obtain vibration acceleration information. The velocity sensor is based on the principle of electromagnetic induction to measure the vibration velocity. Then, the signal conditioning module amplifies and filters the weak electrical signals output by the sensors to remove noise interference. The data acquisition card converts the analog signals into digital signals and collects them at a set sampling frequency (at least twice the highest vibration frequency component of the mill). Disc speed data: Use a speed sensor installed near the rotating shaft of the mill disc to monitor the disc speed in real time. By detecting the pulse signal frequency emitted by the sensor, calculate the real-time speed of the disc, record the real-time data of the disc speed, and combine it with the historical memory module to store the speed data at different time nodes, providing data support for subsequent threshold update and load analysis.

[0010] S2. Data transmission module: Use shielded cables to transmit the data collected by each sensor to the data processing center.

[0011] S3. Data analysis module: Vibration analysis model: Conduct time-domain and frequency-domain analyses on the vibration data. In the time domain, assume that the vibration data is given in the form of a discrete sequence x[n], where n = 0, 1, …, N - 1, and N is the total number of data points. Then calculate the amplitude, mean value, and peak statistical parameters of the vibration. In the frequency domain, obtain the spectral characteristics of the vibration through fast Fourier transform, identify the amplitude changes of different frequency components, establish a relationship model between vibration parameters and load according to the normal operating conditions of the mill, determine the change range of the vibration amplitude in different load intervals, and the corresponding relationship between specific frequency components and abnormal load conditions, so as to predict the change trend of the speed. Dynamically update the speed threshold according to the usage time and wear degree of the disc. When the disc speed exceeds or is lower than the threshold range, combine the vibration amplitude parameters to judge the load state of the mill. When the disc speed suddenly drops and the vibration amplitude increases simultaneously, it indicates that the mill load has increased.

[0012] S4. Alarm module: According to the analysis and diagnosis results, when the operating parameters of the mill exceed the normal range or potential faults are detected, transmit the data to the display interface of the device, and prompt the staff that the mill is overloaded.

[0013] S5. Historical memory and update module: Long-term store all the collected operation data and analysis results, establish a historical database. As the operation time of the mill increases and the working conditions change, regularly update and optimize the multi-parameter model using new historical data.

[0014] In a possible implementation, in the data analysis module, amplitude: The peak-to-peak value is the difference between the maximum and minimum values of the vibration signal in one cycle, reflecting the fluctuation range of the signal. Its formula is:

[0015] A P-P = max(x[n]) - min(x[n])

[0016] Mean: The mean represents the average level of the vibration signal, which is obtained by summing all data points and dividing by the total number of data points. The formula is:

[0017]

[0018] Peak value: The peak value is the maximum value in the vibration signal. The formula is:

[0019] A peak = max(x[n]).

[0020] In a possible implementation, in the data analysis module, for a discrete-time sequence x[n] of length N, where n = 0, 1,..., N - 1, its discrete Fourier transform is defined as:

[0021]

[0022] where k = 0, 1,..., N - 1,

[0023] Obtaining spectral characteristics and amplitude: Assume that the result obtained by the fast Fourier transform is X[k], which is a complex number array. For each k value, the corresponding frequency f k can be calculated by the following formula (assuming the sampling frequency is f s ): f s is the number of data points collected per second;

[0024] Amplitude spectrum: It represents the amplitude size of each frequency component. Since X[k] is a complex number, its amplitude |X[k]| is calculated by the following formula:

[0025]

[0026] where Re(X[k]) is the real part of |X[k]| and Im(X[k]) is the imaginary part of X[k];

[0027] To obtain the correct amplitude ratio, it is usually necessary to scale the result. If the original signal x[n] has not been specially processed, for the actually collected vibration data, the first half of the amplitude spectrum (k = 0 to ) contains all the effective frequency information (because the spectrum of a real signal has symmetry), and |X[0]| and do not require additional scaling, while needs to be multiplied by for scaling to obtain the spectral amplitude corresponding to the amplitude of the original signal. The formula is:

[0028]

[0029] Through the above calculations, different frequency components f can be obtained. k The corresponding amplitude A[k] can be obtained, thereby obtaining the spectral characteristics of the vibration, observing the amplitude changes of different frequency components. When the amplitude of the natural frequency of the mill or the characteristic frequency related to the fault shows an abnormal increase, it may predict that there is a fault in the mill.

