An edge node-based transmission device safety hazard prediction method

Abstract

The application provides an edge node-based transmission equipment safety hazard prediction method, wherein a controller collects component information at the same position of multiple sets of transmission equipment, detects equipment types for different targets, selects appropriate parameters from displacement parameters, wear, temperature, working noise decibels and rotating speed to set a prediction model for the corresponding type of equipment, compares parameters of components at the same position under the same working condition by using the prediction model after pre-processing based on the environment, downloads all historical data of the same type of equipment from the cloud, establishes a special model in combination with the historical data, and determines equipment safety hazards (fault early warning). The application can investigate the consumption of components at the same position under the same working condition, efficiently utilize data by establishing a special model in combination with historical data or a distributed diagnostic expert system, predict the use of equipment in advance, and finely manage the safety status of equipment.

Description

Technical Field

[0001] This invention belongs to the field of monitoring technology, and in particular relates to a method for predicting safety hazards in transmission equipment based on edge nodes. Background Technology

[0002] Currently, the monitoring of transmission equipment usually involves collecting parameters and then having engineers manually determine whether there are any safety hazards, or judging whether there is a risk of failure based on its working efficiency, and judging the magnitude of the safety hazard based on the model. The above methods mainly rely on experience and data as an aid, and the level of intelligence is not high. In addition, the working conditions in different regions are different, and the applicability of the model needs to be adjusted. The model can only be used after it has been learned. At this time, if the safety hazards are not identified in time, the probability of failure will increase.

[0003] Therefore, there is an urgent need for a method to predict safety hazards in transmission equipment based on edge nodes, which can investigate the consumption of components at the same location under the same working conditions. By combining historical data to build a dedicated model, data can be used efficiently to predict equipment usage in advance and to manage equipment safety status in a refined manner. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method for predicting safety hazards in transmission equipment based on edge nodes. In this method, the controller collects information on components at the same location in multiple sets of transmission equipment. For different types of target detection equipment, appropriate parameters are selected from displacement parameters, wear, temperature, operating noise decibels, and rotational speed to set the prediction model for the corresponding type of equipment. After environmental preprocessing, the prediction model is used to compare the parameters of components at the same location under the same working conditions. After downloading all historical data of the same type of equipment from the cloud, a dedicated model is established based on this historical data to determine equipment safety hazards. This invention can investigate the consumption of components at the same location under the same working conditions. By establishing a dedicated model based on historical data, data can be used efficiently to predict equipment usage in advance and to manage equipment safety status in a refined manner.

[0005] The method performs the following steps: S1. Multiple transmission devices of the same model [C1, C2, ..., Cn] are respectively connected to their respective field edge nodes [L1, L2, ..., Ln]; S2. A temperature sensor, a noise monitoring and acquisition sensor, a displacement sensor, and a speed sensor are respectively installed on each transmission device; the noise monitoring and acquisition sensor collects the operating noise decibels of each transmission device. S3. For the transmission device C1, the temperature data, noise decibel data, displacement data and rotation speed data of the transmission device C1 are periodically collected by the field edge node L1, and temperature data sequence T1, noise decibel data sequence X1, displacement data sequence W1 and rotation speed data sequence Z1 are generated. S4. The L1 downloads the transmission equipment operation prediction model from the cloud and inputs the temperature data sequence T1, the noise decibel data sequence X1, the displacement data sequence W1, and the rotational speed data sequence Z1 into the transmission equipment operation prediction model downloaded from the cloud. S5. Output the predicted working results of the transmission equipment of C1.

[0006] Further, S6. Based on the steps of S1-S3, obtain [T2, T3, Tn], [X2, X3, Xn], [W2, W3, Wn], and [Z2, Z3, Zn] corresponding to [C2, C3, Cn].

[0007] Further, after downloading the operation prediction models of different transmission devices for S7. [L2, L3, ..., Ln] respectively, [T2, T3, ..., Tn], [X2, X3, ..., Xn], [W2, W3, ..., Wn], and [Z2, Z3, ..., Zn] corresponding to [C2, C3, ..., Cn] are respectively input into the downloaded models to obtain the operation prediction results of the transmission devices [C2, C3, ..., Cn].

