Fault cause diagnosis method, device and equipment and readable storage medium

Abstract

Embodiments of the present application disclose a fault cause diagnosis method, device and equipment and a readable storage medium. The fault cause diagnosis method comprises the following steps: determining a directed topological relationship between a plurality of devices included in a system to be analyzed; there is causality between any two devices having an upstream and downstream relationship in the directed topological relationship, which indicates that a device at the upstream will affect the working state of a device at the downstream; determining an upstream device of a fault device from the plurality of devices according to the directed topological relationship, the fault device being a device having a fault in the system to be analyzed, and the upstream device being a device at the upstream of the fault device in the directed topological relationship; determining whether each upstream device has an abnormality; and determining a fault cause according to the abnormality of each upstream device. Through the present application, the problem of low efficiency in the process of fault cause diagnosis in the related art is solved.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis, and in particular to a method, apparatus, device, and readable storage medium for diagnosing fault causes. Background Technology

[0002] Modern production systems often include a large number of devices of various types, which are interconnected (e.g., linked or interacting) to ensure the orderly operation of the entire system. In actual operation, a change in the operating state of any one device can have a ripple effect, affecting downstream devices and potentially causing some to exceed threshold limits, leading to equipment malfunctions and alarms. When a malfunction is detected, the cause can be difficult to diagnose because it could be caused by changes in the state of any device in the system.

[0003] For example, the inlet air temperature of a data center server rack has a significant impact on the normal operation of the IT equipment inside. Real-time monitoring and rapid response to anomalies can effectively suppress the escalation of risks and prevent downtime caused by high temperatures. However, a variety of factors can lead to increased inlet air temperature in server racks. Malfunctions in various components of the data center cooling system can also cause this type of fault to rise, making the diagnosis of such a failure quite difficult.

[0004] In the process of diagnosing the cause of a fault in related technologies, operators need to check and analyze the relevant equipment that may cause the fault one by one to finally determine the cause of the fault. This method of diagnosing the cause of a fault is very inefficient and makes it difficult to quickly locate the cause of the fault and eliminate the fault.

[0005] Therefore, the related technologies suffer from low efficiency in diagnosing fault causes. Currently, no effective solution has been proposed to address these issues.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background art described herein. Therefore, the background art may contain information that would not be considered part of the prior art by those skilled in the art. Summary of the Invention

[0007] This invention provides a method, apparatus, device, and readable storage medium for diagnosing fault causes, thereby at least addressing the problem of low efficiency in the fault cause diagnosis process in related technologies.

[0008] According to a first aspect of the present invention, a fault cause diagnosis method is provided, comprising: determining a directed topological relationship between multiple devices in a system to be analyzed, wherein any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, and the causal relationship indicates that the upstream device affects the working state of the downstream device; determining, from the multiple devices, an upstream device of a faulty device based on the directed topological relationship, wherein the faulty device is the device that has failed in the system to be analyzed, and the upstream device is the device upstream of the faulty device in the directed topological relationship; determining whether each upstream device has an anomaly; and determining the cause of the fault based on the anomaly of each upstream device.

[0009] Furthermore, based on the abnormal conditions of each upstream device, the cause of the fault is determined as follows: based on the directed topology and the abnormal conditions of each upstream device, the root cause device of the fault is determined, which is one or more of the upstream devices with abnormal conditions; the cause of the fault is determined to be the abnormality of the root cause device.

[0010] Furthermore, based on the directed topology and the abnormal conditions of each upstream device, the root cause device of the fault is determined as follows: starting from the faulty device, for each link in the directed topology, the upstreammost device that is abnormal is identified as a root cause device of the fault.

[0011] Furthermore, based on the directed topology and the abnormal conditions of each upstream device, the root cause device of the fault is determined by using a depth-first traversal algorithm, starting from the faulty device.

[0012] Furthermore, after determining the cause of the fault based on the abnormal conditions of each upstream device, the fault cause diagnosis method also includes: determining at least one faulty link based on the directed topology and the root cause device of the fault, wherein the faulty link is a link from the root cause device of the fault to the faulty device, and the faulty link includes multiple devices; and outputting information of at least one faulty link.

[0013] Furthermore, determining whether any upstream equipment is abnormal includes: for each upstream equipment, obtaining time-series data of preset indicators within a preset time period, where the preset indicators are indicators related to the fault of the faulty equipment; and determining whether any upstream equipment is abnormal based on the time-series data.

[0014] Furthermore, determining whether there is an anomaly in the upstream equipment based on the time series data includes: determining whether there is a subsequence of the target type in the time series data, wherein the subsequence of the target type is a data sequence related to the fault of the faulty equipment; and determining that there is no anomaly in the upstream equipment if there is no subsequence of the target type in the time series data.

[0015] Furthermore, determining whether there is an anomaly in the upstream equipment based on the time series data includes: if there is a subsequence of the target type in the time series data, determining whether there is an anomaly in the upstream equipment based on the subsequence of the target type and the historical time series data of the upstream equipment.

[0016] Furthermore, determining whether there is an anomaly in the upstream device based on the subsequence of the target type and the historical time series data of the upstream device includes: determining that there is no anomaly in the upstream device if the subsequence of the target type appears periodically in the historical time series data of the upstream device; and determining that there is an anomaly in the upstream device if the subsequence of the target type does not appear periodically in the historical time series data of the upstream device.

[0017] Furthermore, determining whether there is a target type of subsequence in the time series data includes: dividing the time series data into multiple subsequences with the objective of minimizing the sum of squared deviations of each subsequence, each subsequence being a data sequence composed of a portion of the data in the time series data; and determining whether there is a target type of subsequence among the multiple subsequences according to a preset classification rule.

[0018] According to a second aspect of the present invention, a fault cause diagnosis device is also provided, comprising: a first determining unit, configured to determine a directed topological relationship between multiple devices in a system to be analyzed, wherein any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, and the causal relationship indicates that the upstream device can affect the working state of the downstream device; a second determining unit, configured to determine, based on the directed topological relationship, the upstream device of the faulty device from among the multiple devices, wherein the faulty device is the device that has failed in the system to be analyzed, and the upstream device is the device upstream of the faulty device in the directed topological relationship; a third determining unit, configured to determine whether each upstream device has an anomaly; and a fourth determining unit, configured to determine the cause of the fault based on the anomaly of each upstream device.

