A method and system for dimensionality reduction analysis of thermal response data of medium-deep downhole heat exchangers
Through cluster analysis and retention calculation, the problem of inaccurate state of the middle and deep underground heat exchanger is solved, and more efficient data dimensionality reduction and processing are achieved, reducing data consumption.
Patent Information
- Application Number
- CN202411183240.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-27
AI Technical Summary
The state of the medium and deep underground heat exchanger is affected by a variety of factors. The traditional dimensionality reduction method cannot accurately reflect the heat exchanger status, resulting in inaccurate data analysis and large consumption of data processing time and storage space.
By obtaining the dimensional data sequence of the middle and deep underground heat exchanger, performing cluster analysis, calculating cluster expression and time expression, correcting and obtaining retention, dimensionality reduction operations are performed based on retention, and the data sequence after dimensionality reduction is generated.
It realizes more accurate reflection of the heat exchanger status, reduces database consumption, and improves data processing efficiency and accuracy.
Smart Images

Figure CN118708953B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a dimensionality reduction analysis method and system for thermal response data of a medium-deep downhole heat exchanger. Background Art
[0002] The state of medium-deep underground heat exchangers is affected by many factors, including geological conditions, groundwater temperature, heat exchanger material and structure, changes in groundwater flow rate, and instability of geological conditions. These factors have an important impact on the long-term heat transfer performance of the heat exchanger and need to be considered during the design and operation process; but a large amount of data requires a lot of storage space, and the analysis of the data also requires more time, so it is necessary to perform dimensionality reduction operations on a large number of data types.
[0003] Traditional dimensionality reduction operations mainly screen and retain data types based on the extent of change in the data type within a certain period of time. However, since there are some data types that change not due to changes in the state of the heat exchanger, the retained data types obtained by traditional methods cannot accurately reflect the state of the heat exchanger. Summary of the invention
[0004] In order to solve the above problems, the present invention provides a dimensionality reduction analysis method and system for thermal response data of a medium-deep downhole heat exchanger.
[0005] An embodiment of the present invention provides a method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger, the method comprising the following steps:
[0006] Obtaining dimensional data sequences of mid-depth downhole heat exchangers at several sampling times;
[0007] Cluster all dimensional data in the dimensional data sequence of the medium-deep downhole heat exchanger at each sampling time to obtain several clusters at each sampling time; obtain the cluster expression degree of each dimensional data in each cluster at each sampling time according to the data value change of each dimensional data in the cluster to which it belongs and the difference in data value change with other clusters;
[0008] According to the difference between the cluster expression degrees of each dimensional data in the cluster clusters at adjacent sampling moments, the moment expression degree of each dimensional data in the cluster cluster at each sampling moment is obtained; according to the moment expression degrees of different dimensional data in the cluster cluster, the moment expression degree of each dimensional data is corrected to obtain the moment expression adjustment value of each dimensional data in the cluster cluster at each sampling moment; the retention degree of each dimensional data in each cluster cluster at the last sampling moment is obtained through the moment expression adjustment value;
[0009] Based on the retention degree, a dimensionality reduction operation is performed on the dimensional data sequence of the medium-deep downhole heat exchanger to obtain the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction.
[0010] Preferably, the method of obtaining the cluster expression of each dimensional data in each cluster at each sampling time according to the data value change of each dimensional data in the cluster to which it belongs and the difference in data value change with other clusters includes the following specific methods:
[0011] For The sampling time The first Dimensional data, according to The dimension data is The data value changes in the clusters are obtained. The degree of change of the value of the dimensional data;
[0012] According to The dimension data is The difference between the data values in the cluster and other clusters is obtained. The data range gap of the dimensional data;
[0013] The first The product of the numerical change degree of the dimensional data and the data range gap is taken as the first The cluster expression of dimensional data.
[0014] Preferably, the The dimension data is in The data value changes in the clusters are obtained. The numerical change degree of the dimensional data includes the following specific methods:
[0015] The first The first The maximum value of the data of the first dimension is The first The normalized value between the minimum values of the data of the dimension data is recorded as The numerical change factor of the dimensional data;
[0016] The first The numerical change factor of the first dimension data is related to the The ratio of the maximum values of the numerical change factors of all dimensional data in the clusters is taken as the The degree of change of the value of the dimensional data.
