CN118167790A

CN118167790A - Monitoring protection method and system for speed reducer

Info

Publication number: CN118167790A
Application number: CN202410600526.7A
Authority: CN
Inventors: 徐婧; 李泽民; 李柏君
Original assignee: Modori Intelligent Transmission Jiangsu Co ltd
Current assignee: Modori Intelligent Transmission Jiangsu Co ltd
Priority date: 2024-05-15
Filing date: 2024-05-15
Publication date: 2024-06-11

Abstract

The application provides a monitoring protection method and a system of a speed reducer, which relate to the technical field of equipment monitoring operation and maintenance, and the method comprises the following steps: synchronously receiving the monitoring data flow through a twin model of the speed reducer, and determining an abnormal monitoring data set and an abnormal trend set; performing simulated fault prediction to generate simulated fault prediction information; performing network fault prediction to generate network fault prediction information; fusing to obtain fault prediction information; delay analysis is carried out on the fault type and the fault strength, the prediction delay time is determined to compensate the fault prediction node, and a compensation prediction node is generated; and if the compensation prediction node does not meet the preset node, overhauling and maintaining are carried out. The method and the device can solve the technical problems that potential fault threats of the speed reducer and fault occurrence nodes can not be found in time due to low precision of fault prediction, can improve the precision of fault prediction of the speed reducer, and effectively avoid production interruption and loss caused by sudden shutdown or occurrence of serious faults of equipment.

Description

Monitoring protection method and system for speed reducer

Technical Field

The application relates to the technical field of equipment monitoring operation and maintenance, in particular to a monitoring protection method and system of a speed reducer.

Background

The speed reducer is widely applied to various industrial equipment, such as production lines, conveying belts, mechanical arms and the like, and in the scenes, the speed reducer plays roles of reducing speed and increasing torque, so that the working requirements of the equipment are met, and once the speed reducer fails, the speed reducer is often in a malignant fault and can directly lead to production line or equipment to stop, production interruption and serious loss due to the fact that failure precursor characteristics of the speed reducer in the operation process are hidden and are not easy to perceive and judge.

At present, the existing speed reducer monitoring protection method has the technical problems that potential fault threats of the speed reducer cannot be found in time and fault occurrence nodes are accurately predicted due to low fault prediction accuracy, so that equipment operation risks are large.

Disclosure of Invention

The application aims to provide a monitoring protection method and system for a speed reducer, which are used for solving the technical problems that the potential fault threat of the speed reducer cannot be found in time and the fault occurrence node is accurately predicted due to lower precision of fault prediction in the traditional speed reducer monitoring protection method, so that the equipment running risk is larger.

In view of the above problems, the present application provides a method and a system for monitoring and protecting a speed reducer.

In a first aspect, the present application provides a method for monitoring and protecting a speed reducer, where the method is implemented by a monitoring and protecting system for a speed reducer, and the method includes: taking attribute information and a working scene of a target speed reducer as constraints, and carrying out networking retrieval to determine a high-frequency fault data set, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault correlation feature sets of the high-frequency fault types; building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream; synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set, and determining an abnormal data trend set; utilizing a twin sub-model of the speed reducer twin model to perform simulated fault prediction based on the anomaly monitoring dataset and the data trend set, and generating simulated fault prediction information; performing network fault prediction based on the anomaly monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm; fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node; reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the predicted deferral time is determined, and the fault prediction node is compensated based on the predicted deferral time, so that a compensation prediction node is generated; and if the compensation prediction node does not meet the preset node, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.

In a second aspect, the present application further provides a monitoring and protecting system for a speed reducer, for executing the monitoring and protecting method for a speed reducer according to the first aspect, where the system includes: the high-frequency fault data set determining module is used for determining a high-frequency fault data set by taking attribute information and a working scene of a target speed reducer as constraints through networking retrieval, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault associated feature sets of the high-frequency fault types; the target speed reducer monitoring module is used for building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream; the abnormal data trend set determining module is used for synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set and determining an abnormal data trend set; the simulated fault prediction module is used for carrying out simulated fault prediction based on the anomaly monitoring data set and the data trend set by utilizing a twin sub-model of the speed reducer twin model to generate simulated fault prediction information; the network fault prediction module is used for predicting network faults based on the abnormal monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm; the fault prediction information obtaining module is used for fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node; the compensation prediction node generation module is used for reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the prediction deferral time is determined, compensation is carried out on the fault prediction node based on the prediction deferral time, and a compensation prediction node is generated; and the speed reducer overhauling and maintaining module is used for executing the shutdown of the source equipment of the target speed reducer if the compensation prediction node does not meet the preset node, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.

One or more technical schemes provided by the application have at least the following technical effects or advantages:

