Bulk cargo terminal operation and maintenance situation awareness method and system

Abstract

The present application belongs to bulk cargo terminal operation situation awareness technical field, and relates to a bulk cargo terminal operation situation awareness method and system, which uses the equipment influence factor of different equipment in a set time period to predict the equipment health index through the health state analysis model, calculates the equipment residual life through the residual life probability distribution prediction model, and predicts the equipment failure rate through the situation awareness model according to the equipment health index and the equipment residual life, so as to obtain the equipment alarm signal of different levels, generate and execute the auxiliary management scheme according to the equipment alarm signal, reasonably plan the equipment maintenance time, and improve the equipment utilization and the on-site production efficiency. The present application realizes the health state monitoring and operation situation awareness of the bulk cargo terminal, improves the equipment utilization and the on-site production efficiency, and improves the turnover rate and the turnover efficiency of the terminal cargo.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of bulk cargo terminal operation and maintenance management, and relates to a bulk cargo terminal operation and maintenance situation awareness method and system. BACKGROUND

[0002] With the continuous expansion of modern bulk cargo terminals, the continuous improvement of informatization level, and the continuous increase of operation and maintenance business, the operation safety, operation and maintenance efficiency, and production continuity of bulk cargo terminal equipment have become an important issue. The common operation mode of bulk cargo terminals is linear, and the operation process mainly focuses on loading and unloading equipment (for example: car dumper, stacker, loader, ship transfer machine, etc.) and conveying equipment (for example: belt conveyor, etc.). The traditional on-line monitoring method of equipment can only evaluate the health degree according to the monitoring state of the equipment at that time, and cannot evaluate the health degree according to the self-wear of the equipment, resulting in unexpected equipment failure, inability to coordinate production and operation work, and reduction of port efficiency. At the same time, when the operation personnel selects the operation process line, the health degree of the process cannot be considered, and only the preset process can be started fixedly, which will cause the whole production line to be unable to proceed when a fault occurs at a key node, greatly affecting the operation efficiency. SUMMARY

[0003] The present application provides a bulk cargo terminal operation and maintenance situation awareness method and system to solve the above problems such as low operation efficiency in the prior art. The device influence factor of different devices in a set time period is obtained, the device health degree index is predicted by a health state analysis model constructed, the probability distribution of the device residual life is predicted by a residual life probability distribution prediction model constructed, the risk failure time is determined according to the probability distribution of the device residual life, the device residual life is calculated according to the risk failure time and the current time, the device failure rate is predicted according to the device health degree index and the device residual life through a constructed situation awareness model, and then different levels of device alarm signals are obtained. According to the device alarm signal, an auxiliary management scheme is generated and executed, the device maintenance time is reasonably planned, the device utilization rate and on-site production efficiency are improved, and the turnover rate and turnover efficiency of bulk cargo terminal goods are improved.

[0004] In a first aspect, the present application provides a bulk cargo terminal operation and maintenance situation awareness method, comprising:

[0005] Data acquisition step: acquiring device influence factors of different devices in a set time period, wherein the device influence factors include device environment parameters and device state parameters;

[0006] Health degree prediction step: evaluating the device health degree index according to the device influence factors of the set time period through a pre-constructed health state analysis model to obtain the device health degree index at different times;

[0007] The remaining life prediction step: predicting the probability distribution of the remaining life of the equipment according to the equipment influence factor of the set time period through the pre-constructed remaining life probability distribution prediction model, determining the risk failure time according to the probability distribution of the remaining life of the equipment, and calculating the remaining life of the equipment according to the risk failure time and the current time;

[0008] The situation prediction step: predicting the equipment failure rate according to the equipment health index and the equipment remaining life through the pre-constructed situation awareness model;

[0009] The evaluation step: generating different levels of equipment alarm signals according to the equipment failure rate;

[0010] The equipment management step: generating an auxiliary management scheme according to the equipment alarm signal, and controlling the equipment to be managed by executing the auxiliary management scheme.

[0011] In some embodiments, the construction method of the health state analysis model is:

[0012] Obtain the historical equipment influence factor and the historical equipment health index of different equipment to form a data set, and divide the data set into a training set and a validation set;

[0013] Based on the regression model, the health state analysis model is constructed with the equipment influence factor as the input and the equipment health index as the output. The health state analysis model is represented as:

[0014]

[0015] In the formula, represents the equipment health index, represents the equipment factory health index, is the standardized value of the nth equipment influence factor index, is the weight of the nth equipment influence factor;

[0016] The health state analysis model is trained and verified through the training set and the validation set.

[0017] In some embodiments, when training the health state analysis model, the weight region is trained through the training set, and the health state analysis model weight is updated by using a statistical method.

[0018] In some embodiments, the construction method of the remaining life probability distribution prediction model is:

[0019] Obtain the historical equipment influence factor and the historical equipment remaining life probability distribution of different equipment to form a data set, and divide the data set into a training set and a validation set;

[0020] A remaining useful life probability distribution prediction model is constructed based on a long short-term memory network (LSTM), with a device impact factor as input and a device remaining useful life probability distribution as output, and is expressed as:

[0021]

[0022] wherein, is a risk prediction function according to a risk failure time t, and represents an instantaneous failure risk of a covariate ; is a baseline risk function, and represents a natural aging trend of a device; is a covariate coefficient, and represents an impact weight of a device impact factor on a failure risk; is a covariate of a device impact factor; is a number of device impact factors;

[0023] The remaining useful life probability distribution prediction model is trained and verified through a training set and a verification set.

[0024] In some embodiments, in the situation prediction step, a method for predicting a device failure rate according to a device health index and a device remaining useful life through a pre-constructed situation awareness model is as follows:

[0025] A device health index mean and a device health index standard deviation within a set time period are calculated according to health indexes at different time points;

[0026] A normalized device remaining useful life is obtained by normalizing a device remaining useful life;

[0027] A device failure rate is predicted through the situation awareness model according to the device health index mean, the device health index standard deviation, and the normalized device remaining useful life.