[0017] Preferably, the The dimension data is The difference between the data values in the cluster and other clusters is obtained. The data range gap of the dimensional data includes the following specific methods:
[0018] For In all clusters except Any dimension data other than the first dimension data will be The absolute value of the difference between the maximum value of the data of the first dimension data and the maximum value of the data of the dimension data is recorded as the upper limit factor of the data range of the dimension data; The absolute value of the difference between the data minimum value of the dimensional data and the data minimum value of the dimensional data is recorded as the data range lower limit factor of the dimensional data;
[0019] The sum of the lower limit factor and the upper limit factor of the data range of the dimension data is recorded as the data range gap factor of the dimension data; In all clusters except The sum of the data range gap factors of all dimension data except the first dimension data is taken as the The data range gap of these dimensional data.
[0020] Preferably, the method of obtaining the moment expression degree of each dimension data in the cluster cluster at each sampling moment according to the difference between the cluster expression degrees of each dimension data in the cluster clusters at adjacent sampling moments includes the following specific methods:
[0021] The first The sampling time The first The first dimension data and The sampling time The first The number of data with the same data value between the two dimensional data is recorded as the amount of overlapping data at adjacent moments;
[0022] The first The sampling time The first The first dimension data and The sampling time The first The reciprocal of the absolute value of the difference in cluster expression between the dimensional data is recorded as the cluster expression difference factor;
[0023] The first The sampling time The first The product of the cluster expression degree of the dimensional data, the amount of overlapping data at adjacent moments, and the cluster expression difference factor is used as the first The moment expression of dimensional data.
[0024] Preferably, the moment expression degree of each dimension data is corrected according to the moment expression degrees of different dimension data in the clustering cluster to obtain the moment expression adjustment value of each dimension data in the clustering cluster at each sampling time, including the specific method of:
[0025] According to The sampling time The moment expression of different dimensional data in the clusters is obtained. The sampling time The abnormality of the heat exchanger of each cluster;
[0026] The abnormality of the heat exchanger is The sampling time The first The moment expression of the dimensional data is modified to obtain the The moment expression adjustment value of the dimensional data.
[0027] Preferably, the The sampling time The moment expression of different dimensional data in the clusters is obtained. The sampling time The specific methods for calculating the abnormality of heat exchangers of clusters are as follows:
[0028] The first In all clusters except The normalized value of the cumulative sum of the momentary expression of all dimensional data except the dimensional data is recorded as The abnormal factors of the heat exchangers of the clusters; 1 and The sum of the heat exchanger abnormal factors of the clusters is taken as the The sampling time The abnormality of the heat exchanger of each cluster.
[0029] Preferably, the method of obtaining the retention degree of each dimension data in each cluster at the last sampling time by expressing the adjustment value at the time includes the following specific methods:
[0030] All sampling moments The first The cumulative sum of the momentary expression adjustment values of the dimensional data is recorded as The retention factor of the dimensional data;
[0031] The first The retention factor of the first dimension data is The normalized value of the ratio between the maximum values of the retention factors of all dimensional data in the cluster is taken as the first The first The degree of retention of dimensional data.
[0032] Preferably, the dimensionality reduction operation is performed on the dimensional data sequence of the medium-deep downhole heat exchanger based on the retention degree to obtain the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction, including the specific method of:
[0033] Preset threshold parameters For any dimension data in any cluster at the last sampling time, if the retention degree of this dimension data is greater than the threshold parameter , record this kind of dimensional data as the retained dimensional data; and take the data sequence composed of all the retained dimensional data in all clusters as the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction.
[0034] The present invention also proposes a thermal response data dimensionality reduction analysis system for a medium-deep downhole heat exchanger, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of the thermal response data dimensionality reduction analysis method for a medium-deep downhole heat exchanger.