[0019] Furthermore, the fourth determining unit includes: a first determining module, used to determine the root cause device of the fault based on the directed topology relationship and the abnormality of each upstream device, wherein the root cause device of the fault is one or more of the upstream devices that are abnormal; and a second determining module, used to determine that the cause of the fault is the abnormality of the root cause device.

[0020] The first determination module is used to: starting from the faulty device, for each link in the directed topology, determine the upstream device that is the most upstream and has an anomaly as a root cause device of the fault;

[0021] The first determination module is used to: starting from the faulty device, use a depth-first traversal algorithm to determine the root cause device of the fault;

[0022] The fault cause diagnosis device also includes: a fifth determining unit, used to determine at least one fault link based on the directed topology relationship and the fault root source device after determining the fault cause according to the abnormal situation of each upstream device, wherein the fault link is a link from the fault root source device to the fault device and the fault link includes multiple devices; and an output unit, used to output information of at least one fault link.

[0023] The third determining unit includes: an acquisition module, used to acquire time-series data of preset indicators within a preset time period for each upstream device, wherein the preset indicators are indicators related to the fault of the faulty device; and a third determining module, used to determine whether there is an anomaly in the upstream device based on the time-series data.

[0024] The third determining module includes: a first determining submodule, used to determine whether there is a target type subsequence in the time series data, wherein the target type subsequence is a data sequence related to the fault of the faulty device; and a second determining submodule, used to determine that there is no abnormality in the upstream device when there is no target type subsequence in the time series data.

[0025] The third determination submodule also includes: the third determination submodule, used to determine whether there is an anomaly in the upstream device based on the target type subsequence and the historical time series data of the upstream device when there is a target type subsequence in the time series data;

[0026] The third determination submodule is used to: determine that there is no anomaly in the upstream device when the subsequence of the target type appears periodically in the historical time series data of the upstream device; and determine that there is anomaly in the upstream device when the subsequence of the target type does not appear periodically in the historical time series data of the upstream device.

[0027] The first determination submodule is used to: divide the time series data into multiple subsequences with the goal of minimizing the sum of squared deviations of the data in each subsequence, and each subsequence is a data sequence composed of a portion of the data in the time series data; and determine whether there is a subsequence of the target type among the multiple subsequences according to the preset classification rules.

[0028] According to a third aspect of the present invention, a method for diagnosing the cause of a temperature rise alarm in a data center is also provided, comprising: determining a directed topological relationship between multiple devices included in a data center, wherein the multiple devices include multiple devices to be cooled and multiple working devices in a cooling system, and any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, wherein the causal relationship indicates that the upstream device will affect the working state of the downstream device; determining the upstream device of the faulty device from the multiple devices according to the directed topological relationship, wherein the faulty device is the device that triggered the temperature rise alarm among the multiple devices to be cooled, and the upstream device is the device upstream of the faulty device in the directed topological relationship; determining whether each upstream device has an anomaly; and determining the cause of the temperature rise alarm of the faulty device according to the anomaly of each upstream device.

[0029] According to a fourth aspect of the present invention, a readable storage medium is also provided, on which computer instructions are stored, wherein when the computer instructions are executed by a processor, the above-described fault cause diagnosis method is implemented, or when the computer instructions are executed by a processor, the above-described data center temperature rise alarm cause diagnosis method is implemented.

[0030] According to a fifth aspect of the present invention, a fault cause diagnosis device is also provided, including a memory and a processor. The memory stores computer instructions, wherein when the computer instructions are executed by the processor, they implement the above-described fault cause diagnosis method, or when the computer instructions are executed by the processor, they implement the above-described data center temperature rise alarm cause diagnosis method.

[0031] The fault cause diagnosis method of this invention first determines the directed topological relationship between multiple devices in the system to be analyzed during the fault cause diagnosis process. This directed topological relationship represents the upstream and downstream connections between multiple devices during operation. That is, devices that are interconnected and interact with each other during operation are reflected in the topological relationship. There is a causal relationship between any two devices with upstream and downstream connections in the directed topological relationship. Causality means that the upstream device will affect the working state of the downstream device. After obtaining the topological relationship, the upstream devices of the faulty device can be more accurately found. These upstream devices are the possible causes of the faulty device. After accurately locating the upstream devices that may cause the faulty device to fail, it is determined whether each upstream device has any abnormalities. The cause of the fault can then be located based on the abnormalities of each upstream device. The fault cause diagnosis method of this invention, based on the directed topological relationship containing causality to determine upstream devices, can accurately identify the upstream devices related to the fault, and then analyze the abnormalities of the upstream devices in a targeted manner to find the cause of the fault. This effectively improves the efficiency of fault cause diagnosis and solves the problem of low efficiency in the fault cause diagnosis process in related technologies. Attached Figure Description

[0032] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0033] Figure 1 This is a flowchart illustrating a fault cause diagnosis method provided in an embodiment of the present invention;

[0034] Figure 2 This is a schematic diagram of a fault cause diagnosis device provided in an embodiment of the present invention;

[0035] Figure 3 This is a schematic diagram illustrating the process of diagnosing temperature rise faults at the data center terminal using the fault cause diagnosis method of this invention. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this invention are used to distinguish different objects, rather than to limit a specific order.

[0038] Figure 1 This is a fault cause diagnosis method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0039] Step S102: Determine the directed topological relationship between multiple devices in the system to be analyzed. There is a causal relationship between any two devices that have an upstream-downstream connection. Causality indicates that the upstream device will affect the working state of the downstream device.

[0040] Step S104: Based on the directed topology, determine the upstream device of the faulty device from among multiple devices. The upstream device is the device that is upstream of the faulty device in the directed topology.

[0041] Step S106: Determine if any abnormalities exist in the upstream devices;

[0042] Step S108: Determine the cause of the fault based on the abnormal conditions of each upstream device.