[0028] In some embodiments, a method for constructing the situation awareness model is as follows:

[0029] Historical device health indexes, historical device remaining useful lives, and historical device failure rates of different devices are obtained;

[0030] A historical device health index mean and a historical device health index standard deviation are calculated according to historical device health indexes, and a normalized historical device remaining useful life is obtained by normalizing a historical device remaining useful life;

[0031] The normalized historical device remaining useful life, the historical device health index mean, the historical device health index standard deviation, and the historical device failure rate are formed into a data set, and are divided into a training set and a verification set;

[0032] Based on a long short-term memory network LSTM, a situational awareness model is constructed with the mean of the equipment health index, the standard deviation of the equipment health index and the normalized equipment remaining life as input and the equipment failure rate as output, and the situational awareness model is represented as:

[0033]

[0034] In the formula, is the equipment failure rate, is the mean of the equipment health index, is a sensitivity constant, is the standard deviation of the equipment health index, is the normalized equipment remaining life.

[0035] The situational awareness model is trained and verified through the training set and the verification set.

[0036] In a second aspect, the present application provides a bulk cargo terminal operation and maintenance situational awareness system for implementing the bulk cargo terminal operation and maintenance situational awareness method of the first aspect of the present application, comprising:

[0037] A data acquisition module is configured to acquire equipment influence factors of different equipment in a set time period, and historical equipment influence factors, historical equipment health indexes, historical equipment remaining life probability distributions, historical equipment remaining life and historical failure rates of different equipment.

[0038] A model generation module is configured to generate a health state analysis model according to the historical equipment influence factors and the historical equipment health indexes of different equipment, generate a remaining life probability distribution prediction model according to the historical equipment influence factors and the historical equipment remaining life probability distributions of different equipment, and generate a situational awareness model according to the historical equipment health indexes, the historical equipment remaining life and the historical failure rates of different equipment.

[0039] A prediction module is configured to predict the equipment health index and the probability distribution of the equipment remaining life through the health state analysis model and the remaining life probability distribution prediction model respectively according to the equipment influence factors of different equipment in the set time period, determine the risk failure time according to the probability distribution of the equipment remaining life, calculate the equipment remaining life according to the risk failure time and the current time, and predict the equipment failure rate through the situational awareness model according to the equipment health index and the equipment remaining life.

[0040] An evaluation module is configured to generate equipment alarm signals of different levels according to the equipment failure rate.

[0041] A device management module is configured to generate an auxiliary management scheme according to the equipment alarm signal, and control the equipment to be managed by executing the auxiliary management scheme.

[0042] In some embodiments, the model generation module comprises:

[0043] a first model generation module configured to generate a health state analysis model according to historical device impact factors and historical device health indexes of different devices;

[0044] a second model generation module configured to generate a residual life probability distribution prediction model according to historical device impact factors and historical residual life probability distributions of different devices;

[0045] a third model generation module configured to generate a situational awareness model according to historical device health indexes, historical residual life and historical failure rates of different devices.

[0046] In some embodiments, the first model generation module comprises:

[0047] a first data set generation submodule configured to form a first data set according to historical device impact factors and historical device health indexes of different devices, and divide the first data set into a first training set and a first validation set;

[0048] a first model construction submodule configured to construct a health state analysis model based on a regression model, with device impact factors as input and device health indexes as output;

[0049] a first model training submodule configured to train the health state analysis model according to the first training set;

[0050] a first model verification submodule configured to verify the performance of the health state analysis model according to the first validation set.

[0051] In some embodiments, the second model generation module comprises:

[0052] a second data set generation submodule configured to form a second data set according to historical device impact factors and historical residual life probability distributions of different devices, and divide the second data set into a second training set and a second validation set;

[0053] a second model construction submodule configured to construct a residual life probability distribution prediction model based on a long short-term memory network (LSTM), with device impact factors as input and device residual life probability distributions as output;

[0054] a second model training submodule configured to train the residual life probability distribution prediction model according to the second training set;

[0055] a second model verification submodule configured to verify the performance of the residual life probability distribution prediction model according to the second validation set.

[0056] In some embodiments, the third model generation module comprises:

[0057] The first calculation submodule is configured to calculate a historical equipment health index mean and a historical equipment health index standard deviation according to a historical equipment health index;

[0058] The normalization submodule is configured to normalize the historical equipment residual life to obtain normalized historical equipment residual life;

[0059] The third data set generation submodule is configured to form a third data set by combining the normalized historical equipment residual life, the historical equipment health index mean, the historical equipment health index standard deviation, and the historical equipment failure rate, and divide the third data set into a third training set and a third validation set;

[0060] The third model construction submodule is configured to construct a situational awareness model based on a long short-term memory network (LSTM), with the equipment health index mean, the equipment health index standard deviation, and the normalized equipment residual life as inputs and the equipment failure rate as output;

[0061] The third model training submodule is configured to train the situational awareness model according to the third training set;

[0062] The third model verification submodule is configured to verify the performance of the situational awareness model according to the third validation set.

[0063] In some embodiments, the prediction module comprises:

[0064] The health state prediction submodule is configured to predict the equipment health index of different equipment in a set time period by a health state analysis model according to the equipment influence factor of the set time period;

[0065] The probability distribution prediction submodule is configured to predict the probability distribution of the equipment residual life by a residual life probability distribution prediction model according to the equipment influence factor of the set time period;

[0066] The second calculation submodule is configured to determine a risk failure time according to the probability distribution of the equipment residual life, and calculate the equipment residual life according to the risk failure time and the current time;

[0067] The situation prediction submodule is configured to predict the equipment failure rate by the situational awareness model according to the equipment health index and the equipment residual life.

[0068] Compared with the related art, the bulk cargo terminal operation situation perception method and system provided by the present application utilize the constructed health state analysis model and residual life probability distribution prediction model, predict the equipment health degree index and the probability distribution of the equipment residual life according to the obtained equipment influence factor in a set time period, determine the risk failure time according to the probability distribution of the equipment residual life, calculate the equipment residual life according to the risk failure time and the current time, predict the equipment failure rate according to the equipment health degree index and the equipment residual life through the constructed situation perception model, generate equipment alarm signals of different levels according to the equipment failure rate, generate an auxiliary management scheme according to the equipment alarm signals, and control the equipment to be managed by the auxiliary management scheme. The present application realizes the perception of the utilization rate, health state and operation situation of the bulk cargo terminal equipment within a certain time (which can be preset according to actual needs), uniformly plans the bulk cargo terminal equipment according to the operation plan, the production plan, the equipment health state and the operation situation, and improves the cargo turnover rate and turnover efficiency of the bulk cargo terminal. BRIEF DESCRIPTION OF DRAWINGS