1. The method comprises the steps of determining a high-frequency fault data set by taking attribute information and a working scene of a target speed reducer as constraints through networking retrieval, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault associated feature sets of the high-frequency fault types; building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream; synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set, and determining an abnormal data trend set; utilizing a twin sub-model of the speed reducer twin model to perform simulated fault prediction based on the anomaly monitoring dataset and the data trend set, and generating simulated fault prediction information; performing network fault prediction based on the anomaly monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm; fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node; reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the predicted deferral time is determined, and the fault prediction node is compensated based on the predicted deferral time, so that a compensation prediction node is generated; and if the compensation prediction node does not meet the preset node, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength. That is, the sensing monitoring matrix is built based on the fault association characteristic set of the target speed reducer, and the target speed reducer is monitored in real time to obtain a monitoring data stream; then, identifying the monitoring data flow by utilizing a synchronously operated speed reducer twin model, determining abnormal monitoring data, and tracing to determine the change trend of the abnormal data; then, carrying out simulated fault prediction based on the abnormal monitoring data and the abnormal data change trend through a twin sub-model of the speed reducer twin model, and determining simulated fault prediction information; on the other hand, a preset fault prediction network is built based on the BP neural network, the training process of the preset fault prediction network is optimized through an optimizing algorithm, and network fault prediction is carried out through the optimized preset fault prediction network, so that network fault prediction information is generated; further fusing the simulated fault prediction information and the network fault prediction information to obtain fault prediction information, wherein the fault prediction information comprises fault types, fault intensities and fault prediction nodes; performing deferral analysis on the fault type and the fault strength according to a preset fault deferral scheme, generating a predicted deferral time to compensate the fault prediction node, and generating a compensated prediction node; and finally judging whether the compensation prediction node meets a preset node, if not, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer. The method can improve the accuracy of predicting the faults of the speed reducer, discover potential fault threat and accurately predict fault occurrence nodes of the speed reducer in time, effectively avoid production interruption and loss caused by sudden shutdown or major faults of equipment, and ensure the reliability and stability of equipment operation.

2. The fault prediction network of the speed reducer is built based on the BP neural network, and the initial weight and the initial bias of the fault prediction network are optimized by utilizing the optimizing algorithm, so that the model precision of the fault prediction network can be remarkably improved, and the precision and the accuracy of the fault prediction of the speed reducer network are further improved.

The foregoing description is only an overview of the present application, and is intended to be implemented in accordance with the teachings of the present application in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present application more readily apparent. It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the application or to delineate the scope of the application. Other features of the present application will become apparent from the description that follows.

Drawings

In order to more clearly illustrate the application or the technical solutions of the prior art, the following brief description will be given of the drawings used in the description of the embodiments or the prior art, it being obvious that the drawings in the description below are only exemplary and that other drawings can be obtained from the drawings provided without the inventive effort for a person skilled in the art.

FIG. 1 is a schematic flow chart of a method for monitoring and protecting a speed reducer according to the present application;

FIG. 2 is a schematic flow chart of a method for monitoring and protecting a speed reducer according to the present application for determining a high-frequency fault data set through networking search;

Fig. 3 is a schematic structural diagram of a monitoring protection system of a speed reducer according to the present application.

Reference numerals illustrate:

the system comprises a high-frequency fault data set determining module 11, a target speed reducer monitoring module 12, an abnormal data trend set determining module 13, a simulated fault predicting module 14, a network fault predicting module 15, a fault predicting information obtaining module 16, a compensation predicting node generating module 17 and a speed reducer overhauling and maintaining module 18.

Detailed Description

The application provides a monitoring protection method and a monitoring protection system for a speed reducer, which solve the technical problems that the potential fault threat of the speed reducer cannot be found in time and the fault occurrence node is accurately predicted due to lower accuracy of fault prediction, so that the equipment running risk is larger in the existing monitoring protection method for the speed reducer. The method can improve the accuracy of predicting the faults of the speed reducer, discover potential fault threat and accurately predict fault occurrence nodes of the speed reducer in time, effectively avoid production interruption and loss caused by sudden shutdown or major faults of equipment, and ensure the reliability and stability of equipment operation.

In the following, the technical solutions of the present application will be clearly and completely described with reference to the accompanying drawings, and it should be understood that the described embodiments are only some embodiments of the present application, but not all embodiments of the present application, and that the present application is not limited by the exemplary embodiments described herein. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application. It should be further noted that, for convenience of description, only some, but not all of the drawings related to the present application are shown.

Example 1

Referring to fig. 1, the application provides a monitoring protection method for a speed reducer, wherein the method is applied to a monitoring protection system for the speed reducer, and specifically comprises the following steps:

Step one: taking attribute information and a working scene of a target speed reducer as constraints, and carrying out networking retrieval to determine a high-frequency fault data set, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault correlation feature sets of the high-frequency fault types;

Specifically, firstly, acquiring attribute information and a working scene of a target speed reducer, wherein the attribute information comprises speed reducer type, speed reducer size, structural characteristics, working principle and performance parameters, and the performance parameters comprise transmission efficiency, transmission ratio and the like; the working scene comprises source equipment, working state and working environment, wherein the source equipment refers to industrial equipment of target speed reducer application, such as: mechanical arm, production line, etc., the operating condition refers to conventional state information in the running process of the speed reducer, for example: load information, high-frequency transmission ratio state and the like, and the working environment refers to environment information in the running process of the speed reducer, and comprises information such as environment temperature, environment humidity, dust and the like. The attribute information and the working scene can be set according to the actual condition of the target speed reducer.

Industrial big data is a series of technologies and methods for mining and displaying the value contained in industrial mass data, and the technologies comprise technical means of data planning, data acquisition, analysis mining and the like. Based on the industrial big data technology, taking attribute information and a working scene of a target speed reducer as retrieval constraint conditions, carrying out information retrieval in a networking mode, and determining a high-frequency fault data set, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types, and the high-frequency fault types refer to frequently occurring speed reducer fault types, such as: gear failure of the speed reducer, bearing failure, etc.; and further obtaining a fault associated feature set of the high frequency fault type, where the fault associated feature set corresponds to the high frequency fault type one-to-one, for example: the fault associated characteristics corresponding to the bearing faults comprise characteristics such as abnormal temperature, abnormal vibration frequency, abnormal amplitude and the like. The frequent fault type and the corresponding fault association characteristics of the target speed reducer can be clarified by acquiring the high-frequency fault data set, and a basis is provided for constructing a sensing monitoring matrix in the next step.