[0008] Furthermore, in S21, the relative positions of different sensors in the transmission device C1 are consistent with the relative positions of different sensors in the transmission device [C2, C3, ..., Cn].

[0009] Further, S31. The working temperature data sequence T1 of C1 is specifically [t11, t12, ..., t1m].

[0010] Further, S32. The noise decibel data sequence X1 of C1 is specifically [x11, x12, ..., x1m].

[0011] Further, S33. The displacement data sequence W1 of C1 is specifically [w11, w12, ..., w1m].

[0012] Further, S33. The rotational speed data sequence Z1 of C1 is specifically [z11, z12, ..., z1m].

[0013] Furthermore, the cloud-downloaded transmission equipment operation prediction models all have self-learning capabilities. After any transmission equipment applies the model, it outputs results and directly feeds the results back to the cloud. The cloud then adjusts the model based on the model and the output results.

[0014] Advantages of this invention: 1. By periodically and automatically collecting four parameters—working temperature, noise level, displacement, and rotational speed—and conducting a preliminary evaluation at the edge nodes, the transmission equipment working prediction model is downloaded from the cloud. The collected parameters are input into the model, which then performs calculations and outputs the results. This invention employs a two-level evaluation: a first-level evaluation at the edge nodes, followed by a second-level evaluation based on the evaluation results, selecting different models to predict the working condition of the transmission equipment. This eliminates the need for manual evaluation and allows for automatic prediction of the transmission equipment's working condition based on a self-learning transmission equipment working prediction model.

[0015] 2. A central control center is set up to control various edge devices on site. After evaluating the different data collected by each edge device, four evaluation models are set up for operating temperature, noise level, displacement, and rotation speed. Different models are selected for operation prediction based on different needs, which can accurately manage transmission equipment. At the same time, multiple transmission equipment operation prediction models can be downloaded simultaneously for the same transmission equipment to predict different types of data, enabling a comprehensive evaluation of equipment operation.

[0016] 3. The solution of the present invention can simultaneously predict safety hazards for multiple transmission devices on site, which is highly efficient. By setting edge nodes, it is easy to accurately locate the transmission device when the safety hazard prediction result is not ideal, and it is easy to find safety hazard information.

[0017] 4. The four models for operating temperature, operating noise decibels, displacement, and rotational speed are all different, with different algorithms and judgment methods. This allows for targeted model design based on the characteristics of different parameters. In addition, the operating temperature model requires the use of rotational speed-related parameters for judgment, while the operating noise decibel model requires the use of displacement parameters. By using different types of data as references and as key factors in prediction and judgment, the safety, reliability, and accuracy of model predictions can be improved.

[0018] 5. In each model judgment, multiple thresholds are set. These thresholds can be continuously adjusted during the model's self-learning process, and the accuracy of safety hazard prediction is continuously increased as the number of samples increases. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of a method for predicting safety hazards in transmission equipment based on edge nodes. Detailed Implementation To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] like Figure 1 As shown, the method performs the following steps: S1. Multiple transmission devices of the same model [C1, C2, ..., Cn] are respectively connected to their respective field edge nodes [L1, L2, ..., Ln]; S2. A temperature sensor, a noise monitoring and acquisition sensor, a displacement sensor, and a speed sensor are respectively installed on each transmission device; the noise monitoring and acquisition sensor collects the operating noise decibels of each transmission device. S3. For the transmission device C1, the temperature data, noise decibel data, displacement data and rotation speed data of the transmission device C1 are periodically collected by the field edge node L1, and temperature data sequence T1, noise decibel data sequence X1, displacement data sequence W1 and rotation speed data sequence Z1 are generated. S4. The L1 downloads the transmission equipment operation prediction model from the cloud and inputs the temperature data sequence T1, the noise decibel data sequence X1, the displacement data sequence W1, and the rotational speed data sequence Z1 into the transmission equipment operation prediction model downloaded from the cloud. S5. Output the predicted working results of the transmission equipment of C1.

[0021] Further, S6. Based on the steps of S1-S3, obtain [T2, T3, Tn], [X2, X3, Xn], [W2, W3, Wn], and [Z2, Z3, Zn] corresponding to [C2, C3, Cn].