[0069] The accompanying drawings, which are included to provide a further understanding of the present application and constitute a part of this application, illustrate certain illustrative embodiments of the present application and together with the description serve to explain the present application. In the drawings:

[0070] Figure 1 A flowchart of the bulk cargo terminal operation situation perception method according to the embodiment of the present application is shown in the figure;

[0071] Figure 2 A flowchart of the construction method of the health state analysis model according to the embodiment of the present application is shown in the figure;

[0072] Figure 3 A flowchart of the construction method of the residual life probability distribution prediction model according to the embodiment of the present application is shown in the figure;

[0073] Figure 4 A flowchart of the method of predicting the equipment failure rate according to the equipment health degree index and the equipment residual life through the pre-constructed situation perception model according to the embodiment of the present application is shown in the figure;

[0074] Figure 5 A flowchart of the construction method of the situation perception model according to the embodiment of the present application is shown in the figure;

[0075] Figure 6 A structural block diagram of the bulk cargo terminal operation situation perception system according to the embodiment of the present application is shown in the figure;

[0076] Figure 7 A structural block diagram of the model generation module according to the embodiment of the present application is shown in the figure;

[0077] Figure 8 A structural block diagram of the first model generation module according to the embodiment of the present application is shown in the figure;

[0078] Figure 9 This is a structural block diagram of the second model generation module in an embodiment of the present invention;

[0079] Figure 10 This is a structural block diagram of the third model generation module described in an embodiment of the present invention;

[0080] Figure 11 This is a structural block diagram of the prediction module described in an embodiment of the present invention.

[0081] In the diagram: 1. Data Acquisition Module; 2. Model Generation Module; 21. First Model Generation Module; 211. First Dataset Generation Submodule; 212. First Model Construction Submodule; 213. First Model Training Submodule; 214. First Model Validation Submodule; 22. Second Model Generation Module; 221. Second Dataset Generation Submodule; 222. Second Model Construction Submodule; 223. Second Model Training Submodule; 224. Second Model Validation Submodule; 23. Third Model Generation Module; 231. First Calculation Submodule; 232. Normalization Submodule; 233. Third Dataset Generation Submodule; 234. Third Model Construction Submodule; 235. Third Model Training Submodule; 236. Third Model Validation Submodule; 3. Prediction Module; 31. Health Status Prediction Submodule; 32. Probability Distribution Prediction Submodule; 33. Second Calculation Submodule; 34. Situation Prediction Submodule; 4. Evaluation Module; 5. Equipment Management Module. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments provided by this invention without inventive effort are within the scope of protection of this invention.

[0083] Obviously, the accompanying drawings described below are merely some examples or embodiments of the present invention. Those skilled in the art can apply the present invention to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this invention, modifications to design, manufacturing, or production based on the technical content disclosed in this invention are merely conventional technical means and should not be construed as insufficient disclosure of the present invention.

[0084] Reference to "an embodiment" or "the embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase "in one embodiment" or "in an embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily all referring to a particular embodiment logically separated from the other embodiments of the application. It is explicitly contemplated that embodiments described in the application can be combined with each other in their individual aspects, without conflict.

[0085] Unless otherwise defined, technical terms or scientific terms used in the present application shall have the ordinary meanings used in the art to which the present application pertains. The terms "a", "an", "one", "the", and similar terms in the present application do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced item. The terms "include", "comprise", "have", and any variations thereof in the present application are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a list of steps or modules (units) is not limited to the listed steps or units, but can further include other steps or units not listed or can further include other steps or units inherent to such process, method, product, or device. The terms "connect", "connected", "coupling", and similar terms in the present application are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The term "multiple" in the present application means two or more. The term "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects. The terms "first", "second", "third", and the like in the present application are merely to distinguish similar objects, and do not represent a specific order for the objects.

[0086] In a first aspect of the embodiments of the present application, a bulk cargo terminal operation situation awareness method is provided. Figure 1 is a flow chart of the bulk cargo terminal operation situation awareness method of the embodiments of the present application, and the steps of the bulk cargo terminal operation situation awareness method include:

[0087] S1, a data acquisition step: acquiring device influence factors of different devices in a set time period, the device influence factors being multi-dimensional time sequence data reflecting current and historical operation conditions of the devices.

[0088] Specifically, the device influence factors are divided into real-time state parameters (including high-frequency signals collected by temperature, vibration, current, voltage, and other sensors), operation condition parameters (including cumulative operation time, start-stop times, load rate, maintenance records, and other low-frequency operation and maintenance data), and environmental parameters (including external environmental monitoring values such as temperature and humidity, dust concentration, and salt fog level).

[0089] In an embodiment of the present application, the device impact factor includes but is not limited to sensor data, device operating parameters, device environment parameters. The device includes monitoring devices (such as various sensors such as voltage and current sensors, temperature sensors, vibration sensors, ultrasonic sensors, infrared sensors, liquid level sensors, etc.) and production devices (such as car dumpers, stacker-reclaimers, car loaders, ship loaders, conveyors, etc.). The sensor data includes but is not limited to device temperature, device noise, device voltage, device current, etc. The device operating parameters include but are not limited to device operating time, device carrying capacity, device average start-up time, device start-stop cycle, device maintenance cycle, device maintenance cycle, device wear rate, device operating speed, device factory-set service life, device actual production time, device fault burst rate, etc. The device environment parameters include but are not limited to weather conditions (such as rain, smog, etc.). The specific setting can be made according to the actual demand.

[0090] S2, health degree prediction step: according to the device impact factor of the set time period, the health state analysis model is constructed in advance to evaluate the device health degree index to obtain the device health degree index at different times.

[0091] Specifically, in an embodiment of the present application, referring to Figure 2 , the construction method of the health state analysis model is:

[0092] S21, obtaining historical device impact factors and historical device health degree indexes of different devices to form a data set, and dividing the data set into a training set and a validation set.

