A method for health assessment of industrial production equipment

Abstract

The present invention relates to a method for evaluating the health of industrial production equipment, comprising the following steps: real-time collection of equipment feature data, the equipment feature data including equipment operating status data and equipment environmental parameters; data preprocessing of the equipment feature data; extracting key features from the preprocessed data by multivariate regression analysis, and determining the priority of each key feature's impact on equipment health by correlation analysis; constructing an equipment health evaluation model using the degree of deviation between the equipment feature data and the ideal value and the impact priority; inputting the equipment feature data to be evaluated into the equipment health evaluation model to obtain the overall health score of the equipment. The present invention collects the equipment's operating status data, historical data, and environmental factors, and uses advanced data analysis technology and machine learning algorithms to achieve accurate evaluation of the equipment's health status, providing a scientific basis for preventive maintenance and fault prediction of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment health evaluation, and in particular to an industrial production equipment health evaluation method. Background Art

[0002] With the rapid development of industrialization and informatization, industrial production equipment has become one of the core elements in various production links. Equipment health evaluation refers to the real-time monitoring, evaluation and preventive suggestions of the equipment operation status in the industrial production process, so as to optimize the equipment maintenance strategy and ensure production efficiency and equipment safety. The normal operation of equipment is not only directly related to the quality and production efficiency of the product, but also to production safety and the cost-effectiveness of the enterprise. However, traditional equipment maintenance methods are mostly based on post-maintenance, that is, repairing or replacing the equipment after it fails. This method will not only lead to production interruptions and affect production efficiency, but also may cause irreversible damage to the equipment and even cause safety accidents.

[0003] In order to solve the limitations of the traditional post-maintenance model, preventive maintenance came into being. Preventive maintenance refers to ensuring that the equipment is always in good working condition and avoiding equipment failure through regular inspection, maintenance and care before the equipment fails. Compared with post-maintenance, preventive maintenance can significantly reduce the failure rate and downtime of equipment, thereby improving production efficiency and reducing equipment maintenance costs. In addition, preventive maintenance can also extend the service life of equipment and reduce the occurrence of safety accidents.

[0004] With the development of sensing technology, information technology and artificial intelligence technology, equipment health evaluation has gradually become a key means to achieve intelligent maintenance. Equipment health evaluation collects various operating data of the equipment in real time, such as vibration, temperature, pressure, current, etc., through sensors installed on the equipment. Then, through data analysis technology, the operating status of the equipment is analyzed, potential signs of failure are identified, and the possible failure time of the equipment is predicted. This process can not only provide a basis for preventive maintenance of the equipment, but also realize predictive maintenance, that is, targeted maintenance is carried out before the equipment is about to fail, thereby minimizing the failure risk and downtime of the equipment. At present, there is an equipment health evaluation method put into use in coal preparation plants. By modeling the temperature, current, vibration and other data of the equipment, the fault prediction model is used to predict and analyze the characteristic data of the equipment regularly. However, the data indicators concerned by this method are not aggregated at the overall level of the equipment, nor do they reflect the effects of different indicators on the health of the equipment. Therefore, how to comprehensively evaluate the health of different characteristic data and their impact on the health of the equipment has become a technical problem that needs to be solved by technicians in this field, which is of practical significance for ensuring the normal operation of the equipment, extending the life of the equipment, and improving production efficiency. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies in the prior art and provide a method for evaluating the health of industrial production equipment. By collecting the operating status data, historical data and environmental factors of the equipment and using advanced data analysis technology and machine learning algorithms, an accurate assessment of the health status of the equipment can be achieved, providing a scientific basis for preventive maintenance and fault prediction of the equipment.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for evaluating the health of industrial production equipment comprises the following steps:

[0008] S1. Collecting device characteristic data in real time through monitoring equipment, the device characteristic data includes device operation status data and device environment parameters;

[0009] S2. Performing data preprocessing on the device characteristic data, wherein the data preprocessing includes filtering and denoising;

[0010] S3. Extract key features from the preprocessed data through multivariate regression analysis, and determine the priority of each key feature on the health of the equipment through correlation analysis;

[0011] S4. According to the impact priority of key features on equipment health, the equipment health assessment model is constructed by using the degree of deviation between equipment feature data and ideal values ​​and the impact priority;

[0012] S5. Input the characteristic data of the device to be evaluated into the device health evaluation model to obtain the overall health score of the device.

