Machine room fault prediction method and device, nonvolatile storage medium and electronic equipment

Abstract

The invention discloses a machine room fault prediction method and device, a nonvolatile storage medium and electronic equipment. The method comprises the steps that temperature detection data are acquired, the temperature detection data comprise at least one of detection time, a detection place, a temperature value and a machine room identifier, and the machine room identifier is an identifier of a machine room corresponding to the temperature detection data; determining the prediction priority of the temperature detection data according to the machine room identifier of the temperature detection data; inputting the temperature detection data into a temperature prediction model according to a prediction priority sequence, and obtaining a temperature prediction value output by the temperature prediction model; and when the temperature prediction value is higher than the preset temperature threshold value, determining that the detection site corresponding to the temperature detection data has a fault. The technical problems that the detection workload is large and the fault is not found in time due to the fact that an existing machine room fault detection technology depends on manpower are solved.

Description

Technical Field

[0001] The present application relates to the field of communication room operation, and specifically, to a method, device, non-volatile storage medium and electronic equipment for predicting room faults. Background Art

[0002] With the vigorous development of IDC (Internet Data Center) business, more and more core businesses have entered IDC machine buildings. These machine buildings bear provincial and even regional business loads. The load of a single rack has soared from the traditional 3-4KW to more than 12KW. High power means high heat generation. The time from the generation of temperature rise alarms in the equipment room to the equipment downtime due to high temperature is getting shorter and shorter.

[0003] The timely generation and prediction of temperature rise alarms for equipment in the computer room has always been an important part of ensuring stable business operations. Traditional sites have been formulating emergency plans when various temperature rise alarms occur, so as to reduce the ambient temperature of business equipment as soon as possible to ensure that equipment business is not interrupted.

[0004] In traditional operations, the prediction of temperature rise in the station room is mainly based on manual correlation analysis. After correlation with the changes in outdoor ambient temperature before and after, the deployment of air conditioners inside the station, and on-site experience, it is possible to predict in advance that a temperature rise alarm may occur, and make plans in advance to reduce reaction time, help maintenance personnel respond quickly, and ensure stable operation of business equipment. However, this method requires a lot of manual experience and is combined with weather conditions for prediction. The prediction method is highly dependent on the professional ability of the maintenance personnel themselves and their familiarity with the layout of the room. It cannot cover a large range and can only be specially deployed in key machine buildings. There is an urgent need for a room temperature rise prediction method that can be quickly, simply, and batch deployed to assist maintenance personnel in timely discovering hidden dangers and preparing in advance.

[0005] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiments of the present application provide a method, device, non-volatile storage medium and electronic device for predicting a computer room fault, so as to at least solve the technical problems of heavy detection workload and untimely fault detection caused by the existing computer room fault detection technology relying on manual work.

[0007] According to one aspect of an embodiment of the present application, a method for predicting a computer room fault is provided, comprising: obtaining temperature detection data, wherein the temperature detection data comprises at least one of the following: detection time, detection location, temperature value, and computer room identification, wherein the computer room identification is an identification of the computer room corresponding to the temperature detection data; determining a prediction priority of the temperature detection data based on the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the order in which a temperature prediction model processes the temperature detection data, the temperature prediction model being trained based on historical temperature detection data, and the historical temperature detection data being stored in a data lake; inputting the temperature detection data into the temperature prediction model based on the prediction priority order, and obtaining a temperature prediction value output by the temperature prediction model; and determining, when the temperature prediction value is higher than a preset temperature threshold, that a fault exists at the detection location corresponding to the temperature detection data.

[0008] Optionally, determining the predicted priority of the temperature detection data based on the computer room identification of the temperature detection data includes: obtaining computer room priority data, wherein the computer room priority data includes the computer room identification and the computer room priority corresponding to the computer room identification, and the computer room priority is determined based on the importance of each computer room and the historical failure probability of the computer room; determining the computer room priority corresponding to the temperature detection data based on the computer room identification; and using the computer room priority corresponding to the temperature detection data as the priority of the temperature detection data.

[0009] Optionally, after obtaining the temperature detection data, the method also includes: removing null values ​​in the temperature detection data; testing the temperature detection data according to a preset value interval, and removing the temperature detection data that exceeds the preset value interval; performing a unit root test on the temperature detection data; and performing differential processing on the temperature detection data whose unit root test result is that the unit root test passes.

[0010] Optionally, before inputting the temperature detection data into the temperature prediction model according to the prediction priority order, the method also includes: inputting historical temperature detection data into the temperature prediction model for training, wherein the temperature prediction model is an autoregressive moving average model; obtaining the temperature prediction result output by the temperature prediction model, and ending the training of the temperature prediction model when the accuracy of the temperature prediction result is greater than a preset threshold.

[0011] Optionally, before inputting the historical temperature detection data into the temperature prediction model, the method also includes: generating an autocorrelation graph and a partial autocorrelation graph based on the historical temperature detection data, wherein the autocorrelation graph is used to determine a first input parameter of the temperature prediction model, and the partial autocorrelation graph is used to determine a second input parameter of the temperature prediction model.

[0012] Optionally, after obtaining the temperature prediction value output by the temperature prediction model, the method also includes: obtaining an actual temperature value corresponding to the temperature prediction value, wherein the prediction time point of the temperature prediction value and the detection time point of the actual temperature value are the same; evaluating the prediction success rate of the prediction model based on the error between the temperature prediction value and the actual temperature value, wherein the error includes a mean square error and a root mean square error; when the error is less than a preset error threshold, determining that the evaluation result of the prediction model is a correct prediction; when the error is less than a preset error threshold, determining that the evaluation result of the prediction model is an incorrect prediction.

[0013] Optionally, the method further includes: retraining the temperature prediction model when the proportion of prediction errors in the evaluation results is greater than a preset error threshold.

[0014] Optionally, after determining that a fault exists at a detection location corresponding to the temperature detection data, the method further includes: sending fault alarm information to a hidden danger management platform, wherein the hidden danger management platform is used to generate an inspection work order based on the fault alarm information.