[0043] This fault cause diagnosis method, employing this approach, first determines the directed topological relationships between multiple devices in the system under analysis. These relationships represent the upstream and downstream connections between devices during operation; devices that are interconnected and interact with each other are reflected in the topological relationship. A causal relationship exists between any two devices in the directed topological relationship, meaning that an upstream device influences the operating state of a downstream device. After obtaining the topological relationship, the upstream devices of the faulty device can be more accurately located. The faulty device can be any device in the system under analysis that has failed; these upstream devices represent the possible causes of the fault. After accurately locating the upstream devices that may have caused the fault, it is determined whether each upstream device exhibits any anomalies. Based on these anomalies, the cause of the fault can be pinpointed. This fault cause diagnosis method, based on a directed topological relationship containing causality, accurately identifies upstream devices related to the fault, allowing for targeted analysis of anomalies in these upstream devices to find the cause of the fault. This effectively improves the efficiency of fault cause diagnosis and solves the problem of low efficiency in related technologies.

[0044] In directed topology, "directed" indicates that the topological connections between these devices have a directionality, which can be used to determine the upstream and downstream relationships between devices. The upstream device of the faulty device can be an adjacent device in the directed topology or a non-adjacent device (i.e., separated from the faulty device by one or more other devices). A directed topology is a directed connection between devices. For example, if the output of device A is the input of device B, and the output of device B is the input of device C, then connecting devices A, B, and C sequentially with directed line segments can be considered a directed topology relationship between them. Here, device A is the upstream device of devices B and C, and device C is the downstream device of devices A and B. The criteria for determining whether an upstream device is faulty can be flexibly determined based on the actual situation, thus allowing for a more targeted identification of the cause of the fault. The aforementioned faults, as the names suggest, indicate that the equipment has malfunctioned. These faults can take many forms, such as system crashes or alarms. An anomaly, however, has a broader scope, indicating that the equipment's operating state is not normal. Specifically, it can include two situations: one is a significant anomaly that meets the fault criteria, in which case it is considered a fault state; the other is a minor anomaly, perhaps just slightly higher or lower than the expected data, but not meeting the fault criteria. Both of these situations constitute equipment anomalies.

[0045] In some real-world scenarios, the number and / or connection relationships of devices in the system under analysis may change, rendering the previously determined directed topology relationships inaccurate. To address this issue, in a preferred embodiment, after determining the directed topology relationships between the multiple devices in the system under analysis, the fault cause diagnosis method further includes: receiving topology update data; and updating the directed topology relationships based on the topology update data. This allows for timely updates of the directed topology relationships between devices at the required frequency (e.g., once a day), ensuring the accuracy of fault cause diagnosis.

[0046] The process of determining the directed topological relationships among multiple devices in the system to be analyzed requires ensuring that there is a causal relationship between any two devices that are upstream or downstream in the determined directed topological relationship. Causality means that the upstream device will affect the working state of the downstream device. Therefore, the focus of constructing the directed topological relationship is to find the causal relationships between the various devices in the system to be analyzed. In practice, the process of determining the directed topological relationship can be implemented in different ways. For example, the directed topological relationship can be determined directly based on the causal relationships between the devices. Another example is to first determine the topological relationship, and then adjust the topological relationship based on the causal relationships between the interconnected devices in the topological relationship, thereby obtaining the directed topological relationship.

[0047] In a preferred embodiment, determining the directed topological relationships among multiple devices in the system to be analyzed includes: determining the connection relationships among the multiple devices in the system to be analyzed; for any two interconnected devices, determining the causal relationship between them, whereby the causal relationship characterizes whether either of the two devices will affect the working state of the other, i.e., characterizes whether there is causality between the two devices; and determining the directed topological relationship based on the connection relationships and the causal relationships. In this embodiment, in order to conveniently and accurately construct the directed topological relationship, the actual connection relationships between the devices are first determined, and then, for any two connected devices, their causal relationship is determined, thereby allowing the directed topological relationship to be determined based on the connection relationships and the causal relationships.

[0048] For any two interconnected devices among a plurality of devices, determining the causal relationship between them can be achieved in various ways, such as theoretical analysis and experimental verification. In a preferred embodiment, determining the causal relationship between any two interconnected devices among a plurality of devices includes: obtaining historical state data of the two interconnected devices, where historical state data represents the historical states of the two devices; and determining the causal relationship between them based on the historical state data. In this embodiment, the causal relationship between the two devices can be easily determined by analyzing their historical state data. For example, if a change in the state of one device in the historical state data leads to a change in the state of the other device, then a causal relationship can be determined. If the changes in the state data of the two devices are not related, then a causal relationship can be determined. Determining the directed topological relationship based on the connection relationship and the causal relationship includes: for any two interconnected devices, if there is no causal relationship between the two devices, then the connection relationship between them is deleted.

[0049] The fault cause diagnosis method of this invention can be applied to diagnose fault causes in various types of systems on the infrastructure side, as long as there are some devices in the system that have mutual influence and interaction. Examples include data center HVAC systems, refrigeration systems, and power systems.

[0050] In practice, the causes of equipment failure can be varied. It could be due to multiple device malfunctions leading to the final failure, a single device malfunction causing the failure, or a failure in communication or media transmission between devices. However, the most likely scenario is that one or more devices malfunction, leading to the failure of the final device. Because of the aforementioned causal directed topological relationship between devices, an upstream device malfunction may incidentally cause malfunctions in its downstream devices. In this embodiment, determining the cause of failure based on the malfunctions of each upstream device includes: identifying the root cause device based on the directed topological relationship and the malfunctions of each upstream device; and determining the cause of failure as an malfunction of the root cause device. In other words, this embodiment, in determining the cause of failure, searches upstream based on the directed topological relationship to identify the root cause device, thereby more effectively locating the root cause of the failure.

[0051] In one embodiment, determining the root cause device based on the directed topology and the anomalies of each upstream device includes: starting from the faulty device, for each link in the directed topology, identifying the upstreammost device with an anomaly as the root cause device. In other words, the process of determining the root cause device starts from the faulty device and identifies the upstreammost device with an anomaly as the root cause device for each link in the directed topology. This method of identifying the root cause device is more reasonable. When it malfunctions, its downstream devices are likely to malfunction under its influence, and these downstream devices are likely to be operating normally. Therefore, the root cause of the faulty device is very likely to be the upstreammost device in the directed topology, thus achieving rapid and effective location of the root cause.