[0035] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the moment expression of each dimensional data in the cluster cluster at each sampling moment according to the difference between the cluster expression degrees of each dimensional data in the cluster clusters at adjacent sampling moments; corrects the moment expression of each dimensional data according to the moment expression degrees of different dimensional data in the cluster cluster, and obtains the moment expression adjustment value of each dimensional data in the cluster cluster at each sampling moment; obtains the retention degree of each dimensional data in each cluster cluster at the last sampling moment through the moment expression adjustment value; performs dimensionality reduction operation on the dimensional data sequence of the medium-deep downhole heat exchanger based on the retention degree, and obtains the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction; thereby, the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction can more accurately reflect the state of the heat exchanger, while reducing the consumption of the database. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0037] Figure 1 It is a flowchart of the steps of a method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger of the present invention;
[0038] Figure 2 The present invention is a characteristic relationship flow chart of a method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger. DETAILED DESCRIPTION
[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the thermal response data dimensionality reduction analysis method and system of a medium-deep downhole heat exchanger proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0041] The following is a detailed description of a method and system for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger provided by the present invention in conjunction with the accompanying drawings.
[0042] See also Figure 1 , which shows a flowchart of a method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger provided by an embodiment of the present invention, the method comprising the following steps:
[0043] Step S001: Obtain dimensional data sequences of medium-deep downhole heat exchangers at several sampling times.
[0044] It should be noted that the state of medium-deep underground heat exchangers is affected by many factors, mainly including heat exchanger operating parameters, geothermal geological parameters and pipe well parameters; these factors have an important impact on the long-term heat exchange performance of the heat exchanger, but a large amount of data requires a large amount of storage space, and analyzing a large amount of data also requires more time; so a large amount of data of different dimensions should be reduced in dimension; among them, heat exchanger operating parameters include voltage data, current data and power data; geothermal geological parameters include groundwater temperature data, groundwater flow rate data, geothermal fluid density data and thermal conductivity data; pipe well parameters include heat transfer coefficient data, heat transfer area data, heat load data, pressure drop data and thermal efficiency data.
[0045] Specifically, it is necessary to first collect the dimensional data sequence of the deep-downhole heat exchanger at several sampling times. The specific process is as follows:
[0046] Every 3 minutes is a sampling moment. Each time, the corresponding sensors are used to collect 12 types of dimensional data, including voltage data, current data, power data, groundwater temperature data, groundwater flow rate data, geothermal fluid density data, thermal conductivity data, heat transfer coefficient data, heat transfer area data, heat load data, pressure drop data and thermal efficiency data of the medium-deep downhole heat exchanger in turn. The data is collected for a total of 24 hours. The data sequence composed of all kinds of dimensional data at each sampling moment is used as the dimensional data sequence of the medium-deep downhole heat exchanger at each sampling moment.
[0047] So far, the dimensional data sequence of the medium-deep downhole heat exchanger at several sampling times is obtained through the above method.
[0048] Step S002: Cluster all dimensional data in the dimensional data sequence of the medium-deep downhole heat exchanger at each sampling moment to obtain several clusters at each sampling moment; obtain the cluster expression degree of each dimensional data in each cluster at each sampling moment according to the data value change of each dimensional data in the cluster to which it belongs, and the difference in data value change with other clusters.
[0049] It should be noted that dimensionality reduction is a method for processing high-dimensional feature data. Dimensionality reduction is to retain the most important features of high-dimensional data and remove noise and unimportant features, so as to achieve the purpose of improving data processing speed. Therefore, it is necessary to perform dimensionality reduction operations on the dimensional data sequence of medium-deep downhole heat exchangers. By clustering all dimensional data in the dimensional data sequence of medium-deep downhole heat exchangers, analysis is performed based on the clustering results, and the cluster expression of each dimensional data in each cluster cluster at each sampling time is obtained.