[0093] Specifically, the historical device impact factor includes but is not limited to weather conditions (such as rain, smog, etc.), device operating time, device carrying capacity, stop-start time ratio, carrying cargo type ratio, device generated planned life, device maintenance cycle, device repair cycle, device fault burst rate, natural wear and tear, etc. Among them: the device operating time represents the working time of the device since production, the device carrying capacity represents the load condition of the device in the transportation process, the stop-start time ratio represents the ratio of idle time to actual working time of the device, the carrying cargo type ratio represents the proportion of different types of carrying cargo in the total carrying cargo, the device generated planned life represents the expected service life given in the design stage of the device, the device maintenance cycle represents the time of regular maintenance and maintenance, the device repair cycle represents the use cycle after fault repair, the device fault burst rate represents the frequency of past fault events, and the natural wear and tear represents the mechanical wear and tear caused by long-term operation.

[0094] In the embodiments of the present application, the training set and the validation set are divided according to a set proportion. For example, the training set and the validation set are divided according to a proportion of 8:2. It should be noted that the proportion of the training set and the validation set can be set according to actual needs. The fixed proportion segmentation algorithm is used to divide the data set, so as to ensure the fairness of the data set division and avoid the occurrence of features in the training set that are not covered in the validation set.

[0095] S22, based on the regression model, the device environment parameter and the device state parameter are taken as inputs, and the device health index is taken as output to construct a health state analysis model.

[0096] The health state analysis model is represented as:

[0097]

[0098] In the formula, represents the device health index, represents the factory health index of the device, is the standardized value of the nth device influence factor index, is the weight of the nth device influence factor.

[0099] S23, the health state analysis model is trained and verified through the training set and the validation set.

[0100] Specifically, in an embodiment of the present application, when training the health state analysis model, the weight region is trained through the training set, and the health state analysis model weight is updated by using a statistical method (for example, expert scoring method, regression analysis method and principal component analysis method, etc.).

[0101] It should be noted that in the process of training the health state analysis model through the training set, the health state analysis model learns the features and patterns in the data to optimize the parameters of the health state analysis model and improve the prediction ability of the health state analysis model on the input data. The use of the training set can effectively avoid data leakage and model overfitting, and ensure that the health state analysis model has good fitting ability in the training stage.

[0102] The performance of the trained health state analysis model is verified through the validation set. The role of the validation set is to evaluate the generalization ability of the health state analysis model, that is, the performance of the health state analysis model on unseen data. Through the test of the validation set, the accuracy and other key indicators of the health state analysis model can be accurately judged, so as to determine whether the health state analysis model can meet the actual application requirements.

[0103] In the embodiments of the present application, based on the training and verification mechanism of historical data, through scientific and reasonable data set division, model training and performance verification process, the accuracy and generalization ability of the health state analysis model are effectively improved. In practical application, the health state analysis model can better adapt to the equipment health state evaluation in different scenarios, and quickly and accurately evaluate the equipment health index according to real-time input data.

[0104] In some specific implementations in the embodiments of the present application, the K-fold cross-validation method is used to train and verify the health state analysis model. Taking the five-fold cross-validation method as an example, the historical equipment influence factors and historical equipment health degree indexes of different equipment are formed into a data set. In the training and verification, the data set is divided into five subsets, and any one subset is selected as the verification set and the other subsets are selected as the training set. In the training and verification process, the health state analysis model is verified for five rounds, the results of the five rounds of verification are summarized, the average performance index of the five rounds is calculated, and the performance index of the health state analysis model is obtained.

[0105] In the embodiments of the present application, the K-fold cross-validation method is used to train and verify the health state analysis model, which can fully utilize each sample in the data set, and at the same time avoid the dependence of the health state analysis model on specific data division, so as to more accurately evaluate the generalization ability of the health state analysis model.

[0106] According to the calculated equipment health degree index The equipment health degree level can be divided.

[0107] For example, the health degree level is divided as follows:

[0108] , which indicates that the health state of the equipment is good, and the equipment can continue to operate normally.

[0109] , which indicates that the equipment is in a sub-healthy state, and it is recommended to perform routine inspection.

[0110] , which indicates that the equipment needs attention, and it is recommended to perform maintenance inspection.

[0111] , which indicates that the equipment has a high risk of failure, and it is recommended to immediately stop and perform comprehensive inspection and repair.

[0112] Traditional equipment health assessment may rely on manual inspection or simple threshold judgment, which is difficult to realize continuous and accurate quantification. Through the health degree analysis model, on the one hand, the health degree index of the equipment can be updated in real time or regularly according to the continuous data (such as hourly, daily sensor data, etc.) in a set time period, reflecting the change trend of the equipment state; on the other hand, the complex state of the equipment (such as vibration, temperature, pressure, etc. Multidimensional data) is converted into a single health degree index, so that the equipment health state is intuitive and visible, which is convenient for quick understanding and decision-making.

[0113] S3, remaining life prediction step: according to the equipment influence factor of the set time period, the remaining life probability distribution of the equipment is predicted through the pre-constructed remaining life probability distribution prediction model, the risk failure time is determined according to the remaining life probability distribution of the equipment, and the remaining life of the equipment is calculated according to the risk failure time and the current time.

[0114]

[0115] In the formula, is the remaining life of the equipment, is the risk failure time, is the current time.

[0116] Specifically, the method for determining the risk failure time according to the probability distribution of the remaining life of the equipment is to select the quantile of the probability distribution of the remaining life of the equipment at a preset confidence level (such as 90%) as the risk failure time.

[0117] Specifically, in an embodiment of the application, referring to Figure 3 , the construction method of the remaining life probability distribution prediction model is:

[0118] S31, obtaining historical equipment influence factors and historical equipment remaining life probability distribution of different equipment to form a data set, and dividing the data set into a training set and a validation set.

[0119] Specifically, the historical equipment influence factors include equipment historical operation data (such as vibration, temperature, current, voltage, etc.), equipment historical full life cycle operation records (including failure time label) and equipment environment data (such as humidity, load, running mode, etc.).

[0120] In an embodiment of the application, the training set and the validation set are divided according to a set proportion. For example, the training set and the validation set are divided according to a proportion of 9:1. It should be noted that the proportion of the training set and the validation set can be set according to actual needs. The fixed proportion segmentation algorithm is used to divide the data set, which ensures the fairness of the data set division and avoids the occurrence of features in the training set in the validation set.

[0121] S32, based on the long short-term memory network LSTM, taking the device influence factor as input and taking the device residual life probability distribution as output to construct a residual life probability distribution prediction model.

