CN114645844B

CN114645844B - Method, computing device and computer medium for determining flow state of air compression station

Info

Publication number: CN114645844B
Application number: CN202210560040.6A
Authority: CN
Inventors: 周子叶; 沈国辉; 陈欢
Original assignee: Mogulinker Technology Shenzhen Co Ltd
Current assignee: Guangdong Mushroom Iot Technology Co ltd
Priority date: 2022-05-23
Filing date: 2022-05-23
Publication date: 2022-07-19
Anticipated expiration: 2042-05-23
Also published as: CN114645844A

Abstract

The invention provides a method, a computing device and a computer medium for determining a flow state of an air compression station. The air compression station comprises a plurality of air compressors and a main pipe connected with the air compressors. The method comprises the following steps: the method comprises the steps of obtaining real-time power of each of a plurality of air compressors, real-time pressure and real-time flow of a main pipe at a group of detection moments; estimating a start-stop pressure curve included angle of the air compressor based on the power difference before and after start-stop operation of the air compressor and the real-time pressure of a main pipe by using a single-machine pressure model of each air compressor; estimating the actual displacement of the air compressor based on the real-time power of the air compressor and the real-time flow of the main pipe by using a single-machine displacement model of each air compressor; and determining the flow state of the air compression station by using a main pipe abnormity diagnosis model of the main pipe based on the real-time pressure and the real-time flow of the main pipe at the group of detection moments and the start-stop pressure curve included angle and the actual air displacement of each air compressor at the group of detection moments.

Description

Method, computing device and computer medium for determining flow state of air compression station

Technical Field

The present invention relates generally to the field of industrial control, and more particularly, to a method, computing device, and computer-readable storage medium for determining a flow state of an air compression station.

Background

The air compressor is a device for compressing gas, and is widely used in construction, steel, mining and chemical plants. In situations where gas demand is high, an air compression station including a plurality of air compressors is typically employed to provide compressed gas to the gas end. The gas end may be, for example, a factory or a workshop. In an air compression station, a plurality of air compressors are connected to the same main pipe to deliver compressed air to the air consuming side.

In the digital operation and maintenance of the air compressor station, the flow meter is an important sensor, and many important operating state parameters need to be inferred based on the flow, such as pipeline pressure loss, actual exhaust amount of each air compressor and energy efficiency level. The method is limited by factors such as site station houses, pipeline structures and construction cost, and usually only a flowmeter is installed on a main pipe, so that the measurement accuracy and the fault state of the main pipe flowmeter have great influence on the control accuracy of the air compressor.

However, the measurement accuracy and stability of the flow meter are insufficient with respect to the electric meter and the pressure meter, and thus it is necessary to evaluate the measurement accuracy of the main pipe flow meter and perform intelligent fault diagnosis using predictive maintenance. For example, the abnormal reading of the flow meter of the main pipe may be caused by the deterioration of the performance of the pipeline, the deterioration of the performance of the air compressor, or the malfunction of the flow meter, and therefore, the abnormality can be accurately solved only by accurately determining the reason for the abnormal reading of the flow meter.

One solution is to add one or more redundant flow meters to the parent pipe to perform double or multiple checks. However, this solution will significantly increase the hardware cost, increase the construction difficulty, increase the operation and maintenance difficulty, and the newly added flow meter will also increase the cost of data acquisition, transmission, storage and calculation, and this method is also not suitable for the case where the original flow meter and the redundant flow meter are failed at the same time.

Another solution is to optimize the internal structure of the flow meter to enable fault diagnosis and measurement bias compensation. However, this solution only modifies the internal structure of the flow meter, and still needs to further solve the problem of data transmission to an external control system, and this method cannot realize real-time status monitoring, and cannot timely find and synchronize if a fault occurs during the use.

Disclosure of Invention

In order to solve the problems, the invention provides a method for evaluating the measurement accuracy of a main pipe flowmeter and intelligently diagnosing the fault type when the main pipe flowmeter is abnormal by utilizing existing equipment on the main pipe and an air compressor, such as a main pipe flowmeter and an electric meter and a pressure meter of the air compressor, through a plurality of AI models.

According to one aspect of the invention, a method of determining a traffic status of an air compression station is provided. The air compression station comprises a plurality of air compressors and a main pipe connected with the plurality of air compressors. The method comprises the following steps: acquiring real-time power of each air compressor in the plurality of air compressors, real-time pressure and real-time flow of the main pipe at a group of detection moments; estimating a start-stop pressure curve included angle of each air compressor based on the power difference before and after the start-stop operation of the air compressor and the real-time pressure of the main pipe by using a single-machine pressure model of each air compressor; estimating the actual displacement of each air compressor based on the real-time power of the air compressor and the real-time flow of the main pipe by using a single-machine displacement model of each air compressor; and determining the flow state of the air compression station by using a main pipe abnormity diagnosis model of the main pipe based on the real-time pressure and the real-time flow of the main pipe at the group of detection moments and the start-stop pressure curve included angle and the actual air displacement of each air compressor at the group of detection moments.

In some embodiments, estimating the start-stop pressure curve included angle of the air compressor based on the power difference before and after the start-stop operation of the air compressor and the real-time pressure of the main pipe comprises: determining whether the air compressor is started or stopped at one detection moment in the group of detection moments; if the fact that the air compressor is started and stopped at the detection moment is determined, two real-time power values before and after the air compressor is started and stopped are obtained; determining a power difference before and after starting and stopping operation of the air compressor based on the two real-time power values; and estimating the start-stop pressure curve included angle of the air compressor based on the single-machine pressure model of the air compressor and the power difference.

In some embodiments, estimating the actual displacement of the air compressor based on the real-time power of the air compressor and the real-time flow of the main pipe comprises: determining whether or not other air compressors than the air compressor are in an operating state at one detection time of the set of detection times; and if it is determined that other air compressors except the air compressor are not in the running state at the detection moment, estimating the actual air displacement of the air compressor based on a single-machine air displacement model of the air compressor and the real-time flow of the main pipe.

In some embodiments, determining the flow status of the air compression station comprises: determining pressure data and air displacement data of the air compression station and equipment data of each air compressor based on the start-stop pressure curve included angle and the actual air displacement of each air compressor; and determining the flow state of the air compression station based on the pressure data, the air displacement data and the equipment data of each air compressor by using a main pipe abnormity diagnosis model of the main pipe.

In some embodiments, the determining the pressure data, the air displacement data and the equipment data of each air compressor based on the start-stop pressure curve included angle and the actual air displacement of each air compressor comprises: combining start-stop pressure curve included angles output by the single-machine pressure models of the plurality of air compressors to acquire pressure data of the air compression station; combining actual air displacement output by the single air displacement models of the plurality of air compressors to obtain air displacement data of the air compression station; and combining the start-stop pressure curve included angle output by the single-machine pressure model of each air compressor with the actual air displacement output by the single-machine air displacement model to acquire the equipment data of the air compressors.