[0015] According to another aspect of an embodiment of the present application, a computer room fault prediction device is also provided, including: a first acquisition module, used to acquire temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, and the computer room identification is the identification of the computer room corresponding to the temperature detection data; a first determination module, used to determine the prediction priority of the temperature detection data based on the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by the temperature prediction model, and the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in a data lake; the first prediction module, used to input the temperature detection data into the temperature prediction model according to the prediction priority order, and obtain the temperature prediction value output by the temperature prediction model; the second determination module, used to determine that there is a fault at the detection location corresponding to the temperature detection data when the temperature prediction value is higher than a preset temperature threshold.

[0016] According to another aspect of an embodiment of the present application, a non-volatile storage medium is provided, in which a program is stored, wherein when the program is running, a device where the non-volatile storage medium is located is controlled to execute a computer room fault prediction method.

[0017] According to another aspect of an embodiment of the present application, there is further provided an electronic device, including: a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the computer room fault prediction method is executed when the program is run.

[0018] According to another aspect of an embodiment of the present application, a computer program product is also provided, including a computer program, and when the computer program is executed by a processor, a method for predicting a computer room failure is implemented.

[0019] In an embodiment of the present application, temperature detection data is obtained, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, wherein the computer room identification is the identification of the computer room corresponding to the temperature detection data; the prediction priority of the temperature detection data is determined based on the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by the temperature prediction model, and the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in a data lake; the temperature detection data is input into the temperature prediction model according to the prediction priority order, and the temperature prediction value output by the temperature prediction model is obtained; when the temperature prediction value is higher than the preset temperature threshold, it is determined that there is a fault at the detection location corresponding to the temperature detection data, and the purpose of discovering the temperature rise fault of the station in advance is achieved by automatically collecting data and predicting the temperature rise according to the priority based on the ARIMA model, thereby achieving the technical effect of reducing manual intervention, improving prediction accuracy and fault response speed, and thus solving the technical problems of large detection workload and untimely fault discovery caused by the reliance on manual labor in the existing computer room fault detection technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0021] Figure 1 is a structural schematic diagram of a computer terminal provided according to an embodiment of the present application;

[0022] Figure 2 It is a flow chart of a method for predicting a computer room failure according to an embodiment of the present application;

[0023] Figure 3 It is a comparison diagram of the original sample data and the effects after first-order and second-order differences provided in an embodiment of the present application;

[0024] Figure 4 is an ACF diagram of sample data provided according to an embodiment of the present application;

[0025] Figure 5 is a PACF diagram of sample data provided according to an embodiment of the present application;

[0026] Figure 6 is a schematic diagram of original and predicted data of sample data provided according to an embodiment of the present application;

[0027] Figure 7 It is a structural schematic diagram of a device for predicting a computer room fault according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] In order to better understand the embodiments of the present application, the technical terms involved in the embodiments of the present application are explained as follows:

[0031] ARIMA model: A statistical method for analyzing and predicting time series. It can capture autoregression, differencing, and moving average in time series. Its full name is the difference autoregression moving average model, which consists of three main components: autoregression (AR), differencing (I), and moving average (MA).

[0032] Temperature rise: Temperature rise refers to the temperature of each component in an electronic device above the ambient temperature. After the device is powered on, due to the thermal effect of the current, the temperature of the conductor surface will gradually rise until it stabilizes. Generally, the temperature rise in a stable state refers to the temperature difference before and after the test point within a certain time interval.

[0033] The traditional method of monitoring temperature rise in the equipment room has the following three problems:

[0034] 1. Relying on the personnel's own ability and experience, the workload is large:

[0035] Currently, operators discover equipment temperature rise mainly based on their own experience. They first find the local high-temperature points in each machine building, then conduct manual inspections, combine traditional network management manual measurement methods, and rely on long-term station maintenance experience to determine whether there will be temperature rise risks in the future. In this way, the frequency of manual inspections is increased, for example, 3 or 6 times a day, so as to intervene as soon as possible when temperature rise occurs. However, this will result in a large manual workload, especially when there are multiple local high-temperature points in traditional old machine buildings. Manual inspections are prone to negligence, which can easily result in real hidden dangers not being reported in a timely manner, delaying the best processing time.

[0036] 2. Without a scientific and objective model to support it, the success rate is unstable:

[0037] The maintenance personnel of the machine room have long used their own experience and used conventional text operation tools such as xls tables and word to perform limited logical calculations. Moreover, since everyone's experience is different, errors are inevitable even after case sharing, resulting in each machine room having a unique and personalized logic. These models formed based on their own experience are purely based on experience and are not supported by scientific and objective models, so they are not very stable. As a result, the maintenance personnel increased the manpower but found that no temperature rise faults were found, resulting in a waste of manpower. The limited manpower cannot be invested in the inspection and maintenance work that really needs attention, resulting in a waste of personnel.

[0038] 3. Without system support, it is impossible to iterate on a large scale and the effect cannot be improved:

[0039] In the current hidden danger system, dispatching is based on certain values, and the predicted data is not used for production. It is impossible to store effective comparison sample data to help the model to continuously iterate and improve its efficiency and success rate. Therefore, a prediction model that can support autonomous configuration, rapid deployment, storage, and presentation is needed, which can truly help maintenance personnel to continuously learn and continuously improve the success rate in their day-to-day maintenance work, and ultimately be used for production prediction.

[0040] Due to the above reasons, the current methods for discovering temperature rise faults in substations have the problems of high success rate reliance on experience and large workload, low success rate without scientific models, and inability to iterate and upgrade without system support.

[0041] In order to solve the above problems, relevant solutions are provided in the embodiments of the present application, which are described in detail below.

[0042] According to an embodiment of the present application, a method embodiment of a computer room fault prediction method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0043] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a method for predicting a computer room failure. Figure 1 As shown, the computer terminal 10 may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0044] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10. As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0045] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the computer room fault prediction method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned computer room fault prediction method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0046] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0047] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0048] In the above operating environment, the present application embodiment provides a method for predicting a computer room failure, such as Figure 2 As shown, the method comprises the following steps:

[0049] Step S202, obtaining temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, where the computer room identification is an identification of the computer room corresponding to the temperature detection data.

