Resin grinding wheel factory production management system and method

Abstract

The invention provides a resin grinding wheel factory production management system and method, relates to the field of intelligent management, and analyzes hydraulic oil pressure time sequence data, hydraulic oil temperature time sequence data and hydraulic oil flow time sequence data of a target monitoring oil press to know the operation state of an oil press hydraulic system. Meanwhile, the vibration signal of the target monitoring oil press is analyzed to know the operation state of the mechanical part of the oil press, and whether the target monitoring oil press needs to be maintained or not is judged by comprehensively considering the operation state of the hydraulic system of the oil press and the operation state of the mechanical part of the oil press. In this way, accurate maintenance of the oil press is facilitated, the downtime is shortened, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent management, and more specifically, to a production management system and method for a resin grinding wheel factory. Background Art

[0002] Resin grinding wheels are common industrial production consumables in people's lives and are the main type of grinding tools in grinding processing, with a large market demand. In a resin grinding wheel factory, a hydraulic press is a key equipment for producing resin grinding wheels, and its operating state directly affects product quality and production efficiency. Under the traditional production management mode, the maintenance of this key production equipment, the hydraulic press, mainly depends on a predetermined schedule or after a failure occurs. This method may lead to untimely maintenance, resulting in unexpected shutdowns and production interruptions, and irregular or excessive maintenance will increase unnecessary maintenance costs.

[0003] Therefore, an optimized production management solution for a resin grinding wheel factory is needed. Summary of the Invention

[0004] In order to solve the above technical problems, the present application is proposed. Embodiments of the present application provide a production management system and method for a resin grinding wheel factory.

[0005] According to one aspect of the present application, a production management system for a resin grinding wheel factory is provided, which includes:

[0006] A hydraulic press monitoring data acquisition module for obtaining the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow time series data, and vibration signals of a target monitored hydraulic press measured by a sensor group;

[0007] A hydraulic press hydraulic system operation analysis module for comprehensively analyzing the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow time series data of the target monitored hydraulic press to obtain a hydraulic press hydraulic system operation state feature vector;

[0008] A hydraulic press mechanical component operation analysis module for analyzing the vibration signals to obtain a hydraulic press mechanical component operation state feature vector;

[0009] A hydraulic press operation state monitoring feature fusion module for fusing the hydraulic press hydraulic system operation state feature vector and the hydraulic press mechanical component operation state feature vector to obtain a hydraulic press operation state monitoring feature vector;

[0010] A maintenance result generation module for judging whether the target monitored hydraulic press needs to be maintained according to the information contained in the hydraulic press operation state monitoring feature vector.

[0011] According to another aspect of the present application, there is provided a production management method for a resin grinding wheel factory, which includes:

[0012] Obtaining the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow rate time series data and vibration signals of the target monitored hydraulic press measured by the sensor group;

[0013] Performing comprehensive time series analysis on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to obtain the operation state feature vector of the hydraulic press hydraulic system;

[0014] Performing signal analysis on the vibration signal to obtain the operation state feature vector of the mechanical components of the hydraulic press;

[0015] Fusing the operation state feature vector of the hydraulic press hydraulic system and the operation state feature vector of the mechanical components of the hydraulic press to obtain the operation state monitoring feature vector of the hydraulic press;

[0016] Judging whether the target monitored hydraulic press needs to be maintained according to the information contained in the operation state monitoring feature vector of the hydraulic press.

[0017] Compared with the prior art, the resin grinding wheel factory production management system and method provided by the present application analyze the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to understand the operation state of the hydraulic press hydraulic system. At the same time, by analyzing the vibration signal of the target monitored hydraulic press to understand the operation state of the mechanical components of the hydraulic press, and comprehensively considering the operation state of the hydraulic press hydraulic system and the operation state of the mechanical components of the hydraulic press to judge whether the target detected hydraulic press needs to be maintained. In this way, it helps to achieve precise maintenance of the hydraulic press, reduce downtime and lower maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings. In the drawings:

[0019] Figure 1 It is a system block diagram of a resin grinding wheel factory production management system according to an embodiment of the present application.

[0020] Figure 2 It is a block diagram of a hydraulic press hydraulic system operation analysis module in a resin grinding wheel factory production management system according to an embodiment of the present application.

[0021] Figure 3It is a block diagram of the operation analysis module of the hydraulic press mechanical components in the resin grinding wheel factory production management system according to an embodiment of the present application.

[0022] Figure 4 It is a block diagram of the maintenance result generation module in the resin grinding wheel factory production management system according to an embodiment of the present application.

[0023] Figure 5 It is a flowchart of the resin grinding wheel factory production management method according to an embodiment of the present application. Detailed implementation manners

[0024] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0025] As mentioned in the above background art, the resin grinding wheel, as a common industrial production consumable, plays an indispensable role in grinding processing, belongs to one of the main types of grinding tools, and has a wide range of demands in the market. In the manufacturing process of the resin grinding wheel, the hydraulic press, as the core production equipment, its operating state is directly related to the quality of the final product and the production efficiency. However, in the traditional production management mode, the maintenance strategy for such key equipment usually relies on a pre-set schedule for regular inspections or only takes measures after the equipment fails. This passive maintenance method often fails to detect potential problems in a timely manner, resulting in an increased risk of unexpected shutdowns and production interruptions, seriously affecting the production plan and delivery time. In addition, irregular or excessive maintenance not only wastes resources and increases unnecessary maintenance costs, but may also damage the equipment itself due to frequent disassembly and reinstallation. Therefore, an optimized resin grinding wheel factory production management solution is expected.