In some embodiments, the main pipe abnormality diagnosis model includes a pressure abnormality detection model, an exhaust gas amount abnormality detection model, and an equipment abnormality detection model, and a fault diagnosis model, and wherein determining the flow state of the air compressing station based on the pressure data, the exhaust gas amount data, and the equipment data of each air compressor using the main pipe abnormality diagnosis model of the main pipe includes: determining whether the pressure data of the air compression station is abnormal pressure data by using a pressure abnormality detection model based on a novel detection algorithm; determining whether the air displacement data of the air compression station is abnormal air displacement data by using an air displacement abnormal detection model based on a novel detection algorithm; determining whether the equipment data of the air compressor is abnormal equipment data or not by using an equipment abnormality detection model based on a novelty detection algorithm; and determining a flow rate state of the air compression station using the failure diagnosis model based on results of the pressure abnormality detection model, the exhaust gas amount abnormality detection model, and the equipment abnormality detection model.

In some embodiments, determining the flow state of the air compression station using the fault diagnosis model comprises: if the abnormal pressure data, the abnormal air displacement data and the abnormal equipment data do not exist, determining that the flow state of the air compression station is normal; and determining that a flow state of the air compression station is abnormal if it is determined that at least one of abnormal pressure data, abnormal air displacement data, and abnormal equipment data exists.

In some embodiments, the method further comprises: if the equipment data of at least one air compressor in the plurality of air compressors is determined to be abnormal equipment data and the equipment data of other air compressors is not abnormal equipment data, determining that the reason that the flow state of the air compression station is abnormal is that the performance of the at least one air compressor is abnormal; if the exhaust gas volume data of the air compression station are determined to be abnormal exhaust gas volume data and the pressure data are determined to be abnormal pressure data, determining that the reason that the flow state of the air compression station is abnormal is that the pipeline performance of the main pipe is abnormal; and if it is determined that the air displacement data of the air compression station are all abnormal air displacement data but the pressure data are not abnormal pressure data, determining that the cause of the abnormal flow state of the air compression station is a flow meter failure of the main pipe.

In some embodiments, before estimating the start-stop pressure curve included angle of the air compressor based on the power difference before and after the start-stop operation of the air compressor and the real-time pressure of the main pipe, the method further includes: acquiring a power sample set, a pressure sample set and a flow sample set of each air compressor; training the stand-alone pressure model based on the variation data in the power sample set and the pressure sample set of the master pipe; and training the single machine displacement model based on stable data in the power sample set and the flow sample set of the main pipe.

In some embodiments, determining the flow state of the air compression station further comprises: and training a main pipe abnormity diagnosis model based on the start-stop pressure curve included angle sample output by the single-machine pressure model, the air displacement sample output by the single-machine air displacement model and the equipment sample of each air compressor.

In some embodiments, obtaining a set of pressure samples for the parent tube comprises: acquiring operation records of the plurality of air compressors and the main pipe; screening recording points including starting and stopping operations of the air compressor from the operation records and eliminating other data in first preset time before and after the recording points; for each recording point, determining whether the recording point of the start-stop operation of other air compressors exists within second preset time before and after the recording point; and if it is determined that no recording point of the start-stop operation of other air compressors exists in second preset time before and after the recording point of the start-stop operation of the air compressors, taking two pieces of power data before and after the start-stop operation of the air compressors as a group of change data in a power sample set of the air compressors and taking two pieces of main pipe pressure data before and after the start-stop operation of the air compressors as a group of pressure samples in a pressure sample set of the main pipes.

In some embodiments, training the stand-alone pressure model based on the varying data in the set of power samples and the set of pressure samples of the parent tube comprises: determining a power difference of the air compressor before and after the start-stop operation based on a group of variation data in a power sample set of the air compressor; determining a pressure slope of the main pipe and a pressure included angle before and after starting and stopping operation of the air compressor on the basis of a group of pressure samples corresponding to the group of change data in the pressure sample set of the main pipe; and training the single-machine pressure model by taking the power difference of the air compressor before and after the start-stop operation and the pressure included angle of the main pipe before and after the start-stop operation of the air compressor as a training sample.

In some embodiments, the single-machine pressure model includes a linear regression model, and when the air compressor is a variable frequency air compressor, the model parameters of the single-machine pressure model of the air compressor are determined based on a power difference of the air compressor before and after the start-stop operation and a pressure included angle of the main pipe before and after the start-stop operation of the air compressor.

In some embodiments, the single-machine pressure model includes a linear regression model, and when the air compressor is a power frequency air compressor, the model parameters of the single-machine pressure model of the air compressor are determined based on the rated power of the air compressor and the pressure included angle of the main pipe before and after the start-stop operation of the air compressor.

In some embodiments, obtaining a set of flow samples for the parent pipe comprises: acquiring operation records of the plurality of air compressors and the main pipe; screening record points without air compressor operation within third preset time from the operation records; selecting recording points which meet a preset stable condition from the screened recording points; and taking the power data of the air compressor at the recording point as stable data in a power sample set of the air compressor and taking the flow data of the main pipe at the recording point as a flow sample in a flow sample set of the main pipe.

In some embodiments, the predetermined stable condition comprises: the flow difference of the main pipe between the recording point and the adjacent recording point is smaller than a preset flow difference threshold value; the sum of the power of the plurality of air compressors is less than a preset power and a threshold value; and the sum of the specific powers of the air compressors is smaller than a preset specific power threshold.

In some embodiments, the stand-alone displacement model comprises a linear regression model, and wherein training the stand-alone displacement model based on the stable data in the set of power samples and the set of flow samples for the parent comprises: and training the single-machine air displacement model based on stable data in the power sample set of each air compressor and the flow of the main pipe corresponding to the stable data.

According to another aspect of the invention, a computing device is provided. The computing device includes: at least one processor; and at least one memory coupled to the at least one processor and storing instructions for execution by the at least one processor, the instructions when executed by the at least one processor causing the computing device to perform the method as described above.

According to yet another aspect of the present invention, a computer-readable storage medium is provided, having stored thereon computer program code, which, when executed, performs the method as described above.

Because the pressure gauge and the flowmeter of the main pipe are generally in standard configuration, the scheme of the invention can detect the state of the flowmeter of the main pipe in real time and analyze abnormal reasons in time to early warn without increasing the cost caused by increasing or modifying the flowmeter of the main pipe or needing additional data acquisition cost.

Drawings

The invention will be better understood and other objects, details, features and advantages thereof will become more apparent from the following description of specific embodiments of the invention given with reference to the accompanying drawings.

Fig. 1 shows a schematic view of an air compression station for implementing a method for determining a traffic status of the air compression station according to an embodiment of the invention.

Fig. 2 illustrates a flow diagram of a method for determining a traffic status of an air compression station, according to some embodiments of the invention.

Fig. 3 shows a schematic block diagram of a model for determining the flow state of an air compression station according to an embodiment of the present invention.

Fig. 4A illustrates a further flow chart of a process for estimating a start-stop pressure curve angle of an air compressor according to some embodiments of the present invention.

FIG. 4B illustrates a schematic of a pressure curve and a flow curve of a parent pipe according to some embodiments of the invention.

FIG. 5 illustrates a further flow diagram of a process for determining a cause of a flow anomaly in a parent pipe according to some embodiments of the present invention.

FIG. 6A illustrates a schematic structural diagram of a parent pipe anomaly diagnostic model according to some embodiments of the present invention.

Fig. 6B illustrates a further flow diagram of a process for determining a traffic status of an air compression station in accordance with some embodiments of the present invention.