Step two: building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream;

Specifically, determining a fault feature type and abnormal feature positioning based on the fault associated feature set, wherein the abnormal feature positioning refers to a component where an abnormal feature is located, such as abnormal bearing temperature, abnormal gear vibration and the like; then, selecting an adaptive monitoring sensor based on the fault characteristic type, if the abnormal characteristic is temperature abnormality, determining a plurality of adaptive monitoring sensors if the sensor type is temperature sensor; and then distributing a plurality of adaptive monitoring sensors according to the abnormal characteristic positioning to construct a sensing monitoring array. Under a preset time node, the target speed reducer is subjected to sensing monitoring through the sensing monitoring matrix, and the preset time node can be set based on actual conditions, such as: setting the time node to be 1 minute, namely carrying out sensing data acquisition once every 1 minute to obtain a real-time monitoring data stream, wherein the real-time monitoring data stream comprises sensing monitoring data under a plurality of same monitoring nodes. By obtaining the real-time monitoring data stream, data support is provided for the recognition of the abnormal monitoring data of the next step.

Step three: synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set, and determining an abnormal data trend set;

Specifically, the digital twin is a technology integrating multiple physical, multi-scale and multidisciplinary attributes, has the characteristics of real-time synchronization, faithful mapping and high fidelity, integrates multiple technologies such as collaborative modeling, data mining, man-machine interaction and the like, realizes collaborative interaction between the physical world and the digital world, and has the advantage that compared with the traditional fault prediction method, the fault prediction method based on the digital twin has the advantages of improving the prediction precision and accuracy. Based on a digital twin technology, performing simulation modeling on the target speed reducer according to the attribute information and the working scene, generating a speed reducer twin model, and performing synchronous operation on the speed reducer twin model and the target speed reducer.

And synchronously receiving the real-time monitoring data stream through the twin model of the speed reducer, comparing the real-time monitoring data stream with the simulated operation data under the same node, and determining an abnormal monitoring data set based on the comparison result, wherein the abnormal monitoring data is data with larger deviation. And tracing the abnormal monitoring data set, namely acquiring historical monitoring data sets of the abnormal monitoring data under a plurality of monitoring nodes, analyzing the variation trend of the abnormal data based on the historical monitoring data sets, and determining an abnormal data trend set, wherein the abnormal monitoring data and the abnormal data trend are in one-to-one correspondence. By obtaining the abnormal monitoring data set and the abnormal data trend set, a basis is provided for the fault prediction of the next step target speed reducer.

Step four: utilizing a twin sub-model of the speed reducer twin model to perform simulated fault prediction based on the anomaly monitoring dataset and the data trend set, and generating simulated fault prediction information;

Specifically, the twin model of the speed reducer is subjected to twin replication based on a twin network, and a twin sub-model is generated, wherein the twin model of the speed reducer is also used for synchronously monitoring a target speed reducer. And inputting the anomaly monitoring data set and the data trend set into the twin sub-model to perform simulated fault prediction to obtain simulated fault prediction information, wherein the simulated fault prediction information comprises a simulated fault type, simulated fault intensity and simulated fault prediction nodes, and the simulated fault prediction nodes refer to simulated fault occurrence time. By obtaining the simulated fault prediction information, support is provided for subsequent comprehensive fault prediction evaluation.

Step five: performing network fault prediction based on the anomaly monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm;

specifically, the BP neural network is a typical multi-layer feedforward network, and consists of two parts of forward propagation and backward propagation, in the forward propagation stage, an input layer provides information for the network, a signal is learned through each neuron, finally transmitted to an output layer and outputs useful information, and if a predicted value is inconsistent with an actual value, error analysis is needed; in the backward propagation phase, the error is transmitted back to the input layer, and the weights and thresholds are further updated to reduce the error. Building an initial structure of a fault prediction network based on a BP neural network, wherein the fault prediction network is a neural network model which can be subjected to iterative optimization in machine learning and is obtained through supervision training; and then optimizing the initial weight and the initial bias of the fault prediction network by utilizing an optimizing algorithm to obtain the optimal initial weight and the optimal initial bias, configuring the fault prediction network according to the optimal initial weight and the optimal initial bias, and performing supervision training to a convergence state by utilizing a sample data set to obtain a preset fault prediction network.

And then inputting the anomaly monitoring data set and the data trend set into the preset fault prediction network to perform fault prediction to obtain network fault prediction information, wherein the network fault prediction information comprises network prediction fault types, network prediction fault intensities and network fault prediction nodes, and the network fault prediction nodes refer to fault prediction occurrence time. The model precision of the fault prediction network can be remarkably improved by constructing the fault prediction network of the speed reducer based on the BP neural network and optimizing the initial weight and the initial bias of the fault prediction network by utilizing an optimizing algorithm, so that the precision and the accuracy of the fault prediction of the speed reducer network are further improved.

Step six: fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node;

Specifically, a preset fusion strategy is obtained, wherein the preset fusion strategy comprises weight duty ratios of the simulated fault prediction information and the network fault prediction information, the weight duty ratios can be set based on the credibility of the simulated fault prediction and the network fault prediction, and the higher the credibility of which fault prediction is, the larger the corresponding weight duty ratio is, and the weighting can be performed based on the existing variation coefficient method. And then carrying out weighted calculation on the simulated fault prediction information and the network fault prediction information based on the preset fusion strategy, and taking a weighted calculation result as fault prediction information to obtain the fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node.

By fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy, the advantages of digital twin fault prediction and BP neural network fault prediction can be combined, and therefore accuracy and reliability of obtaining the fault prediction information are further improved.