[0026] In recent years, deep learning and neural networks have been widely applied in fields such as computer vision, natural language processing, and text signal processing. In addition, deep learning and neural networks have also shown levels close to or even exceeding those of humans in fields such as image classification, object detection, semantic segmentation, and text translation. The development of deep learning and neural networks provides new solutions and ideas for the production management of resin grinding wheel factories.

[0027] Figure 1 It is a system block diagram of the resin grinding wheel factory production management system according to an embodiment of the present application. As Figure 1As shown in the figure, in the resin grinding wheel factory production management system 100, it includes: a hydraulic press monitoring data acquisition module 110, which is used to obtain the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow rate time series data, and vibration signals of the target monitored hydraulic press measured by the sensor group; a hydraulic press hydraulic system operation analysis module 120, which is used to perform comprehensive time series analysis on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to obtain the operation state feature vector of the hydraulic press hydraulic system; a hydraulic press mechanical component operation analysis module 130, which is used to perform signal analysis on the vibration signals to obtain the operation state feature vector of the hydraulic press mechanical components; a hydraulic press operation state monitoring feature fusion module 140, which is used to fuse the operation state feature vector of the hydraulic press hydraulic system and the operation state feature vector of the hydraulic press mechanical components to obtain the operation state monitoring feature vector of the hydraulic press; a maintenance result generation module 150, which is used to judge whether the target monitored hydraulic press needs maintenance according to the information contained in the operation state monitoring feature vector of the hydraulic press.

[0028] In the embodiment of the present application, the hydraulic press monitoring data acquisition module 110 is used to obtain the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow rate time series data, and vibration signals of the target monitored hydraulic press measured by the sensor group. It should be understood that the hydraulic oil pressure time series data records the pressure change of the hydraulic oil in the hydraulic system during the operation of the hydraulic press. By analyzing this data, the load condition of the hydraulic system, whether there are abnormal pressure fluctuations or continuous high-pressure states can be understood, which may be caused by wear, blockage or other mechanical problems of hydraulic components. The hydraulic oil temperature time series data reflects the temperature change of the hydraulic oil during operation. Too high or too low temperature of the hydraulic oil may affect its performance and the efficiency of the hydraulic system. For example, too high temperature may cause the hydraulic oil to become thinner, reduce the lubrication effect, and accelerate component wear; while too low temperature may increase the viscosity of the hydraulic oil, resulting in problems such as difficult starting. The hydraulic oil flow rate time series data shows the change of the flow rate of the hydraulic oil in the system. The hydraulic oil flow rate is directly related to the working efficiency and response speed of the hydraulic system. Abnormal flow rate changes (such as sudden increase or decrease) may indicate leakage, pump failure or other system problems. The vibration signal provides important information about the operation state of the hydraulic press mechanical components. Any vibration deviating from the normal mode (such as changes in frequency and amplitude) may be an early warning signal of mechanical failures (such as bearing wear, imbalance, etc.). Generally speaking, by real-time monitoring and in-depth analysis of these data, the overall health status of the hydraulic press can be evaluated more accurately, and a decision on whether maintenance is required can be made based on this.

[0029] In an embodiment of the present application, the hydraulic press hydraulic system operation analysis module 120 is configured to perform comprehensive time series analysis on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to obtain the operation state feature vector of the hydraulic press hydraulic system. Specifically, Figure 2 It is a block diagram of the hydraulic press hydraulic system operation analysis module in the resin grinding wheel factory production management system according to an embodiment of the present application. As Figure 2 shown, the hydraulic press hydraulic system operation analysis module 120 includes: a hydraulic press hydraulic system data time series encoding unit 121, configured to perform time series feature extraction on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to obtain a hydraulic oil pressure time series feature vector, a hydraulic oil temperature time series feature vector, and a hydraulic oil flow rate time series feature vector; a hydraulic press hydraulic system data feature integration unit 122, configured to perform two-dimensional arrangement on the hydraulic oil pressure time series feature vector, the hydraulic oil temperature time series feature vector, and the hydraulic oil flow rate time series feature vector to obtain an operation state input matrix of the hydraulic press hydraulic system; a hydraulic press hydraulic system operation state feature generation unit 123, configured to perform operation state feature extraction on the operation state input matrix of the hydraulic press hydraulic system to obtain an operation state feature matrix of the hydraulic press hydraulic system; a hydraulic press hydraulic system operation state feature dimensionality reduction unit 124, configured to expand the operation state feature matrix of the hydraulic press hydraulic system to obtain the operation state feature vector of the hydraulic press hydraulic system.

[0030] In the embodiment of the present application, the hydraulic press hydraulic system data time series encoding unit 121 is used to perform time series feature extraction on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press to obtain a hydraulic oil pressure time series feature vector, a hydraulic oil temperature time series feature vector, and a hydraulic oil flow rate time series feature vector. Specifically, in the embodiment of the present application, the hydraulic press hydraulic system data time series encoding unit is used to: respectively pass the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press through a time series feature analyzer based on a forward LSTM model to obtain the hydraulic oil pressure time series feature vector, the hydraulic oil temperature time series feature vector, and the hydraulic oil flow rate time series feature vector. It should be understood that the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data are all sequence data that change over time. These data contain a large amount of noise and redundant information. Directly using the original data for analysis will increase the computational complexity and it is difficult to capture the key information in the data. In order to remove the noise and redundant information of the original data, better capture the key information in the data, and at the same time deeply explore the laws and patterns hidden in these data to better understand the operating state of the hydraulic press hydraulic system, in the present application, it is necessary to perform time series feature extraction processing on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press. In particular, considering that the forward LSTM (Long Short-Term Memory Network) controls the information flow by introducing a "gate" mechanism, allows the network to learn when to forget past information and when to update new information, and processes data from front to back along the sequence, it can effectively capture the long-term dependencies in the time series data. Based on this, in a specific embodiment of the present application, the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow rate time series data of the target monitored hydraulic press are respectively passed through a time series feature analyzer based on a forward LSTM model to respectively capture the hydraulic oil pressure time series feature, hydraulic oil temperature time series feature, and hydraulic oil flow rate time series feature, thereby generating the corresponding hydraulic oil pressure time series feature vector, hydraulic oil temperature time series feature vector, and hydraulic oil flow rate time series feature vector.