Fig. 7 illustrates a flow chart of a process for determining a cause of an anomaly in a traffic state of an air compression station, according to some embodiments of the present invention.

FIG. 8 illustrates a flow diagram of a process for training a model according to some embodiments of the invention.

FIG. 9A illustrates a further flow diagram of a process of obtaining a set of pressure samples for a parent tube according to some embodiments of the invention.

FIG. 9B illustrates a further flow diagram of a process for training a stand-alone pressure model according to some embodiments of the invention.

FIG. 10 illustrates a further flow diagram of a process of obtaining a flow sample set for a parent pipe according to some embodiments of the invention.

FIG. 11 illustrates a block diagram of a computing device suitable for implementing embodiments of the present invention.

Detailed Description

Preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

In the following description, for the purposes of illustrating various inventive embodiments, certain specific details are set forth in order to provide a thorough understanding of the various inventive embodiments. One skilled in the relevant art will recognize, however, that the embodiments may be practiced without one or more of the specific details. In other instances, well-known devices, structures and techniques associated with this application may not be shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

Throughout the specification and claims, the word "comprise" and variations thereof, such as "comprises" and "comprising," are to be understood as an open, inclusive meaning, i.e., as being interpreted to mean "including, but not limited to," unless the context requires otherwise.

Reference throughout this specification to "one embodiment" or "some embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiments is included in at least one embodiment. Thus, the appearances of the phrases "in one embodiment" or "in some embodiments" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

Furthermore, the terms first, second and the like used in the description and the claims are used for distinguishing the objects for clarity, and do not limit the size, other sequence and the like of the described objects.

Fig. 1 shows a schematic illustration of an air compression station 1 for implementing a method for determining a traffic state of the air compression station according to an embodiment of the invention. As shown in fig. 1, the air compressing station 1 may include a plurality of air compressors 10 (5 air compressors 10-1, 10-2, 10-3, 10-4, and 10-5 are exemplarily shown in fig. 1) and a mother pipe 20, wherein the plurality of air compressors 10 transmit the generated compressed gas to a gas using end 30 through the mother pipe 20. Each air compressor 10 comprises or is equipped with an electricity meter 12 for measuring the operating power of the air compressor 10 at various times. The parent pipe 20 includes or is configured with a flow meter 22 and a pressure gauge 24, the flow meter 22 being configured to measure flow data through the parent pipe 20 at each time, and the pressure gauge 24 being configured to sense pressure data generated on the parent pipe by the gas flowing in the parent pipe 20 at each time. Air compressor station 1 may also include or be connected to a computing device 40, which computing device 40 may obtain, via wired or wireless communication links, power data measured by electric meter 12 of each air compressor 10, flow data measured by flow meter 22 of main pipe 20, and pressure data measured by pressure gauge 24. The computing device 40 may be located in the air compression station, or may be located remotely from the air compression station, such as a cloud-based device.

The computing device 40 may include at least one processor 42 and at least one memory 44 coupled to the at least one processor 42, the memory 44 having stored therein instructions 46 executable by the at least one processor 42, the instructions 46 when executed by the at least one processor 42 performing at least a portion of a method as described below. The specific structure of computing device 40 may be described, for example, in connection with FIG. 11, as follows.

Fig. 2 illustrates a flow diagram of a method 200 for determining a traffic status of an air compression station, according to some embodiments of the invention. Method 200 may be performed by computing device 40 in system 1 shown in fig. 1. The method 200 is described below in conjunction with fig. 1-11.

As shown in fig. 2, method 200 includes a block 210 in which computing device 40 may obtain real-time power, real-time pressure, and real-time flow of each air compressor 10, and of manifold 20 at a set of test times.

Depending on the rated sampling frequency or the upper-level configuration of the electricity meter 12, the flow meter 22, and the pressure gauge 24, the electricity meter 12, the flow meter 22, and the pressure gauge 24 may respectively detect the power of the corresponding air compressor 10 and the flow rate and pressure of the main pipe 20 at a certain detection interval (or detection frequency). The detection intervals of the meter 12, the flow meter 22 and the pressure gauge 24 may be the same or different. For convenience, it is assumed herein that the electricity meter 12, the flow meter 22, and the pressure meter 24 perform measurements at the same measurement intervals to obtain real-time power, real-time flow rate, and real-time pressure at each measurement instant. In the event that the detection intervals of the electric meter 12, the flow meter 22, and the pressure gauge 24 are different, the computing device 40 may further sample the actual measured data to obtain real-time power, real-time flow, and real-time pressure at each detection time. For example, the detection time may be a time of the least common multiple of the respective detection intervals.

In this way, the computing device 40 can obtain real-time flow data and real-time pressure data of a set of continuous mother pipes at the detection time and real-time power data of each air compressor 10, and can determine whether the flow state of the air compression station 1 is normal and the abnormal reason in the case of abnormal state by using a trained artificial intelligence model according to the data.

Fig. 3 shows a schematic block diagram of a model 300 for determining the flow conditions of an air compression station according to an embodiment of the invention. As shown in FIG. 3, the model 300 may have a two-layer model structure, where a first layer includes a stand-alone pressure model 312 and a stand-alone displacement model 314, and a second layer includes a parent pipe anomaly diagnostic model 320. The single-machine pressure model 312 is used for determining the start-stop pressure curve included angle of each air compressor 10 based on the power of each air compressor 10 and the pressure of the main pipe 20, and the single-machine displacement model 314 is used for determining the actual displacement of each air compressor 10 based on the power of each air compressor and the flow of the main pipe 20. The various models of the various layers of model 300 will be described in greater detail below in conjunction with fig. 2-11.

At block 220, the computing device 40 may estimate a start-stop pressure curve angle of each air compressor 10 based on the power difference between the air compressor 10 before and after the start-stop operation and the real-time pressure of the main pipe 20 by using the single-machine pressure model 312 of the air compressor 10.

Fig. 4A illustrates a further flow chart of a process for estimating a start-stop pressure curve angle for air compressor 10 (i.e., block 220) according to some embodiments of the present invention. FIG. 4B illustrates a schematic of a pressure curve 402 and a flow curve 404 of the parent pipe 20, according to some embodiments of the invention. As shown in fig. 4B, at a time t2, a start-stop operation of one of the air compressors 10 occurs, for example, the air compressor is changed from a power-off state to a power-on state, so that the flow rate of the main pipe 20 is increased instantaneously and the pressure is increased from the reduced pressure. The slope of the pressure curve 402 before and after time t2 forms an angle α, referred to herein as the start-stop pressure curve angle. The purpose of block 220 is to estimate the pressure curve angle of the air compressor 10 by using the power change before and after the start-stop operation, so as to estimate the displacement.

As shown in fig. 4A, at block 222, the computing device 40 may determine whether a start-stop operation of the air compressor 10 occurred at one detection time of the set of detection times. Specifically, it may be determined whether the turning on and off or the loading and unloading operation of each air compressor 10 has occurred around each detection time, for example, based on the operation record of each air compressor 10. If it is determined that the air compressor 10 is turned on or off or the loading and unloading operation is performed near a detection time, it may be determined that the air compressor 10 is turned on or off at the detection time. When the air compressor 10 starts or stops, the actual gas production of the air compressor 10 changes, so that the pressure measured by the main pipe 20 also changes.