Step seven: reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the predicted deferral time is determined, and the fault prediction node is compensated based on the predicted deferral time, so that a compensation prediction node is generated;

Specifically, a preset fault deferral scheme is obtained, where the preset fault deferral is a method capable of optimizing current abnormal characteristics and reducing abnormal change trend, so as to defer fault occurrence time, and the method can be set according to actual conditions, for example: increasing lubrication strength of the speed reducer, reducing load of the speed reducer and the like. Then carrying out deferral analysis on the fault type and the fault strength according to the preset fault deferral scheme to obtain a predicted deferral time, wherein the predicted deferral time refers to a time period in which the occurrence of the fault can be deferred; and further compensating the fault prediction node based on the prediction delay time, namely summing the prediction delay time and the fault prediction node, and taking the summation result as a compensation prediction node.

Step eight: and if the compensation prediction node does not meet the preset node, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.

Specifically, a preset node is acquired, where the preset node refers to a preset equipment downtime or a preset overhaul time node, for example: every 72 hours of continuous operation of the equipment, the equipment is stopped or overhauled once, and the equipment can be set according to actual conditions; and judging the compensation prediction node based on the preset node, if the compensation prediction node is earlier than the preset node, representing that the fault is earlier than the preset equipment downtime, executing the source equipment downtime of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength, thereby effectively avoiding production interruption and loss caused by the sudden downtime of the equipment or major faults, and ensuring the reliability and stability of the operation of the equipment.

The monitoring protection method of the speed reducer is applied to a monitoring protection system of the speed reducer, and can solve the technical problems that potential fault threats of the speed reducer and accurately predicting fault occurrence nodes cannot be found in time due to low accuracy of fault prediction in the existing monitoring protection method of the speed reducer, so that equipment running risk is high. Firstly, taking attribute information and a working scene of a target speed reducer as constraints, and carrying out networking retrieval to determine a high-frequency fault data set, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault associated feature sets of the high-frequency fault types; then, a sensing monitoring matrix is built based on the fault association characteristic set, and the target speed reducer is monitored through the sensing monitoring matrix to obtain a real-time monitoring data stream; then, synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set, and determining an abnormal data trend set; then, utilizing a twin sub-model of the speed reducer twin model to conduct simulated fault prediction based on the anomaly monitoring dataset and the data trend set, and generating simulated fault prediction information; then, carrying out network fault prediction based on the anomaly monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm; further, fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node; in addition, a preset fault deferral scheme is read, deferral analysis is carried out on the fault type and the fault intensity based on the preset fault deferral scheme, prediction deferral time is determined, compensation is carried out on the fault prediction node based on the prediction deferral time, and a compensation prediction node is generated; and finally, if the compensation prediction node does not meet the preset node, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength. The method comprises the steps of building a sensing monitoring matrix based on a fault association characteristic set of a target speed reducer, and monitoring the target speed reducer in real time to obtain a monitoring data stream; then, identifying the monitoring data flow by utilizing a synchronously operated speed reducer twin model, determining abnormal monitoring data, and tracing to determine the change trend of the abnormal data; then, carrying out simulated fault prediction based on the abnormal monitoring data and the abnormal data change trend through a twin sub-model of the speed reducer twin model, and determining simulated fault prediction information; on the other hand, a preset fault prediction network is built based on the BP neural network, the training process of the preset fault prediction network is optimized through an optimizing algorithm, and network fault prediction is carried out through the optimized preset fault prediction network, so that network fault prediction information is generated; further fusing the simulated fault prediction information and the network fault prediction information to obtain fault prediction information, wherein the fault prediction information comprises fault types, fault intensities and fault prediction nodes; performing deferral analysis on the fault type and the fault strength according to a preset fault deferral scheme, generating a predicted deferral time to compensate the fault prediction node, and generating a compensated prediction node; and finally judging whether the compensation prediction node meets a preset node, if not, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer. The method can improve the accuracy of predicting the faults of the speed reducer, discover potential fault threat and accurately predict fault occurrence nodes of the speed reducer in time, effectively avoid production interruption and loss caused by sudden shutdown or major faults of equipment, and ensure the reliability and stability of equipment operation.

Further, with the attribute information and the working scene of the target speed reducer as constraints, the high-frequency fault data set is determined by networking searching, as shown in fig. 2, and the first step of the application comprises the following steps:

the attribute information comprises a speed reducer type, a speed reducer size, structural characteristics, a working principle and performance parameters, and the working scene comprises source equipment, a working state and a working environment;

Based on industrial big data, taking the attribute information and the working scene as constraint conditions, and obtaining sample fault record data through networking retrieval;

clustering the sample fault record data according to the fault type, and calculating a fault duty ratio of the fault type;

Extracting a fault type with the fault duty ratio larger than a fault duty ratio threshold value to be set as a high-frequency fault type, and acquiring a fault associated feature set of the high-frequency fault type based on the sample fault record data;

And establishing a mapping relation between the high-frequency fault type and the fault association feature set, and obtaining the high-frequency fault data set based on the mapping relation.

Specifically, the attribute information includes a speed reducer type, a speed reducer size, structural characteristics, a working principle and performance parameters, wherein the performance parameters include transmission efficiency, transmission ratio and the like; the working scene comprises source equipment, working state and working environment, wherein the source equipment refers to industrial equipment of target speed reducer application, such as: mechanical arm, production line, etc., the operating condition refers to conventional state information in the running process of the speed reducer, for example: load information, high-frequency transmission ratio state and the like, and the working environment refers to environment information in the running process of the speed reducer, and comprises information such as environment temperature, environment humidity, dust and the like. The attribute information and the working scene can be set according to the actual condition of the target speed reducer.