[0050] Optionally, full-volume data collection is performed on the relevant machine buildings, stations, equipment, and signals at regular intervals, and an ARIMA prediction model is established to automatically predict temperature rise hazards. In the application embodiment of this scheme, the new generation of configurable capabilities is utilized to flexibly and autonomously configure one's own atomic capabilities in operation and maintenance, and preprocess, change, and verify the data. Ultimately, early warning of temperature rise hazards in stations can be achieved, and early warnings can be graphically presented in the intelligent workbench. In the early stage, it is necessary to ensure that the data flow interfaces between the new generation systems are connected to the network so that the interfaces can achieve barrier-free mutual access. The following will be combined with the example of the analysis, prediction, and presentation of the temperature rise in one of the machine rooms to explain this method in detail. The specific implementation process is as follows:

[0051] 1. Extract the station computer room information from the big data lake fixed asset table as Table 1, including: computer room name, computer room code, station to which it belongs, level, whether it is IDC, and star rating information.

[0052] For example: A site XX room, NJ.250JSNJXX01, A1 site_B_XX office, B, no, none;

[0053] 2. Obtain the computer room topology information from the big data lake computer building construction design table as Table 2, including: computer room name, computer room code, affiliated station, area, drawings, and air conditioning location information.

[0054] For example: A site XX room, NJ.250JSNJXX01, A1 site_B_XX office, 300 square meters, xxx.jpg, xxx.jpg;

[0055] 3. Obtain the monitoring device location information from the Big Data Lake Network Resource Fusion Library XML table as Table 3, including: device name, device code, collector MAC, and IP address.

[0056] For example: intelligent temperature sensor 01, ZNWDCGQ01, XX:DD:WE:SS:EW, XXX.146.124.23;

[0057] 4. Obtain the standardized table of dynamic environment monitoring points from the big data lake standardized configuration specification as Table 4, including: signal name, SPID, device to which it belongs, and system to which it belongs.

[0058] For example: ambient temperature, 231200010010, intelligent temperature sensor 01, XX bureau XX computer room system 01;

[0059] 5. Associate the above-mentioned Table 1 and Table 2 through the computer room code to form a priority definition table for computer room maintenance. The fields in the table include: computer room name, computer room code, affiliated station, level, whether it is IDC, star rating, area, drawings, and air conditioning location.

[0060] For example: A site XX room, NJ.250JSNJXX01, A1 site_B_XX office, B, no, none, 300 square meters, xxx.jpg, xxx.jpg;

[0061] 6. Associate Table 3 and Table 4 according to the device code to form a collection input parameter table with the device as the dimension. Currently, the attributes in this table include: signal name, SPID, system, device name, device code, collector MAC, IP, which are used as input parameters for real-time and historical data.

[0062] For example: ambient temperature, 231200010010, XX computer room system 01, intelligent temperature sensor 01, XX:DD:WE:SS:EW, XXX.146.124.23;

[0063] 7. Establish a station temperature acquisition module, use the collection input parameter table in the previous step, call the unified acquisition and control data to collect the temperature data of each temperature sensor in the station (i.e., temperature detection data), and store it in the collection table A. The collection table attributes are: signal SPID, collector MAC, computer room name (computer room identification), station name (detection location), collection time (detection time), and temperature value.

[0064] For example: 231200010010, XX:DD:WE:SS:EW, A site XX computer room, A1 site_B_XX bureau, 2024-07-01, 29.3;

[0065] In the technical solution provided in step S202, after obtaining the temperature detection data, the method also includes: removing null values ​​in the temperature detection data; testing the temperature detection data according to a preset value interval, and removing the temperature detection data that exceeds the preset value interval; performing a unit root test on the temperature detection data; and performing differential processing on the temperature detection data whose unit root test result is that the unit root test passes.

[0066] Optionally, a data cleaning preprocessing module is established to process missing values ​​in the collection table A and remove null values. The data in table A is further processed for outliers, abnormally large or abnormally small values ​​are removed, and finally an ADF test (unit root test) is performed to verify whether it is stable.

[0067] For example: The output result of the ADF test for this sample is ADF Statistic: -4.012176, p-value: 0.001348;

[0068] Optionally, calculate the first and second order differences of collection table A, and observe the stable trend again. You can use the following statement to call it and use matplotlib.pyplot to draw a comparison chart.

[0069] For example, the local sample uses the first-order difference as follows:

[0070] 2022-07-19NaN

[0071] 2022-07-20 0.9

[0072] 2022-07-21-1.4

[0073] 2022-07-22 0.6

[0074] 2022-07-23 0.9 ...

[0076] 2024-07-11 0.4

[0077] 2024-07-12-0.2

[0078] 2024-07-13 0.4

[0079] 2024-07-14-0.3

[0080] 2024-07-15 0.0

[0081] 10. Based on the sample data obtained this time, the comparison chart is as follows: Figure 3 As shown, it is found that after the first-order difference is performed, the data is more stable and is basically consistent with the second-order difference.

[0082] Step S204, determining the prediction priority of the temperature detection data according to the room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by the temperature prediction model, the temperature prediction model is trained based on the historical temperature detection data, and the historical temperature detection data is stored in the data lake.

[0083] In the technical solution provided in step S204, determining the predicted priority of the temperature detection data based on the computer room identification of the temperature detection data includes: obtaining computer room priority data, wherein the computer room priority data includes the computer room identification and the computer room priority corresponding to the computer room identification, and the computer room priority is determined based on the importance of each computer room and the historical failure probability of the computer room; determining the computer room priority corresponding to the temperature detection data based on the computer room identification; and using the computer room priority corresponding to the temperature detection data as the priority of the temperature detection data.