[0050] Preferably, in some implementations of the embodiments of the present invention, the DBSCAN clustering algorithm is a typical density-based spatial clustering algorithm, and the DBSCAN clustering algorithm can be applied to both convex sample sets and non-convex sample sets; since the significant advantages of the DBSCAN clustering algorithm are fast clustering speed and the ability to effectively process noise points and discover spatial clusters of arbitrary shapes; therefore, the DBSCAN clustering algorithm is used to cluster all dimensional data in the dimensional data sequence of the deep downhole heat exchanger at each sampling moment, and the specific method for obtaining several cluster clusters at each sampling moment is as follows:
[0051] Preset a radius parameter and a neighborhood parameter , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;
[0052] The sample space is constructed based on 12 dimensional data types, including voltage data, current data, power data, groundwater temperature data, groundwater flow rate data, geothermal fluid density data, thermal conductivity data, heat transfer coefficient data, heat transfer area data, heat load data, pressure drop data and thermal efficiency data. The parameter Eps radius value is set to , the number of points in the MinPts neighborhood is , the DBSCAN clustering algorithm is used to cluster the dimensional data sequence of the medium-deep downhole heat exchanger at each sampling moment in the sample space, and several clusters are obtained at each sampling moment.
[0053] Preferably, in some implementations of the embodiments of the present invention, since in each cluster, the data value ranges of different dimensional data are different, when the data range of one dimensional data is larger, it proves that the data of this data dimension has a greater cluster expression degree for the state of the heat exchanger expressed by the cluster to which it belongs; when a special situation occurs in the heat exchanger, the change of the dimensional data that has a greater impact on the special situation should be more obvious; for example, when a tube bundle leaks in the heat exchanger, the change range of the pressure data in the cluster reflecting the heat exchanger leakage will be larger than the change range of the groundwater temperature data and the groundwater flow rate data; that is, the data range value of the dimensional data that has a greater impact on it in the heat exchanger state is more different from the data range value of the same dimensional data in other heat exchanger states; for example, the pressure drop data range value when the heat exchanger tube bundle leaks is larger than the pressure data range value when the heat exchanger is in a normal state; therefore, according to the data value change of each dimensional data in the cluster to which it belongs, and the difference in data value change with other clusters, the specific method for obtaining the cluster expression degree of each dimensional data in each cluster at each sampling time is:
[0054] For The sampling time The first Dimensional data, according to The dimension data is in The data value changes in the clusters are obtained. The degree of change of the value of the dimensional data;
[0055] According to The dimension data is The difference between the data values in the cluster and other clusters is obtained. The data range gap of the dimensional data;
[0056] The first The product of the numerical change degree of the dimensional data and the data range gap is taken as the first The cluster expression of dimensional data.
[0057] The specific formula is: In the formula, Indicates The sampling time The first The cluster expression of the dimensional data; Indicates The sampling time The first The degree of change of the value of the dimensional data; Indicates The sampling time The first The data range gap of these dimensional data.
[0058] It should be noted that The greater the change in the value of the first dimension data, the The greater the cluster expression of the first dimension data; The larger the data range gap of the first dimension data is, the The greater the cluster expression of the dimensional data.
[0059] Preferably, in some implementations of the embodiments of the present invention, obtaining the The specific method for calculating the degree of change of the numerical value of the dimensional data is as follows:
[0060] The first The first The maximum value of the data of the first dimension is The first The normalized value between the minimum values of the data of the dimension data is recorded as The numerical change factor of the dimensional data;
[0061] The first The numerical change factor of the first dimension data is related to the The ratio of the maximum values of the numerical change factors of all dimensional data in the clusters is taken as the The degree of change of the value of the dimensional data.
[0062] The specific formula is: In the formula, Indicates The sampling time The first The degree of change of the value of the dimensional data; Indicates The sampling time The first The maximum value of the data of the dimension data; Indicates The sampling time The first The minimum value of the data of the dimension; Indicates The sampling time The maximum value of the numerical change factor of all dimensional data in a cluster; represents the linear normalization function.
[0063] Preferably, in some implementations of the embodiments of the present invention, obtaining the The specific method for measuring the data range gap of the dimensional data is as follows:
[0064] For In all clusters except Any dimension data other than the first dimension data will be The absolute value of the difference between the maximum value of the data of the first dimension data and the maximum value of the data of the dimension data is recorded as the upper limit factor of the data range of the dimension data; The absolute value of the difference between the data minimum value of the dimensional data and the data minimum value of the dimensional data is recorded as the data range lower limit factor of the dimensional data;
[0065] The sum of the lower limit factor and the upper limit factor of the data range of the dimension data is recorded as the data range gap factor of the dimension data; In all clusters except The sum of the data range gap factors of all dimension data except the first dimension data is taken as the The data range gap of these dimensional data.