If it is determined at block 222 that the air compressor 10 has started or stopped at the detected time, at block 224, the computing device 40 may obtain two real-time power values before and after the start or stop operation of the air compressor 10. Here, the two real-time power values may be determined from the real-time power of each air compressor 10 at a set of detection times acquired in block 210. On the contrary, if the start-stop operation does not occur at the detection timing, the data of the detection timing (not shown in the figure) is not used.

At block 226, computing device 40 may determine a power difference between before and after the start-stop operation of air compressor 10 based on the two real-time power values. The power difference is equal to the difference between the two real-time power values. For example, in the case where the start-stop operation is switched from off to on, the air compressor 10 is power-up, and therefore the power difference is positive and negative.

At block 228, the computing device 40 may estimate a start-stop pressure curve angle for the air compressor 10 based on the stand-alone pressure model of the air compressor 10 and the power difference determined at block 226. As shown in fig. 4B, when each air compressor 10 is started or stopped, the included angle of the pressure curve 402 mainly depends on the actual gas production rate of the air compressor 10, and the single-machine pressure model 312 is used to calculate the pressure included angle that can represent the gas production performance of the air compressor 10.

Here, the stand-alone pressure model 312 of the air compressor 10 is a trained stand-alone pressure model, as described in detail below in conjunction with fig. 9A and 9B. The stand-alone pressure model 312 may be, for example, a linear regression model, and through a training process, a stand-alone pressure model of each air compressor 10 can be obtained, wherein the stand-alone pressure model shows a relationship between a power difference of each air compressor 10 before and after the start-stop operation occurs and a start-stop pressure curve included angle in a pressure curve of the main pipe 20.

Continuing with fig. 2, at block 230, the computing device 40 may estimate the actual displacement of each air compressor 10 based on the real-time power of the air compressor 10 and the real-time flow rate of the main 20 using the single-machine displacement model 314 for that air compressor 10. Unlike the single-machine pressure model 312 using variation data, the single-machine displacement model 314 mainly uses data (i.e., stable data) of the air compressor 10 in a stable state to estimate the actual displacement, so as to avoid unstable measurement accuracy of the flow meter due to severe flow variation caused by unstable operation state of the air compressor.

In some embodiments, the computing device 40 may determine whether other air compressors than the air compressor 10 are in operation at one detection time in the set of detection times, and estimate the actual displacement of the air compressor 10 based on the stand-alone displacement model 314 of the air compressor 10 and the real-time flow rate of the main pipe 20 when it is determined that none of the other air compressors is in operation at a certain detection time.

Here, the single-unit displacement model 314 of the air compressor 10 is a single-unit displacement model trained using stable data of the main line pressure and the main line flow, as described in detail below with reference to fig. 10. The single-machine displacement model 314 may be, for example, a linear regression model, and through a training process, a single-machine displacement model of each air compressor 10 can be obtained, which shows the relationship between the power of each air compressor 10 and the flow rate of the main pipe 20 when no other air compressor is in an operating state. Therefore, at the detection timing, when it is determined that no other air compressor is in operation, the flow value of the header 20 can be estimated as the estimated value of the actual displacement of the air compressor 10 based on the stand-alone displacement model and the power of the air compressor 10.

Continuing with fig. 2, at block 240, the computing device 40 may determine a reason for the flow abnormality of the main duct 20 based on the real-time pressure and the real-time flow of the main duct 20 at the set of detection times, the start-stop pressure curve included angle of each air compressor 10 at the set of detection times determined at block 220, and the actual air displacement determined at block 230, using the main duct abnormality diagnosis model 320 of the main duct 20.

FIG. 5 illustrates a further flow chart of the process of determining the cause of the flow anomaly in the parent pipe 20 (i.e., block 240) according to some embodiments of the present invention.

As shown in fig. 5, block 240 may further include block 242, wherein computing device 40 may determine pressure data, displacement data, and equipment data for each air compressor 10 for air compressor plant 1 based on the start-stop pressure curve angle and actual displacement for each air compressor 10 estimated at

blocks

220 and 230.

Specifically, in some embodiments, the computing device 40 may combine the start-stop pressure curve included angles output by the single-machine pressure models of the plurality of air compressors 10 in the air compression station 1 to obtain the pressure data of the air compression station 1, combine the actual air displacement output by the single-machine air displacement models of the plurality of air compressors 10 in the air compression station 1 to obtain the air displacement data of the air compression station 1, and combine the start-stop pressure curve included angles output by the single-machine pressure models of each air compressor 10 and the actual air displacement output by the single-machine air displacement models to obtain the device data of the air compressors 10.

As described above, at

blocks

220 and 230, the calculation device 40 estimates the start-stop pressure curve included angle and the actual displacement of each air compressor 10 at each detection time, and at block 240, the calculation device 40 may combine the start-stop pressure curve included angle and the actual displacement of each air compressor 10 according to the time dimension as data for analyzing the air compression station 1. In addition, the calculation device 40 may combine the start-stop pressure curve included angle and the actual displacement of each air compressor 10 into the device data of the air compressor 10 according to the dimension (i.e., the device dimension) of the air compressor 10.

At block 244, the computing device 40 may determine a flow state of the air compressor station 1 based on the pressure data, displacement data, and plant data for each air compressor 10 of the air compressor station 1 using the parent pipe anomaly diagnostic model 320 of the parent pipe 20. Here, the flow rate state of the air compressing station 1 specifically refers to whether the flow rate (i.e., the flowmeter reading) of the parent pipe 20 is normal or abnormal, and in the case where the flow rate of the parent pipe 20 is abnormal, the cause of the flow rate abnormality of the parent pipe 20 is further determined.

FIG. 6A illustrates a schematic structural diagram of a parent pipe anomaly diagnostic model 320, according to some embodiments of the invention. As shown in fig. 6A, the parent pipe abnormality diagnosis model 320 may be a two-layer structure, with a lower layer including a pressure abnormality detection model 612, an exhaust amount abnormality detection model 614, and a device abnormality detection model 616, and an upper layer including a fault diagnosis model 620. Among them, the pressure abnormality detection model 612 is used to detect whether the pressure data of the air compressor station 1 is normal, the air displacement abnormality detection model 614 is used to detect whether the air displacement data of the air compressor station 1 is normal, the equipment abnormality detection model 616 is used to detect whether each equipment (i.e., the air compressor 10) is normal, and the fault diagnosis model 620 may comprehensively determine whether the flow rate state of the air compressor station 1 is normal and the cause of abnormality in the case of abnormal flow rate state based on the detection results of the pressure abnormality detection model 612, the air displacement abnormality detection model 614, and the equipment abnormality detection model 616.