[0013] According to the above technical solution, preferably, in step S1, the equipment operation status data includes current, temperature, vibration, number of alarms, number of maintenances and failure rate, and the equipment environmental parameters include ambient temperature, humidity and air pressure.

[0014] According to the above technical solution, preferably, step S2 includes:

[0015] Using linear interpolation to fill in missing data in the collected device characteristic data;

[0016] After the missing data is filled, the Kalman filter algorithm is used to denoise the data as a whole.

[0017] According to the above technical solution, preferably, in step S3, a set of device characteristic data X1, X2, ..., X n To predict the health status Y of the equipment,

[0018] Y=β0+β1X1+β2X2+…+β n X n

[0019] Among them, Y is the equipment health status indicator, X1, X2…X n is the characteristic data of each device status, β0 represents the initial value of the device health index when there is no characteristic data influence, β1, β2…β n is the regression coefficient of each feature, indicating the influence weight of each feature on the equipment health index.

[0020] According to the above technical solution, preferably, in step S3, the multivariate regression analysis model is minimized.

[0021]

[0022] Among them, Y i is the actual device health indicator, is the health status predicted by the multiple regression analysis model, and m is the number of data points.

[0023] According to the above technical solution, preferably, step S4 includes:

[0024] Using an error scoring function to score each device feature data;

[0025] After obtaining the evaluation score of each device feature data, the overall health score of the device is calculated based on the priority of the impact of key features on device health.

[0026] The beneficial effects of the present invention are:

[0027] The present invention uses a machine learning algorithm to extract and analyze equipment feature data, determine the degree of influence of different feature data on equipment health, and then determine the influence weights of various feature data, and integrate them into the equipment health evaluation model, combining different feature data with their weights to evaluate the health of the equipment as a whole. The present invention can help on-site personnel clearly understand the health status of each device, improve the accuracy and efficiency of equipment health assessment, provide a scientific basis for preventive maintenance and fault prediction of equipment, reduce equipment failure rate and maintenance costs, and improve production efficiency and safety. It is applicable to various types of equipment and has strong versatility and scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the implementation flow of the method of the present invention.

[0029] Figure 2 It is a schematic diagram of data processing in an embodiment of the present invention.

[0030] Figure 3The evaluation results of each indicator in the embodiment of the present invention are Figure 1 .

[0031] Figure 4 The evaluation results of each indicator in the embodiment of the present invention are Figure 2 .

[0032] Figure 5 The evaluation results of each indicator in the embodiment of the present invention are Figure 3 . DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the technical solution of the present invention, the actual production process of a domestic coal preparation plant is taken as an example for further detailed description. Based on the embodiments of the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0034] As shown in the figure, the present invention comprises the following steps:

[0035] S1. Collect device characteristic data in real time through monitoring equipment, wherein the device characteristic data includes device operation status data and device environment parameters.

[0036] Specifically, the operating status data of the equipment, such as temperature, vibration, current, etc., is collected in real time through monitoring devices such as sensors; at the same time, the historical data of the equipment is obtained, including the number of alarms, the number of maintenance times, and the failure rate, etc. In addition, the environmental factors around the equipment, namely temperature, humidity, air pressure, etc., must also be considered.

[0037] S2. Perform data preprocessing on the device characteristic data, wherein the data preprocessing includes filtering and denoising to improve the efficiency and effectiveness of the data analysis process.

[0038] Specifically, linear interpolation is first used to fill in the missing data in the original data:

[0039]

[0040] Among them, (x i ,y i ) and (x k ,y k ) is the adjacent known data point, x is the position of the interpolation point, located at x i and x j y is the specific inserted value at the x position. Interpolation processing can effectively deal with small missing data. If the data is missing significantly (greater than 30%), the collection of PLC points should be checked first to improve the quality of the original data.