[0066] The specific formula is: In the formula, Indicates The sampling time The number of clusters in the cluster The data range gap of the dimensional data; Indicates The sampling time In all clusters except The number of all dimension data except the dimension data; Indicates The sampling time The first The maximum value of the data of the dimension data; Indicates The sampling time The first The minimum value of the data of the dimension; Indicates The sampling time The first The maximum value of the data of the dimension data; Indicates The sampling time The first The minimum value of the data of the dimension; Indicates taking the absolute value.
[0067] At this point, the cluster expression of each dimension of data in each cluster at each sampling time is obtained through the above method.
[0068] Step S003: According to the difference between the cluster expression degrees of each dimensional data in the cluster clusters at adjacent sampling moments, the moment expression degree of each dimensional data in the cluster clusters at each sampling moment is obtained; according to the moment expression degrees of different dimensional data in the cluster clusters, the moment expression degree of each dimensional data is corrected to obtain the moment expression adjustment value of each dimensional data in the cluster cluster at each sampling moment; and the retention degree of each dimensional data in each cluster cluster at the last sampling moment is obtained through the moment expression adjustment value.
[0069] It should be noted that, with the update of the sampling time, when the newly added data forms a new clustering cluster, the deviation of the heat exchanger state expressed by the new and old clustering clusters represents the similarity between the added data and the original clustering cluster; when the newly added data is re-clustered, the clustering result may change; the smaller the data change between the clustering cluster where the new data after re-clustering is located and the previous clustering cluster, it proves that the heat exchanger state represented by the current data is more consistent with the cluster expression degree, so the moment expression degree of each dimension data in the clustering cluster at each sampling time is obtained by the difference between the cluster expression degrees of each dimension data in the clustering clusters at adjacent sampling times; when the dimensional data in the clustering cluster at a certain sampling time has a greater impact on the heat exchanger state, it proves that the state of the heat exchanger at the current sampling time is more abnormal, and it is necessary to obtain the moment expression adjustment value of each dimensional data in the clustering cluster at each sampling time, and then obtain the retention degree of each dimensional data in each clustering cluster at the last sampling time through the moment expression adjustment value.
[0070] 1. Obtain the moment expression of each dimension of data in the cluster at each sampling moment.
[0071] It should be noted that the data changes between the clustering cluster where the new dimensional data after re-clustering is located and the previous clustering cluster can be reflected by the cluster expression of the dimensional data in the two clustering clusters and the number of data with the same data value between the two clustering clusters.
[0072] Preferably, in some implementations of the embodiments of the present invention, according to the difference between the cluster expression degrees of each dimensional data in the cluster clusters at adjacent sampling moments, the specific method for obtaining the moment expression degree of each dimensional data in the cluster cluster at each sampling moment is:
[0073] The first The sampling time The first The first dimension data and The sampling time The first The number of data with the same data value between the two dimensional data is recorded as the amount of overlapping data at adjacent moments;
[0074] The first The sampling time The first The first dimension data and The sampling time The first The reciprocal of the absolute value of the difference in cluster expression between the dimensional data is recorded as the cluster expression difference factor;
[0075] The first The sampling time The first The product of the cluster expression degree of the dimensional data, the amount of overlapping data at adjacent moments, and the cluster expression difference factor is used as the first The moment expression of dimensional data.
[0076] The specific formula is: In the formula, Indicates The sampling time The first The moment expression of the dimensional data; Indicates The sampling time The first The first dimension data and The sampling time The first The number of data with the same data value between the dimensions; Indicates The sampling time The first The cluster expression of the dimensional data; Indicates The sampling time The first The cluster expression of the dimensional data; Indicates taking the absolute value.
[0077] It should be noted that The sampling time The first The greater the cluster expression of the dimensional data, the The greater the degree of expression of the heat exchanger state by the dimensional data; The smaller the value is, the more similar the two clusters are before and after the sampling time is updated, and the smaller the data change between the cluster where the new data after re-clustering is located and the previous cluster, which proves that the heat exchanger state represented by the current data is more consistent with the expression of the cluster.
