Elevator on-demand maintenance time sequence prediction method and system based on LSTM

Abstract

The invention discloses an elevator on-demand maintenance time sequence prediction method and system based on LSTM. The elevator on-demand maintenance time sequence prediction method comprises the steps that all feature data of an elevator are collected and preprocessed; key features of elevator maintenance are selected from the preprocessed data, weighted values are given according to the importance degrees of the features, true values of all the features are obtained, and a data set is constructed; constructing and training a multivariable LSTM model, and capturing a mutual relation and influence between key features; the input of the model is time sequence data of each feature, and the output of the model is prediction of the maintenance state of the next time step; and elevator characteristic data are collected in real time, preprocessed and then input into the trained multivariable LSTM model, a time sequence prediction result is output, and an elevator maintenance plan is made according to the prediction result. According to the method, the elevator maintenance time sequence is accurately predicted by collecting and analyzing multiple elevator characteristics and combining the time sequence prediction capacity of the multivariable LSTM model.

Description

Technical Field

[0001] The present invention relates to the field of elevator on-demand maintenance, and in particular to an LSTM-based elevator on-demand maintenance time series prediction method and system. Background Art

[0003] Elevator on-demand maintenance is a maintenance method that flexibly arranges maintenance plans based on actual usage and the operating status and needs of the elevator. The traditional regular inspection and maintenance method, which is carried out every 15 days, cannot fully consider the actual operating status and needs of the elevator, which may lead to waste of resources and poor maintenance results. Elevator on-demand maintenance can make personalized maintenance arrangements based on the actual situation of the elevator, and the maintenance cycle can be up to 3 months, thereby improving maintenance results and resource utilization efficiency.

[0004] At present, there are also studies on the prediction of elevator maintenance based on big data analysis. For example, Chinese patent application CN201711168356.6 discloses an elevator on-demand maintenance system and method based on big data analysis, which predicts the risk of failure through elevator fault data and performs on-demand maintenance of elevators based on fault warning. This method provides an energy-saving and time-saving intelligent maintenance idea, but it also has some technical and feasibility defects. First, the fault risk calculation model proposed in the above patent uses less data: the data 1 hour before the fault time point is brought into the fault analysis prediction model to obtain the fault risk coefficient within 5 minutes before the fault occurs; secondly, there are many fault factors involved in the elevator, and the mechanical fault analysis needs to consider multi-factor dependence. A single fault data analysis method will cause the safety deviation of the entire elevator maintenance. Therefore, there is a need for a method that can predict the maintenance time sequence based on the multi-dimensional maintenance factors of the elevator and the long-term actual operation status data, so as to improve the maintenance efficiency and reduce the risk of failure. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a method and system for predicting elevator on-demand maintenance time series based on LSTM.

[0006] The specific technical solutions are as follows:

[0007] An LSTM-based elevator on-demand maintenance time series prediction method includes the following steps:

[0008] S1: Collect various characteristic data of the elevator and perform preprocessing;

[0009] S2: Select the key features of elevator maintenance from the preprocessed data and assign weight values ​​according to the importance of the features to obtain the true value of each feature and construct a data set;

[0010] S3: Build a multivariate LSTM model to capture the interrelationships and impacts between key features. The input of this model is the time series data of each feature, and the output is the prediction of the maintenance status at the next time step.

[0011] S4: Use the data set constructed in S2 to train the multivariate LSTM model to obtain a trained multivariate LSTM model;

[0012] S5: Collect elevator feature data in real time, pre-process it using the method in S1, input it into the trained multivariate LSTM model, output the time series prediction results, and formulate an elevator maintenance plan based on the prediction results.

[0013] Furthermore, in S1, the various characteristic data of the elevator include: machine room environment, operating device status, drive host vibration, brake status, encoder status.

[0014] Furthermore, in S1, the preprocessing includes: data cleaning, missing value processing, and outlier processing.