[0022] Further, after downloading the operation prediction models of different transmission devices for S7. [L2, L3, ..., Ln] respectively, [T2, T3, ..., Tn], [X2, X3, ..., Xn], [W2, W3, ..., Wn], and [Z2, Z3, ..., Zn] corresponding to [C2, C3, ..., Cn] are respectively input into the downloaded models to obtain the operation prediction results of the transmission devices [C2, C3, ..., Cn].

[0023] Furthermore, in S21, the relative positions of different sensors in the transmission device C1 are consistent with the relative positions of different sensors in the transmission device [C2, C3, ..., Cn].

[0024] Further, S31. The working temperature data sequence T1 of C1 is specifically [t11, t12, ..., t1m].

[0025] The operating temperature data sequences corresponding to [C1, C2, ..., Cn] can form an operating temperature data matrix:

[0026] Further, S32. The noise decibel data sequence X1 of C1 is specifically [x11, x12, ..., x1m].

[0027] The noise decibel data sequences corresponding to [C1, C2, ..., Cn] can form a noise decibel data matrix:

[0028] Further, S33. The displacement data sequence W1 of C1 is specifically [w11, w12, ..., w1m].

[0029] The displacement data sequences corresponding to [C1, C2, ..., Cn] can form a displacement data matrix:

[0030] Further, S33. The rotational speed data sequence Z1 of C1 is specifically [z11, z12, ..., z1m].

[0031] The rotational speed data sequences corresponding to [C1, C2, ..., Cn] can form a rotational speed data matrix:

[0032] Furthermore, the cloud-downloaded transmission equipment operation prediction models all have self-learning capabilities. After any transmission equipment applies the model, it outputs results and directly feeds the results back to the cloud. The cloud then adjusts the model based on the model and the output results.

[0033] Furthermore, the displacement data represents the change in the relative distance between the various connecting components on the transmission device.

[0034] Furthermore, S71. The central control center L controls the [L1, L2, ..., Ln]; S72. The [L1, L2, ..., Ln] are four data matrices formed based on the four data sequences of working temperature, noise decibels, displacement, and rotational speed corresponding to [C1, C2, ..., Cn], respectively, and the changing trends of the four data sequences of working temperature, noise decibels, displacement, and rotational speed collected for each of them are evaluated. S73. Set the operating temperature threshold Yt, noise decibel threshold Yx, displacement threshold Yw, and rotational speed threshold Yz; S74. For any of the aforementioned [L1, L2, ..., Ln], First operating condition: When there are data points in the collected working temperature data where tqp≥Yt is greater than the first threshold, the temperature data sequence is sent to the central control center L, and the Lq that uploaded the abnormal data is notified that the working temperature prediction model needs to be downloaded; where n≥q≥1, m≥p≥1; Second operating condition: When there are data points in the collected noise decibel data where xqp≥Yt is greater than the second threshold, the temperature data sequence is sent to the central control center L, and the abnormal data is uploaded to Lq, which is informed that the noise decibel prediction model needs to be downloaded. Third working condition: When there are data points in the collected displacement data where wqp≥Yt is greater than the third threshold, the temperature data sequence is sent to the central control center L, and the Lq that uploaded the abnormal data is notified that the displacement prediction model needs to be downloaded. Fourth operating condition: When there are data points in the collected rotational speed data where zqp≥Yt is greater than the fourth threshold, the temperature data sequence is sent to the central control center L; and the Lq that uploaded the abnormal data is notified that the rotational speed prediction model needs to be downloaded. The first threshold, the second threshold, the third threshold, and the fourth threshold are set manually.

[0035] If multiple operating conditions exist among the first to fourth operating conditions, the data that matches each operating condition will be categorized and sent to the central control center L. S75. The central control center feeds back to the [L1, L2, ..., Ln] the model identifier and download path of the same or different types of transmission equipment working prediction models that need to be downloaded; There are situations where Lp needs to download the operation prediction models of Class 1, Class 2, Class 3, or Class 4 transmission equipment simultaneously.

