Pipeline risk assessment method based on integrity management

Abstract

The invention relates to the technical field of pipeline risk assessment, and discloses a pipeline risk assessment method based on integrity management, comprising the following steps: obtaining multiple risk factors of pipeline failure and the risk probability corresponding to each risk factor based on historical data, encoding based on the risk probability corresponding to each risk factor, so that the associated risk factors have the same encoding rule; using a tree structure iteration module to construct an evaluation model for pipeline failure risk, so that the evaluation model has sub-nodes formed according to the encoding and the encoding rule, and each sub-node forms a different node level according to different risk probabilities; obtaining pipeline operation monitoring data, and using the monitoring data as a basis to evaluate the pipeline failure risk in the pipeline operation process with an evaluation model; the evaluation model for pipeline failure risk of the invention includes a plurality of sub-nodes of a single risk assessment model, and there is correlation between the sub-nodes, so as to complete the integrity evaluation of pipeline management.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline risk assessment, and in particular to a pipeline risk assessment method based on integrity management. Background Art

[0002] At present, pipeline risk assessment is performed according to different risk factors to obtain different risk assessment values. For example, the publication number "CN113128803B" discloses a method, device and computer equipment for determining oil and gas pipeline risks, including: obtaining pipeline construction period data, associating the construction period data with the basic risk assessment indicators of the initial stage of the pipeline according to the data type; for multiple monitoring locations on the pipeline, based on the construction period data of different indicators of each monitoring location, calculating the risk assessment value of each indicator; for each monitoring location, performing weighted summation on the acquired risk assessment values, and using the obtained sum as the initial risk value of the monitoring location; based on the initial risk values ​​of each monitoring location on the pipeline, determining an initial risk baseline and an initial baseline risk point of the pipeline, the initial risk baseline is used The invention relates to a method for identifying a unified risk measurement standard at each monitoring position on the pipeline, obtaining operation period data of the pipeline, wherein the operation period data refers to dynamic data of the pipeline during operation; obtaining real-time risk assessment values ​​of multiple indicators of multiple monitoring positions of the pipeline according to the operation period data of the pipeline; determining risk values ​​of the multiple monitoring positions based on the real-time risk assessment values ​​of multiple indicators of the multiple monitoring positions of the pipeline; comparing the risk values ​​of the multiple monitoring positions with the initial risk value of the pipeline, wherein one indicator is associated with at least one type of operation period data, and the initial risk value of the pipeline includes the initial risk baseline and the initial baseline risk point of the pipeline; and displaying the multiple monitoring positions differently in a visual interface according to the different risk levels of the multiple monitoring positions according to the comparison results.

[0003] Since there are many risk factors involved in pipelines, and there is a certain correlation between different risk factors, for example, the risk of pipeline decompression is related to the pressure of the fluid in the pipeline, and is also related to factors such as pipeline leakage. Therefore, there is a correlation between risk factors, and the risk of the pipeline cannot be reflected by a single one or more indicators. Summary of the invention

[0004] In view of this, the main purpose of the present invention is to provide a pipeline risk assessment method based on integrity management.

[0005] The technical solution adopted by the invention is as follows: a pipeline risk assessment method based on integrity management, comprising the following steps: obtaining multiple risk factors of pipeline failure and the risk probability corresponding to each risk factor based on historical data, encoding based on the risk probability corresponding to each risk factor, so that the associated risk factors have the same encoding rules; using a tree-structured iterative module to construct an evaluation model for pipeline failure risk, so that the evaluation model has sub-nodes constructed according to the encoding and encoding rules, and each sub-node forms a different node level according to the different risk probabilities; obtaining monitoring data of pipeline operation, and using the monitoring data as a basis, evaluating the pipeline failure risk during pipeline operation through an evaluation model.

[0006] Furthermore, the risk factors include: pipeline decompression risk and leakage risk caused by exceeding the limit of pipeline flow data and pipeline pressure data; pipeline corrosion risk formed by pipeline corrosion status and pipeline velocity data; pipeline damage risk caused by exceeding the limit of pipeline vibration data; and pipeline crack risk obtained from eddy current detection data.