[0052] In practical implementation, there are various specific search methods during the fault finding process along each link in the directed topology. In one optional implementation, determining the root cause device based on the directed topology and the anomalies of each upstream device includes: starting from the faulty device, using a depth-first traversal algorithm to determine the root cause device.

[0053] In this embodiment, a depth-first search (DFS) algorithm is used to locate the root cause of the fault. This method of searching one link to the end before searching other links can quickly trace back to the upstream device with the anomaly, thereby facilitating faster location of the root cause of the fault and improving the efficiency of fault diagnosis.

[0054] Of course, in addition to using the depth-first traversal method mentioned above to find the root cause of the fault, other search methods can be used if the search efficiency requirement is not high, such as breadth-first search (BFS).

[0055] After determining the cause of the fault, directly providing the cause to the user can lead to poor interpretability, making it difficult for the user to understand the fault's occurrence process. In a preferred embodiment, after determining the fault cause based on the anomalies of each upstream device, the fault cause diagnosis method further includes: determining at least one fault link based on the directed topology and the root cause device. A fault link is a link from the root cause device to the faulty device, and the fault link includes multiple devices; and outputting information about at least one fault link. After determining the fault cause, to help users understand the cause of the faulty device's failure more clearly, at least one fault link is determined based on the root cause device and the directed topology, and information about at least one fault link is output. Here, a fault link is a link from the root cause device to the faulty device, and it includes multiple devices. The process of determining the fault link is also simple: based on the directed topology, the corresponding fault link can be determined by searching along the direction from the root cause device to the faulty device. By outputting the fault link information, users can understand the cause of the fault more clearly and intuitively, i.e., which root cause device failed, which devices subsequently became abnormal, and ultimately, which device failed.

[0056] After identifying the specific upstream devices of the faulty device based on the directed topology, determining whether each upstream device exhibits anomalies becomes a crucial step affecting fault diagnosis. In practice, there are various ways to determine whether each upstream device exhibits anomalies, such as manual on-site inspection. During the inspection, judgments can be made based on various equipment, inspection techniques, or experience to determine the presence of anomalies. In a preferred embodiment, to improve the accuracy of judging the anomalies of each upstream device, determining whether each upstream device exhibits anomalies includes: for each upstream device, acquiring time-series data of preset indicators within a preset time period. These preset indicators are those related to the fault of the faulty device; and determining whether the upstream device exhibits anomalies based on the time-series data. Time-series data, collected sequentially over time, can effectively characterize the working status of upstream devices. This embodiment, by collecting time-series data of indicators related to the fault of the faulty device, can accurately determine whether the upstream device exhibits anomalies based on the analysis of the time-series data.

[0057] Determining whether there is an anomaly in the upstream equipment based on time-series data is a crucial step in the entire fault diagnosis process. The accuracy of this determination directly affects the precision of subsequent fault location. In a preferred embodiment, determining whether there is an anomaly in the upstream equipment based on time-series data includes: determining whether there is a target type subsequence in the time-series data, where the target type subsequence is a data sequence related to the fault of the faulty equipment; if there is no target type subsequence in the time-series data, it is determined that there is no anomaly in the upstream equipment.

[0058] In this embodiment, it is determined whether a special subsequence exists in the time-series data. A special subsequence is a subsequence of a target type, which is a sequence composed of a portion of the data in the time-series data. The target type subsequence can be of any type, and its classification criteria can be flexibly determined, such as rising type, falling type, fluctuating type, types with rising rates greater than a certain threshold, types with falling rates greater than a certain threshold, etc. By correlating the fault of the faulty equipment with specific subsequence types, the anomaly of upstream equipment can be easily determined. Specifically, if a target type subsequence appears in the time-series data, the upstream equipment may be abnormal; if no target type subsequence appears, the upstream equipment is not abnormal. For example, if the fault of the faulty equipment is excessively high temperature, and the upstream equipment to be determined to be abnormal is the fan on the cold air outlet side of the refrigeration system, and the collected time-series data is the fan speed, and a subsequence appears in the time-series data where the fan speed gradually decreases, it may lead to insufficient air supply to the faulty equipment, thereby causing the temperature of the faulty equipment to rise. If there is no data in the time-series data showing a decrease in fan speed, then the possibility of the upstream equipment being faulty can be ruled out.

[0059] As mentioned above, the presence of a target-type subsequence in the time-series data indicates a potential anomaly in the upstream device, requiring further assessment to confirm its existence. To facilitate accurate assessment, in this embodiment, determining the presence of an anomaly in the upstream device based on the time-series data includes: when a target-type subsequence exists in the time-series data, determining whether the upstream device is abnormal based on the target-type subsequence and the historical time-series data of the upstream device. In this embodiment, the historical time-series data of the upstream device is used for auxiliary assessment, i.e., judging the presence of anomalies based on historical experience, thereby obtaining a reliable assessment result. Of course, besides this method of combining historical experience, other methods can also be used for assessment, such as using a trained deep learning network.

[0060] Specifically, determining whether an upstream device exhibits anomalies based on the target type's subsequences and historical time-series data of the upstream device includes: if the target type's subsequences appear periodically in the upstream device's historical time-series data, the upstream device is determined to be free of anomalies; if the target type's subsequences do not appear periodically in the upstream device's historical time-series data, the upstream device is determined to be abnormal. In combining historical time-series data to determine whether an upstream device is abnormal, it is determined whether the target type's subsequences appear periodically in the upstream device's historical time-series data. If these subsequences appear periodically, it indicates that they are a normal phenomenon in the upstream device's operation, and the conclusion that the upstream device is not abnormal can be drawn with a relatively accurate assessment. Conversely, if these subsequences do not appear periodically, it indicates that their appearance is abnormal and accompanied by a malfunction in the faulty device, thus confirming that the upstream device is abnormal. This method of determining whether an upstream device is abnormal by combining historical time-series data has better reliability.

[0061] In a specific embodiment, determining whether a corresponding upstream device is abnormal based on time-series data includes: storing directed topological relationships in a graph database; determining whether a target type subsequence exists in the time-series data, where the subsequence is a data sequence composed of at least a portion of the data in the time-series data, and the target type subsequence is a sequence related to the fault of the faulty device; adding attribute labels to the corresponding device nodes in the graph database based on the determination result of whether a target type subsequence exists in the time-series data, where the attribute labels represent whether a target type subsequence exists in the time-series data; subsequently, during the process of searching for the root cause device of the fault using a depth-first traversal algorithm, when the corresponding upstream device is found, it can be determined whether the upstream device is abnormal based on the attribute labels.