[0041] After the missing data is filled, the Kalman filter algorithm is used to denoise the data as a whole. The Kalman filter consists of two stages, namely prediction and update. The prediction steps are as follows:

[0042]

[0043] in, is the predicted value of the equipment status at time k, based on the equipment status data at time k-1 And the state transfer matrix F k To make predictions, the state transfer matrix describes the impact of the state at the previous moment on the state at the current moment, which is determined by production experience. k is the external control input, i.e. the environmental factors affecting the equipment, B k is the control input matrix used to control u k The degree of control can be set according to the actual situation of the equipment. k|k―1 is the predicted value of the covariance matrix at time k, which is used to estimate the error of the equipment status data. The covariance matrix is ​​also given based on manual experience at the beginning. k is the covariance of the process noise in the equipment status data, which is given by prior knowledge.

[0044] The update steps are as follows:

[0045]

[0046] P k|k =(I-K k H k ) k|k―1

[0047] Among them, K k is the Kalman gain matrix, which weighs the trust between the device's predicted state and the observed state. The role of the gain is to determine how to balance the weight of the observed information and the model prediction. If the observed data is unreliable (large observation noise, R k If the observed data is reliable (R k If the data is small, it will rely more on observational data. is the observation matrix, which corresponds to the type of device status data, R k is the covariance matrix of the observation noise, that is, the measurement error of the sensor, which is given by artificial production and measurement experience. is the updated state estimate, which is determined by the predicted state estimate and the observation residual with Kalman gain Combination of z kis the device status data observation vector at time k, representing the device status observation data obtained from the PLC. k|k is the updated covariance matrix. It represents the new uncertainty. After the state is updated, the covariance usually decreases because the new observations reduce the uncertainty of the state, where I is the identity matrix.

[0048] After the device status data collected by PLC is processed by the above-mentioned Kalman filter, the data quality can be greatly improved, providing a good data quality foundation for subsequent feature processing.

[0049] S3. Use the multivariate regression method to extract key features from the preprocessed data, and perform correlation fitting on the key features obtained through feature extraction to determine the priority of each key feature's impact on equipment health.

[0050] Specifically, the key equipment status features are extracted from the preprocessed data through multivariate regression analysis, and the priority of each key feature on equipment health is determined through correlation analysis. The goal of multivariate regression is to use a set of equipment feature data X1, X2, …, X n To predict the health status Y of the equipment, the formula is as follows:

[0051] Y=β0+β1X1+β2X2+…+β n X n

[0052] Among them, Y is the equipment health status indicator, represented by the failure rate, X1, X2…X n is the characteristic data of each device state, represented by current, temperature, and vibration. β0 represents the initial value of the health index of the device when there is no characteristic data influence. β1, β2…β n is the regression coefficient of each feature, indicating the influence weight of each feature on the equipment health index.

[0053] The current, temperature, and vibration data of each device are selected as the main features. At the same time, in order to determine the impact of environmental factors on the health of the equipment, the temperature, humidity, and air pressure of the environment in which the equipment is located are selected for another set of multiple regression analysis. This is to separate the strong correlation features from the weak correlation features during the analysis process to prevent the weak correlation features from being diluted by the strong correlation features. After that, the multiple regression analysis model is minimized:

[0054]

[0055] Among them, Y i is the actual device health indicator, is the health status predicted by the multiple regression analysis model, and m is the number of data points.

[0056] After training, we can get the influence degree of each equipment characteristic data on its health index β i The larger the absolute value, the greater the impact of this feature on the health of the device.

[0057] S4. According to the priority of the impact of key features on equipment health, the equipment health assessment model is constructed by using the degree of deviation between the equipment feature data and the ideal value and the impact priority. In the process of model construction, the correlation between different features and their impact on the health of the equipment also needs to be considered.

[0058] Specifically, the error scoring function is first used to score each device feature data:

[0059]

[0060] Among them, S i is the score of each feature index, x r is the actual value of the characteristic index during the evaluation period, x p It is the expected value of the characteristic index. This method is used to calculate the scores of the current, temperature, and vibration characteristic data of the equipment.

[0061] After obtaining the evaluation score of each device feature, the device feature data priority determined in step S3 is used to calculate the overall health score of the device, which represents the overall health of the device, as shown in the following formula:

[0062]

[0063] Among them, S is the total health score of the device, S i is the score of each sub-item feature data, β i is the corresponding weight, from which the final health score of the equipment can be calculated. The threshold and standard of the health score can be set according to the actual situation, and the specific fault indicators can be determined when the equipment is in poor health, so that on-site personnel can quickly locate the root cause of the problem.