[0030] In a possible implementation manner, in the data analysis module, according to the normal operating conditions of the mill, a relationship model between vibration parameters and load is established to determine the change range of the vibration amplitude in different load intervals, and the corresponding relationship between specific frequency components and abnormal load conditions.

[0031] Rotation speed analysis model: Combining the real-time data and historical data of the grinding disc rotation speed, using the time series analysis method, with the time series of the grinding disc rotation speed y t (t = 1, 2, …, T, where T is the length of the time series), using the autoregressive model to assume that the value at the current moment linearly depends on the values of several past moments, and its formula is: y t is the value of the grinding disc rotation speed at time t; p is the autoregressive order, that is, the model considers the values of the past p moments to predict the current value; is the autoregressive coefficient, measuring the influence degree of the value at the i-th past moment on the current value; ∈ t is the white noise error term, representing the random fluctuation that cannot be explained by the past values, and usually assuming that its mean value is 0.

[0032] In a possible implementation manner, in the data transmission module, the mill sensor collects key operation data, and there is a lot of electromagnetic interference in the industrial environment. Due to its special structure, the shielding layer of the shielded cable can direct the external electromagnetic interference to the ground, reduce the influence on the transmitted signal, ensure the accurate transmission of data, provide a reliable data basis for the mill load analysis, and help to judge the operating state of the mill and fault warning.

[0033] In a possible implementation, in the historical memory and update module, in the mill load analysis system, all the collected operation data and analysis results are stored for a long time to establish a historical database. The historical data therein includes, on the one hand, the parameter values of real-time monitoring, such as vibration data (vibration acceleration, vibration velocity) and the rotational speed of the grinding disc, which directly reflect the immediate state of the mill operation. On the other hand, it also covers the load state information obtained through multi-parameter model analysis, such as whether the mill is in a light load, full load or overload state, which enables the staff to clearly understand the load level of the mill. In addition, the fault diagnosis records are also included, which record the time of fault occurrence, phenomena, possible causes and countermeasures for subsequent reference. As the running time of the mill increases, the working conditions will also change continuously. To make the multi-parameter model fit the actual operation conditions of the mill, the system will regularly update and optimize the model using new historical data. Specifically, the system will, according to a set period (set according to the running stability of the mill and the frequency of working condition changes, such as weekly, monthly or quarterly), incorporate the new historical data into the model for retraining. In the vibration analysis model, the relationship between vibration parameters and the mill load is re-determined based on the new data, and the relevant coefficients are adjusted. In the rotational speed analysis model, the parameters of time series analysis are optimized by combining the wear of the grinding disc and the change in rotational speed. Through retraining, the model can analyze the mill load more accurately, improve the accuracy of fault diagnosis, and detect potential faults in a timely manner.

[0034] Beneficial effects compared with the prior art:

[0035] 1. In this solution, comprehensive and accurate mill load analysis is achieved through the collaborative work of multiple modules. The data acquisition module uses a variety of sensors to comprehensively collect vibration and grinding disc rotational speed data, providing rich and accurate basic data for subsequent analysis. The vibration data is collected by acceleration and velocity sensors and processed through signal conditioning and a data acquisition card to ensure data quality. The grinding disc rotational speed data is obtained in real time through a rotational speed sensor at a specific position and stored in combination with the historical memory module. The data analysis module conducts time-domain and frequency-domain analysis on the vibration data to establish a relationship model with the load, and at the same time uses time series analysis for the grinding disc rotational speed to dynamically update the threshold to judge the load state. This multi-parameter data fusion and analysis method can reflect the actual situation of the mill load more comprehensively and accurately compared with single-parameter analysis, which helps to detect potential fault hazards in a timely manner, avoid equipment damage caused by abnormal load of the mill, and ensure the continuity and stability of production.