[0062] In a preferred embodiment, determining whether a target type of subsequence exists in the time-series data includes: dividing the time-series data into multiple subsequences with the objective of minimizing the sum of squared deviations of each subsequence; each subsequence is a data sequence composed of a portion of the data in the time-series data; and determining whether a target type of subsequence exists among the multiple subsequences according to a preset classification rule. In determining whether a target type of subsequence exists in the time-series data, the time-series data is divided with the objective of minimizing the sum of squared deviations of each subsequence, resulting in multiple subsequences. These multiple subsequences are homogeneous, meaning that the data within the same subsequence either increases or decreases sequentially. This increase or decrease in data is directly related to causing anomalies, therefore, the subsequences obtained in this way can be better used for anomaly judgment of the upstream equipment. It is easy to understand that for a specific type of fault, the change type corresponding to the time-series data of a specific indicator should be definite; that is, only a specific type of subsequence can cause anomalies in the upstream equipment, thereby leading to faults in the downstream equipment. For example, if the faulty equipment is overheating, and the upstream equipment to be checked for an anomaly is the fan on the cold air outlet side of the refrigeration system, the collected time-series data is the fan speed. The time-series data shows two different types of subsequences: gradually increasing data and gradually decreasing data. Clearly, an increase in fan speed will lead to an increase in the airflow to the faulty equipment, but will not cause the faulty equipment to heat up. Therefore, the subsequence type of the fan speed data corresponding to the faulty equipment overheating should be gradually decreasing fan speed. Therefore, after dividing the time-series data into various subsequences, a preset classification rule is used to determine whether there are subsequences of the target type among these subsequences. For example, based on the classification rule of increasing and decreasing time-series data, it is determined whether there are subsequences that conform to the classification rule of decreasing time-series data. If so, it is determined that there are subsequences of the target type among multiple subsequences.

[0063] The fault cause diagnosis method of the present invention will be described below with reference to a specific embodiment. Figure 3 This is a schematic diagram illustrating the process of diagnosing temperature rise faults at data center terminals using the fault cause diagnosis method of this invention:

[0064] In this embodiment, the fault diagnosis method is used to diagnose terminal temperature rise faults within a data center. Terminal temperature within a data center refers to the temperature at the cold air duct outlet / inlet air temperature of the server rack. Server racks house IT equipment such as servers, which generate significant heat during operation. This heat is generated by the cooling system and blown into the racks through the cold air duct outlet to cool the equipment. Therefore, terminal temperature directly affects the normal operation of IT equipment. Real-time monitoring and rapid response to anomalies can effectively suppress the escalation of risks and prevent downtime caused by overheating. The causes of terminal temperature anomalies are varied, and the signs are complex. Currently, intelligent alarm algorithms are very effective at detecting terminal temperature anomalies. However, identifying the root cause still requires on-site personnel to investigate. After a terminal temperature rise alarm is generated, on-site maintenance personnel will investigate and manually intervene to control temperature changes. In the complex data center infrastructure system, the root cause of terminal temperature alarms often comes unexpectedly. Relying on years of accumulated operational experience to quickly resolve superficial problems may not address the root cause. Analyzing data from various measuring points (specific indicators) within the cooling system is time-consuming. Therefore, related fault diagnosis methods lack intelligent root cause localization capabilities, hindering efficiency improvements in large-scale operations and maintenance. In other words, diagnosing abnormal temperature alarms at the data center's edge requires manual analysis and fault location by operators, resulting in low efficiency.

[0065] This embodiment employs intelligent root cause diagnosis technology to pinpoint the root cause of temperature rise alarms at the data center's endpoints. The main steps consist of three stages: constructing the cooling system topology, detecting the status of key measurement points, and performing root cause diagnosis.

[0066] For the part about constructing the refrigeration system topology:

[0067] First, multiple devices in the data center can be obtained from the equipment archive database, such as end-point cold aisles, server racks, and equipment and sensors in the refrigeration system such as air conditioners, water pumps, chillers, and cooling towers. These devices are stored as nodes in the graph database, and the device type, location, name, and unit group information are stored as node attributes.

[0068] Secondly, the refrigeration system involves the circulation of chilled water, cooling water, and air. To model the impact of the refrigeration system on the terminal devices, a unidirectional link for cooling is selected to construct the topological connections between the devices, i.e., to determine the directed topological relationships, and these connections are then stored in a graph database. After completing the construction of the directed topological relationships, to ensure the accuracy of the topological relationships, device nodes, connections, and attributes can be updated daily through offline scheduling.

[0069] For the key measurement point status detection section:

[0070] First, when an alarm occurs at the end of the device, the alarmed device is the faulty device. To determine the cause of the fault, it is necessary to perform status detection based on the recent indicator data of key measurement points (key measurement points are preset indicators) of upstream devices. Specifically, based on the location information of the alarm node at the end, the upstream devices related to that node are queried from the graph database; recent time-series data of measurement points that affected cooling before the alarm of these upstream devices are retrieved from the device indicator library; the time-series data is divided into homogeneous subseries by minimizing the sum of squared deviations of each subseries through change point detection; the features of the subseries are extracted to identify the status changes of the measurement point data; the subseries that will cause the temperature rise are first screened from multiple subseries, and then periodic fluctuation detection is performed on them based on historical time-series data to avoid the appearance of periodic fluctuations interfering with the root cause inference; the extracted status information is synchronized to the graph database as the attribute items of the corresponding upstream devices.

[0071] Regarding the root cause diagnosis section:

[0072] First, causal relationships between upstream and downstream devices are learned based on equipment measurement data. This involves verifying the causality of the constructed topological relationships, ensuring that causality exists between devices with upstream and downstream connections within the directed topological relationships. Then, using the causal-based directed topological relationships as a knowledge graph, a depth-first search is performed, adding nodes that meet the corresponding attribute conditions to the inference subtree. The search for that causal link stops when a node that does not meet the requirements is encountered. Based on the inference subtrees obtained from the depth-first search, they are organized into root cause links. Based on the influence of each root cause link on the downstream device, at least one root cause link is selected for output, leading to a root cause diagnosis conclusion.