[0007] Furthermore, the method for making the associated risk factors have the same coding rule is: obtain each risk factor and the risk probability corresponding to each risk factor; extract the risk probability corresponding to each risk factor through the first thread, and sort them from large to small according to the risk probability to form a comparison library; extract each risk factor through the second thread, and input the risk factors into the association library in turn for risk association analysis, and confirm the risk correlation between multiple risk factors; call the comparison library to determine whether multiple risk factors with risk correlation are arranged from large to small according to the risk probability. If so, the multiple risk factors are graded according to the set risk level range to form a graded definition result, and the graded definition result is encoded, so that the associated risk factors have the same first coding rule according to the graded definition result; if not, they are encoded according to the set risk level range, wherein the risk factors in the same risk level range have the same second coding rule.

[0008] Furthermore, the encoding is performed according to a set encoding program.

[0009] Furthermore, the coding program has: a coding area for setting differentiation codes according to a set risk level range; a first rule area for defining a first coding pattern for risk factors having the same risk level range, and forming a second coding rule with the first coding pattern; and a second rule area for defining a second coding pattern for risk factors within the same grading definition result, and forming a first coding rule with the second coding pattern.

[0010] Furthermore, the first coding pattern has a first level bit for indicating a risk level and a first distinction bit for distinguishing risk factors.

[0011] Furthermore, the second coding pattern comprises a grading delimitation bit for indicating a grading delimitation result, a second grade bit for indicating a risk grade, and a second distinguishing bit for distinguishing risk factors.

[0012] Furthermore, the iteration module has: multiple iteration units, each iteration unit uses a neural network model to perform an iteration operation, and is used to iteratively train the historical data of the corresponding risk factors through the neural network model; an association management unit, which is used to correspond multiple risk factors to multiple iteration units one by one according to the encoding and encoding rules, and use a tree structure to arrange the hierarchy of the iteration units.

[0013] Furthermore, the iteration unit comprises: a memory for storing historical data assigned to corresponding risk factors; and an executor for calling a neural network model according to an executive operation to perform iterative training on the historical data of corresponding risk factors.

[0014] Furthermore, the executive operation refers to simultaneously turning on the actuators provided in each iteration unit through an executive switch to perform synchronous iterative training.

[0015] Furthermore, the tree structure obtained by the association management unit is used to construct an evaluation model for pipeline failure risk.

[0016] Furthermore, the evaluation model has sub-nodes formed according to the iteration unit, and each sub-node forms a different node level according to the different risk probabilities.

[0017] Furthermore, the method for constructing a pipeline failure risk evaluation model using a tree structure specifically includes:

[0018] Obtaining the codes and coding rules of multiple risk factors, and obtaining the corresponding relationship between each risk factor and each iteration unit;

[0019] Parsing the coding rules to obtain a first level bit for indicating the risk level and a first distinction bit for distinguishing the risk factors in the first coding rule, and a grading demarcation bit for indicating the grading demarcation result, a second level bit for indicating the risk level and a second distinction bit for distinguishing the risk factors in the second coding rule;

[0020] A tree structure is established with the first level bit, the classification boundary bit and the second level bit, and the iteration units are sequentially embedded into the tree structure according to the corresponding relationship between the iteration units and the risk factors;

[0021] A neural network model is configured for each child node of the tree structure to construct an evaluation model for pipeline failure risk, wherein a memory and an executor are configured for each neural network model, the memory is used to store historical data assigned to corresponding risk factors, and the executor is used to call the neural network model according to executive operations to iteratively train the historical data of the corresponding risk factors.