[0036] 2. In this solution, the dynamic optimization and adaptive adjustment of the model are achieved through the historical memory and update module; all operating data and analysis results of the system are stored in the historical database for a long time, covering rich information such as real-time parameter values, load status, and fault diagnosis records; as the mill operates, the working conditions change continuously, and the system regularly retrains the multi-parameter model with new historical data; this enables the model to better adapt to changes in the actual operating conditions of the mill, such as the influence of grinding plate wear, material property changes, etc. on load analysis; through continuous optimization, the accuracy of load analysis and fault diagnosis is improved, providing a more reliable basis for formulating the mill operation and maintenance strategy, reasonably arranging resources, reducing maintenance costs, and extending the service life of the mill. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The above description is only an overview of the technical solution of the present invention. In order to understand the technical means of the present invention more clearly and implement it according to the content of the specification, the following describes the preferred embodiments of the present invention in detail with reference to the accompanying drawings.

[0038] Figure 1 It is a schematic diagram of the system framework structure of the present invention;

[0039] Figure 2 It is a schematic diagram of the system operation process of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] The preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention can be implemented in various different forms, so the present invention is not limited to the embodiments described below;

[0041] The technical solutions in the embodiments of the present application are to solve the problems in the above background technology, and the general idea is as follows:

[0042] Embodiment 1:

[0043] Please refer to Figure 1 and Figure 2 As shown, this embodiment introduces a mill load analysis system based on a multi-parameter model, including the following modules;

[0044] S1. Data acquisition module

[0045] Vibration data: Install an acceleration sensor and a velocity sensor. The acceleration sensor is selected from piezoelectric or piezoresistive types to obtain vibration acceleration information; the velocity sensor is based on the principle of electromagnetic induction to measure vibration velocity, and then the signal conditioning module amplifies and filters the weak electrical signals output by the sensors to remove noise interference. The data acquisition card converts the analog signals into digital signals and collects them at a set sampling frequency (at least twice the highest vibration frequency component of the mill or more).

[0046] Disc grinding speed data: A speed sensor is installed near the rotating shaft of the grinding disc of the mill to monitor the rotational speed of the grinding disc in real time. By detecting the frequency of the pulse signal emitted by the sensor, the real-time rotational speed of the grinding disc is calculated, and the real-time data of the grinding disc speed is recorded. Combining with the historical memory module, the speed data at different time nodes is stored to provide data support for subsequent threshold updates and load analysis.

[0047] S2: Data transmission module:

[0048] For mills that are relatively close to the control center, shielded cables are used to transmit the data collected by each sensor to the data processing center; the mill sensors collect key operating data, and there is a lot of electromagnetic interference in the industrial environment; due to its special structure, the shielding layer of the shielded cable can direct the external electromagnetic interference to the ground, reducing the impact on the transmitted signal, ensuring accurate data transmission, providing a reliable data basis for the load analysis of the mill, and helping to judge the operating status and fault warning of the mill.

[0049] S3: Data analysis module:

[0050] Vibration analysis model: Perform time-domain and frequency-domain analysis on the vibration data. In the time domain, assuming that the vibration data is given in the form of a discrete sequence x[n], where n = 0, 1,..., N - 1, and N is the total number of data points, the amplitude, mean, and peak statistical parameters of the vibration are calculated.

[0051] Amplitude: The peak-to-peak value is the difference between the maximum and minimum values of the vibration signal in one cycle, reflecting the fluctuation range of the signal. Its formula is:

[0052] A P-P = max(x[n]) - min(x[n])

[0053] Mean: The mean represents the average level of the vibration signal, which is obtained by summing all data points and dividing by the total number of data points. Its formula is:

[0054]

[0055] Peak: The peak is the maximum value in the vibration signal. Its formula is:

[0056] A peak = max(x[n])

[0057] In the frequency domain, the spectral characteristics of the vibration are obtained through the fast Fourier transform to identify the amplitude changes of different frequency components.