[0073] Secondly, such as Figure 2As shown, embodiments of the present invention also provide a fault cause diagnosis device, comprising: a first determining unit, configured to determine the directed topological relationship between multiple devices in the system to be analyzed, wherein any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, and the causal relationship indicates that the upstream device will affect the working state of the downstream device; a second determining unit, configured to determine the upstream device of the faulty device from among the multiple devices according to the directed topological relationship, wherein the faulty device is any device in the system to be analyzed that has failed, and the upstream device is the device upstream of the faulty device in the directed topological relationship; a third determining unit, configured to determine whether each upstream device has an anomaly; and a fourth determining unit, configured to determine the cause of the fault based on the anomaly of each upstream device. In this fault cause diagnosis device, during the fault cause diagnosis process, the first determining unit first determines the directed topological relationship between the multiple devices in the system to be analyzed. This directed topological relationship represents the upstream and downstream connections of the multiple devices during operation, that is, devices that are interconnected and interact with each other during operation are reflected in the topological relationship, and any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, meaning that the upstream device will affect the working state of the downstream device. After obtaining the topological relationship, the second determining unit can more accurately locate the upstream devices of the faulty device, which are the possible causes of the fault. After accurately locating the upstream devices that may have caused the fault, the third determining unit determines whether each upstream device has any abnormalities. Finally, the fourth determining unit can locate the cause of the fault based on the abnormalities of each upstream device. The fault cause diagnosis device of this embodiment of the invention, which determines upstream devices based on a directed topological relationship containing causality, can accurately identify the upstream devices related to the fault, and then analyze the abnormalities of the upstream devices in a targeted manner to find the cause of the fault. This effectively improves the efficiency of fault cause diagnosis and solves the problem of low efficiency in the fault cause diagnosis process in related technologies.

[0074] In one specific embodiment, the fourth determining unit includes: a first determining module, used to determine the root cause device of the fault based on the directed topology relationship and the abnormality of each upstream device, wherein the root cause device of the fault is one or more of the upstream devices that are abnormal; and a second determining module, used to determine that the cause of the fault is an abnormality of the root cause device.

[0075] The first determination module is used to: starting from the faulty device, for each link in the directed topology, determine the upstream device that is the most upstream and has an anomaly as a root cause device of the fault;

[0076] The first determination module is used to: starting from the faulty device, use a depth-first traversal algorithm to determine the root cause device of the fault;

[0077] The fault cause diagnosis device also includes: a fifth determining unit, used to determine at least one fault link based on the directed topology relationship and the fault root source device after determining the fault cause according to the abnormal situation of each upstream device, wherein the fault link is a link from the fault root source device to the fault device and the fault link includes multiple devices; and an output unit, used to output information of at least one fault link.

[0078] The third determining unit includes: an acquisition module, used to acquire time-series data of preset indicators within a preset time period for each upstream device, wherein the preset indicators are indicators related to the fault of the faulty device; and a third determining module, used to determine whether there is an anomaly in the upstream device based on the time-series data.

[0079] The third determining module includes: a first determining submodule, used to determine whether there is a target type subsequence in the time series data, wherein the target type subsequence is a data sequence related to the fault of the faulty device; and a second determining submodule, used to determine that there is no abnormality in the upstream device when there is no target type subsequence in the time series data.

[0080] The third determination submodule also includes: the third determination submodule, used to determine whether there is an anomaly in the upstream device based on the target type subsequence and the historical time series data of the upstream device when there is a target type subsequence in the time series data;

[0081] The third determination submodule is used to: determine that there is no anomaly in the upstream device when the subsequence of the target type appears periodically in the historical time series data of the upstream device; and determine that there is anomaly in the upstream device when the subsequence of the target type does not appear periodically in the historical time series data of the upstream device.

[0082] The first determination submodule is used to: divide the time series data into multiple subsequences with the goal of minimizing the sum of squared deviations of the data in each subsequence, and each subsequence is a data sequence composed of a portion of the data in the time series data; and determine whether there is a subsequence of the target type among the multiple subsequences according to the preset classification rules.

[0083] Furthermore, embodiments of the present invention also provide a method for diagnosing the cause of a data center temperature rise alarm, comprising: determining a directed topological relationship between multiple devices included in a data center, wherein the multiple devices include multiple devices to be cooled and multiple working devices in a cooling system, and any two devices with upstream and downstream connections in the directed topological relationship have a causal relationship, wherein the causal relationship indicates that the upstream device will affect the working state of the downstream device; determining the upstream device of the faulty device from among the multiple devices according to the directed topological relationship, wherein the faulty device is the device among the multiple devices to be cooled that triggered the temperature rise alarm, and the upstream device is the device upstream of the faulty device in the directed topological relationship; determining whether each upstream device has an anomaly; and determining the cause of the temperature rise alarm of the faulty device based on the anomaly of each upstream device. In this data center temperature rise alarm cause diagnosis method, during the process of diagnosing the cause of a data center temperature rise alarm, the directed topological relationship between multiple devices in the data center is first determined, wherein the multiple devices include multiple devices to be cooled and multiple working devices in a cooling system, and the directed topological relationship indicates the upstream and downstream connections of the multiple devices during operation, that is, devices that are interconnected and interact with each other during operation are reflected in the topological relationship. For any two devices in a topological relationship to have an upstream-downstream connection, a causal requirement must be met. Causality means that an upstream device will affect the operating state of a downstream device. After obtaining the topological relationship, the upstream devices of the faulty device can be more accurately identified. These upstream devices are the possible causes of the temperature rise alarm of the faulty device. After accurately locating the upstream devices that may be causing the faulty device, it is determined whether each upstream device has any anomalies. Based on the anomalies of each upstream device, the cause of the temperature rise alarm of the faulty device can be located. The data center temperature rise alarm cause diagnosis method of this invention, based on a directed topological relationship containing causality to determine upstream devices, can accurately identify the upstream devices related to the fault, and then analyze the anomalies of the upstream devices in a targeted manner to find the cause of the temperature rise alarm. This effectively improves the efficiency of data center temperature rise alarm cause diagnosis and solves the problem of low efficiency in the process of data center temperature rise alarm cause diagnosis in related technologies.