In some embodiments of the present invention, the pressure abnormality detection model 612, the exhaust gas amount abnormality detection model 614, and the device abnormality detection model 616 are constructed using a novel detection (novelty detection) algorithm, respectively. The novel detection algorithm is an abnormal point detection algorithm, can be realized by using a single-Class Support Vector Machine (or called single-Class SVM), and is a semi-supervised learning algorithm. The novelty detection algorithm adopts a hypersphere to perform space division on input sample data, obtains a spherical boundary around the sample data in a feature space and enables the volume of the hypersphere to be minimum. The novelty detection algorithm can be expressed as finding a hypersphere that meets the following condition:

where o is the center of the hyper-sphere,ris the radius of the hyper-sphere,V(r) Is the volume of the hyper-sphere,x _iis the sample data of the mobile terminal and is,

is the slack variable corresponding to the sample data,Cis the corresponding penalty factor for the number of bits,mis the number of sample data. The objective of the above-described novelty detection algorithm is to find a way to make a collection of sample data

Minimized center o and radiusr. Note that the meaning of the sample data is different for different models, and the relaxation variables and penalty coefficients are also different. In the pressure abnormality detection model 612, the sample data is the pressure data of the air compressor station 1 (as described above in conjunction with block 242), in the exhaust gas amount abnormality detection model 614, the sample data is the exhaust gas amount data of the air compressor station 1 (as described above in conjunction with block 242), and in the equipment abnormality detection model 616, the sample data is the equipment data of each air compressor 10 (as described above in conjunction with block 242).

In the model training process, the center and radius parameters of the corresponding hyper-sphere are searched by using a set of various sample data (pressure data, air displacement data and equipment data). Here, the set of sample data may be obtained based on measured values of the air compressing station 1 at a plurality of consecutive time instants during operation. During model use, it may be determined whether newly acquired real-time sample data (pressure data, displacement data, equipment data) is located within the hyper-sphere. If the acquired real-time sample data is located in the hypersphere, the real-time sample data is determined to be normal data, and the flow state of the air compression station 1 is normal, otherwise, the real-time sample data is determined to be abnormal data (abnormal pressure data, abnormal air displacement data and abnormal equipment data), and the flow state of the air compression station 1 is abnormal.

Fig. 6B illustrates a further flow diagram of a process for determining a flow state of an air compression station (i.e., block 244) in accordance with some embodiments of the invention.

As shown in fig. 6B, at block 2442, a determination is made as to whether the pressure data for the air compressor station 1 determined at block 242 above is anomalous pressure data using the pressure anomaly detection model 612 based on the singularity detection algorithm.

At block 2444, computing device 40 may determine whether the exhaust gas amount data for air compressor station 1 determined at block 242 above is abnormal exhaust gas amount data using an exhaust gas amount abnormality detection model 614 based on a singularity detection algorithm.

At block 2446, the computing device 40 may determine whether the plant data for each air compressor 10 is anomalous plant data using the plant anomaly detection model 616 based on the novelty detection algorithm.

Those skilled in the art will appreciate that while the above blocks 2442 through 2446 are described in the order illustrated, the blocks need not be performed in the order illustrated and described, but may be performed in parallel or in other orders.

At block 2448, the computing device 40 may determine a flow state of the air compressor station 1 using the fault diagnosis model 620 based on the results of the pressure abnormality detection model 612, the displacement abnormality detection model 614, and the device abnormality detection model 616.

In some embodiments, if it is determined that there is no abnormal pressure data, abnormal displacement data, and abnormal equipment data, the computing device 40 may determine that the flow state of the air compression station 1 is normal at block 2448.

Conversely, if it is determined that at least one of the abnormal pressure data, the abnormal displacement data, and the abnormal device data is present, the computing device 40 may determine that the flow state of the air compression station 1 is abnormal at block 2448.

In some embodiments, upon determining that the flow state of the air compression station 1 is abnormal, the computing device 40 may further determine a cause of the flow state abnormality. Fig. 7 illustrates a flow chart of a process 700 for determining a cause of a traffic condition anomaly of the air compression station 1 according to some embodiments of the present invention.

As shown in fig. 7, if it is determined that the equipment data of at least one air compressor 10 of the plurality of air compressors 10 is abnormal equipment data (i.e., the air displacement or pressure of the air compressor 10 is abnormal) and the equipment data of the other air compressors 10 is not abnormal equipment data at block 710, it may be determined that the cause of the abnormality in the flow state of the air compressor station 1 is the performance abnormality of the at least one air compressor 10 at block 720.

If it is determined that the air displacement data of the air compression station 1 are both abnormal air displacement data and the pressure data are both abnormal pressure data at block 730, it may be determined that the cause of the abnormal flow state of the air compression station 1 is the abnormal pipe performance of the parent pipe 20 at block 740.

If it is determined at block 750 that the exhaust gas amount data of the air compression station 1 are all abnormal exhaust gas amount data but the pressure data is not abnormal pressure data, it may be determined at block 760 that the cause of the abnormal flow state of the air compression station 1 is a flow meter failure of the parent pipe 20.

As previously described, the stand-alone pressure model 312, the stand-alone displacement model 314, and the main pipe anomaly diagnosis model 320 in the model 300 are trained artificial intelligence models. For example, the stand-alone pressure model 312 and the stand-alone displacement model 314 may be constructed using linear regression models. In some embodiments according to the invention, a process of training the models is also included.

FIG. 8 illustrates a flow diagram of a process 800 for training the model 300 according to some embodiments of the invention. The process 800 may be performed, for example, prior to block 220 of the method 200, either as part of the method 200 or independent of the method 200.

As shown in fig. 8, at block 810, the computing device 40 may obtain a set of power samples, a set of pressure samples, and a set of flow samples for each air compressor 10, the manifold 20.

At block 820, the computing device 40 may train the stand-alone pressure model 312 based on the varying data in the power sample set of the air compressor 10 and the pressure sample set of the main duct 20. As shown in fig. 3, the inputs of the single-machine pressure model 312 are the power 301 of each air compressor 10 at each successive time and the pressure 302 of the main pipe 20 at the corresponding time, and the output is the start-stop pressure curve angle 304 of the air compressor 10. The acquisition of the change data is, for example, as described below with reference to fig. 9A.

At block 830, the computing device 40 may train the stand-alone displacement model 314 based on the stable data in the power sample set of the air compressor 10 and the flow sample set of the main duct 20. As shown in fig. 3, the inputs to the stand-alone displacement model 314 are the power 301 of each air compressor 10 at each successive time and the flow 303 of the mother pipe 20 at the corresponding time, and the output is the displacement 305 of that air compressor 10. The acquisition of the stable data is, for example, as described below with reference to fig. 10.

FIG. 9A illustrates a further flow diagram of the process of obtaining a set of pressure samples for the parent tube 20 (block 810) according to some embodiments of the invention.

As shown in fig. 9A, at block 811, the computing device 40 may obtain a record of the operation of the plurality of air compressors 10 and the main pipe 20 in the air compression station 1.

At block 812, the computing device 40 may filter the operation records to include the start-stop operation of each air compressor 10 and reject other data before and after the record within a first predetermined time. The first predetermined time may be several minutes, for example 3 to 5 minutes.

At block 813, the computing device 40 may determine, for each recording point, whether there is a recording point of the start-stop operation of another air compressor 10 within a second predetermined time before and after the recording point. The second predetermined time may also be several minutes, for example 3 to 5 minutes.

If it is determined that there is no recorded point of the start-stop operation of another air compressor 10 within the second predetermined time before and after the recorded point of the start-stop operation of one air compressor 10, at block 814, the computing device 40 may use two power data before and after the start-stop operation of the air compressor 10 as a set of variation data in the power sample set of the air compressor 10 and use two main pipe pressure data before and after the start-stop operation of the air compressor 10 as a set of pressure samples in the pressure sample set of the main pipe 20. Otherwise, the data of the recording point is not used as sample data (not shown in the figure).