[0015] Furthermore, the multivariate LSTM model includes five layers: an input layer, an LSTM layer one, an LSTM layer two, a fully connected layer, and an output layer;

[0016] The input layer is used to accept time series data of multiple features as an input sequence;

[0017] The LSTM layer 1 includes multiple LSTM units for capturing the basic temporal relationship between features; the number of LSTM units is between 20 and 100;

[0018] The LSTM layer 2 includes multiple LSTM units for further extracting deeper temporal relationships between features; the number of LSTM units is between 20 and 100;

[0019] The fully connected layer is used to extract potential interaction relationships between input features;

[0020] The output of the fully connected layer is connected to the output layer, and the number of neurons in the output layer is determined according to actual needs, including using one neuron for binary classification, the basis of which is whether maintenance is required; or using multiple neurons for multi-classification, the basis of which is maintenance priority.

[0021] Furthermore, in S4, during the training process of the multivariate LSTM model, a mean square error loss function and a stochastic gradient descent optimization algorithm are used to optimize model parameters.

[0022] Furthermore, the elevator maintenance plan includes: maintenance time nodes, maintenance contents, and maintenance priorities.

[0023] Furthermore, the maintenance priorities include three categories: high priority, medium priority, and low priority; for high-priority elevator maintenance needs, personnel are arranged to complete the maintenance of the elevator within 24 hours; for medium-priority elevator maintenance needs, personnel are arranged to complete the maintenance of the elevator within one week; for low-priority elevator maintenance needs, personnel are arranged to complete the maintenance of the elevator within two weeks.

[0024] An LSTM-based elevator on-demand maintenance time series prediction system, used to implement the LSTM-based elevator on-demand maintenance time series prediction method, including: a data collection and preprocessing module, an elevator maintenance demand prediction module, a maintenance plan formulation module, and a multivariate LSTM model update module;

[0025] The data collection and preprocessing module is used to collect user elevator characteristic data in real time and perform preprocessing;

[0026] The elevator maintenance demand prediction module is used to input the data output by the data collection and preprocessing module into the trained multivariate LSTM model to obtain a time series prediction result; the trained multivariate LSTM model is the output of the multivariate LSTM model update module;

[0027] The maintenance plan formulation module is used to formulate an elevator maintenance plan according to the time series prediction results, including maintenance time nodes, maintenance contents and maintenance priorities;

[0028] The multivariate LSTM model update module is used to periodically obtain user historical elevator maintenance data from the database, input it into the multivariate LSTM model after preprocessing, and output the trained multivariate LSTM model after optimizing and updating the model.

[0029] The beneficial effects of the present invention are:

[0030] (1) The present invention can achieve accurate prediction: by collecting and analyzing multiple elevator features and combining the time series prediction capability of the multivariate LSTM model, accurate prediction of the elevator maintenance time series can be achieved.

[0031] (2) The present invention can realize on-demand maintenance: the maintenance timing is predicted according to the actual operating status of the elevator, thus avoiding the resource waste and low efficiency of fixed-cycle maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the LSTM-based elevator on-demand maintenance time series prediction method proposed in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The present invention will be described in detail below according to the accompanying drawings and preferred embodiments, and the purpose and effect of the present invention will become more clear. The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0034] like Figure 1 As shown, a method for predicting elevator maintenance time series based on LSTM includes the following steps:

[0035] S1: Collect various characteristic data of the elevator and perform preprocessing.

[0036] The characteristic data of the elevator include: machine room environment, operating device status, drive host vibration, brake status, encoder status, etc.

[0037] Preprocessing includes: data cleaning, missing value processing, outlier processing, etc. It is used to ensure the quality and integrity of the data.

[0038] In this embodiment, the data in the quarterly maintenance list of an elevator company is taken as an example to illustrate the content of the present invention. Among them, the maintenance items are used as characteristic data, and the results of each item are divided into three situations: normal (meets the requirements, the test and experimental functions are normal), abnormal (found to be non-compliant, need to adjust or replace parts for processing), and none (the elevator does not have this item or function). According to the needs of elevator maintenance, the characteristic value of the characteristic data with abnormal results is scored as 1, and the characteristic value of the characteristic data with normal or none results is scored as 0. That is, if the item score reaches 1, it means that the item needs maintenance.