[0031] In the embodiment of the present application, the hydraulic press hydraulic system data feature integration unit 122 is configured to perform two-dimensional arrangement on the hydraulic oil pressure time series feature vector, the hydraulic oil temperature time series feature vector, and the hydraulic oil flow rate time series feature vector to obtain an input matrix of the operating state of the hydraulic press hydraulic system. Correspondingly, considering that the time series feature vectors of the hydraulic oil pressure, temperature, and flow rate each contain information about different aspects of the hydraulic press hydraulic system. In order to integrate this scattered information into a data structure for subsequent overall analysis and processing of the operating state of the hydraulic system, and comprehensively consider the condition of the hydraulic system from multiple dimensions, in the present application, it is necessary to perform two-dimensional arrangement on the hydraulic oil pressure time series feature vector, the hydraulic oil temperature time series feature vector, and the hydraulic oil flow rate time series feature vector. Compared with only analyzing each vector separately to obtain the characteristics and changes of a single factor, after combining these vectors into a two-dimensional matrix, the mutual connection and co-variation relationship between the three factors of hydraulic oil pressure, temperature, and flow rate can be revealed through the structure of the matrix and the relationship between elements, thereby helping to more comprehensively understand the operating state of the hydraulic press hydraulic system. Each row of data in the obtained input matrix of the operating state of the hydraulic press hydraulic system refers to the combination of the characteristic values of the three factors of hydraulic oil pressure, temperature, and flow rate at a specific moment, and each column of data refers to the three different characteristic dimensions of hydraulic oil pressure, temperature, and flow rate.

[0032] In the embodiment of the present application, the hydraulic press hydraulic system operation state feature generation unit 123 is used to extract the operation state feature of the hydraulic press hydraulic system operation state input matrix to obtain the hydraulic press hydraulic system operation state feature matrix. Specifically, in the embodiment of the present application, the hydraulic press hydraulic system operation state feature generation unit is used to: pass the hydraulic press hydraulic system operation state input matrix through the hydraulic press hydraulic system operation state capturer to obtain the hydraulic press hydraulic system operation state feature matrix. It should be understood that although the hydraulic press hydraulic system operation state input matrix has integrated the time series features of hydraulic oil pressure, temperature, and flow, these features may still be at a relatively shallow level. Through the operation state feature extraction, it is intended to dig out deeper, more abstract, and more representative features of the hydraulic system operation state from the input matrix, so as to more accurately describe the actual operation status of the hydraulic system. Specifically, the hydraulic press hydraulic system operation state input matrix is ​​processed through the hydraulic press hydraulic system operation state capturer to generate a hydraulic press hydraulic system operation state feature matrix containing the key operation features of the hydraulic press hydraulic system. In particular, the hydraulic system operating state capturer described in the present application is a convolutional neural network model in which adjacent layers use mutually transposed convolution kernels. This model structure in which adjacent layers use mutually transposed convolution kernels enables it to better capture local feature relationships. Specifically, when processing matrix data, ordinary convolutional neural networks mainly extract local features through convolution kernels. However, the convolutional neural network model in which adjacent layers use mutually transposed convolution kernels can transmit and fuse feature information between different layers in a special way. When processing the input matrix of the hydraulic system operating state of the hydraulic press, this structure can more effectively capture the local relationship between hydraulic oil pressure, temperature, flow and other features at different time steps and different feature dimensions, such as the correlation between the change of pressure in adjacent time steps and the temperature in the same time step, etc., so that the local characteristics of the hydraulic system operating state can be more carefully portrayed. And because this structure can extract the characteristics of the hydraulic system operating state more flexibly and comprehensively, the model can better adapt and learn the hydraulic system data of the hydraulic press under different working conditions and different operating conditions, thereby improving the generalization ability of the model.

[0033] In the embodiment of the present application, the dimensionality reduction unit 124 of the operating state characteristics of the hydraulic press system is configured to expand the matrix of the operating state characteristics of the hydraulic press system to obtain the vector of the operating state characteristics of the hydraulic press system. Accordingly, considering that the data in the form of a two-dimensional matrix of the matrix of the operating state characteristics of the hydraulic press system has a relatively high complexity when performing certain calculations. In order to reduce the computational complexity and facilitate the rapid analysis and processing of the model, in the present application, it is necessary to expand the matrix of the operating state characteristics of the hydraulic press system to obtain the vector of the operating state characteristics of the hydraulic press system. It is worth mentioning that although the form is simplified from a matrix to a vector, through a reasonable expansion method, the key feature information contained in the matrix of the operating state characteristics of the hydraulic press system can be retained. These key information are important representations of the operating state of the hydraulic press system obtained through the previous data processing, which can provide a data basis for accurately judging whether the hydraulic press needs maintenance in the future, and important state information will not be lost due to the change of the data form.