In this way, the variation data in the power sample set of each air compressor 10 can be selected from the operation records of the air compressors 10 and the main pipe 20, so that the variation data can be further used for training the single-machine pressure model 312.

FIG. 9B illustrates a further flow diagram of the process of training the stand-alone pressure model 312 (block 820) according to some embodiments of the invention.

As shown in fig. 9B, at block 822, the computing device 40 may determine the power difference of the air compressor 10 before and after the start-stop operation based on a set of changing data in the power sample set of the air compressor 10. Note that block 822 is directed to the operation of the training phase of the stand-alone pressure model 312, and block 226 in FIG. 4A is directed to the operation of the use phase of the stand-alone pressure model 312, which are substantially the same and will not be described herein again.

At block 824, the computing device 40 may determine the pressure slope of the mother pipe 20 and the included pressure angle before and after the start-stop operation of the air compressor 10 based on the set of pressure samples corresponding to the set of variation data in the set of pressure samples of the mother pipe 20. Note that block 824 is directed to the operation of the training phase of the stand-alone pressure model 312, and block 228 in FIG. 4A is directed to the operation of the use phase of the stand-alone pressure model 312, which are substantially the same and will not be described herein again.

At block 826, the computing device 40 may train the stand-alone pressure model 312 using the power difference of the air compressor 10 before and after the start-stop operation and the pressure angle of the main pipe 20 before and after the start-stop operation of the air compressor 10 as a training sample.

In the case of using a linear regression model, the stand-alone pressure model 312 can be expressed as:

wherein

The included pressure angle before and after the start-stop operation of the air compressor 10 is shown,

respectively represents the power difference before and after the start-stop operation of each of the n air compressors 10,

are model parameters and need to be obtained through training.

Each air compressor machine 10 in the air compression station 1 can be power frequency air compressor machine or frequency conversion air compressor machine, and the power value of these two types of air compressors has different change law, consequently adopts different modes to the model training of these two types of air compressors.

Particularly, for the power frequency air compressor(e.g., the ith air compressor 10), since the power during its operation is substantially constant, when the start-stop operation occurs, the power will be changed from the rated power directly to 0 (stop operation) or from 0 directly to the rated power (start operation). In this case, the model parameters of the single-machine pressure model of the air compressor 10 may be determined based on the rated power of the air compressor 10 and the pressure included angle of the main pipe 20 before and after the start-stop operation of the air compressor 10ω _i. That is to say, in the case of a power frequency air compressor, the power difference between before and after the start-stop operation of the air compressor is equal to the rated power of the air compressor.

On the other hand, for the variable frequency air compressor (for example, the jth air compressor 10), the power during the operation thereof may significantly vary, so that the power difference of the air compressor 10 before and after the start-stop operation can be calculated as above, and the model parameter of the single-machine pressure model of the air compressor 10 can be determined based on the power difference and the pressure included angle of the main pipe 20 before and after the start-stop operation of the air compressor 10ω _j。

The output of the single-machine pressure model 312 is the pressure included angle of the main pipe when the air compressor is started and stopped.

For industrial frequency air compressors (e.g., the ith air compressor 10), the output of the single-machine pressure model 312 is the power difference of the industrial frequency air compressorx _iAnd (4) the pressure included angle when the power difference of other air compressors 10 is 0 is rated power.

For an inverter air compressor (e.g., the jth air compressor 10), the output of the single-machine pressure model 312 is the power difference of the inverter air compressorx _iAnd (3) the power difference of other air compressors 10 is the pressure included angle when the power difference is 0.

In some embodiments, block 820 may also include training the stand-alone pressure model 312 with a gradient descent algorithm. Specifically, the model parameters are obtained by training the stand-alone pressure model 312 with a training sample in block 826ω _iThe model parameters may then be modeled using a loss function based on the output values (i.e., estimated values of the power differences) of the stand-alone pressure model 312 and their actual power differencesω _iAnd performing iterative updating.

For example, assuming the sum of the squared mean errors as a function of the loss for the stand-alone pressure model 312, it can be expressed as:

whereinα' _iIs the output value of the stand-alone pressure model 312,α _iis the actual power difference.

Then, the loss function loss is determined with respect to each model parameterω _iPartial derivative of (2)

And updating the model parameters based on the partial derivatives (and the learning step size)ω _i。

The above steps are repeated until each model parameter converges, thereby obtaining each model parameter of the trained stand-alone pressure model 312.

FIG. 10 illustrates a further flow diagram of the process of obtaining a sample set of flows for the parent pipe 20 (block 810), according to some embodiments of the invention.

As shown in fig. 10, as in fig. 9A, at block 811, computing device 40 may obtain a record of the operation of the plurality of air compressors 10 and the parent pipe 20 in air compressor station 1.

At block 815, the computing device 40 may filter the operation records for which there is no operation of the air compressor for a third predetermined time. The third predetermined time may be several minutes, for example 3 to 5 minutes.

At block 816, the computing device 40 may select recording points from the recording points screened at block 815 that meet a predetermined stability condition, and at block 817 may use the power data of the air compressor 10 at the recording points as a stable data in the set of power samples of the air compressor 10 and the flow data of the main pipe 20 at the recording points as a flow sample in the set of flow samples of the main pipe 20.

Here, the recording point meeting the predetermined stable condition selected in block 816 may be, for example, a recording point satisfying the following condition:

the flow difference of the main pipe between the adjacent recording points is smaller than a preset flow difference threshold value;

the sum of the powers of the air compressors 10 of the air compression station 1 is less than a predetermined power and a threshold value; and

the sum of the specific powers of the plurality of air compressors 10 of the air compression station 1 is less than a predetermined specific power threshold.

Here, the predetermined flow rate difference threshold value, the predetermined power and threshold value, and the predetermined specific power threshold value may be determined according to the configuration of the air compressing station 1, which may be some empirical values.

In this way, the stable data in the power sample set of each air compressor 10 can be selected from the operation records of the air compressors 10 and the main pipe 20, so that the single machine displacement model 314 can be further trained by using the stable data.

In the case of using a linear regression model, the single-machine displacement model 314 may be expressed as:

wherein

The flow rate of the main pipe is shown,

respectively represent the power of n air compressors 10,

is a parameter of the model that is,bare bias terms that need to be derived through training.

In training the stand-alone displacement model 314 in block 830, the stand-alone displacement model 314 may be trained based on one stable data in the set of power samples for each air compressor 10 and the main pipe flow corresponding to the stable data as one training sample.

The output of the single-unit displacement model 314 is the actual displacement of each air compressor 10. For the ith air compressor 10, the output of the single machine displacement model 314 is the power of the air compressor 10x _iWhen the power of the other air compressors 10 is 0, the actual displacement of the air compressor 10 is obtained.

In some embodiments, block 830 may also include training the stand-alone displacement model 314 using a gradient descent algorithm. Specifically, in block 830, the model parameters are obtained by training the single-machine displacement model 314 using a training sampleω _iThereafter, the model parameters may be modeled using a loss function based on the output value of the stand-alone displacement model 314 (i.e., the estimated value of the displacement) and its actual displacementω _iAnd performing iterative updating.