[0039] S2: Select the key features of elevator maintenance from the preprocessed data and assign weight values ​​according to the importance of the features. According to the scoring rules given in S1, multiply the weight value by the score as the true value of the feature (i.e., label value). The set of true values ​​of all features is used to construct the data set.

[0040] In this embodiment, 45 key features of elevator maintenance are extracted, as shown in Table 1 below.

[0041] Table 1 Elevator maintenance feature number table

[0042]

[0043] In Table 1, the requirements for judging the "machine room and pulley room environment" numbered 1 as normal are cleanliness, intact doors and windows, and normal lighting; the requirements for the "manual emergency operation device status" numbered 2 are that the device is complete and located at the specified position; the requirements for the "drive host status" numbered 3 are that there is no abnormal vibration and abnormal sound during operation; the requirements for the "brake clearance status" numbered 5 are that there should be no friction between the brake lining and the brake wheel when it is opened, and the clearance value meets the requirements of the manufacturer; the "brake self-monitoring status" numbered 6 specifically refers to the self-monitoring status of the brake when it is used as the stopping subsystem of the car accidental movement protection device; The requirement for "encoder status" No. 7 is to be clean and firmly installed; the requirement for "oil cup status on guide shoe" No. 13 specifically refers to whether the oil amount in the oil cup is appropriate and whether there is any leakage; the requirement for "engagement length of floor door locking element" No. 29 is to be no less than 7mm; the requirement for "pit environment status" No. 30 is to be clean, free of water seepage and accumulation, and normal lighting; the requirement for "reducer lubricating oil status" No. 33 is to have an appropriate amount of lubricating oil and no leakage; the requirement for "traction wheel groove and suspension device cleanliness, wire rope status" No. 37 is to be free of severe grease, have uniform tension, and meet the requirements of the manufacturing unit.

[0044] According to the importance of these key features to elevator maintenance, they are divided into three levels: A, B, and C. The weight is 0.2 for Class A features, 0.5 for Class B features, and 1 for Class C features, to reflect the importance of the features to maintenance time series prediction. The classification is shown in Table 2 below.

[0045] Table 2 Elevator maintenance feature classification and weight assignment table

[0046]

[0047] S3: Build a multivariate LSTM model to capture the interrelationships and impacts between the selected key features of elevator maintenance; use the time series data of each feature as the input sequence to predict the maintenance status of the next time step.

[0048] When building a multivariate LSTM model, it is first necessary to determine the number of layers of the network and the number of LSTM units (i.e., the number of neurons) in the LSTM layer, because a deeper network and more neurons can usually better capture the temporal relationship between features, but also require more computing resources and longer training time. In this embodiment, the multivariate LSTM model includes five layers in sequence: an input layer, an LSTM layer 1, an LSTM layer 2, a fully connected layer, and an output layer.

[0049] Input layer: used to accept time series data with multiple features as input sequence.

[0050] LSTM layer 1: contains multiple LSTM units, which are used to capture the basic temporal relationship between features. The number of LSTM units can be set according to needs, generally between 20 and 100.

[0051] LSTM layer 2: also contains multiple LSTM units, which are used to further extract deeper temporal relationships between features; the number of LSTM units can be the same as or different from that of LSTM layer 1, and the number of LSTM units is also between 20 and 100.

[0052] Fully connected layer: used to extract potential interactions between input features.

[0053] Output layer: Connect the output of the fully connected layer to the output layer. The number of neurons in the output layer is determined according to actual needs. For example, one neuron can be used for binary classification (need maintenance and no maintenance) or multiple neurons can be used for multi-classification (maintenance priority).