[0058] Fast Fourier transform: For a discrete-time sequence x[n] of length N, where n = 0, 1,..., N - 1, its discrete Fourier transform is defined as:

[0059]

[0060] where \(k = 0, 1, \ldots, N - 1\)

[0061] Obtain the spectral characteristics and amplitude: Assume that the result obtained by fast Fourier transform is \(X[k]\), which is a complex number array. For each value of \(k\), the corresponding frequency \(f\) k can be calculated by the following formula (assuming the sampling frequency is \(f\) s ):

[0062]

[0063] \(f\) s is the number of data points collected per second.

[0064] Amplitude spectrum: It represents the amplitude size of each frequency component. Since \(X[k]\) is a complex number, its amplitude \(|X[k]|\) can be calculated by the following formula:

[0065]

[0066] where \(Re(X[k])\) is the real part of \(|X[k]|\), and \(Im(X[k])\) is the imaginary part of \(X[k]\);

[0067] To obtain the correct amplitude ratio, it is usually necessary to scale the result. If the original signal \(x[n]\) has not been specially processed, for the actually collected vibration data, the first half of the amplitude spectrum (\(k = 0\) to ) contains all the effective frequency information (because the spectrum of a real signal has symmetry), and \(|X[0]|\) and do not require additional scaling, while needs to be multiplied by for scaling to obtain the spectral amplitude corresponding to the amplitude of the original signal. The formula is:

[0068]

[0069] Through the above calculations, the amplitude \(A[k]\) corresponding to different frequency components \(f\) k can be obtained, thereby obtaining the spectral characteristics of the vibration and observing the amplitude changes of different frequency components. When the amplitude of the natural frequency of the mill or the characteristic frequency related to the fault shows an abnormal increase, it may predict that there is a fault in the mill;

[0070] According to the normal operating conditions of the mill, establish a relationship model between the vibration parameters and the load, determine the change range of the vibration amplitude in different load intervals, and the corresponding relationship between specific frequency components and abnormal load conditions.

[0071] Rotational speed analysis model: Combining the real-time data and historical data of the rotational speed of the grinding disc, using the time series analysis method, with the time series of the grinding disc rotational speed y t (t = 1, 2, …, T, where T is the length of the time series), using the autoregressive model to assume that the value at the current moment linearly depends on the values at several past moments, and its formula is:

[0072]

[0073] y t is the value of the grinding disc rotational speed at time t; p is the autoregressive order, that is, the model considers the values at the past p moments to predict the current value; is the autoregressive coefficient, measuring the influence degree of the value at the i-th past moment on the current value; ∈ t is the white noise error term, representing the random fluctuations that cannot be explained by the past values, and usually assuming its mean value is 0.

[0074] to predict the change trend of the rotational speed. According to the usage time and wear degree of the grinding disc, dynamically update the rotational speed threshold. When the grinding disc rotational speed exceeds or is lower than the threshold range, combine the vibration amplitude parameter to judge the mill load state. When the grinding disc rotational speed suddenly drops and at the same time the vibration amplitude increases, it indicates that the mill load increases.

[0075] S4: Alarm module

[0076] According to the analysis and diagnosis results, when the operating parameters of the mill exceed the normal range or a potential fault is detected, transmit the data to the display interface of the device, and it will prompt the staff that the mill is overloaded.