[0022] This application obtains multiple risk factors that cause pipeline failure through historical data of pipeline monitoring, and obtains the risk probability corresponding to each risk factor, and encodes the risk factor based on the risk probability corresponding to each risk factor. When encoding, the correlation between different risk factors is fully considered, and the associated risk factors have the same encoding rules. At the same time, a tree-structured iterative module is used to construct an evaluation model for pipeline failure risk. In this way, the evaluation model for pipeline failure risk actually contains several sub-nodes of a single risk evaluation model; and there is correlation between these sub-nodes, which can complete the integrity evaluation of pipeline management. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The following drawings are only used to illustrate and explain the present invention, and are not used to limit the scope of the present invention, wherein:

[0024] Figure 1 is a flow chart of the method of the present invention;

[0025] Figure 2 It is a schematic diagram of encoding according to a set encoding program in an embodiment of the present invention;

[0026] Figure 3 A flow chart of a method for making associated risk factors have the same coding rules in the present invention;

[0027] Figure 4 A flow chart of a method for constructing a pipeline failure risk evaluation model using a tree structure in the present invention;

[0028] Figure 5 It is a schematic diagram of the structure of the pipeline risk assessment system based on integrity management of the present invention;

[0029] Figure 6 Schematic diagram of the neural network model structure in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution, design method and advantages of the present invention more clear, the present invention is further described in detail by specific examples in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0031] Example 1, reference Figure 1 and Figure 2 The present application discloses a pipeline risk assessment method based on integrity management, comprising the following steps: obtaining multiple risk factors of pipeline failure and the risk probability corresponding to each risk factor based on historical data, encoding based on the risk probability corresponding to each risk factor, so that the associated risk factors have the same encoding rule; constructing an evaluation model of pipeline failure risk using a tree-structured iterative module, so that the evaluation model has sub-nodes constructed according to the encoding and encoding rules, and each sub-node forms a different node level according to the different risk probabilities; obtaining monitoring data of pipeline operation, and using the monitoring data as a basis, evaluating the pipeline failure risk during pipeline operation through the evaluation model.

[0032] In the above, the above historical data are various monitoring data obtained during the operation of the pipeline, including pressure data obtained by a pressure sensor installed on the pressure pipeline, flow data obtained by a flow sensor, pipeline corrosion monitoring data, pipeline vibration data and eddy current detection data; wherein, the pipeline corrosion monitoring data can utilize a combination of one or more of a resistance probe, an inductance probe and an electrochemical probe, the pipeline vibration data adopts a pipeline instantaneous pulsation analysis method, and utilizes an eddy current monitoring method to monitor pipeline surface defects, cracks, etc.

[0033] In the above, the pipeline transient pulsation analysis method can be pressure transient pulsation. Pressure transient pulsation should exclude the influence caused by the pressure fluctuation of the fluid itself. Generally, the pressure fluctuation of the fluid itself may be caused by the unstable pressure during the fluid pressurization and transportation process, or it may be caused by pipeline leakage. When excluding, the following method can be used: obtain the duration and fluctuation frequency of the pressure transient pulsation, and determine whether the fluctuation frequency decreases in a descending order with time as a reference. If such a pattern exists, it is determined to be caused by pipeline vibration.

[0034] In the above, since the pipeline vibrates with a frequency from large to small (excluding the possibility of continuous vibration at a certain frequency, such as the vibration of the pipeline caused by continuous hitting the pipeline with a certain impact force, generally the pipeline needs to vibrate continuously, the impact force must be large enough, excluding human factors, and it is impossible to happen under natural factors, so this situation is just an assumption), the pressure in the pipeline also fluctuates with a frequency from large to small. For example, when the pipeline bursts, a huge pressure fluctuation will be formed at the next monitoring point of the pipeline burst, and then the pressure fluctuation will gradually change with a certain pressure loss law due to the pressure loss.

[0035] In the above, the risk factors include: pipeline decompression risk and leakage risk caused by exceeding the limit of pipeline flow data and pipeline pressure data; pipeline corrosion risk formed by pipeline corrosion status and pipeline velocity data; pipeline damage risk caused by exceeding the limit of pipeline vibration data; pipeline crack risk obtained by eddy current detection data.