[0034] In the embodiment of the present application, the operating analysis module 130 of the mechanical components of the hydraulic press is configured to perform signal analysis on the vibration signal to obtain the vector of the operating state characteristics of the mechanical components of the hydraulic press. Specifically, Figure 3 It is a block diagram of the operating analysis module of the mechanical components of the hydraulic press in the resin grinding wheel factory production management system according to the embodiment of the present application. As Figure 3 shown, the operating analysis module 130 of the mechanical components of the hydraulic press includes: an operating state characteristic generation unit 131 of the mechanical components of the hydraulic press, configured to obtain the matrix of the operating state characteristics of the mechanical components of the hydraulic press by using a vibration signal analyzer based on a non-overlapping convolution kernel for the waveform diagram of the vibration signal; and a dimensionality reduction unit 132 of the operating state characteristics of the mechanical components of the hydraulic press, configured to expand the matrix of the operating state characteristics of the mechanical components of the hydraulic press to obtain the vector of the operating state characteristics of the mechanical components of the hydraulic press.

[0035] In the embodiment of the present application, the hydraulic press mechanical component operation state feature generation unit 131 is configured to obtain the operation state feature matrix of the hydraulic press mechanical component by passing the waveform diagram of the vibration signal through a vibration signal analyzer based on non-overlapping convolutional kernels. It should be understood that the waveform diagram of the vibration signal reflects the vibration modes of the mechanical components of the hydraulic press during operation, and these modes include frequency component information, amplitude change information, phase relationship, etc. In order to extract this information from the original waveform diagram, in the present application, the waveform diagram of the vibration signal needs to be processed through a vibration signal analyzer based on non-overlapping convolutional kernels to generate an operation state feature matrix of the hydraulic press mechanical component that includes the vibration frequency domain features and time domain features of the hydraulic press mechanical component. In particular, in the present application, the vibration signal analyzer based on non-overlapping convolutional kernels is a convolutional neural network model using non-overlapping convolutional kernels. Specifically, the waveform diagram of the vibration signal contains a large amount of local feature information, such as the vibration peak value at a specific moment, the mutation point of the waveform, etc. The convolutional neural network model using non-overlapping convolutional kernels can effectively extract these local features without redundancy or confusion in feature extraction caused by the overlap of convolutional kernels. And since each convolutional kernel is only responsible for processing a non-overlapping local area, it can extract the unique features of this area more attentively, avoiding interference from the features of adjacent areas, making the extracted local features more obvious and accurate, which is conducive to the subsequent judgment of the operation state of the hydraulic press mechanical component.

[0036] In the embodiment of the present application, the hydraulic press mechanical component operation state feature dimensionality reduction unit 132 is configured to expand the operation state feature matrix of the hydraulic press mechanical component to obtain the operation state feature vector of the hydraulic press mechanical component. It should be understood that the operation state feature matrix of the hydraulic press mechanical component has a relatively high dimension, including information in multiple rows and columns. Expanding it into an operation state feature vector of the hydraulic press mechanical component simplifies the data dimension to one-dimensional in form, making it easier to understand and process, helping to reduce the complexity of data processing, and improving the calculation efficiency and model training speed.

[0037] In an embodiment of the present application, the hydraulic press operation status monitoring feature fusion module 140 is configured to fuse the operation status feature vectors of the hydraulic system of the hydraulic press and the operation status feature vectors of the mechanical components of the hydraulic press to obtain a hydraulic press operation status monitoring feature vector. It should be understood that the hydraulic system and mechanical components of the hydraulic press are interrelated and work together. The hydraulic system provides power and pressure for the mechanical components, and the operation status of the mechanical components also affects the operation of the hydraulic system. The operation status feature vector of the hydraulic system alone can only reflect the hydraulic aspects, such as parameters such as the pressure, temperature, and flow rate of the hydraulic oil; the operation status feature vector of the mechanical components alone mainly focuses on the vibration and other conditions of the mechanical components. Only by fusing the two can the overall operation status of the hydraulic press be comprehensively and completely reflected from multiple dimensions. That is, by fusing these two feature vectors, the operation status of the hydraulic press can be monitored more comprehensively, and potential problems that may be overlooked when analyzing the hydraulic system or mechanical components separately can be detected in a timely manner. For example, when there are abnormal fluctuations in the pressure of the hydraulic system and the vibration of the mechanical components also exceeds the normal range, the fused feature vector can more accurately reflect that this may be a serious fault hidden danger that requires timely maintenance, rather than making misjudgments or omissions based on a single feature vector.

[0038] In an embodiment of the present application, the maintenance result generation module 150 is configured to determine whether the target monitored hydraulic press needs maintenance according to the information included in the hydraulic press operation status monitoring feature vector. Specifically, Figure 4 The block diagram of the maintenance result generation module in the resin grinding wheel factory production management system according to the embodiment of the present application is as follows. As Figure 4 shown, the maintenance result generation module 150 includes: a hydraulic press operation status monitoring feature optimization unit 151, configured to perform feature adaptive selection and adjustment based on hidden space mapping on the hydraulic press operation status monitoring feature vector to obtain an optimized hydraulic press operation status monitoring feature vector; a hydraulic press operation status monitoring optimized feature analysis unit 152, configured to pass the optimized hydraulic press operation status monitoring feature vector through a maintenance requirement classifier to obtain a maintenance classification result, and the maintenance classification result is used to indicate whether the target monitored hydraulic press needs maintenance.