[0036] The temperature prediction model is as follows: For any single transmission device, the average speed is

[0037] In the sequence [zq1, zq2, ..., zqm], the value greater than... The number of data points is a; The average operating temperature is

[0038] In the range [tq1, tq2, ..., tqm], the value greater than... The number of data points is b; A is obtained by taking the absolute value of the difference between a and b. If A ≤ temperature prediction threshold, and the two sequences [zq1, zq2···zqm] and [tq1, tq2···tqm] have the same trend, and all data in the sequence [tq1, tq2···tqm] are less than the alarm temperature, then no abnormal result is output; otherwise, an abnormal alarm is issued to the central control center L and the field edge node Lq.

[0039] The noise level prediction model is as follows: If the noise decibel prediction model is invoked, it is determined whether there are outliers in [wq1, wq2, ..., wqm] corresponding to [xq1, xq2, ..., xqm]. If there are abnormal values ​​in [wq1, wq2, ..., wqm], and if xqp is abnormal, the corresponding wqp is also abnormal, then notify the maintenance personnel to inform them that the equipment connection gap is abnormal.

[0040] The displacement prediction model is as follows: If the displacement prediction model is enabled, maintenance personnel will be directly called to retrieve the geographical location of the corresponding transmission equipment and the field edge connection Lq to which it is connected, query the location of the displacement sensor, and check the equipment.

[0041] The speed prediction model is as follows: The control log of the transmission equipment is retrieved. If the speed change changes accordingly with the control command, no abnormal result is output. Otherwise, an abnormal alarm is issued to the central control center L and the field edge node Lq.

[0042] S76. [L1, L2, ..., Ln] Download the model according to the model identifier and download path of the transmission equipment working prediction model, and input the four types of collected data to obtain the output results.

[0043] Optionally, a distributed diagnostic expert system can be used to replace historical data to build a dedicated model, which can efficiently utilize data, predict equipment usage in advance, and manage equipment safety status in a refined manner.

[0044] Advantages of this invention: 1. By periodically and automatically collecting four parameters—working temperature, noise level, displacement, and rotational speed—and conducting a preliminary evaluation at the edge nodes, the transmission equipment working prediction model is downloaded from the cloud. The collected parameters are input into the model, which then performs calculations and outputs the results. This invention employs a two-level evaluation: a first-level evaluation at the edge nodes, followed by a second-level evaluation based on the evaluation results, selecting different models to predict the working condition of the transmission equipment. This eliminates the need for manual evaluation and allows for automatic prediction of the transmission equipment's working condition based on a self-learning transmission equipment working prediction model.

[0045] 2. A central control center is set up to control various edge devices on site. After evaluating the different data collected by each edge device, four evaluation models are set up for operating temperature, noise level, displacement, and rotation speed. Different models are selected for operation prediction based on different needs, which can accurately manage transmission equipment. At the same time, multiple transmission equipment operation prediction models can be downloaded simultaneously for the same transmission equipment to predict different types of data, enabling a comprehensive evaluation of equipment operation.

[0046] 3. The solution of the present invention can simultaneously predict safety hazards for multiple transmission devices on site, which is highly efficient. By setting edge nodes, it is easy to accurately locate the transmission device when the safety hazard prediction result is not ideal, and it is easy to find safety hazard information.

[0047] 4. The four models for operating temperature, noise level, displacement, and rotational speed are all different, with different algorithms and judgment methods. This allows for targeted model design based on the characteristics of different parameters. In addition, the operating temperature model requires the use of rotational speed parameters for judgment, while the noise level model requires the use of displacement parameters. By using different types of data as references and as key factors in prediction and judgment, the safety, reliability, and accuracy of model predictions can be improved.

[0048] 5. In each model judgment, multiple thresholds are set. These thresholds can be continuously adjusted during the model's self-learning process, and the accuracy of safety hazard prediction is continuously increased as the number of samples increases.