For example, assuming a sum of squared average errors as a loss function of the single-machine displacement model 314, it can be expressed as:

whereiny' _iIs the output value, y, of the stand-alone displacement model 314_iIs the actual displacement.

Then, the loss function loss is determined with respect to each model parameterω _iPartial derivatives of

And partial derivative with respect to deviation term b

And updating the model parameters based on the partial derivatives (and the learning step size)ω _iAnd a deviation term b.

The above steps are repeated until each model parameter converges, thereby obtaining each model parameter of the trained single-machine displacement model 314.

Continuing with FIG. 8, at block 840, the computing device 40 may train the main pipe anomaly diagnosis model 320 based on the start-stop pressure curve included angle samples output by the stand-alone pressure model 312, the displacement samples output by the stand-alone displacement model 314, and the device samples for each air compressor.

Here, as described above, the main pipe abnormality diagnostic model 320 may be a two-layer structure, with the lower layer including the pressure abnormality detection model 612, the displacement abnormality detection model 614, and the equipment abnormality detection model 616, and the upper layer including the failure diagnostic model 620. As described above in connection with fig. 6A, in the case where the pressure abnormality detection model 612, the displacement abnormality detection model 614, and the device abnormality detection model 616 are constructed using the singularity detection algorithm, respectively, hyperspheres that meet the respective conditions may be found, and lagrange dual solution may be used for training.

In addition, in further embodiments, individual models may also be updated.

In some embodiments, an incremental update method may be used for the stand-alone pressure model 312 and/or the stand-alone displacement model 314 to iteratively update the stand-alone pressure model 312 and/or the stand-alone displacement model 314 every predetermined period of time with newly generated data for the predetermined period of time.

As for the pressure abnormality detection model 612, the exhaust gas amount abnormality detection model 614, and the equipment abnormality detection model 616, sample data whose outputs are normal samples (normal pressure samples, normal exhaust gas amount samples, and normal equipment samples) may be iteratively updated as new training data.

FIG. 11 illustrates a block diagram of a computing device 1100 suitable for implementing embodiments of the present invention. The computing device 1100 may be, for example, the computing device 40 in the air compression station 1 as described above.

As shown in fig. 11, computing device 1100 may include one or more Central Processing Units (CPUs) 1110 (only one of which is schematically shown) that may perform various suitable actions and processes in accordance with computer program instructions stored in a Read Only Memory (ROM) 1120 or loaded from a storage unit 480 into a Random Access Memory (RAM) 1130. In the RAM 1130, various programs and data required for operation of the computing device 1100 may also be stored. The CPU 1110, ROM 1120, and RAM 1130 are connected to each other by a bus 1140. An input/output (I/O) interface 1150 is also connected to bus 1140.

A number of components in the computing device 1100 connect to the I/O interface 1150, including: an input unit 1160 such as a keyboard, a mouse, or the like; an output unit 1170 such as various types of displays, speakers, and the like; a storage unit 1180 such as a magnetic disk, optical disk, or the like; and a communication unit 1190 such as a network card, modem, wireless communication transceiver, or the like. The communication unit 1190 allows the computing device 1100 to exchange information/data with other devices over a computer network, such as the internet, and/or various telecommunications networks.

Various methods or blocks described above may be performed, for example, by the CPU 1110 of one or more computing devices 1100. For example, in some embodiments, the methods or blocks may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 1180. In some embodiments, part or all of the computer program may be loaded onto and/or installed onto computing device 1100 via ROM 1120 and/or communications unit 1190. When the computer program is loaded into RAM 1130 and executed by CPU 1110, one or more operations of the methods described above may be performed. Further, the communication unit 1190 may support wired or wireless communication functions.

Those skilled in the art will appreciate that the computing device 1100 illustrated in FIG. 11 is merely illustrative. In some embodiments, computing device 1100 may contain more or fewer components than shown in FIG. 11.

By using the scheme of the invention, on one hand, the accuracy of the main pipe flowmeter can be detected in real time by using the main pipe pressure gauge and the flowmeter which are in standard configuration and the air compression station ammeter, and no additional hardware cost and data acquisition cost are required to be increased. On the other hand, the reason of the abnormal flow can be rapidly and accurately judged under the condition that the flow of the main pipe (the air compression station) is abnormal, so that early warning can be timely carried out.

The method for determining the flow state of the air compression station and the computing device capable of realizing the method are described in the specification by combining the attached drawings. It will be appreciated by those skilled in the art, however, that the performance of the various blocks of the method or portions thereof described above is not limited to the order shown in the figures and described above, but may be performed in any other reasonable order. Further, the computing device 1100 also need not include all of the components shown in FIG. 11, it may include only some of the components necessary to perform the functions described in the present invention, and the manner in which these components are connected is not limited to the form shown in the figures.

The present invention may be methods, apparatus, systems and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied therein for carrying out aspects of the present invention.

In one or more exemplary designs, the functions described herein may be implemented in hardware, software, firmware, or any combination thereof. For example, if implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium.

The units of the apparatus disclosed herein may be implemented using discrete hardware components, or may be integrally implemented on a single hardware component, such as a processor. For example, the various illustrative logical blocks, modules, and circuits described in connection with the invention may be implemented or performed with a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein.

Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm blocks described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both.

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

Claims

1. A method for determining a flow state of an air compressor station, wherein the air compressor station includes a plurality of air compressors and a main connecting the plurality of air compressors, the method comprising:

acquiring real-time power of each air compressor in the plurality of air compressors, real-time pressure and real-time flow of the main pipe at a group of detection moments;

estimating a start-stop pressure curve included angle of each air compressor based on the power difference before and after start-stop operation of the air compressor and the real-time pressure of the main pipe by using a single-machine pressure model of each air compressor;

estimating the actual displacement of each air compressor based on the real-time power of the air compressor and the real-time flow of the main pipe by using a single-machine displacement model of each air compressor; and

and determining the flow state of the air compression station by using a main pipe abnormity diagnosis model of the main pipe based on the real-time pressure and the real-time flow of the main pipe at the group of detection moments and the start-stop pressure curve included angle and the actual air displacement of each air compressor at the group of detection moments.

2. The method of claim 1, wherein estimating the start-stop pressure curve angle of the air compressor based on the power difference before and after the start-stop operation of the air compressor and the real-time pressure of the main pipe comprises:

determining whether the air compressor is started or stopped at one detection moment in the group of detection moments;

if the fact that the air compressor is started and stopped at the detection moment is determined, two real-time power values before and after the air compressor is started and stopped are obtained;

determining a power difference before and after starting and stopping operation of the air compressor based on the two real-time power values; and

and estimating the start-stop pressure curve included angle of the air compressor based on the single-machine pressure model of the air compressor and the power difference.

3. The method of claim 1, wherein estimating the actual displacement of the air compressor based on the real-time power of the air compressor and the real-time flow rate of the main comprises:

determining whether or not other air compressors than the air compressor are in an operating state at one detection time of the set of detection times; and

and if the fact that other air compressors except the air compressor are not in the running state at the detection moment is determined, estimating the actual air displacement of the air compressor based on a single-machine air displacement model of the air compressor and the real-time flow of the main pipe.