[0078] At this point, the moment expression of each dimension of data in the cluster at each sampling moment is obtained.
[0079] 2. Obtain the time expression adjustment value of each dimension data in the cluster at each sampling time.
[0080] It should be noted that when the heat exchange temperature of the heat exchanger is abnormal, the heat exchanger load will increase, the efficiency will decrease, the fluid flow rate will decrease, and other dimensional data will change; in order to record the state of the heat exchanger at the current sampling moment more clearly, more dimensional data should be retained, and thus the moment expression of each dimensional data should be adjusted according to the current abnormality of the heat exchanger to retain more dimensional data; the more serious the current abnormality of the heat exchanger, the greater the expression of the heat exchanger state by each dimensional data should be adjusted, so the moment expression of each dimensional data is corrected according to the moment expression of different dimensional data in the clustering cluster, and the moment expression adjustment value of each dimensional data in the clustering cluster at each sampling moment is obtained.
[0081] Preferably, in some implementations of the embodiments of the present invention, the moment expression degree of each dimension data is corrected according to the moment expression degree of different dimension data in the cluster, and the specific method for obtaining the moment expression adjustment value of each dimension data in the cluster at each sampling time is:
[0082] According to The sampling time The moment expression of different dimensional data in the clusters is obtained. The sampling time The abnormality of the heat exchanger of each cluster;
[0083] The abnormality of the heat exchanger is The sampling time The first The moment expression of the dimensional data is modified to obtain the The moment expression adjustment value of the dimensional data.
[0084] In the formula, Indicates The sampling time The first The moment expression adjustment value of the dimensional data; Indicates The sampling time The first The moment expression of the dimensional data; Indicates The sampling time The abnormality of the heat exchanger of each cluster.
[0085] Preferably, in some implementations of the embodiments of the present invention, obtaining the The sampling time The specific method for calculating the abnormality of the heat exchanger of a cluster is:
[0086] The first In all clusters except The normalized value of the cumulative sum of the momentary expression of all dimensional data except the dimensional data is recorded as The abnormal factors of the heat exchangers of the clusters; 1 and The sum of the heat exchanger abnormal factors of the clusters is taken as the The sampling time The abnormality of the heat exchanger of each cluster.
[0087] The specific formula is: In the formula, Indicates The sampling time The abnormality of the heat exchanger of each cluster; Indicates The sampling time In all clusters except The number of all dimension data except the dimension data; Indicates The sampling time The first The moment expression of the dimensional data; represents the linear normalization function.
[0088] At this point, the time expression adjustment value of each dimension data in the cluster at each sampling time is obtained.
[0089] 3. Obtain the degree of retention of each dimension of data in each cluster.
[0090] It should be noted that since the main purpose of the dimensionality reduction operation is to reduce the dimension of the data while retaining the information of the original data as much as possible to reduce the computing cost, improve the efficiency and accuracy of the model, and reduce the redundancy and noise in the data; therefore, this embodiment selects and deletes the dimensional data by judging the cluster expression of different dimensional data at different sampling times to achieve the purpose of dimensionality reduction; the cluster expression of different dimensional data at a single sampling time does not fully reflect the state of the heat exchanger, and cannot accurately express the performance of different dimensional data on different states of the heat exchanger; therefore, the retention degree of each dimensional data in each cluster cluster at the last sampling time is obtained through the time expression adjustment value; all dimensional data in the cluster cluster at the last sampling time are screened and retained based on the retention degree.
[0091] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the retention degree of each dimension data in each cluster at the last sampling time by using the time expression adjustment value is:
[0092] All sampling moments The first The cumulative sum of the momentary expression adjustment values of the dimensional data is recorded as The retention factor of the dimensional data;
[0093] The first The retention factor of the first dimension data is The normalized value of the ratio between the maximum values of the retention factors of all dimensional data in the cluster is taken as the first The first The degree of retention of dimensional data.