[0036] In the above, the risk of pipeline decompression can be judged as follows: from historical data, when the flow rate is constant, the pressure in the pipeline is set within a certain flow range as a reference value. In this way, when the pressure in the pipeline exceeds the set reference value, there will be a risk of pipeline decompression. For the leakage risk, from historical data, when the set flow rate and pressure are constant, the set flow rate and pressure are used as reference values. When the flow rate fluctuates under the reference value, and the pressure in the pipeline is also under the reference value at this time, it is considered that there is a leakage risk. At the same time, the probability distribution of pipeline decompression risk and leakage risk can also be obtained.

[0037] Similarly, from the historical monitoring data, the corrosion status of multiple corrosion points at different times can be obtained. By judging the corrosion status at different times, the theoretical pipeline rate can be obtained, and the pipeline corrosion risk under the theoretical state can be calculated based on the corrosion rate. At the same time, the probability of pipeline corrosion risk can also be obtained based on the current corrosion status and corrosion rate.

[0038] In the above, since there is a vibration frequency from large to small when the pipeline vibrates, the pressure in the pipeline is also accompanied by a fluctuation frequency from large to small. For example, when the pipeline bursts, a huge pressure fluctuation will be formed at the next monitoring point of the pipeline burst, and then the pressure fluctuation will gradually change with a certain pressure loss law due to the pressure loss. Therefore, by monitoring the instantaneous changes in pressure and flow, when the instantaneous changes are large and cannot be restored to the set flow and pressure values ​​in the later monitoring, it means that there is a risk of pipeline damage. At the same time, the probability of pipeline damage risk can also be obtained according to the above judgment method.

[0039] In the above, the pipeline crack risk can be obtained through eddy current detection data obtained at different times, and the theoretical curve of pipeline crack risk can be formed by eddy current detection data obtained at different times. According to the theoretical curve, the upper limit of pipeline crack risk can be roughly obtained. At the same time, the probability of pipeline cracks can also be obtained by the current pipeline crack state and the theoretical curve.

[0040] In the above, pipeline decompression risk and leakage risk are generally used as the second level of the tree structure. This is because pipeline decompression risk includes factors such as pipeline corrosion, damage, pipeline vibration, and pipeline cracks, and is not just caused by unstable pressurization during fluid pressurization. Other risk factors can be determined at which level based on risk probability.

[0041] In the above, the method for making the associated risk factors have the same coding rule is: obtain each risk factor and the risk probability corresponding to each risk factor; extract the risk probability corresponding to each risk factor through the first thread, and sort them from large to small according to the risk probability to form a comparison library; extract each risk factor through the second thread, and input the risk factors into the association library in turn for risk association analysis, and confirm the risk correlation between multiple risk factors; call the comparison library to determine whether multiple risk factors with risk correlation are arranged from large to small according to the risk probability. If so, the multiple risk factors are graded according to the set risk level range to form a graded definition result, and the graded definition result is encoded, so that the associated risk factors have the same first coding rule according to the graded definition result; if not, they are encoded according to the set risk level range, wherein the risk factors in the same risk level range have the same second coding rule.

[0042] In the above, the coding is performed according to a set coding procedure. Furthermore, the coding procedure has: a coding area for setting differentiation codes according to a set risk level range; a first rule area for defining a first coding pattern for risk factors having the same risk level range, and forming a second coding rule with the first coding pattern; a second rule area for defining a second coding pattern for risk factors within the same grading definition result, and forming a first coding rule with the second coding pattern. Among them, the first coding pattern has a first level position for indicating the risk level and a first differentiation position for distinguishing risk factors. The second coding pattern has a grading definition position for indicating the grading definition result, a second level position for indicating the risk level, and a second differentiation position for distinguishing risk factors.