[0084] In a specific embodiment, determining the cause of the temperature rise alarm of the faulty device based on the abnormality of each upstream device includes: determining the root cause device of the fault based on the directed topology relationship and the abnormality of each upstream device, wherein the root cause device of the fault is one or more of the upstream devices with abnormality; and determining the cause of the temperature rise alarm of the faulty device as the abnormality of the root cause device.

[0085] Based on the directed topology and the anomalies of each upstream device, the root cause device of the fault is determined as follows: starting from the faulty device, for each link in the directed topology, the upstreammost device that is also abnormal is identified as a root cause device of the fault.

[0086] Based on the directed topology and the abnormal conditions of each upstream device, the root cause device of the fault is determined by starting from the faulty device and using a depth-first traversal algorithm.

[0087] After determining the cause of the temperature rise alarm of the faulty device based on the abnormal situation of each upstream device, the data center temperature rise alarm cause diagnosis method also includes: determining at least one faulty link based on the directed topology relationship and the root cause device of the fault. The faulty link is a link from the root cause device of the fault to the faulty device, and the faulty link includes multiple devices; and outputting the information of at least one faulty link.

[0088] Determining whether there are any abnormalities in each upstream device includes: for each upstream device, obtaining time-series data of preset indicators within a preset time period, where the preset indicators are indicators related to the fault of the faulty device; and determining whether there are any abnormalities in the upstream device based on the time-series data.

[0089] Determining whether there is an anomaly in the upstream equipment based on time series data includes: determining whether there is a subsequence of the target type in the time series data, where the subsequence of the target type is a data sequence related to the fault of the faulty equipment; and determining that there is no anomaly in the upstream equipment if there is no subsequence of the target type in the time series data.

[0090] Determining whether there is an anomaly in the upstream equipment based on time series data includes: if there is a subsequence of the target type in the time series data, determining whether there is an anomaly in the upstream equipment based on the subsequence of the target type and the historical time series data of the upstream equipment.

[0091] Determining whether an upstream device is abnormal based on the subsequence of the target type and the historical time series data of the upstream device includes: if the subsequence of the target type appears periodically in the historical time series data of the upstream device, it is determined that the upstream device is not abnormal; if the subsequence of the target type does not appear periodically in the historical time series data of the upstream device, it is determined that the upstream device is abnormal.

[0092] Determining whether a target type of subsequence exists in time series data includes: dividing the time series data into multiple subsequences with the objective of minimizing the sum of squared deviations of each subsequence; each subsequence is a data sequence composed of a portion of the data in the time series data; and determining whether a target type of subsequence exists among the multiple subsequences according to a preset classification rule.

[0093] Furthermore, this embodiment of the invention also provides a readable storage medium storing computer instructions thereon, wherein when the computer instructions are executed by a processor, they implement the above-mentioned fault cause diagnosis method, or when the computer instructions are executed by a processor, they implement the above-mentioned data center temperature rise alarm cause diagnosis method.

[0094] Finally, this embodiment of the invention also provides a fault cause diagnosis device, including a memory and a processor. The memory stores computer instructions, wherein when the computer instructions are executed by the processor, they implement the above-mentioned fault cause diagnosis method, or when the computer instructions are executed by the processor, they implement the above-mentioned data center temperature rise alarm cause diagnosis method.

[0095] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0096] The sequence numbers of the above embodiments of the present invention are merely for descriptive purposes and do not represent the superiority or inferiority of the embodiments. Moreover, the steps shown in the flowcharts 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 flowcharts, in some cases the steps shown or described may be performed in a different order than that shown here.

[0097] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0099] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0101] If the integrated unit is implemented as 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 invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0102] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fault cause diagnosis method, comprising: determining a directed topological relationship between a plurality of devices included in a system to be analyzed, comprising: determining a connection relationship between the plurality of devices included in the system to be analyzed; for any two interconnected devices in the plurality of devices, determining a causality relationship between the any two interconnected devices, the causality relationship representing whether any one of the two devices will affect the working state of the other device, i.e., whether there is causality between the two devices; and determining a directed topological relationship according to the connection relationship and the causality relationship, wherein there is causality between any two devices having an upstream-downstream relationship in the directed topological relationship, and the causality represents that the device in the upstream among the two devices will affect the working state of the device in the downstream; determining, from the plurality of devices, an upstream device of a fault device according to the directed topological relationship, the fault device being a device in the system to be analyzed that has failed, and the upstream device being the device in the upstream of the fault device in the directed topological relationship; for each of the upstream devices, obtaining time series data of a preset index in a preset time period, the preset index being an index related to the failure of the fault device; and determining, according to the time series data, whether the upstream device has an abnormality; determining a fault cause according to the abnormality of each of the upstream devices.

2. The failure cause diagnosis method according to claim 1, wherein The determining of the fault cause according to the abnormality of each of the upstream devices comprises: determining a fault root device according to the directed topological relationship and the abnormality of each of the upstream devices, the fault root device being one or more of the upstream devices having the abnormality; and determining the fault cause as the abnormality of the fault root device. The determining of the fault root device according to the directed topological relationship and the abnormality of each of the upstream devices comprises:

3. The failure cause diagnosis method according to claim 2, wherein determining, from the fault device, for each link in the directed topological relationship, the most upstream upstream device having the abnormality as a fault root device. The determining of the fault root device according to the directed topological relationship and the abnormality of each of the upstream devices comprises:

4. The failure cause diagnosis method according to claim 3, wherein determining, from the fault device, the fault root device by using a depth-first traversal algorithm. After the determining of the fault cause according to the abnormality of each of the upstream devices, the fault cause diagnosis method further comprises:

5. The failure cause diagnosis method according to any one of claims 2 to 4, wherein, determining at least one fault link according to the directed topological relationship and the fault root device, the fault link being a link from the fault root device to the fault device, and the fault link including a plurality of the devices; and outputting information of at least one of the fault links. The determining of whether the upstream device has the abnormality according to the time series data comprises: determining whether there is a target type of subsequence in the time series data, the target type of subsequence being a data sequence related to the failure of the fault device; and in a case where there is no target type of subsequence in the time series data, determining that the upstream device has no abnormality.