Then based on industrial big data, taking the attribute information and the working scene as retrieval constraint conditions, and obtaining sample fault record data through networking retrieval, wherein the sample fault record data comprises data such as sample fault types, fault association characteristics and the like; and clustering the sample fault record data according to the fault types, namely gathering the sample fault record data corresponding to the same fault type into one type, and calculating and obtaining the fault duty ratio of the fault type, wherein the fault duty ratio is the ratio of the occurrence frequency of the fault type to the sum of the occurrence frequencies of all the fault types, and the larger the fault duty ratio is, the higher the occurrence frequency of the fault type is represented.

Then extracting a fault type with the fault duty ratio larger than a fault duty ratio threshold value, wherein the fault duty ratio threshold value can be set according to actual conditions, and the fault type is set as a high-frequency fault type; further acquiring a fault-associated feature set of the high-frequency fault type based on the sample fault record data; and then establishing a mapping relation between the high-frequency fault type and the fault association feature set, wherein the high-frequency fault type corresponds to the fault association feature set one by one, and constructing the high-frequency fault data set based on the mapping relation.

Further, the third step of the present application includes:

using a digital twin technology to carry out simulation modeling on the target speed reducer according to the attribute information and the working scene to generate a speed reducer twin model;

synchronously operating the twin model of the speed reducer, performing traversal comparison on the real-time monitoring data stream and the simulation operation data stream of the same node, determining abnormal monitoring data which does not meet the tolerance interval of the data deviation, and generating the abnormal monitoring data set;

Extracting historical monitoring data flow of the abnormal monitoring data in the abnormal monitoring data set, and analyzing data change trend of the abnormal monitoring data based on the historical monitoring data flow to obtain the abnormal data trend set, wherein the abnormal monitoring data and the abnormal monitoring trend are in one-to-one correspondence.

Specifically, based on a digital twin technology, in a three-dimensional visual simulation platform, simulation modeling is carried out on the target speed reducer according to the attribute information and the working scene, and a speed reducer twin model is obtained. Then synchronously operating the twin model of the speed reducer, and synchronously receiving the real-time monitoring data stream through the twin model of the speed reducer; then, performing traversal comparison on the real-time monitoring data stream and the simulation operation data stream of the same node to determine a data deviation set; and screening the data deviation set based on a deviation tolerance section, wherein the deviation tolerance section refers to an allowable data deviation range, the allowable data deviation range can be set according to actual conditions, monitoring data corresponding to data deviation larger than the deviation tolerance section in the data deviation set is extracted and set as abnormal monitoring data, and an abnormal monitoring data set is obtained.

Then extracting a historical monitoring data stream of the abnormal monitoring data in the abnormal monitoring data set, wherein the historical monitoring data stream comprises historical monitoring data under a plurality of monitoring nodes; and then carrying out data change trend analysis on the abnormal monitoring data according to the historical monitoring data flow, wherein the data change trend analysis comprises a data change trend and a data change amplitude, the data change trend comprises an ascending trend, a descending trend and the like, and the abnormal data trend set is obtained, and the abnormal monitoring data and the abnormal monitoring trend are in one-to-one correspondence.

The abnormal monitoring data is subjected to data change trend analysis based on the historical monitoring data flow to obtain an abnormal data trend set, data support is provided for subsequent speed reducer fault prediction, and meanwhile accuracy of speed reducer fault prediction can be further improved.

Further, the predetermined fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm, and the method comprises the following steps:

Setting up an initial fault preset network based on a BP neural network, and determining a weight threshold and a bias threshold of the initial fault preset network;

Taking the attribute information and the working scene as constraints, and obtaining a sample training data set in a networking way, wherein the sample training data comprises a sample abnormal data set, a sample data trend set and sample fault prediction information;

Dividing the sample training data set into a first sample training set and a second sample training set according to a preset data dividing proportion;

optimizing the initial weight and the initial bias based on the first sample training set by taking the weight threshold and the bias threshold as optimizing spaces to obtain an optimized initial weight and an optimized initial bias;

specifically, a BP neural network is a typical multi-layer feed-forward network, consisting of two parts, forward propagation and backward propagation. Setting up an initial fault preset network based on a BP neural network, and determining a weight threshold and a bias threshold of the initial fault preset network, wherein the weight threshold and the bias threshold default to be more than or equal to-1 and less than or equal to 1. And then taking the attribute information and the working scene as constraints, and acquiring a sample training data set based on industrial big data networking retrieval, wherein the sample training data comprises a sample abnormal data set, a sample data trend set and sample fault prediction information.

Obtaining a preset data dividing ratio, wherein the preset data dividing ratio can be set according to the actual sample data quantity, and the first ratio is far smaller than the second ratio, for example: the first ratio may be set to 10% and the second ratio to 90%; dividing the sample training data set into a first sample training set and a second sample training set according to the preset data dividing proportion; and further taking the weight threshold and the bias threshold as optimizing spaces of initial weights and initial biases, and optimizing the initial weights and the initial biases based on the first sample training set to obtain optimized initial weights and optimized initial biases.

Further, the application also comprises the following steps:

extracting the weight threshold and the bias threshold based on a preset step length to obtain a plurality of initial weights and a plurality of initial biases;

Randomly fusing the initial weights and the initial biases to obtain a plurality of initial parameter sets, wherein the initial parameter sets comprise any initial weight and any initial bias;

setting the initial fault preset network based on the plurality of initial parameter sets respectively to obtain a plurality of set fault preset networks;

Respectively performing supervision training on the plurality of preset fault preset networks through the first sample training set to obtain a plurality of first sample training comprehensive error sets;

And optimizing the initial weight and the initial bias based on the plurality of first sample training comprehensive error sets by taking the weight threshold and the bias threshold as constraints to obtain the optimized initial weight and the optimized initial bias.