[0122] The device residual life probability distribution prediction model based on LSTM outputs the probability distribution (such as normal distribution, Weibull distribution, or interval distribution obtained through quantile regression) of the residual life of the device by inputting the device influence factors (such as sensor data, operating parameters, environmental variables, etc.), and has a more comprehensive uncertainty expression capability compared with the traditional point prediction model.

[0123] It should be noted that, on the one hand, when the traditional LSTM model predicts the residual life, it usually outputs a single-point estimate (such as a residual life of 10 days), and cannot reflect the confidence of the prediction result (such as a 90% probability within 8-12 days). The probability distribution prediction model can quantitatively express the uncertainty by outputting the probability density function PDF or the cumulative distribution function CDF of the residual life. For example, if the prediction result is that the residual life follows a normal distribution with a mean of 10 and a standard deviation of 2, it can be inferred that there is a 68% probability that the device will fail within 8-12 days and a 95% probability that the device will fail within 6-14 days.

[0124] On the other hand, the degradation process of the device is often affected by multiple factors (such as load fluctuation, environmental corrosion, component aging, etc.), and there is significant nonlinearity and randomness. The traditional LSTM model can effectively capture long-term temporal dependencies through its gating mechanism (forget gate, input gate, output gate), and the probability distribution output further adapts to the randomness of the degradation process. For example, when predicting the residual life of a bearing, the model can learn the temporal changes of the influence factors such as vibration signals and temperature, and output the Weibull distribution of the residual life (Weibull distribution is often used to describe the "bathtub curve" characteristics of device life), which is more consistent with the actual degradation law than single-point prediction.

[0125] The historical device influence factors and the actual residual life of the key part roller and the liner plate of the conveyor system are collected, the historical device influence factors are formed into corresponding data sets, and the residual life prediction is performed on the roller data set and the liner plate data set through the traditional LSTM model and the residual life probability distribution model described in the present application. The prediction results are shown in Table 1.

[0126] Table 1

[0127]

[0128] Experiments show that compared with the traditional LSTM model point prediction, the residual life probability distribution model of the present application reduces the MAE by 38% on multiple device data sets (such as the roller data set and the liner plate data set), and the accuracy is improved more significantly especially in the "late stage" when the device is close to failure.

[0129] Specifically, in an embodiment of the present application, the residual life probability distribution prediction model is represented as:

[0130]

[0131] In the formula, is a risk prediction function according to the risk failure time t, representing the instantaneous failure risk of the covariate ; is a baseline risk function, representing the natural aging trend of the device; is a covariate coefficient, representing the influence weight of the device impact factor on the failure risk; is the covariate of the device impact factor; is the number of device impact factors.

[0132] S33, training and verifying the residual life probability distribution prediction model through the training set and the verification set.

[0133] In an embodiment of the present application, based on the training and verification mechanism of historical data, through scientific and reasonable data set division, model training and performance verification process, the accuracy and generalization ability of the residual life probability distribution prediction model are effectively improved. In practical application, the residual life probability distribution prediction model can better adapt to the residual life probability distribution prediction of the device in different scenarios, and quickly and accurately evaluate the residual life probability distribution according to real-time input data.

[0134] In an embodiment of the present application, by learning the time sequence degradation mode of the device state parameters, the residual life probability distribution prediction model is trained, so as to quantify the time distribution of the device from failure.

[0135] In some specific implementation manners in an embodiment of the present application, K-fold cross-validation is used to evaluate the performance of the model, so that each sample participates in training and verification, effectively reducing the evaluation bias, avoiding the dependence of the residual life probability distribution prediction model on specific data division, and thus more accurately evaluating the generalization ability of the residual life probability distribution prediction model.

[0136] S4, situation prediction step: predicting the device failure rate through the pre-constructed situation awareness model according to the device health index and the device residual life.

[0137] In an embodiment of the present application, based on the device health index and the device residual life, the device failure rate is predicted through the situation awareness model, which realizes the accurate quantification and forward-looking prediction of the failure risk, and solves the pain points of traditional failure rate prediction, such as "depending on experience, ignoring dynamic state, and poor timeliness".

[0138] Specifically, in an embodiment of the present application, referring toFigure 4 The method for predicting the device failure rate through the pre-constructed situational awareness model according to the device health index and the device remaining life is:

[0139] S411, calculating the device health index mean and the device health index standard deviation in a set time period according to the health indexes at different time points.

[0140] For example, taking the current time point as the end point, the device health index sequence in a preset time window (such as 7 days) is backtracked, and the mean and the standard deviation of the sequence are calculated to represent the stability of the device health state.

[0141] S412, normalizing the device remaining life to obtain the normalized device remaining life.

[0142] Specifically, the device remaining life is normalized by using the maximum design life related to the device type, so as to ensure the scale consistency between different devices.

[0143] S413, predicting the device failure rate according to the device health index mean, the device health index standard deviation, and the normalized device remaining life through the situational awareness model.

[0144] Specifically, in an embodiment, referring to Figure 5 The construction method of the situational awareness model is:

[0145] S421, obtaining the historical device health index, the historical device remaining life, and the historical device failure rate of different devices.

[0146] S422, calculating the historical device health index mean and the historical device health index standard deviation according to the historical device health index, and normalizing the historical device remaining life to obtain the normalized historical device remaining life;

[0147] S423, forming a data set by using the normalized historical device remaining life, the historical device health index mean, the historical device health index standard deviation, and the historical device failure rate, and dividing the data set into a training set and a validation set.

[0148] S424, constructing a situational awareness model based on a long short-term memory network (LSTM), taking the device health index mean, the device health index standard deviation, and the normalized device remaining life as inputs, and taking the device failure rate as output; the situational awareness model is represented as:

[0149]

[0150] In the formula, is the device failure rate; is the device health index mean; is a sensitivity constant, which is set by expert experience; is a device health index standard deviation; is a normalized device remaining life.

[0151] For example, the device health index mean is the device health index mean in the past 30 days, the device health index standard deviation is the device health index standard deviation in the past 30 days, and the device health index mean is taken as .

[0152] S425, training and verifying the situational awareness model through the training set and the verification set.

[0153] In the embodiments of the present application, based on the training and verification mechanism of historical data, through scientific and reasonable data set division, model training and performance verification process, the accuracy and generalization ability of the situational awareness model are effectively improved. In practical application, the situational awareness model can better adapt to the prediction of device failure rate in different scenarios, and quickly and accurately evaluate the device failure rate according to real-time input data.