[0084] Optionally, the temperature detection data is associated with a priority definition table (computer room priority data) maintained by the computer room based on the computer room name (computer room identification) in the collection table attributes (temperature detection data), and the level (i.e., priority) of the computer room corresponding to the temperature detection data is determined based on the computer room identification; the computer room priority corresponding to the temperature detection data is used as the priority of the temperature detection data.

[0085] Optionally, first analyze the importance of each computer room, which is usually based on the type of business carried by the computer room, the business volume, and the regionality and timeliness of the business. For example, computer rooms that carry provincial core networks, regional data centers, or key businesses will be given a higher level of importance. Subsequently, examine the historical failure records of the computer room, and calculate the historical failure probability of the computer room by analyzing the failure frequency, failure type, and impact of the failure on the business of each computer room in the past year. These two indicators are combined according to the preset weight calculation to form a comprehensive evaluation system for the priority of the computer room.

[0086] Take a core machine building as an example. This machine building carries most of the province's Internet backbone network access services, has a large business volume, and has extremely high requirements for business continuity. Therefore, it is given the highest level of importance. At the same time, based on historical data analysis, this machine building has caused two downtime incidents due to temperature rise problems in the past year, and the probability of failure is relatively high. Combining these two factors, the machine building has a high overall priority score. When predicting the temperature through the temperature prediction model, the data of this machine building will be processed first to ensure that the temperature rise hidden dangers can be discovered and preventive measures can be taken in the early stage to avoid business interruption. Through this data-driven priority setting, it is possible to more accurately respond to the maintenance needs of different machine rooms, optimize resource allocation, and improve the operating efficiency and stability of the entire system.

[0087] Optionally, construct ACF (autocorrelation) and PACF (partial autocorrelation) graphs for collection table A to observe whether there is tailing and determine the final prediction template and p, q input parameters (autoregressive term order, moving average term order). After constructing the observation graph, it can be seen that the ACF graph and PACF graph are tailing, and finally it is determined to use the ARIMA model for prediction.

[0088] For example: Figure 4 The ACF diagram of this sample data is shown. Figure 5 The PACF diagram of this sample data is shown as Figure 4 and Figure 5 As shown, the ACF graph is tailing and the PACF graph is also tailing.

[0089] Since the ACF graph is tailing, and the number of vertical lines beyond the blue line is 2, the input parameter p=2 is taken. Further observation of the PACF graph tailing, the number of vertical lines beyond the horizontal blue line -1 is equal to 1, and finally the input parameter q=1 is determined. Since the first-order difference is used, d=1. The above parameters are used to input the ARIMA model.

[0090] Step S206, inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model.

[0091] Optionally, create an ARIMA model, fit the model, further obtain the forecast values ​​for the next 10 points, print and present them graphically.

[0092] For example, the 10 prediction point values ​​obtained this time are:

[0093] 2024-07-17 28.705828

[0094] 2024-07-18 28.706757

[0095] 2024-07-19 28.707106

[0096] 2024-07-20 28.707227

[0097] 2024-07-21 28.707265

[0098] 2024-07-22 28.707277

[0099] 2024-07-23 28.707281

[0100] 2024-07-24 28.707281

[0101] 2024-07-25 28.707282

[0102] Optionally, create a visualization of the original data and predicted data presentation to graphically and intuitively view the prediction results.

[0103] For example: The original and predicted data of this sample data are as follows: Figure 6 shown.

[0104] In the technical solution provided in step S206, before the temperature detection data is input into the temperature prediction model according to the prediction priority order, the method also includes: inputting the historical temperature detection data into the temperature prediction model for training, wherein the temperature prediction model is an autoregressive moving average model; obtaining the temperature prediction result of the output of the temperature prediction model, and ending the training of the temperature prediction model when the accuracy of the temperature prediction result is greater than a preset threshold.

[0105] Optionally, relying on the modular deployment method of each data processing component of the big data lake, the historical temperature collection data of each core machine building and IDC machine building is obtained, and the experience of the machine room maintenance personnel is electronically marked and solidified in combination with the actual situation of the machine room, and the local temperature rise area is automatically focused on to form a wide table of historical data of station temperature rise, a wide table of real-time data of the station, and a wide table of relative positions of the machine room. After that, a relatively mature ARIMA time series prediction model is introduced, and the sorted sample wide table is trained. The training results are objectively evaluated by the real value of the next day, and the model parameters are continuously optimized and estimated using the maximum likelihood estimation method. Finally, after the sample data similarity evaluation reaches 95% (i.e., the preset threshold) or more, the model application can be deployed for production operation.

[0106] Optionally, before inputting the historical temperature detection data into the temperature prediction model, the method also includes: generating an autocorrelation graph and a partial autocorrelation graph based on the historical temperature detection data, wherein the autocorrelation graph is used to determine a first input parameter of the temperature prediction model, and the partial autocorrelation graph is used to determine a second input parameter of the temperature prediction model.

[0107] Optionally, construct ACF and PACF diagrams for the collection table A to determine whether there is tailing and determine p and q input parameters (i.e., the first input parameter and the second input parameter), where p is the order of the autoregressive term and q is the order of the moving average term. After constructing the observation diagram ( Figure 4 ,5), it can be seen that its ACF graph is tailing and PACF graph is also tailing, and it is finally determined that the ARIMA model is used for prediction.

[0108] Since the ACF graph is tailing and the number of vertical lines beyond the blue line is 2, all input parameters p=2 are taken. The PACF graph is tailing and the number of vertical lines beyond the horizontal blue line -1 is equal to 1. Finally, the input parameter q=1 is determined. Since it uses the first-order difference, d=1.

[0109] Optionally, after obtaining the temperature prediction value output by the temperature prediction model, the method also includes: obtaining an actual temperature value corresponding to the temperature prediction value, wherein the prediction time point of the temperature prediction value and the detection time point of the actual temperature value are the same; evaluating the prediction success rate of the prediction model based on the error between the temperature prediction value and the actual temperature value, wherein the error includes a mean square error and a root mean square error; when the error is less than a preset error threshold, determining that the evaluation result of the prediction model is a correct prediction; when the error is less than a preset error threshold, determining that the evaluation result of the prediction model is an incorrect prediction.