[0043] In the above, the first level digit is used to represent and distinguish risk factors with the same risk level range, such as Figure 2 As shown, the pipeline decompression risk and the leakage risk have the same level, therefore, the pipeline decompression risk and the leakage risk have the same first level bit when they are coded, and the first level bit can be represented by one or more combinations of letters, symbols, and numbers, and the same first level bit has the same coding rule. Similarly, the first distinction bit, the graded boundary bit, the second level bit, and the second distinction bit can be coded in an equivalent manner.

[0044] In the above, the iteration module has: a plurality of iteration units, each iteration unit uses a neural network model {the neural network model is a RNN neural network (Recurrent Neural Network, recursive neural network), the recursive neural network mainly includes: an input layer, a hidden layer, an output layer and a softmax layer, such as Figure 6As shown, specifically, the historical data of the corresponding risk factors are input into the input layer, the input layer receives the historical data of the corresponding risk factors and passes it to the hidden layer, and the hidden layer models the historical data of the corresponding risk factors; the neurons in the hidden layer receive their own outputs after a fixed delay (one or more iterations); the output layer converts the historical data of the corresponding risk factors into probability distribution through the softmax layer by setting weights and thresholds of different types of risk probabilities, and determines the distribution of historical data;

[0045] Among them, the output expression after softmax regression processing is:

[0046]

[0047] In the formula, y i The probability of pipeline failure caused by various reasons;

[0048] Among them, y i =y g xy t xy a ;

[0049] In the formula, y i --The probability of pipeline failure caused by various reasons;

[0050] y g --General failure probability, the average failure probability of pipelines obtained by counting a large number of accident cases;

[0051] y t --The proportion of each failure mode, including small pipeline leakage, large pipeline leakage and rupture;

[0052] y a --Correction coefficient;

[0053] where y g and t It is obtained based on the historical failure database, y a It is the correction coefficient obtained by comparing the actual situation of the pipeline with the actual situation of the pipeline in the historical database;

[0054] Among them, the weight can be set according to the probability of occurrence of different types of risk factors in the historical data, the threshold can be understood as the limit value of a certain risk factor at different risk levels, and can be set according to the historical data} to perform iterative operations (the iterative operation can be understood as inputting the historical data of the corresponding risk factor into the input layer, the input layer receives the historical data of the corresponding risk factor and passes it to the hidden layer, the neurons in the hidden layer will receive its own output after a fixed delay, the output layer determines the distribution of historical data by setting the weights and thresholds of different types of risk probabilities, and the process of inputting historical data into the recursive neural network for calculation is an iterative operation), and is used to iteratively train the historical data of the corresponding risk factor through the neural network model; the association management unit is used to correspond multiple risk factors to multiple iteration units one by one according to the encoding and encoding rules, and use a tree structure to hierarchically arrange the iteration units. The iteration unit has: a memory for storing the historical data assigned to the corresponding risk factor; an actuator for calling the neural network model according to the executive operation to iteratively train the historical data of the corresponding risk factor. The executive operation refers to turning on the actuators set in each iteration unit at the same time through the executive switch to perform synchronous iterative training.

[0055] In the above, the tree structure obtained by the association management unit is used to construct the pipeline failure risk evaluation model. The evaluation model has child nodes formed according to the iteration unit, and each child node forms a different node level according to the different risk probabilities, and a tree structure is formed through the node level. The specific tree structure is as follows: Figure 2 As shown; further, the method for constructing an evaluation model for pipeline failure risk with a tree structure specifically includes: obtaining the codes and coding rules of multiple risk factors, and obtaining the corresponding relationship between each risk factor and each iteration unit; parsing the coding rules to obtain a first level bit for indicating the risk level and a first distinction bit for distinguishing the risk factors in the first coding rule, and a grading boundary bit for indicating the grading boundary result, a second level bit for indicating the risk level and a second distinction bit for distinguishing the risk factors in the second coding rule; establishing a tree structure with the first level bit, the grading boundary bit and the second level bit, and correspondingly embedding the iteration units into the tree structure in sequence according to the corresponding relationship with the risk factors; configuring a neural network model for each child node of the tree structure to construct an evaluation model for pipeline failure risk, wherein a memory and an executor are configured for each neural network model, the memory is used to store the historical data assigned to the corresponding risk factor, and the executor is used to call the neural network model according to the execution operation to iteratively train the historical data of the corresponding risk factor.