Specifically, the method for optimizing the initial weight and the initial bias to obtain the optimized initial weight and the optimized initial bias includes the following steps that firstly, a preset step length is obtained and can be set according to actual requirements, wherein the higher the required precision is, the smaller the preset step length is, for example, the preset step length can be set to be 0.05; and extracting the weight threshold and the bias threshold based on a preset step length, namely selecting one initial weight or initial bias every other preset step length to obtain a plurality of initial weights and a plurality of initial biases.

Based on the initial weights and the initial biases, randomly fusing any initial weight and any initial bias to obtain a plurality of initial parameter sets, wherein the initial parameter sets comprise any initial weight and any initial bias; and then, respectively carrying out initial weight and initial bias configuration on the initial fault preset network based on the initial parameter sets to obtain a plurality of setting fault preset networks. And further performing supervision training on the multiple preset fault preset networks through the first sample training set respectively, calculating sample error data of first sample data and network fault prediction results in the training process, accumulating and summing absolute values of the multiple sample errors, and taking the summed result as a first sample training comprehensive error set, wherein the first sample training comprehensive error set characterizes the comprehensive error of any preset fault preset network to obtain multiple first sample training comprehensive error sets, and the first sample training comprehensive error sets are in one-to-one correspondence with the preset fault preset networks.

And then optimizing the initial weight and the initial bias based on the plurality of first sample training comprehensive error sets by taking the weight threshold and the bias threshold as constraints to obtain the optimized initial weight and the optimized initial bias.

Further, the application also comprises the following steps:

determining a plurality of optimization fitness based on the plurality of first sample training integrated error set analyses, wherein the first sample training integrated error set is inversely related to the optimization fitness;

According to the optimizing fitness, arranging the initial parameter sets from large to small to obtain an initial parameter set sequence;

extracting first K initial parameter sets of the initial parameter set sequence to be set as an optimal solution, and setting last J initial parameter sets to be a secondary solution, wherein J is larger than K, and the sum of J and K is the number of the initial parameter sets;

Clustering J secondary solutions based on K optimal solutions to obtain K clustering solution sets, and adjusting the secondary solutions based on preset updating amplitude in the K clustering solution sets to obtain K updating clustering solution sets;

Identifying the optimal solutions in the K updated cluster solutions and the updated secondary solutions, and updating the optimal solutions based on the secondary solutions if the optimizing fitness of the updated secondary solutions is greater than or equal to the optimizing fitness of the optimal solutions;

iterative optimization is continuously carried out until the threshold value of the optimizing times is met, and the current K clustering solution sets are output;

And respectively calculating the comprehensive optimizing fitness of the K cluster solution sets, and selecting the cluster solution set optimal solutions with the optimal comprehensive optimizing fitness as an optimal parameter set to obtain the optimal initial weight and the optimal initial bias.

Specifically, the method for optimizing the initial weight and the initial bias based on the plurality of first sample training integrated error sets with the weight threshold and the bias threshold as constraints includes that firstly, a plurality of optimizing fitness is determined based on the plurality of first sample training integrated error set analysis, wherein the first sample training integrated error set and the optimizing fitness are inversely related, that is, the smaller the first sample training integrated error set is, the larger the corresponding optimizing fitness is.

And further based on the optimizing fitness, arranging the initial parameter groups from large to small according to the optimizing fitness to generate an initial parameter group sequence. And then extracting the first K initial parameter sets of the initial parameter set sequence to be set as an optimal solution, and setting the last J initial parameter sets to be a sub-solution, wherein the sum of J and K is the number of the initial parameter sets, J is larger than K, namely the number of the sub-solutions is larger than the number of the optimal solutions, and the default number of the sub-solutions is more than 10 times of the number of the optimal solutions. And then carrying out random clustering on J secondary solutions based on the K optimal solutions to obtain K clustering solution sets, wherein the number of the secondary solutions in each clustering solution set is the same. And then in the K cluster solution sets, adjusting the secondary solutions in the cluster solution sets based on a preset update amplitude by taking the optimal solutions as optimizing directions, wherein the preset update amplitude is the adjustment amplitude of each time, and the K update cluster solution sets can be obtained based on actual conditions.

Then, the weight threshold and the bias threshold are used as constraints, updated secondary solutions in the K updated cluster solutions are judged, and if the updated secondary solutions do not meet the weight threshold or the bias threshold, the secondary solutions are not updated; and further identifying the optimal solutions in the K updated cluster solutions and the updated sub-solutions, and updating the optimal solutions based on the sub-solutions if the optimizing fitness of the updated sub-solutions is greater than or equal to the optimizing fitness of the optimal solutions. By utilizing the method, iterative optimization is continuously carried out until the threshold value of the optimizing times is met, the current K clustering solution sets are output, and the threshold value of the optimizing times can be set based on optimizing precision, wherein the higher the optimizing precision is, the larger the threshold value of the optimizing times is. And then respectively calculating comprehensive optimizing fitness of the K cluster solution sets, wherein the comprehensive optimizing fitness is the sum of optimizing fitness of the cluster solution sets and optimizing fitness of a plurality of sub-solutions, setting the cluster solution set optimal solution with the largest comprehensive optimizing fitness as an optimal parameter set, and obtaining the optimizing initial weight and the optimizing initial bias based on the optimal parameter set.

By optimizing the initial weight and the initial bias by using the optimizing algorithm, the global searching capability of the algorithm is strong, so that the condition of being in a local optimal state in the optimizing process can be avoided, and the comprehensiveness, accuracy and reliability of optimizing the initial weight and the initial bias are improved.