[0064] S5. Input the characteristic data of the device to be evaluated into the device health evaluation model to obtain the overall health score of the device. According to the scoring result, the health status of the device can be judged and a corresponding maintenance strategy can be formulated.

[0065] Specifically, the characteristic data of the device to be evaluated is input into the constructed health assessment model to obtain the overall health score of the device. Figure 2 The health evaluation results of the three indicators of current, temperature and vibration of the equipment are as follows: Figure 3As shown in the figure, the over-limit situation and slope trend situation are used to evaluate various indicators of the equipment, and the health status of each indicator is obtained. Then all indicators are weighted to make an overall score for the equipment. According to the scoring results and recommended details, the health status of the equipment can be judged and corresponding maintenance strategies can be formulated.

[0066] The present invention first performs feature lossless filtering on the equipment's operating data, extracts its health evaluation indicators according to different types of equipment analysis, and uses multivariate regression analysis to determine its parameter priority. Then, an error evaluation function is used to obtain the health score of each equipment characteristic parameter, and finally a comprehensive health evaluation of the equipment is given.

[0067] The present invention proposes and analyzes the equipment characteristic data to determine the characteristic data with the greatest impact on the health of the equipment, and uses multiple regression analysis to determine the priority of each type of characteristic data. Finally, the error evaluation function is used to evaluate each characteristic data within a period of time, and the total score of the equipment health evaluation is given according to the priority weight. The scheme disclosed by the present invention can help on-site personnel to clearly understand the health status of each device, and conduct targeted inspection and maintenance work in the event of an abnormality, effectively improve the monitoring effect of the status of each device in the coal preparation plant, and has practical significance for ensuring the normal operation of the equipment, extending the life of the equipment, and improving production efficiency.

[0068] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for evaluating the health of industrial production equipment, characterized in that: The steps include: S1. Collecting device characteristic data in real time through monitoring equipment, the device characteristic data includes device operation status data and device environment parameters; S2. Performing data preprocessing on the device characteristic data, wherein the data preprocessing includes filtering and denoising; S3. Extract key features from the preprocessed data through multivariate regression analysis, and determine the priority of each key feature on the health of the equipment through correlation analysis; S4. According to the impact priority of key features on equipment health, the equipment health assessment model is constructed by using the degree of deviation between equipment feature data and ideal values ​​and the impact priority; S5. Input the characteristic data of the device to be evaluated into the device health evaluation model to obtain the overall health score of the device.

2. According to claim 1, a method for evaluating the health of industrial production equipment is characterized in that: In step S1, the equipment operation status data includes current, temperature, vibration, alarm times, maintenance times and failure rate. The equipment environment parameters include ambient temperature, humidity and air pressure.

3. According to claim 1, a method for evaluating the health of industrial production equipment is characterized in that: Step S2 includes: Using linear interpolation to fill in missing data in the collected device characteristic data; After the missing data is filled, the Kalman filter algorithm is used to denoise the data as a whole.

4. According to claim 1, a method for evaluating the health of industrial production equipment is characterized in that: In step S3, a set of device characteristic data X1, X2, ..., X n To predict the health status Y of the equipment, Y = β0 + β1X1 + β2X2 + ... + β n X n Among them, Y is the equipment health status indicator, X1, X2…X n is the characteristic data of each device status, β0 represents the initial value of the device health index when there is no characteristic data influence, β1, β2…β n is the regression coefficient of each feature, indicating the influence weight of each feature on the equipment health index.

5. According to claim 4, a method for evaluating the health of industrial production equipment is characterized in that: In step S3, the multivariate regression analysis model is minimized. Among them, Y i is the actual device health indicator, is the health status predicted by the multiple regression analysis model, and m is the number of data points.

6. The method for evaluating the health of industrial production equipment according to any one of claims 1 to 5, characterized in that: Step S4 includes: Using an error scoring function to score each device feature data; After obtaining the evaluation score of each device feature data, the overall health score of the device is calculated based on the priority of the impact of key features on device health.

Images