[0049] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting safety hazards in transmission equipment based on edge nodes, characterized in that: S1. Multiple transmission devices of the same model [C1, C2, ..., Cn] are respectively connected to their respective field edge nodes [L1, L2, ..., Ln]; S2. A temperature sensor, a noise monitoring and acquisition sensor, a displacement sensor, and a speed sensor are respectively installed on each transmission device; the noise monitoring and acquisition sensor collects the operating noise decibels of each transmission device. S21. In the transmission device C1, the relative positions between different sensors are consistent with the relative positions between different sensors in the transmission device [C2, C3, ..., Cn]; S3. For the transmission device C1, the temperature data, noise decibel data, displacement data and rotation speed data of the transmission device C1 are periodically collected by the field edge node L1, and temperature data sequence T1, noise decibel data sequence X1, displacement data sequence W1 and rotation speed data sequence Z1 are generated. S31. The working temperature data sequence T1 of C1 is specifically [t11, t12, ..., t1m]; S32. The noise decibel data sequence X1 of C1 is specifically [x11, x12, ..., x1m]; S33. The displacement data sequence W1 of C1 is specifically [w11, w12, ..., w1m]; the rotational speed data sequence Z1 of C1 is specifically [z11, z12, ..., z1m]; S4. The L1 downloads the transmission equipment operation prediction model from the cloud and inputs the temperature data sequence T1, the noise decibel data sequence X1, the displacement data sequence W1, and the rotational speed data sequence Z1 into the transmission equipment operation prediction model downloaded from the cloud. S5. Output the predicted working result of the transmission equipment of C1; S6. Based on the steps of S1-S3, obtain [T2, T3, Tn], [X2, X3, Xn], [W2, W3, Wn], and [Z2, Z3, Zn] corresponding to [C2, C3, Cn]. S7. After downloading the operation prediction models for different transmission devices [L2, L3, ..., Ln] respectively, input [T2, T3, ..., Tn], [X2, X3, ..., Xn], [W2, W3, ..., Wn], and [Z2, Z3, ..., Zn] corresponding to [C2, C3, ..., Cn] into the downloaded models respectively to obtain the operation prediction results of the transmission devices [C2, C3, ..., Cn]. S71. The central control center L controls the [L1, L2, ..., Ln]; S72. The [L1, L2, ..., Ln] are four data matrices formed based on the four data sequences of working temperature, noise decibels, displacement, and rotational speed corresponding to [C1, C2, ..., Cn], respectively, and the changing trends of the four data sequences of working temperature, noise decibels, displacement, and rotational speed collected for each of them are evaluated. S73. Set the operating temperature threshold Yt, noise decibel threshold Yx, displacement threshold Yw, and rotational speed threshold Yz; S74. For any of the aforementioned [L1, L2, ..., Ln], First operating condition: When there are data points in the collected working temperature data where tqp≥Yt is greater than the first threshold, the temperature data sequence is sent to the central control center L, and the Lq that uploaded the abnormal data is notified that the working temperature prediction model needs to be downloaded; where n≥q≥1, m≥p≥1; Second working condition: When there are data points in the collected noise decibel data that xqp≥Yt are greater than the second threshold, the noise decibel data sequence is sent to the central control center L, and the Lq that uploaded the abnormal data is notified that the noise decibel prediction model needs to be downloaded. Third working condition: When there are data points in the collected displacement data that are greater than the third threshold, the displacement data sequence is sent to the central control center L, and the Lq that uploaded the abnormal data is notified that the displacement prediction model needs to be downloaded. Fourth operating condition: When there are data points in the collected rotational speed data that are greater than the fourth threshold, the rotational speed data sequence is sent to the central control center L; and the Lq that uploaded the abnormal data is notified that the rotational speed prediction model needs to be downloaded. The first threshold, the second threshold, the third threshold, and the fourth threshold are set manually. If multiple operating conditions exist among the first to fourth operating conditions, the data that matches each operating condition will be categorized and sent to the central control center L. S75. The central control center feeds back to the [L1, L2, ..., Ln] the model identifiers and download paths of the same or different types of transmission equipment operation prediction models that need to be downloaded.

2. The method for predicting safety hazards in transmission equipment based on edge nodes according to claim 1, characterized in that: The cloud-downloaded transmission equipment operation prediction models all have self-learning capabilities. After any transmission equipment applies the model, it outputs results and directly feeds the results back to the cloud. The cloud then adjusts the model based on the model and the output results.

Images