[0110] Optionally, obtain the predicted data and store it in Table B, establish an analysis module, and store the predicted data table in the following format: signal name, acquisition value, test time, predicted value, actual observed value, affiliated equipment, affiliated computer room, affiliated station. When the actual observed temperature value is obtained the next day, compare the predicted result with the actual value to evaluate the prediction success rate of the model.

[0111] Optionally, the mean square error (MSE) and root mean square error (RMSE) are used to comprehensively evaluate the effect of the prediction model.

[0112] For example: Test MSE of this sample data: 0.288, Test RMSE: 0.536;

[0113] Optionally, the method further includes: retraining the temperature prediction model when the proportion of prediction errors in the evaluation results is greater than a preset error threshold.

[0114] Step S208: When the temperature prediction value is higher than the preset temperature threshold, it is determined that a fault exists at the detection location corresponding to the temperature detection data.

[0115] In the technical solution provided in step S208, after determining that a fault exists at the detection location corresponding to the temperature detection data, the method further includes: sending fault alarm information to a hidden danger management platform, wherein the hidden danger management platform is used to generate an inspection work order based on the fault alarm information.

[0116] Optionally, use the smart workbench page presentation function to periodically output the above results to the interface for maintenance personnel to present at a glance. Maintenance personnel can understand the various predicted values ​​at any time in this interface, and enter the priority level to predict important station data first, so as to obtain hidden dangers as quickly as possible and shorten the hidden danger handling time.

[0117] Optionally, the hidden danger handling platform can be connected to push the predicted temperature rise hidden dangers to the hidden danger management platform in real time to realize real-time closed-loop dispatch processing.

[0118] Optionally, an embodiment of the present application provides a method for predicting computer room faults: by creating various timing process modules in a big data lake, expert experience is digitized and solidified in the program, and a large amount of data that requires manual inspection and sorting is automatically formed into a wide table every day, reducing the workload of manual repetitive work; relying on the mature ARIMA time series differential autoregressive moving average model, more than 1,000 historical temperature points that change according to time are effectively used as sample data, input into the model for training, and evaluated in combination with actual data. When the similarity reaches more than 95%, the model configuration is applied for production operations; in the model configuration, a new generation of modular configuration ideas are used to program and configure each module such as data collection, data analysis, and data presentation separately to improve the independence of each module, and collect effective samples for model retraining on a monthly basis, and continuously iterate and upgrade to continuously improve the stability of the model operation, thereby improving the success rate of station temperature rise prediction. Each step of the model is sequentially configured to each platform through a new generation of configurable modules, and each stage such as collection, cleaning, analysis, and presentation is programmed and implemented respectively. By utilizing the high configurability and flexibility of the new generation, the model is continuously upgraded and iterated to improve the accuracy of model operation, so that the temperature rise hidden dangers can be automatically predicted in advance to ensure the safety of cooling supply in the station. Specifically, the method includes the following steps:

[0119] (1) Relying on the data acquisition capabilities of the big data lake, establish a multi-dimensional wide table:

[0120] Through the huge data acquisition capabilities of the big data lake, necessary sample data is collected, including but not limited to: temperature data, station room data, topology data, air conditioning data, network management configuration data, etc., which are associated and merged to eventually form two wide table data, one for defining priorities and the other as a sample set of historical temperature performance data and real-time temperature performance data.

[0121] The specific process of analyzing and establishing a wide table is as follows:

[0122] 1. Extract the station computer room information from the big data lake fixed asset table as Table 1, including: computer room name, computer room code, station to which it belongs, level, whether it is IDC, and star information;

[0123] 2. Obtain the computer room topology information from the big data lake computer building construction design table as Table 2, including: computer room name, computer room code, affiliated bureau station, area, drawings, and air conditioning location information;

[0124] 3. Obtain the monitoring equipment location information from the Big Data Lake Network Resource Fusion Library XML table as Table 3, including: equipment name, equipment code, collector MAC, IP;

[0125] 4. Obtain the standardized table of dynamic environment monitoring points from the big data lake standardized configuration specification as Table 4, including: signal name, SPID, device to which it belongs, and system to which it belongs;

[0126] 5. Associate the above-mentioned Table 1 and Table 2 through the computer room code to form a priority definition table for computer room maintenance. The fields in the table include: computer room name, computer room code, affiliated station, level, whether it is IDC, star rating, area, drawings, and air conditioning location.

[0127] 6. Associate Table 3 and Table 4 according to the device code to form a collection input parameter table with the device as the dimension. Currently, the attributes in this table include: signal name, SPID, system, device name, device code, collector MAC, IP, which are used as input parameters for real-time and historical data.

[0128] Through the above steps, two wide tables, the priority definition table and the signal input parameter table, are formed.

[0129] (2) Establish an ARIMA model based on the configurable capabilities of the new generation system.

[0130] Through the new generation of acquisition and control system, the signal input table is used as the input parameter, and the timed temperature acquisition program with hours as the dimension is configured to obtain historical temperature performance data. Through the energy adjustment system, the ARIMA model is established, and the collected performance data is exported to the model as the input parameter, and data cleaning is completed (mainly dealing with missing values ​​and outliers). The cleaned data is differentially processed, the data is converted to stability, and the first-order difference and second-order difference effects are observed, and the first-order difference is finally selected. Use the autocorrelation function and partial autocorrelation function graph to identify the number of autoregressive (p) and moving average (q) terms, and finally confirm the p and q input parameters. Use the predicted value to compare with the real data of the next day to verify the sample prediction output result. When the comparison success rate reaches 95%, the parameters of the model are finally determined for fitting. After modular deployment, it can be used for station production temperature rise prediction.

[0131] The specific process is as follows:

[0132] 1. Establish a station temperature acquisition module, use the acquisition input table in the previous step, call the unified acquisition and control data to collect the temperature data of each temperature sensor in the station, and store it in the acquisition table A. The attributes of the acquisition table are: signal SPID, collector MAC, computer room name, station name, acquisition time, and temperature value.