[0094] The specific formula is: In the formula, Indicates The first The degree of retention of data in different dimensions; Indicates The sampling time The first The moment expression adjustment value of the dimensional data; represents the number of all sampling moments; Indicates The maximum value of the retention factor of all dimensional data in the cluster; represents the linear normalization function.
[0095] So far, the retention degree of each dimension data in each cluster is obtained through the above method.
[0096] Step S004: performing a dimensionality reduction operation on the dimensional data sequence of the medium-deep downhole heat exchanger based on the retention degree to obtain the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction.
[0097] Preferably, in some implementations of the embodiments of the present invention, a dimensionality reduction operation is performed on the dimensional data sequence of the medium-deep downhole heat exchanger based on the retention degree, and a specific method for obtaining the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction is:
[0098] Preset threshold parameters , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;
[0099] For any dimension data in any cluster at the last sampling time, if the retention degree of this dimension data is greater than the threshold parameter , record this kind of dimensional data as the reserved dimensional data; take the data sequence composed of all the reserved dimensional data in all clusters as the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction; store the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction in the database.
[0100] See also Figure 2 , which shows a characteristic relationship flow chart of a dimensionality reduction analysis method for thermal response data of a mid-depth downhole heat exchanger.
[0101] Through the above steps, a dimensionality reduction analysis method for thermal response data of a medium-deep downhole heat exchanger is completed;
[0102] The present invention also proposes a dimensionality reduction analysis system for thermal response data of a medium-deep downhole heat exchanger, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, steps S001 to S004 are implemented to implement a dimensionality reduction analysis method for thermal response data of a medium-deep downhole heat exchanger.
[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A dimensionality reduction analysis method for thermal response data of a medium-deep downhole heat exchanger, characterized in that: The method comprises the following steps: Obtaining dimensional data sequences of mid-depth downhole heat exchangers at several sampling times; Cluster all dimensional data in the dimensional data sequence of the medium-deep downhole heat exchanger at each sampling time to obtain several clusters at each sampling time; obtain the cluster expression degree of each dimensional data in each cluster at each sampling time according to the data value change of each dimensional data in the cluster to which it belongs and the difference in data value change with other clusters; According to the difference between the cluster expression degrees of each dimensional data in the cluster clusters at adjacent sampling moments, the moment expression degree of each dimensional data in the cluster cluster at each sampling moment is obtained; according to the moment expression degrees of different dimensional data in the cluster cluster, the moment expression degree of each dimensional data is corrected to obtain the moment expression adjustment value of each dimensional data in the cluster cluster at each sampling moment; the retention degree of each dimensional data in each cluster cluster at the last sampling moment is obtained through the moment expression adjustment value; Performing a dimensionality reduction operation on the dimensional data sequence of the medium-deep downhole heat exchanger based on the retention degree to obtain the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction; The data value changes of each dimension data in the cluster to which it belongs, as well as the data values in other clusters According to the difference in value changes, the cluster expression of each dimension data in each cluster at each sampling time is obtained, including the specific methods: For the j-th dimension data in the k-th cluster at the p-th sampling time, according to the data value change of the j-th dimension data in the k-th cluster, obtain the value change degree of the j-th dimension data; According to the difference in data value changes between the j-th dimension data in the k-th cluster and other clusters, the data range gap of the j-th dimension data is obtained; The product of the numerical change degree of the j-th dimension data and the data range gap is used as the cluster expression degree of the j-th dimension data; According to the difference between the cluster expression of each dimension data in the cluster clusters at adjacent sampling times, the The time expression of each dimension data in the cluster at the sampling time includes the following specific methods: The number of data with the same data value between the j-th dimension data in the k-th cluster at the p-th sampling time and the j-th dimension data in the k-th cluster at the p-1-th sampling time is recorded as the amount of overlapping data at adjacent moments; The reciprocal of the absolute value of the difference between the j-dimensional data in the k-th cluster at the p-th sampling time and the j-dimensional data in the k-th cluster at the p-1-th sampling time is recorded as the cluster expression difference factor; The product of the cluster expression of the j-th dimension data in the k-th cluster at the p-th sampling time, the amount of overlapping data at adjacent moments, and the cluster expression difference factor is taken as the moment expression of the j-th dimension data; The retention of each dimension data in each cluster at the last sampling time is obtained by expressing the adjustment value at the time The specific methods include: The cumulative sum of the time expression adjustment values of the j-th dimension data in the k-th cluster at all sampling times is recorded as the retention factor of the j-th dimension data; The normalized value of the ratio between the retention factor of the j-th dimensional data and the maximum value of the retention factors of all dimensional data in the k-th cluster is taken as the retention degree of the j-th dimensional data in the k-th cluster.