[0077] S5: Historical memory and update module

[0078] Historical data storage: In the mill load analysis system, all the collected operating data and analysis results will be stored for a long time to establish a historical database; the historical data here, on the one hand, includes the parameter values of real-time monitoring, such as vibration data (vibration acceleration, vibration speed) and the rotational speed of the grinding disc, which directly reflect the immediate state of the mill operation. On the other hand, it also covers the load state information obtained from the multi-parameter model analysis, such as whether the mill is in a light load, full load or overload state, which can enable the staff to clearly understand the mill load level. In addition, the fault diagnosis records are also included, which record the fault occurrence time, phenomenon, possible causes and countermeasures for subsequent reference;

[0079] Model Update: As the running time of the mill increases, the operating conditions will also change continuously. To make the multi-parameter model fit the actual operating conditions of the mill, the system will regularly update and optimize the model using new historical data. Specifically, the system will incorporate new historical data into the model for retraining at a set period (set according to the running stability of the mill and the frequency of changes in operating conditions, such as weekly, monthly, or quarterly). In the vibration analysis model, the relationship between vibration parameters and mill load will be re-determined based on the new data, and the relevant coefficients will be adjusted. In the rotational speed analysis model, the parameters of time series analysis will be optimized by combining the wear of the grinding disc and the change in rotational speed. Through retraining, the model can analyze the mill load more accurately, improve the accuracy of fault diagnosis, and detect potential faults in a timely manner.

[0080] Finally, it should be noted that: Obviously, the above embodiments are only examples for clearly illustrating the present invention and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. And the obvious changes or modifications derived therefrom are still within the protection scope of the present invention.

Claims

1. A mill load analysis system based on a multi-parameter model, characterized in that: Includes the following modules: S1. Vibration data: Install acceleration sensors and velocity sensors. Acceleration sensors are piezoelectric or piezoresistive, which are used to obtain vibration acceleration information. Velocity sensors measure vibration velocity based on the principle of electromagnetic induction, and then amplify and filter the weak electrical signals output by the sensor through the signal conditioning module to remove noise interference. The data acquisition card converts the analog signal into a digital signal and collects it at the set sampling frequency. Grinding disc speed data: Use a speed sensor, which is installed near the rotating shaft of the grinding disc of the mill to monitor the speed of the grinding disc in real time. By detecting the frequency of the pulse signal emitted by the sensor, the real-time speed of the grinding disc is calculated, and the real-time data of the grinding disc speed is recorded. In combination with the historical memory module, the speed data at different time nodes is stored to provide data support for subsequent threshold updates and load analysis. S2, data transmission module: using shielded cables to transmit the data collected by each sensor to the data processing center; S3, data analysis module: vibration analysis model: perform time domain and frequency domain analysis on vibration data. In the time domain, assume that the vibration data is given in the form of a discrete sequence x[n], n=0, 1, ..., N-1, where N is the total number of data points, so as to calculate the amplitude, mean, and peak statistical parameters of the vibration. In the frequency domain, obtain the spectral characteristics of the vibration through fast Fourier transform, identify the amplitude changes of different frequency components, and establish a relationship model between vibration parameters and loads according to the normal operating conditions of the mill. Determine the range of vibration amplitude changes in different load intervals, as well as the correspondence between specific frequency components and load abnormalities, to predict the change trend of the speed. According to the use time and wear degree of the grinding disc, dynamically update the speed threshold. When the grinding disc speed exceeds or falls below the threshold range, the mill load state is judged in combination with the vibration amplitude parameter. When the grinding disc speed suddenly drops and the vibration amplitude increases at the same time, it indicates that the mill load has increased. S4, alarm module: according to the analysis and diagnosis results, when the mill operating parameters exceed the normal range or a potential fault is detected, the data is transmitted to the display interface of the device to prompt the staff that the mill is overloaded; S5. Historical memory and update module: All collected operating data and analysis results are stored for a long time to establish a historical database. As the mill operating time increases and the operating conditions change, the multi-parameter model is regularly updated and optimized using new historical data.