[0154] In some specific implementation manners in the embodiments of the present application, the K-fold cross-validation method is used to train and verify the situational awareness model. Through the K-fold cross-validation method, the situational awareness model is trained and verified, so that each sample in the data set can be fully utilized, and the dependence of the situational awareness model on specific data division is avoided, so that the generalization ability of the situational awareness model can be more accurately evaluated.

[0155] S5, evaluation step: generating device alarm signals of different levels according to the device failure rate.

[0156] In the embodiments of the present application, based on the device failure rate, multi-level alarms are divided (such as: first-level alarm ≤5% (low risk), 5%<second-level alarm ≤15% (medium risk), 15%<third-level alarm ≤30% (high risk), fourth-level alarm >30% (extremely high risk)), different levels correspond to different alarm modes (such as: first-level only system record, second-level pop-up reminder, third-level sound and light alarm, fourth-level emergency shutdown notification). The alarm level is dynamically updated with the failure rate (such as when the device failure rate increases from 8% to 18%, the second-level alarm is automatically upgraded to the third-level alarm), avoiding the alarm lag caused by the static threshold, and at the same time, through the hierarchical filtering of low-risk alarms, the invalid response of the operation and maintenance personnel is reduced.

[0157] S6, device management step: generating an auxiliary management scheme according to the device alarm signal, and controlling the execution of the auxiliary management scheme to assist in managing the device.

[0158] The application realizes rapid decision-making through the linkage of alarm levels and preconfigured auxiliary management schemes. Different alarm levels are preset with standardized auxiliary management schemes (for example, the second alarm corresponds to "increasing the inspection frequency to once every 2 hours + real-time monitoring of key parameters"; the third alarm corresponds to "arranging maintenance within 48 hours + allocating spare parts"; and the fourth alarm corresponds to "emergency shutdown and maintenance + starting standby equipment"). When an alarm is triggered, the scheme is automatically pushed to the operation and maintenance management system, and the core execution points (such as maintenance priority, required spare part model, and operation process) are marked, without the need for manual additional research and judgment. The response time is shortened from the traditional several hours to minutes, and the fault handling efficiency is significantly improved.

[0159] It should be noted that the steps shown in the above process or the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

[0160] The bulk cargo terminal operation and maintenance situation perception method of the embodiment of the application predicts the instantaneous health state and residual life of the equipment, predicts the equipment failure rate based on the health state and residual life, and realizes accurate early warning, rapid response and effective control of equipment risk through hierarchical alarm driven by the equipment failure rate, linkage of alarm and management scheme, and closed-loop control of the execution process, so as to finally achieve the core goal of "reducing failure loss, improving operation and maintenance efficiency, and ensuring production safety", and provide core support for fine management of the whole life cycle of the equipment.

[0161] The second aspect embodiment of the application provides a bulk cargo terminal operation and maintenance situation perception system for realizing the bulk cargo terminal operation and maintenance situation perception method of the first aspect embodiment of the application. Figure 6 The structure block diagram of the bulk cargo terminal operation and maintenance situation perception system of the embodiment of the application is shown in Figure 6 As shown in the figure, the bulk cargo terminal operation and maintenance situation perception system comprises:

[0162] The data acquisition module 1 is configured to acquire the equipment influence factors of different equipment in a set time period, and the historical equipment influence factors, the historical equipment health index, the historical equipment residual life probability distribution, the historical equipment residual life and the historical failure rate of different equipment.

[0163] The model generation module 2 is configured to generate a health state analysis model according to the historical equipment influence factors and the historical equipment health index of different equipment, generate a residual life probability distribution prediction model according to the historical equipment influence factors and the historical equipment residual life probability distribution of different equipment, and generate a situation perception model according to the historical equipment health index, the historical equipment residual life and the historical failure rate of different equipment.

[0164] The prediction module 3 is configured to predict the equipment health index and the probability distribution of the equipment residual life of different equipment according to the equipment influence factor of the set time period through the health state analysis model and the residual life probability distribution prediction model respectively, determine the risk failure time according to the probability distribution of the equipment residual life, calculate the equipment residual life according to the risk failure time and the current time, and predict the equipment failure rate through the situation awareness model according to the equipment health index and the equipment residual life.

[0165] The evaluation module 4 is configured to generate equipment alarm signals of different levels according to the equipment failure rate.

[0166] The equipment management module 5 is configured to generate an auxiliary management scheme according to the equipment alarm signal, and control the equipment to be managed by the auxiliary management scheme.

[0167] In some embodiments, referring to Figure 7 , the model generation module 2 comprises:

[0168] The first model generation module 21 is configured to generate the health state analysis model according to the historical equipment influence factor and the historical equipment health index of different equipment.

[0169] The second model generation module 22 is configured to generate the residual life probability distribution prediction model according to the historical equipment influence factor and the historical equipment residual life probability distribution of different equipment.

[0170] The third model generation module 23 is configured to generate the situation awareness model according to the historical equipment health index, the historical equipment residual life and the historical failure rate of different equipment.

[0171] In some embodiments of the present application, referring to Figure 8 , the first model generation module 21 comprises:

[0172] The first data set generation sub-module 211 is configured to form a first data set according to the historical equipment influence factor and the historical equipment health index of different equipment, and divide the first data set into a first training set and a first verification set.

[0173] The first model construction sub-module 212 is configured to construct the health state analysis model based on a regression model, taking the equipment influence factor as the input and the equipment health index as the output.

[0174] The first model training sub-module 213 is configured to train the health state analysis model according to the first training set.

[0175] The first model verification sub-module 214 is configured to verify the performance of the health state analysis model according to the first verification set.

[0176] In some embodiments of the present application, referring to Figure 9 , the second model generation module 22 comprises:

[0177] The second data set generation submodule 221 is configured to form a second data set by using historical device impact factors and historical device residual life probability distributions of different devices, and divide the second data set into a second training set and a second verification set;

[0178] The second model construction submodule 222 is configured to construct a residual life probability distribution prediction model based on a long short-term memory network (LSTM), take a device impact factor as an input, and take a device residual life probability distribution as an output.

[0179] The second model training submodule 223 is configured to train the residual life probability distribution prediction model according to the second training set.

[0180] The second model verification submodule 224 is configured to verify the performance of the residual life probability distribution prediction model according to the second verification set.