[0133] 2. Establish a data cleaning preprocessing module to process missing values ​​in the collection table A and remove null values. Further process the data in table A for outliers, remove abnormally large or small values, and finally perform an ADF test to verify whether it is stable.

[0134] 3. Calculate the first-order and second-order differences of collection table A, and observe the stable trend again. You can use the following statement to call it, and use matplotlib.pyplot to draw a comparison chart.

[0135] 4. Through the sample data obtained this time, it was found that after the first-order difference was performed, the data stability was relatively high and basically consistent with the second-order difference.

[0136] 5. Construct ACF and PACF graphs for collection table A to observe whether there is tailing and determine the final prediction template and p, q input parameters. After constructing the observation graph, it can be seen that the ACF graph and PACF graph are tailing, and finally it is determined to use the ARIMA model for prediction. The code for constructing ACF and PACF graphs is as follows:

[0137] 6. Since the ACF graph has a tail and the number of vertical lines beyond the blue line is 2, the input parameter p=2 is taken. Further observation of the PACF graph tailing shows that the number of vertical lines beyond the horizontal blue line -1 is equal to 1. Finally, the input parameter q=1 is determined. Since it uses the first-order difference, that is, d=1.

[0138] 7. Create an ARIMA model, fit the model, further obtain the predicted values ​​for the next 10 points, print and present them graphically.

[0139] 8. Obtain the predicted data and store it in Table B. Establish an analysis module and save the predicted data table in the following format: signal name, acquisition value, test time, predicted value, actual observed value, affiliated equipment, affiliated computer room, affiliated station. When the actual observed temperature value is obtained the next day, compare the predicted result with the actual value to evaluate the prediction success rate of the model.

[0140] 9. In this paper, mean square error (MSE) and root mean square error (RMSE) are used to comprehensively evaluate the effect of the prediction model.

[0141] 10. Create a visualization of the original data and predicted data to visualize the prediction results. The specific presentation code is as follows.

[0142] (3) Rapidly implement configuration of each module through the new generation of configurable platform capabilities.

[0143] Based on the powerful configuration capability of the new generation platform, each module in the second step is independently configured for easy daily maintenance and use. It is generally divided into: data acquisition module, data preprocessing module, ARIMA establishment module, sample training module, and model effect verification module. Then, a customizable new generation maintenance presentation interface is configured to assist on-site maintenance personnel to appropriately establish models to analyze the temperature rise of the computer room according to the degree of importance, thereby improving the overall efficiency of computer room maintenance.

[0144] The specific implementation process is as follows:

[0145] 1. In the configuration platform of the operation and maintenance center, configure the atomic acquisition capability, subscribe to the dcoos interface of the new generation acquisition and control system, and realize the ability to obtain the temperature rise value of the computer room in real time by using the collector IP, device ID, and SPID as input parameters. Use the job plan timing function to set the acquisition time to 5 o'clock, and finally store the collected data in the wide table A of the big data lake.

[0146] 2. Establish the atomic capability of data preprocessing, use the job planning timing module to automatically run at 6 o'clock, preliminarily remove the missing values, null values ​​and other abnormal values ​​in table A, and further use the first-order difference to transform the data smoothly.

[0147] 3. Build a visualization interface in the smart workbench, and present the ACF and PACF graphs for the wide table A that has completed the smooth transformation. At this time, observe the tailing or truncation of the graph, determine the p and q values, output the modeling input parameters to the model for formal prediction, establish the ARIMA fitting model, and obtain the predicted values ​​of the next 10 points and store them in Table B.

[0148] 4. Use the smart workbench page presentation function to regularly output the above results to the interface for maintenance personnel to present at a glance. Maintenance personnel can understand the various predicted values ​​at any time in this interface, and enter the priority level to prioritize the prediction of important station data, so as to obtain hidden dangers as quickly as possible and shorten the time for hidden danger handling.

[0149] 5. Connect to the hidden danger handling platform and push the predicted temperature rise hidden dangers to the hidden danger management platform in real time to realize real-time closed-loop dispatch processing.

[0150] Through the above steps, a method for early detection of station temperature rise faults based on ARIMA can be implemented, and a set of component-based hidden danger early detection systems with independent modules can be established. The independence of each module can independently complete data collection, cleaning preprocessing, model establishment, sample training, data evaluation, model optimization, deployment application, monitoring operation and other tasks, realizing a flexible, self-configured, easy-to-maintain and optimized time series prediction system. Specifically, the method embodiment of the present application has the following advantages:

[0151] (1) Relying on the data acquisition capability of the big data lake, the establishment of a multi-dimensional wide table effectively solves the problem of relying on the ability and experience of personnel themselves and the heavy workload. This patent uses the huge data acquisition capability of the big data lake to collect necessary sample data, including but not limited to: temperature data, station room data, topology data, air conditioning data, network management configuration data, etc., and associates and merges them to eventually form two wide table data, one for defining priorities, and the other for historical temperature performance data and real-time temperature performance data sample sets. It effectively solves the two problems of manual collection of system data and electronicization by on-site personnel, improves the efficiency of data acquisition, reduces the workload of on-site maintenance personnel, and ensures that on-site maintenance personnel can spend their time on ensuring power supply safety.

[0152] (2) Based on the configurable capabilities of the new generation of systems, the establishment of an ARIMA model effectively solves the problem of unstable success rate due to the lack of a scientific and objective model support. Through the new generation of acquisition and control system, the signal input parameter table is used as the input parameter, and a timed temperature acquisition program with an hour dimension is configured to obtain historical temperature performance data. Through the energy adjustment system, an ARIMA model is established, and the collected performance data is exported to the model as an input parameter. After data cleaning, differential processing, and smoothing processing are completed, the autocorrelation function and partial autocorrelation function graph are used to identify the input parameter and finally predict the station temperature rise. This effectively solves the pain point that the branch company's on-site maintenance cannot discover hidden dangers before the failure, reduces the maintenance personnel's traditional method of checking the machine room by manual inspection and discovering hidden dangers by physical sense of suitability, reduces the number of temperature failure work orders during high temperatures, and ensures the safe and stable operation of the station.