6. The failure cause diagnosis method according to claim 1, wherein The determining of whether the upstream device has the abnormality according to the time series data comprises: ​ ​ 7. The failure cause diagnosis method according to claim 6, wherein ​ In a case where the sub-sequence of the target type exists in the time series data, determining, according to the sub-sequence of the target type and historical time series data of the upstream device, whether the upstream device is abnormal.

8. The failure cause diagnosis method according to claim 7, wherein Determining, according to the sub-sequence of the target type and historical time series data of the upstream device, whether the upstream device is abnormal comprises: In a case where the sub-sequence of the target type periodically appears in the historical time series data of the upstream device, determining that the upstream device is not abnormal; In a case where the sub-sequence of the target type does not periodically appear in the historical time series data of the upstream device, determining that the upstream device is abnormal.

9. The failure cause diagnosis method according to claim 6, wherein Determining whether a sub-sequence of a target type exists in the time series data comprises: Dividing the time series data to obtain a plurality of the sub-sequences, each of which is a data sequence composed of part of the data in the time series data, with the objective of minimizing the sum of squared deviations of data of each of the sub-sequences; According to a preset classification rule, determining whether a sub-sequence of the target type exists in the plurality of the sub-sequences.

10. A fault cause diagnosis apparatus, comprising: A first determining unit configured to determine a directed topological relationship between a plurality of devices included in a system to be analyzed, including: determining a connection relationship between the plurality of devices included in the system to be analyzed; for any two of the plurality of devices connected to each other, determining a causal relationship between the any two of the devices connected to each other, the causal relationship representing whether any one of the two devices will affect a working state of the other device, i.e., representing whether there is causality between the two devices; and determining the directed topological relationship according to the connection relationship and the causal relationship, there being causality between any two of the devices having an upstream-downstream relationship in the directed topological relationship, the causality representing that the device in the upstream of the two devices will affect the working state of the device in the downstream; A second determining unit configured to determine, according to the directed topological relationship, an upstream device of a fault device from the plurality of devices, the fault device being a device in the system to be analyzed that has a fault, and the upstream device being the device in the upstream of the fault device in the directed topological relationship; A third determining unit configured to, for each of the upstream devices, acquire time series data of a preset index in a preset time period, the preset index being an index related to the fault of the fault device; and determine, according to the time series data, whether the upstream device is abnormal; A fourth determining unit configured to determine a fault cause according to abnormal conditions of each of the upstream devices.

11. The fault cause diagnosis apparatus according to claim 10, wherein The fourth determining unit includes: a first determining module configured to determine, according to the directed topological relationship and the abnormal conditions of each of the upstream devices, a fault root device, the fault root device being one or more of the upstream devices that are abnormal; and a second determining module configured to determine that the fault cause is the abnormality of the fault root device. The first determining module is configured to determine, from the faulty device, for each link in the directed topological relationship, the most upstream abnormal upstream device as a faulty root cause device; The first determining module is configured to determine, from the faulty device, the faulty root cause device by using a depth-first search algorithm; The fault cause diagnosis apparatus further includes a fifth determining unit configured to, after determining the fault cause according to the abnormal conditions of the upstream devices, determine at least one faulty link according to the directed topological relationship and the faulty root cause device, the faulty link being a link from the faulty root cause device to the faulty device, and the faulty link including a plurality of devices; and an output unit configured to output information of the at least one faulty link. The third determining module includes a first determining submodule configured to determine whether a target type of subsequence exists in the time series data, the target type of subsequence being a data sequence related to the fault of the faulty device; and a second determining submodule configured to, in a case where the target type of subsequence does not exist in the time series data, determine that the upstream device is not abnormal. The third determining submodule is further configured to, in a case where the target type of subsequence exists in the time series data, determine whether the upstream device is abnormal according to the target type of subsequence and historical time series data of the upstream device. The third determining submodule is configured to, in a case where the target type of subsequence periodically appears in the historical time series data of the upstream device, determine that the upstream device is not abnormal; and in a case where the target type of subsequence does not periodically appear in the historical time series data of the upstream device, determine that the upstream device is abnormal. The first determining submodule is configured to divide the time series data to obtain a plurality of subsequence, each of the subsequence being a data sequence composed of part of the time series data, by minimizing a sum of squared deviations of data of each of the subsequence as a target, and determine whether the target type of subsequence exists in the plurality of subsequence according to a preset classification rule.

12. A data center temperature rise alarm cause diagnosis method, comprising: determining a directed topological relationship between a plurality of devices included in a data center includes: determining a connection relationship between a plurality of devices included in a system to be analyzed; for any two interconnected devices in the plurality of devices, determining a causal relationship between the any two interconnected devices, the causal relationship representing whether any one of the two devices will affect the working state of the other device, i.e., whether there is causality between the two devices; determining a directed topological relationship according to the connection relationship and the causal relationship, the plurality of devices including a plurality of heat dissipation devices and a plurality of working devices in a refrigeration system, there being causality between any two devices having an upstream and downstream relationship in the directed topological relationship, the causality representing that the device in the upstream among the two devices will affect the working state of the device in the downstream; According to the directed topological relationship, an upstream device of a fault device is determined from the plurality of devices, the fault device being a device of the plurality of devices to be cooled that has issued a temperature rise alarm, the upstream device being a device upstream of the fault device in the directed topological relationship; For each of the upstream devices, time series data of a preset index in a preset time period is obtained, the preset index being an index related to the fault of the fault device; and whether the upstream device has an abnormality is determined according to the time series data; A temperature rise alarm cause of the fault device is determined according to the abnormality of each of the upstream devices.

13. A readable storage medium having stored thereon computer instructions, wherein, The computer instructions, when executed by the processor, implement the fault cause diagnosis method in any one of claims 1 to 9, or the computer instructions, when executed by the processor, implement the data center temperature rise alarm cause diagnosis method in claim 12.

14. A failure cause diagnosis apparatus comprising a memory and a processor, the memory having stored thereon computer instructions, wherein, The computer instructions, when executed by the processor, implement the fault cause diagnosis method in any one of claims 1 to 9, or the computer instructions, when executed by the processor, implement the data center temperature rise alarm cause diagnosis method in claim 12.

Images