[0039] In the embodiments of the present application, the hydraulic press operation state monitoring feature optimization unit 151 is used to perform feature adaptive selection and adjustment based on hidden space mapping on the hydraulic press operation state monitoring feature vector to obtain an optimized hydraulic press operation state monitoring feature vector. In particular, considering that the operation state of the hydraulic press is affected by various factors, such as different workloads, ambient temperatures, hydraulic oil quality, etc. If the training data fails to cover these diverse scenarios, the generalization ability of the model will be limited when facing new operating conditions. At the same time, the model mainly relies on the statistical characteristics of the data itself for learning and fails to actively incorporate the experience and prior knowledge of domain experts, which may lead to misjudgments of the model when facing certain complex hydraulic press states, and thus the maintenance decision-making is not accurate enough. Based on this, the present application needs to perform feature adaptive selection and adjustment based on hidden space mapping on the hydraulic press operation state monitoring feature vector to obtain an optimized hydraulic press operation state monitoring feature vector.

[0040] Specifically, in the embodiments of the present application, the hydraulic press operation state monitoring feature optimization unit is used to: extract the hydraulic press operation prior knowledge matrix; perform key feature distillation on the hydraulic press operation prior knowledge matrix to obtain a set of hydraulic press operation core prior information coding vectors, and this process can be expressed by the formula:

[0041]

[0042] where U represents the set of hydraulic press operation core prior information coding vectors, v1, v2, v m respectively represent the first, second, and m-th hydraulic press operation core prior information coding vectors, T represents the transpose operation, CoreExtraction represents key feature distillation, M p represents the hydraulic press operation prior knowledge matrix, Λ represents the hydraulic press operation diagonal matrix, and λ1, λ m respectively represent the values at the first and m-th positions on the diagonal of the hydraulic press operation diagonal matrix;

[0043] Construct a hydraulic press operation prior information response coding hidden space matrix between the hydraulic press operation state monitoring feature vector and each hydraulic press operation core prior information coding vector in the set of hydraulic press operation core prior information coding vectors to obtain a set of hydraulic press operation prior information response coding hidden space matrices, and this process can be expressed by the formula:

[0044]

[0045] where x o represents the hydraulic press operation state monitoring feature vector, l i (x o ) represents x oPerform a linear transformation. The eigenvectors after the linear transformation have the same eigen-scale as the corresponding prior information coding vectors of the hydraulic press operation core, v i represents the i-th prior information coding vector of the hydraulic press operation core, represents matrix multiplication, L represents the length of the prior information coding vector of the hydraulic press operation core, MR i represents the i-th prior information response coding hidden space matrix of the hydraulic press operation;

[0046] Calculate the prior information response significant descriptors of each prior information response coding hidden space matrix in the set of prior information response coding hidden space matrices of the hydraulic press operation to obtain a set of prior information response significant descriptors of the hydraulic press operation. This process can be expressed by the formula:

[0047] S i = ||MR i || F

[0048] where ||·|| F represents the Frobenius norm of the matrix, S i represents the i-th prior information response significant descriptor of the hydraulic press operation;

[0049] Based on the set of prior information response significant descriptors of the hydraulic press operation, perform non-uniform projection integration on the set of prior information response coding hidden space matrices of the hydraulic press operation to obtain a prior information response projection coding matrix of the hydraulic press operation. This process can be expressed by the formula:

[0050]

[0051] where softmax represents the normalized exponential function, and P represents the prior information response projection coding matrix of the hydraulic press operation;

[0052] Map the hydraulic press operation state monitoring eigenvector to the feature space of the prior information response projection coding matrix of the hydraulic press operation to obtain the optimized hydraulic press operation state monitoring eigenvector. This process can be expressed by the formula:

[0053]

[0054] where x opt represents the optimized hydraulic press operation state monitoring eigenvector.

[0055] To address the above technical problems, in the technical solution of this application, the feature vector for monitoring the operating state of the hydraulic press is adaptively selected and adjusted based on latent space mapping to obtain an optimized feature vector for monitoring the operating state of the hydraulic press. This process begins with the extraction of the prior knowledge matrix of the hydraulic press operation. Specifically, the experience accumulated by domain experts in the long-term maintenance and operation of the hydraulic press, such as the relationship between the wear of each component of the hydraulic press and the operating duration under specific production processes, is transformed into a prior knowledge matrix. This enables the model to no longer be limited to the statistical features on the surface of the data and can deeply understand the complex causal relationships behind the operation of the hydraulic press. When judging whether the hydraulic press needs maintenance, by combining this prior knowledge, the model can make more informed and practical decisions, reducing misjudgments caused by simply relying on data statistics.

[0056] Next, key feature distillation needs to be performed on the prior knowledge matrix of the hydraulic press operation to obtain a set of core prior information coding vectors for the hydraulic press operation. It should be understood that during the construction of the prior knowledge matrix of the hydraulic press operation, due to the possible inclusion of a large amount of information from various sources, including the summary of experiences under various working conditions and the correlations between different monitoring indicators, the dimension may be relatively high and redundant information may be included. Performing key feature distillation can remove those redundant parts that have little effect on judging the operating state of the hydraulic press. For example, some weakly correlated information that only appears under specific rare working conditions has no practical value in most normal operating scenarios. After removing it, the matrix dimensions involved in the subsequent calculation process are significantly reduced, and both the computing resource occupancy and computing time can be significantly optimized, making the entire analysis process more efficient. Moreover, the existence of redundant information may cause the model to show large fluctuations when facing different datasets or minor data changes. For example, when training the model, if there is a large amount of redundant information in the prior knowledge matrix, when there are some small fluctuations or outliers in the training data, the model may be interfered by this redundant information and produce a large deviation. After key feature distillation, the information relied on by the model is more stable and crucial, and it has stronger anti-interference ability against small fluctuations and outliers in the data. Even when facing different batches of monitoring data, the model can give relatively consistent and accurate judgments based on the set of core prior information coding vectors for the hydraulic press operation obtained after processing, thereby enhancing the stability of the model.