[0153] (3) Through the new generation of configurable platform capabilities, the configuration of each module can be quickly realized, effectively solving the problems of no system support, inability to iterate on a large scale, and inability to improve the effect. Based on the powerful configuration capabilities of the new generation platform, each processing module is configured independently, and is generally divided into: data acquisition module, data preprocessing module, ARIMA establishment module, sample training module, and model effect verification module. The customizable function of the optimization module supports on-site maintenance personnel to freely establish model analysis according to the importance and their own needs, which improves the flexibility of the system and the convenience of later iterations. It ensures that on-site maintenance personnel are not only users, but also configurators, connecting operations.

[0154] (5) Ensure that on-site maintenance personnel can find hidden dangers in the first place and make emergency operations in time before the alarm work order occurs. In this example, a medium-sized station has about 30 temperature measurement points. If manual statistics and predictions are performed, it will take 150 minutes and the prediction effect cannot be achieved. This method can automatically perform real-time statistics of all measurement points in the station in seconds and detect temperature hazards in advance. If they can be handled in time, the number of temperature work orders can be reduced by 30%. If it is promoted to the whole province, it can effectively save the xls statistical operations of on-site maintenance personnel. It is estimated that 8 man-days can be saved every day, and the automation rate of temperature statistics can be increased by about 60%.

[0155] (6) The problem of insufficient experience of on-site maintenance personnel has been solved. In this example, the expert experience is digitized, that is, the collected data can be automatically processed and predicted. Regardless of whether the on-site personnel are novices or veterans, they only need to process the results in a timely manner, which reduces the reliance on the professional knowledge of on-site maintenance personnel. The temperature hazards of the station can be discovered in real time, and emergency plans can be made in time to prevent failures.

[0156] The present application embodiment provides a device for predicting a computer room failure. Figure 7 is a schematic diagram of the structure of the device, such as Figure 7 As shown, the device includes: a first acquisition module 70, used to acquire temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, and the computer room identification is the identification of the computer room corresponding to the temperature detection data; a first determination module 72, used to determine the prediction priority of the temperature detection data according to the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by the temperature prediction model, and the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in the data lake; a first prediction module 74, used to input the temperature detection data into the temperature prediction model according to the prediction priority order, and obtain the temperature prediction value output by the temperature prediction model; a second determination module 76, used to determine that there is a fault at the detection location corresponding to the temperature detection data when the temperature prediction value is higher than a preset temperature threshold.

[0157] In some embodiments of the present application, the first determination module 72 determines the predicted priority of the temperature detection data based on the computer room identification of the temperature detection data, including: obtaining computer room priority data, wherein the computer room priority data includes the computer room identification and the computer room priority corresponding to the computer room identification, and the computer room priority is determined based on the importance of each computer room and the historical failure probability of the computer room; determining the computer room priority corresponding to the temperature detection data based on the computer room identification; and using the computer room priority corresponding to the temperature detection data as the priority of the temperature detection data.

[0158] In some embodiments of the present application, after the first acquisition module 70 acquires the temperature detection data, the method also includes: removing null values ​​in the temperature detection data; testing the temperature detection data according to a preset value interval, and removing the temperature detection data that exceeds the preset value interval; performing a unit root test on the temperature detection data; and performing differential processing on the temperature detection data whose unit root test result is that the unit root test passes.

[0159] In some embodiments of the present application, before the first prediction module 74 inputs the temperature detection data into the temperature prediction model according to the prediction priority order, the method also includes: inputting historical temperature detection data into the temperature prediction model for training, wherein the temperature prediction model is an autoregressive moving average model; obtaining the temperature prediction result of the output of the temperature prediction model, and ending the training of the temperature prediction model when the accuracy of the temperature prediction result is greater than a preset threshold.

[0160] In some embodiments of the present application, before the first prediction module 74 inputs the historical temperature detection data into the temperature prediction model, the method also includes: generating an autocorrelation graph and a partial autocorrelation graph based on the historical temperature detection data, wherein the autocorrelation graph is used to determine the first input parameter of the temperature prediction model, and the partial autocorrelation graph is used to determine the second input parameter of the temperature prediction model.

[0161] In some embodiments of the present application, after the first prediction module 74 obtains the temperature prediction value output by the temperature prediction model, the method also includes: obtaining the actual temperature value corresponding to the temperature prediction value, wherein the prediction time point of the temperature prediction value and the detection time point of the actual temperature value are the same; evaluating the prediction success rate of the prediction model based on the error between the temperature prediction value and the actual temperature value, wherein the error includes the mean square error and the root mean square error; when the error is less than a preset error threshold, determining that the evaluation result of the prediction model is a correct prediction; when the error is less than the preset error threshold, determining that the evaluation result of the prediction model is an incorrect prediction.

[0162] In some embodiments of the present application, the method further includes: retraining the temperature prediction model when the proportion of prediction errors in the evaluation results is greater than a preset error threshold.

[0163] In some embodiments of the present application, after the second determination module 76 determines that there is a fault at the detection location corresponding to the temperature detection data, the method also includes: sending fault alarm information to a hidden danger management platform, wherein the hidden danger management platform is used to generate an inspection work order based on the fault alarm information.

[0164] It should be noted that the various modules in the above-mentioned computer room fault prediction device can be program modules (for example, a set of program instructions that implement a certain specific function) or hardware modules. For the latter, it can be expressed in the following forms, but is not limited to this: the expression form of each of the above-mentioned modules is a processor, or the functions of each of the above-mentioned modules are implemented by a processor.

[0165] An embodiment of the present application provides a non-volatile storage medium, in which a program is stored, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the following computer room fault prediction method: obtaining temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, and the computer room identification is the identification of the computer room corresponding to the temperature detection data; determining the prediction priority of the temperature detection data according to the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by a temperature prediction model, and the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in a data lake; inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model; when the temperature prediction value is higher than a preset temperature threshold, determining that there is a fault at the detection location corresponding to the temperature detection data.