[0054] S4: Use the data set constructed in S2 to train the multivariate LSTM model to obtain a trained multivariate LSTM model. The model is trained to learn the temporal relationship between features to accurately predict the maintenance needs of elevators. Specifically, the time series data of each feature is used as an input sequence and fed into the multivariate LSTM model; in the multivariate LSTM, the sequence of each feature is input into a different input channel or feature channel, so that the model can consider the temporal relationship between multiple features at the same time.

[0055] Furthermore, the dataset is divided into a training set and a validation set in a ratio of 7:3. The multivariate LSTM model is trained using the data from the training set. This process uses appropriate loss functions and optimization algorithms, such as the mean squared error (MSE) loss function and the stochastic gradient descent (SGD) optimization algorithm. During the training process, the network parameters are updated through the backpropagation algorithm based on the input features and true values ​​of the training set, so that the model gradually learns the temporal relationship between the features. The validation set is used to evaluate the performance and accuracy of the model for tuning and verification.

[0056] Among them, the MSE loss function is used to measure the difference between the model prediction value and the actual value. The expression is as follows:

[0057]

[0058] In the formula, n represents the number of training samples, y predRepresents the predicted value of the model, y true Represents the true value of the sample.

[0059] The SGD optimization algorithm is a commonly used optimization algorithm used to update the parameters of the model to minimize the loss function. The expression is as follows:

[0060]

[0061] In the formula, θ represents the parameters of the model, θ′ represents the updated model parameters, and learning rate represents the learning rate, L represents the loss function, namely the MSE loss function; Represents the gradient of the loss function with respect to the model parameters.

[0062] Back propagation algorithm: It is used to calculate the gradient of the loss function to the model parameters in order to update the parameters. It transfers the gradient layer by layer through the chain rule, and calculates the gradient backward from the output layer to the input layer.

[0063] The training process is to input the data of the training set into the model for forward propagation and calculate the predicted value for each training iteration (epoch). According to the predicted value and the actual value, the value of the loss function is calculated. The gradient is calculated using the back propagation algorithm, and the parameters of the model are updated according to the optimization algorithm. The above steps are repeated until the preset number of iterations is reached or the loss function converges.

[0064] Validation Process: After each training iteration, the validation set data is used to input the model for forward propagation and calculate the predicted value. Based on the predicted value and the actual value of the validation set, the value of the loss function on the validation set is calculated to evaluate the performance of the model. If the loss function is less than the threshold, the performance of the model is considered sufficient, otherwise it is trained again.

[0065] S5: Collect elevator feature data in real time, pre-process it using the method in S1, input it into the trained multivariate LSTM model, output the time series prediction results, and formulate an elevator maintenance plan based on the prediction results. The elevator maintenance plan includes: maintenance time node, maintenance content, maintenance priority, etc.

[0066] In this embodiment, the multivariate LSTM model classifies elevator maintenance needs into three categories according to priority: high priority, medium priority, and low priority. For high-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within 24 hours; for medium-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within one week; for low-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within two weeks.

[0067] In order to realize the above-mentioned LSTM-based elevator on-demand maintenance time series prediction method, an embodiment of the present invention also proposes an LSTM-based elevator on-demand maintenance time series prediction system, including: a data collection and preprocessing module, an elevator maintenance demand prediction module, a maintenance plan formulation module, and a multivariate LSTM model update module.

[0068] The data collection and preprocessing module is used to collect user elevator characteristic data in real time and perform preprocessing.

[0069] The elevator maintenance demand prediction module is used to input the data output by the data collection and preprocessing module into the trained multivariate LSTM model to obtain the time series prediction results; the trained multivariate LSTM model is the output of the multivariate LSTM model update module.

[0070] The maintenance plan making module is used to make elevator maintenance plans based on the time series prediction results, including maintenance time nodes, maintenance contents and maintenance priorities.

[0071] The multivariate LSTM model update module will periodically obtain the user's historical elevator maintenance data from the database, input it into the multivariate LSTM model after preprocessing, and optimize and update the model.