[0056] This application obtains multiple risk factors that cause pipeline failure through historical data of pipeline monitoring, and obtains the risk probability corresponding to each risk factor, and encodes the risk factor based on the risk probability corresponding to each risk factor. When encoding, the correlation between different risk factors is fully considered, and the associated risk factors have the same encoding rules. At the same time, a tree-structured iterative module is used to construct an evaluation model for pipeline failure risk. In this way, the evaluation model for pipeline failure risk actually contains several sub-nodes of a single risk evaluation model; and there is correlation between these sub-nodes, which can complete the integrity evaluation of pipeline management.

[0057] Embodiment 2. The present application also discloses a pipeline risk assessment system based on integrity management, including: a risk factor determination unit, which obtains multiple risk factors of pipeline failure and the risk probability corresponding to each risk factor based on historical data; a coding unit, which encodes based on the risk probability corresponding to each risk factor so that the associated risk factors have the same coding rules; an evaluation model construction unit, which is used to construct an evaluation model of pipeline failure risk using an iterative module of a tree structure, so that the evaluation model has sub-nodes constructed according to the coding and coding rules, and each sub-node forms a different node level according to the different risk probabilities; an evaluation driving unit, which is used to obtain monitoring data of pipeline operation in real time, and evaluate the pipeline failure risk during pipeline operation through an evaluation model based on the monitoring data.

[0058] In the above, the risk factors generally include: pipeline decompression risk and leakage risk caused by exceeding the limit of pipeline flow data and pipeline pressure data; pipeline corrosion risk formed by pipeline corrosion status and pipeline velocity data; pipeline damage risk caused by exceeding the limit of pipeline vibration data; pipeline crack risk obtained by eddy current detection data.

[0059] It should be noted that in addition to the above-mentioned risks, other risk factors are also included. This application exemplarily describes the main risk factors of pipelines. Although the probability of other factors is lower, they are also within the scope of consideration of this application and therefore also fall within the scope of protection of this application.

[0060] In the above, the encoding unit includes: a data loading unit, which is used to obtain each risk factor and the risk probability corresponding to each risk factor; an encoding driving unit, which is connected to the data loading unit, and the encoding driving unit is used to generate a first thread, a second thread and a cache unit for temporary data storage according to the driver therein, wherein the risk probability corresponding to each risk factor is extracted by the first thread, and the risk probability is sorted from large to small to form a comparison library; each risk factor is extracted by the second thread, and the risk factor is sequentially input into the association library for risk association analysis, the risk correlation between multiple risk factors is confirmed, and the obtained comparison library and risk correlation are stored by the cache unit; a judgment and processing unit, which is connected to the driving unit, and the judgment and processing unit is used to call the comparison library to determine whether multiple risk factors with risk correlation are arranged from large to small according to risk probability, if so, the multiple risk factors are graded according to the set risk level range to form a graded definition result, and the graded definition result is encoded, and the associated risk factors have the same first encoding rule according to the graded definition result; if not, they are encoded according to the set risk level range, wherein the risk factors in the same risk level range have the same second encoding rule.

[0061] In the above, the coding is performed according to a set coding procedure. Furthermore, the coding procedure has: a coding area for setting differentiation codes according to a set risk level range; a first rule area for defining a first coding pattern for risk factors having the same risk level range, and forming a second coding rule with the first coding pattern; a second rule area for defining a second coding pattern for risk factors within the same grading definition result, and forming a first coding rule with the second coding pattern. The first coding pattern has a first level bit for indicating the risk level and a first differentiation bit for distinguishing risk factors. The second coding pattern has a grading boundary bit for indicating the grading definition result, a second level bit for indicating the risk level, and a second differentiation bit for distinguishing risk factors.