4. The method of claim 1, wherein determining the flow status of the air compression station comprises:

determining pressure data and air displacement data of the air compression station and equipment data of each air compressor based on the start-stop pressure curve included angle and the actual air displacement of each air compressor; and

and determining the flow state of the air compression station based on the pressure data, the air displacement data and the equipment data of each air compressor by using a main pipe abnormity diagnosis model of the main pipe.

5. The method of claim 4, wherein determining the pressure data, the displacement data, and the equipment data for each air compressor based on the start-stop pressure curve angle and the actual displacement for each air compressor comprises:

combining start-stop pressure curve included angles output by the single-machine pressure models of the plurality of air compressors to acquire pressure data of the air compression station;

combining actual air displacement output by the single air displacement models of the plurality of air compressors to obtain air displacement data of the air compression station; and

and combining the start-stop pressure curve included angle output by the single-machine pressure model of each air compressor and the actual air displacement output by the single-machine air displacement model to acquire the equipment data of the air compressors.

6. The method of claim 4, wherein the parent pipe anomaly diagnostic model includes a pressure anomaly detection model, an air displacement anomaly detection model, and a device anomaly detection model, and a fault diagnostic model, and wherein determining, using the parent pipe anomaly diagnostic model of the parent pipe, a flow state of the air compressor station based on pressure data, air displacement data, and device data of each air compressor comprises:

determining whether the pressure data of the air compression station is abnormal pressure data by using a pressure abnormality detection model based on a novel detection algorithm;

determining whether the air displacement data of the air compression station is abnormal air displacement data by using an air displacement abnormal detection model based on a novel detection algorithm;

determining whether the equipment data of the air compressor is abnormal equipment data or not by using an equipment abnormality detection model based on a novelty detection algorithm; and

determining a flow rate state of the air compression station using the failure diagnosis model based on results of the pressure abnormality detection model, the displacement abnormality detection model, and the equipment abnormality detection model.

7. The method of claim 6, wherein determining the flow state of the air compression station using the fault diagnosis model comprises: if the abnormal pressure data, the abnormal air displacement data and the abnormal equipment data do not exist, determining that the flow state of the air compression station is normal; and

determining that a flow state of the air compression station is abnormal if it is determined that at least one of abnormal pressure data, abnormal displacement data, and abnormal equipment data exists.

8. The method of claim 7, further comprising:

if the equipment data of at least one air compressor in the plurality of air compressors is determined to be abnormal equipment data and the equipment data of other air compressors is not abnormal equipment data, determining that the reason that the flow state of the air compression station is abnormal is that the performance of the at least one air compressor is abnormal;

if the exhaust gas volume data of the air compression station are determined to be abnormal exhaust gas volume data and the pressure data are determined to be abnormal pressure data, determining that the reason that the flow state of the air compression station is abnormal is that the pipeline performance of the main pipe is abnormal; and

and if the exhaust gas volume data of the air compression station are determined to be abnormal exhaust gas volume data but the pressure data are not abnormal pressure data, determining that the reason that the flow state of the air compression station is abnormal is the flow meter of the main pipe is in failure.

9. The method of claim 1, wherein before estimating the start-stop pressure curve angle of the air compressor based on the power difference before and after the start-stop operation of the air compressor and the real-time pressure of the main pipe, the method further comprises:

acquiring a power sample set, a pressure sample set and a flow sample set of each air compressor;

training the stand-alone pressure model based on the variation data in the power sample set and the pressure sample set of the master pipe; and

and training the single machine displacement model based on stable data in the power sample set and the flow sample set of the main pipe.

10. The method of claim 9, wherein determining the flow state of the air compression station further comprises:

and training a main pipe abnormity diagnosis model based on the start-stop pressure curve included angle sample output by the single-machine pressure model, the air displacement sample output by the single-machine air displacement model and the equipment sample of each air compressor.

11. The method of claim 9, wherein obtaining the set of pressure samples for the parent tube comprises:

acquiring operation records of the plurality of air compressors and the main pipe;

screening recording points including starting and stopping operations of the air compressor from the operation records and eliminating other data in first preset time before and after the recording points;

for each recording point, determining whether the recording point of the start-stop operation of other air compressors exists within second preset time before and after the recording point; and

and if it is determined that no recording point of start-stop operation of other air compressors exists in second preset time before and after the recording point of the start-stop operation of the air compressors, taking two pieces of power data before and after the start-stop operation of the air compressors as a group of change data in a power sample set of the air compressors and taking two pieces of main pipe pressure data before and after the start-stop operation of the air compressors as a group of pressure samples in a pressure sample set of the main pipes.

12. The method of claim 11, wherein training the stand-alone pressure model based on the varying data in the set of power samples and the set of pressure samples of the parent pipe comprises:

determining a power difference of the air compressor before and after the start-stop operation based on a group of variation data in a power sample set of the air compressor;

determining a pressure slope of the main pipe and a pressure included angle before and after starting and stopping operation of the air compressor on the basis of a group of pressure samples corresponding to the group of change data in the pressure sample set of the main pipe; and

and training the single-machine pressure model by taking the power difference of the air compressor before and after the start-stop operation and the pressure included angle of the main pipe before and after the start-stop operation of the air compressor as a training sample.

13. The method of claim 12, wherein the stand-alone pressure model comprises a linear regression model, and when the air compressor is a variable frequency air compressor, the model parameters of the stand-alone pressure model of the air compressor are determined based on a power difference of the air compressor before and after the start-stop operation and a pressure included angle of the main pipe before and after the start-stop operation of the air compressor.

14. The method of claim 12, wherein the stand-alone pressure model comprises a linear regression model, and when the air compressor is a power frequency air compressor, the model parameters of the stand-alone pressure model of the air compressor are determined based on the rated power of the air compressor and a pressure included angle of the main pipe before and after the start-stop operation of the air compressor.

15. The method of claim 9, wherein obtaining a flow sample set for the parent pipe comprises:

screening record points without air compressor operation within third preset time from the operation records;

selecting recording points which meet a preset stable condition from the screened recording points; and

and taking the power data of the air compressor at the recording point meeting the preset stable condition as one stable data in a power sample set of the air compressor, and taking the flow data of the main pipe at the recording point meeting the preset stable condition as one flow sample in a flow sample set of the main pipe.

16. The method of claim 15, wherein the predetermined stable condition comprises:

the flow difference of the main pipe between the recording point and the adjacent recording point is smaller than a preset flow difference threshold value;

the sum of the power of the plurality of air compressors is less than the preset power and the threshold value; and

the sum of the specific powers of the air compressors is smaller than a preset specific power threshold.

17. The method of claim 15, wherein the stand-alone displacement model comprises a linear regression model, and wherein training the stand-alone displacement model based on the stable data in the set of power samples and the set of bus master flow samples comprises:

and training the single-machine air displacement model based on stable data in the power sample set of each air compressor and the flow of the main pipe corresponding to the stable data.

18. A computing device, comprising:

at least one processor; and

at least one memory coupled to the at least one processor and storing instructions for execution by the at least one processor, the instructions when executed by the at least one processor causing the computing device to perform the method of any of claims 1-17.

19. A computer readable storage medium having stored thereon computer program code which, when executed, performs the method of any of claims 1 to 17.