[0181] In some embodiments, referring to Figure 10 , the third model generation module 23 comprises:

[0182] The first calculation submodule 231 is configured to calculate a historical device health index mean and a historical device health index standard deviation according to a historical device health index.

[0183] The normalization submodule 232 is configured to perform normalization processing on the historical device residual life to obtain normalized historical device residual life.

[0184] The third data set generation submodule 233 is configured to form a third data set by using the normalized historical device residual life, the historical device health index mean, the historical device health index standard deviation, and a historical device failure rate, and divide the third data set into a third training set and a third verification set.

[0185] The third model construction submodule 234 is configured to construct a situational awareness model based on a long short-term memory network (LSTM), take the device health index mean, the device health index standard deviation, and the normalized device residual life as inputs, and take a device failure rate as an output.

[0186] The third model training submodule 235 is configured to train the situational awareness model according to the third training set.

[0187] The third model verification submodule 236 is configured to verify the performance of the situational awareness model according to the third verification set.

[0188] In some embodiments of the present application, referring to Figure 11 , the prediction module 3 comprises:

[0189] The health state prediction submodule 31 is configured to predict the equipment health index of different equipment through the health state analysis model according to the equipment influence factor of the set time period respectively.

[0190] The probability distribution prediction submodule 32 is configured to predict the probability distribution of the equipment residual life through the residual life probability distribution prediction model according to the equipment influence factor of the set time period.

[0191] The second calculation submodule 33 is configured to determine the risk failure time according to the probability distribution of the equipment residual life, and calculate the equipment residual life according to the risk failure time and the current time.

[0192] The situation prediction submodule 34 is configured to predict the equipment failure rate through the situation awareness model according to the equipment health index and the equipment residual life.

[0193] The bulk cargo terminal operation situation awareness system described above in the embodiment of the present application predicts the instantaneous health state and the residual life of the equipment, predicts the equipment failure rate based on the health state and the residual life, and realizes the accurate early warning, the rapid response and the effective control of the equipment risk through the hierarchical alarm driven by the equipment failure rate, the linkage of the alarm and the management scheme, and the closed-loop control of the execution process, so as to finally achieve the core goal of “reducing failure loss, improving operation efficiency, and ensuring production safety”, and provide core support for the fine management of the whole life cycle of the equipment.

[0194] The following will compare the above-mentioned bulk cargo terminal operation situation awareness method and system (hereinafter referred to as the present application) of the present application with the prior art to verify the effectiveness of the present application by combining specific embodiments.

[0195] Embodiment: The historical influence factor, the historical health index, the historical residual life and the historical failure rate of the main vulnerable parts of the bulk cargo terminal conveyor are adopted, the historical influence factor is taken as the conveyor data set, and the failure rate analysis of the conveyor is performed on the data set by the present application and the prior art. The analysis result is shown in Table 2.

[0196] Table 2

[0197]

[0198] As shown in Table 2, compared with the traditional method, the failure rate prediction error of the present application on the conveyor data set is obviously reduced, for example, the MAE is reduced by 18.3%, and the RMSE is reduced by 18.2%, which can effectively identify the equipment failure rate, realize the accurate early warning, the rapid response and the effective control of the equipment risk through the hierarchical alarm driven by the equipment failure rate, the linkage of the alarm and the management scheme, and the closed-loop control of the execution process.

[0199] Any combination of the technical features in the above-described embodiments can be made, and for the sake of brevity, not all possible combinations are described, however, as long as the combination of the technical features does not exist in contradiction, it shall be considered within the scope of the present disclosure.

[0200] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it shall not be understood as a limitation on the patent scope of the present application. It shall be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these shall be within the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A method for situational awareness in the operation and maintenance of bulk cargo terminals, characterized in that, Includes the following steps: Data acquisition steps: Obtain the device impact factors of different devices within a specified time period; Health prediction steps: Based on the equipment impact factors within a set time period, the equipment health index is evaluated using a pre-built health status analysis model to obtain the equipment health index at different times; Remaining life prediction steps: Based on the equipment impact factors within a set time period, predict the probability distribution of the remaining life of the equipment using a pre-built remaining life probability distribution prediction model; determine the risk failure time based on the probability distribution of the remaining life of the equipment; and calculate the remaining life of the equipment based on the risk failure time and the current time. Situation prediction steps: Based on the equipment health index and remaining equipment lifespan, predict the equipment failure rate using a pre-built situation awareness model; Assessment steps: Generate different levels of equipment alarm signals based on the equipment failure rate; Equipment management steps: Generate auxiliary management plans based on equipment alarm signals, and control the execution of auxiliary management plans to assist in equipment management; The method for constructing the situational awareness model is as follows: Obtain historical equipment health index, historical equipment remaining lifespan, and historical equipment failure rate for different devices; The mean and standard deviation of the historical equipment health index are calculated based on the historical equipment health index. The remaining life of the historical equipment is then normalized to obtain the normalized remaining life of the historical equipment. The normalized remaining lifespan of historical equipment, the mean of historical equipment health index, the standard deviation of historical equipment health index, and the failure rate of historical equipment are used to form a dataset, which is then divided into a training set and a validation set. Based on the Long Short-Term Memory (LSTM) network, a situational awareness model is constructed using the mean of the equipment health index, the standard deviation of the equipment health index, and the normalized remaining equipment lifetime as inputs, and the equipment failure rate as output. The situational awareness model is expressed as follows: In the formula, For equipment failure rate, This represents the average health index of the equipment. It is the sensitivity constant. The standard deviation of the equipment health index. This represents the normalized remaining lifespan of the equipment. The situational awareness model was trained and validated using training and validation sets.

2. The bulk cargo terminal operation and maintenance situation awareness method as described in claim 1, characterized in that, The method for constructing the health status analysis model is as follows: The historical device impact factors and historical device health indices of different devices are obtained to form a dataset, and the dataset is divided into a training set and a validation set; Based on a regression model, a health status analysis model is constructed using equipment influencing factors as input and equipment health index as output; the health status analysis model is expressed as follows: In the formula, This indicates the equipment health index. This indicates the equipment's health index before it leaves the factory. Let n be the standardized value of the impact factor index for the nth device. The weight of the influence factor for the nth device; The health status analysis model was trained and validated using training and validation sets.