Setting the initial fault preset network according to the optimized initial weight and the optimized initial bias, and performing supervision training on the initial fault preset network through the second sample training set to obtain the preset fault prediction network conforming to expected training constraint.

Specifically, the initial fault preset network is set according to the optimized initial weight and the optimized initial bias, then the initial fault preset network is subjected to supervision training through the second sample training set until an expected training constraint is met, the current fault prediction network is output to be set as the preset fault prediction network, wherein the expected training constraint is an expected network prediction accuracy rate, and the initial fault prediction network can be set according to actual requirements.

Further, based on the preset fault deferral scheme, deferral analysis is performed on the fault type and the fault intensity, and a predicted deferral time is determined, and the seventh step of the application includes:

The fault type and the fault strength are input into an operation and maintenance resource library to be matched, and a fault deferral mode set is determined;

identifying the fault deferral mode set based on the preset fault deferral scheme, and determining an effective fault deferral scheme;

And carrying out delay analysis on the fault type and the fault strength based on the effective fault delay scheme through the speed reducer twin model, and generating the predicted delay time.

Specifically, a plurality of sample fault types, a plurality of sample fault intensities and a plurality of sample operation and maintenance schemes are obtained based on historical operation and maintenance data of a target speed reducer, wherein the sample fault types, the sample fault intensities and the sample operation and maintenance schemes are in one-to-one correspondence; and then, based on a decision tree principle, taking the sample fault type and the sample fault strength as child nodes, and taking a corresponding sample operation and maintenance scheme as leaf nodes of the child nodes to construct an operation and maintenance resource library.

Then, the fault type and the fault strength are input into an operation and maintenance resource library for matching, an operation and maintenance scheme is determined based on a matching result, and the operation and maintenance scheme is used as the fault deferral mode set; and then identifying the fault deferral mode set based on the preset fault deferral scheme, and determining an effective fault deferral scheme, wherein the effective fault deferral scheme is a fault deferral mode existing in the preset fault deferral scheme, namely a usable fault deferral mode. And finally, carrying out simulation delay analysis on the fault type and the fault strength based on the effective fault delay scheme through the twin model of the speed reducer to obtain a predicted delay time, wherein the predicted delay time is a time period for fault delay.

Further, the step eight of the present application includes:

and if the compensation prediction node meets the preset node, executing the operation optimization of the target speed reducer based on the effective fault deferral scheme.

Specifically, if the compensation prediction node is equal to or later than the preset node, the equipment can be overhauled and maintained within the preset equipment downtime or the preset overhauling time, so that the influence and loss caused by production interruption are avoided, and the production efficiency of the equipment is improved. And then when the subsequent target speed reducer operates, executing operation optimization of the target speed reducer based on the effective fault deferral scheme, so as to deferred the fault occurrence time.

In summary, the monitoring and protecting method for the speed reducer provided by the application has the following technical effects:

1. The method comprises the steps of building a sensing monitoring matrix based on a fault association characteristic set of a target speed reducer, and monitoring the target speed reducer in real time to obtain a monitoring data stream; then, identifying the monitoring data flow by utilizing a synchronously operated speed reducer twin model, determining abnormal monitoring data, and tracing to determine the change trend of the abnormal data; then, carrying out simulated fault prediction based on the abnormal monitoring data and the abnormal data change trend through a twin sub-model of the speed reducer twin model, and determining simulated fault prediction information; on the other hand, a preset fault prediction network is built based on the BP neural network, the training process of the preset fault prediction network is optimized through an optimizing algorithm, and network fault prediction is carried out through the optimized preset fault prediction network, so that network fault prediction information is generated; further fusing the simulated fault prediction information and the network fault prediction information to obtain fault prediction information, wherein the fault prediction information comprises fault types, fault intensities and fault prediction nodes; performing deferral analysis on the fault type and the fault strength according to a preset fault deferral scheme, generating a predicted deferral time to compensate the fault prediction node, and generating a compensated prediction node; and finally judging whether the compensation prediction node meets a preset node, if not, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer. The method can improve the accuracy of predicting the faults of the speed reducer, discover potential fault threat and accurately predict fault occurrence nodes of the speed reducer in time, effectively avoid production interruption and loss caused by sudden shutdown or major faults of equipment, and ensure the reliability and stability of equipment operation.

2. The model precision of the fault prediction network can be remarkably improved by constructing the fault prediction network of the speed reducer based on the BP neural network and optimizing the initial weight and the initial bias of the fault prediction network by utilizing an optimizing algorithm, so that the precision and the accuracy of the fault prediction of the speed reducer network are further improved.

3. By fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy, the advantages of digital twin fault prediction and BP neural network fault prediction can be combined, and therefore accuracy and reliability of obtaining the fault prediction information are further improved.

Example two

Based on the same inventive concept as the monitoring protection method of a speed reducer in the foregoing embodiment, the present application further provides a monitoring protection system of a speed reducer, referring to fig. 3, the system includes:

The high-frequency fault data set determining module 11 is used for determining a high-frequency fault data set by networking searching with the attribute information and the working scene of the target speed reducer as constraints, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault associated feature sets of the high-frequency fault types;

The target speed reducer monitoring module 12 is used for building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream;

The abnormal data trend set determining module 13 is used for synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set and determining an abnormal data trend set;

The simulated fault prediction module 14 is configured to perform simulated fault prediction based on the anomaly monitoring dataset and the data trend set by using a twin sub-model of the speed reducer twin model, and generate simulated fault prediction information;

the network fault prediction module 15 is configured to predict a network fault based on the anomaly monitoring data set and the data trend set through a predetermined fault prediction network, generate network fault prediction information, build the predetermined fault prediction network based on a BP neural network, and obtain the network fault prediction information by using an optimizing algorithm to supervise and train;

The fault prediction information obtaining module 16, where the fault prediction information obtaining module 16 is configured to fuse the simulated fault prediction information and the network fault prediction information based on a preset fusion policy to obtain fault prediction information, where the fault prediction information includes a fault type, a fault strength, and a fault prediction node;

The compensation prediction node generating module 17 is configured to read a preset fault deferral scheme, perform deferral analysis on the fault type and the fault strength based on the preset fault deferral scheme, determine a prediction deferral time, and compensate the fault prediction node based on the prediction deferral time, so as to generate a compensation prediction node;

The speed reducer overhauling and maintaining module 18 is used for executing the source equipment shutdown of the target speed reducer if the compensation prediction node does not meet the preset node, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.