[0062] In the above, the iteration module has: multiple iteration units, each iteration unit uses a neural network model to perform an iteration operation, and is used to iteratively train the historical data of the corresponding risk factors through the neural network model; an association management unit, which is used to correspond multiple risk factors to multiple iteration units one by one according to the encoding and encoding rules, and use a tree structure to arrange the hierarchy of the iteration units.

[0063] In the above, the iteration unit comprises: a memory for storing historical data assigned to corresponding risk factors; and an executor for calling a neural network model according to an executive operation to iteratively train the historical data of the corresponding risk factors.

[0064] In the above, the executive operation refers to simultaneously turning on the actuators provided in each iteration unit through the executive switch to perform synchronous iterative training.

[0065] In the above, the tree structure obtained by the association management unit is used to construct the pipeline failure risk evaluation model. The evaluation model has child nodes formed according to the iteration unit, and each child node forms a different node level according to the different risk probabilities. The method for constructing an evaluation model for pipeline failure risk using a tree structure specifically includes: obtaining codes and coding rules of multiple risk factors, and obtaining the corresponding relationship between each risk factor and each iteration unit; parsing the coding rules to obtain a first level bit for indicating the risk level and a first distinction bit for distinguishing the risk factors in a first coding rule, and a grading boundary bit for indicating the grading boundary result, a second level bit for indicating the risk level and a second distinction bit for distinguishing the risk factors in a second coding rule; establishing a tree structure using the first level bit, the grading boundary bit and the second level bit, and correspondingly embedding the iteration units into the tree structure in sequence according to the corresponding relationship with the risk factors; configuring a neural network model for each child node of the tree structure to construct an evaluation model for pipeline failure risk, wherein a memory and an executor are configured for each neural network model, the memory is used to store historical data assigned to the corresponding risk factor, and the executor is used to call the neural network model according to the execution operation to iteratively train the historical data of the corresponding risk factor.

[0066] It should be noted that the encoding unit also has a processing unit and an updating unit, wherein the processing unit is used to take time as a reference axis, to statistically analyze the probability time nodes of the risk factors during the operation of the pipeline, and to check the changes in the time intervals between two or more consecutive identical risk factors. When the time interval becomes shorter and shorter, it means that the probability of occurrence of this risk factor is gradually increasing. At this time, the risk level of the risk factor can be updated through the updating unit. Based on this, since the structure of the evaluation model is constructed in a tree structure based on the risk probability of the risk factor, the structure of the evaluation model can change according to the change of the risk level of the risk factor.

[0067] In the above, taking pipeline corrosion as an example, in the initial stage of pipeline construction, the probability of pipeline corrosion is almost zero, but as the pipeline operation time increases, the probability of pipeline corrosion will gradually increase. Therefore, as the pipeline operation time changes, the probability of pipeline corrosion may exceed the possibility of other probabilities. At this time, the level of pipeline corrosion risk will change.

[0068] In this specification, example embodiments have been presented in terms of a selected set of details. However, one of ordinary skill in the art will appreciate that many other example embodiments may be practiced that include different selected sets of these details. It is intended that the appended claims cover all possible example embodiments.