[0072] Those skilled in the art can understand that the above are only preferred examples of the invention and are not intended to limit the invention. Although the invention is described in detail with reference to the above examples, those skilled in the art can still modify the technical solutions recorded in the above examples or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, etc. made within the spirit and principle of the invention shall be included in the protection scope of the invention.

Claims

1. A method for predicting elevator maintenance time series based on LSTM, characterized in that: The following steps are involved: S1: Collect various characteristic data of the elevator and perform preprocessing; S2: Select the key features of elevator maintenance from the preprocessed data and assign weight values ​​according to the importance of the features to obtain the true value of each feature and construct a data set; S3: Build a multivariate LSTM model to capture the interrelationships and impacts between key features. The input of this model is the time series data of each feature, and the output is the prediction of the maintenance status at the next time step. S4: Use the data set constructed in S2 to train the multivariate LSTM model to obtain a trained multivariate LSTM model; S5: Collect elevator feature data in real time, pre-process it using the method in S1, input it into the trained multivariate LSTM model, output the time series prediction results, and formulate an elevator maintenance plan based on the prediction results.

2. The elevator on-demand maintenance time series prediction method based on LSTM according to claim 1 is characterized in that: In S1, the various characteristic data of the elevator include: machine room environment, operating device status, drive host vibration, brake status, encoder status.

3. The elevator on-demand maintenance time series prediction method based on LSTM according to claim 1 is characterized in that: In S1, the preprocessing includes: data cleaning, missing value processing, and outlier processing.

4. The elevator on-demand maintenance time series prediction method based on LSTM according to claim 1 is characterized in that: The multivariate LSTM model includes five layers: input layer, LSTM layer 1, LSTM layer 2, fully connected layer, and output layer; The input layer is used to accept time series data of multiple features as an input sequence; The LSTM layer 1 includes multiple LSTM units for capturing the basic temporal relationship between features; the number of LSTM units is between 20 and 100; The LSTM layer 2 includes multiple LSTM units for further extracting deeper temporal relationships between features; the number of LSTM units is between 20 and 100; The fully connected layer is used to extract potential interaction relationships between input features; The output of the fully connected layer is connected to the output layer, and the number of neurons in the output layer is determined according to actual needs, including using one neuron for binary classification, the basis of which is whether maintenance is required; or using multiple neurons for multi-classification, the basis of which is maintenance priority.

5. The elevator on-demand maintenance time series prediction method based on LSTM according to claim 1 is characterized in that: In S4, during the training process of the multivariate LSTM model, a mean square error loss function and a stochastic gradient descent optimization algorithm are used to optimize model parameters.

6. The LSTM-based elevator on-demand maintenance time series prediction method according to claim 1 is characterized in that: The elevator maintenance plan includes: maintenance time nodes, maintenance content, and maintenance priorities.

7. The LSTM-based elevator on-demand maintenance time series prediction method according to claim 6 is characterized in that: The maintenance priorities include three categories: high priority, medium priority, and low priority. For high-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within 24 hours; for medium-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within one week; for low-priority elevator maintenance needs, personnel will be arranged to complete the maintenance of the elevator within two weeks.

8. An elevator on-demand maintenance time series prediction system based on LSTM, used to implement the elevator on-demand maintenance time series prediction method based on LSTM according to any one of claims 1 to 7, characterized in that: include: Data collection and preprocessing module, elevator maintenance demand prediction module, maintenance plan formulation module, multivariate LSTM model update module; The data collection and preprocessing module is used to collect user elevator characteristic data in real time and perform preprocessing; The elevator maintenance demand prediction module is used to input the data output by the data collection and preprocessing module into the trained multivariate LSTM model to obtain a time series prediction result; the trained multivariate LSTM model is the output of the multivariate LSTM model update module; The maintenance plan formulation module is used to formulate an elevator maintenance plan according to the time series prediction results, including maintenance time nodes, maintenance contents and maintenance priorities; The multivariate LSTM model update module is used to periodically obtain user historical elevator maintenance data from the database, input it into the multivariate LSTM model after preprocessing, and output the trained multivariate LSTM model after optimizing and updating the model.

Images