2. According to claim 1, a dimensionality reduction analysis method for thermal response data of a medium-deep downhole heat exchanger is characterized in that: The specific method of obtaining the value change degree of the j-th dimensional data according to the data value change of the j-th dimensional data in the k-th cluster is as follows: The normalized value between the maximum value of the j-th dimension data in the k-th cluster and the minimum value of the j-th dimension data in the k-th cluster is recorded as the numerical change factor of the j-th dimension data; The ratio between the numerical change factor of the j-th dimensional data and the maximum value of the numerical change factors of all dimensional data in the k-th cluster is taken as the numerical change degree of the j-th dimensional data.
3. The method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger according to claim 1 is characterized in that: The specific method of obtaining the data range gap of the j-th dimensional data according to the data value change difference between the k-th cluster and other clusters of the j-th dimensional data is as follows: For any dimension data except the j-th dimension data in the k-th cluster, the absolute value of the difference between the data maximum value of the j-th dimension data and the data maximum value of the dimension data is recorded as the data range upper limit factor of the dimension data; the absolute value of the difference between the data minimum value of the j-th dimension data and the data minimum value of the dimension data is recorded as the data range lower limit factor of the dimension data; The sum of the data range lower limit factor and the data range upper limit factor of the dimensional data is recorded as the data range gap factor of the dimensional data; the cumulative sum of the data range gap factors of all dimensional data in the kth cluster except the jth dimensional data is taken as the data range gap of the jth dimensional data.
4. The method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger according to claim 1 is characterized in that: The method of correcting the moment expression degree of each dimension data according to the moment expression degree of different dimension data in the clustering cluster to obtain the moment expression adjustment value of each dimension data in the clustering cluster at each sampling time includes the following specific methods: According to the time expression of different dimensional data in the kth cluster at the pth sampling time, the abnormality of the heat exchanger of the kth cluster at the pth sampling time is obtained; The moment expression degree of the j-th dimensional data in the k-th cluster at the p-th sampling time is corrected by the abnormality degree of the heat exchanger to obtain the moment expression adjustment value of the j-th dimensional data.
5. The method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger according to claim 4 is characterized in that: According to the time expression of different dimensional data in the kth cluster at the pth sampling time, the first The specific method for calculating the abnormality of the heat exchanger of the kth cluster at p sampling times is as follows: The normalized value of the cumulative sum of the momentary expression degrees of all dimensional data in the kth cluster except the jth dimensional data is recorded as the heat exchanger anomaly factor of the kth cluster; the sum of 1 and the heat exchanger anomaly factor of the kth cluster is taken as the heat exchanger anomaly degree of the kth cluster at the pth sampling moment.
6. The method for dimensionality reduction analysis of thermal response data of a medium-deep downhole heat exchanger according to claim 1, characterized in that: The dimensionality data sequence of the medium-deep downhole heat exchanger is subjected to a dimensionality reduction operation based on the retention degree to obtain the dimensionality data sequence of the medium-deep downhole heat exchanger after dimensionality reduction, including the specific method of: A threshold parameter T is preset. For any dimensional data in any cluster at the last sampling moment, if the retention degree of this dimensional data is greater than the threshold parameter T, this dimensional data is recorded as the retained dimensional data; the data sequence composed of all the retained dimensional data in all clusters is used as the dimensional data sequence of the medium-deep downhole heat exchanger after dimensionality reduction.
7. A dimensionality reduction analysis system for thermal response data of a medium-deep downhole heat exchanger, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of a dimensionality reduction analysis method for thermal response data of a mid-depth downhole heat exchanger as described in any one of claims 1-6 are implemented.