Claims

1. The pipeline risk assessment method based on integrity management is characterized by: The steps include: Acquire multiple risk factors of pipeline failure and the risk probability corresponding to each risk factor based on historical data, and encode them based on the risk probability corresponding to each risk factor so that the associated risk factors have the same encoding rules; The method to make related risk factors have the same coding rules is: Obtain each risk factor and the risk probability corresponding to each risk factor; The risk probability corresponding to each risk factor is extracted through the first thread, and the risk probability is sorted from large to small to form a comparison library; Extract each risk factor through the second thread, input the risk factors into the association library in sequence for risk association analysis, and confirm the risk association degree between multiple risk factors; The comparison library is called to determine whether the multiple risk factors with risk correlation are arranged from large to small according to the risk probability. If so, the multiple risk factors are graded according to the set risk level range to form a graded definition result, and the graded definition result is encoded, and the associated risk factors have the same first coding rule according to the graded definition result; if not, they are encoded according to the set risk level range, wherein the risk factors in the same risk level range have the same second coding rule; An evaluation model of pipeline failure risk is constructed using an iterative module of a tree structure, so that the evaluation model has sub-nodes formed according to coding and coding rules, and each sub-node forms a different node level according to the different risk probabilities; The method of constructing a pipeline failure risk evaluation model using a tree structure specifically includes: Obtaining the codes and coding rules of multiple risk factors, and obtaining the corresponding relationship between each risk factor and each iteration unit; Parsing the coding rules to obtain a first level bit for indicating the risk level and a first distinction bit for distinguishing the risk factors in the first coding rule, and a grading demarcation bit for indicating the grading demarcation result, a second level bit for indicating the risk level and a second distinction bit for distinguishing the risk factors in the second coding rule; A tree structure is established with the first level bit, the classification boundary bit and the second level bit, and the iteration units are sequentially embedded into the tree structure according to the corresponding relationship between the iteration units and the risk factors; A neural network model is configured for each child node of the tree structure to construct an evaluation model for pipeline failure risk, wherein a memory and an actuator are configured for each neural network model; The monitoring data of the pipeline operation is obtained, and based on the monitoring data, the pipeline failure risk during the pipeline operation is evaluated through an evaluation model.

2. The pipeline risk assessment method based on integrity management according to claim 1 is characterized in that: The risk factors include: The risk of pipeline decompression and leakage caused by exceeding the limit of pipeline flow data and pipeline pressure data; Pipeline corrosion risk formed by pipeline corrosion status and pipeline velocity data; The risk of pipeline damage caused by exceeding the limit of pipeline vibration data; Pipeline crack risk obtained from eddy current testing data.

3. The pipeline risk assessment method based on integrity management according to claim 1 is characterized in that: The encoding is performed according to a set encoding program.

4. The pipeline risk assessment method based on integrity management according to claim 3 is characterized in that: The encoding program has: The coding area is used to set the differentiation code according to the set risk level range; A first rule area is used to define a first coding pattern of risk factors having the same risk level range, and to form a second coding rule with the first coding pattern; The second rule area is used to define a second coding pattern of risk factors within the same classification definition result, and to form a first coding rule with the second coding pattern.

5. The pipeline risk assessment method based on integrity management according to claim 4 is characterized in that: The first coding pattern has a first level bit for indicating a risk level and a first distinguishing bit for distinguishing risk factors.

6. The pipeline risk assessment method based on integrity management according to claim 5 is characterized in that: The second coding pattern comprises a grading delimitation bit for indicating a grading delimitation result, a second grade bit for indicating a risk grade, and a second distinguishing bit for distinguishing risk factors.

7. The pipeline risk assessment method based on integrity management according to claim 6 is characterized in that: The iteration module has: A plurality of iteration units, each iteration unit performs an iteration operation using a neural network model and is used to iteratively train historical data of corresponding risk factors through the neural network model; The association management unit is used to correspond multiple risk factors to multiple iteration units one by one according to the coding and coding rules, and use a tree structure to arrange the hierarchy of the iteration units.

8. The pipeline risk assessment method based on integrity management according to claim 7 is characterized in that: The iteration unit has: A memory for storing historical data assigned to corresponding risk factors; The executor is used to call the neural network model according to the executive operation to iteratively train the historical data of the corresponding risk factors.

9. The pipeline risk assessment method based on integrity management according to claim 8 is characterized in that: The executive operation refers to turning on the actuators provided in each iteration unit simultaneously through the executive switch to perform synchronous iterative training.

10. The pipeline risk assessment method based on integrity management according to claim 9 is characterized in that: The tree structure obtained by using the associated management unit is used to construct an evaluation model for pipeline failure risk.

11. The pipeline risk assessment method based on integrity management according to claim 10 is characterized in that: The evaluation model has sub-nodes formed according to the iteration unit, and each sub-node forms a different node level according to the different risk probabilities.

Images