Further, the high frequency fault data set determination module 11 in the system is further configured to:

Further, the abnormal data trend set determination module 13 in the system is further configured to:

Further, the network failure prediction module 15 in the system is further configured to:

Further, the compensation prediction node generating module 17 in the system is further configured to:

Further, the speed reducer overhaul maintenance module 18 in the system is also configured to:

In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the difference from other embodiments, and the monitoring protection method and specific example of a speed reducer in the first embodiment are also applicable to the monitoring protection system of a speed reducer in this embodiment, and by the foregoing detailed description of the monitoring protection method of a speed reducer, those skilled in the art can clearly know the monitoring protection system of a speed reducer in this embodiment, so that, for brevity of the specification, no further details will be given here. For the system disclosed in the embodiment, since the system corresponds to the method disclosed in the embodiment, the description is simpler, and the relevant points refer to the description of the method section.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the present application and the equivalent techniques thereof, the present application is also intended to include such modifications and variations.

Claims

1. The monitoring and protecting method for the speed reducer is characterized by comprising the following steps of:

Taking attribute information and a working scene of a target speed reducer as constraints, and carrying out networking retrieval to determine a high-frequency fault data set, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault correlation feature sets of the high-frequency fault types;

building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream;

Synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set, and determining an abnormal data trend set;

utilizing a twin sub-model of the speed reducer twin model to perform simulated fault prediction based on the anomaly monitoring dataset and the data trend set, and generating simulated fault prediction information;

Performing network fault prediction based on the anomaly monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm;

Fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node;

Reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the predicted deferral time is determined, and the fault prediction node is compensated based on the predicted deferral time, so that a compensation prediction node is generated;

and if the compensation prediction node does not meet the preset node, stopping the source equipment of the target speed reducer, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.

2. The method of claim 1, wherein the networked searching for determining the high frequency failure data set with the attribute information and the operation scenario of the target speed reducer as constraints comprises:

3. The method of claim 2, wherein synchronously receiving the real-time monitoring data stream via a speed reducer twin model, comparing to determine an anomaly monitoring dataset, and tracing the anomaly monitoring dataset to determine an anomaly data trend set, comprises:

4. The method according to claim 2, wherein the predetermined fault prediction network is built based on a BP neural network, and is obtained by supervised training using an optimization algorithm, comprising:

5. The method of claim 4, wherein optimizing the initial weights and initial biases based on the first sample training set with the weight threshold and bias threshold as optimization spaces comprises:

6. The method of claim 5, wherein optimizing the initial weights and initial biases based on the plurality of first sample training integrated error sets subject to the weight threshold and the bias threshold comprises:

7. A method according to claim 3, wherein deferral analysis of the fault type and the fault strength based on the preset fault deferral scheme, determining a predicted deferral time, comprises:

8. The method of claim 7, wherein the method further comprises:

9. A monitoring and protection system for a speed reducer, characterized by the steps for implementing the method according to any one of claims 1 to 8, said system comprising:

the high-frequency fault data set determining module is used for determining a high-frequency fault data set by taking attribute information and a working scene of a target speed reducer as constraints through networking retrieval, wherein the high-frequency fault data set comprises a plurality of high-frequency fault types and fault associated feature sets of the high-frequency fault types;

The target speed reducer monitoring module is used for building a sensing monitoring matrix based on the fault association characteristic set, and monitoring the target speed reducer through the sensing monitoring matrix to obtain a real-time monitoring data stream;

the abnormal data trend set determining module is used for synchronously receiving the real-time monitoring data flow through a twin model of the speed reducer, comparing and determining an abnormal monitoring data set, tracing the abnormal monitoring data set and determining an abnormal data trend set;

The simulated fault prediction module is used for carrying out simulated fault prediction based on the anomaly monitoring data set and the data trend set by utilizing a twin sub-model of the speed reducer twin model to generate simulated fault prediction information;

The network fault prediction module is used for predicting network faults based on the abnormal monitoring data set and the data trend set through a preset fault prediction network to generate network fault prediction information, wherein the preset fault prediction network is built based on a BP neural network, and is obtained through supervision training by utilizing an optimizing algorithm;

The fault prediction information obtaining module is used for fusing the simulated fault prediction information and the network fault prediction information based on a preset fusion strategy to obtain fault prediction information, wherein the fault prediction information comprises a fault type, a fault strength and a fault prediction node;

The compensation prediction node generation module is used for reading a preset fault deferral scheme, deferring analysis is carried out on the fault type and the fault strength based on the preset fault deferral scheme, the prediction deferral time is determined, compensation is carried out on the fault prediction node based on the prediction deferral time, and a compensation prediction node is generated;

And the speed reducer overhauling and maintaining module is used for executing the shutdown of the source equipment of the target speed reducer if the compensation prediction node does not meet the preset node, and overhauling and maintaining the target speed reducer according to the fault type and the fault strength.