3. The bulk cargo terminal operation and maintenance situation awareness method as described in claim 2, characterized in that, When training the health status analysis model, the weight regions are trained using the training set, and statistical methods are used to update the weights of the health status analysis model.

4. The bulk cargo terminal operation and maintenance situation awareness method as described in claim 1, characterized in that, The method for constructing the remaining lifetime probability distribution prediction model is as follows: A dataset is formed by obtaining the historical equipment impact factors and the probability distribution of the remaining lifespan of different equipment, and the dataset is divided into a training set and a validation set. Based on the Long Short-Term Memory (LSTM) network, a remaining lifetime probability distribution prediction model is constructed using the equipment impact factor as input and the equipment remaining lifetime probability distribution as output; the remaining lifetime probability distribution prediction model is expressed as: In the formula, Let be the risk prediction function based on the risk failure time t, representing the covariate... The risk of instantaneous failure; This is the baseline risk function, representing the natural aging trend of the equipment; Let be the covariate coefficient, representing the th . The weight of the impact factor of equipment on failure risk; For the first Covariates of equipment impact factors; The number of equipment impact factors; The remaining lifespan probability distribution prediction model was trained and validated using the training and validation sets.

5. The bulk cargo terminal operation and maintenance situation awareness method as described in claim 1, characterized in that, In the situation prediction step, the method for predicting the equipment failure rate based on the equipment health index and remaining equipment lifespan using a pre-built situation awareness model is as follows: Calculate the mean and standard deviation of the equipment health index within a set time period based on the health index at different times. The remaining equipment life is normalized to obtain the normalized remaining equipment life. The equipment failure rate is predicted using the situational awareness model based on the mean of the equipment health index, the standard deviation of the equipment health index, and the normalized remaining equipment lifespan.

6. A bulk cargo terminal operation and maintenance situation awareness system, used to implement the bulk cargo terminal operation and maintenance situation awareness method as described in any one of claims 1 to 5, characterized in that, include: The data acquisition module is used to acquire the equipment impact factors of different devices within a set time period, as well as the historical equipment impact factors, historical equipment health index, historical equipment remaining life probability distribution, historical equipment remaining life, and historical failure rate of different devices. The model generation module is used to generate a health status analysis model based on the historical equipment impact factors and historical equipment health index of different devices, generate a remaining life probability distribution prediction model based on the historical equipment impact factors and historical equipment remaining life probability distribution of different devices, and generate a situational awareness model based on the historical equipment health index, historical equipment remaining life, and historical failure rate of different devices. The prediction module is used to predict the equipment health index and the probability distribution of the remaining life of equipment based on the equipment impact factors of different equipment within a set time period, using a health status analysis model and a remaining life probability distribution prediction model, respectively; determine the risk failure time based on the probability distribution of the remaining life of equipment; calculate the remaining life of equipment based on the risk failure time and the current time; and predict the equipment failure rate based on the equipment health index and the remaining life of equipment using a situational awareness model. The evaluation module is used to generate different levels of equipment alarm signals based on the equipment failure rate. The equipment management module is configured to generate auxiliary management schemes based on equipment alarm signals and control the execution of these schemes to assist in equipment management.

7. The bulk cargo terminal operation and maintenance situation awareness system as described in claim 6, characterized in that, The model generation module includes: The first model generation module is used to generate a health status analysis model based on the historical equipment impact factors and historical equipment health indices of different devices. The second model generation module generates a remaining life probability distribution prediction model based on the historical equipment impact factors and historical equipment remaining life probability distribution of different equipment. The third model generation module generates a situational awareness model based on the historical equipment health index, historical equipment remaining lifespan, and historical failure rate of different devices.

8. The bulk cargo terminal operation and maintenance situation awareness system as described in claim 7, characterized in that, The first model generation module includes: The first dataset generation submodule is used to form the first dataset based on the historical device impact factor and historical device health index of different devices, and to divide the first dataset into the first training set and the first validation set. The first model construction submodule is based on a regression model, using equipment influencing factors as input and equipment health index as output to construct a health status analysis model. The first model training submodule is used to train the health status analysis model based on the first training set. The first model validation submodule is used to validate the performance of the health status analysis model based on the first validation set. The second model generation module includes: The second dataset generation submodule is used to form the second dataset by the historical device impact factor and the historical device remaining life probability distribution of different devices, and the second dataset is divided into the second training set and the second validation set. The second model construction submodule is based on the Long Short-Term Memory (LSTM) network. It uses the device impact factor as input and the device remaining lifetime probability distribution as output to construct a remaining lifetime probability distribution prediction model. The second model training submodule is used to train the remaining lifespan probability distribution prediction model based on the second training set. The second model validation submodule is used to validate the performance of the remaining lifetime probability distribution prediction model based on the second validation set. The third model generation module includes: The first calculation submodule is used to calculate the mean and standard deviation of the historical equipment health index based on the historical equipment health index. The normalization submodule normalizes the remaining lifespan of historical equipment to obtain the normalized remaining lifespan of historical equipment. The third dataset generation submodule is used to form a third dataset by combining the normalized remaining lifespan of historical equipment with the mean of historical equipment health index, the standard deviation of historical equipment health index, and the failure rate of historical equipment, and divide it into a third training set and a third validation set. The third model construction submodule is used to build a situational awareness model based on the Long Short-Term Memory (LSTM) network, with the mean of the equipment health index, the standard deviation of the equipment health index, and the normalized remaining equipment lifetime as inputs, and the equipment failure rate as output. The third model training submodule is used to train the situational awareness model based on the third training set. The third model verification submodule is used to verify the performance of the situational awareness model based on the third verification set.

9. The bulk cargo terminal operation and maintenance situation awareness system as described in claim 6, characterized in that, The prediction module includes: The health status prediction submodule is used to predict the equipment health index based on the equipment impact factors of different equipment within a set time period using a health status analysis model. The probability distribution prediction submodule is used to predict the probability distribution of the remaining life of equipment based on the equipment impact factors over a set time period using the remaining life probability distribution prediction model. The second calculation submodule is used to determine the risk failure time based on the probability distribution of the remaining lifespan of the equipment, and to calculate the remaining lifespan of the equipment based on the risk failure time and the current time. The situation prediction submodule is used to predict the equipment failure rate based on the equipment health index and the remaining lifespan of the equipment using a situational awareness model.

Images