[0057] Then, construct the prior information response encoding hidden space matrix of the hydraulic press operation state monitoring feature vector and each prior information encoding vector of the core prior information of the hydraulic press operation in the set to obtain a set of prior information response encoding hidden space matrices of the hydraulic press operation. It should be understood that by constructing the hidden space matrix between the hydraulic press operation state monitoring feature vector and the prior information encoding vector of the core prior information of the hydraulic press operation, the feature information actually monitored during the operation of the hydraulic press (such as feature vectors such as hydraulic oil pressure, temperature, flow rate, etc.) can be deeply integrated with the core information extracted from prior knowledge (such as domain expert experience). For example, when an abnormal fluctuation in the hydraulic oil pressure of the hydraulic press is detected, combined with the core information about the normal fluctuation range of the hydraulic oil pressure under different working loads in the prior knowledge, through the mapping of the hidden space matrix, it can be more accurately judged whether the fluctuation is an abnormal situation, so that the model no longer relies solely on a single monitoring data feature, but makes a judgment by integrating prior knowledge, improving the depth and accuracy of the understanding of the hydraulic press operation state, and facilitating better maintenance decision-making. Moreover, in the actual operation of the hydraulic press, the relationships between various features and between features and prior knowledge are often non-linear. Hidden space mapping can learn the non-linear response patterns of the hydraulic press operation state monitoring feature vector to prior knowledge in different aspects. For example, the temperature change of the hydraulic oil may not only be related to the ambient temperature, but also have complex non-linear relationships with various factors such as the working duration and load size of the hydraulic press, and these relationships may be contained in the prior knowledge. By constructing the hidden space matrix, these non-linear relationships can be mined, enabling the model to more comprehensively capture the change laws of the hydraulic press operation state, and thus more accurately predict the possible problems of the hydraulic press and make preparations for maintenance in advance.

[0058] Next, calculate the prior information response significant descriptors of each prior information response encoding hidden space matrix in the set of prior information response encoding hidden space matrices of the hydraulic press operation to obtain a set of prior information response significant descriptors of the hydraulic press operation. It should be understood that in the prior information response encoding hidden space matrix of the hydraulic press operation, there may be a large amount of complex and interrelated information. By calculating the prior information response significant descriptors, the key information can be accurately extracted from these matrices. And the calculation process of the significant descriptors realizes the feature representation reduction of the hidden space matrix, achieving the effect of data dimensionality reduction, which is beneficial to reducing the complexity of subsequent calculations and improving the calculation efficiency.

[0059] Then, based on the set of significant descriptors of the prior information response of the hydraulic press operation, a non-uniform projection integration is performed on the set of prior information response coding latent space matrices of the hydraulic press operation to obtain a prior information response projection coding matrix of the hydraulic press operation. It should be understood that the set of prior information response coding latent space matrices of the hydraulic press operation contains descriptive information on the operation state of the hydraulic press from different prior knowledge perspectives. Through non-uniform projection integration, these multi-angle information can be organically integrated. For example, some latent space matrices reflect the operation characteristics of the hydraulic system, some reflect the vibration conditions of mechanical components, and some reflect the information in terms of electrical control. Non-uniform projection integration can, according to the significant descriptors of the prior information response of the hydraulic press operation, targetedly fuse this information from different aspects to form a comprehensive prior information response projection coding matrix of the hydraulic press operation that comprehensively reflects the operation state of the hydraulic press, avoiding the isolation and one-sidedness of information. It should be particularly noted that due to the existence of sparsity, non-uniform projection integration does not treat all prior information responses equally. It can judge which prior information responses are important and which are relatively unimportant according to the significant descriptors. For important prior information responses, higher weights are given for integration, while for unimportant responses, they are weakened or ignored. This can make the features contained in the finally obtained projection coding matrix more representative and effective, thereby enhancing the feature selection ability and enabling the model to better focus on key features for analysis and judgment.

[0060] Finally, the hydraulic press operation state monitoring feature vector is mapped to the feature space of the prior information response projection coding matrix of the hydraulic press operation to obtain the optimized hydraulic press operation state monitoring feature vector. It should be understood that the prior information response projection coding matrix of the hydraulic press operation is obtained through effective integration of multi-angle prior information, feature selection, etc., and the feature space defined by it contains rich prior knowledge. After mapping the hydraulic press operation state monitoring feature vector to this feature space, the initial feature vector can absorb the information of prior knowledge, thereby optimizing its expression ability. Specifically, in the original feature space, the feature vectors of the hydraulic press under different operation states may not be clearly distinguishable. However, through mapping to the feature space of the projection coding matrix, the optimized feature vector can better reflect the differences between different operation states. For example, for the normal operation and slight wear states of the hydraulic press, the initial hydraulic oil pressure features may have small differences and it is difficult to accurately distinguish them. But in the new feature space, since the projection coding matrix incorporates prior knowledge, such as the change rules of hydraulic oil pressure under different wear degrees, etc., the optimized feature vector can more clearly distinguish these two states, helping to more accurately judge the operation state of the hydraulic press and providing a more reliable basis for maintenance decisions.