2. A mill load analysis system based on a multi-parameter model as claimed in claim 1, characterized in that: In the data analysis module, the amplitude: peak-to-peak value is the difference between the maximum and minimum values ​​of the vibration signal in one cycle, reflecting the fluctuation range of the signal, and its formula is: A P-P =max(x[n])-min(x[n]) Mean: The mean represents the average level of the vibration signal and is obtained by summing all data points and dividing by the total number of data points. The formula is: Peak value: The peak value is the maximum value in the vibration signal. Its formula is: A peak =max(x[n])。 3. A mill load analysis system based on a multi-parameter model as claimed in claim 2, characterized in that: In the data analysis module, for a discrete time series of length N, x[n], n=0, 1…, N-1, its discrete Fourier transform is defined as: in, Obtaining spectral characteristics and amplitude: Assume that the result obtained by fast Fourier transform is X[k], which is a complex array. For each k value, the corresponding frequency f k Calculated by the following formula, assuming the sampling frequency is f s : f s is the number of data points collected per second; Amplitude spectrum: It represents the amplitude of each frequency component. Since X[k] is a complex number, its amplitude |X[k]| is calculated by the following formula: Where Re(X[k]) is the real part of |X[k]|, and Im(X[k]) is the imaginary part of X[k]; In order to obtain the correct amplitude ratio, it is usually necessary to scale the result. If the original signal x[n] is not specially processed, for the actual collected vibration data, the first half of the amplitude spectrum, k = 0 to Contains all valid frequency information, and |X[0]| and No additional scaling is required, and Need to multiply Scaling is performed to obtain the spectrum amplitude corresponding to the original signal amplitude. The formula is: Through the above calculation, we can get different frequency components f k The corresponding amplitude A[k] is used to obtain the spectral characteristics of the vibration and observe the amplitude changes of different frequency components. When the amplitude of the natural frequency of the mill or the characteristic frequency related to the fault increases abnormally, it may be predicted that the mill has a fault.

4. A mill load analysis system based on a multi-parameter model as claimed in claim 3, characterized in that: In the data analysis module, according to the normal operating conditions of the mill, a relationship model between vibration parameters and load is established to determine the variation range of vibration amplitude in different load intervals, as well as the corresponding relationship between specific frequency components and load abnormalities; Speed ​​analysis model: Combine the real-time data and historical data of grinding disc speed, use the time series analysis method, and use the grinding disc speed time series y t (t=1,2,…,T, T is the length of the time series), using the autoregressive model to assume that the value at the current moment is linearly dependent on the values ​​at several moments in the past, and the formula is: y t is the value of the grinding disc speed at time t; p is the autoregressive order, that is, the model considers the values ​​of the past p moments to predict the current value; is the autoregressive coefficient, which measures the influence of the value at the past i-th moment on the current value; ∈ t is a white noise error term, representing random fluctuations that cannot be explained by past values, and is usually assumed to have a mean of 0.

5. The mill load analysis system based on a multi-parameter model according to claim 1, characterized in that: In the data transmission module, the mill sensor collects key operating data, and the industrial environment has a lot of electromagnetic interference. Due to the special structure of the shielded cable, its shielding layer can guide the external electromagnetic interference to the ground, reduce the impact on the transmission signal, ensure accurate data transmission, provide a reliable data basis for mill load analysis, and help judge the mill operation status and fault warning.

6. A mill load analysis system based on a multi-parameter model as claimed in claim 1, characterized in that: In the historical memory and update module, the historical data includes, on the one hand, parameter values ​​monitored in real time, such as vibration data and grinding disc speed, which directly reflect the instantaneous state of the mill operation, and on the other hand, load state information obtained based on the multi-parameter model analysis, such as whether the mill is in a light load, full load or overload state, which allows the staff to clearly understand the mill load level. In addition, fault diagnosis records are also included, which record the time, phenomenon, possible causes and countermeasures of the fault, which is convenient for subsequent reference. As the mill operation time increases, the operating conditions will continue to change. In order to make the multi-parameter model fit the actual operation status of the mill, the system will regularly use new historical data to update and optimize the model. Specifically, the system will incorporate new historical data into the model for retraining according to the set cycle. In the vibration analysis model, the relationship between vibration parameters and mill load is re-determined based on the new data, and the correlation coefficient is adjusted. In the speed analysis model, the parameters of time series analysis are optimized in combination with grinding disc wear and speed changes. Through retraining, the model can more accurately analyze the mill load, improve the accuracy of fault diagnosis, and detect potential faults in a timely manner.

Images