[0166] An embodiment of the present application provides an electronic device, including: a memory and a processor, the processor is used to run a program stored in the memory, wherein the following computer room fault prediction method is executed when the program is running: obtaining temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, and the computer room identification is the identification of the computer room corresponding to the temperature detection data; determining the prediction priority of the temperature detection data according to the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the processing order of the temperature detection data by a temperature prediction model, the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in a data lake; inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model; when the temperature prediction value is higher than a preset temperature threshold, determining that there is a fault at the detection location corresponding to the temperature detection data.

[0167] An embodiment of the present application provides a computer program product, including a computer program, which implements the following computer room fault prediction method when executed by a processor: obtaining temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, and the computer room identification is the identification of the computer room corresponding to the temperature detection data; determining the prediction priority of the temperature detection data based on the computer room identification of the temperature detection data, wherein the prediction priority is used to indicate the order in which the temperature detection data is processed by a temperature prediction model, and the temperature prediction model is trained based on historical temperature detection data, and the historical temperature detection data is stored in a data lake; inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model; when the temperature prediction value is higher than a preset temperature threshold, determining that there is a fault at the detection location corresponding to the temperature detection data.

[0168] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0170] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0171] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0172] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0173] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for predicting a computer room failure, characterized in that: include: Acquire temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, wherein the computer room identification is an identification of the computer room corresponding to the temperature detection data; Determining a prediction priority of the temperature detection data according to a computer room identifier of the temperature detection data, wherein the prediction priority is used to indicate a processing order of the temperature detection data by a temperature prediction model, the temperature prediction model is trained according to historical temperature detection data, and the historical temperature detection data is stored in a data lake; Inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model; When the temperature prediction value is higher than a preset temperature threshold, it is determined that a fault exists at the detection location corresponding to the temperature detection data.

2. The method for predicting computer room failure according to claim 1, characterized in that: Determining the prediction priority of the temperature detection data according to the room identifier of the temperature detection data includes: Acquire computer room priority data, wherein the computer room priority data includes a computer room identifier and a computer room priority corresponding to the computer room identifier, and the computer room priority is determined based on the importance of each computer room and the historical failure probability of the computer room; Determining the computer room priority corresponding to the temperature detection data according to the computer room identification; The computer room priority corresponding to the temperature detection data is used as the priority of the temperature detection data.

3. The method for predicting computer room failure according to claim 1, characterized in that: After acquiring the temperature detection data, the method further includes: Removing null values ​​from the temperature detection data; Checking the temperature detection data according to a preset value interval, and removing the temperature detection data that exceeds the preset value interval; Performing a unit root test on the temperature detection data; The temperature detection data whose unit root test result is that the unit root test passes is subjected to differential processing.

4. The method for predicting computer room failure according to claim 1, characterized in that: Before inputting the temperature detection data into the temperature prediction model according to the prediction priority order, the method further includes: Inputting historical temperature detection data into a temperature prediction model for training, wherein the temperature prediction model is an autoregressive moving average model; A temperature prediction result outputted by the temperature prediction model is obtained, and when the accuracy of the temperature prediction result is greater than a preset threshold, the training of the temperature prediction model is terminated.

5. The method for predicting computer room failure according to claim 4, characterized in that: Before inputting the historical temperature detection data into the temperature prediction model, the method further includes: An autocorrelation graph and a partial autocorrelation graph are generated based on the historical temperature detection data, wherein the autocorrelation graph is used to determine a first input parameter of the temperature prediction model, and the partial autocorrelation graph is used to determine a second input parameter of the temperature prediction model.

6. The method for predicting computer room failure according to claim 1, characterized in that: After obtaining the temperature prediction value output by the temperature prediction model, the method further includes: Acquire an actual temperature value corresponding to the temperature prediction value, wherein a prediction time point of the temperature prediction value and a detection time point of the actual temperature value are the same; Evaluate the prediction success rate of the prediction model according to the error between the temperature prediction value and the actual temperature value, wherein the error includes a mean square error and a root mean square error; When the error is less than a preset error threshold, determining that the evaluation result of the prediction model is a correct prediction; When the error is less than a preset error threshold, it is determined that the evaluation result of the prediction model is a prediction error.

7. The method for predicting computer room failure according to claim 6, characterized in that: The method further comprises: When the proportion of prediction errors in the evaluation results is greater than a preset error threshold, the temperature prediction model is retrained.

8. The method for predicting computer room failure according to claim 1, characterized in that: After determining that a fault exists at the detection location corresponding to the temperature detection data, the method further includes: sending fault alarm information to a hidden danger management platform, wherein the hidden danger management platform is used to generate an inspection work order based on the fault alarm information.

9. A device for predicting machine room failures, characterized in that: include: A first acquisition module is used to acquire temperature detection data, wherein the temperature detection data includes at least one of the following: detection time, detection location, temperature value, and computer room identification, wherein the computer room identification is an identification of the computer room corresponding to the temperature detection data; A first determination module, configured to determine a prediction priority of the temperature detection data according to a computer room identifier of the temperature detection data, wherein the prediction priority is used to indicate a processing order of the temperature detection data by a temperature prediction model, the temperature prediction model is trained according to historical temperature detection data, and the historical temperature detection data is stored in a data lake; A first prediction module, used for inputting the temperature detection data into the temperature prediction model according to the prediction priority order, and obtaining the temperature prediction value output by the temperature prediction model; The second determination module is used to determine that a fault exists at the detection location corresponding to the temperature detection data when the temperature prediction value is higher than a preset temperature threshold.

10. A non-volatile storage medium, characterized in that: The non-volatile storage medium stores a program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the computer room fault prediction method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the computer room fault prediction method according to any one of claims 1 to 8 is executed when the program is run.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for predicting a computer room failure described in any one of claims 1 to 8 are implemented.

Images