[0061] In the embodiment of the present application, the optimization feature analysis unit 152 for monitoring the operating state of the hydraulic press is configured to obtain a maintenance classification result by passing the optimized feature vector for monitoring the operating state of the hydraulic press through a maintenance requirement classifier. The maintenance classification result is used to indicate whether the target monitored hydraulic press needs to be maintained. It should be understood that the optimized feature vector for monitoring the operating state of the hydraulic press contains a large amount of information about the operation of the hydraulic press, but this information is relatively complex and diverse, and it is difficult to directly and intuitively determine whether the hydraulic press needs maintenance based on these feature vectors. Through the maintenance requirement classifier, these complex feature vectors can be converted into clear and quantifiable judgment results, that is, whether maintenance is required, providing a clear and accurate basis for maintenance decisions. Specifically, the maintenance requirement classifier is essentially a model constructed based on deep learning. Before use, it needs to be trained with a large amount of labeled operating data of the hydraulic press, and these labeled data include the actual situation of whether the hydraulic press needs maintenance under different operating states. By learning these data, the maintenance requirement classifier can establish a mapping relationship between the optimized feature vector for monitoring the operating state of the hydraulic press and the maintenance requirement. Based on the learned mapping relationship and the analysis of the newly input features, a final maintenance classification decision is made. The maintenance classification result can directly inform the staff whether the target monitored hydraulic press needs to be maintained. According to this result, the staff can reasonably arrange the maintenance plan, promptly arrange maintenance personnel and resources for the equipment that needs immediate maintenance, and let the equipment that does not need maintenance for the time being continue normal production, which is conducive to improving the timeliness and pertinence of maintenance work.

[0062] In summary, the resin grinding wheel factory production management system 100 based on the embodiment of the present application is clarified. It analyzes the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow time series data of the target monitored hydraulic press to understand the operating state of the hydraulic system of the hydraulic press, and at the same time analyzes the vibration signal of the target monitored hydraulic press to understand the operating state of the mechanical components of the hydraulic press. By comprehensively considering the operating state of the hydraulic system of the hydraulic press and the operating state of the mechanical components of the hydraulic press, it is determined whether the target detected hydraulic press needs to be maintained. In this way, it helps to achieve precise maintenance of the hydraulic press, reduce downtime and lower maintenance costs.

[0063] Figure 5 For the flowchart of the resin grinding wheel factory production management method according to the embodiment of the present application. As Figure 5As shown, in the production management method of a resin grinding wheel factory, it includes: S110, obtaining the hydraulic oil pressure time-series data, hydraulic oil temperature time-series data, hydraulic oil flow time-series data, and vibration signals measured by a sensor group of a target monitoring hydraulic press; S120, performing comprehensive time-series analysis on the hydraulic oil pressure time-series data, hydraulic oil temperature time-series data, and hydraulic oil flow time-series data of the target monitoring hydraulic press to obtain the operation state feature vector of the hydraulic press hydraulic system; S130, performing signal analysis on the vibration signals to obtain the operation state feature vector of the mechanical components of the hydraulic press; S140, fusing the operation state feature vector of the hydraulic press hydraulic system and the operation state feature vector of the mechanical components of the hydraulic press to obtain the operation state monitoring feature vector of the hydraulic press; S150, judging whether the target monitoring hydraulic press needs maintenance according to the information contained in the operation state monitoring feature vector of the hydraulic press.

[0064] Here, those skilled in the art can understand that the specific operations of each step in the above production management method of the resin grinding wheel factory have been introduced in detail in the description of the resin grinding wheel factory production management system above with reference to Figures 1 to 4 and thus, the repeated description thereof will be omitted.

[0065] In summary, the production management method of the resin grinding wheel factory based on the embodiments of the present application is clarified. It understands the operation state of the hydraulic press hydraulic system by analyzing the hydraulic oil pressure time-series data, hydraulic oil temperature time-series data, and hydraulic oil flow time-series data of the target monitoring hydraulic press, and at the same time understands the operation state of the mechanical components of the hydraulic press by analyzing the vibration signals of the target monitoring hydraulic press. Considering the operation state of the hydraulic press hydraulic system and the operation state of the mechanical components of the hydraulic press comprehensively to judge whether the target detection hydraulic press needs maintenance. In this way, it helps to achieve precise maintenance of the hydraulic press, reduce downtime and lower maintenance costs.

[0066] The basic principles of the present application have been described above in combination with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present application are only examples and not limitations. It cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present application. In addition, the specific details of the above application are only for the purpose of illustration and easy understanding, rather than limitations. The above details do not limit the present application to necessarily adopt the above specific details to implement.

Claims

1. A resin grinding wheel factory production management system, characterized in that: include: The hydraulic press monitoring data acquisition module is used to obtain the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow time series data and vibration signal of the target monitored hydraulic press measured by the sensor group; The hydraulic system operation analysis module of the hydraulic press is used to perform a comprehensive time series analysis on the hydraulic oil pressure time series data, the hydraulic oil temperature time series data, and the hydraulic oil flow time series data of the target monitored hydraulic press to obtain a characteristic vector of the hydraulic system operation state of the hydraulic press; A hydraulic press mechanical component operation analysis module, used for performing signal analysis on the vibration signal to obtain an operating state feature vector of the hydraulic press mechanical component; A hydraulic press operation status monitoring feature fusion module, used to fuse the hydraulic press hydraulic system operation status feature vector and the hydraulic press mechanical component operation status feature vector to obtain the hydraulic press operation status monitoring feature vector; The maintenance result generating module is used to determine whether the target monitored hydraulic press needs maintenance according to the information contained in the hydraulic press operation status monitoring feature vector.

2. The resin grinding wheel factory production management system according to claim 1, characterized in that: The oil press hydraulic system operation analysis module includes: A hydraulic press hydraulic system data timing encoding unit, used for performing timing feature extraction on the hydraulic oil pressure timing data, hydraulic oil temperature timing data, and hydraulic oil flow timing data of the target monitored hydraulic press to obtain a hydraulic oil pressure timing feature vector, a hydraulic oil temperature timing feature vector, and a hydraulic oil flow timing feature vector; A hydraulic press hydraulic system data feature integration unit, used for two-dimensionally arranging the hydraulic oil pressure time series feature vector, the hydraulic oil temperature time series feature vector and the hydraulic oil flow time series feature vector to obtain an input matrix of the hydraulic press hydraulic system operation state; An oil press hydraulic system operating state feature generating unit, used for extracting operating state features from the oil press hydraulic system operating state input matrix to obtain an oil press hydraulic system operating state feature matrix; The hydraulic press hydraulic system operating state feature dimension reduction unit is used to expand the hydraulic press hydraulic system operating state feature matrix to obtain the hydraulic press hydraulic system operating state feature vector.

3. The resin grinding wheel factory production management system according to claim 2, characterized in that: The hydraulic press hydraulic system data timing encoding unit is used to: respectively pass the hydraulic oil pressure timing data, hydraulic oil temperature timing data, and hydraulic oil flow timing data of the target monitored hydraulic press through a timing feature analyzer based on a forward LSTM model to obtain the hydraulic oil pressure timing feature vector, the hydraulic oil temperature timing feature vector, and the hydraulic oil flow timing feature vector.

4. The resin grinding wheel factory production management system according to claim 3, characterized in that: The oil press hydraulic system operating state feature generating unit is used to: pass the oil press hydraulic system operating state input matrix through the oil press hydraulic system operating state capturer to obtain the oil press hydraulic system operating state feature matrix.

5. The resin grinding wheel factory production management system according to claim 4, characterized in that: The hydraulic press mechanical parts operation analysis module includes: A hydraulic press mechanical component operation state feature generation unit, used for passing the waveform of the vibration signal through a vibration signal analyzer based on a non-overlapping convolution kernel to obtain an hydraulic press mechanical component operation state feature matrix; The hydraulic press mechanical component operation state feature dimension reduction unit is used to expand the hydraulic press mechanical component operation state feature matrix to obtain the hydraulic press mechanical component operation state feature vector.

6. The resin grinding wheel factory production management system according to claim 5, characterized in that: The hydraulic press hydraulic system operation status capturer is a convolutional neural network model in which adjacent layers use mutually transposed convolution kernels, and the vibration signal analyzer based on non-overlapping convolution kernels is a convolutional neural network model using non-overlapping convolution kernels.

7. The resin grinding wheel factory production management system according to claim 6, characterized in that: The maintenance result generating module comprises: A hydraulic press operation status monitoring feature optimization unit, used for performing feature adaptive selection and adjustment on the hydraulic press operation status monitoring feature vector based on latent space mapping to obtain an optimized hydraulic press operation status monitoring feature vector; The optimized feature analysis unit for monitoring the running state of the hydraulic press is used to pass the optimized feature vector for monitoring the running state of the hydraulic press through a maintenance requirement classifier to obtain a maintenance classification result, wherein the maintenance classification result is used to indicate whether the target monitored hydraulic press needs maintenance.

8. The resin grinding wheel factory production management system according to claim 7, characterized in that: The hydraulic press operation status monitoring feature optimization unit is used to: Extract the prior knowledge matrix of hydraulic press operation; Performing key feature distillation on the hydraulic press operation prior knowledge matrix to obtain a set of hydraulic press operation core prior information encoding vectors; Constructing a hydraulic press operation prior information response coding latent space matrix between the hydraulic press operation state monitoring feature vector and each hydraulic press operation core prior information coding vector in the set of the hydraulic press operation core prior information coding vector to obtain a set of hydraulic press operation prior information response coding latent space matrices; Calculating the significant descriptors of the prior information response of the hydraulic press operation of each hydraulic press operation prior information response encoding latent space matrix in the set of the hydraulic press operation prior information response encoding latent space matrix to obtain a set of significant descriptors of the prior information response of the hydraulic press operation; Based on the set of significant descriptors of the hydraulic press operation prior information response, a set of the hydraulic press operation prior information response encoding latent space matrices is subjected to non-uniform projection integration to obtain a hydraulic press operation prior information response projection encoding matrix; The hydraulic press operation status monitoring feature vector is mapped to the feature space of the hydraulic press operation prior information response projection coding matrix to obtain the optimized hydraulic press operation status monitoring feature vector.

9. A resin grinding wheel factory production management method, characterized in that: include: Acquire the hydraulic oil pressure time series data, hydraulic oil temperature time series data, hydraulic oil flow time series data and vibration signal of the target monitoring hydraulic press measured by the sensor group; Performing a comprehensive time series analysis on the hydraulic oil pressure time series data, hydraulic oil temperature time series data, and hydraulic oil flow time series data of the target monitored hydraulic press to obtain a characteristic vector of the hydraulic system operation state of the hydraulic press; Performing signal analysis on the vibration signal to obtain a characteristic vector of the operating state of the mechanical component of the hydraulic press; Fusion of the hydraulic system operating state feature vector of the hydraulic press and the mechanical component operating state feature vector of the hydraulic press to obtain an operating state monitoring feature vector of the hydraulic press; According to the information contained in the hydraulic press operation status monitoring feature vector, it is determined whether the target monitored hydraulic press needs maintenance.

10. The resin grinding wheel factory production management method according to claim 9, characterized in that: Judging whether the target monitored hydraulic press needs maintenance according to the information contained in the hydraulic press operation status monitoring feature vector includes: Performing feature adaptive selection adjustment based on latent space mapping on the hydraulic press operation status monitoring feature vector to obtain an optimized hydraulic press operation status monitoring feature vector; The optimized hydraulic press operation status monitoring feature vector is passed through a maintenance requirement classifier to obtain a maintenance classification result, and the maintenance classification result is used to indicate whether the target monitored hydraulic press needs maintenance.

Images